CIGRE 467 - Economic Framework for Power Quality

467 Economic Framework For Power Quality

CIGRE/CIRED Joint Working Group C4.107

June 2011

Economic Framework for Power Quality JWG CIGRE-CIRED C4.107

Members Jose Gutierrez Iglesias (ES) - (Convener), Detmar Arlt (DE), Gerhard Bartak (AT), Math Bollen (SE),Dave Byrne (EI), David Chapman (UK), Alice Delahunty (UK), Philippe Eyrolles (FR), Elena Fumagalli (IT), Mats Hager (SE), Zbigniew Hanzelka (PL), Bill Howe (US), Rafaël Jahn (BE), Alex McEachern (US), Ian McMichael (AU), Jovica V. Milanovic (UK), Patxi Pazos (ES), Roman Targosz (PL), MarioTremblay (CN), Jasper Van Casteren (NL), Mathieu Van Den Bergh (US), Raghavan Venkatesh (IN), Paola Verde (IT)

Copyright © 2011 “Ownership of a CIGRE publication, whether in paper form or on electronic support only infers right of use for personal purposes. Are prohibited, except if explicitly agreed by CIGRE, total or partial reproduction of the publication for use other than personal and transfer to a third party; hence circulation on any intranet or other company network is forbidden”. Disclaimer notice “CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties and conditions are excluded to the maximum extent permitted by law”.

ISBN: 978- 2- 85873- 157-2

EXECUTIVE SUMMARY Electric power quality disturbances can have significant economic consequences for many different types of facilities. Although power quality is widely recognized as an important issue, there is no consensus on its total economic impact. Indeed, there is not even consensus on how to measure the impact. A wide range of potential solutions, with varying degrees of cost and effectiveness, are available to mitigate the consequences associated with poor power quality. Power quality solutions can be applied at different levels or locations within the global electrical system. The evaluation of power quality improvement alternatives is an exercise in economics. Facility managers and utility engineers must evaluate the economic impacts of the power quality variations against the costs of improving performance for the different alternatives. The best choice will depend on the costs of the problem and the total operating costs of the various solutions. In general, the costs of these solutions increase as the power level of the load that must be protected increases. This means that economies usually can be achieved if sensitive equipment or controls can be isolated and protected individually from equipment that does not need protection. Each solution technology needs to be characterized in terms of cost and effectiveness. In broad terms, the solution cost should include initial procurement and installation expenses, operating and maintenance expenses, and any disposal and/or compensation value considerations. Improving facility performance during power quality variations can result in significant savings and can be a competitive advantage. Therefore, it is important for customers and suppliers to work together in identifying the best alternative for achieving the required level of performance. A methodology for performing a comparative economic analysis is featured in this report. A joint working group, JWG C4.107, has been formed between CIGRE (electric power transmission emphasis) and CIRED (electric power distribution emphasis) to develop a systematic approach to this issue. The JWG works to develop a framework for analysis of the economics of power quality, and has created a bibliography of existing references. However, gathering specific values and data to assess the economics of power quality is beyond the scope of the Group; the work will be limited to developing a framework. Different technologies are evaluated by estimating the improved performance that can be expected after the technology has been applied. The power quality cost savings are calculated for each technology along with the costs of applying the technology. JWG C4.107 aimed to produce this report that summarizes available information about cost-benefit analysis of power quality, and to propose a framework for how to assess costs, how to assess the economic impact of mitigation, and how to assess the economic impact of immunity.

2

TABLE OF CONTENTS EXECUTIVE SUMMARY ........................................................................................................................2

1. Introduction to the Economics of Power Quality .............................................................................5 1.1. SCOPE OF THE REPORT ...................................................................................................................5 1.1.1. The Economic Importance of Power Quality............................................................................5 1.2. ECONOMIC CONSEQUENCES OF POOR PQ FOR END USERS ..................................................6 1.3. ECONOMICS OF POWER QUALITY FOR POWER NETWORKS .................................................7 1.4. ECONOMICS OF POWER QUALITY FOR SOCIETY .....................................................................8 1.5. ROLE OF REGULATION....................................................................................................................9 1.6. OVERVIEW OF THE DOCUMENT .................................................................................................10

2. Overview of Methodologies for Assessment of Economic Impact – End User Perspective.........12 2.1. METHODOLOGY FOR QUANTIFYING THE ECONOMIC IMPACT OF VOLTAGE SAGS AND SHORT INTERRUPTIONS.......................................................................................................................12 2.1.1. Introduction.............................................................................................................................12 2.1.2. Overview of Existing Methodologies .....................................................................................12 2.1.3. IEEE Guidelines......................................................................................................................14 2.1.4. Analytical Economic Analysis................................................................................................14 2.1.5. Indirect Economic Analysis ....................................................................................................17 2.1.6. Reported PQ-Related Losses from Around the World............................................................19 2.2. METHODOLOGY FOR QUANTIFYING THE ECONOMIC IMPACT OF HARMONICS ...........23 2.2.1. Introduction.............................................................................................................................23 2.2.2. Overview of Existing Methodologies .....................................................................................24 2.3. METHODOLOGY FOR QUANTIFYING THE ECONOMIC IMPACT OF OTHER PQ PHENOMENA...........................................................................................................................................29 2.3.1. Voltage and Current Unbalance ..............................................................................................29 2.3.2. Surges and Transients .............................................................................................................34 2.3.3. Flicker .....................................................................................................................................35 2.4. CONCLUSIONS .................................................................................................................................36

3. Overview of Existing Methodologies for Assessment of Economic Impact – Public Distribution Network Perspective.................................................................................................................................37 3.1. INTRODUCTION...............................................................................................................................37 3.2. REVIEW OF LITERATURE AND DOCUMENTED METHODOLOGIES.....................................38 3.3. COSTS ASSOCIATED WITH PQ .....................................................................................................38 3.3.1. Costs Incurred by the Utility to Mitigate PQ Issues................................................................39 3.3.2. Costs Associated with Improving Reliability but not PQ .......................................................45 3.3.3. Costs for Responding to PQ Issues .........................................................................................46 3.4. SUMMARY ........................................................................................................................................47 3.5. CONCLUSIONS .................................................................................................................................47

4. Methodology for Collecting Power Quality Economic Data..........................................................49 4.1. INTRODUCTION...............................................................................................................................49 4.2. IMPORTANCE AND MOTIVATION ...............................................................................................49 4.3. END-USER PERSPECTIVE ..............................................................................................................50 4.3.1. Technical Data ........................................................................................................................50 4.3.2. Economic Data........................................................................................................................52 4.4. DNO PERSPECTIVE: DATA COLLECTION ..................................................................................60 4.5. CONCLUSIONS .................................................................................................................................62

5. Methodology for the Economic Assessment of Power Quality Solutions......................................63 5.1. INTRODUCTION...............................................................................................................................63 5.2. ECONOMIC ANALYSIS OF THE COSTS OF PQ...........................................................................63 5.2.1. Economic Impact of Power Quality Variations ......................................................................63 5.3. END-USE PQ SOLUTIONS...............................................................................................................71 5.4. CHOOSING THE OPTIMAL PQ SOLUTION ..................................................................................79 5.5. CONCLUSION ...................................................................................................................................80 APPENDIX 1 ............................................................................................................................................81

3

A

COMMON PQ PHENOMENA.......................................................................................................81 A.1 Categories of Power Quality Variations .............................................................................81 B RESPONSE OF SENSITIVE EQUIPMENT TO PQ EVENTS......................................................87 B.1 Data Processing and Communications Equipment .............................................................87 B.2 Variable-Speed Drives........................................................................................................88 B.3 Lighting ..............................................................................................................................88 B.4 Solenoid-Operated Contactors............................................................................................89 C ADDITIONAL LOSSES CAUSED BY POOR PQ ........................................................................89 C.1 Cables .................................................................................................................................89 C.2 Transformers.......................................................................................................................90 C.3 Motors ................................................................................................................................90 APPENDIX 2 ............................................................................................................................................91 A OVERVIEW OF INTERRUPTION COST CALCULATION........................................................91 B PROBABILISTIC VOLTAGE DIP COSTS CALCULATION......................................................92 C OVERVIEW OF EQUIPMENT SENSITIVITY.............................................................................92 D UNCERTAINTY INVOLVED WITH EQUIPMENT SENSITIVITY..........................................93 E COUNTING OF PROCESS TRIPS ................................................................................................94 F COST ASSESSMENT.....................................................................................................................96 G NUMERICAL EXAMPLE.............................................................................................................96 I TYPICAL LOSS VALUES ...........................................................................................................102 J TYPICAL FINANCIAL LOSS VALUES - SUMMARY.............................................................108 K FORMULAE FOR COMPUTING HARMONIC LOSSES FOR THE MAIN ELECTRICAL COMPONENTS.......................................................................................................................................113 L METHODS FOR PROBABILISTIC EVALUATIONS................................................................116 APPENDIX 3 ..........................................................................................................................................123 A COST ASPECTS ...........................................................................................................................123 B HYDRO-QUEBEC-IREQ REPORT FOR ECONOMICAL ASPECT OF HARMONICS ON DISTRIBUTION AND TRANSMISSION SYSTEM .............................................................................128 B.1 Harmonics Power Losses Evaluation .......................................................................................128 B.2 Harmonics Losses Evaluation ..................................................................................................128 B.3 Harmonic Losses Cost Evaluation............................................................................................129 B.4 Conclusion................................................................................................................................129 APPENDIX 4 ..........................................................................................................................................131 A STRUCTURING THE DATA COLLECTION PROCESS ..........................................................131 B EXECUTING DATA COLLECTION PROCESS – END USER PERSPECTIVE ......................133 C CONCLUSIONS ...........................................................................................................................134 APPENDIX 5 ..........................................................................................................................................136 A ILLUSTRATIVE CASE STUDY .................................................................................................136 A.1 Base Case: Facility Data and Base Case Calculations......................................................136 A.2 Case 1: Redundancy in the Utility Supply........................................................................137 A.3 Case 2: Applying a Battery UPS ......................................................................................138 A.4 Case 4: Using Distributed Energy Resources (DER) .......................................................140 B CASE COMPARISON AND SENSITIVITY ...............................................................................141 REFERENCES .......................................................................................................................................143 ACKNOWLEDGMENTS......................................................................................................................150

4

1. Introduction to the Economics of Power Quality 1.1. Scope of the Report Various independent studies have been undertaken by power companies, consultants, regulators, and research organizations to estimate the cost of power quality problems to the power companies and their customers. A good understanding of the basis for determining these costs is important in assessing appropriate interventions (either by the distribution network operator or by the customers themselves). A joint working group has been launched with CIRED, where the question has been a subject of a 2001 round-table discussion. This is supported by the convener of CIRED S2 (EMC & Power Quality). The scope of the proposed JWG was to: •

Review and document the economic implications of the power quality parameters: voltage dips, short interruptions, and voltage waveform quality. Long interruptions were not considered.

•

Review and document methods of assessing these costs that have been used to date, including aspects such as: I. Direct and indirect costs to customers (e.g. production losses and plant damage). II. Energy losses associated with poor power quality. III. Cost of energy not supplied. IV. Methods of collecting customer costs. V. Actual customer costs collected to date for various industry sectors.

•

Propose a standardized method of collecting the above information, based on the experience of various international studies.

•

Recommend a methodology of using this data to cost and motivate power quality interventions on the power system or within the customer plant.

•

Provide indicative costs for specific industry sectors, where possible.

Many professionals, including industry regulators, consultants, system and installation designers, maintenance managers, production managers, and financial mangers, are concerned about the impact of the costs of poor power quality on businesses and how these costs can be managed. Techniques for avoiding or reducing the impact of power quality issues are well known and the cost of their deployment relatively easily determined. However, assessing the potential cost impact of power quality (PQ) issues is difficult because, for example, the incidence of problems, the response of equipment, and the effect on process continuity are statistical in nature and are difficult to quantify. Although there have been numerous case studies, there has been, so far, no consensus on how the calculation or assessment of these costs should be approached. This report provides a methodology for examining the economic framework for PQ. It will enable all interested parties to establish costs and benefits of PQ improvement and mitigation measures in a consistent and open manner.

1.1.1. The Economic Importance of Power Quality “Power quality” is the term generically used to describe the extent to which the electrical power available at the point of use is compatible with the needs of the load equipment connected at that point. The effects of a lack of compatibility are termed PQ problems or PQ issues. Compatibility is a two-sided equation because both the characteristics of the electrical power supply AND the sensitivity of the load equipment are important variables.

5

There are many parameters for which compatibility is necessary, including supply voltage level, voltage stability, waveform distortion due to harmonics and interharmonics, voltage unbalance between the phases, and long- and short-term availability of the supply. When there is a lack of compatibility, end-user equipment may cease to function, may operate erratically or incorrectly, or may operate outside its normal envelope at reduced efficiency or in such a way that its operating life is reduced. The situation is further complicated by the fact that many PQ issues are caused by the operation (or mis-operation) of end-use equipment that is connected to the network. Electrical and electronic equipment rarely operate in isolation. Even the simplest of commercial operations requires the interoperation of several items of equipment—the use of a personal computer usually requires the aid of some communications equipment, a network, and a printer, for example. In other words, the failure of one piece of equipment usually results in the failure of a process that may or may not affect other processes. Regardless, however, when process equipment ceases operation, the result is a financial loss. Depending on factors such as the nature of the business, the organization of the work flow (whether continuous processing or batch production) and the value of the product, this loss may range from the trivially small to the extremely large. There are two obvious approaches to ensuring 100% compatibility between electric power supply and end-use loads: Either design and construct a perfect electric power delivery grid, or make all end-use devices perfectly tolerant of common PQ issues. Unfortunately, for a number of reasons, neither of these approaches represents the economic optimum. Firstly, some loads are relatively insensitive to many PQ phenomena while being rather sensitive to others. Incandescent lighting is insensitive to harmonic distortion but overly sensitive—in combination with the human response—to flicker. On the other hand, electronic equipment is not disturbed by the scale of voltage instability that causes flicker; however, it is very sensitive to voltage dips and to higher levels of harmonic voltage distortion. Making every supply suitable for every load would be expensive and is unnecessary. Secondly, although the cost of designing and manufacturing any individual piece of equipment to be “immune” is not large, that cost is multiplied by the total number of pieces of equipment in use and represents a very large economic burden on consumers. Thirdly, the option of building a very robust, very clean power system would be extremely high and it would be very difficult, if not impossible, to guarantee a minimum performance level at all points of common coupling. Increased penetration of distributed generation will make this even more difficult as generation is added at medium and low voltage levels. Lastly, many PQ issues arise within the consumer’s premises, due to the characteristics of the installed equipment, sub-optimal installation of equipment and cabling, poor electromagnetic compatibility (EMC) performance of earthing systems, etc., so perfection at the point of common coupling is no guarantee of adequacy at the point of use.

1.2. Economic Consequences of Poor PQ for End Users From the descriptions of equipment responses, it is apparent that the economic consequences of poor PQ fall into three broad categories: • • •

Complete or partial loss of one or more processes (e.g. loss of control following a dip) Poor long-term productivity or product quality (e.g. as a result of operator fatigue due to flicker) Increased costs due to reduction of equipment lifetime resulting in premature failure (e.g. overheating of transformers due to harmonics)

These consequences take effect over very different time scales. A process failure, triggered by a PQ event such as a dip, has immediate consequences followed by a period of recovery, during which further costs may be incurred. It is relatively easy to identify the costs that result from a single event or to predict what the costs might be.

6

Continuous or prolonged conditions, such as flicker, can reduce long-term productivity. If the problem is prolonged and widespread, the business may become uncompetitive and may require additional borrowing to sustain it. Premature failure of equipment will usually result in process failure with similar consequences to those caused by single PQ events. The difference is that the causes are in the past and were unrecognized, suggesting that predicting these costs is difficult unless a survey procedure is put in place. Depending on the type of operation in question, the economic consequences may range from trivial to catastrophic. The user can take several approaches: • • • •

Do nothing, and suffer the consequences. Take responsibility by adding mitigation equipment or hardening measures within the installation. Work with a supplier to improve the level of PQ by local measures specifically for, but external to, the installation. Negotiate with a supplier for a guarantee of a defined level of PQ, along the lines of an insurance policy.

“Doing nothing”—business as usual—is viable only for those enterprises that use batch processing for manufacturing and data handling. Process interruptions are limited in their impact and are relatively easily mitigated by, for example, reorganizing work schedules. The economic consequences are not zero but are acceptable to the business. In every other case, the first steps in analysis of the economic impact of PQ on a particular organization or part of an organization include: • • • • •

Obtaining a thorough and continuing measurements of relevant PQ parameters Logging of process failures and their costs and relating their occurrence to PQ events Assessing the likely failure modes and failure rates of processes and items of equipment, bearing in mind the different time scales involved Considering options for redesigning processes to reduce interdependence and reduce the risk of cascading failures Investigate options for hardening process equipment against PQ events and conditions

1.3. Economics of Power Quality for Power Networks Network operators are usually required to maintain a certain quality of service to end-users by local legislation or regulation. Quality of service may be defined by a number of parameters, such as availability and voltage stability. Achieving the required level requires that the operator invest in, for example: a. b. c.

A monitoring program to identify potential failures (e.g. in transformers) so that repair or maintenance work can be planned, and unplanned downtime can be avoided Careful planning of maintenance to avoid excess unavailability Maintenance work to avoid damage to lines, such as tree cutting programs

Network operators need to ensure that consumers are connected appropriately (e.g. at a suitable voltage level) to avoid negative impacts on other local consumers from excessive harmonic currents or voltage disturbances. This usually involves offering consumers pre-connection support so that such issues can be avoided. Some of the measures taken by utilities and consumers to reduce the impact of power quality issues require the installation of additional equipment. Apart from the obvious cost issue, this equipment has environmental consequences; electrical energy efficiency is reduced. and the additional equipment consumes materials and energy for its manufacture. The so-called externalities need to be taken into account.

7

1.4. Economics of Power Quality for Society While the economic impact of power quality for users and for utilities is readily identifiable, the impact on society in general is less tangible. In the recent years, the thrust on sustainable development, clean development mechanisms, and various other global (green) initiatives have necessitated extrapolating the local and short-term aspects to the globe and long-term impacts to identify the true holistic impact of industrial activities. Specific to energy, the focus is shifting from local short-term to global long-term and while computing the impact of energy conservation it makes sense to consider the energy leverage. Electrical energy is only an intermediate form of energy used only for bulk power transmission. Due to this, the role of electrical energy is very critical, and due to the various transformations taking place right from environment to end use, the impact of energy conservation at end use assumes a very high significance. Considering a simple case of power quality improvement (such as power factor improvement or harmonic mitigation), the benefits to various players at different levels are as follows: • • • •

End user – Reduction in utility bill, direct economic benefits Utility – Reduction in T&D loss, better asset management, higher operational efficiencies Power generator – Better asset management, higher operating efficiency Society – Lower carbon foot print, reduced global warming, sustainable development

The investment to improve the power quality could have been made at any level, by any player, but if the holistic benefit is considered, the investment decision is expected to appear better, which is more realistic considering the societal benefits. The need is to drive decision making based on long-term global impacts rather than local short-term benefits. This is expected to influence decision making, and in most cases, the benefit of PQ improvement is expected to be more than what is being considered at present. This is expected to help in selection of an optimal/appropriate solution for a specific PQ issue and in general to make power quality improvements more attractive. While methodologies can be developed to factor the long-term societal impact of power quality and deployment of power conditioning solutions, it is also important to develop a framework that will ensure proper application of the methodology. Players in the arena: • Utility • Users • Manufacturers of equipment • Regulators The roles are: • Good voltage quality at the customer bus is the utility’s responsibility. • Good quality for load current drawn from the bus in the customer’s responsibility. • Developing and supplying cost-effective power conditioning devices and equipment with adequate tolerance to power quality with appropriate technology are the manufacturers’ responsibility. • Ensuring an efficient balance of responsibilities is the role of the regulator. In short, the responsibility of considering long-term societal impact has to be with the regulator, and this framework can be used to: • Drive investment decisions considering long-term societal impact rather than short-term local benefits. This is mainly applicable for what a utility spends on improving power quality. • Drive government policies and investment decisions. • Formulate tariff guidelines as to capture the true cost of power quality, and this indirectly influences power quality improvement initiatives.

8

• •

Formulate standards and guidelines, policies, and power quality norms and ensure compliance based on global impact of power quality. Influence equipment standards as to enhance the compatibility and performance of equipment.

1.5. Role of Regulation Because of the sensitivity of end-user equipment, voltage quality is of primary concern to industrial users, service providers, and especially for parameters such as supply voltage variations, even household consumers. In particular, productivity and competitiveness of the manufacturing and service sectors depend highly on the quality of the electricity supply. Indeed, even after the liberalization of the electricity market, the quality of electricity mainly relies upon transmission and distribution networks, i.e. upon investments and operation practices of regulated firms. As a consequence, voltage disturbances are a crucial issue not only for distribution network operators, transmission system operators, manufacturers of electric appliances, and designers of electrical installations, but also for energy regulators. For example, after years of work devoted primarily to commercial quality and continuity of supply, European energy regulators are becoming increasingly involved with the regulation of voltage quality. However, regulation in this area encounters a main difficulty. Because power quality results from the interaction between the network and the customers’ equipment, a tradeoff exists between investments in the network and higher immunity levels for end-user equipment. From a regulatory perspective, performance standards on equipment immunity are to be defined in close relationship with voltage quality requirements for power networks. In this sense, the proposed introduction of the concept of “responsibility sharing” in technical standards (and in particular in the EN 50160) is fundamental to enable energy regulators to define enforceable requirements for all stakeholders [1]. This idea is easily explained with an example: in South Africa, according to the National Standard NRS 048-2:2003, customer installations are expected to tolerate voltage dips with residual voltage over 70% with duration up to 150 ms, over 80% up to 600 ms, and over 85% for longer durations. For all other dips, the allowed number of events is limited by the National Standards [1,2]. In defining a responsibility-sharing curve, duties and rights of all parties should be taken into account. In other words, the choice of a responsibility-sharing curve should be the result of an agreement between network operators, final customers, equipment manufacturers, and energy regulators. As a result, performance requirements given by regulators will not be in conflict with other technical standards, for instance product, emission, and immunity standards. Moreover, these requirements will not impose unsustainable costs on any stakeholders. European energy regulators have begun to work in this direction, under the aegis of the European Regulatory Group for Electricity and Gas (ERGEG). In fact, ERGEG has already suggested to introduce several revisions to the EN 50160, among which is the introduction of a responsibility-sharing curve [3]. Although the existence of international standards is important, to design a workable regulatory framework, regulators need information on both consumer costs for voltage disturbances as well as the level of voltage quality provided on distribution networks and the cost of providing that quality. Indeed, regulatory standards, for instance on the number of events, should be developed at the national level and allow for differentiations according to network structures, protection schemes, and characteristics of withdrawal. Several European regulators are already engaged in monitoring power quality levels and assessing customer cost. Others, such as the energy regulators of Norway, Hungary, and France, already enforce voltage quality standards, which, in a number of cases, are more demanding than the values given in the EN 50160. The work of European energy regulators on voltage quality is thoroughly described in CEER (Council of European Energy Regulators) Benchmarking Reports on Electricity Supply [4,5,6,7]. Altogether, significant developments are expected in this area of regulation in the incoming years.

9

EMC COORDINATION EMISSION IMMUNITY

1.6. Overview of the Document Chapter 2 provides a general overview of established and new methodologies used to assess the financial losses incurred to industries by power quality (PQ) disturbances, in particular voltage dips, short interruptions, and harmonics. The first part focuses on quantifying the economical damage suffered by industrial customers due to nuisance process trips induced by voltage dips and short interruptions. For this purpose, guidelines provided by IEEE standards are discussed and critically assessed to reveal their major strengths and weaknesses. Next, methodologies recently proposed by researchers for financial loss assessment of voltage dips and short interruptions are gathered and discussed. It is generally found that accurate assessment involves careful consideration of three major factors: voltage dip profile at the busbars involved, customer load susceptibility, and calculation of the losses induced by process interruption. Finally, representative studies conducted in Europe, U.S., and Asia are investigated, with their findings and reported losses presented, to demonstrate the scale of the losses. The second part of the chapter deals with methods and techniques used to economically quantify the effects of harmonics on electrical systems. The economic evaluation includes the increased losses, the premature ageing and the malfunction of the equipment present in the system. The economical value of the losses and premature aging versus the harmonic pollution level can give indications of the amount of costs for equipment to be met/saved for a given increase/decrease of harmonics. Regarding the malfunction, its economical value requires computing the effects of the malfunction on the process in which the equipment is inserted. The analysis in most cases can be conducted adopting the methods valid in the same case as of other disturbances, like micro-interruptions or voltage dips. Finally, the economic consequences of unbalanced voltages and flicker are discussed. Chapter 3 examines the economics of PQ from the network operator’s point of view. It identifies the costs involved in three categories: providing a response to customer issues; mitigation of PQ issues by network design, asset management, and maintenance; and the measures to ensure reliability. Indicative costs of measures are given where appropriate and existing methods of collecting data are reviewed. There is not a great experience from PQ-projects, methods and experiences with the assessment of financial losses due inadequate quality of electricity supply. Faults within an industry supply, or other installation, will cause voltage dips in the local supply system, but in most cases these will not propagate upstream and affect other customers supplied from the same HV- system. When immunity requirements and equipment immunity are discussed for a specific installation and/or process, both sources of dips (faults in the HV supply and “internal” faults within a specific industry) must be considered and equipment performance must be chosen with respect to the actual electrical environment. It is quite obvious that the reduction of faults within the customer’s own

10

plant is the owner’s responsibility, and so is the interest to minimize economical losses due to consequences of all faults. Chapter 3 discusses how to find the total cost contribution of dips due to faults in the distribution and transmission systems affecting several customers. If such “total socioeconomic cost” can be provided (or at least a method of how to calculate it), identifying the most cost-effective mitigation can include actions in the distribution and transmission systems, and perhaps open a wider perspective to the “economics” of voltage-dip immunity and emission. Chapter 4 discusses the methodology for the collection of cost data in four parts: the cost of process interruptions resulting from discrete PQ events, the cost of operation and maintenance of mitigation equipment and of reduced energy efficiency of equipment, the cost of reduced equipment lifetimes, and the capital and installation cost of mitigation equipment. To perform a PQ events cost analysis, information is needed regarding improving/mitigating cost at: • Customer (equipment/installation) level • Utilities (network) level Measurement of power quality can be focused on: • The total number of disturbances in networks, for benchmarking purposes. Thus not all disturbances affect equipment. • Only on the number of disturbances affecting equipment/processes. It is necessary to limit the scope for PQ measurements to perform. Are costs accurate and well founded? Are costs based mainly on broad assumptions and only backed up by sparse data? It is also important to understand to whom data information will be presented. Readers could be: customers, utilities, regulators, manufacturers, and also standardization bodies and government institutions. Chapter 5 provides a methodology for the economic assessment of PQ solutions. It proposes calculation of the net present value of PQ investments, which is calculated using a nominal ten-year lifetime. A case study is described that illustrates practical application on the method. The costs to industrial and commercial electric power end users from unmitigated PQ and reliability phenomena are significant and have been well documented by detailed studies These studies have focused principally on quantifying the actual or reported cost to businesses of PQ and reliability phenomena that result in unplanned businesses losses brought about by such factors as process interruptions, equipment damage, extra labor costs, and increased scrap. Although many of these studies also inquire about mitigation equipment employed by end users to try to minimize the business impact of PQ and reliability phenomena, in general, the numbers given for the “cost of PQ and reliability” focus only on the impact of unmitigated phenomena and exclude the cost of preventing unplanned business losses. As such, an unprotected facility might be said to suffer significant PQ and reliability costs, while a facility protected with, say, a double-redundant uninterruptible power supply (UPS) and N+1 backup generation might be said to suffer no PQ or reliability costs whatsoever—a circumstance that does not reflect true business decision-making wherein the costs of outages are balanced against the costs of mitigation. Because of this, a comprehensive strategy to evaluate optimization of overall PQ-related cost is needed, including: •

Costs to industry and electric power providers based on unmitigated PQ phenomena

•

Costs to industry and electric power providers based on prevention and mitigation of the impacts of PQ phenomena

The key challenge is to balance both of these broad cost categories. Although any number of economic analysis approaches may be employed to arrive at an optimum, this chapter emphasizes a simple 10-year net present value (NPV) approach whereby all costs and benefits may be combined to determine the mitigation scenario that optimizes today’s economic performance.

11

2. Overview of Methodologies for Assessment of Economic Impact – End User Perspective 2.1. Methodology for Quantifying the Economic Impact of Voltage Sags and Short Interruptions 2.1.1. Introduction Voltage dips and short interruptions are major contributors to economic losses incurred by end users in terms of power quality (PQ)-related costs. Although they are not as detrimental as long-duration outages, voltage dips and short interruptions (up to 3 minutes) occur much more frequently. Generally, economic losses are incurred when the supply voltage disturbances cause nuisance trips or malfunction of sensitive equipment, which in turns affects (or completely interrupts) the production processes or service. In case of large industrial and commercial customers, the cost of process disruption can be very high. Reliable information regarding the economic losses incurred because of voltage dips and short interruptions is essential to both customers and the utility. It provides the very basis for cost-benefit analysis for all potential investments in mitigating solutions. The actual incurred economic losses, however, are customer-specific and depend on many factors including customer category (industrial, commercial, etc.), type and nature of activities interrupted, the extent of the interruption (both duration and number of activities interrupted or affected), etc. Therefore, estimating the economic impact of voltage dips and short interruptions is a daunting task that requires careful consideration of many technical and non-technical aspects and usually requires significant deployment of personnel and resources. Over the years, numerous studies have been performed to address the problem of economic losses incurred by end users by supply voltage disturbances. This chapter presents a general overview of existing methodologies to assess the economic losses caused by voltage dips and short interruptions. It also presents a summary of the results of cost estimates from various projects carried out around the world.

2.1.2. Overview of Existing Methodologies Assessing the cost of voltage dips and interruptions is a cumbersome task. Over the years, a number of projects attempted to establish the value of economic consequences of voltage dips. Some proposed methods to obtain network level [8, 9] losses due to voltage dips, while others concentrated on customer plant-level losses [10-14]. Regardless of the level (network or individual plant) that the studies were focused on, precise information about the cost of a single process failure/malfunction is essential for the overall accurate assessment of economic losses. Generally, to accurately assess the economic losses due to these disturbances, a thorough understanding of customer plant and processes involved is essential. Therefore, a good methodology has to take into account all the aspects involved, from in the supply voltage profile at the point of connection of an industrial plant, to equipment and process sensitivity to voltage dips and to all direct and indirect costs associated with process disruption. A voltage-dip profile at the customer busbar provides the information regarding the frequency and characteristics of voltage dips and short interruptions. Normally, this information is obtained from historical data or from site monitoring. However, when there are no records available, or when the monitoring period is too short to draw sufficient conclusions, a voltage-dip profile has to be predicted. The fault positions method [8, 9, 15, 16] is a common method used to determine the expected characteristics of voltage dips resulting from for the short-circuit faults in the network. Statistical processing of existing (limited-duration) monitoring records [12], e.g. extrapolation of data, can also be used to predict future dip performance. The sensitivity/resilience of equipment used in industrial processes to voltage dips and short interruptions directly influences the response of the industrial process to incoming voltage dips and interruptions, and therefore has direct impact on the resulting financial losses. The sensitivity of equipment is normally

12

expressed in terms of the magnitude and duration of the voltage dip. The voltage-tolerance curve for an individual device (equipment) can be obtained either from the equipment manufacturer, from laboratory or field measurements, or if none of the previous is available, a voltage-tolerance curve from existing standards could be used (at least as a guidance). The commonly used standards for characterizing equipment sensitivity are the Computer Business Equipment Manufacturers Association (CBEMA) curve, Information Technology Industry Council (ITIC) curve, and the “semiconductor processing” (SEMIF47) curve [17]. Because different types of equipment exhibit different sensitivities to voltage dips and short interruptions, equipment-specific voltage-tolerance curves [15, 18-20] have been and are being developed from laboratory tests. According to the IEEE Standard 1346-1998 [19], there is a range of uncertainty in the magnitude– duration plane associated with voltage-tolerance curves. To account for this uncertainty, various methods have been developed in the past and used in assessments of equipment sensitivity to voltage dips, including probabilistic methods [15, 16, 21], fuzzy logic [22], and voltage-dip severity indices [23]. On a higher level, process sensitivity depends on many factors, including but not limited to equipment interconnections, composition ratio of equipment, function and significance of each equipment type, and the relationship between equipment failure modes and process operation. To address these factors, various approaches have been attempted by researchers around the world. The approaches include probabilistic methods [15, 16, 20, 21], fault tree analysis [24, 25], fuzzy logic [25], loss of voltage during dip [26], loss of energy during dip [26], and one-parameter characterization method [26]. Once the information about the voltage-dip profile and customer process sensitivity is available, the number of process failures or malfunctions can be determined. Following the estimation of expected number of process failures, the next step is to determine the economic losses associated with each of them and to add up losses associated with individual events in order to come up with the annual plant exposure. Detailed methodologies for calculation of the costs associated with voltage dips have been proposed in [27, 28]. Cost calculation involves careful investigation of all direct and indirect costs caused by voltage dips. Theoretical and mathematical formulae are derived to represent various causes of losses. The cost functions of equipment, sub-processes, and processes are then incorporated into the technical states of the processes to determine the costs of each process and the plant. Determining the cost of voltage dips and interruptions based on previously described calculations would generate a very accurate cost estimation for every dip providing that reliable input data is available. The drawback is that one would require a significant amount of information regarding all direct and indirect costs for every individual sub-process in the plant. These cost figures, however, are very difficult to obtain without a time-consuming detailed investigation, which often involves confidentiality issues. Alternatively, some studies relate the economic losses incurred by voltage dips with customer interruption cost (CIC). CIC is the economic damage to customers caused by power interruptions (outages) of a specified duration. Customer damage functions due to power interruptions are well studied and reasonably well documented, and thus provide a convenient reference for voltage dip-related cost analysis. Basically, CIC can be obtained from survey results obtained from a large number of customers of various industrial sectors. This information is then analyzed, aggregated per sector, and averaged to give plantlevel costs per voltage dip or cost per kW of power per voltage dip (normalized cost) [10] for various industrial sectors. Studies that use cost per event for voltage-dip economic analysis include [8, 9, 15, 16, 29], while [10, 18, 29] use cost per kW power per voltage dip. A PQ index that uses CIC/kWh for financial loss assessment is proposed in [30]. Weighted cost per dip method is used [7, 21, 24, 31-33] where different weighting factors are assigned to different magnitudes of voltage dips. In this way, the cost of severe dips are equal to the cost of interruptions, while less-severe dips incur a fraction of the cost of interruption. There are also indirect ways to estimate the economic impact of voltage dips and short interruptions. Some studies [14] use the power of a customer plant as basis for cost evaluation. The losses incurred

13

because of voltage dips and short interruptions are estimated as a percentage of the annual cost of power consumption. Other common methods of indirect economic analysis are the willingness to pay (WTP) and willingness to accept compensation (WTA) methods [34].

2.1.3. IEEE Guidelines The IEEE Standard 1346-1998 [19] provides guidelines to assess economic losses at customer facilities due to voltage dips. The aspects considered there include the voltage-dip performance at the utility and the industrial plant level, the equipment susceptibility to voltage dips, and economic evaluation of the losses incurred because of process-disruptive dips. A method of representing a voltage-dip profile at the customer facility using contour lines and comparing it with equipment voltage-tolerance curves to obtain the number of disruptive dips is presented. In terms of financial aspect, a list of all direct and indirect costs was given in a standard cost of disruption evaluation form [19] to aid economic assessment of the losses incurred. Initially, a voltage-dip profile of the facility concerned is acquired from either utility data, measurements, monitoring, or prediction. Using this dip data, the supply dip performance contours are drawn, where each contour represents the number of voltage dips per year. Next, equipment sensitivities (voltage-tolerance curves) are overlaid on the supply dip performance contours to form dip-coordination charts. The sensitivity of the process is defined by the most sensitive component, with the knee point located at the upper most left hand portion of the chart. The subsequent step involves cost estimation of process disruption. All losses involved are listed in a cost of disruption evaluation form, which should be completed by those who are familiar with the operational impact of process stoppage (frontline workers, supervisors), finance, accounting, sales, and marketing personnel to ensure that all aspects of economic losses are considered. Briefly, the costs of disruption in industrial processes are made up of downtimerelated costs (lost production, idled labor, equipment damage, recovery cost), product quality-related costs (scrap and rework costs), and other indirect costs (customer dissatisfaction, employee and customer safety, fines and penalties). Finally, the total financial losses of the facility are obtained by multiplying the cost of process disruption and the number of disruptive dips per year. The method proposed by this standard is useful for estimation of economic losses due to voltage dips. However, there are a few important issues yet to be addressed. These issues include: •

• • •

The sensitivity of the entire industrial process is determined by the most sensitive equipment in the process. This assumption may not be appropriate because the process sensitivity also depends on the function and significance of the equipment involved. Tripping of the most sensitive equipment does not necessarily disrupt the entire process. The interconnections between equipment and sub-processes could have significant impact on process operation, but these are not considered in this standard. It is shown that all equipment types have a range of voltage-tolerance curves. This range (of uncertainty) is not considered in the method when evaluating the number of disruptive dips. The cost values used for economic assessment are based on historical data or experience; this may not be useful for evaluation of new industries at the planning stage.

2.1.4. Analytical Economic Analysis In the past decade, new methodologies have been continually developed with the promise of improved accuracy in assessment. In analytical economic analysis, losses due to PQ disturbances are often calculated or estimated through detailed assessment processes. These assessment processes may consider the probability of PQ events occurring, characteristics of events, equipment and process sensitivities to events, cost of process disruption, cost and benefit of mitigations, and other indirect costs subsequent to the event. 2.1.4.1. Assessment of Equipment and Process Sensitivity For most PQ disturbances, economic losses are incurred when equipment or industrial processes are tripped or disrupted. Hence, equipment and process response to PQ disturbances directly influence the magnitude of economic losses. However, predicting equipment and process response to disturbances is not entirely straightforward due to various uncertainties involved in equipment sensitivity. Therefore, properly representing these uncertainties helps reduce assessment error.

14

Probabilistic Assessment of Financial Losses due to Interruptions and Voltage Dips This methodology [15, 16] uses a probabilistic approach to assess the economic losses due to interruptions and voltage dips. The cost of interruptions and voltage dips are assessed separately and then combined to estimate the total economic losses in the network. This methodology can be used to assess both customer-level losses and network level losses. It takes into account the uncertainties associated with voltage-dip calculation, equipment sensitivity, interconnection of equipment within an industrial process, and customer type and location of the process in the network. The fault positions method is used to obtain voltage-dip performance of the system. Process equipment is grouped into four categories of equipment types, and the voltage-tolerance curves of these equipment types are obtained through a series of laboratory tests. The main equipment types tested are personal computer (PC), programmable logic controllers (PLCs), adjustable-speed drives (ASDs), and AC contactors. The impact of voltage dips at a particular site in the network is estimated through three basic steps. They include fault analysis, voltage-dip analysis, and economic analysis. Fault analysis is typically performed using the fault positions method to simulate various types of faults at various locations throughout the system network. The corresponding voltage magnitude and duration during of each fault is determined. Voltage-dip analysis is performed at the point of common coupling (PCC) between the network and the buses of interest. The historical fault performance (fault per kilometre per year) of all network buses, overhead lines, and underground cables is then used to determine frequency of dips of specified magnitude and duration over a period of interest. Dip durations depend on fault-clearing times of protection devices used in the network. The economic analysis is performed in two stages. First, sensitive equipment is classified into various categories based on device type. The voltage tolerance characteristic of four main equipment types, namely personal computer (PC), PLC, ASD, and AC contactors are determined through a series of laboratory tests. General voltage-tolerance characteristic is used to represent each equipment type. Next, dip performance charts of the network buses of interest are prepared using the results from voltage-dip analysis (Step 2). The dip-performance charts are compared with the equipment voltage-tolerance curves to determine equipment response (failure probability of equipment) to voltage dips. After obtaining the failure probabilities of equipment, the probability of a process trip is calculated. Finally, the total economic losses can be determined using (2.1).

Total financial loss = Total process trips × Cost per trip

(2.1)

A major advantage of this method is that the uncertainties regarding equipment sensitivity are represented using probability density functions. Probabilistic representation is more realistic and efficient as compared to the deterministic approach, especially when a large number of equipment is to be evaluated. Furthermore, this methodology provides the flexibility for different equipment sensitivity levels to be represented using different probability density functions. This methodology is probably one of the most comprehensive methodologies developed so far that takes into account many aspects of the system. An example of its application is given in Appendix 2-A. However, there are still some problems that were not fully resolved even with this relatively comprehensive methodology. They include the choice of appropriate probability distribution functions for equipment sensitivity evaluation (which are yet to be determined) and interdependence between equipment controlling a process or sub-process. Prob-A-Dip Method The Prob-A-Dip method [21] manipulates two-dimensional arrays to represent all parameters for economic loss management. This method allows equipment sensitivity to be represented using both discrete states (on or off state) and probabilistic values. Different cost values can be assigned to voltage dips of different characteristics, which enables more realistic evaluation of losses as costs such as equipment damage that occurs only for a certain characteristic of voltage dip. The method also takes into account the interconnections between equipment in a probabilistic manner and the effect of mitigation devices.

15

It is worth noting that Prob-A-Dip is more of a management tool rather than an assessment tool. All input values have to be acquired through other means before they can be processed by this Prob-A-Dip. Basically, the user has to acquire information regarding equipment sensitivity, annual dip frequency, process-related information, and the consequential economic loss associated with different voltage dips. With this information, the Prob-A-Dip method can be used to determine sensitivity of the industrial process, the frequency of dip-induced plant interruptions, and the total economic losses incurred due to voltage dips and short interruptions. The main advantages of this method include the following [21]: • • • • • •

All quantities are presented in a uniform format. It is applicable to all environments from a single customer to a complete power distribution system. It delivers flexible accuracy, from a rough estimate to exact values. It allows probabilistic processing of data. It can be implemented in power system software platforms. It enables assessment of the effectiveness of mitigating solutions to a certain degree.

Estimating the Economic Impact of Voltage Dips The methodology for estimating the economic impact of voltage dips proposed in [24] was built on the assumption that different voltage dips have different impacts on customer process. It is postulated that the behavior of process equipment varies with the severity of voltage dips, thus causing different failure modes of the customer plant. First, in order to characterize process equipment behavior when exposed to a voltage dip, the voltagetolerance curves of different equipment are obtained and converted into a component behavior function. The function represents the state (On or Off) of the equipment when subjected to a voltage dip with specific dip magnitude and duration. Next, the behavior of the customer load is categorized into different failure modes, where certain combinations of equipment behaviors and dip conditions (causes) will lead to certain failure modes (consequence). The “cause” and “consequence” are related using fault trees. A cost function is assigned to each failure mode. The economic impact of a voltage dip is estimated by combining the cost function with information regarding voltage-dip frequency. The use of fault tree analysis provides more space for subjective judgement in process sensitivity evaluation. The user would not need to deal with complex equipment interconnections to determine the consequences of equipment failure. Using different failure modes for different voltage-dip levels would also yield more realistic results. However, it is worth noting that the equipment, even of the same type and brand, exhibit very different responses to a voltage dip. So, it is virtually impossible to generalize equipment behavior into a common working state. Also, the number of failure modes increases rapidly with size of customer plant, and hence increases the complexity of evaluation. PQ Index Based on Equipment Sensitivity, Cost, and Network Vulnerability The ideas of load drop index (LDI) and load drop cost (LDC) [18] are proposed with the objective of capturing load vulnerability and the cost impact of voltage dips. These indices are calculated using customer equipment composition data, load information, equipment sensitivity curves, and historicalderived cost data. Basically, voltage dips are first categorized into various duration classes (instantaneous, momentary, temporary, and sustained interruption) consistent with the classification of interruption events given in IEEE Standard 1159 [35]. For each duration class, various regions of voltage dip class areas are defined based on the voltage-tolerance curves of sensitive equipment involved. Next, the historical voltage-dip profile of the bus of interest is processed to obtain the number of events that fall in each defined area. With this information, the load drop index for each duration class k is calculated using (2.2) [18]. r

LDI ( k ) = ∑ N kj Lkj ,

j = 1, 2,...r

(2.2)

j =1

Where Lkj represents load composition ratio and Nkj represents the number of events that falls in the area defined by duration class k and sensitivity curve of load type j. r is the number of load type.

16

Different cost indices are derived for different duration classes based on average cost of interruption in each duration class. These cost indices are multiplied by the corresponding LDI of each duration class to obtain LDC [18]. 4

LDC = ∑ Ck LDI ( k )

(2.3)

k =1

Where Ck is the cost index reflecting the average cost of interruption in a given duration class. The average cost of power interruptions reported in [18] are given in Table I-1 of Appendix 2-I. LDC is particularly useful for assessment of economic losses due to voltage dips and interruption events. As a PQ index, it can be easily translated into cost figures for evaluation of all general industrial plants. The merits of using LDI and LDC also include: • • • • •

Instead of using the most sensitive equipment to define process sensitivity, the impact of a voltage dip on all equipment types is considered. Composition ratios of equipment are considered. Capable of processing probabilistic values of equipment sensitivity. Different economic impacts of different voltage-dip severities (dip duration) are considered. Only involves data processing and does not require additional instrumentation.

It is worth noting that the effects of equipment interconnections and the importance level of individual equipment in process operation are not modeled in LDI and LDC. This might prevent accurate assessment because equipment interconnections and importance are significant factors that affect process vulnerability to voltage dips. Unified Reliability and PQ Index The unified reliability and PQ index method proposed in [36] combines the costs incurred by interruption, voltage dip, voltage deviation, and harmonics into a unified reliability and PQ index. In terms on voltagedip cost, the factors considered include voltage dip rate, the load size at customer busbar, and sector customer damage function for voltage dips (SCDF(dip)) at customer busbar. The dip rate is calculated utilizing sustained interruption rate and momentary interruption rate at the customer busbar. Two types of system configuration are considered (loop and radial), both protected by a reclosing system. SCDF(dip) depends on the sector where the cost is to be assessed. Seven sectors are classified, namely large user, industrial, commercial, agricultural, residential, government installation, and office buildings. This methodology is suitable for fast estimation of economic losses at network level. Due to the fact that many important factors are not considered, the accuracy of estimation is not too high. 2.1.4.2. The Cost of Process Interruption Regardless of the type of disturbance in an industrial process (voltage dip, transients, short interruption, or long interruption), economic losses are incurred every time the process trips. The cost per trip should include only those costs that are above and beyond the normal production costs, net of potential savings. An example of how economic losses for an industrial customer can be determined is suggested in [37] and elaborated on in more detail in Appendix 2.

2.1.5. Indirect Economic Analysis When the information required for analytical economic analysis is not available, indirect economic analysis is the only option to estimate the financial losses due to PQ disturbances. Common ways of analysis include the customer willingness to pay method [34], customer willingness to accept method [34], and cost estimation from the size and value of mitigating solutions. 2.1.5.1. Customer’s Willingness to Pay The customer’s willingness to pay (WTP) method has been used in several studies [34, 38, 39] to obtain the costs of power supply interruptions. Usually, customers are given several hypothetical outage scenarios and asked to express the amount of money that they are willing to pay in order to avoid each outage scenarios. In terms of PQ, customers are asked to express their willingness to pay for different levels of PQ improvements. Though the WTP method may not be as technical as the analytical approaches, it reflects the value customers place on electricity supply and PQ.

17

However, one should anticipate the amount a customer is willing to pay to be lower than the actual financial damage caused by a PQ disturbance [34]. This is because economic benefit could only be achieved if the financial damage due to power interruption is more than the amount paid to avoid the damage. Therefore, from the customers’ point of view, the WTP amount will always be less than the actual damage due to PQ disturbances. Besides, the WTP method makes sense only when customers understand the damaging effects of power supply interruptions on their processes. Usually, the effects of a total power interruption are more apparent and well known. However, the effects of other PQ disturbances such as voltage dips that cause partial disruption of processes are not straightforward. In most cases, customers do not know the financial damages due to PQ disturbances, and therefore cannot place an accurate WTP value on them. In the customer’s willingness to accept (WTA) method, electricity users are given various imaginary outage scenarios and asked to estimate the amount of compensation that they are willing to accept for each outage scenario. The WTA is similar to the WTP method because they both require customers to place a monetary value on hypothetical outage scenarios. However, in most cases, the WTA method gives substantially larger values compared to the WTP method. According to [34], the reason behind this is that customers consider electricity supply as a social right rather than a market commodity. It is also recommended in [34] that the two methods can be used together to produce upper and lower limits for power interruption costs. Both WTP and WTA methods are heavily dependent upon the customer’s subjectivity in placing a value on PQ costs, and may be influenced by other considerations, such as the customer’s perception of the electricity supply, their knowledge on PQ disturbances, and their ability to pay. 2.1.5.2. Cost Estimation from Historical Events Over the years, numerous surveys have been carried out around the world to gather information regarding economic losses of various industrial, agricultural, commercial, and even residential customers. By reviewing the past studies, the financial loss information can be gathered and aggregated to represent different customer types and sizes. This information can be conveniently used to estimate PQ-related costs of a particular customer. To obtain a realistic cost estimation from historical events, one would have to use historical values from the customer type that best resemble the customer of concern. Ideally, the historical values used should be obtained from customers of similar type and size, and within the same geographical region as the customer of concern. Unfortunately, information gathered from historical events available today is still insufficient to meet the abovementioned requirements. Most studies produced cost values for total power interruption, not considering the impact of other PQ disturbances. Some studies managed to produce cost values of voltage dips but have yet to obtain cost values for different severity levels of voltage dips. Overall, cost estimates from historical events without considering sensitive equipment in a customer’s plant and process sensitivities will not produce accurate financial loss values. In [29] an original approach was used to estimate economic losses for industrial users. The authors designed a questionnaire, in the form of a journal. The main innovative aspects of this instrument are: 1. 2. 3.

The questionnaire was not of the usual scenario type. Instead, it entailed the registration, for a specified time period (at least three months), of the consequences experienced by the end-user during process disruptions caused by very short interruptions and voltage dips; The questionnaire required the end-users to provide a structured description of “what happened” at the production site during the voltage disturbance (see below), together with per-unit economic data (for instance, hourly wages). It did not request direct cost estimations from the respondents. The questionnaire did not demand to identify precisely what type of voltage disturbance brought the process to a halt. This information was extracted, at a later stage, from the data recorded by a power quality recorder, installed at the customer’s connection point.

The questionnaire included a technical section, to be filled in by personnel working on the production line(s). The description requested was structured in a time sequence: •

At the occurrence of the event: damaged equipment and defective WIP (and its destination: recycling, second-hand goods, etc.).

18

• • •

During downtime: duration and number of workers inactive (or engaged in restarting the process). During the restarting of the process: defective WIP and lost production (if the process requires time to return to the nominal production quality and speed). After the event: time and number of workers necessary to recover lost production (or other means to recover it).

The questionnaire included also an economic section, to be filled in by a manager or by an accountant. This section investigated the per-unit costs of several production inputs, such as raw material, labor, and energy (but also the cost of repairing the damaged equipment). With this approach, the authors were able to retain full control over the cost assessment: they did not ask industrial users to estimate their costs, because it is normally done in surveys, but to provide a structured description of the process disruptions, together with “per-unit” economic information. The calculation of costs was performed by the authors according to a standardized methodology. Overall, the approach resulted in an improvement in terms of feasibility and robustness with respect to previous surveys.

2.1.6. Reported PQ-Related Losses from Around the World In the past decade, many studies have been conducted around the world to determine the cost of voltage dips and short interruptions. The experience gained from these studies is very valuable and can be used for carrying out similar studies in the future. This section summarizes the major findings of those studies. A few studies that focused on obtaining the cost of interruptions (outages) only are also included in this summary as the information gathered can be further postprocessed and used for the assessment of cost of voltage dips and short interruptions. Numerical data from the surveys is reported in Appendix 2. The surveys confirm high sensitivity to short interruptions and voltage dips in many processes, with particularly high losses in the production of electrical and electronic equipment, chemical products, food products, and motor vehicles. Generally, survey results are presented in the following ways: • • • • •

Direct cost per kW of plant per disturbance (Table -J-1) Direct cost per kVA of plant per disturbance (Table J-2) Direct cost per disturbance event (Table J-3) Annual cost of disturbance (Table J-4) Cost per hour of process interruption (Table J-5)

Though high losses, processes are commonly identified in most surveys, the magnitude of the losses is rather inconsistent. For example, huge differences in losses can be seen in different surveys reported for chemical products and electrical products manufacturing. This disparity is due to the difference in circumstances while conducting the surveys. In particular, there are differences in the country in which the surveys are conducted, the categorization of industries, the type of disturbances included, the year of survey, the size of the industries involved, and the base currency used for loss representation. These differences prevent the surveys from being compared effectively and meaningfully. With increasing need for accurate loss estimation for the industrial sector, a common standard in conducting surveys is crucial to ensure a consistent outcome in future surveys. In the meantime, a methodology capable of grouping and analyzing the surveys is urgently required.

19

2.1.6.1. Studies in Europe In 2007, the Leonardo PQ Initiative (LPQI) team published results of a pan-European PQ survey [40] comprised of 62 face-to-face interviews across eight European countries. A total of 16 industrial and services sectors were covered in the survey, which essentially represents 38% of the EU-25 turnover and 70% of the region’s final electricity consumption. The costs of all major PQ disturbances (dips, swells, short and long interruptions, harmonics, flicker, surges, transients, unbalance earthing, and EMC problems) were obtained considering direct and indirect cost components. It was found that dips and short interruptions account for 60% of the overall cost for industrial samples and 57% for the total sample. Further regression analysis concluded that PQ cost is directly correlated to the annual turnover of the affected customer, with industrial and services customers wasting around 4% and 0.142% of their annual turnover respectively to PQ disturbances. Major findings of this study are summarized in Fig. I-1 and I-2 of Appendix 2-I. Indeed, it is considered a good practice to put PQ costs in context. Moreover, it is a necessary step when one is interested in understanding the significance of the problem over the economy of a country (relevant information, for instance, for regulators) or society in general. Several studies compare PQ-related costs with the costs of long interruptions (e.g., [41]), others with other economical parameters such as the annual electricity expenses (e.g., [14]), or the cost of annual sales or value added (e.g., [29]). The methodology for projecting plant-level costs (the usual outcome of a survey) to the national economy should be addressed with care. Normally the sample of respondents in a survey is neither large nor stratified for any relevant dimension so as to be representative of the universe. It follows that a great caution should be used in reading the results of the analysis. A survey [42] conducted by UMIST, UK since October 1992 assessed the outage cost of various customer categories due to electricity supply interruption. The survey covered three regional electrical company areas, and customer sectors are categorized as residential, commercial, industrial, and large user. A customer interruption cost (CIC) was defined as the perceived individual customer or average sector customer costs resulting from electricity interruption [33]. The survey provided CIC values for each customer sectors for various duration of interruption, as given in Table I-2 of Appendix 2-I. A separate study by researchers of UMIST investigated the influence of process equipment composition on economic losses due to voltage dips [28]. Detailed formulas were proposed to calculate the direct and indirect damage costs associated with industrial process disruption due to voltage dips. The study was simulated on a generic distribution system consisting of 295 buses. Four types of sensitive equipment are considered, namely personal computers (PC), PLCs, ASDs, and AC contactors. It was observed that different load compositions at customer plant sites result in significant variation in dip costs. In year 2000, researchers from Helsinki University of Technology, Finland conducted studies [8, 9] to estimate the annual frequency and cost of voltage dips for customers of five Finnish distribution companies (three rural and two urban networks). Customers were divided into five categories of domestic, agricultural, industrial, commercial services, and public services. The method of fault positioning [8] was applied for the calculation of voltage-dip frequency. Economic consequences were obtained by multiplying the dip frequency, the cost of a single voltage dip, and the number of customers. The cost of a single voltage dip was taken from a survey in the mid-1990s in three Nordic countries. Different cost values were used for different customer categories. Results obtained indicated much higher losses than expected. It was also suggested that more accurate results can be obtained by more precise representation of customers’ dip-related inconvenience and actual economic losses. Results of the study are given in Fig. I-3 of Appendix 2-I Aggregated data is given in [41], as the results of a new Norwegian survey carried out in the years 2001-2003. A report by STRI AB, Sweden [43] presented a structural way to investigate voltage dip immunity of industrial processes and their related costs. Basically, a cost index and a fault index are assigned to each industrial process. A cost index indicates process contribution to the overall cost due to voltage dips, while a fault index represents the fault frequency of the process. A process box is used to represent the process, with information regarding cost index and fault index clearly stated in the box. The economic losses of a process due to voltage dips can then be evaluated using (2.4) [34].

Si = Ci ⋅ Fi + ∑ Ck ⋅ Fi

(2.4)

(k )

20

Where Ci is the cost index (€/voltage dip) of process i, Fi is the fault index of process i, and k is the processes affected by tripping of process i. This method can be effectively used to determine the economic losses of processes due to voltage dips and identify the most sensitive or the most expensive process. It can be further used to include the interconnections of sub-processes inside a process. Besides presenting methods to calculate costs, the report also summarized general voltage dip-related cost for different industries, as presented in Fig. I-4 of the Appendix 2-I. In year 2000, Italian researchers [14] published estimates of the costs associated with poor PQ. These estimates were built on a survey conducted by a semiconductor and pharmaceutical facilities construction company. The survey included around 30 production plants located in Europe, the USA, and the Far East that do not have any measures in place to mitigate against PQ disturbances. Having analyzed the results of the survey, three categories of voltage-dip profiles were determined as the most meaningful for estimation of costs. The categories are: • • •

Category A – includes 10 or fewer voltage dips per year with residual voltage less than 40% of nominal and dip duration shorter than 100 ms. Category B - includes 10 or fewer voltage dips per year with residual voltage less than 40% of nominal and dip duration shorter than 100 ms, and five or fewer voltage dips per year with residual voltage less than 70% of nominal and duration ranging from 100 ms to 300 ms. Category C – includes one interruption with duration of three minutes or more.

Estimated costs for the industrial sectors considered in the survey are given in percentage of the total yearly power cost, as can be seen in Table I-3, Appendix 2-I. As for Italy, researchers performed a survey in different areas of the North-East part of the country between year 1999 and 2002 [10]. The survey focused on 200 small industrial customers of various sectors. The costs due to voltage dips are presented in normalized cost per voltage dip per kW power to ease comparison between industrial sectors and sizes. It was found that most sensitive plants have normalized cost per dip in the range of 0.25-1.5 Euro/kW. Detailed results of this survey are given in Fig. I-5 of the Appendix 2-I. Based on this survey, the same group of researchers proposed a method for computation of the interruption costs caused by supply voltage dips and interruptions in small industrial plants [27]. The assumptions made were that industrial plants have only one shut-down model, and that each voltage dip or interruption that trips the process requires equal restart time. This further implies that severe voltage dips and momentary interruptions cause equal interruption costs. The error introduced by these assumptions is thought to be reasonably low. A large portion of the paper focuses on producing equations for cost calculation. The costs considered include cost of lost production during supply disruption and restart time, cost of wasted materials, imperfect product, damaged equipment, and extra maintenance resulting from the disturbance. The savings on raw material, energy not consumed, and recovery of lost production were also considered. Using this method, a production plant in the plastic sector was investigated [10, 27]. It was found that the cost of a nuisance voltage dip is 517.5 Euro. This value is about 66% of the losses due to a one-hour unplanned interruption. In 2007, Politecnico di Milano of Italy [29] conducted a field survey on 50-MV industrial customers in 13 (potentially sensitive) manufacturing sectors, to determine the direct costs due to voltage dips and momentary interruptions (less than 1 second). The first objective of the work was to obtain cost indicators for sensitive manufacturing sectors. Results are presented in terms of plant-level cost indicators that are normally found in the literature: annual direct costs per kW, direct cost per event per kW, annual direct cost per production plant, and direct cost per event per production plant. Table I-4 in Appendix 2-I gives a sub-sample that excludes responses with zero costs. A second objective was to estimate the weight of these costs on the Italian economy. Country-level direct costs were estimated by projecting plant-level cost indicators to the Italian economy. Indirect costs (i.e., costs for protecting the production plant) were assessed through market-based analysis, whereby the annual amortization costs for mitigation (UPS) were used as an indicator of cost. The total annual costs sustained by the Italian production system was estimated to be comprised between a minimum of 449 and a maximum of 809 million Euros. The impact of direct costs on the economy of sensitive sectors is certainly not negligible. The incidence of direct costs on the level of sales of these sectors is 1.5-1.7 € for every 1,000 € of sales (6.8-8.1 € for every 1,000 € of value added). Note that these costs are more than

21

four times higher than those in a generic sector. Finally, it emerges that direct costs are a very specific and concentrated problem. In other words, very short interruptions and voltage dips result in important direct costs only for a portion of the whole Italian production system. Sensitive sectors account for 16.97% of national sales (and 14.98% of value added). Based on the report published by the Copper Development Association and sponsored by the copper producers and fabricators, [44], a 10-month study carried out by a major generator in Europe on 12 sites of low-technology manufacturing operations logged a total financial loss of €600,000. Some of the findings of this report are summarized in Table I-5 in the Appendix 2-I. 2.1.6.2. Studies in the USA An on-site survey of 299 U.S. large commercial and industrial customers was carried out in 1992 to determine the financial losses incurred by interruption and voltage dips [45]. Interruption costs for the following scenarios were investigated: • • • • • •

A 1-hour interruption starting at 3 p.m. on a afternoon without advance notice. A 1-hour interruption starting at 3 p.m. on a afternoon with 1 hour advance notice. A 4-hour interruption starting at 3 p.m. on a afternoon without advance notice. A 2-hour interruption starting at 7 a.m on a winter without advance notice. A 1- to 2-second momentary interruption on a afternoon in clear weather. A 10% to 20% voltage dip for 15 cycles.

summer summer summer morning summer

Summary of the survey results is given in Table I-6 in the Appendix 2-I. In year 1993, Clemmensen [46] provided the first-ever PQ cost estimate for U.S. manufacturing sector. The estimate derived that annual spending on industrial equipment due to PQ problems could sum up to $26 billion dollars for the U.S. manufacturing sector. It was estimated that for every manufacturing sales dollar, 1.5 to 3 U.S. cents (i.e., 1.5% - 3%) are spent to mitigate PQ problems. A few years later in 1998, Swaminathan and Sen [46], in a Sandia National Laboratory report, estimated that U.S. annual power interruption cost reaches $150 billion. This estimate was based on a 1992 Duke Power outage cost survey in the U.S. that manipulated industrial electricity sales as the basis for the estimate. Later in year 2001, EPRI’s Consortium for Electric Infrastructure to Support a Digital Society (CIEDS) [47] produced a report based on a Primen survey in the United States. The report identified three sectors of the U.S. economy that are particularly sensitive to power disturbances: • • •

The Digital Economy (DE): telecommunications, data storage and retrieval services, biotechnology, electronics manufacturing, and the financial industry. Continuous Process Manufacturing (CPM): paper, chemicals, petroleum, rubber and plastic, stone, clay and glass, and primary metals. Fabrication and Essential Services (F&ES): all other manufacturing industries, plus utilities and transportation facilities.

These three sectors collectively lose $45.7 billion a year due to outages and another $6.7 billion a year due to other PQ phenomena. It is estimated that the U.S. economy losses between $104 billion to $164 billion due to outages and another $15 billion to $24 billion due to PQ phenomena. In the meantime, EPRI Solutions (formerly EPRI PEAC) [48] conducted PQ investigations on the continuous process manufacturing (CPM) sector of U.S. industries to identify industry-specific cost data resulting from PQ disturbances. CPM involves manufacturing facilities that continuously feed raw material at high temperature. Results of this investigation are summarized in Fig. I-6 in the Appendix 2-I. According to [49], a consulting firm specializing in evaluating technology markets, estimated over $20

22

billion of annual voltage disturbance cost by U.S. industries. Estimated losses for various industries per voltage dip are also provided in the study, as shown in Table I-7 in the Appendix 2-I. A comprehensive summary of the outage cost is given in an EnerNex Corporation report [50] in year 2004. It includes outage costs obtained from different surveys. Detailed results are given in Table I-8 in Appendix 2-I. 2.1.6.3. Studies in Asia Survey results of interruption costs for 284 high-tech industries in Taiwan were published in [51] in year 2001. Six categories of high-tech industries were studied. They included semiconductor (SC), computer and peripherals (CP), telecommunications (TC), optoelectronics (OE), precision machinery (PM), and biotechnology (BT). The report also compares the obtained interruption costs with the interruption costs from other countries. Summary of the results is given in Table I-9 in Appendix 2-I. The results of this survey were also presented in a separate paper published in 2006 [32]. The cost of interruptions was represented as a customer damage function, which gives interruption costs as a function of interruption duration. These customer damage functions are given in Fig. I-7 in the Appendix 2-I. The same paper also presented results of a PQ survey conducted on the same industries. Financial analysis for voltage dips used weighting factors for different voltage-dip magnitudes. Besides, voltage-dip sensitivity factors were derived based on the survey results. It is concluded that high-tech industries are sensitive to supply quality, and that the semiconductor industry suffers the highest losses for interruptions of less than three seconds. The dip-sensitivity factors are given in Table I-10 in Appendix 2-I. In South Korea, the Korea Electrotechnology Institute in cooperation with Gallup Korea conducted an interviews-based survey on 660 industrial customers [52] of various sizes and sectors. The survey resulted in successful estimation of interruption costs for the industries surveyed. 2.1.6.4. Other Reported Losses A case study on two industrial plants in Egypt was published in 2004 [11]. It was reported that for each voltage dip costs manufacturing plant (size of 1MVA) a cost of $5,800 and $8,060 a 200kVA polyester factory. The data gathered by the ABB [53] related to different industries and sensitive loads shows that the financial consequences of voltage disturbances can range between 3 and 120 $/kVA per event. More details are shown in Table I-13 in the Appendix 2-I. The results of the similar assessment provided in EPRI’s PQ Applications Guide for Architects and Engineers are summarized in Table I-14 and those from U.S. Department of Energy [37] in Table I-15 and Table I-16 in Appendix 2-I.

2.2. Methodology for Quantifying the Economic Impact of Harmonics 2.2.1. Introduction Economically quantifying the effects of the harmonics in an electrical system requires the computation of all the consequences that the harmonics of current and of voltage have on all the equipment and components. The effects of the voltage and current distortion on any equipment or component fall in three classes: additional energy losses, premature aging, and malfunction . The term “additional” means that these losses are superimposed to the ones at the fundamental; the term “premature” refers to the possibility of aging rate accelerated by the rise of stress level with respect to the nominal service conditions. The term “malfunction” pertains to the loss of the equipment performance in respect to the nominal conditions. For each class, the methods to be followed can be of two types: deterministic or probabilistic. Deterministic methods are adequate when all the items of the analysis, from the operating conditions of the system to the discount rate value, are known without uncertainty. This can be the case of ex post analyses performed on existing systems whose operating conditions are repetitive and well stated. Some real cases can refer to industrial systems. Probabilistic methods are instead needed when some of the problem variables are affected by

23

uncertainties. This clearly happens for non-existing systems or also for existing systems where some expansions have to be planned. However, technicians are often involved in estimating the costs to face for the future operation of existing systems when both cash flows and operating conditions of the system vary over a range and thus introduce a degree of uncertainty. First proposals of methods to economically quantify the harmonics in a system [50-52] dealt with the deterministic evaluation of the cost to the electric utility to contend with the harmonic pollution. The costs include the total active power losses value as well as the capital invested in the design and construction of filtering systems. Even if these first studies recognize the premature aging of the equipment can lead to potential additional costs, it was not included. Several successive studies [53-55] proposed probabilistic methods, also extending the costs due to harmonics, to take into account the premature aging of the equipment. Unfortunately, few contributions can be found regarding the economics of malfunction [56, 57] that recently [58] has been addressed, approaching it like reliability costs.

2.2.2. Overview of Existing Methodologies The effects of the voltage and current distortion on the equipment that can be economically quantified the energy losses, premature aging, and malfunction. The related economic value represents the searched quantities. Sections 2.2.2.1 to 2.2.2.3 give an overview of deterministic approach, while section 2.2.2.4 discusses the main steps involved in probabilistic analysis of economic losses due to harmonics. 2.2.2.1. Deterministic Methodologies Economical Values of Energy Losses To compute the economic value of the additional energy losses arising for an operating period in question, it is required to have the information on the following: • • •

System operating conditions in the study period, i.e. network configurations, typical duration of system states, and so on. Type, operating conditions, and absorbed power level of linear and nonlinear loads. The rate of variation of the electric energy unit cost and present value of discount rate.

As aforementioned, when uncertainties affect some of the variables involved in the economical analysis, a probabilistic approach is needed. In the following, we firstly recall deterministic methods because they offer the frame of study also useful for probabilistic methods, outlined in paragraph 2.2.2.4. Let us initially refer to the case of a single electrical component continuously subject to hmax harmonics h1 h max in the time interval ∆T . The operating costs Dwk are: of voltage or current G ,..., G

(

)

Dwk = Dwk G h1 ,..,G h max = Kw Pk ( G h1 ,..,G h max )∆T

(2-5)

Where Kw is the unit cost of electrical energy and Pk ( G h1 ,..,G h max ) represents the losses due to the h1 h max on the kth component. Appendix 2-K reports the formula useful to compute harmonics G ,..., G such losses for the main electrical components

The operating costs in ∆T for the whole system, in which m components and equipment operate, can be computed as the sum of each one: m

Dw = ∑ Dwk

(2-6)

k =1

To evaluate the operating costs of the system components with reference to more years—usually the electrical system life—it is then necessary to take into account both the variation of the unit cost of the

24

electric energy in the years and the present worth of the costs taking place in every year of the system life. The following relationships can be assumed for the variation of the electric energy unit cost:

( Kw) n = ( Kw) 1 (1 + β ) n −1

(2-7)

and for the present worth value:

(Dw)npw =

Dwn (1 + α ) n −1

(2-8)

Where β is the variation rate of the electric energy unit cost and α is the present worth discount rate. Finally, the present worth expected value of the operating costs of harmonic losses, referred to the whole electrical system period of NT years, is: NT

NT

Dwn , n −1 n =1 (1 + α )

Dw =∑ (Dw) pw =∑ n

n =1

(2-9)

Where Dwn is the sum of the operating costs for the whole system in all the time intervals ∆Tj that occurred in the generic year n. The present worth expected value of the operating costs of harmonic losses, referred to the whole electrical system period of NT years, is: Nr

Dw = ∑ n =1

Dwn

(1 + α )n−1

,

(2.10)

Where Dwn is the sum of the operating costs for the whole system in all the time intervals ∆Tj that occurred in the generic year n and α is the present worth discount rate.. The recalled relations evidence that computing the economical values of losses due to harmonics imposes h1 h max on each component the knowledge of several quantities; among them, the harmonics G ,..., G are the currents and/or voltages. For most real cases, the main contributions to the total economical value of harmonics are due to the current harmonics flowing into series components of the system like cables or overhead lines. However, dielectric losses linked to voltage harmonics can play a not negligible role, for example in transformers or also in medium-voltage (MV) cables when the thickness of insulation material is particularly large [59, 62, 63, 86]. 2.2.2.2. Economical Value of Premature Aging To compute the economic value of substituting damaged components due to their premature aging, it is required to know: •

The system operating conditions in the study period, i.e., network configurations, typical duration of system states, and so on.

•

The type, operating conditions, and absorbed power level of linear and nonlinear loads.

•

The appropriate “life models” of equipment and components in order to estimate the failure times of their electrical insulation.

•

The costs of the replacement (new) components together with the cost variation rate.

The premature aging caused by the harmonic pollution involves incremental investment costs to face during the observation period. Also for this analysis, it is useful to firstly recall deterministic methods, assuming all problem data is known without uncertainty. Referring initially to a single component, let these incremental costs be defined as the aging costs Da k :

Da k = C k ,ns − C k ,s

(2.11)

25

In (2.11), Ck,s and Ck,ns are the present worth value of the total investment costs for buying the kth component during the system life in sinusoidal and non-sinusoidal operating conditions, respectively. The values of Ck,s and Ck,ns can be at once evaluated when the useful lives Ls and Lns are known. In fact, once they are known, both the number of times that the component has to be bought in the system life and the years in which the purchases have to be done are fully estimated. The useful lives Ls and Lns of an insulated electrical apparatus can be estimated summing the relative life losses ∆L, which come in succession until reaching the unity. It is important to highlight that electrical power system components are subjected to different service stresses (electrical, thermal, mechanical, and so on), which can lead to degradation of electrical insulation. The degradation (aging) of solid-type insulation, like it is for MV/LV distribution system components, is an irreversible process, involving failure and, thus, breakdown or outage of the whole component. However, electrical and thermal stresses (i.e., voltage and temperature) are, in general, the most significant for insulation in MV/LV power systems. Moreover, the interaction between electrical and thermal stresses can lead to a further increase of electrothermal aging rate with respect to the effect of these stresses applied separately; such a phenomenon is called stress synergism [59]. Aging rate can be accelerated by the rise of the stress level with respect to the nominal service conditions. This can be due just to voltage and current harmonics, which may lead to an increase of electrical and thermal stresses on the insulation, thus shortening insulation time-to-failure, i.e., component useful life. In a distorted regime, the life models of equipment and components can take into account only thermal stress, leading to Arrhenius law-based models, or can take into account also electrical stresses, leading to more complex life model. Let us first assume that the useful life of an insulated device is only linked to the thermal degradation of the insulation materials. Thermal degradation can be represented by the well-known reaction rate equation of Arrhenius [58, 59] when the absolute temperature of the materials is constant. From the Arrhenius relationship, in [62, 63] it has been demonstrated that the thermal loss of life of the k-th t component ∆Lk in a time period Tc characterized by different operating conditions, each at given temperature and of given duration, can be expressed as the summation of relative losses of life: q  t i ,k

  i = 1 Λ (θ i ,k )

∆Ltk = ∑ 

(2.12)

Where q is the number of operating conditions in Tc; tik is the duration of operating condition of the k-th component at constant temperature θi,k; and finally

Λ (θi ,k ) is the useful life of the k-th component at

temperature θik, obtained from the Arrhenius model. The temperature of each insulated components θi,k can be determined considering the heat balance relationships, in which the losses at the fundamental and at the harmonics are the forcing terms. The present worth value of the additional aging costs arising in the whole system for N components, is computed as the sum of the cost of each component: N

Da = ∑ Dak

(2.13)

k =1

Where the value of Dak is calculated, via (2.6), starting from the knowledge of the useful lives of the various components. When both thermal and electrical stresses have to be accounted for, the procedure does not modify, but the life models to be used in relation (2.12) change in electrothermal model. In such a condition, the et relative loss of life of the MV/LV power system component in the time period Tc, ∆Lk , can be again expressed as the summation of fractional losses of life:

26

q 

 ti  i = 1 L(Ei ,θ i ) 

∆Let k = ∑ 

(2.14)

Where L(Ei ,θ i ) is the life that the component would experience if constant values of electric and thermal stresses Ei and θ i were continuously applied until failure. In the literature, the electrothermal models of the most common equipment and components of MV and LV systems can be found; in particular, the electrothermal models that explicitly account for voltage and current harmonics are in [53, 64-66]. 2.2.2.3. Economic Value of Misoperation The economical evaluation of the misoperation is the most complex subject and, maybe, the least explored one. The complexity of the cost estimation is strongly linked to the absence of exact knowledge of the cause-effect linkage between harmonics and degradation of performance of equipment for the real difficulty of the concrete discrimination of harmonics as the only cause of the disturbance. The harmonics as the origin of several degradations of the equipment performance remain obscure for all the equipment life. Indeed, in [67, 68] there are reported some categories for which the performance degradation due to harmonics can be more easily discriminated: electronic equipment operating with voltage zero crossing, meters, lighting devices. Generally, the economical impact of misoperation involves financial analysis of all the effects that misoperation has on the process/activity where the equipment is inserted. Typically, the misoperation costs can be estimated for existing systems whose duty cycle is well known. Unexpected tripping of protections, for example, can result in stopping a whole industrial process. The cost of such an event includes several items, like cost of downtime, cost of restoring/repairing, cost for replacing the equipment, where applicable. Some interesting values can be found in [68], which mostly referred to existing systems in Spain, where extensive investigation was carried out among a wide range of commercial and industrial sectors. The findings of the research confirm that to estimate the misoperation cost requires deep knowledge of: • • •

The equipment malfunctioning in presence of harmonics. The process/activity where the equipment is inserted. The economical value of all the items involved in lower productivity.

Looking at the problem of evaluating the misoperation costs with this point of view, it is evident that several analogies arise with the problem of evaluating the economical effects of micro-interruptions or of voltage dips. At least for all the cases where the lower productivity is due to partial or complete stoppage of the process, the methods and the components of the financial analysis are the same. Recently [67] proposed to include the supply unreliability costs into the category of misoperation. This interesting proposal is based on the concept of sector customer damage and allows estimating the misoperation cost in function of well-stated figures in reliability studies, like sustained failure rate and momentary failure rate. The model is particularly suitable for distributors that, in the planning stage, can use the economical metrics to choose the best solution among future alternatives. 2.2.2.4. Main Steps Involved in Probabilistic Evaluations When faced with uncertainty, it often unavoidably affects the input data in real systems for changes of linear load demands, of network configurations, and of operating modes of nonlinear load. It is needed to translate the economical models on a probabilistic ground. This implies the introduction of random variables and application of probabilistic techniques of analysis, as well as other terms to introduce probabilistic methods. The first step in a probabilistic approach is to recognize that output economic figures to be computed are statistical quantities. In the most general cases, their probability density functions (PDFs) completely describe their statistical features. However, for the sake of estimating the economic value of losses and premature aging due to harmonics, it is adequate referring to the total expected value as:

27

E ( D) = E ( Dw) + E ( Da)

(2.15)

Where symbol E(.) indicates the expected value of the quantities already introduced in the paragraphs 2.2.2.1 and 2.2.2.2. When estimating expected values for a period of time, it is needed to consider their present worth values as:

E ( D) pw = E ( Dw) pw + E ( Da) pw

(2.16)

The present worth expected economical value of losses due to harmonics losses, E ( Dw) the whole electrical system life of NT years, is: NT

pw

, referred to

NT

E ( Dw) n n −1 n =1 (1 + α )

E ( Dw) pw =∑ E (Dw)n =∑ pw

n =1

(2.17)

pw Where E (Dw)n is the present worth expected value of the harmonic losses in the nth year, and

E ( Dw) n is computed summing the economical value of harmonic losses of each component in each jth combination characterized by mj components operating in the same time period ∆Tj: mj

E (Dw) j = ∑ E (Dw)k , j .

(2.18)

k =1

For the gncombinations taking place in year n, it is: gn

gn m j

j =1

j =1 k =1

E (Dw)n = ∑ E( Dw ) j = ∑ ∑ E( Dw )k , j

(2.19)

It is clear from relation (2.19) that it is necessary to compute the expected value of harmonic losses for each component of the system, that is E( Dw )k , j . Considering each single electrical component

,G h 2 ,..,G h max h max , E( Dw )k , j .

continuously subject to an hmax harmonics of voltage or current harmonic G

∆T

h1

by the joint PDF f h1 G ,..,G For the most common components of industrial energy systems, the harmonic losses Pk , j ( G h1 ,..,G h max ) , with their expressions reported in Appendix 2 - L (L-7), can be obtained by

characterized in the time interval

summing up the losses due to each harmonic. pw , is evaluated summing the The present worth economical value of premature aging in (2.16), E ( Da ) present worth expected value of the aging costs of each of the N components of the system:

N

E( Da ) pw = ∑ E( Da )kpw

(2.20)

k =1

pw Where the value of E( Da )k is calculated starting from the knowledge of the useful lives of the

various components by the relation:

28

E(Da)kpw = E(Cns )kpw - E(Cs )pw k

(2.21)

pw pw Where E (C s ) k and E (Cns ) k are the present worth expected value of the costs for buying the

component during the system life in sinusoidal and non-sinusoidal operating conditions, respectively. The actualization of the costs can be effected in a similar way as in the previous equations (L-11) and (L12) of Appendix 2 - L. considering both the discount rate and the cost variation for buying the component; the expected value of cost to be met for buying each component at year n in sinusoidal and non-sinusoidal regimes is linked to the expected value of the component life in these conditions, respectively. To estimate these figures again, the cumulative damage theory can be applied, as in the case of deterministic methods. In such a case, we have to refer to the expected value of relative loss of life in the study period, E[∆RL]. A more complete analytical formulation is reported in Appendix 2-L.

2.3. Methodology for Quantifying the Economic Impact of Other PQ Phenomena 2.3.1. Voltage and Current Unbalance The sensitivity of electrical equipment to unbalance differs from one appliance to another. A short overview of the most common problems is given below: Induction machines: The magnitude of the internally induced rotating magnetic field in induction machines (IMs) is proportional to the amplitude of the direct and/or inverse components. The rotational sense of the field of the inverse component is opposite to the field of the direct component. Hence, in the case of an unbalanced supply, the total rotating magnetic field becomes “elliptical” instead of circular and consequently could lead to three types of problems in IM operation. First, the machine cannot produce its full torque as the inversely rotating magnetic field of the negative-sequence system causes a negative braking torque that has to be subtracted from the base torque linked to the normal rotating magnetic field. Secondly, the bearings may suffer mechanical damage because of induced torque components at double system frequency. Finally, the stator and, especially, the rotor are heated excessively, possibly leading to faster thermal aging. This heat is caused by induction of significant currents by the fast rotating (in the relative sense) inverse magnetic field, as seen by the rotor, and is also accompanied by vibrations. To be able to deal with this extra heating, the motor must be derated, which may require a machine of a larger power rating to be installed. In general, if voltage unbalance is permanently higher than 2%, the losses of fully loaded IM are likely to cause damage. Synchronous generators: Synchronous generators are exposed to similar stress as IM when subjected to unbalance. However, they mainly suffer from excess heating. Special care must therefore be devoted to the design of stabilizing damper windings on the rotor, where the currents are induced by the indirect and homopolar components. Capacity of transformers, cables, and lines: The capacity of transformers, cables, and lines is reduced due to negative-sequence components. The operational limit is in fact determined by the RMS rating of the total current, being partially made up of “useless” non-direct-sequence currents as well. This has to be considered when setting trigger points of protection devices, operating on the total current. The maximum capacity can be expressed by a derating factor, to be supplied by the manufacturer, which can be used to select a larger system, capable of handling the load. Transformers: Transformers subject to negative-sequence voltages transform them in the same way as positive-sequence voltages. The behavior with respect to homopolar voltages depends on the primary and secondary connections and, more particularly, the presence of a neutral conductor. If, for instance, one side has a three-phase four-wire connection, neutral currents can flow. If at the other side the winding is delta-connected, the homopolar current is transformed into a circulating (and heat-causing) current in the delta winding. The associated homopolar magnetic flux passes through constructional parts of the transformer, causing parasitic losses in parts such as the tank, sometimes requiring an additional derating.

29

Electronic power converters: These are present in many modern devices such as adjustable-speed drives, PC power supplies, efficient lighting, etc., and the amount of power electronic converters is bound to increase further in the future. As a consequence of unbalanced supply, they can be faced with additional uncharacteristic harmonics, although, in general, the total harmonic distortion remains more or less constant. The design of passive filter banks dealing with these harmonics must take this phenomenon into account. The table below specifies percent of extra losses from load unbalance as a function of neutral current resulting from unbalance to average phase current. Ratio of neutral current to average phase current 0,5 1,0 1,5 2,0 3,0

Table 2-1 Extra losses due to unbalance % of additional losses from load unbalance Transformers Low-voltage lines 6-8 40-50 15-20 70-140 35-50 140-260 70-90 200-500 150-200

2.3.1.1. Classification of Unbalance Costs Economic losses due to voltage and current unbalance, i.e., the economic costs of unbalance ( K as ), can ''

'

be divided into two categories, namely technological losses ( K as ) and electromagnetic losses ( K as ). The technological losses include losses resulting mostly from changes in the slip and torque of induction motors and consequential decrease in the output of motor-driven production equipment, a decrease in the induction motor’s maximum torque, reduced efficiency of single-phase electrical heating equipment, and reduced efficiency and lower quality of production due to changes in electric lighting. They also depend on the load type and should be calculated taking into account specific features of production processes. The electromagnetic losses associated with voltage unbalance result mainly from an increase in active power losses, as well as increase in the active and reactive power demand, reduction of capacitors and synchronous machines’ reactive power with respect to the required value, accelerated aging of insulation, and reduced in-service time of light sources. '

The annual costs of losses K as due to unbalance can be expressed as the sum: m

m

m

j =1

j =1

j =1

K as' = ∑ ∆K Pj + ∑ ∆K Aj + ∑ K Rj + K Q + K 0 Where

(2.22)

∆K Pj = additional cost of power losses in the j element (equipment, load) in the considered ∆K Aj

facility due to voltage and current unbalance = additional cost of energy losses in the j element due to voltage and current unbalance

K Rj

= costs of restoration of the j element caused by aging of its insulation due to voltage

KQ

and current unbalance = cost of the reactive power reduction due to unbalance

K0

= cost of light sources replenishment due to detrimental effects of voltage unbalance '

The overall unbalance costs ( K as ) include costs of the negative-sequence unbalance ( K as 2 = K as 2 +

K as' ' 2 ) and zero-sequence unbalance.

30

2.3.1.2. Additional Costs of Power Losses and Electric Energy Losses Annual cost of additional losses in the j element (loads, the series and shunt transmission, and distribution equipment) is:

∆K Pj = k pj ∆P2 j Where: ∆P2 j

k pj

(2.23) = maximum additional losses in a year, caused by voltage unbalance (loads, shunt equipment, no-load losses in transformers, etc.) or current unbalance (transmission and distribution series equipment) = unit cost of power losses at the level of the power system in which the j1 element is connected

Annual costs of additional energy losses in the j element are:

∆K Aj = k Aj ∆A2 j Where: ∆A2 j

k Aj

(2.24)

= annual additional cost of energy losses in the j element caused by the voltage or current unbalance = unit cost of energy losses at the level of the power system in which the j element is connected

For practical purposes, ∆A2 is often calculated from the formula:

∆A2 j = τ j ∆P2 j

(2.25)

Where τ j is the annual duration of maximum losses ∆P2 j . The relative power losses ∆P2 = *

∆P2 100 , where ∆PN is the nominal losses, under permissible ∆PN

voltage unbalance conditions (i.e., the measured negative-sequence unbalance factor 2%) are negligible. For example, according to [100] these losses are: (a) 6-kV and 10-kV induction motors with rated powers above 100 kW ∆P2 = 2.4% ; (b) synchronous motors with rated powers above 100 kW ― 4.2%; (c) *

transformers in industrial networks 1 to 4%. 2.3.1.3. Costs of Equipment Restoration The voltage or current unbalance causes additional heating of electrical equipment components that results in shortened in-service time of the equipment insulation. The equipment in-service life will also be shortened because of the intensification of ionizing processes caused by the voltage increase due to unbalance. Under balanced supply conditions, the equipment in-service time equals TS (in years). The operability costs of this equipment during time TS , i.e., total annual costs of expanded reproduction and operating

1

The method of determining the losses ∆P2 for various types of loads and transmission and distribution equipment is outside the scope of this report.

31

costs, are K us , and discounted costs (constant over time) are K uar . Under unbalanced conditions, the equipment in-service time equals Ta < TS , total annual costs are K ua , and annual discounted costs are

K uar : K us = TS K usr = TS pS K i K ua = Ta K uar = Ta pa K i Where: K i

(2.26a) (2.26b)

= investment expenditures associated with the equipment installation

pS and pa = coefficients of the equipment reproduction (fixed operating costs taken into account); pS ≠ pa because of different depreciation periods in each case. After elapse of time Ta , the equipment should be repaired or new identical equipment should be installed. The expected cost of repair or installation of new equipment is K m . Total annual costs incurred during the period ( Ta - TS ) are:

∆K R = (TS − Ta ) pn K ni

(2.27)

Where: pn = expanded reproduction coefficient as pS and pa but related to the new or restored equipment. Total discounted costs associated with premature replacement or necessary repair of equipment, the socalled annual costs of equipment restoration, are:

KR =

∆K R TS − Ta = pn K ni TS TS

(2.28)

In practical calculations, pn ≈ pS ≈ pa can be assumed. As evident from the formula (2.28), the relative time of shortening the equipment life due to voltage and current unbalance has considerable influence on the costs K R 2:

∆T * =

∆T TS − Ta = TS TS

(2.29)

As follows from research [69], under voltage unbalance conditions and the voltage unbalance factor of 2%, the average values of time ∆T are: (a) induction motors – 9.1%; (b) synchronous motors – 10.2%; *

(c) distribution transformers – 2.3%; transmission transformers – 3.4%; converters – 3.4%; power capacitors – 20 to 25%.

2.3.1.4. Cost of Reduction of the Reactive Power Value The reactive power of a capacitor bank is changing as a result of the voltage unbalance. Compared to the reactive power under the balanced supply voltage conditions, this power often decreases by ∆QK . The unbalance of currents in synchronous machines reduces their inductive reactive power by ∆QG . 2

The method of determining the time ∆T or ∆T * for various types of loads is outside the scope of this report.

32

The deficit in reactive power ∆Q = ∆QK + ∆QG should be replenished by means of additional compensation equipment, e.g. installation of additional capacitor banks. Total annual discounted costs associated with the installation and operation of additional capacitor banks are:

K Q = pQ K Qi + K PA + K ZQ + K pQ + K RQ where:

(2.30)

K Qi = investment expenditures for the equipment to compensate the reactive power ∆Q pQ = the expanded reproduction installment, including fixed operating costs K PA = cost of power losses and energy losses in the compensation equipment K ZQ = costs of undependability caused by the compensation equipment unreliability

K pQ = other costs associated with installation of the compensation equipment (positive or negative) concerning e.g. consequences of changes in the PQ parameters

K RQ = costs of restoration of the compensation equipment resulting from the voltage unbalance effects The costs K Qi are essentially dependent on the compensation equipment, i.e. , K Qi = f ( ∆Q ) . Under the voltage unbalance conditions, the reactive power of a capacitor bank ( QK ) can be larger or smaller than its rated power ( QKN ) or the power under the balanced supply voltage conditions ( QKs ). The value of ∆QK = QKN − QKs can therefore be positive or negative. Consequently, the costs K Q can also be positive (additional loss) or negative (an extra profit). For a synchronous machine operated under 2% voltage unbalance and the system p.u. negative-sequence reactance equal 0.24, the negative-sequence symmetrical component of machine currents is 8% [69], which for some types of machines (turbogenerators) is intolerable. This problem occurs particularly in industrial cogeneration plants with unbalanced load. In such cases, it is necessary to reduce the reactive power generated by synchronous machines. With voltage unbalance exceeding 3% at the terminals of a synchronous motor, both the motor current and the generated reactive power shall be reduced. Under voltage unbalance equal to 2%, the reduction of reactive power generated by a synchronous motor is 5 to 23% [69]. 2.3.1.5. Costs of Replenishment of Light Sources Voltage unbalance in lighting installations causes the voltage rise in one or sometimes in two phases. It results in shorter in-service time of lamps, increased active power demand, and in the case of discharge lamps, increased reactive power. A reduced voltage (in one or two phases) results in reduction of the luminous flux, reduction of power, and losses in lighting installation. Total luminous flux ( Φ ns ) of all light sources, supplied from different phases under the voltage unbalance conditions, may differ or differ insignificantly from the luminous flux ( Φ s ) in the case of balanced voltages. If the flux under unbalanced conditions is lower, additional light sources with power

∆Pd 0 and luminous flux output ∆Φ = Φ S − Φ ns shall be installed in order to provide a luminous flux

33

required by standards. The annual discounted costs associated with this installation, i.e., annual costs of additional light sources, are:

K d 0 = f p (∆Pd 0 ) = f Φ (∆Φ ) = pd 0 K d 0i + K inne

(2.31)

Where: K d 0i = the cost of installation of additional light sources

pd 0 = coefficient of expanded reproduction (operating costs taken into account) K inne = other costs components (the cost of power and electric energy, the cost of power and electric energy losses, etc.) The costs of replacement of light sources ( K w 0 )3 are the cost incurred in connection with premature replacement of incandescent or fluorescent lamps, or even entire luminaries, due to shortened in-service time. The number of light sources (lamps) to be prematurely replaced ( ∆L ) is associated with the considered facility and depends essentially on the form of voltage unbalance. The annual discounted costs of replacement of lamps can be calculated in a similar way as costs of compensation equipment (capacitor bank) restoration because of their premature wear-out due to voltage unbalance, i.e.:

K w0 =

∆T0 pw 0 K w 0 i T0 s

Where: T0 s

(2.32)

- in-service time of lamps supplied with balanced phase voltages

K w0i - costs of replacement of lamps (light sources) pw0 - coefficient of expanded reproduction (operating costs taken into account) The costs of replenishment of light sources ( K 0 ) are the sum of additional light sources ( K d 0 ) and costs of replacement of light sources ( K w 0 ):

K 0 = K d 0 + K w0

(2.33)

The issue of economic losses due to voltage and current unbalance still remains an open issue that requires analyses and experimental research.

2.3.2. Surges and Transients Most transients arise from the effects of lightning strikes or switching of heavy or reactive loads. Because of the high frequencies involved, they are considerably attenuated as they propagate through the network so that those occurring close to the point of interest will be much larger than those originating further away. Protective devices in the network ensure that transients are generally kept to a safe level, and most problems arise because the source of the transient is close to or within the installation.

3

The method of determining the shortening of in-service time ( ∆T0 ) for various light sources under voltage unbalance conditions is outside the scope of this report.

34

The damage that results may be instantaneous, such as the catastrophic failure of electrical plant or appliances, or the corruption of data within computers or in network cabling, or it may be progressive with each event doing a little more damage to insulation materials until catastrophic failure occurs. The cost of replacing the failed equipment and the cost of the downtime involved must be considered. If a surge or transient does not pause the process, still it could create dielectric stress in cables, which accumulate in the form of extra insulation aging. Similarly, the same could happen to capacitors. Such effects are, however, difficult to assess. The approach described in [18] can be used. Surges and transients even if not halting the production process may cause some process and equipment cost. Surge arresters need maintenance and replacement of activated parts, production equipment and also protection and controls need checks or resets and sometimes additional maintenance. This can be addressed by process and equipment costs approach as discussed above.

2.3.3. Flicker The voltage of an electrical network varies all the time under the influence of various switching operations of electrical equipment connected to the supply network. It can be slow or fast, depending on whether it is a progressive variation of the total load supplied by the grid or it is an abrupt variation of a large load. The level of voltage variations emitted by connected electrical equipment into the supply network depends on the network impedance. With increasing impedance, the level of voltage variations will increase. The variations of the voltage creates flicker, a perturbation which affects the lighting equipment and creates an impression of unsteadiness of the visual sensation. Voltage fluctuations in the power systems cause a number of harmful effects of technical and ergonomic nature. Both types of effects may involve additional costs in the production process. Several selected adverse effects of voltage fluctuation are shortly described. Also, frequently occurring, irregular operation of contactors and relays should be mentioned, as their economic effects could be damaging. Electric machines: Voltage fluctuations at the induction motor terminals cause changes in torque and slip; as a consequence, they influence technical processes. In the worst of cases they may lead to excessive vibrations and therefore to a reduction of mechanical strength and shortening the motor service life. Voltage fluctuations at the terminals of synchronous motors and generators give rise to hunting and premature wear of rotors; they also cause additional torque, changes in power, and increase in losses. Static rectifiers: A change of supply voltage in phase-controlled rectifiers with DC side parameters control usually results in a lower power factor and generation of non-characteristic harmonics and interharmonics. In the case of a drive braking in an inverter mode, it can result in a switching failure, with consequent damage to the system components. Electrolysers: Here the equipment useful life can be shortened and the efficiency of technical processes can decrease. Elements of the high-current line become significantly degraded, and there exists a real risk of increased maintenance and/or repair costs. Electroheat equipment: In this case the efficiency is lessened—for example, with the arc furnace due to a longer melt time—but it is noticeable only when the magnitude of a voltage fluctuation is significant. Light sources: A change in the supply voltage magnitude results in change of the luminous flux of a light source, known as flicker. It is a subjective visual impression of unsteadiness of a light flux, whose luminescence or spectral distribution fluctuates with time. Excessive flicker can cause migraines and is responsible in some instances of the so-called “sick building syndrome.” One has to mention though that the complaints due to flicker are usually a localized problem. As a consequence, routine measurement campaigns are not carried out often. On the other hand, the available data confirms that the long-term and short-term flicker levels are commonly below those levels that might give rise to complaints. Values in excess of those prescribed in standard EN 50160 do occur, however. In remote areas in particular (weak networks), flicker levels might increase up to critical values, as demonstrated by many measurement results.

35

Due to the localized nature of complaints arising from flicker, excessive flicker values tend to be found in the framework of measurements that are targeted specifically at areas of complaint. Given the immediate visibility of the phenomenon and the severe human discomfort that can be caused, each case of complaint must be taken very seriously. In order to prevent flicker becoming a widespread problem, appropriate emission limitation is essential, with due allowance for the cumulative effect across the network levels (different from the harmonic cumulative effect) .

2.4. Conclusions Major studies around the world concluded that voltage dips and short interruptions cause significant financial losses to customers of various sectors. Over the years, customer surveys have been the most common method used to economically quantify the losses incurred by voltage dips and short interruptions. New methodologies are being continually developed to consider more and more factors that contribute to financial losses. Although theoretically, these methods promise better process/system representation and improved accuracy, their effectiveness is yet to be proven due to the fact that none of them are being tested in actual processes/network. The harmonics can increase operating and investment costs of a power system. In dependence on data and system information availability, deterministic or probabilistic methods allow the cost quantification. The reported methods fall in the category of analytical methods that need deep knowledge of the system structure, electrical components, electrical devices characteristics, and functions.

36

3. Overview of Existing Methodologies for Assessment of Economic Impact – Public Distribution Network Perspective 3.1. Introduction The cost of power quality to consumers is well documented, and a range of techniques for calculating the costs exist, as detailed in Chapter 2. Utilities, however, do not generally collect the costs associated with different PQ phenomena. Instead, they are accounted for in routine operation and maintenance figures. In this chapter, utility costs associated with PQ are identified and their relevance defined. A review of existing methods for collecting economic data is also provided. The term “utility” can mean many things; it is therefore necessary to define what the term “utility” means in the context of this chapter. A definition of a utility is “an organization responsible for maintaining the infrastructure of a public service.” A public services could include gas, electricity, water, sewage, telephone, and Internet. In the electricity industry alone, “utility” could be interpreted in many different ways, i.e. a vertically integrated energy company, an electricity supplier, a network operator, or a generator. All are affected by PQ in different ways, as illustrated in Table 3.1. Table 3.1 Effect of power quality Utility Type Economic impact of voltage dips Generator (Wholesale) Generators tend to sell less electricity in the hours following a voltage dip while the industrial processes restart. Transmission Network TNOs incur costs associated with mitigating faults, which may lead to Operator (TNO) problems further down the system. Supplier (Retail) Suppliers are generally the first point of contact for consumers and therefore suffer additional strain on call centers. Distribution Network In most countries, the DNO is responsible for PQ and incurs high costs Operator (DNO) associated with mitigation, resolution, and investigations. Vertically Integrated A vertically integrated company will be affected in all of the ways described above. This chapter is concerned only with the cost of PQ to the DNO, i.e. the owner and operator of the network of towers, cables, substations, etc. that bring electricity from the National Transmission Network to homes and businesses in a particular region, i.e. the medium-voltage (MV) and low-voltage (LV) networks. They are neither the organizations that sell electricity to the end consumer nor the company responsible for generating electricity. In the UK, where the industry is privatized, they are also responsible for allocating meter point administration numbers (MPANs), providing new electricity connections and resolving power outages in their area. The UK is split into 14 distribution network regions, and they each distribute electricity at following voltage levels: 132 kV to 33 kV, 33 kV to 11 kV, and 11 kV to 400 V. In France, transmission operates at voltages greater than 63 kV, while the distribution system operates at voltages of 20 kV or less. According to ANSI C84.1 [167], Chapter 3 System voltage classes and Table 1, in the USA, transmission is described as high voltage (HV) and is considered to be voltages above 100 kV, and distribution is medium voltage (MV) with voltages between 1 kV and 100 kV or low voltage (LV) which is anything less than 1 kV.. In Ireland, the transmission system operator (TSO) is responsible for the planning operation and control of the transmission network, while the transmission asset owner (TAO) is responsible for the construction and maintenance of the transmission assets in accordance with the development plans and maintenance policies issued by the TSO. The transmission network consists of the 400 kV, 220 kV, and the vast majority of the 110 kV network. There is only one distribution system operator (DSO), which is responsible for the planning, development construction, operation, and maintenance of the distribution

37

network, which consists of the 38-kV, MV (20 kV and 10 kV) and LV networks and the noninterconnected 110-kV circuits. In Italy the distribution system operates at low (equal to or less then 1 kV) and medium voltages (above 1 kV and equal to or less than 35 kV). Part of the distribution system operates also at high voltage (above 35 kV and equal to or less than 150 kV). In Spain, transmission operates at voltages greater than 132 kV (namely at 230 kV and 400 kV), while the distribution system operates at voltages of 132 kV or less. I.e., HV from 45 kV to 132 kV and MV less than 36 kV. According to European Standard EN 50160 [101], the limit between medium voltage and high voltage public networks is established at 35 kV. Nevertheless in a recent new edition 2010 it has been changed such limit up to 36 kV.

3.2. Review of Literature and Documented Methodologies Although the cost of PQ to customers is well documented and understood, very little has been published in relation to the cost of PQ to a DNO. The literature relates to determining pricing schemes for delivering power of higher quality and reliability to specific customers or to calculating the insurance policies relating to these supply contracts. In [102], an economic analysis has been carried out using the cost-benefit ratio as a basis of comparison to determine the mitigating equipment with best economic benefits. The process involves determining the total annual cost for each alternative, including both the costs associated with the power quality variations and the costs of implementing the solution. The cost per interruption is known, so the financial savings can be calculated using the net present value (NPV). Comparing the annual costs of different power quality solution alternatives identifies the solution with the lowest cost that warrants more detailed investigation. The costs associated with purchasing and installing various solution technologies are onetime, up-front costs that can be annualized using an appropriate interest rate and assumed lifetime or evaluation period. The paper concludes that the two parameters that have the greatest influence on the cost-benefit ratio of the mitigating equipment are the equipment costs and the cost of an interruption due to voltage dips. Reference [103] looks at a power quality service pricing approach considering treatment expenses and stop-loss insurance. The paper goes through the theory of stop-loss insurance before looking at the pricing of PQ. The pricing of PQ service is based on the correct understanding of PQ level, which is evaluated by a feasible method. A formula is defined to work out the fee of PQ service: F = C(s) + P(s) Where F is the fee of PQ service, s is the parameter of PQ level, C is the treatment expense charged by PSCOM (power supply company) for fulfilling the customer’s demand PQ level of s, and P is the premium charged for the stop-loss insurance related to s.

3.3. Costs Associated with PQ In this section, the costs incurred by a DNO in relation to PQ are listed. For each cost, the relevance to PQ is explained, indicative figures are provided, and an indication of how the costs could be established is given. The costs are split into two categories, depending on whether they are incurred as a result of a PQ incident or are associated with activities that mitigate the occurrence of PQ events. The costs associated with responding to and investigating complaints are the most easily quantified and attributed to PQ. Resolving PQ issues can vary hugely in nature and are therefore difficult to attribute an indicative cost to. Costs associated with mitigating PQ incidents are rarely separated from those associated with improving reliability and general network maintenance. There are, however, some activities that improve reliability but not PQ, and these are specifically noted because they should not be included in any attempt to quantify the cost of PQ to a DNO. Power quality is an inherent element of the basic electricity

38

connection. Increasing voltage quality or regulation above the basic design level cannot be done in isolation for particular customers because the unreliability of other network sections would simply be tagged on to more reliable circuits.

3.3.1. Costs Incurred by the Utility to Mitigate PQ Issues For utilities, quality of supply generally covers continuity and low voltage, and it is for these cases that investments are made, rather than for harmonics/dips/swells etc. It is also the case that network investment for the purposes of providing increases in capacity will also (as a byproduct) improve continuity and voltage regulation. The types of network reinforcement possible can be assembled into different groupings with their main drivers, which will also dictate where their costs are allocated; e.g. surge arresters are installed with all pole-mounted transformers to protect the transformer and may also result in an improvement in voltage swells/dips, but they are not installed for power quality improvement, and their costs will be allocated to new supplies or network refurbishment rather than power quality. Harmonics Harmonic limits are set on the system by national standards, and in order that these are not breached, lower limits are imposed on customer connections. If the customer’s connection is likely to result in a harmonic limit breach, then the method of connection is either changed or the customer installs mitigation measures at their own expense. Harmonic filters would usually be installed by customers rather than the utility. If background harmonics were excessive, the utility might be required to install harmonic filters. Zig-zag transformers can be used to cancel out harmonics. However, on an HV system, a delta winding is installed on transformers giving the same or similar effect. The extra cost of providing the delta winding is minimal when compared to the total cost of the transformer. The neutral in a balanced three-phase system should be lightly loaded, but with increased amounts of loads such as office fluorescent lighting and switched-mode power supplies, the harmonic currents in the neutral become much larger. Neutral currents can be up to 170% higher than the phase currents [102]. The utilities that had undersized their neutral might be forced to up-rate it to the same size as the phase conductor, except where it can be shown that a smaller conductor will suffice. This extra cost would amount to about €2 per meter on a main incoming LV cable. Direct Cost Calculation The presence of harmonics on a network can have a detrimental effect on assets in the medium and long term; these include but are not limited to: • • • • • •

Equipment is subjected to voltages and currents at frequencies that it was not designed to withstand. Derating of network equipment, such as cables and overhead lines, due to the additional harmonic load. Derating and overheating of transformers, particularly due to saturation effects in the iron core. Premature aging of network equipment, e.g. insulation materials and electronic components. Neutral conductor overload. Additional losses in the conductors and transformers.

Joule losses in aerial and underground power networks can be calculated according to the following simplified method: ∞

PCU

=

∑ 3·R ·I n

n =2

2 n

·L·t·E (euros)

(3-2)

where:

39

PCU = power losses (quantified in euros/year) Rn = impedance at harmonic n (Ω/km) In = averaged current at harmonic n (A) L = total lengths of line (km) t = hours E = price energy (euros/kWh) The power losses can be determined for different areas of the network and added together to get an indication of the system losses due to harmonics. For DNOs that are regulated, the value of these losses to the company will be determined by the regulatory framework. In the UK, DNOs are not penalized for losses, but they are rewarded for reducing them. The cost of reducing harmonics would therefore need to be less than the reward available for a DNO. The costs associated with premature aging and derating of assets are not easily quantified because DNOs do not generally maintain records of operating temperatures and harmonic levels at their assets. The effect on the asset is therefore unknown. More detailed calculation methods of are described in detail in Appendix 2-K. Flicker Voltage fluctuations in the power systems cause a number of harmful effects of technical and ergonomic nature. Both kinds of effects may involve additional costs in the production process. Several selected adverse effects of voltage fluctuation are shortly described. Also, frequently occurring, irregular operation of contactors and relays should be mentioned, because their economic effects could be damaging. •

Electric machines: Voltage fluctuations at the induction motor terminals cause changes in torque and slip; as a consequence, they influence technical processes. In the worst of cases, they may lead to excessive vibrations and therefore to a reduction of mechanical strength and shortening the motor service life. Voltage fluctuations at the terminals of synchronous motors and generators give rise to hunting and premature wear of rotors; they also cause additional torque, changes in power, and increase in losses.

•

Static rectifiers: A change of supply voltage in phase-controlled rectifiers with DC side parameters control usually results in a lower power factor and generation of non-characteristic harmonics and interharmonics. In the case of a drive braking in an inverter mode, it can result in a switching failure, with consequent damage to the system components.

•

Electrolysers: Here the equipment useful life can be shortened, and the efficiency of technical processes can decrease. Elements of the high-current line become significantly degraded, and there exists a real risk of increased maintenance and/or repair costs.

•

Electroheat equipment: In this case, the efficiency is lessened—for example, with the arc furnace, due to a longer melt time—but it is noticeable only when the magnitude of a voltage fluctuation is significant. •

Light sources: A change in the supply voltage magnitude results in change of the luminous flux of a light source, known as flicker. It is a subjective visual impression of unsteadiness of a light flux, whose luminescence or spectral distribution fluctuates with time. Excessive flicker can cause migraines and is responsible in some instances for the so-called “sick building syndrome.”

The voltage of an electrical network varies all the time under the influence of various switching on-and-off operations of electrical equipment connected to the supply network. The voltage variation can be slow or fast, depending on whether it is a progressive variation of the total load supplied by the grid, or it is an abrupt variation of a large load. The level of voltage variations

40

emitted by an electrical equipment into the supply network to which it is connected depends on the network impedance. With increasing impedance, the level of voltage variations will increase. The variations of the voltage create flicker, a perturbation that affects the lighting equipment and creates a impression of unsteadiness of the visual sensation. Complaints due to flicker are usually a localized problem. As a consequence, routine measurement campaigns are not carried out often. On the other hand, the available data confirm that the long-term and short-term flicker levels are commonly below those levels that might give rise to complaints. Values in excess of the EN 50160 value do occur, however. Especially in remote areas, flicker levels might increase up to critical values, as demonstrated by other measurement results. Given the localized nature of complaints arising from flicker, excessive flicker values tend to be found in the framework of measurements that are targeted specifically at areas of complaint. Given the immediate visibility of the phenomenon and the severe human discomfort that can be caused, each case of complaint must be taken very seriously. In order to prevent flicker from becoming a widespread problem, appropriate emission limitation is essential, with due allowance for the cumulative effect across the network levels (different from the harmonic cumulative effect) .

Lex = ∑ ni × ( Li + Oi ) × SDR i

Underground Cables Underground cables are less prone to transient faults than overhead, but it would not be effective unless the remainder of the network was also underground. UG cables are installed in areas where it is not feasible to install overhead lines, or where it is more economic to do so because it is a site that has not previously been developed and the ground is already open. However, it is worth noting that in cases where the underground cable is old and fault-prone, it can give rise to a significant impact on continuity as it is usually located in urban areas feeding large numbers of customers. Usually the proportion of faulty cable is low (depending on life cycle stage) and may be deemed poor enough to warrant replacement because of its condition. It is not possible to give an indicative cost for replacing overhead lines with underground cable, but it is likely to be prohibitively expensive in the vast majority of cases given the need to bury the entire network for transient faults to be reduced. According to some reports, 100% of underground cable would reduce the occurrences of dips by 67%, but due to higher losses of supply the end costs would be reduced by only 1%. Increased Sectionalizing Increased sectionalizing reduces the number of customers impacted by a fault and thereby increases overall reliability. Where networks are in close proximity and can be interconnected and sectionalized, this is done as a matter of design to improve continuity. There is also the spin-off benefit in reducing the impact of dips. Dividing a network into two halves would result into a reduction of voltage dips by 50% each. However, this would also result in reduced redundancy, increased restrictions on switching of network parts, and therefore reduced security of supply [113]. Indicative costs for (statistically) avoiding one voltage dip by splitting a network is given in the following example: MV network Length: 4539 km Number of customers: 616,000 Measurements conducted in 30 substations on this network showed an average of 21.2 voltage dips per substation per year and a statistical occurrence of 0.14 disturbances per km. Two-busbar operation enables a reduction of occurring voltage dips to half, i.e. 10.1 per substation a year. This modification would cost around €15.5million for 30 Peterson coils and 8 transformers. This would

41

result in (statistically 10.1 x 30) 318 voltage dips per year being avoided. Therefore, the cost per voltage dip avoided would be approximately €50,000. Insulate Overhead Lines Insulation of an overhead line can provide protection against transient earth faults caused by trees/branches rubbing against the line. In isolated neutral circuits, the line will not trip and the dip will be minor, but in directly earthed circuits there will be a much more significant impact on continuity. There are four ways in which this can be addressed: 1. Vegetation management 2. Installation of an arc-suppressed system 3. Installation of faulty phase earthing on current direct earthed 20-kV system 4. Insulation of covered conductors Vegetation management has low yearly costs even when very extensive work is required and is effective at reducing dips and outages caused by trees. Installation of an arc-suppressed system would depend on the suitability of the network to accommodate it. Use of insulated conductors on new lines is expensive and unnecessary where the line is clear of vegetation. However, most lines are not new, just extensions of existing circuits, so there is often little point in insulating one area when another is left open. Restringing (reconductoring) lines in covered/insulated conductors would be a major exercise, and to be effective in improving continuity would have to be done for the full circuit, assuming that the dip on the busbar feeding station for a fault on adjoining lines was not severe enough to offset the benefits. Conductor Spacing Modification/Animal Guards A network is normally designed in such a way that conductor clashing does not occur and thereby avoids unnecessary outages. Typically, in order to have faults due to conductor spacing, the line spacing must be such that it can be bridged by a bird’s wingspan, or in the case of bushings on pole mounted equipment, that they are close enough for vermin to bridge the gap between bushings. These possibilities are usually foreseen in the design, so that conductor spacing is adequate, and where the line is in the path of migrating birds with large wing spans, bird guards can be fitted on the erection or at a later date at low cost, because they simply clip onto the line and can be installed from the ground using a hotstick. For equipment such as transformers or reclosers mounted on poles, guards are placed over the bushings where the clearances are close. The installation of bird guards is a relatively low-cost solution that is adopted to reduce faults, which are normally sever in that they result in fuse blowing and power cuts. Lightning Protection Lightning causes problems either by hitting a line, in which case permanent damage is caused and cannot be prevented, or striking near the line, causing induced currents, whose effects can be mitigated to some extent. The probability of lightning strikes depends on location, but within location is strongly dependent on the structure’s height—the higher the structure the greater the likelihood of being affected by lightning strikes. High-voltage lines are much taller than LV or MV lines, so there is a higher probability of such lines being affected. There are a number of mitigation measures that can be taken, the most common being the installation of ground wires above the phase conductors, which results in the lightning strike being earthed via numerous towers so that the rise in voltage at each tower is insufficient to backflash onto the phase conductor and cause a dip in voltage. For lines held by wood pole construction, accommodating a ground wire costs about 20% more, as well as resulting in a structure that is visually intrusive and less acceptable to the local population. An

42

alternative solution that can be used is to place surge arresters on each phase so that a lightning strike causes a short dip in voltage but is less likely to damage equipment, meaning that the circuit will still be available after the strike. Pole and tower grounding are important in order to avoid tower voltage rise and risk of flashover, which would cause a voltage dip in ground-wire-protected lines. For an HV network, the precautions against lightning strikes arise from the need to protect equipment from damage and have the spin-off effect of reducing the impact of voltage dips. HV networks are optimized at design so that usually there is little improvement that can be made if the line has been properly designed from the start. MV lines tend not to have ground wires due to excessive cost and the lesser impact and probability of a lightning strike. Using surge arrestors on MV lines is standard practice at each transformer and at other pole-mounted equipment such as reclosers, as well as at singlephase tee off’s and cable line interconnections. If a lightning strike occurs, the surge arrestor would conduct, resulting in a voltage dip. In such cases, there is no method to prevent against dips from lightning strikes because the mechanism is there to protect against equipment damage. Fast Switching with Instantaneous Protection Fast switching requires sophisticated relays and circuit breakers that can also operate quickly. It also requires a meshed system or else the fast switching would result in an outage. On transmission networks, fast switching would be normal, but not at lower voltages, where the network is usually radial. Typical costs for the equipment involved are: Static switch with backup feeder Static switches

$100/kVA for low-voltage applications. $60/kVA for medium-voltage applications

$600k-$700k for very fast TS: within 0.25 cycles (11 kV, 10 MW) $125k for fast TS: within 2 cycles $75k for regular TS: within 6-7 s (10 MVA)

Solid-state transfer switch (SSTS) From $75k (several seconds) to $700k (1/4 cycle) Fuse Replacements Correctly sizing fuses so that faults are confined to the section with the fault is good design. However, the closeness to which fuses can be coordinated with each other can be limited by the network. There are new intelligent fuses that are actually single-phase reclosers that can be more correctly set so that they trip in the event of a fault before the recloser on the main line trips. These cost around €2,000 per installation. Fault-Current Limiting These are reactors that are installed so as to limit the fault current and hence reduce voltage dips. In new installations, there is little difficulty in accommodating them, but in existing installations, it would often be difficult to find room for them and to make the necessary connections, particularly if compact metalclad switchgear is used. Reactors can be installed in the coupler bay between two transformers; they have little current through them and so low losses. They do not come into play unless there is a fault on either transformer,

43

whereupon the fault current from the second is limited. Reactors can also be placed on individual lines so as to limit the fault current so that the voltage dip on other lines is limited. It does mean, however, that the voltage dip on the line with the reactor will be greater, although the dip on adjoining lines will be less. The cost of a 10% impedance reactor on the MV side of a 20-MVA transformer would be about €50,000 installed in a new build situation. Capacitors for Voltage Regulation Voltage regulating transformers rather than capacitors are used for voltage regulation because they are active and can either reduce or boost the voltage around a given set point. They cost around €25,000 per set and two sets are used in open delta for regulation on MV. Improving the power factor of the line by supply Vars is one way in which shunt capacitors can be used, but suitable tariffs that require the customer power factor to be between 0.95 and 1 are more effective, as these are spread over the network and the Vars supplied locally. Placing capacitors in series with the line to reduce inductance is not done at MV because overvoltages can arise, which damage the capacitor. Protection Coordination Modifications Regular reviews of protection settings take place to ensure that proper protection coordination is achieved to ensure that downstream devices (circuit breakers/fuses) operate before upstream devices to ensure that the section of network effected by the fault is minimized. At the main feeding substations, the individual feeder breakers should trip before the transformer breakers and similarly the protection devices on the network should trip before the upstream protection on the feeder trips. However, it is also recognized that protection systems are complex and attempts to optimize too closely could result in reducing the impact of the protection. Replacement of Old Feeders/Transformers Transformers normally operate correctly (excluding tap changer) until they fail; i.e., they do not malfunction— they catastrophically fail. So unless there is an on load tap changer issue, the only reason to change a transformer is to avoid the risk of it failing unexpectedly. It will operate correctly up to the time it fails, without giving rise to dips/poor voltage. Replacement of old overhead feeders can be effective as problems due to cracked insulators are eliminated and usually the line capacity will be increased, improving voltage regulation. Replacing an old overhead conductor is the same as building a new line along the same route, as the poles and line headgear will also have deteriorated. LV or MV lines cost between €14,000 and €20,000 per km but can increase due to terrain. Reliability will be improved because a new line will be less likely to be damaged in storm conditions, but in terms of dips, there should be little difference unless cracked insulators were the cause. In such cases, replacement of the insulators concerned would be a more effective way of eliminating dips. For poor voltage where the cause is an overloaded line, a rebuild in a larger construction is one option and a booster/sectionalizing another. FACTS devices FACTS (Flexible AC Transmission System) devices such as dynamic and static Var compensators and DSMES (Distributed Superconducting Magnetic Energy Storage) are capable of reducing the number of voltage dips or reducing the severity of dips experienced by end users. They are also used to mitigate

44

harmonics. FACTS devices are expensive mitigation devices with the main cost being capital and maintenance costs incurred every year of their life time. Maintenance and operation costs can be assumed to be anywhere between 5% and 15% of the capital cost. [104] Load Rebalancing Unbalance voltages are responsible for polyphase motor heating and torque pulsations. This implies premature aging of the winding insulation material and the mechanical degradation of the ball bearing. Extra-heating is also responsible for the derating of the motor nominal load capacity. This problem is apparent mainly in multi-grounded distribution systems as in North America and is caused by unevenly distributed single-phase load along feeders. The problem could come also from defective switch gear of voltage regulators or from blown fuses of a capacitor bank. The most common solution for voltage unbalance is to rebalance the load along the feeder. This implies a relocation of some of the single-phase loads to a less loaded phase or a phase permutation along the feeder. The increase of short-circuit capacity of the feeder could reduce the voltage unbalance. This is possible in changing substation transformer or in changing conductor size. Maintenance Costs Other regular maintenance costs such as insulator washing and equipment inspections can have a positive impact on PQ but cannot be separated from routine network management.

3.3.2. Costs Associated with Improving Reliability but Not PQ Reclosing Schemes The use of reclosing schemes on three-phase overhead lines improves reliability by isolating the fault. However, it also imposes voltage dips on other customers on the line, so there is a tradeoff between increased numbers of voltage dips for all customers but decreased outages for all and particularly those customers on fused spurs who would otherwise be shed during a transient fault. Reclosing schemes improve reliability at the expense of PQ, and hence the cost associated with them should not be attributed to improving power quality. Redundant Feeders and Loop Schemes Redundant feeders would have no effect on voltage dips but would allow a standby feed in the event of loss of the primary feeder. Generally, redundant feeders would not be used for standby except where a customer paid specifically for a dual radial feed whereby half the load was fed from one feeder and half from the other, in which case only half would receive a dip in the event of a fault. Loop schemes will provide greater reliability whether they are in a closed mesh or in a standby arrangement. Where the network is dense, the most effective use of the capacity in the network requires looping of new connections, which also reduces losses and maximizes loading between feeders. It also provides standby supply in the event of a fault, but this is a byproduct, and a looped feeder in a less dense network would normally be provided. Redundant feeders improve reliability but not PQ; therefore, their cost should not be attributed to improving PQ. Feeder Design Modification to Improve Reliability This is done to a certain extent in that on a cable network, separation will be maintained between circuits so that accidental damage to one will not affect the other cable, providing standby (i.e. avoiding commonmode failure mechanisms).

45

3.3.3. Costs for Responding to PQ Issues The previous sections outlined the costs associated with mitigating or preventing PQ incidents. In this section, the costs incurred after a PQ event are presented, as well as the ongoing costs associated with facilities, which will be required should an event occur, for example call centers. Call Centers PQ enquires cover a wide range of power problems from large-scale outages due to storms down to nuisance dips in voltage lasting a few cycles. Most of a call centre’s costs are related to the need to plan for responses to large-scale outages, so the cost of responding to less serious and less frequent complaints is a marginal cost of the operator’s time and equipment. The cost of a call centre operator in dealing with a voltage complaint will be the cost of their time, which depends on the duration of the call, which will be low (estimated at €10 for a typical complaint). The cost of the operator’s equipment (telephone, headset, computer, database for logging calls) is a business need and is not directly attributed to the cost of one PQ inquiry nor are operating service costs of the call centre such as gas, water, electricity, and telephone. Responding Crew Not all complaints are valid and some may be resolved over the phone. However, those that can’t be will require a responding crew to investigate. A responding crew will normally install a voltage recorder for a period of time at the problem site. The cost of the crew’s time will depend on how much time is spent travelling to the site, installing recorders, and returning to their office. There are also the overhead costs of the crew’s vehicle and fuel. An estimate would be 1.5 hours for travel and installation, which would double if a return journey was needed to collect the device(s). So an average response would cost €170, including overheads. A voltage recorder does not always need to be installed by a responding crew; it could be posted with setup instructions, thus saving cost. Consultation Once the voltage recorders have been retrieved, the data that they have collected needs analyzing. The costs involved with this process can vary depending on the time needed to establish the reason for a PQ problem. If the PQ problem is intermittent, then it is likely the analysis process would take longer. On average, a PQ problem will be allocated 20 hours for collecting data, analyzing, and report follow-up. Resolution When the reason for the PQ problem has been established, recommendations of work to be carried out would be made to resolve the problem. In many cases, work is required because of the condition of the connection itself, which due to its age would have required changing anyway. Hence while the work may be triggered by a voltage complaint, it would probably have been required in any case under planned maintenance. The cost of resolutions can range from hundreds of Euros to millions. Whether or not all these costs are attributed purely to PQ can only be decided by the DNO on a case-by-case basis. Occasionally, PQ issues are introduced to the network by a consumer; for example, equipment with a high startup current (outside normal limits) can cause voltage dips, large nonlinear loads can inject unacceptable high levels of harmonics into the network, arc-furnaces, arc-welders, and similar facilities can cause voltage flicker, etc. In these cases, the customers would be asked to disconnect the equipment or to pay for network reinforcements. Typically, voltage complaints requiring upgrade have been resolved by replacing the existing transformer with a larger, low-impedance module, which would cost around €2,000 to €3,000 per customer, although other customers sharing this connection would also benefit. Compensation Financial compensation for poor PQ is rare; it is therefore not possible to give a typical cost. However, a DNO seeking to quantify the cost of PQ to their business should include the full costs associated with any cases resulting in compensation, including legal fees.

46

3.4. Summary Facility managers and utility engineers must evaluate the economic impacts of the power quality variation against the cost of improving performance. It is desirable for utilities to optimize equipment maintenance subject to limited budgets. Investment and economic losses depend on numerous parameters, and this is itself an ambiguity and complexity factor in comparison issues regarding power quality improvement projects. Calculation of financial damages related to system quality problems widely various from customers and facility managers prospective. Meanwhile, the rate of estimated damage depends on the quality problems at the related region. In general, the main power quality issue can be identified as: - Voltage dips - Harmonic distortion - Voltage variation - Voltage unbalance - Voltage fluctuation - Transients This is why from an economic perspective, voltage variations, harmonics and, particularly, voltage dips are counted as the most important power quality problems in a system.

ACCUMULATIVE EFFECT OF DISTURBANCES

3.5. Conclusions Consumers of electrical energy are always interested in having access to a reliable source of energy. On the other hand, facility managers manifest high attraction to this issue based upon many reasons. However, this desire varies for different customers such that demand for electrical energy quality is not comparable between residential customers and an industrial plant. One of the main tasks of any electricity utility is to provide a reliable electricity supply to its customers at reasonable prices. The more reliable the electricity supply, the higher the price. However, if system

47

reliability is low, power interruptions and voltage dips tend to occur more often and will result in cost impacts to the utility and to the end customers. The balance between economical and technical considerations is therefore necessary for the utility’s operation. The optimal reliability level will be at the balanced point between the total cost of supply and the benefits to the customers. The DNO managers attempt to solve PQ problems for two main reasons. First, they are bound under current laws and standards to maintain power supply quality for customers within an appropriate range [101]. Second, DNO managers are interested in maintaining connected load, , and thus disturbances that interrupt end use processes are a hindrance. In the event that the network operator is not made aware of a customer’s load that is causing a breach of the power quality limits until after the load has been installed, the network operator will require that the customer take steps to bring the power quality parameter(s) back within acceptable limits. This can be achieved either by paying for the network operator to upgrade his network or by the customer taking mitigation measures within his installation. The latter option is most often the most cost effective since it can be focused on the offending item(s) of equipment, rather than the network as a whole. The principle that the action of one customer should not unduly interfere with the supply to another customer(s) is actually enshrined in national legislation of some EU members. Therefore the network operators have a statutory duty to ensure that the provision of a service to a particular customer will not cause the quality of other services to fall outside of recognized limits. In a small number of EU countries energy regulators set standards that are different from those indicated in the EN 50160. Moreover, EU energy regulators also impose standards for potentially disturbing customers, as well as for requests by customers of individual voltage quality verifications. For details, refer to CEER Benchmarking Reports (2005, 2008) [4, 5]. Electricity utilities are investing regularly on refurbishment and improvement projects on the power network to reduce voltage dips. For each of these investments, a detailed cost justification and business case is needed. For this to be done, it is essential that the benefits of improved power quality to both utilities and the end customer be quantified.

48

4. Methodology for Collecting Power Quality Economic Data 4.1. Introduction This chapter deals with methods for collecting economical data, separating end user perspective from distribution network operator (DNO). For both of them, the type of economical data and the methodology to collect them vary in dependence of multiple factors, like activity sector, type of disturbance, PQ phenomena under consideration, and required level of assessment accuracy.

4.2. Importance and Motivation The knowledge of economic data is at the basis of the economical analyses on different aspects of power quality (PQ). Any method used for estimating the economical costs, for deciding solutions for improving PQ levels, for investing in reinforcement of the network starts from economical data that has to be properly collected. Both end users and DNO are affected. For both these categories the first question to answer should be: Why do we need information on PQ economic data? Answering this question allows focusing on the type of data to collect, the most adequate format, the time horizon of the collection, and the most efficient way for collecting them. This last aspect is generally neglected assuming that, as in other economical investigations, the mode to obtain the information has no effect on the results. Instead, in the specific case of PQ economical analysis, the way can strongly affect the activity success. For example, sending by mail a questionnaire to an industrial facility that has not yet experienced a major PQ event can fail. In fact, the PQ costs can remain hidden until a disruptive incident, and the industry management does not perceive the importance of the analysis. This frequently happens for quasi-stationary disturbances like harmonics or low-frequency voltage variations. Moreover, also in the case of customers conscious of the importance of the investigation, the questionnaire must reach the right persons of that facility to obtain the right information. The most common economical analysis related to PQ is the quantification of the costs of disturbance effects. Economical quantification of the effects of PQ events that have already happened or estimation of future effects of PQ events that have not yet occurred are more and more frequent for both users and DNO. Further economical analyses can be based on the following reasons: • Becoming conscious of the magnitude of PQ costs, which practically may or practically may not affect productivity of a company. • Statistics and previous experiences may be helpful, but a very well known principle is that two companies operating in the same sector will hardly be equally vulnerable to PQ disturbances, and thus a cost survey is needed. • More and more often, PQ becomes a subject to contract between a user and a supplier. Costs of PQ are needed to quantify so-called willingness to pay, which is a measure of a value of improved PQ for which the user is going to pay a premium price. • In case of a failure caused by a PQ event that lays responsibility at the supplier according to contract for electric power supply, the supplier should compensate for the losses incurred. These losses will be calculated ex post, but an earlier cost survey may help in preparation to this calculation and its precision. • Finally, the knowledge on the cost of PQ will help to minimize these costs and optimize PQ cost with the cost of mitigation. Efforts to minimize PQ cost are always welcome, and sometimes they do not require any substantial investment. However, to start these efforts, the knowledge is needed where to focus exploration. While increasing the immunity of a single relay, which may be responsible for costly process interruptions, is not really an investment, sometimes costly efforts are needed to secure uninterrupted production processes. The methodology leading to unmitigated PQ cost and mitigation cost optimization is usually called investment analysis into PQ solutions.

49

Collecting PQ cost data is not trivial process. The process can be either retrospective or planned for the future. Planning this process depends on the activity sector that the company represents and other factors described in this chapter.

4.3. End-User Perspective As far as end users are concerned, the activity sector plays an important role in preparation for data collection because it requires understanding of the production or activity process and particular focus on certain cost categories and the data-collection method. The economic data that should be collected is not limited to financial consequences of PQ problems but includes also “savings,” which are a side effect of no use of electricity, other raw material, or staff resources during process interruptions. More economic data refers to the cost of PQ mitigation/prevention, e.g. maintenance cost of existing UPS, disposal costs, etc. Moreover, an appropriate long-term analysis should include indirect consequences such as contractual penalties, loss of reputation / brand, loss of traditional strategic customers, etc. Historical information on a site turnover, profitability, assets value, capital cost, electricity bills, electrical system maintenance cost, and past investments in PQ mitigation cost will be helpful to extend practical use of data and data analysis. Also, other than monetary data like employment, working time system, electrical system configurations, power and energy quantities, and loading, data is necessary to complete some calculations but also to analyze and benchmark the PQ impact on site activity. Data can be collected either on certain time interval basis, most practically annually (or pro rata annual basis when the frequency of disturbances is less than once per year), or on cost-per-event basis, mainly depending on the type of disturbance. For all PQ consequences, the methods to be followed can be of two types: deterministic or probabilistic. Deterministic methods are adequate when all the items of the analysis, from the operating conditions of the system to the discount rate value, are known without uncertainty. This can be the case of ex post analyses performed on existing systems whose operating conditions are repetitive and well-stated. Some real cases can refer to industrial systems. Probabilistic methods are instead needed when some of the problem variables are affected by uncertainties. This clearly happens for non-existing systems or also for existing systems where some expansions have to be planned. However, technicians are often involved in estimating the costs for the future operation of existing systems when both cash flows and operating conditions of the system vary over a range and thus introduce a degree of uncertainty. In the following the main technical and economic data to collect are summarized, the reader can be helped also by the Appendix 4 that indicates a way for structuring the data collection process.

4.3.1. Technical Data The collection of the technical data depends on the method used for estimating the costs due to PQ disturbances. Following the approach shown in Chapter 2, it is suggested to separate the methods in function of the type of disturbance. In the following, events from variations are distinguished, and in particular we’ll refer to the technical data needed for the following PQ disturbances: • • •

Voltage dips and short interruptions Harmonics Current and voltage unbalances

With reference to voltage dips and short interruptions, all the methods shown in Chapter 2 involve the estimation of the voltage-dip performance of the supply system and the evaluation of the effects of voltage dips on components and equipment. The performance of the supply system in terms of voltage dips can be assessed by measurements and by simulation. Regarding the measurements, the actual standards IEC 61000-4-30 [117] and IEC 61000-4-7 [116] define the class of instruments and the methods to detect, measure, and post-process the data. The most

50

important output of such an analysis is in the form of a voltage-dip performance chart (Fig. 4.1), tables that furnish the number of voltage dips in function of duration and amplitude (Fig. 4.2), voltage-dip pattern with time of the day (Fig. 4.3), and plots of daily variation in number of voltage dips (Fig. 4.4).

Fig.4.1: Voltage-dip performance chart [122] Fig 4.2 reports the national average number of voltage dips measured per Voltage Quality Recorder (VQR) in Italy in 2007, by the QUEEN monitoring system [118]. Residual voltage u (%) 90 > u ≥ 80 80 > u ≥ 70 70 > u ≥ 40 40 > u ≥ 5 5 > u ≥1 Total

10 - 200

200 - 500

Duration (ms) 500 - 5000

37,7 19,9 38,8 12,5 0,3 109,2

5000 - 60000

5,5 2 4,1 0,7 6,6 0,8 2,6 0,4 0 0 18,8 3,9 Year 2007, around MV 400 measuring points Fig.4.2: Voltage-dip performance chart [118]

0,1 0 0,1 0 0 0,2

Total 45,3 24,7 46,3 15,5 0,3 132,1

Fig. 4.3: Voltage-dip pattern with time of the day [119]

51

Fig. 4.4: Daily variation in number of voltage dips [119] When measurements are not available, simulation methods can be used to obtain the needed technical data. The two main methods are the Critical Distance method and the Fault Position method. Both methods are based on the simulation of the system in short-circuit conditions for the assigned position of fault. Consequently, to apply any of these methods, all the technical data for short circuit analysis is needed. Regarding the effects of voltage dips and short interruptions on the equipment and on the process, two main types of information are needed: the voltage-tolerance curve of the equipment and the connection of the equipment inside the process. This information allows someone to ascertain the most critical equipment with respect to the stopping of the process that is the most important effect of voltage dips inside an industrial premise. Regarding the effects of harmonics, again, the technical data can be obtained by measurements or by simulations. Measurements have to be done with respect of EN 61000-4-30 [117] and 61000-4-7 [116]; simulations can be performed in deterministic scenarios (usually selecting the worst condition) or in probabilistic scenarios. In the latter case, the probabilistic description of voltage and current harmonics in terms of probability density functions is needed. As evidenced in Chapter 2, usually several simplifications can be introduced to avoid excessive complexity of the study. The main information is: current and voltage harmonics (amplitude of each single harmonic component) applied at all the equipment of the system, the type of equipment exposed to harmonics (cable line, capacitor, transformer, electrical motor), and all the characteristics of the component needed to use formulas reported in Chapter 2 (see details in the Appendix 2.J). Regarding the effects of current and voltage unbalances, the following data are needed: • • • •

Power losses in the j element (equipment, load) in the considered facility due to voltage and current unbalance. Energy losses in the j element due to voltage and current unbalance. Aging of element insulation due to voltage and current unbalance. Reactive power reduction due to unbalance.

4.3.2. Economic Data The economical data that has to be collected to estimate the costs of PQ disturbances can be divided into two main categories with respect to the final effects of the PQ disturbance: i) cases when the PQ disturbance does not cause the stop of a process; ii) cases when the PQ disturbance causes the stop of a process.. If the PQ disturbance does not cause the stop of a process, then only the costs of additional losses and the premature replacement of damaged components have to be computed. These computations are straight

52

forward and can be deduced from formulas in Chapter 2. Mainly the economic data needed in this last case are: -

for the additional losses: price of energy, variation rate of energy price, discount rate; for the premature ageing: price of damaged component to substitute; variation rate of the component price; discount rate.

If the PQ disturbance causes the stop of a process, then the corresponding cost has to be computed. This quantity can be named in different ways ; for example in [123] it is called COD that is Cost of Down time; in the IEEE Std 1346 [124] it is called Cost of disruption, in [57] it is called Economical Damage. In the following two main models, called in the following as model A and model B, are presented that evidence all the economic data to collect. The first model is more general and can be applied to continuous and non continuous processes; the second method is more adequate for continuous process. Model A The process interruption cost, namely PIC, for each event is expressed as:

PIC = A1 + A2 + A3 + A4 + A5 + A6 − A7

(4.1)

with the following meaning of the symbols : A1 = cost of work in progress; A2 = cost of the labor; A3 = cost of process slow down; A4 = cost of process restart; A5 = cost of equipment; A6 = other cost; A7 = savings. In the following each cost component in (4.1) is described to evidence which are the economic data to collect. A1 Work in progress Work in progress (WIP) represents the lost or wasted work due to a process disruption; the corresponding cost CW is given by:

CW = W 1 + W 2

(4.2)

where: W1 is the cost of unrecoverable WIP; W2 is the cost of reworking recoverable WIP to a usable standard. When production and services in progress are partially wasted, W1 describes only that part which cannot be recovered; irrecoverable means that the (semi-finished) product of an interrupted process will not be repaired, used in a further process or sold as a lower quality product. The model to use for computing W1 is:

W 1 = M + E + L WIP1

(4.3)

where: M: unrecoverable material lost, consisting of purchase price cost plus overhead cost of purchase plus site transportation, less scrap or residual value; E: energy cost unrecoverably lost in WIP in €/kWh; LWIP1: labor cost defined as

53

P

LWIP1 = ∑ ni × ( Li + Oi )

(4.4)

i =1

where: P: ni : Li: Oi:

number of the i different processes/operations affected by unrecoverable lost; hours of wasted labor consumed in each ith process; hourly depreciation rates of fixed assets; hourly overheads which express rental cost of equipment owned or used by the company.

With reference to the item Oi, it is important to evidence that, in the simplest case or where differentiation into processes makes only negligible calculation difference, this will be the number of wasted labor hours multiplied by the cost of labor, including overhead costs. Both M and Li are expressed in monetary value. For that part of the WIP that can be recovered, W2 represents the cost of labor required to complete the product to normal standard. Π

W 2 = ∑ ni 2 × ( Li 2 + Oi 2)

(4.5)

i 2=1

where: Π: ni2 : Li2: Oi2:

number of the i2 different processes/operations affected by recoverable lost; hours of labor needed in each i2th process to complete the product; hourly depreciation rates of fixed assets; hourly overheads which express rental cost of equipment owned or used by the company.

It is crucial to assess how much of WIP can be recovered; during surveys staff may underestimate both W1 and W2. In reality there is often enough spare capacity to allow production to return to normal levels within a reasonably short time, but this may not be the case for time-sensitive or seasonal goods or where fresh or perishable produce is being processed. Establishing W1 and W2 requires an in depth analysis of cost elements. One practical approach would be to monitor production costs for a period following an outage and compare them with costs for a similar period with no outages. The balance between W1 and W2 is sometimes difficult to assess because the decision to reuse or scrap WIP is subjective and may depend on conditions at the time; for example, if raw material were in short supply or on long lead times, WIP would be more likely to be reused. A2 Labor The cost of lost or idle labor is the cost of staff who are unable to work due to a process interruption, starting from the moment of interruption and ending when normal process activity resumes. It is indicated with the symbol LOUT as:

LOUT = ∑ (ni 3 × ( Li 3 + Oi 3) ± n'i 3 ×( L' i 3 + O'i 3))

(4.6)

i3

The meaning of symbols in (4.6) is analogous to the one of the formula (4.5). Symbols marked with a prime (′) indicate terms which account for the redeployment of labor to other tasks (not related to the stoppage) or the employment of additional labor to aid recovery of the process. If the staff responsible for the i3th operation or process is completely idle and no additional labor cost will be required to compensate for lost time, the prime value will be zero. If the staff is redeployed to other tasks which are necessary but are not stoppage related, the difference in the working time, labor and overhead rates are accounted for in the prime terms which would be

54

subtracted from the non-prime elements. If additional labor was required to recover from the stoppage, the prime terms would be added to the non-prime terms. A3 Process slow down This cost subcategory can be used either as a supplementary element to process interruption costs or as a specific alternative approach for some sectors. It is a supplementary element, for example, when a process is restrained by the failure of another. In the case of this cost component is used as an alternative approach to the cost of interruption, care must be taken to avoid overlap with other process interruption costs. If equipment or a process is affected by a power quality disturbance, activity may be reduced, e.g. because only a fraction of parallel processes are operating, operate at a slower rate or some fraction of the product is out of specification. The value of the related economic cost, P, is estimated per single event as:

P = P1 + P 2 .

(4.7)

In the equation (4.7), P1 is the consequential loss caused by the reduction in efficiency below normal due to reduced production due to limitations on capacity or speed, temporarily loss of synchronization, additional restarts and resets, re-calibration, repair and maintenance and increased defect rate; P2 is the cost of dealing with out-of-specification product by, e.g., repair, rework, recycling or scrapping. The component P1 is given by:

P1 =

T n ∆Eff WT

(4.8)

where: T:

the annual cost of sales (i.e. turnover minus profit). In cases where no raw materials (including energy) are lost, the annual ‘added value’ should be used. If the effect is simply to reduce the efficiency of labor, the annual cost of labor should be used; normal annual working time of organization. This is average time in a year when company is WT: working, taking account of shift patterns, holidays, etc. n: number of hours for which efficiency is reduced ∆Eff : percentage reduction of level of activity. This is the best assessment of the reduction in performance (as a sole result of PQ disturbance) compared to normal activity from the broadest possible perspective. A4 Process restart When a process is interrupted, other ancillary processes, such as heating, cooling, ventilation and filtration, may also trip. These processes must be re-established and verified before the main process can restart, requiring additional time and labor. Some checking procedures, such as cleaning in paper or food production, may also be required. The corresponding process restart cost, namely PR, typically takes into account different cost items like materials, consumables (calculated directly in monetary units) and labor calculated using equation (4.4). In cases where the interrupted process is restarted from an independent power supply until normal power supply conditions are restored, all the operating costs of the generating equipment form part of the PR. A5 Equipment When a process is interrupted the shutdown occurs in a disorderly manner and it is possible that equipment will be damaged as a result. Damage may be instantaneous (e.g. damage by mechanical collision) or incremental (e.g. by overheating due to loss of coolant) leading to shorter equipment life, increased maintenance, etc. The value of the equipment damage. Namely E, is estimated per single event as:

55

E = E1 + E 2

(4.9)

Where E1 is the main equipment damage cost, and E2 account for additional maintenance, repair, material and consumables costs as E1 consists of: • • • • •

cost of equipment and tools damaged beyond repair and scrapped. Scrap or recovered value should be deducted cost of repair, adjustment and calibration of damaged equipment and tools cost of installation of new equipment, parts and tools cost of hiring replacement equipment other indirect costs of equipment damage, e.g. additional costs for backup equipment, extra (compared to normal scenario) cost of additional overhaul in future.

This category typically includes transformers, capacitors, motors, cables, contactors, relays, protection and control, computers, lights, tools. E2 account for all the additional costs as a consequence of the process interruption. Examples are bearings, pads, fuses, compressed air, water, oil. Value of this cost should be expressed directly in monetary units. A6 Other cost Other process interruption related, usually indirect, costs, include: • • • • • • •

•

penalties due to contract non-delivery or late delivery environmental fines / penalties cost of evacuation of personnel and equipment (also external) costs of personnel injury (additional inability to work) increased insurance rates (equipment, personnel health, liability) compensation paid out hidden costs from loss of: o competitiveness o reputation o customer satisfaction and, as a consequence, lost opportunity of subsequent revenues o employee tolerance Other non-specified indirect or direct costs.

A7 Savings It is likely that interrupted operations can save money or defer expenditure. This usually is defined in terms of ‘unused raw materials’, ‘unpaid wages for contracted/temporary staff’ or ‘savings due to reduced energy usage’. Because of idle time resulting from PQ related disruption, some ‘savings’ might be generated. The value of these savings as a consequence of the particular PQ disturbance can be: • • • •

savings from unused materials or inventory savings from wages that were not paid savings on energy bill other specific savings.

Model B The cost of disturbance, namely COD, that causes the stop of a process can be computed using the following generalized formula: COD = DC + RC + HC

(4.10)

56

COD is the cost of downtime time as a result of a disturbance, DC is the direct cost, RC is the restart cost, and HC is the hidden cost. The tool presented here builds on the work reported in [123] and experience gained through discussions with pharmaceutical manufacturing plant personnel. The proposed COD estimation tool is strictly applicable to aseptic manufacturing processes. However, general principles of the developed methodology are applicable to any continuous manufacturing process. DC component in a COD context refers to production cost accrual at a given instance of disturbance, and thus is a function of time and process activity. Most manufacturing sectors involve the following direct cost components: raw material, energy, labor overheads, outage savings, and profit lost. A brief discussion on each of these cost components is presented below. Raw Material. Manufacturing process disruption, either partial or complete, involves a significant amount of raw material wastage (usually referred as scrap), some cost savings of the material that would be otherwise added to achieve a finished product, and cost of recycling of the affected product. The damage to the product is not always observable. In the case when this happens though, product damage can be costly if the damage is subtle and the effects take time to surface [123]. Additionally, high disturbance frequency increases the burden on manufacturers to store excess raw material, leading to an increase in warehouse use, space, storage, and maintenance cost. Energy. Although electrical energy is most commonly used among industrial sectors to power/run processes, other forms of energy consumption such as steam, gas, and coal are also widely used. The energy cost of product damage is a sum of a plant’s base energy use (e.g. lighting, PCs, etc.) and progressive energy consumption cost until the instance of plant disruption leading to product damage [123]. Labor. The labor cost in context of direct cost is the money paid for the labor to work on the product until the instance of process disruption. Labor contracts, usually annual or seasonal contracts, are generally inclusive of reserve hours that may be required to make up for downtime scenarios. In the event when there is little or no use (due to fewer process disruptions or problem mitigation) of reserve hours, the company can claim back monies paid for unused hours. Overheads. Overhead includes marketing and sales cost, administrative cost, annual plant maintenance cost (e.g. equipment repair due to wear and tear, consultants, electrical contractors, etc.), and site service cost. Overhead cost is part of the direct cost and does not include the disturbance-related repair or damage, or restart costs, which are treated as a separate cost in this study. Lost Opportunity. Process disruption leads to interrupted sales or severely impacted revenue flow, resulting in delayed production schedules [123, 125]. This is usually an identifiable or observable cost. Penalties. Occasionally, damaged product due to PQ disturbances can cause companies to be penalized for not delivering the order on time, because it might affect the customer’s production line. And in some cases, this might even upset company shares and reputation. Memory chip-maker Samsung Electronic Co. in August 2001 reported that it could have incurred a total loss of $54.19 million in damage as a result of a power outage [126, 127]. As a consequence, the company shares dropped more than 2% in value from previous closing price [127]. th

th The computation of the DC component for n process activities, j failure and u product variant

th

processed at each i process activity is described in the following. For sake of generality, the costs are here expressed in Unity Money (UM); the same quantities are expressed in pound (£) in the original version [6].

•

phui : Amount of product handled in % at i th process. th rui : Cumulative raw material cost in UM at i process. osui : Outage savings accrued for product handled in UM at i th process, following a

•

eui :

• •

complete/partial process disruption. th

Cumulative energy cost in UM at i process.

57

lui : oui : prui :

• • •

th

Cumulative labor cost in UM at i process. th

Cumulative overhead cost in UM at i process. Profits lost for product handled in UM at i process disruption.

th

process, following a complete/partial

peui : Penalties accrued for product handled in UM at i th process. prmui : Progressive raw material cost in UM at i th process ( phui × rui ). th sui : Progressive outage savings accrued for product handled in UM at i process, following

• • •

a complete/partial process disruption.

pec ui : Progressive energy cost in UM at i th process ( ph × e ). ui ui plc ui : Progressive labor cost in UM at i th process ( ph × l ). ui ui poc ui : Progressive overhead cost in UM at i th process ( phui × oui ). pplui : Progressive profits lost for product handled in UM at i th process, following a complete/partial process disruption ( phui × prui ). ppaui : Progressive penalties accrued for product handled in UM at i th process ( ph × pe ).

• • • • •

ui

Direct cost in UM at i

th

ui

process is given as,

dc ui = phui (rui + eui + lui + oui + prui + peui ) − sui

(4.11)

Total direct cost is given as, y

n

DC = ∑ ∑ dc ui

(4.12)

u =1 i =1

For the RC component, the costs include damage assessment cost accrued as a result of hiring either internal or external consultants or contractors, equipment and production material and consumables lost, damage, repair and replacement cost, wasted energy, and finally idle, restart, and overtime labor cost to recover for lost production time. Each of these costs is discussed briefly in the following subsections. Expert Damage Assessment. Occasionally, internal or external expert damage assessment is required through consultants, contractors, etc. Lost, Damage, Repair, and Replace. This category includes costs due to loss, damage, repair, and replacement of manufacturing equipment, consumables (e.g. radiated sterile packs of pens, etc.), or production material (e.g. plastic containers holding materials). Restart Energy. This is the energy consumed by all or part of the plant from the moment of failure until the system is brought back to normal operation. Idle and Restart Labor. Labor hire can be hourly, seasonal, or annual, depending on the nature of industrial sector, product, and the individual manufacturing plant’s labor hire practices. Seasonal and annual hire contracts usually take into account certain number of lost hours in account, which is either claimed back or left unclaimed. Either claimed or unclaimed, this is still an additional cost of downtime, which is not clearly observable. However, when the hire is usually hourly or in scenarios where the plant needs labor in addition at an hourly rate to restart or regain normal operation, the cost of this additional labor is clearly observable. The computation of RC component can be effected as follows. th edauij : Expert damage assessment cost in UM for j th failure at i process activity.

58

ldrruij : Lost ( lo ), damage ( da ), repair ( re ) and replace ( rp ) of parts, production material etc, for j th th

failure at i process activity, in UM ( louij + dauij + reuij + rpuij ).

enqij : Energy cost in UM consumed from instance of failure to restart for j th failure at i th process activity. rlcqij : Idle labor cost ( il ), restart labor cost ( rl ), labor overtime to recover at later date ( rlo ) in UM for j

th

th

failure at i process activity ( il uij + rluij + rlouij ).

Cost of restart for j th failure at

i th process activity is given as,

rcuij = edauij + ldrruij + enuij + rlcuij

(4.13)

Total restart cost at any given instance for j given as,

q

m

th

failure selected/assessed at each i

n

th

process activity is

(4.14)

RC = ∑ ∑ ∑ rc uij u =1 j =1 i =1

For the HC component, the costs usually result from damage, or losses, not immediately or readily observed [6]. One method of quantifying this factor is through surveys conducted among plant personnel and customers, and comparing availability score among competitors. Decreased Competitiveness, Reputation, and Customer Dissatisfaction. High frequency of process disruption leads to poor product quality and reduced availability (usually quantified using Overall Equipment Effectiveness index, which is a product of equipment availability, performance, and yield), which in some cases can lead to delayed production schedules. These shortcomings certainly decrease competitiveness, reputation, customer satisfaction, and loss of customer loyalty that can prove very costly [123] and difficult to quantify. Employee Annoyance as a Result of Stoppages. PQ damages occasionally cause annoyance among employees, especially disruptions leading to significant personnel involvement, cleaning, overtime work schedules to recover lost time, etc. This factor is not readily quantifiable in terms of reduced efficiency. The computation of HC component is as follows. •

rct ui :

th

Retained competitiveness in p.u. from nominal as a result of lost product at i process

activity. •

rrt ui :

th

Retained reputation in p.u. from nominal as a result of lost product at i process

activity. • •

rcsui :

Retained customer satisfaction in p.u. from nominal as a result of lost product at i process activity.

ret ui :

Retained employee tolerance in p.u. from nominal as a result of lost product at i process activity.

Hidden cost factor for j

th

th

th

th

failure at i process activity is given as,

hcf uij = rct uij × rrt uij × rcsuij × ret uij

(4.15)

Total hidden cost at any give failure instance is given as,

59

q

m

n

HC = ∏∏∏ hcf uij u =1 j =1 i =1

Other factors can influence the COD estimation, as listed in the following: Hit Rate and Miss Rate. The typical definitions for “hit rate” and “miss rate” are not readily available in manufacturing literature. The following definitions were adopted following correspondence with typical continuous manufacturing plant personnel. Hit rate is the ratio of intended use of resources used and sum of intended and un-intended (e.g. process failures) use of resources. Miss rate is the ratio of unintended use of resources used and sum of intended and unintended use of resources. Thus miss rate is given as, Miss rate = 1 – Hit rate Pass Rate and Fail Rate. As before, typical definition of “pass rate” and “fail rate” is not readily available in manufacturing literature. However, use in various research publications [128, 129] suggests the following definitions of these terms. Pass rate is the ratio of product number or batches that passed a set criterion and total number of product number or batches initiated. Fail rate is the ratio of product number or batches that did not pass a set criterion and total number of product number or batches initiated. Thus fail rate is given as, Fail rate = 1 – Pass rate

4.4. DNO Perspective: Data Collection As far as DNOs are concerned, the first economic aspects linked to PQ are the consequences of noncompliance to certain PQ levels defined by a contract with an end user. The costs can be directly identified in the contract as a penalty to be recognized to the customer but also include further elements. For solving the customer complaints, the utilities must face several costly activities that include personnel for communication, measurement campaign, data analysis, and so on. Despite these, they also suffer from losses of unsupplied energy when end users are not using electricity for their activities because of interrupted power supply as a consequence of inadequate PQ. The capital and operating expenses of mitigation equipment and systems improving PQ are further elements of cost that have to be properly collected. The objectives for data-collection for DNO are: • • • •

Quantify the existing level of power quality. Identify potential power quality issues. Identify potential improvement opportunities. Quantify costs of poor power quality.

The data to be collected are very copious and different. Appropriate forms help to collect these data. In the following several tables are shown examples of possible forms. Some preliminary information on the DNO are useful; some of the most important are: Costs incurred while answering/responding to customer PQ enquiries/complaints, including costs of: - Call centers - Responding crew - Inspection - Monitoring - Consultation - Mitigation

60

- Follow-ups Costs incurred to maintain/improve quality of supply - Fuse replacements (reliability) - Reclosing schemes (reliability) - Fast switching with instantaneous protection (voltage sags) - Pole and tower grounding improvements (voltage sag) - Increased sectionalizing (reliability and voltage sags) - Surge arrestors and transient voltage surge suppressors (transients) - Lightning protection – shield wires (transients, voltage sags and reliability) - Conductor spacing modification (reduce faults) - Insulate/cover overhead conductors (reduce faults) - Underground cables (reduce faults) - Harmonic filters (harmonics) - Increase size of neutral conductor (harmonics) - Zigzag transformer (harmonics) - Redundant feeders (voltage sags, reliability) - Fault current limiting (voltage sags) - Capacitors for voltage regulation (voltage regulation) - FACTS devices (voltage regulation and other) - Animal guards (reduce faults) - Arc suppression coil earthing with time grading protection - Feeder design modification to increase reliability - Protection co-ordination modification to increase reliability - Loop schemes - New/replacement of old feeders/transformers to improve reliability and power quality: . Maintenance costs - Tree trimming - Insulator washing - Cable and transformer maintenance - Switchgear maintenance: Costs to provide standard quality of supply to customer. All costs involved in order to provide a certain/standard quality of supply to a particular customer, which could be avoided if the customer was not connected to the network. - Switchgear (circuit breakers, ring main units, auto reclosers, switch disconnectors, sectionalizers, expulsion fuses) - Transformer - Cable - Overhead conductors (insulated, uninsulated, and covered) - Steel tower and wood pole structures (struktura dalekvodnih stubova I bandera) Other data to evaluate • number of individual consumers (revenue meters); • annual revenues; • size of population served (or similar information like concentration areas); • number and type of employees. Regarding the last point (number and type of employees), it can be useful to adopt a table like the following. Function Management Corporate services Finance Management Information System (MIS) Operations Engineering

Employees

61

Procurement Design Maintenance Human resources Customer support Research & development Total Assets • • • •

Total circuit length (if possible give details for each voltage level) Number of distribution transformers (< 1600 kVA) Number of medium-power transformers (1600 - 5000 kVA) Number of large power transformers (> 5000 kVA)

4.5. Conclusions The approach presented in this chapter facilitates the collection of technical and economical data for the economical evaluation of costs both for end-users and for DNO. In both cases it is necessary to know the characteristics not only of the disturbances but also the characteristics of the electrical system suffering for PQ degradation. The chapter gives all the terms to be collected for the cost evaluation of a process stop due to disturbances. Two approaches are described in this chapter; the first one is more general and the second one is more tied for continuous process. Both the presented approaches are useful to clarify all the economic items to collect for computing the overall costs.

62

5. Methodology for the Economic Assessment of Power Quality Solutions 5.1. Introduction The costs to industrial and commercial electric power end users from unmitigated power quality (PQ) and reliability phenomena are significant and have been well documented by detailed studies [131]. These studies have focused principally on quantifying the actual or reported cost to businesses of PQ and reliability phenomena that result in unplanned businesses losses brought about by such factors as process interruptions, equipment damage, extra labor costs, and increased scrap. Although many of these studies also inquire about mitigation equipment employed by end users to try to minimize the business impact of PQ and reliability phenomena, in general, the numbers given for the “cost of PQ and reliability” focus only on the impact of unmitigated phenomena and exclude the cost of preventing unplanned business losses. As such, an unprotected facility might be said to suffer significant PQ and reliability costs, while a facility protected with, say, a double-redundant uninterrupted power supply (UPS) and N+1 backup generation might be said to suffer no PQ or reliability costs whatsoever—a circumstance that does not reflect true business decision-making wherein the costs of outages are balanced against the costs of mitigation. Because of this, a comprehensive strategy to evaluate optimization of overall PQ-related cost is needed, including: • Costs to industry and electric power providers of unmitigated PQ phenomena • Costs to industry and electric power providers of prevention and mitigation of the impacts of PQ phenomena The key challenge is to balance both of these broad cost categories. Although any number of economic analysis approaches may be employed to arrive at an optimum, this chapter emphasizes a simple 10-year Net Present Value (NPV) approach, whereby all costs and benefits may be combined to determine the mitigation scenario that optimizes today’s economic performance.

5.2. Economic Analysis of the Costs of PQ Power quality and reliability continue to grow in importance with deregulation of the electric power industry. The factory automation and wide-spread use of electronic goods has put the electric utilities under severe stress to improve the quality of supply and service to customers. As plant operations and processes are becoming more and more automated, the need to keep the equipment operating is of utmost importance. Any downtime can be directly correlated to lost production, revenue, and profits.

5.2.1. Economic Impact of Power Quality Variations Out of various power quality disturbances, voltage dips are the most frequent [131] and cause the greatest loss of revenue because they result in frequent malfunction of equipment. Therefore, the ability of modern industrial process equipment to ride through voltage dips is becoming more important than never before. Especially, the equipment used in modern industrial processes, e.g. programmable logic controllers (PLCs), adjustable-speed drives (ASDs), computers, and motor-contactors, are highly sensitive to voltage dips [132]. Another useful data point is an analysis conducted by EPRI based on the company’s more than 500 investigations of PQ and reliability-related problems at end-user facilities. These were prompted primarily by utilities responding to customer complaints, and the investigations have spanned the spectrum of standard industry classifications (SIC) codes ranging from commercial buildings to transportation to food

63

processing to plastics, printing, and chemical processing. Figure 5.1 shows a snapshot of the types of PQ phenomena that have, upon investigation, turned out to be the culprit in these investigations. In EPRI’s experience, the single most potent cause of end-user PQ problems is voltage dips or swells. Given the number of these sorts of events documented by the EPRI DPQ study, this should come as no surprise. The second most frequent contributor is harmonics (unwanted frequencies in a facility’s voltage or current waveforms). The next largest contributor is grounding and other wiring issues. Collectively, these three PQ phenomena account for more than 85% of PQ investigations conducted by ERPI over the years.

Root Causes From 500 EPRI PEAC Investigations Electromagnetic Interference

2%

Other

2% 5%

Load Interaction

6%

Cap. Switching Transients

15%

Grounding

22%

Harmonics

48%

Voltage Sags & Swells

0%

10%

20%

30%

40%

50%

60%

Percent of Total Figure 5.1 Breakdown of the power quality phenomena found in more than 500 EPRI investigations Figure 5.2 illustrates a compilation of the sources of PQ-related complaints from customers. It is interesting to note that only 5% of complaints were the result of complete interruptions of power, with 95% precipitated by the combined effects of voltage dips and surges.

The Most Common End User PQ Problems Voltage Swells

1%

Interruptions Transient Over-voltages

5% 7%

Voltage Sags 0%

87% 20%

40%

60%

80%

100%

Percent of Reported Events Figure 5.2 Common PQ problems as reported by a large North American utility4

4

“Basics of Power Quality and Surge Protection,” Florida Power Corp., 1998.

64

5.2.1.1. Assessment and Prediction of Voltage Dips Clearly the monitoring of voltages at power system buses is the best way to assess voltage-dip performance. If high accuracy of the monitoring data is required, however, it may take quite a long time, typically several years [149]. In practice, fault location and the type of fault may vary with time depending on weather conditions and geographical location. Furthermore, power system networks, distribution networks in particular, are changing (e.g., changing network topology, different maintenance practice, aging of equipment). Therefore, conclusions based on historical monitoring results could yield unreliable assessment of voltage dip performance [150]. The other possibility to assess voltage-dip performance is by using a stochastic approach based on computer simulations. This is generally the most suitable way to assess voltage dips during the power system planning stage when the actual system (or part of it) may not exist yet or to assess system dip performance for different operating scenarios and loading conditions. Moreover, it does not take many years of monitoring to obtain the required accuracy of the dip-performance data. These are clearly advantages over the monitoring approach. Along with the stochastic approach, the method of fault positions and the concept of the area of vulnerability are generally used to assess and understand the system voltage-dip performance. The method of fault positions is a simple way to determine the expected number of dips and their characteristics at power system buses. In this method, system fault analysis including symmetrical and asymmetrical faults is initially performed at each fault position. Each fault position represents faults in a specific part of the system. Remaining voltages and phase angles at each bus in the system during the fault are determined; and the related phase angle jumps are computed. After taking into account corresponding fault rate at each fault position and fault clearing time, the expected number of dips as a function of magnitude, duration, and phase shift is calculated [151, 152]. Accuracy of this method depends on modelled fault positions in the system. Different fault positions will result in different magnitudes of dips according to system impedance, fault impedance, and type of fault. For the higher accuracy of the results, more fault positions need to be used [153]. The network model and the reliability of the historical fault statistics data used in the analysis are the other factors that will govern accuracy of the results. The methodology used is based on the guidelines recommended by the IEEE Std 493-1997 (Gold Book). Two different approaches are used to calculate the total expected number of dips due to the faults inside the area of vulnerability and due to the faults along the boundary crossing lines. Dip performance within the area is assessed by using equations (5-3) to (5-5). A combination of the method of fault positions and exponential fault distribution pattern along the boundary crossing lines is used in determining the number of voltage dips following the faults. The fault rates at each location along the boundary crossing lines are calculated using the exponential distribution pattern and then applied to the corresponding fault positions. After taking into account calculated fault rates, the expected number of dips due to the faults along the boundary crossing lines is calculated. 4

n

∑∑

NSBF = i=1 4

j=1

B × BFR

∑∑

NSLF = i=1 j=1 L × LL × LFR With probabilistic occurrence of a fault included, equations (5.1) and (5.2) become: 4

n

3

j=1

k=1

n

3

∑ ∑∑

NSBF = i=1 4

B × PO × BFR

∑∑∑

NSLF = i=1 j=1 k=1 L × LL × PO × LFR CNS = NSBF + NSLF Where B L

(5.1)

n

(5.2)

(5.3)

(5.4) (5.5)

= Bus inside the area of vulnerability = Line inside the area of vulnerability

65

LL PO BFR LFR CNS NSBF NSLF i n j k

= Length of the line inside the area of vulnerability = Probabilistic occurrence of fault (1 for symmetrical fault and 1/3 for asymmetrical faults) = Bus fault rate = Line fault rate = Cumulative number of dips = Number of dips due to the bus faults = Number of dips due to the line faults = Type of fault = Total number of buses or lines inside the area of vulnerability = Bus or line inside the area of vulnerability = Number of phases

5.2.1.2 Overview of Equipment Sensitivity In order to establish the consequences of voltage dips at a given point of common coupling (PCC), voltage-dip characteristics are compared with the equipment voltage-tolerance (sensitivity) curves to assess its performance (i.e., whether the equipment would trip/malfunction or ride-through the dip of specified characteristics). This analysis consists of preparing a dip-performance chart (i.e., magnitudeduration chart) for a particular bus in the system and superimposing it on the equipment voltage-tolerance curves and thereby obtaining a single graphic display [134-139] from which equipment ride-through capabilities could be determined. The information about the sensitivity of individual equipment can be obtained either from the equipment manufacturer, available standards, or through laboratory tests. In case of non-availability of prior information, testing is the most reliable and efficient way to identify equipment sensitivity to voltage dips. However, determining equipment sensitivity is the most difficult task when assessing voltage-dip consequences because different categories of industrial equipment have different sensitivities to voltage dips [136]. Furthermore, different devices, even belonging to the same equipment category, do not exhibit the same sensitivity to voltage dips [131, 136-147]. On the other hand, it is not reasonable to test all the sensitive devices in customer facilities. Therefore, devices are classified into various categories based on device type. Testing of an adequate number of devices representing one equipment category justifies generalization of the acquired results. Because different brands of the same equipment type and even different models of the same brand often have different sensitivity, typical sensitivity data with appropriate statistical deviation and error parameters can be determined for a particular equipment type. The sensitivity information so obtained needs to be updated continuously as and when more test results are made available. The evaluation of the impact of voltage dips at particular site in the network involves three basic steps fault analysis, voltage-dip analysis, and economic analysis. In fault analysis, the method of fault-positions [148] is used in which various types of faults (symmetrical and asymmetrical) are simulated at numerous locations throughout the system network and corresponding expected voltage magnitudes and durations (assuming 100% reliable primary protection, i.e., the duration of voltage dips is determined by the primary protection settings) are determined at various network buses. In subsequent voltage-dip analysis performed at a point of common coupling (PCC), the frequency of dips of specified dip magnitude and duration over a period of interest is determined by associating it with the historical fault performance (fault per km per year) of all network buses, overhead lines, and underground cables. This information is generally available from historic data obtained through long-term monitoring at respective locations in the network. The corresponding duration of voltage dips depends on fault-clearing times of protective devices used in the power system network. (In this analysis, it is assumed that the primary protection is 100% reliable and that all the faults are cleared by the primary protection.) The final and the most crucial step for the economic assessment of power quality requires the information about the consequences of expected voltage dips on the performance of industrial processes. In order to determine whether the equipment will trip/malfunction or ride-through the dip of specified magnitude and duration, expected voltage dips are compared with the sensitivity of process equipment connected at a given bus. This procedure requires preparing a dip-performance chart for a particular bus in the system

66

and coordination of the customer equipment responses with these voltage dips on a single graphic display [135]. For this purpose, the precise information about the equipment sensitivity is required for the accurate quantification of their nuisance trips due to voltage dips. The information about the equipment sensitivity may be gathered from the equipment manufacturer or by testing. The testing of each and every piece of equipment is neither justifiable nor possible. Therefore, sensitive industrial equipments are classified into various equipment categories based on equipment types, and then the testing is performed on a suitable number of equipment picked up randomly from each category. However, even though the equipment may belong to the same equipment category, it might not exhibit the same sensitivity to voltage dips [132]. This makes it difficult to develop a single standard that defines the sensitivity of process equipment. In addition to this, it is also possible that a process may get disrupted due to tripping of individual equipment or it may require the tripping of a group of equipment depending upon their interconnections. Therefore, for any assessment of financial losses incurred in a customer facility due to voltage dips, the precise counting of process (not individual equipment) trips is essential. The probabilistic assessment of the number of process trips incorporates the uncertainty coupled with the equipment sensitivity as well as the uncertainty associated with possible connections of various equipment involved in an industrial process. The equipment sensitivity to voltage dips is usually expressed only in terms of the magnitude and duration of the voltage dip. For this purpose, the rectangular voltage-tolerance curve (as shown in Fig. 5.3) can be used. It indicates that a voltage dip deeper than the specified voltage magnitude threshold (Vmin) and longer than the specified duration threshold (Tcrit) will cause malfunction (or trip) of the equipment. However, in practice, most of the equipment, e.g., motor-contactors and household electronics items, would have non-rectangular voltage-tolerance characteristics [144-146]. Other two parameters, which may be detrimental to sensitivity of some of the industrial equipment (though to a lesser extent than voltage-dip magnitude and duration) such as motor contactors, are point-on-wave of dip initiation and phase-shift during the dip [135, 144-146]. V 1.0 Normal operation Vmin Malfunction / Trip 0

Tcrit

t (ms)

Figure 5.3 Equipment voltage-tolerance curve 5.2.1.3 Uncertainty Involved with Equipment Sensitivity After laboratory tests, when the voltage-tolerance characteristics of all the equipment from the same equipment type are drawn on a two-dimensional (magnitude/duration) chart, it is found that all the equipment belonging to a particular equipment category do not exhibit the same sensitivity against voltage dips [144-147]. Even the same equipment acquires different sensitivities depending on power system conditions and loading of the equipment. However, most of the equipment exhibit, more or less, perfect rectangular characteristics as a first approximation. Voltage magnitude-threshold and durationthreshold may vary between Vmin and Vmax and between Tmin and Tmax, respectively. Therefore, the voltage-tolerance curves of these equipment may occur anywhere inside the shaded region on a voltagedip magnitude v/s duration chart shown in Fig. 5.4, such that the knee point of curve always remains inside sub-region C.

67

Voltage (%)

Vm ax C

B

V m in A 0

T m in

Tm ax

D u ra tio n (m s .)

Figure 5.4 The region of uncertainty for voltage tolerance curves of sensitive equipment 5.2.1.4 Counting of Equipment Trips In counting the number of equipment trips/malfunctions, the number of dips occurring at the customer site below the voltage-tolerance characteristic of the equipment should be taken into account. The random behavior of the equipment (represented by the shaded area in Fig. 5.4) poses a problem; i.e., which voltage-tolerance characteristic should one consider as the equipment may not have a single sensitivity curve but a family of curves inside the associated (shaded) region of uncertainty? This uncertainty of voltage-tolerance curve location inside the region can be taken care of by knowing the likelihood (i.e., assigning certain probability to each possible curve) of the individual sensitivity curve location inside the possible range. Method 1: Ordinary Probability Approach The equipment sensitivity is a bivariate random variable (T,V) where T and V are two statistically independent discrete random variables. (Note: In calculating voltage dips at different buses, the operation of protection system was typically not modeled. It is assumed that all faults are cleared by primary protection, and that fault clearing times are fixed for specific voltage levels, i.e., 80 ms at 132 kV, 150 ms at 33 kV, and 300 ms at 11 kV. Based on this, one can assume that T and V are independent). T is the voltage duration-threshold varying between Tmin and Tmax, and V is the voltage magnitude-threshold varying between Vmin and Vmax. Therefore, if fX(T) and fY(V) are the probability density functions for random variable T and V respectively, then the joint probability density function for the bivariate random variable (T, V) is given by Bayes rule [154] as: fXY (T, V) = fX(T) fY(V)

(5-6)

For the equipment having rectangular voltage-tolerance characteristic, the knee of all characteristics resides inside the sub-region C (see Fig. 5.4). This means that the total sum of probabilities (fXY (T, V)) of occurrence of the knee of the equipment characteristics being inside the sub-region C is unity. The general trend, i.e., the location of the voltage-tolerance curve inside the shaded region for particular equipment or equipment-type (i.e., whether the equipment has high, low, moderate, or uniform sensitivity), can be represented by various types of probability density functions [155] for the two random variables V and T as follows: a) Uniform sensitivity: If there is equal probability that the equipment voltage tolerance curve may assume any location within the region of uncertainty, it can be represented by assuming fX(T) and fY(V) to be uniform probability density functions for V and T within their respective ranges. b) Moderate sensitivity: This type of sensitivity can be expressed by assuming fX(T) and fY(V) to be normal probability density functions so that there is higher probability that knee of the equipment’s sensitivity curve will occur in the centre of the region of uncertainty, i.e., sub-region C. c) High sensitivity: If probabilities are assumed in exponentially decreasing order from high voltagethreshold to low voltage-threshold and from low duration-threshold to high duration-threshold, it will represent highly sensitive equipment having very poor ride-through capabilities against voltage dips.

68

d) Low sensitivity: Exponential distributions assumed opposite to the previous case (reverse exponential distributions) will represent equipment with low sensitivity, i.e., having very good ride-through capabilities against voltage dips. After calculating different joint probability density functions using (5-6), the expected number of trips (ENT) of particular equipment (considering one type of sensitivity (a) – (d) at a time) can be determined as follows:

ENT (T , V ) = f XY (T , V ) ⋅ N (T , V )

ENT = ∑ ∑ ENT (T , V ) T

Where,

V

Tmin ≤ T ≤ Tmax and V min ≤ V ≤ V max

(5-7)

(5-8)

Where fXY (T, V) is the joint probability density function for the knee of a specific voltage-tolerance curve inside the region of uncertainty such that ΣfXY=1and N (T, V) is the number of expected equipment trips (with corresponding voltage tolerance curve) as a result of voltage dips at a given location (PCC). Only one voltage-tolerance curve is considered at a time. (Example: Assume that the probability of equipment having a voltage-tolerance curve defined by 60% magnitude and 100 ms duration is f100,60 = 0.4. Assume further that based on calculated dip performance at a particular bus (PCC), it is found that there will be N (100, 60) = 50 trips of particular equipment having that sensitivity curve. The expected number of trips of this particular equipment therefore will be ENT(100,60) = 20. Total expected number of trips of a particular equipment-type at a given bus (PCC) is obtain by summation using (5-8) over the whole range of dip magnitudes and durations defined by the region of uncertainty given in Fig. 5.4) Method 2: Cumulative Probability Approach Method 1, discussed in the previous section, sums up all the trip contributions made by various possible sensitivity curves for a particular equipment-type, considering one sensitivity curve at a time and multiplying it with its respective probability of occurrence. Method 2, on the other hand, works on the premise that on the occurrence of a voltage dip, the sensitivity acquired by the equipment at that moment will decide whether the equipment will trip or ride-through that voltage dip. This makes use of cumulative probabilities instead of simple probabilities as in the case of the Method 1. The reasons for considering cumulative probability are discussed below.

1.0 1

No trip: p = 0

A2

1

0

Voltage (%)

C3

A1

0

C2

C1

B

?: 0
Trip: p = 1

Duration (ms.)

Figure 5.5 Expected behavior of sensitive equipment against voltage dips of different characteristics There is an uncertainty involved with equipment sensitivity (as discussed in section II-B) because equipment may not have a single sensitivity curve but a family of curves inside the region of uncertainty associated with the equipment-type. A piece of equipment may have any voltage-tolerance characteristic inside permissible range. This paves the way for the stochastic assessment of likelihood of equipment having a particular sensitivity inside the permissible range at the time of occurrence of a voltage dip and consequential impact on the equipment operation.

69

Let us consider six different voltage-dip events namely, A1, A2, B, C1, C2, and C3 on the voltage-dip chart as depicted in Fig. 5.5. It is obvious from the figure that voltage dips A1 and A2 will never cause any malfunction or trip of the equipment, and therefore the probability of equipment trip is zero. Similarly, voltage dip event B will certainly cause the tripping of the equipment and hence the probability of the equipment trip is unity. However, the behavior of the equipment for voltage dips C1-C3 will depend on the actual sensitivity characteristics of the equipment at the time of these voltage dips. It implies that there is a certain probability of equipment either surviving these voltage dips or tripping when exposed to them. In the case of voltage dip C1, any sensitivity characteristics occurring on the left of voltage dip C1 will result in equipment trips. Therefore, one has to consider the probability of all these sensitivity characteristics. This can be taken care of by considering the cumulative probability in such a way that its value increases gradually from zero (for the left-most characteristic, i.e., left hand side thick solid line) to one (corresponding to the right-most characteristic, i.e., right hand side thick solid line). Similarly, all the sensitivity characteristics occurring above the voltage dip C2 will lead to tripping of the equipment, and thus, one has to consider cumulative probability (instead of probability) of occurrence of these characteristics such that its value increases from zero (corresponding to top-most characteristic) to unity (corresponding to bottom-most characteristic). In the case of voltage dip C3, all the characteristics falling on the left, and above C3 will cause the tripping of the equipment. Having in mind this discussion, the variation in equipment sensitivity can be represented in terms of univariate random variable (T) in sub-region A, uni-variate random variable (V) in sub-region B, and bivariate random variable (T, V) in sub-region C (see Fig. 5.4) where T and V are assumed to be two statistically independent discrete random variables. T is the voltage duration-threshold varying between Tmin and Tmax (determined by the protection settings) and V is the voltage magnitude-threshold varying between Vmin and Vmax. Therefore, if pX(T) and pY(V) are the probability distribution functions for random variable T and V, respectively, then the joint probability distribution function for the bivariate random variable (T, V) in sub-region C is given by Bayes rule [156] as follows: pXY (T, V) = pX(T) pY(V)

(5-9)

Tmin ≤ T ≤ Tmax , V min ≤ V ≤ V max The general trend of sensitivity (e.g. high, moderate, uniform, or good ride-through) of a particular equipment or equipment type can be represented by assuming various types of probability density functions [155] in the sub-regions of uncertainty for one/two random variable(s), i.e., voltage threshold V and/or duration threshold T as discussed in Method 1. For example, two types of sensitive behavior— highly sensitive and moderately sensitive—are considered in this study with the help of exponential and normal probability density function, respectively, as depicted in Fig. 5.5. After calculating probability distribution functions as discussed above, the expected number of equipment trips, considering one type of equipment sensitivity at a time, can be determined as follows: Total equipment trips =

∑∑ p T

XY

(T , V ) ⋅ N (T , V )

(5-10)

V

Where p XY (T , V ) is the trip probability of the equipment as defined in (5-9) against the voltage dips with dip magnitude V and dip duration T, and N (T, V) is the number of such voltage dips expected at the specified site over specified period of time. By comparing equations (5-7) and (5-8) with equation (5-10), it is to be noticed that both formulations look almost identical, but equation (5-7) and (5-8) uses probability density function fXY (T, V), whereas equation (5-10) uses probability distribution function pXY (T, V). Similarly, N(T,V) in equation (5-7) stands for the number of expected equipment trips (with corresponding voltage tolerance curve) as a result of voltage dips at a given location (PCC), whereas in equation 10, N(T,V) represents the number of voltage dips expected (with dip magnitude V and dip duration T) at the specified site over a specified period of time. In the absence of any proper information about the type and the nature of operation of the sensitive customers connected at the selected network buses, several assumptions can be made. The buses can be ranked in the decreasing order of their total connected load and then classified into several different

70

groups, e.g., Group-I consisting of buses with high loads (>2MW); Group-II consisting of buses with medium loads (between 1MW and 2MW); and Group-III consisting of buses with loads up to 1 MW. Then, the distribution of total connected load at respective buses among different categories of customers and the corresponding cost per dip can be assumed, e.g., for Group-I buses, 70% of the total load connected can be assumed to be large customer loads that run continuous automated industrial processes like chip-manufacturing plants etc. and remaining 30% comprised of general industrial loads. Group-II buses mainly supply general industrial loads (70%), some large user load, representing steel plants, packaging plants, bottling plants, dairies, etc. (20%), and a small amount (10%) of commercial load out of which 5% represent large users (e.g., banks), who report huge financial losses due to voltage dips. For Group-III buses, it can be assumed that 50% of the total load connected is residential load whose financial losses due to voltage dips are generally small, and therefore they were not counted in the economic assessment of the total incurred costs. Further, 20% of the connected load comprised of the industrial load, whereas the remaining 30% is commercial load, out of which 5% represent large users (e.g., banks) who report huge financial losses due to voltage dips, etc. To improve further the accuracy of the economic assessment, the general trends of various customer types are also considered (see Table 5.1). The total number of process trips after comparing their sensitivities against voltage dips experienced at a specific location was therefore multiplied by a suitable correction factor to get the actual number of process trips attributed to each customer category at a given bus over a year. For example, the commercial establishments generally remain closed at least for one day in a week—either on Sundays or on Fridays and are open only for 10 hours a day, from 10 a.m. to 8 p.m. Therefore, a correction factor of [{(36552)/365}*10/24]= 0.3573 is used to get actual number of process trips affecting commercial activities (i.e., a voltage dip occurrence when the commercial establishment is closed is not going to disrupt any process). Similarly, to prevent frequent process disruption and consequential huge financial losses, large industries (like chip-manufacturing companies or financial organizations) generally install mitigation devices (e.g., UPS etc.), which provide ride-through for over 95% of the voltage dips. Therefore, only about 5% of the total voltage dips per year at such customer’s location will still be able to disrupt their processes. Finally, the upper limit of maximum one trip per day is enforced on the actual number of process trips attributed to a particular customer type (i.e. maximum number of process trips per year experienced by a customer type is 365), whereas the initial assumption was that each trip causes 24 hours of disruption of a production process. Table 5.1: Consideration of customer activities Customer type Working trend of customer Correction factor Residential Commercial N = NT* • One day off per week 0.3573 • 10 hr/day Industrial N = NT* • Two days off 0.2384 • 8 hr/day Large users Continuous process N = NT* Installed mitigation devices 0.05 correct 95% of PQ disturbances. (NT – Total number of process trips at the customer site before correction N – Actual number of process trips at the customer site before correction) After the above-mentioned corrections for the process trips and cost criteria, total voltage dip costs for the system considering processes with different sensitive equipment can be calculated.

5.3. End-Use PQ Solutions A wide range of technologies are available for the mitigation of PQ phenomena. The most common power quality mitigating technologies in use today are chemical-battery-based uninterruptible power

71

supplies (UPSs) and transient voltage surge suppression (TVSS), but a great many other types exist and are widely used today. Uninterruptible Power Supplies Uninterruptible power supplies (UPS) contain stored energy (usually in the form of conventional chemical batteries) that can provide replacement power in the event that utility power is interrupted. Some newer UPS topologies use alternative energy storage such as flywheels, ultracapacitors, or advanced chemical batteries. The annual U.S. market for small (<100 kVA) UPSs is generally estimated at about $2 billion per year, and growing. There are basically three different types of devices that are commonly referred to as UPSs: offline (standby), line-interactive, and online. Figure 5.6 provides a schematic view of these three common UPS topologies.

Figure 5.6 The three principal UPS topologies5 1. Offline or Standby UPS This UPS topology refers to a supply where, under normal circumstances, power is derived directly from the utility without conditioning. When utility power fails, a battery-powered inverter turns on and supplies power. The batteries are then charged when utility power is again available. The pricing on this type of UPS is generally less than other topologies. Disadvantages include the switchover time—the time required for the inverter to come online. This time can vary from one model of standby UPS to another. Because loads are normally connected directly to utility power, this type of UPS provides relatively poor protection from line noise, surges, frequency variations, and brownouts. 2. Line-Interactive UPS This form of UPS conditions utility power by employing a ferroresonant transformer. It is this transformer that acts as an automatic voltage regulator (AVR) standing between the AC input and AC output. The AVR maintains a constant output voltage regardless of input voltage, drawing energy from its internal battery when utility voltage is interrupted or drops too low. 5

PQTN Brief #26: UPS Mitigation of Capacitor-Switching Transients, EPRI, Palo Alto, CA: September 1995, PB-105575.

72

Besides also providing good protection against line noise, a major advantage of the line-interactive UPS is that its ferroresonant transformer can compensate for brief voltage dips, which make up the majority of voltage problems. 3. Online UPS The “traditional” UPS topology, this form of UPS conditions all utility power through a doubleconversion (AC to DC, DC to AC) rectifier/inverter process. The upside to this UPS topology is that there is no transition required when utility power is interrupted. Therefore, online UPSs can provide digitalquality power not possible with offline UPS systems, and generally provide the best isolation from other utility power problems. Disadvantages are that these are more expensive, generate more heat, and are less energy efficient than the other types of UPSs. Ferroresonant or Constant-Voltage Transformers Commercial power systems often contain electronic equipment sensitive to variations in the AC input voltage. A typical way to desensitize such equipment is to install a constant-voltage transformer—or ferroresonant transformer. Unlike conventional transformers, the ferroresonant transformer is designed such that its magnetic core becomes saturated with magnetic flux under normal operating conditions, which maintains a relatively constant output voltage during input voltage variations such as undervoltages, overvoltages, and harmonic distortion. The magnetic shunt shown in Figure 5.7 provides a path from the primary winding to the secondary winding for flux that would be lost in the core of conventional transformers. The resonating winding and capacitor shape the transformer output into a sinusoidal waveform.

Figure 5.7 Typical components of a ferroresonant or constant-voltage transformer6 Transient Voltage Surge Suppression TVSS devices are used to protect other equipment from the dangers of brief but potentially harmful voltage surges, which can be caused by lightning and the switching of inductive or capacitive devices. TVSS devices are best understood as high-voltage, high-power semiconductors. At normal voltages, they conduct little or no electrical current. As voltage climbs, they begin conducting current. For example, a surge arrester designed for a 3 kV system may conduct 1 milliamp at 5 kV and 10,000 amps at 10 kV—a 7-fold increase in current magnitude after a mere doubling of applied voltage. Non-Chemical Energy-Storage Technologies In recent years, a number of energy-storage techniques have been developed that store energy in ways other than through chemical reactions, as do lead-acid batteries. Table 5.2 lists many of these energy-

6

PQTN Brief #13: Ferro-Resonant Transformer Output Performance Under Varying Supply Conditions, EPRI, Palo Alto, CA: May 1993.

73

storage technologies, their general sizing, and ride-through time (at rated load), as well as some of their most attractive applications. Table 5.2 Capabilities of emerging PQ technologies7 Mitigation Device Power Time Large-scale motor generator 500 KVA 12 seconds sets (500 KVA) 4160/480 Low-speed flywheel - DC 250 kW 12 seconds output 500V DC Low-speed flywheel - AC output

250 kW 480 VAC

12 seconds

High-speed flywheel - DC output

200 kW 600V DC

20 seconds

Rotating voltage restorer

3 seconds

Micro turbines AC output

2000 kW 12,470/12,470 100 kW 100 kW 600 V DC 200 kW 480 VAC 35 kVA

Dynamic dip corrector AC output

30 kVA 1∅ 300 kVA 3∅

PQ 2000 battery UPS - AC output

1 MW 2 MW 4160 VAC 1 MVA 2 MVA 4160/12,470 Fraction kVA 1∅ to 1 MVA 3∅ 230 / 480V

Ultracapacitor DC output Fuel cells - AC output

Micro SMES - AC output UPS (single conversion) AC output

10 seconds 5 seconds Continuous – No short-term rating Continuous – no short-term rating 2 seconds in 1 minute 3 second first event 10 seconds 10 seconds 1 second 1 second 15 minutes battery 10 seconds flywheel

Application Service entry units can be paralleled bearing maintenance. Support for ASDs, UPSs, and computer systems and 4-wire system plant-wide support for sections of large installation. Support for ASDs, UPSs, and computer systems and 4-wire system plant-wide support for sections of bearing maintenance. Support for ASDs, UPSs, and computer systems. No bearing issues. Service entry for whole plant protection. Bearing maintenance. Support for ASDs, UPS, and computers systems. Sections of load that requires continuous power. Sub-sections of an industrial plant requiring ultimate security of supply units can be paralleled. Regular maintenance. 4-wire system critical subsections of industrial plant. Service entry for whole plant protection. Battery life in question. Service entry for whole plant protection. Single loads through to complete system. Battery life in question.

Table 5.3 Field evaluation and price of emerging technologies Mitigation Device Field Evaluation Price $/KVA Where Applicable Large scale motor Product in manufacture 350 generators (500 KVA) Low-speed flywheel Successful application at 160 DC output BG&E with UPS Low-speed flywheel Successful application 320 AC output High-speed flywheel Successful application - 200 DC output 17year life predicted Rotating voltage restorer Successful application 250 Ultracapacitor Successful application 200 7

Combined by EPRI PEAC from Myriad Internal and Industry Sources

74

DC Output Fuel cells - AC output Micro turbines - AC output Dynamic dip corrector AC output PQ 2000 Battery UPS AC output Micro SMES - AC output UPS single conversion AC output

Successful application Successful application

400

Successful application for both single and 3phase Successful application

160

Successful application

600

Many successful applications

300

400

Rotary UPS A rotary UPS uses a rotating mass—most often a heavy flywheel or other form of weighted shaft—to help stabilize and maintain rotation of an output generation, similar in concept to how a pottery wheel works. These systems tend to be large and heavy. The rotating mass often shares a shaft with a hot standby generator that is started the instant that utility power is lost. One example of a rotary UPS is the Holec/HiTec system.8 Other versions of this concept include rotating AC restorers and some forms of motor-generator sets. Flywheel Energy Storage Flywheels have been in common use for many years, but only recently were they coupled to electronics for electrical energy storage and retrieval. Low-speed flywheels (such as Active Power9) usually contain heavy steel flywheels rotating at 10,000 rpm or less when fully charged and can provide a few seconds of ride-through power at rated load. High-speed flywheels (such as Beacon Power10 or AFS Trinity11) are the emerging high-tech versions, with composite rotors operating at 50,000 to 100,000 rpm when fully charged. Superconducting Magnetic Energy Storage (SMES) In a SMES system (such as American Superconductor12), energy is stored within a cryogenically cooled superconducting magnet that is capable of releasing megawatts of power within a fraction of a cycle to replace a sudden loss in line power. In standby mode, the current continually circulates. A power supply provides a small trickle charge to replace the power lost in the non-superconducting part of the circuit. When a voltage disturbance is sensed, the controller directs real and reactive power from the inverter to the load, while automatically opening the solid-state isolation switch within two milliseconds. When the voltage across the capacitor bank reaches a pre-set level, the switch closes. This sequence repeats until normal voltage from the utility feeder is restored. Dynamic Voltage Correction or Restoration (DVC, DVR) Dynamic voltage restorers (such as S&C Purewave13 and SST DySC14) focus on augmenting voltage during voltage dips by adding missing voltage at critical times. The theory is that because end users suffer many more voltage dips than outages, having a device that mitigates only dips can do so more costeffectively than conventional technologies. Ultracapacitors

8

http://www.hitecups.com/, June 7, 2002 http://www.activepower.com/ 10 http://www.beaconpower.com 11 http://www.trinityflywheel.com/ 12 http://www.amsuper.com 13 http://www.sandc.com/products/powerquality.asp 14 http://www.softswitch.com/dysc.htm 9

75

Capacitors store energy in the form of a voltage field created by electric charge on two large and closely separated surfaces. Ultracapacitors (such as Maxwell Technologies15) employ materials that allow an extraordinary amount of stored electrical charge (and therefore energy) in a very small space. Other PQ-Enhancing Technologies There are a handful of other PQ-enhancing technologies that bear mention. Transfer Switches Power transfer switches allow an end user to switch between alternate sources of power, whether they are onsite generation or a second utility distribution feeder. These technologies are generally grouped into “slow” switches (such as ASCO16), which aren’t capable of protecting the load from interruptions, and “fast” switches (such as Mitsubishi17), which attempt to switch quickly enough so that loads are not interrupted. Extent of Use of PQ Mitigation Technologies A number of studies have asked end users what sorts of PQ mitigation technologies they have purchased. In all cases, the chemical-battery-based UPS has been the dominant player, usually followed by TVSS technologies. A recent survey conducted by EPRI PEAC Corporation in the California market bears this out (see Figure 5.8). A conventional UPS is almost universally applied in commercial and industrial settings. Surge suppression (TVSS) is the second most common, used at roughly 1/3 of all sites. Currently, the use of DG for PQ purposes is quite limited, but as costs for DG technologies drop, the potential PQ benefits of DG will likely cause them to be more popular. The rates of use in Figure 5.8 have been generally mirrored in other studies covering both broad and narrow geographic areas in the United States.

Use of PQ Mitigation Technologies 8%

Harmonic filters

12%

Generators Transfer switches

20%

Capacitors

20% 30%

Surge protection / TVSS

92%

Battery-based UPS

0%

20%

40%

60%

80%

100%

Percent of Respondents Figure 5.8 Survey results: Use of PQ mitigation technologies18 The cost-analysis approach proposed in this chapter is intended to facilitate a balancing of the costs of unmitigated PQ and reliability phenomena with the costs of preventive measures, with the goal of achieving the highest net present value (NPV) of overall benefits and costs. Although complete prevention of all business costs due to PQ and reliability phenomena would be a happy outcome for any 15

http://www.maxwell.com/ultracapacitors/index.html http://www.asco.com/home.htm 17 http://www.meppi.com/eng_analysis.html 18 EPRI PEAC Corporation, Unpublished survey, May 2002 16

76

end user, it is seldom economically feasible—only a few highly specialized industries commonly take this approach, with data centers, financial facilities, and military facilities being perhaps the best known. Facility planners would like to know how the improved reliability translates into more savings (or less expenditure). Therefore, QRA methodology incorporates the NPV technique for engineering cost analysis. There are many benefits for using an NPV approach for evaluating PQ solutions and, from the perspective of the authors, no significant detriments. Any element that results in a cost or profit, revenue in-flow or out-flow, can be considered in the NPV analysis. The resulting combination of mitigation techniques employed (including the “do nothing” option of no mitigation) can be weighed and the best alternative selected or at least identified. The key to the success of the NPV approach is that all scenarios be treated equally, with the same set of costs or benefits used for each scenario. The number and type of cost and benefit categories that can be considered for inclusion in the NPV analysis are virtually limitless. Some likely categories of costs that should be considered include: • • •

Cost in lost productivity etc. of unmitigated PQ events Capital cost of PQ solutions Maintenance costs for PQ solutions

Some likely categories of benefits that should be considered include: • • •

Reduced maintenance costs Reduced scrap and material replacement costs Reduced loss of “good will,” contract defaults, etc.

The number of configurations for possible PQ solutions is almost limitless. Table 5.4 illustrates some examples compiled by EPRI. Table 5.4 Examples of power quality problems, costs, and solutions in various industries Customer Type of Type of Solution Cost of PQ Problem Segment Equipment Power-Line Problem Disturbance Commercial – Losses (Number/Cost) Excessive Transient Installed Amusement lamps (81,255 overvoltages TVSSs on Equip 1987 1989 Park lamps on site), due to electrical Lamps 33,45 8,570 electric motors lightning panels in 1989 2$8,3 2143 (149 motors on 63 site), computer Total installed Motor 83 2 network boards cost of TVSSs: s (1,540 boards $0.14 per park Board 563 2 on site) failures operating hour s Electr 92.33 57.45 ical per per Maint hour hour enanc e Healthcare – Failure of Transient Installed Cost of Pressure Research pressure overvoltage TVSS on panel Transducer replacement: Centre transducer on powering $1,000 + Labor autoclave autoclave Commercial – Office Building Healthcare – Hospital

Failure of fax machines

Transient overvoltage

Installed pointof-use TVSS

Over $100 repair for each fax machine

X-ray tube failures

Conducted emissions internal to X-

Installed tunable filters to dampen

$80,000 for replacement cost of one X-ray tube

Source of Information Preventative Maintenance Survey: Transient Electrical Voltage and Lightning Suppression for HBJ Parks Orlando, FL

Utility, Healthcare Customer, and Autoclave Service Company Utility Utility and Healthcare Customer

77

Industrial – Chemical Plant Industrial – Cement Plant Commercial – Office Building

Equipment shutdowns Computer and programmable logic controller failures Failure of fax machine power supplies

ray equipment Power-line disturbances Voltage dips

Transient overvoltages

emissions Information not available UPSs, higherquality relays with better ride-through Installed pointof-use TVSS

Healthcare – Hospital

Control system failures, circuit board failures

Voltage dips and transient overvoltages

Customized UPS system for imaging system

Commercial – Office Building

Fluorescent lighting ballast failures

Transient overvoltages

Corrected grounding problem

Picture-Tube Manufacturer Industrial

Downtime, loss of production due to process interruption

Voltage dips and momentary interruptions

Electronics Parts for Automotive Industrial Aircraft Engine Manufacturer Industrial

Downtime, loss of production due to process interruption Downtime, loss of production due to process interruption

Voltage dips

Protecting sensitive control circuits throughout the plant MV static switch/DVR

Paper & Pulp Industrial

Loss of productivity + equipment damage + startup time Turnover losses

Voltage dips

Downtime due to computer losing track of

Wiring/Grounding inside plant

Paper & Pulp Industrial

Fish Product FactoryIndustrial

Voltage dips

Voltage dips and momentary interruptions

Loss = $1,000,000/minute Loss = $1,000,000/plant shutdown

Utility

Power Supply Repair: $127 Cost of off-site faxing: $25 Loss of time due to repair: $250 – Total: $402 per fax machine failure Loss: $10,000/hour Cost of UPS: $84,500 Cost of circuit board replacement: $20,000/board

Utility and Commercial Customer

Loss: $45.00/ballast + cost of electrician to remove and install new ballast X 80 ballasts: $8,000 $100,000/momentary interruption, $40,000/dip

Utility and Industrial Customer

Providing Fluoroscopy In The CCL During Electrical Power Interruptions, Biomedical Instrumentation & Technology, 1995 Customer – EPRI/PEAC Study Customer – EPRI/PEAC Study

$25,000 per process interruption

Customer – EPRI/PEAC Study

Dipproofing/CVT and Coil Lock circuit for sensitive controls Dipproofing/CVT for sensitive controls

$100,000 per process interruption

Customer – EPRI/PEAC Study

$492,750 for 7 process interruptions in 1996, includes $100,000 equipment damage

Customer – EPRI/PEAC Study

Large-scale reactive power compensation/U PS,CVT/dipproofing/cogeneration/islan ding Isolation transformer

Total $2,740,000 turnover losses at 8 paper and pulp mill in South Africa in one year

Customer – South African Power Quality Conference

$4000-7500 lost of production

Paper, PQ’89, page 90

78

South Africa All Customer Segment Electronics manufacturing - Industrial Semiconductor Fab processing - Industrial

batch status Impact of PQ on all South African customers Component damage + loss of production Wafer losses

Voltage dips and momentary interruptions Voltage dips and momentary interruptions

Holec ridethrough Units

Total impact estimated to be $300 million per year; largest portion related to voltage dips $50,000/year for board replacement and $50,000/year in loss of productivity Annual wafer losses $3.1 million

Detail Study conducted by ESKOM EPRI PQ Database From Customer – Silicon Systems Inc.

5.4. Choosing the Optimal PQ Solution The basic approach for application of the NPV cost analysis method for evaluation of PQ mitigation options is straightforward: 1. 2. 3. 4. 5. 6.

The annual cost of unmitigated PQ events for a facility is identified through application of the methods detailed in Chapter 4. The capital and, where appropriate, ongoing operations and maintenance costs for different PQ mitigation approaches are identified. A duration (t) for the NPV analysis is identified (the authors recommend 10 years, although more or less may be appropriate for specific industries). A discount rate (r) and inflation rate (i) to reflect the time value of money are identified, These parameters are combined in the NPV equation provided below for each mitigation scenario. The scenario producing the highest (or least negative) NPV is identified for final analysis and, one hopes, implementation at the facility.

The costs and benefits associated with the NPV approach are spread over the entire life span of the equipment used. However, there is a time value associated with money. In order to make a decision, it is important to alter the values of costs and benefits over the years to reflect at a common point in time. In almost all engineering cost analyses, the common point in time is “the present,” i.e., the planning year [3]. Present worth analysis is used to determine the present value of future money receipts and disbursements. The following equation (5-11) converts a stream of annual benefits and costs (annuities) to a present monetary value reflective of assumptions for rates for inflation, discount, and escalation: n

NPV = CI + ∑ t =0

(Ctb + Ctc ) × (1+ e)

t

(5-11)

[(1+ r)×(1+i)]

t

Where: NPV = CI = Ctb = Ctc = n= e= r= i=

Net present value of a series of cost or benefit components, Ct Initial capital investment (usually expressed as a negative amount) Benefit component occurring at the beginning of time period t Cost component occurring at beginning of time period t (usually expressed as a negative amount) Number of time periods, usually in years Escalation rate of C (normally same value as inflation) Discount rate adjusted for inflation Inflation rate

79

NPV can be compared for different PQ mitigation options. In general, the mitigation option with the highest (or least negative) NPV would be the most beneficial choice.

5.5. Conclusion This chapter presents a Net Present Value (NPV) approach as recommended for evaluating different approaches for mitigation of power quality. Although a 10-year interval for this evaluation is recommended and used in the examples provided, it will be incumbent upon each enterprise or organization applying this method to determine what time interval is most appropriate. Regardless, however, the longer the interval, the more factors other than initial capital cost (such as maintenance, etc.) will figure into the economic evaluation—an approach generally encouraged by the authors.

80

APPENDIX 1 A

Common PQ Phenomena

PQ phenomena are well documented elsewhere; the following paragraphs give a brief description of the most economically disruptive phenomena.

A.1

Categories of Power Quality Variations

The recent proliferation of electronic equipment and microprocessor-based controls has caused electric utilities to redefine PQ in terms of the quality of voltage supply rather than availability of power. In this regard, the Institute of IEEE Std. 1159-1995, Recommended Practice for Monitoring Electric Power Quality, has defined a set of terminologies and their characteristics to describe the electrical environment in terms of voltage quality. A description of the PQ categories as defined by IEEE Std.1159-1995 is provided in the following subsections. A.1.1

Voltage Dips and Swells

A voltage dip is a short-duration decrease of the RMS voltage lasting from 0.5 cycle to two minutes in duration. These events are caused by faults on the power system or by the starting current of a relatively large motor or other large load. A voltage swell may accompany a voltage dip. A voltage swell occurs when a single line-to-ground fault on the system results in a temporary voltage rise on the unfaulted phases. Removing a large load or adding a large capacitor bank can also cause voltage swells, but these events tend to cause longer-duration changes in the voltage magnitude and will usually be classified as long-duration variations. A voltage dip is a short-term reduction in voltage. It is specified in terms of duration and magnitude of retained voltage. Voltage dips are the result of increased voltage drop in the system caused by increased current flow either as the result of the addition of a large load, for example the starting current of a large motor, or due to a fault current. Those caused by large loads are generally localized, often within the same installation, while those due to network fault currents can be widely distributed and affect a large number of consumers. The characteristics of network dips—magnitude and duration—depend many factors, including the voltage level at which the fault is located, the response time of the protective devices, the degree of network meshing, the number and configuration of transformers, etc. Most fault-induced dips are caused by single-phase or two-phase (phase-to-phase) faults. Because three-phase faults are less common than single- and two-phase faults, so too are three-phase dips less likely.

Figure A-1 Classification of Voltage Variations A.1.2

Momentary Voltage Interruption

A voltage interruption is the complete loss of electric voltage. Interruptions can be short-duration (lasting less than 2 minutes) or long-duration. A disconnection of electricity causes an interruption—usually by

81

the opening of a circuit breaker, line recloser, or fuse. For example, if a tree comes into contact with an overhead electricity line, a circuit breaker will clear the fault (short circuit) and the customers who receive their power from the faulted line will lose power and experience an interruption. The causes of interruptions are generally the same as the causes of voltage dips and swells.

Figure A-2 Examples of Voltage Waveforms for Short-Duration Voltage Variations System Faults Customers located on a faulted feeder will experience one or more interruptions, depending on the type of fault and the reclosing practices of the utility. For a temporary fault, one or two reclosing operations may be required before normal power is restored. For a permanent fault, a number of reclosing operations (usually no more than three) will occur before the breaker “locks out” because of the permanent fault condition. In this case, the customers will experience a sustained interruption. Note that the interruptions associated with successive operations of the breaker may be of varying duration, depending on relay characteristics. This gives the fault multiple opportunities to clear. The multiple operations also give sectionalizers the opportunity to operate. These devices typically open during the dead time after counting a certain number of consecutive incidents of fault current within a short time period. The number of fault currents is typically two, although it could be one if the sectionalizer is at the head of an underground cable where all faults are assumed to be permanent. Reclosing practices vary from utility to utility and, perhaps, from circuit to circuit. Feeders that are mostly underground will typically not have any reclosing operations because most faults are permanent. Some utilities are experimenting with faster reclosing times (0.3 to 0.5 seconds) for the first reclosing operation to solve residential customer problems with momentary interruptions. (Residential electronic equipment such as clock radios, VCRs, microwaves, and televisions can often ride through 0.5-second interruptions but cannot ride through longer-duration interruptions.) At medium voltage levels, it usually takes a minimum of 10 to 12 cycles of dead time to ensure that the ionized gases from faults are dispersed. Customers located on parallel feeders (that is, feeders that are supplied from the same bus as the faulted feeder) will experience a voltage dip for the duration that the fault remains on the line. On mediumvoltage systems, nearly all faults are cleared within one second and can be cleared in as short as three cycles, depending on the relay settings and the magnitude of the fault current. This means that customers on parallel feeders will experience at least one voltage dip lasting from three cycles to approximately one second and possibly additional voltage dips if reclosing operations are required. Voltage dips are much less severe than interruptions, and the duration of interest is only the period of time that the fault is on the line. If there are more than two feeders supplied from a common distribution bus, voltage dips will occur more frequently than actual interruptions because a fault on any one feeder will cause voltage dips on all of the other feeders. Customers fed directly from the high-voltage system (that is, transmission-fed or large industrial customers) usually have more than one line supplying the facility, and actual interruptions should be very infrequent for these customers. However, these customers will experience voltage dips during fault conditions over a wide range of the transmission system. Voltage dips caused by faults on a high-voltage system generally have more consistent characteristics. The faults that originate in the medium- and lowvoltage systems tend to have more variation.

82

Because voltage dips can be much more frequent than actual interruptions, it is important to consider the impacts and possible remedies for voltage dips separately from the required solutions for complete interruptions. A.1.3

Overvoltages and Undervoltages

Long-duration voltage variations that are outside the normal limits (that is, too high or too low) are most often caused by unusual conditions on the power system. For example, out-of-service lines or transformers sometimes cause undervoltage conditions. These types of RMS voltage variations are normally short-term, lasting less than one or two days. Voltage variations lasting for a longer period of time are normally corrected by adjusting the voltage with a different setting of a step voltage regulating transformer tap. In addition, voltage can be reduced intentionally in response to a shortage of electric supply.

Figure A-3 Example RMS Measurement of Undervoltage During One Day19 The root cause of most voltage regulation problems is that there is too much impedance in the power system to properly supply the load. The load draws the current that gives a voltage drop across the system impedance. The resistive drop is in phase with the current, and the reactive drop is 90 degrees out of phase. Therefore, the load voltage drops low under heavy load. High voltages can come about when the source voltage is boosted to overcome the impedance drop and the load suddenly diminishes. A.1.4

Voltage Flicker

A waveform may exhibit voltage flicker if its waveform amplitude is modulated at frequencies less than 25 Hz, which the human eye can detect as a variation in the lamp intensity of a standard bulb. Voltage flicker is caused by an arcing condition on the power system. The arcing condition may be a normal part of a production process (for example, a resistance welder or an electric arc furnace. Voltage step changes greater than 3%, usually caused by the starting of large motors, may also cause complaints of light flicker. However, these events are not a frequency modulation of the voltage amplitude. Flicker problems can be corrected with the installation of filters, static VAR systems, or distribution static compensators.

19

Assessing the Impact of Power Quality on California Industries, EPRI, Palo Alto, CA: 2002. Year.

83

Figure A-4 Example Voltage Waveform Showing Flicker Created by an Arc Furnace20 Voltage Fluctuations The primary cause of voltage fluctuations is the time variability of the reactive power component of fluctuating loads such as, for example, arc furnaces, rolling mill drives, main winders, etc. In general, these loads have a high rate of change of power with respect to the short-circuit capacity at the point of connection to the supply. The magnitude of the fluctuations is usually such that the supply voltage remains within the permitted voltage tolerance band, but the cyclic nature of the variation, combined with the characteristics of lamps and the response of the human eye and brain, lead to a sensation of flicker. Flicker is an impression of unsteadiness of illumination that can cause loss of concentration, headaches, and, in some cases, epileptic fits. A.1.5

Harmonic Distortion

Harmonic distortion is the presence of frequencies at integer multiples of the fundamental system frequency, which is 50 Hz in Europe and 60 Hz for the North American power system. Electronic loads and saturable devices generate harmonic distortion. In commercial facilities, computers, lighting, and electronic office equipment generate harmonic distortion. In industrial facilities, adjustable-speed drives and other power electronic loads can generate significant amounts of harmonics. It is generally safe to assume that the sine wave voltage generated in central power stations is very good. In most areas, the voltage found on transmission systems typically has much less than 1% percent distortion. However, the distortion may reach 5 to 8% as we move closer to the load. At some loads, the current waveforms will barely resemble a sine wave. Electronic power converters can chop the current into a variety of waveforms. Most distortion is periodic, or harmonic. That is, it is nearly the same cycle after cycle, changing very slowly, if at all. This has given rise to the widespread use of the term “harmonics” to describe perturbations in the waveform. This term must be carefully qualified to make sense.

Figure A-5 Example Voltage Waveform with 3rd Harmonics and 17.42% Total Harmonic Distortion21 20

Assessing the Impact of Power Quality on California Industries, EPRI, Palo Alto, CA: 2002. Year. Assessing the Impact of Power Quality on California Industries, EPRI, Palo Alto, CA: 2002. Year.

21

84

Solutions to problems caused by harmonic distortion include installing active or passive filters at the load or bus, or taking advantage of transformer connections that enable cancellation of zero-sequence components. Harmonic frequencies are integral multiples of the fundamental supply frequency. Current harmonics are generated within the system and in consumer loads by the nonlinear behavior of magnetic materials, rectifiers, and electronic converters. Although harmonic frequencies have always been present in the electricity system, the increase in the number of equipment using electronic power control in recent years has lead to increased levels. Voltage harmonics are created by harmonic currents as they flow through system impedances. Nonlinear loads draw harmonic currents from the supply, thereby producing a corresponding harmonic voltage drop in the impedance of the supply network. As a result, all consumers see harmonic voltage distortion on the supply voltage. Standards have been introduced to limit the emission of harmonic current by individual items of equipment and by installations in an attempt to limit the overall level of harmonic distortion on supply networks. Fortunately, the design of networks tends to mitigate some of the effects of harmonic load current—delta transformer windings sink the third and ninth harmonic currents emitted by single-phase loads, for example. Electronic converters also introduce other frequencies into the supply, known as interharmonics. So far, the magnitude of interharmonic voltages is small. A.1.6

Unbalance

A three-phase supply system is said to be balanced if the three-phase voltages and currents have the same amplitude and are separated by 120 degrees with respect to each other. If either of these conditions is not met, the system is said to be unbalanced. Unbalanced supply voltage usually arises because of unequal loading of the phases at the low voltage level, where most of the loads are single phase. Other causes are the asymmetry of the distribution system and the connection of large non-three-phase loads, such as railway connections and arc furnaces. A.1.7

Voltage Notching

Voltage notching is caused by the commutation of power electronic rectifiers. It is an effect that can raise PQ issues in any facility where solid-state rectifiers (for example, variable-speed drives) are used. The effect is caused by the switching action of the drive’s input rectifier. When the drive DC link current is commutated from one rectifier thyristor to the next, a line-to-line short circuit occurs at the input terminals to the rectifier. With this disturbance, any given phase voltage waveform will typically contain four notches per cycle caused by a six-pulse electronic rectifier.

Figure A-6 Example Waveform with Notching22 A.1.8

Transient Disturbances

Transient disturbances are caused by the injection of energy by switching or by lightning. The disturbance may either be unidirectional or oscillatory. Lightning, electrostatic discharge, load switching, or capacitor switching may cause a unidirectional transient. It is characterized by its peak value and rise time. An 22

Assessing the Impact of Power Quality on California Industries, EPRI, Palo Alto, CA: 2002. Year.

85

oscillatory transient is characterized by its frequency content and may be caused by a switching operation such as the energizing of a capacitor bank, distribution line or cable, or interruption of current to an inductive load. The switching of a load may cause high-frequency oscillations with principal frequencies greater than 2 kHz. Common solutions to problems caused by transients include the application of surge arresters, passive and active filters, and isolation transformers.

Figure A-7 Example Oscillatory Transient Waveform Caused by Energizing a Capacitor23

Figure A-8 Example Impulsive Transient Waveform24

23 24

Assessing the Impact of Power Quality on California Industries, EPRI, Palo Alto, CA: 2002. Year. Assessing the Impact of Power Quality on California Industries, EPRI, Palo Alto, CA: 2002. Year.

86

Table A-1 Categories of Power Quality Variation—Institute of Electrical and Electronics Engineers (IEEE) 1159-1995 1.0

2.0

3.0 4.0

5.0 6.0 7.0 8.0

Categories Transients 1.1 Impulsive 1.1.1 Voltage 1.1.2 Current 1.2 Oscillatory 1.2.1 Low Frequency 1.2.2 Medium Frequency 1.2.3 High Frequency Short-Duration Variations 2.1 Sags 2.1.1 Instantaneous 2.1.2 Momentary 2.1.3 Temporary 2.2 Swells 2.1.1 Instantaneous 2.1.2 Momentary 2.1.3 Temporary Long-Duration Variations 3.1 Overvoltages 3.2 Undervoltages Interruptions 4.1 Momentary 4.2 Temporary 4.3 Long-Term Waveform Distortion 5.2 Voltage 5.3 Current Waveform Notching Flicker Noise

Spectral Content

Typical Duration

> 5 kHz > 5 kHz

< 200 µs < 200 µs

< 500 kHz 300–2 kHz > 2 kHz

< 30 cycles < 3 cycles < 0.5 cycle

Typical Magnitudes

0.5–30 cycles 0.1–1.0 pu 30–120 cycles 0.1–1.0 pu 2 sec–2 min 0.1–1.0 pu 0.5–30 cycles 0.1–1.8 pu 30–120 cycles 0.1–1.8 pu 2 sec–2 min 0.1–1.8 pu

0–100th Harmonic 0–100th Harmonic 0–200 kHz < 30 Hz 0–200 kHz

> 2 min > 2 min

0.1–1.2 pu 0.8–1.0 pu

< 2 sec 2 sec–2 min > 2 min

0 0 0

steady-state steady-state steady-state intermittent intermittent

0–20% 0–100% 0.1–7%

Transient disturbances are high-frequency events with durations much less than one cycle of the supply. Causes include switching or lightning strikes on the network and switching of reactive loads on the consumer’s site or nearby sites. Transients can have magnitudes of several thousand volts and so can cause serious damage to both the installation and the equipment connected to it. Non-damaging transients can cause severe disruption due to data corruption.

B

Response of Sensitive Equipment to PQ Events

B.1

Data Processing and Communications Equipment

This type of equipment operates internally from a low voltage DC supply derived from the AC supply by a rectifier and electronic converter. It is insensitive to moderate levels of harmonic distortion and can be made immune to most transients, but it is sensitive to voltage dips. When the supply voltage drops during a dip, the amount of energy delivered to the load is reduced. The ability of the equipment to “ride through” the dip depends on the amount of stored energy available from the internal power supply capacitor and the instantaneous energy requirement of the device. A personal computer (PC) will have a better ride-through capability while processing than it would have while writing to an optical drive, for example. IT equipment dip performance is described by curves such as the Computer and Business Equipment Manufacturers Association (CBEMA) curve and its more modern Information Technology Industry Council (ITIC) replacement. These curves show the safe operational envelope of the equipment on a nominal voltage/time plot. If the duration and retained voltage during a dip lie above the boundary, the equipment is likely to continue to operate normally.

87

% nominal voltage

200 180 160 140 120 100 80

Im m unity area

60

Dip

40 20 0 0,0001

Interruption

0,001

0,01

0,1

1

10

100 seconds

Figure B.1 ITIC Curve In reality, these curves simply specify typical equipment performance; they do not imply that the equipment will survive the dips that typically occur on the network.

B.2

Variable-Speed Drives

Variable-speed drives (VSDs) use an electronic converter to produce a variable-frequency motor drive voltage from the fixed supply frequency. Using VSDs is much more energy-efficient than using belts and gearboxes to change speed or using throttle valves to control fluid flows. They are used extensively in industrial processing, materials handling, and building management. During a dip, the amount of energy supplied by the electrical system is reduced and may be below that required by the process, resulting in loss of control. Because motor-controlled processes rarely operate in isolation, this can result in loss of synchronization with other parts of the process and uncoordinated shut down. VSDs usually include a number of measures in order to protect the electronics and the motor from abnormal conditions, such as undervoltage or loss of a phase voltage, that may trigger shutdown in the event of a dip. VSDs draw harmonic currents from the supply. Many drives are designed to minimize or eliminate these currents. VSDs are not affected by normal levels of harmonic distortion.

B.3

Lighting

Any change in supply voltage magnitude may cause a change in the luminous flux or spectral distribution of a light source. Incandescent light sources are particularly sensitive, as the luminous flux is approximately proportional to the cube of the applied voltage. They are susceptible to “flicker,” which is a subjective visual impression of unsteadiness of a light’s flux, when its luminance or spectral distribution fluctuates with time. The human eye-brain response to variation of luminous flux produces fatigue and loss of concentration for relatively small variations in light intensity at frequencies of about 2 to 20 Hz.

88

Gas discharge lighting is less sensitive to traditional flicker, partly because it is often electronically controlled, but is affected by a flicker effect due to jitter caused by voltage variation due to interharmonic voltage distortion. Gas discharge lighting is sensitive to dips: if a dip is deep enough to extinguish the discharge, a hot lamp may not re-strike when the voltage returns to normal.

B.4

Solenoid-Operated Contactors

Solenoid-operated contactors and relays are used in large numbers in process control systems, and they are particularly sensitive to voltage dips. “Hardened” devices are available but are relatively rarely used.

C

Additional Losses Caused by Poor PQ

Additional losses may be the result of harmonic currents. Load-generated harmonic currents flow in installations and in the distribution system; they do not transfer energy but do cause additional loss in cables, transformers, and in motors.

C.1

Cables

In the presence of harmonic currents, the RMS current is higher than that required to energize the load because harmonic currents do not transfer energy. This has to be taken into account in sizing conductors. Zero-sequence harmonics, i.e., those with a harmonic number that is a multiple of three, do not cancel in the neutral of a three-phase supply. This is important in three-phase cables, which provide supplies to single-phase nonlinear equipment where the combined neutral current can exceed the phase currents.

89

C.2

Transformers

Transformers are affected by harmonic currents. Part of the load loss of a transformer is due to eddy current losses in the windings; usually around 5 to 8% of the loss is due to eddy currents and the remainder due to conductor resistance. Eddy current losses are proportional to the square of frequency, so harmonic currents have a serious effect on the heat generated within the transformer, leading to higher operating temperatures and significant reduction in transformer lifetime.

C.3

Motors

Directly connected motors (i.e., those without a VSD) are affected by harmonic voltage, due to the presence of zero-, positive-, and negative-sequence harmonics. The result is excess heating, increased mechanical stress, and reduced lifetime.

90

APPENDIX 2 A

Overview of Interruption Cost Calculation

The evaluation of customer outage costs (COC) for a particular service area utilizes three models [36] System model, which describes the performance of the power system serving the area, Load model, which describes the load connected in the service area at various load points, and Cost model, which represents the costs due to supply interruptions as a function of interruption duration for a particular customer mix. The method of Failure Modes and Effect Analysis (FMEA) [70] is normally used for preparing the system model. A system model is obtained in terms of a failure rate, average outage duration, and annual outage time at each load point in the system. Table A-1 illustrates the system model parameters of for two buses in a general test system. TABLE A-1: SYSTEM MODEL Index

Bus “n”

Bus “k”

0.48

0.46

r (hours)

0.95

0.99

U (hours/ yr)

0.46

0.46

λ (faults/yr)

The load model used can be in the form of either average load or actual time-dependant load curves at each load point. This is used to obtain the annual energy consumed and peak demands for a customer or sector at a given load point. A typical load model is as shown in Table A-2 [42, 71].

Sector y Residential Commercial Industrial Total

TABLE A-2: LOAD MODEL E Pmax (MWh) (MW) 8700 2.43 14600 3.97 9800 2.00 33100 8.40

Load Factor 0.409 0.420 0.559 0.450

The preparation of the cost model requires the customer survey of the service area to collect perceived costs of interruptions of various durations for the customer mix supplied. The cost model is then prepared for each load point in the area as the normalized costs due to supply interruptions as a function of interruption duration. A typical cost model is shown in Table A-3 [42, 71]. TABLE A-3: COST MODEL C(ri) (£/MWh) Duration (ri) Mom. 0.50 1 min. 0.52 20 min. 1.34 1h 2.84 4h 9.32 8h 17.2 24h 22.91 Using the above models, the COC at a load point j supplying ny sectors can be calculated as follows:

 ny  COC j =  ∑ E jy  × C j (rj ) × λ j  y =1 

(in £)

(A-1)

91

Where Ejy is the annual energy consumed by sector y. A summation of the COC at all the relevant load points b of a sector gives the annual COC due to supply interruptions SCOC as follows: b

SCOC = ∑ COC j

(in £)

(A-2)

j∈b

B

Probabilistic Voltage Dip Costs Calculation

The evaluation of the impact of voltage dips at a particular site in the network involves three basic steps: fault-analysis, voltage dip analysis, and economic analysis. In fault analysis, the method of fault-positions [100] is often used in which various types of faults (symmetrical and asymmetrical) are simulated at numerous locations throughout the system network, and corresponding expected voltage magnitudes and durations are determined at various network buses. In a subsequent voltage dip analysis performed at a point of common coupling (PCC), the frequency of dips of specified dip magnitude and duration over a period of interest is determined by associating it with the historical fault performance (fault per km per year) of all network buses, overhead lines, and underground cables. This information is generally available from historic data obtained through long-term monitoring at respective locations in the network. The corresponding duration of voltage dips depends on fault-clearing times of protective devices used in the power system network. The final and the most crucial step for the economic assessment of PQ requires the information about the consequences of expected voltage dips on the performance of industrial processes. This procedure requires preparing a dip-performance chart for a particular bus in the system and coordination of the customer equipment responses with these voltage dips on a single graphic display [72]. The information about the equipment sensitivity may be gathered from the equipment manufacturer or by testing. Sensitive industrial equipment are classified into various equipment categories based on equipment-types, and then the testing is performed on a suitable number of equipment picked randomly from each category. However, even though the equipment may belong to the same equipment category, it might not exhibit the same sensitivity to voltage dips [73]. This makes it difficult to develop a single standard that defines the sensitivity of process equipment. In addition to this, it is also possible that a process may be disrupted due to tripping of individual equipment or it may require the tripping of a group of equipment depending upon their interconnections. The only way to deal with those uncertainties is to apply probabilistic calculations relying on expert advice and limited number of field/laboratory tests related to equipment/process sensitivity to voltage dips. The main emphasis of this example, therefore, is to illustrate a probabilistic approach for quantification of process trips incorporating the uncertainty involved with equipment sensitivity and consequently with the process sensitivity.

C

Overview of Equipment Sensitivity

Equipment sensitivity to voltage dips is usually expressed only in terms of the magnitude and duration of the voltage dip. For this purpose, the rectangular voltage-tolerance curve (as shown in Fig. C-1) is used. It indicates that voltage dips deeper than the specified voltage magnitude threshold (Vmin) and longer than the specified duration threshold (Tmax) will cause malfunction (or trip) of the equipment. However, in practice, some equipment like motor contactors and household electronics items has non-rectangular voltage-tolerance characteristics [74-77]. Other two parameters, which may be detrimental to sensitivity of some of the industrial equipment such as motor contactors, are point-on-wave of dip initiation and phase-shift during the dip [75-78].

92

Fig C-1. Equipment Voltage-Tolerance Curve Equipment is therefore classified into various categories based on device type, and testing of an adequate number of devices representing one equipment category justifies generalization of the acquired results. Because different brands of the same equipment type and even different models of the same brand often have different sensitivities, a typical sensitivity data with appropriate statistical deviation and error parameters can be determined for the equipment type. The sensitivity information so obtained needs to be updated continuously as and when more test results become available.

D

Uncertainty Involved with Equipment Sensitivity

It is found that all the equipment belonging to a particular equipment category do not exhibit same sensitivity against voltage dips [74-79]. However, all equipment types except motor contactors exhibit, more or less, perfect rectangular characteristics. Voltage magnitude-threshold and duration-threshold of three equipment types, namely PLCs, ASDs, and PCs, may vary between Vmin and Vmax and between Tmin and Tmax, respectively. The values of these parameters (obtained in tests) are different for different categories of equipment. The Vmin and Vmax are 30% and 90%, respectively, for PLCs [78, 80], 46% and 63%, respectively, for PCs [79], and 59% and 71%, respectively, for ASDs [86], and corresponding Tmin and Tmax are 20 ms and 400 ms for PLCs, 40 ms and 205 ms for PCs, and 15 ms and 175 ms for ASDs, respectively. Therefore, the voltage-tolerance curves of these equipment may occur anywhere inside the shaded region on Voltage dip magnitude v/s duration chart shown in Fig. D-1, such that the knee point of curve always remains inside sub-region C.

Fig. D-1. The region of uncertainty for sensitivity curves of PCs, PLCs, and ASDs

Similarly, the area of uncertainty related to the AC contactors’ sensitivity can be represented by the shaded region shown in Fig. D-2. Their voltage-tolerance curves may appear anywhere in the shaded region acquiring non-rectangular form for 00 point-on-wave of dip initiation and rectangular form for 900 point-on-wave of dip initiation [ 87-89].

93

Fig D-2. Probable regions for voltage-tolerance curve of contactors

E

Counting of Process Trips

The quantification of expected process trips due to voltage dips over a specified period of time requires the knowledge about the mutual connection of sensitive equipment that control the process. Sometimes, tripping of a single equipment may disrupt a complete industrial process, i.e., all the participating equipment are assumed to be connected in series. On the other hand, the process might be disrupted only when more than one equipment mal-function/trip, i.e., parallel connection of participating equipment. The overall probability of process trip can be determined by knowing the probability of trip of individual equipment and their mutual connections. For example, consider a simple process consisting of four sensitive devices having mutual connections as shown in Fig. E-1.

Fig. E-1 Typical connections of sensitive equipment participating in a process The overall probability of process trip is given by

Ptrip = 1 − (1 − p1 ) ⋅ (1 − p 2 p3 ) ⋅ (1 − p 4 )

(E-1)

Where pi, (i = 1, 2, 3, 4) is the cumulative probability of tripping of ith device. In general, the probability of a process trip can be written as n  m   Ptrip = 1 − ∏ 1 − ∏ pi , j  j =1   i =1 

(E-2)

Where m is the number of series-connected equipment/equipment groups and n is the number of parallelconnected equipment in ith equipment group. pi , j is the cumulative probability of tripping of jth equipment of the ith serially connected equipment group.

94

Fig. E-2. Expected behavior of sensitive equipment against voltage dips of different characteristics A piece of equipment may have any voltage-tolerance characteristic inside permissible range. This paves way for the stochastic assessment of likelihood of equipment having a particular sensitivity inside the permissible range at the time of occurrence of a voltage dip and consequential impact on the equipment operation. Consider six different voltage dips, namely A1, A2, B, C1, C2, and C3 on the voltage-dip chart as depicted in Fig. E-2. It is obvious from the figure that voltage dips A1 and A2 will not cause any malfunction or trip of the equipment, and therefore the probability of equipment trip is zero. Similarly, voltage dip B will certainly cause the tripping of the equipment and hence the probability of the equipment trip is unity. However, the behavior of the equipment for voltage dips C1-C3 will depend on the actual sensitivity characteristics of the equipment at the time of these voltage dips. It implies that there is a certain probability of equipment either surviving these voltage dips or tripping when exposed to them. The variation in equipment sensitivity can be represented in terms of uni-variate random variable (T) in sub-region A, uni-variate random variable (V) in sub-region B, and bivariate random variable (T, V) in sub-region C (see Fig. D-1), where T and V are assumed to be two statistically independent discrete random variables. T is the voltage duration-threshold varying between Tmin and Tmax (determined by the protection settings) and V is the voltage magnitude-threshold varying between Vmin and Vmax. Therefore if pX(T) and pY(V) are the probability distribution functions for random variable T and V respectively, and then the joint probability distribution function for the bivariate random variable (T, V) in sub-region C is given by Bayes rule [81] as follows:

pXY (T, V) = pX(T) pY(V);

(E-3)

Tmin ≤ T ≤ Tmax , V min ≤ V ≤ V max The general trend of sensitivity (e.g. high, moderate, uniform, or good ride-through) of a particular equipment or equipment type can be represented by assuming various types of probability density functions [82] in the sub-regions of uncertainty for one/two random variable(s), i.e., voltage threshold V and/or duration threshold T.

95

Fig. E-3. Representation of contactors’ sensitivity with the combinations of uniform and/or normal probability distributions in various sub-regions of uncertainty

After calculating probability distribution functions as discussed above, the expected number of process trips, considering one type of equipment sensitivity at a time, can be determined as follows: Total process trips =

∑∑ P

trip

T

(T , V ) ⋅ N (T ,V )

(E-4)

V

Where Ptrip (T , V ) is the trip probability of the process as defined in (E-2) against the voltage dips with dip magnitude V and dip duration T, and N (T, V) is the number of such voltage dips expected at the specified site over specified period of time.

F

Cost Assessment

As evident from equation (E-4), total number of trips of a given industrial process is dependent on the location of the process in the system network and the overall sensitivity of the process against voltage dips. For economic assessment of financial losses due to voltages dips, it is pre-requisite to have the information about the type of industrial/commercial process, customer type, mitigation devices installed and the associated damage cost per dip. The total costs incurred due to voltage dips and interruptions should be added together in order to come up with total network financial losses for a given network topology. The implementation aspects of the Bayesian methodology for the assessment of financial losses due to voltage dips described above are given in the sequel.

G

Numerical Example

The results presented here are based on calculations performed on a generic distribution system (GDS) comprising four 275-kV transmission infeeds, 132-kV and 33-kV sub-transmission networks (predominantly meshed), and 11 kV distribution network (predominantly radial) [83, 99]. The GDS consists of 295 buses, 296 overhead lines, and underground cables and a large number of switches and circuit breakers in order to alter the network topology for preventive control and better reliability of the system. A large number of transformers having different (Yd, Yy, etc.) winding connections (generally present in the UK distribution networks) are also modeled. The base case topology consists of 18 switches in their open position. Additional 40 topologies were derived more or less arbitrarily from the base case topology by opening/closing of some of the open switches for the purpose of PQ-cost comparison. Some of the actions taken to generate different network topologies are illustrated in Table G-1. For all these topologies, both interruption and voltage dip costs are

96

determined. Different network designs/topologies are considered in order to compare and minimize the total financial losses in the system. TABLE G-I: VARIOUS NETWORK TOPOLOGIES CONSIDERED Topology Switching action performed 1 Base case 2 All switches closed 3 Line 75-231 closed . . 40 Lines 3-74, 51-52, 85-87, 75-231, 79-223, 123-129 closed 41 Lines 60-64, 61-62, 61-55, 65-55, 66-67, 3-74, 179-26, 25-26, 27-28, 44-222, 36-37, 10-12, 51-52, 85-87, 123-129, 215-225 closed; Lines 249235, 250-236 open. The input data about the customer interruption costs for different categories of customers is adopted from [42] and shown in Table G-2. For the interruption cost calculation as previously described, the MonteCarlo simulation is used. The total system costs due to supply interruptions experienced by customers over a period of one year for the various topologies considered are shown in Table A-VI in the decreasing order.

Customer Type Res. Com. Ind. L. user

Rank 1 2 3 …

Mom. 11.47 1.2k 216k

TABLE G-2: CUSTOMER INTERRUPTION COSTS (£) VALUES CIC(£) for an interruption duration of : 1 min. 20 min. 1h 4h 8h 0.19 0.70 4.78 11.74 49.12 106 345 719 1.5k 2.9k 4.3k 7.6k 12.0k 216k 219k 233k 329k 413k

TABLE G-3: TOTAL INTERRUPTION COSTS FOR DIFFERENT TOPOLOGIES Cost Topology Cost Rank (M£) No. (M£) 685.40 2 39 316.17 649.67 38 40 311.87 630.75 37 41 309.01

24h 1.0k 16.3k 581k Topology No. 6 3 1

On the basis of the interruption costs only, it is clear from Table G-3 that topology 2 results in huge financial losses, whereas the topology 1 is the most economic one. All other topologies result in the interruption costs in between these two extremes. For the voltage-dip assessment, the method of fault positions is used considering fault positions on network buses and transmission lines (one fault position per bus and six fault positions per line) [83]. Voltage-dip magnitudes at the network buses are calculated for symmetrical and asymmetrical faults at these fault positions. A conservative approach is adopted for counting process trips due to voltage dips; i.e., the lowest of all phase voltages is considered as the dip magnitude, and it was assumed that the sensitive equipment is connected to the phase experiencing the lowest dip magnitude. The duration of voltage dips is determined by the protection fault-clearing time assuming that all faults are cleared by the primary protection (i.e., 100% reliable primary protection system). The adopted fault rates and durations are given in [84, 85]. Ten network buses are selected arbitrarily as the buses of interest at which sensitive industrial processes are running. Out of these ten buses, the first eight are 11-kV buses, whereas the last two are 33-kV buses. For the stochastic assessment of process trips taking into account the voltage-dip performance at the site and the sensitivity of individual equipment participating in the industrial process, six different generic process configurations, as shown in Fig. G-1, comprising of series/parallel connections of four pieces of commonly used industrial equipment—PLCs, ASDs, PCs, and AC contactors—are considered. From

97

these six basic process configurations, 37 distinct processes are derived with two additional assumptions: 1) four sensitive devices participating in a process belong to the same equipment-type; 2) all four sensitive devices participating in a process belong to different equipment types. The tripping probabilities for these individual devices are as shown [15].

Fig. G-1: Six typical configurations considered for industrial processes Because these four sensitive devices are connected in series/parallel combinations, the overall sensitivity of a particular process depends on the equipment type(s), expected range of variation of the sensitivity of individual equipment type, and their mutual connections. In order to get a realistic cost assessment (in the absence of exact information about process distribution among network buses), it is decided to allocate any ten randomly selected processes (out of 37 available) among ten network buses of interest such that one process is connected to each bus. (Note: This assumption allows for the same process to be connected to more than one bus at the same time.) To achieve this, Monte Carlo simulation is used. About 10,000 trials were made for the random allocation of processes among ten network buses to get the maximum and minimum range for the system-wide nuisance process trips, once assuming high sensitivity of participating equipment and then assuming moderate sensitivity. After the random allocation of processes, their nuisance trips are determined and used for the evaluation of financial losses, assuming different categories of customers connected at respective network buses. For the economic assessment of customer losses due to voltage dips, it is assumed (conservatively again) that every nuisance trip of an industrial process requires 24 hours of restoration time. The damage costs reported by various categories of customers for 24 hours long interruption [42] are taken as the damage costs for process trips due to voltage dips. These costs for different categories of customers are shown in Table G-7. In the absence of the proper information about the type and the nature of operation of the sensitive customers connected at the selected network buses, several assumptions are made in order to account for those in the most realistic way. The selected ten buses of the network are ranked in the decreasing order of their total connected load and then classified into three different groups: Group-I consisting of buses with high loads (>2MW); Group-II consisting of buses with medium loads (between approximately 1MW and 2MW); and Group-III consisting of buses with loads up to approximately 1 MW as shown in Table G-8. Then, the distribution of the total connected load at respective buses among different categories of customers and the corresponding costs per dip are assumed as shown in Table G-9. For Group-I buses, 70% of the total load connected is assumed to be large customer loads, which runs continuous automated industrial processes like chip manufacturing plants, glass manufacturing, etc., and remaining 30% comprised of general industrial load. Group-II buses mainly supply general industrial load (70%), some large user load, e.g., packaging plants, bottling plants, dairies (20%), and a small amount (10%) of commercial load out of which 5% represents users (e.g., banks, data centers) who report huge financial losses due to voltage dips (see Table G-9). For Group-III buses it was assumed that 50% of the total load connected is residential load whose financial losses due to voltage dips are generally small and therefore they were not counted in the economic assessment of the total incurred costs. Further, 20% of the connected load comprised of the industrial load whereas the remaining 30% is commercial load out of which 5% represents users who report huge financial losses (as above) due to voltage dips (see Table G9). To improve further the accuracy of the economic assessment, the general working trends of various customer types also are considered (see Table G-10). The total number of process trips after comparing

98

their sensitivities against voltage dips experienced at a specific location was therefore multiplied by a suitable correction factor to get the actual number of process trips attributed to each customer category at a given bus over a year. For example, the commercial establishments generally remain closed at least for one day in a week—either on Sunday or on Friday—and are open only for 10 hours a day—from 10 a.m. to 8 p.m. Therefore, a correction factor of {(365-52)/365}*10/24 = 0.3573 is used to get the actual number of process trips affecting a commercial facility (i.e., a voltage dip occurrence when the commercial facility is closed is not going to disrupt any process). Similarly, to prevent frequent process disruption and consequential huge financial losses, large industries (like chip-manufacturing companies or financial organizations) generally install mitigation devices (e.g., UPS, etc.), which provide ridethrough for over 95% of the voltage dips.

TABLE G-7. ASSUMED COSTS PER VOLTAGE DIP [42] Type of Dip Cost/event (£) assuming one Customer Load day-long interruption of production Residential Commercial 1,000 Industrial 16,300 Large User 581,000 TABLE G-8. CATEGORIZATION OF NETWORK BUSES Peak Rank Bus Group Load no. No. 1 247 51.19 I 2 243 34.79 3 89 1.44 4 66 1.15 5 137 1.15 II 6 40 1.05 7 34 0.87 8 76 0.87 9 111 0.15 III 10 16 0.04 TABLE G-9. DISTRIBUTION OF COSTS AND LOAD TYPES AMONG CUSTOMER CATEGORIES Group Type of load Load (%) Cost/dip (£) Large User 70 581 k I Industrial 30 16.3 k Large User 20 581 k Industrial 70 16.3 k II 9.5 (95%) 1k Commercial 0.5 (5%) 581 k Residential 50 0 Industrial 20 16.3 k III 28.5 (95%) 1k Commercial 1.5 (5%) 581 k TABLE G-10. CONSIDERATION OF CUSTOMER ACTIVITIES Customer type Residential Commercial Industrial Large Users

Working trend of customer • One day off per week • 10 hr/day • Two days off • 8 hr/day • Continuous process • Installed mitigation devices correct 95% of PQ disturbances

Correction factor N = NT* 0.3573 N = NT* 0.2384 N = NT* 0.05

(NT – TOTAL NUMBER OF PROCESS TRIPS AT THE CUSTOMER SITE BEFORE CORRECTION N – ACTUAL NUMBER OF PROCESS TRIPS AT THE CUSTOMER SITE BEFORE CORRECTION)

99

Therefore, only about 5% of the total voltage dips per year at such a customer’s location will still be able to disrupt their processes. Finally, the upper limit of the maximum one trip per day is enforced on the actual number of process trips attributed to a particular customer type (i.e., the maximum number of process trips per year experienced by a customer type is 365), because the initial assumption was that each trip causes a 24-hour disruption of a production process. After the above-mentioned corrections for the process trips and cost criteria, total voltage dip costs for the system considering a processes with highly sensitive equipment and then with moderately sensitive equipment are calculated. The variation in voltage-dip costs for the system obtained in first 12 trials with highly sensitive and moderately sensitive equipment is shown in Fig. G-3. The variation in the voltage dip costs for one selected topology (topology 20) is shown in Fig. G-4. It can be seen that the variation in costs due to voltage dips can be very high (between £0.14M and approx. £61M in case of moderately sensitive equipment) depending on the allocation of processes to different system buses. The figure also shows that the sensitivity of the equipment involved in process can increase the costs up to 50% (e.g. total costs with moderately sensitive equipment for trial 1 are about £61M and with highly sensitive equipment about £88M). Finally, the interruption costs and voltage dip costs for 10 different network topologies are added together, and the results are shown in Fig. G-3. It can be seen that the voltage dip costs, when added to the interruption costs, may alter the total financial losses in the system and in some cases alter the ranking of the topologies based initially on interruption costs only. (e.g., topology 20 with dip costs included becomes “more expensive” than topology 39 for about £3.3M per year). The numerical results are illustrated in Fig. G-5. By comparing the total network losses due to voltage dips with those of interruptions it was found that voltage dip costs account for up to about 23% of the total network losses due to interruptions. 60

HS MS

50

Cost (M£/yr.)

40

30

20

10

0 0

1

2

3

4

5

6

7

8

9

10

11

Trial

Fig G-3. Variation in dip costs due to process trips for the whole system (HS – highly sensitive equipment; MS – medium-sensitive equipment)

100

Cost (M£) 100 90

HS MS

80 70 60 50 40 30 20 10 0 1

5

9

13

17 21

25

29 33

37 41

45

49 53

57 61

65

69 73

77

81 85

89

93

97

Trial

Fig. G-4. Variation in total voltage dip cost for network topology The example illustrated the methodology for comprehensive assessment of financial losses incurred to individual customers and the whole network over a specified period of time due to two main PQ disturbances: interruptions and voltage dips. The study performed considers modeling of uncertainties involved with the equipment and process sensitivity and their influence on the quantification of process trips due to voltage dips. In the absence of the accurate data related to equipment and process sensitivity and corresponding trip/failure costs, which is a common and wide-spread occurrence in this type of study, a probabilistic approach is applied that leads to an estimate of the expected number (range) of process trips and consequential financial losses. The estimated range of financial losses due to voltage dips compared to the losses due to outages is in agreement with the reported results based on field surveys. The example further compares total financial losses in the network incurred by interruptions and voltage dips for various network topologies. It is observed that the costs to individual customers and the whole network due to voltage dips alone could be quite substantial (depending on the equipment and process sensitivity) and therefore could have great influence on the total PQ costs.

900

800

Int. cost Int. + Min. sag cost Int. + Max sag cost

778.04 736.77

700

666.55

Cost (M£/Yr.)

600

500

475.9

472.56

430.6

413.75

398.84

400

365.1 329.69

300

200

100

Topology

Fig. G-5. Comparison of interruption and voltage dip costs for the system: Influence of dip costs on total financial losses for the system with various topologies considered

101

1

6

7

4

10

27

14

18

13

23

19

24

30

26

36

28

31

32

35

37

2

0

I

Typical Loss Values Table I-1 Average cost of power interruption. Adopted from [18]. Average cost Cost index K IEEE 1159 class (US$/kW) Ck 1 Instantaneous 0.078 1 2 Momentary 0.176 2 3 Temporary 1.22 15 4 Sustained 3.63 47

Table I-2 CIC (£) values. Adopted from [42]. CIC (£) for an interruption of duration

Sector Residential Commercial Industrial Large user

Momentary 11.47 1.2k 216k

1 min 11.47 1.5k 216k

20 min 0.19 49.12 2.9k 219k

1 hour 0.7 106 4.3k 233k

4 hour 4.78 345 7.6k 329k

8 hour 719 12.0k 413k

24 hour 1.0k 16.3k 581k

Table I-3 Estimated costs for industrial sectors. Adopted from [14]. Voltage Dip Cost (% of total yearly power cost) Industrial Process Category A Category B Category C Semiconductor 0 to 2 2 to 10 5 to 6 Pharmaceutical 0 to 0.8 1 to 5 2 to 4 Chemical 0 to 1 1 to 3 2 to 4 Petrochemical 0 to 1 2 to 5 1.5 to 3.5 Manufacturing 0 to 0.2 0 to 1 0.8 to 1 Metallurgy 0 to 0.2 0 to 1.5 1 to 1.5 Food 0 to 0.5 0 to 1.5 0 to 2 Table I-4 Direct cost per event per kW. Politecnico di Milano. Adopted from [29]. [€/kW-event] All sectors Per NACE codes DA – Food products DB – Textiles DE – Paper DF – Refined petroleum products DG – Chemicals and man-made fibers DH – Plastic products DI – Glass and ceramic products DJ – Metals products DL – Electrical equipment DM – Auto and auto components

Entire sample (sub-sample) Median 0.8 (1.1)

Mean 2.8 (3.3)

Interval 0 (0.1) - 30

0.6 3.2 0.8 (0.9) 13.3 0.6 (0.7) 1.8 0.8 1.1 (4.9) 9.3 2.9

5.9 3.2 0.9 (1.0) 13.3 0.5 (0.7) 2.2 0.9 3.3 (4.9) 10.6 2.9

0.2 – 30 3.2 0.1 – 2.2 13.3 0 (0.6) – 0.8 0.1 – 4.2 0.1 – 2.3 0 (1.1) – 8.7 0.1 – 22.4 0.7 – 5.0

Table I-5 Financial losses due to voltage dips. Adopted from [44]. Industry Typical financial loss per event (€) Semiconductor production 3,800,000 Financial trading 6,000,000 per hour Computer centre 750,000 Telecommunications 30,000 per minute

102

Steel works Glass industry

350,000 250,000

Table I-6 Financial losses of large commercial and industrial customer for various disturbances. Adopted from [45]. Scenario

Financial Losses ($)

4 hour outage without notice 1 hour outage without notice 1 hour outage with notice Voltage dip Momentary outage

74,835 39,459 22,973 7,694 11,027

Table I-7 Impact of voltage dip on industry. Adopted from [49]. Industry

Loss per voltage dip ($)

Paper manufacturing Chemical industry (plastic, glass, etc.) Automobile industry Equipment manufacturing Credit card processing Semiconductor industry

30,000 50,000 75,000 100,000 250,000 2.5 million

Table I-8 Summary of all outage cost studies. Adopted from [50]. Cost per interrupted kW Cost per event Study Average cost per hour or kWh Population Research $61,949 for large industrial Systems and commercial All regions - $59,983 Northwest - $28,609 Southwest - $51,908 Southeast - $86,477 West - $52,734 Midwest - $28,735 ASCO Cellular – $41k Telephone – $72k Airline reservation – $90k EDF $0.67/kW $8/kWh up to 30MWh $17.4/kWh from 30 to 50 MWh ESOURCE $583k over 800 commercial and industrial customer over 1 year IEEE 493-1997 Industrial - $6.43/kW + $9.11/kWh Commercial – $21.77/kWh CEIDS EPRI $7795 for digital establishments $14,746 for continuous process manufacturing Primen Mass Survey $21,688 for 19 businesses surveyed ICF Consulting 80 to 100 times the cost of retail electricity Table I-9 Comparison of interruption costs of industrial customers (in year 2000 US$/kW). Adopted from [51].

103

Study/Duration 2 second Canada 1.07 (small industrial) England 14.49 (industrial) USA (industrial) Nepal (industrial Greece 2.10 (industrial) Taiwan 37.03 (high-tech)

1 min

20 min

1 hour

2 hour

4 hour

8 hour

24 hour

2.55

3.65

7.71

13.68

28.13

52.06

82.87

15.24

33.62

59.5

-

170.1

283

354.3

-

-

9.64

-

-

-

-

0.11

0.23

0.42

0.58

1.50

3.00

10.99

2.55

7.35

12

16.75

21.80

-

46.86

55.15

60.90

87.6

118.1

167.1

242.4

425.2

Table I-10 Voltage-dip sensitivity factors for different industries. Adopted from [32]. Category Dip sensitive factor Semiconductor (SC) 1 Computer and peripherals (CP) 0.4 Telecommunications (TC), and 0.4 Optoelectronics (OE) 0.6 Precision machinery (PM) 0 Biotechnology (BT) 0 Table I-11 Industries surveyed. Adopted from [52]. Industry Number of samples Ratio (%) Food and beverages 49 7.4 Textile and apparel 55 8.3 Pulp and paper products 36 5.8 Chemical and products 127 19.2 Basic/fabricated metal 52 7.9 Other machinery and 49 7.4 equipment Electric and electronic 82 12.4 equipment Electric machinery 53 8.0 Audio visual equipment 48 7.3 Motor vehicles 51 7.7 Other transport equipment 56 8.5 Table I-12 Interruption cost by duration (unit: Won). Adopted from [52]. Interruption cost per average kW ($/kW) Industry type Below 3 seconds Below 1 minutes Below 5 minutes Below 30 minutes Food and beverages 22.783 44.747 78.020 128.504 Textile and apparel 8.421 8.724 9.500 13.935 Pulp and paper 1.660 1.678 1.781 2.100 products Chemical and 39.805 50.284 52.042 61.505 products Basic/Fabricated 12.886 18.706 33.359 63.288 metal Other machinery 11.594 15.950 26.605 59.443 and equipment Electric and electronic 80.335 120.718 174.493 230.076 equipment Electric machinery 7.700 13.634 21.470 45.794 Audio visual 9.547 12.709 23.045 53.517 equipment

104

Motor vehicles Other transport equipment

23.699

36.683

49.706

83.612

9.316

12.862

15.782

39.420

Table I-13 Expected losses due to voltage disturbance. Adopted from [53]. Losses due to voltage disturbance Industry ($/kVA per event) Semiconductors 80 - 120 Glass 10 - 15 Automotive 6 - 10 Plastics 4-7 Textile 3-8 Table I-14 Cost per event of interruption Industry

Cost per Event of Interruption

Paper industry $10,000 - $30,000 Textile industry $10,000 - $40,000 Data processing $10,000 - $40,000 Plastic industry $10,000 - $50,000 Semiconductor industry $10,000 - $50,000 Automotive manufacturing $15000 Source: EPRI – PQ Applications Guide for Architects and Engineers Table I-15 Average cost of outages. Adopted from [37]. Average Industry cost of downtime ($/hour) Mobile communications 41,000 Telephone ticket sales 72,000 Airline reservation 90,000 Credit card operations 2,580,000 Brokerage operations 6,480,000 Source: U.S. Department of Energy’s Strategic Plan for Distributed Energy Resources (2000) Table I-16 Estimated voltage dip costs. Adopted from [37]. Industry UK steel work US glass plant US computer centre US car plant South Africa

Duration

Cost/dip

30% for 3.5 cycles Less than 1 second 2 second Annual exposure Annual exposure

£250k $200k $600k $10M $3B

105

Fig. I-1 Annual costs due to power quality disturbances for the industrial sector in EU-25 [40]

Fig. I-2 Annual costs due to power quality disturbances for the services sector in EU-25 [40]

106

Fig. I-3 Annual costs due to voltage dips for five Finnish distribution companies [8]

Fig. I-4 Voltage dip-related cost in different industries. Adopted from [43].

Fig. I-5 Normalized cost per dip as a function of plant power. Adopted from [10]

107

Fig. I-6 Industry-specific costs of PQ. Adopted from [48].

Fig. I-7 Customer damage functions for different high-tech industry categories. Adopted from [32].

J

Typical Financial Loss Values - Summary Section Manufacturing

TABLE J-1 DIRECT COST PER KW PER EVENT Financial Division Activities Currency Loss General

Disturbance Type

Small Industrial (Canada)

2.55

US$

1-minute power interruption

Industrial (England)

15.24

US$

1-minute power interruption

Industrial (Nepal)

0.11

US$

1-minute power interruption

Industrial (Greece)

2.55

US$

1-minute power interruption

108

Food products and beverages (17, 18)

Textiles (20)

Paper and paper products (24)

High-tech industry (Taiwan)

55.15

US$

Food products (Italy)

5.9

Euro

Food

8

US$

Food and Beverages (South Korea)

44.75

US$

Textiles (Italy)

3.2

Euro

Textiles

11.7

US$

Textiles (South Korea)

8.72

US$

Paper (Italy)

0.9

Euro

Paper

1.7

US$

Paper (South Korea)

1.67

US$

13.3

Euro

0.5

Euro

Chemical

20.6

US$

General cost of power quality

Chemical and petrochemical (South Korea)

50.28

US$

1-minute power interruption

Plastic products (Italy)

2.2

Euro

Very short interruptions and voltage dips

Plastic products

3

US$

General cost of power quality

Glass and ceramic products (Italy)

0.9

Euro

Very short interruptions and voltage dips

Coke and refined Refined petroleum petroleum products products (Italy) (26) Chemicals and man-made fibers (Italy) Chemical and chemical products (27)

Rubber and plastic products (29)

Non-metallic mineral products (30)

1-minute power interruption Very short interruptions and voltage dips General cost of power quality 1-minute power interruption Very short interruptions and voltage dips General cost of power quality 1-minute power interruption Very short interruption and voltage dips General cost of power quality 1-minute power interruption Very short interruptions and voltage dips Very short interruptions and voltage dips

109

Basic/fabricated metals (31, 32)

Computer, electronic and optical products (33)

Electrical equipment (34)

Machinery and equipment (35) Motor vehicles, trailers and semitrailers (36) Other transport equipment (37) Transport and storage

Glass products

8

US$

General cost of power quality

Primary metal

15.5

US$

General cost of power quality

Basic/ fabricated Metal (South Korea)

18.71

US$

1 minute power interruption

Metal products (Italy)

3.3

Euro

Very short interruptions and voltage dips

Electronic

58.3

US$

General cost of power quality

12.71

US$

1-minute power interruption

120.72

US$

1-minute power interruption

13.63

US$

Electrical equipment (Italy)

10.6

Euro

Other Machinery and Equipment (South Korea)

15.95

US$

Auto and auto components (Italy)

2.9

Euro

36.68

US$

12.86

US$

1-minute power interruption

10

US$

General cost of power quality

Audio and Visual Equipment (South Korea) Electrical and Electronic Equipment (South Korea) Electric Machinery (South Korea)

Motor Vehicles (South Korea) Other Transport Equipment (South Korea)

Transportation (55All transportation 57)

1-minute power interruption Very short interruptions and voltage dips 1-minute power interruption Very short interruptions and voltage dips 1-minute power interruption

Information and communication

Communications (62-66, 86)

Communications

28.6

US$

General cost of power quality

Financial and insurance activities

Financial service activities (87)

Business services

3.7

US$

General cost of power quality

TABLE J-2 DIRECT COST PER KVA PER EVENT Section

Division (NACE code)

Activities

Financial Loss

Currency

110

Textiles (20)

Textile

3-8

US$

Rubber and plastic products (29)

Plastics

4-7

US$

Non-metallic mineral products (30)

Glass

10 - 15

US$

80 - 120

US$

6 - 10

US$

Manufacturing Computer, electronic and optical products Semiconductors (336) Motor vehicles, trailers and semi-trailers (36)

Automotive

TABLE J-3 DIRECT COST PER EVENT Section

Division

General Textiles (20) Paper and paper products (24) Chemical and chemical products (27)

Activities

Financial Loss Currency

Large User (UK)

216,000

£

Large industrial and commercial (US)

7694

US$

Industrial (UK)

1200

£

Voltage dip 1-minute power interruption

US$ US$

Process interruption Voltage dip

Textile Industry 10,000-40,000 Paper manufacturing (US) 30,000 Paper industry

10000 - 30000

US$

Process interruption

Chemical industry (US)

50,000

US$

Voltage dip

US$ Euro US$ Euro US$ Euro

Process interruption Voltage dip Voltage dip Voltage dip Voltage dip Voltage dip

Rubber and plastic products (29)

Plastic Industry 10,000-50,000 250,000 Non-metallic mineral Glass industry (Europe) products (30) Glass plant (US) 200,000 Steel works (Europe) 350,000 Basic metals (31) Steel works (UK) 250,000 Semiconductor (Europe) 3,800,000

Computer, electronic and optical products (33) Machinery and equipment (35)

Manufacturing

Motor vehicles, trailers and semitrailers (36)

Wholesale and retail trade

(51-53)

Disturbance type 1-minute power interruption

Semiconductor (US, Europe and Far East)

2,500,000

US$

Voltage dip

Semiconductor

10,000-50,000

US$

Process interruption

Equipment manufacturing (US) Automobile industry (US)

100,000 75,000

US$ US$

Voltage dip Voltage dip

Automotive

15,000

US$

Process interruption

Commercial (UK)

11.7

£

1-minute power

111

interruption Telecommunications (64)

Telecommunications (Europe) Computer centre (Europe) US computer centre (US)

30,000 750,000 600,000

Euro Euro US$

Voltage dip Voltage dip Voltage dip

Data processing

10,000-40,000

US$

Process interruption

Credit card processing (US)

250,000

US$

Voltage dip

Information and Information service communication activities (66) Activities auxiliary to Financial and financial services and insurance insurance activities activities (88)

TABLE J-4 ANNUAL COST Section

Division (NACE code)

Activities

Financial Loss

General

Manufacturing

0-1

Food products and beverages (17, 18)

Food

0-2

Currency % of total yearly power cost % of total yearly power cost

Coke and refined petroleum products Petrochemical (26)

0-5

% of total yearly power cost

Chemical and chemical products (27)

Chemical

0-4

% of total yearly power cost

Basic pharmaceutical products and pharmaceutical preparations (28)

Pharmaceutical

0-5

% of total yearly power cost

Basic metals (31)

Metallurgy

0 - 1.5

% of total yearly power cost

Computer, electronic and optical products (33)

Semiconductor

0 - 10

% of total yearly power cost

Motor vehicles, trailers and semitrailers (36)

U.S. car plant

10,000,000

US$

Manufacturing

Other

General

South Africa 3,000,000,000 total

US$

TABLE J-5 COST PER HOUR OF INTERRUPTION Section

Financial and insurance activities

Information and communication Wholesale and retail trade

Division (NACE code)

Activities auxiliary to financial services and insurance activities (88)

Telecommunications (64)

Activities Brokerage operations Credit card operations Financial trading (Europe) Mobile communications

Retail trade, except of motor Airline reservation vehicles and motorcycles Telephone ticket (53) sales

Financial Loss

Currency

6,480,000

US$

2,580,000

US$

6,000,000

Euro

41,000

US$

90,000

US$

72,000

US$

112

K Formulae for Computing Harmonic Losses for the Main Electrical Components The harmonic losses PT for the transformers (joule and core losses) can be computed as [57-59] :

PT = 3

∑ (I )

h max

h 2

h = h1

V h R + P ∑  1 h = h1 V h T

1 co

h max

  

mT

1 , h 2.6

(K-1)

Where: Ih RTh Vh

= = = = =

1

Pco

mT

current harmonic of order h equivalent resistance of the transformer at the harmonic of order h voltage harmonic of order h core losses at the fundamental frequency numerical coefficient

The harmonic losses PM for the induction motors (joule and core losses) can be computed as [5759]:

PM

2

h max

h max V h Vh  1 = 3 ∑  h  R hM + Pco ∑  1   h = h1  Z M  h = h1 V

   

mM

1 h 0.6

(K-2)

Where: h ZM

= equivalent impedance of the motor at the harmonic of order h = equivalent resistance of the motor at the harmonic of order h = numerical coefficient

h RM

mM

Harmonic losses PC for the condensers can be computed as [57-59]:

PC = 3 ω C

h max

( )

2

∑ h V h tgδh

h = h1

(K-3)

Where:

ω

= angular frequency of system at the fundamental C = capacitance of the condenser tgδ h = loss factor at the harmonic of order h

The harmonic losses PCa of three-conductors cables (joule and dielectric losses) can be computed as [5759]:

PCa = 3

h max

( )

2

hmax

( )

h + 3ω CCa ∑ h tgδh V h ∑ Ih R Ca

h = h1

h = h1

2

(K-4)

Where: h RCa =

alternating current resistance of one conductor of the cable

113

CCa = capacitance per core; = angular frequency of system at the fundamental.

ω

Other formulas are proposed for a precise evaluation of the equipment loss of life. It is necessary to consider the operating condition instead of the nominal one. The operating temperature rise of the hottest point ( ∆TO ) at operating condition is evaluated by the following formula.

PO

∆TO =

PN

⋅ ∆TN

(K-5)

Where :

∆TO ∆TN PO PN

= Expecting temperature rise of the hottest point under operating condition = Temperature rise of the hottest point under nominal operating condition = Operating power = Nominal power

This formula considers that the equipment is at operating condition since enough time to reach the equilibrium temperature. The temperature of the hottest point is given by adding the ambient temperature ( TA ) to the temperature rise.

TO = ∆TO + TA

(K-6)

And

TN = ∆TN + TA

(K-7)

The evaluation of the ambient temperature could be problematic. We should consider the cooling system used. The equipment is inside or outside? Is it in a temperature control environment? The expecting life under operating power condition ( tO ) could be evaluated with Arrhenius function [58, 59, 166] knowing the nominal temperature the hottest point and the lifespan of the equipment from the manufacturer.

tO =t N ⋅e

 ∆T  E   −    K   TN (TN + ∆T ) 

(K-8)

Where :

t N = Expecting life span under nominal condition tO = Expecting life span under operating condition TN = Expecting temperature of the hottest point under nominal condition (ºK) ∆T =TO − TN (most of time negative) The formula presented in chapter 2.2.2.2 will be modified in the following form:

114

Ph

∆Th ≈

PO

th =tO ⋅e

⋅ ∆TO

 E −  K

 ∆Th      TO (TO + ∆Th ) 

(K-9)

(K-10)

Where:

∆Th = Ph = th =

temperature rise of the hottest point cause by harmonics content harmonics content Expecting life span under polluted harmonics condition

There could be more loss of life at operating condition than at rated condition because the life expectancy tO will be greater than t N . Utilities could loss more money in reduction of equipment useful life in harmonics condition when their equipments are operating below their nominal rating. To do the economic we should find the actual cost for the replacement, in the future, of the equipment. The time used will be modulated by the lifespan of the equipment. This implies the use of the present value formula as presented in chapter 2: (C tb − C tc ) × (1 + e) t t =0 [(1 + r )(1 + i )]t n

PV = ∑

(K-11)

This is modified in the following form: PVE =

C E × (1 + e) tO [(1 + r )(1 + i )]tO

(K-12)

Where: PVE CE tO

= = =

Present value of future equipment replaced in sinusoidal condition Actual cost for replacing the equipment Expecting life span of the actual equipment

This formula gives the actual cost for a future The same formula is used to calculate the present cost for buying new equipment in t h years representing the expected lifespan under harmonics condition. PVEh =

C E × (1 + e) th [(1 + r )(1 + i )]th

(K-13)

Where: PVEh = present cost for buying new equipment in t h years representing the expected lifespan under harmonic condition t h = Expecting life span of the actual equipment under harmonics condition

The extra cost for the lifespan reduction due to harmonics will be given by the following formula: C ELR = PVE − PVEh

(K-14)

This procedure is generic and could be apply for the cost of lifespan reduction for any kind of perturbation.

115

A global cost evaluation for loss life for distribution system equipment could be perform by ordering equipment in categories representing the kind of equipment and the nominal power in order to reduce the amount of calculation. Time of the day and date of the year could be used to evaluate different load level (operating condition) and fluctuating ambient temperature.

L

Methods for Probabilistic Evaluations

The first step in a probabilistic approach is to recognize that output economical figures to be computed are statistical quantities. In the most general cases, their probability density functions (PDFs) completely describe their statistical features. However, for the sake of estimating the economical value of losses and premature aging due to harmonics, it is adequate referring to the total expected value as:

E ( D) = E ( Dw) + E ( Da)

(L-1)

where symbol E(.) indicates the expected value of the quantities already introduced. When estimating expected values for a period of time, it is needed to consider their present worth values as:

E ( D) pw = E ( Dw) pw + E ( Da) pw

(L-2)

The present worth expected economical value of losses due to harmonics losses, E (Dw ) the whole electrical system life of NT years, is: NT

NT

n =1

n −1 n = 1 (1 + α )

E ( Dw) pw = ∑ E (Dw)npw = ∑

pw

, referred to

E ( Dw) n

(L-3)

pw Where E (Dw)n is the present worth expected value of the harmonic losses in the nth year, and

E ( Dw) n is computed by summing the economical value of harmonic losses of each component in each jth combination characterized by mj components operating in the same time period ∆Tj: mj

E (Dw) j = ∑ E (Dw)k , j .

(L-4)

k =1

For the gncombinations taking place in year n, it is: gn

gn m j

j =1

j =1 k =1

E (Dw)n = ∑ E( Dw ) j = ∑ ∑ E( Dw )k , j

(L-5)

It is clear from relation (L-5) that it is necessary to compute the expected value of harmonic losses for each component of the system, that is E( Dw )k , j . Considering each single electrical component

,G h 2 ,..,G h max h max , E( Dw )k , j is computed as :

continuously subject to an hmax harmonics of voltage or current harmonic G characterized in the time interval ∞ ∞

∞

∆T

by the joint pdf f h1 G ,..,G

h1

E (Dw)k , j = ∫0 ∫0 ..∫0 Dwk , j ( G h1 ,.., G h max ) f h1 dG h1 ..dG h max G ,..,G h max

(L-6)

with

116

(

)

Dwk , j = Dwk , j G h1 ,..,G h max = Kw Pk , j ( G h1 ,..,G h max )∆T

(L-7)

For the most common components of industrial energy systems, the harmonic losses Pk , j ( G h1 ,..,G h max ) in (L-7) can be obtained by summing up the losses due to each harmonic so that the integral in (L-6) can be strongly simplified as: h max ∞

E (Dw)k , j = ∑ ∫0 Dw(G h ) f h dG h G

(L-8)

h = h1

In spite of the apparent complexity of models from (L-3) to (L-8), it is necessary to evidence that the methods practically require the estimation of losses due to harmonics for each component of the system, paying attention to preliminarily ascertain definite states of operating conditions. h The computation of losses, Dw(G ) in (L-8), does not present particular difficulties; several studies in literature addressed this subject for the most common components and equipment like transformers, cable line, capacitors, and so on [89-92]; also the formulas shown in Appendix 2-L are valid. Main difficulties can arise for deriving in each state the PDFs of voltage and current harmonics. For existing systems, this can be obtained both from measurements and from simulations adopting well-stated probabilistic methods of harmonic analysis [92-98]. pw , is evaluated by summing the The present worth economic value of premature aging in (L-2), E ( Da ) present worth expected value of the aging costs of each of the N components of the system: N

E( Da ) pw = ∑ E( Da )kpw

(L-9)

k =1

pw Where the value of E( Da )k is calculated starting from the knowledge of the useful lives of the

various components by the relation:

E(Da)kpw = E(Cns )kpw - E(Cs )pw k

(L-10)

pw pw where E (C s ) k and E (Cns ) k are the present worth expected value of the costs for buying the

component during the system life in sinusoidal and non-sinusoidal operating conditions, respectively. The actualization of the costs can be effected in a similar way considering both the discount rate and the cost variation for buying the component; the expected value of cost to be met for buying each component at year n in a sinusoidal and non-sinusoidal regime is linked to the expected value of the component life in these conditions, respectively. To estimate these figures, again the cumulative damage theory can be applied, as in the case of deterministic methods. In such a case, we have to refer to the expected value of relative loss of life in the study period; E[∆RL] computes as: ∞ ∞

E [ ∆RL ] = Tc ∫ ... ∫ 0

f x1 x 2 ..x n

n

∏ dxi

0 L( x1 , x2 ,.., xn ) 1

(L-11)

117

where f x x ..x is the joint PDF of the n random variables on which the component life L depends. The 1 2 n successive estimation of the useful life can be carried out, as previously mentioned, by summing the expected values of the relative losses of life until reaching the unity. The main critical point of this method is linked to the complexity of computing the N dimensional integral of (L-11) and, overall, to assign the joint PDF f x x ..x . Indeed, some simplifications 1 2

n

introduced by life models of actual insulated components can greatly help. First of all, in most cases it is adequate to consider electrothermal stress models. Moreover, it is demonstrated that they can be reduced to an even simpler model like: −n p

L = L0 ' K p

exp (-B cθ )

(L-12)

Where L0' is life at nominal sinusoidal voltage and reference temperature; cθ =1/θ0 - 1/θ is the so-called conventional thermal stress, θ is absolute temperature, θ0 is a reference temperature (generally the room temperature); np and C are model parameters. In particular, np is the coefficient related to the effect of the peak of the distorted voltage waveform on life (thus, the larger this coefficient, the stronger the influence of peak voltage). Using model (L-12), the general equation (L-11) becomes:

E [ ∆RL ] = T c

∫∫

DT D K p

f K pθ L( K p ,θ )

dK p dθ

(L-13)

Where f K pθ is the joint PDF of the peak factor Kp and of the equipment temperature θ, defined in the time interval Tc, DK and Dθ are the variation domains of Kp and θ, respectively, and L(Kp, θ) represents p

the equipment life model expressed by (L-12). Equation (L-13) still can present some difficulties in deriving the joint PDF of the random variables Kp and θ. This joint PDF is generally not directly available. Even in the case in which the statistical characterization of the variables is known, the computation of (L-13) is not immediate, mainly due to the fact that Kp cannot be expressed in closed form as a function of voltage harmonics and fundamental component (infinite different combinations of harmonic vectors can provide a given value of Kp). Then, the application of (L-13) in real cases requires the use of Monte Carlo simulation procedures. Some simplifications can be pursued only in particular cases: 1. As an example, having the aim to highlight only the influence of voltage and current harmonics on component life, the voltage and current at fundamental frequency, the ambient temperature and the elements of the system admittance matrices at the fundamental, and at harmonic frequencies can be assumed deterministic quantities. Under this assumption, the voltage harmonics are directly linked to the current harmonics injected by nonlinear loads via the elements of the system harmonic admittance matrix. In such a case, the expected value E[∆RL] of all the MV/LV power system components are a function of only the PDF of the magnitude and phase of the current harmonics injected by nonlinear loads, thus reducing the number of random variables to be accounted for. 2. Further simplifications can be achieved computing the life reduction in the worst condition, i.e., that occurring when the peak voltage is the arithmetic sum of the voltage harmonic peaks. Applying this simplification, there is no need to know the PDF of the phase of harmonic currents injected by the nonlinear loads; moreover, in the presence of only one group of nonlinear loads as the main cause of harmonic pollution, the useful life can be evaluated also with closed form relations, with the simplified procedure proposed in [54].

118

In conclusion, to estimate the economical damage due to harmonic losses for each component of the system in the study, the following procedure can be observed. A) Evaluate the expected value of the operating costs due to harmonic losses E[Dw] as follows: i. Let n1,...,NT be the years of the system study. Let year n1 be assigned to the annual count NN. ii. Let h1,..., H be the harmonics present in the NNth year. Let harmonic h1 be assigned to the harmonic count NH. iii. Evaluate the expected value of the operating cost due to the actual NHth harmonic, [Dw(GNH)]NN, computing each integral in (L-8) for known fGNH . iv. Update the harmonic count NH. If NN > NT, go to step v; otherwise go to step iii. v. Sum the integrals obtained in step iii to estimate the operating costs due to harmonic losses E[Dw]NN in NNth year. vi. Update the annual count NN. If NN > NT, go to step vii; otherwise go to step ii. vii. Calculate the expected value of the total operating costs, E[Dw], summing the actualized values of each considered year E[Dw]NN. A flow-chart of this procedure is shown in Fig. L-1.

119

Start n1,...,NT h1,..., H [Dw(GNH)]N

NH= NH+1

NH > H

no

yes E[Dw]NN

NN= NN+1

NN > NT

no

yes E[Dw]

Stop Fig. L-1. Flow-chart of the procedure to evaluate the expected value of the operating costs due to harmonic losses B) Evaluate the expected value of the aging costs due to harmonic losses E[Da], as follows: i.

Evaluate the expected value of the thermal loss of life E[∆RL] in sinusoidal operating conditions by the integral in (L-13). ii. Sum the E[∆RL] coming in succession until their sum reaches the unity, so establishing the i-th year in which the component must be substituted. iii. Evaluate the purchase cost of the component at the i-th year, taking into account the cost variation to buy it.

120

iv. Evaluate the expected value of the aging costs in sinusoidal operating condition, E[Cs], summing the purchase costs obtained in step iii, taking into account the present worth discount rate. v. Repeat steps i. to iv. in non sinusoidal operating conditions to evaluate E[Cns]. vi. Calculate the difference between E[Cns] and E[Cs].

A flow-chart of this procedure is shown in Fig.L-2.

121

Start E[∆ RL]

ΣE[∆R L]

ΣE[∆R L]=1

no

yes purchase cost of the component

E[C s]

E[∆R L]ns ΣE[∆R L]ns

ΣE[∆ RL]ns=1

no

yes purchase cost of the component

E[Cns] E[Cns]-E[C s]

Stop Fig. L-2. Flow-chart of the procedure to evaluate the expected value of the aging costs due to harmonic losses

122

APPENDIX 3 A

Cost Aspects

The economic effect of voltage dips and short supply interruptions may differ depending on the time of occurrence related to a specific process [120]. It may be without any difference to a control system whether a loss of supply is lasting for 100 ms or for one or more hours; depending on the kind of manufacturing process and its vulnerability, regarding the before-mentioned consequences and/or service costs for re-establishing a related manufacturing process, one case of data loss may result in efforts of up to several thousands of Euros. Depending on the branch, voltage dips or supply interruptions result into costs in a range from 10.000,-- € (paper, plastic, glass manufacturing) to 700.000,-- € (semiconductor manufacturing) per event [121]. Mathematical Model [106] Mathematic modeling of costs of loss caused by supply interruptions and voltage dips can be done by considering these costs consisting of two components: a fix component, that one independent from the duration of voltage loss/reduction; another component proportionate to the duration of voltage loss/reduction [109], what leads to the related specific costs kA affecting the customers:

k A = kW ⋅ t + k P

(A-1)

Where

kA

kW

specific costs of voltage loss/reduction

[€/kW]

energy-specific, constant-cost component, independent from the duration

[€/kWh]

t

duration of supply interruption/voltage dip [h]

kP

power-specific costs for voltage dips

[€/kW]

Some figures on mean cost components Based on study results in different European countries, mean values for related costs/kW have been calculated [106], resulting into the following mean cost components dependent on the duration of supply interruption/voltage dip: Table A-1 Mean cost components Duration “0” min Household Costs Agriculture 0,006 (mean values) Commerce 1,59 [€/kW] Industry 2,6

1 min 0,03 1,92 3,5

15 min 0,11 0,16 3,99 8,7

1 h 2,28 16,8 13 17

4 h 8,2 62 46,5 49,7

8 h 25,4 96,2 80,1 80,3

Applying these values to a diagram and, based on linear interpolation, adding trend lines results into the following functions for the mean values for costs of supply interruptions:

123

€/kW

Agriculture

y = 12 ,51 x + 1,53

Industry

y = 9,72 x + 5,49

Commercial

y = 9,97 x + 2,47

Household

y = 2,94 x

10 0

80

60

40

20

0 0

1

2

3

4

5

6

7

8

Duration [h]

Fig. A-1 - Diagram of mean cost components While the gradient of the trend lines is characteristic for the specific costs of interruptions, the starting values at “0” h give the specific costs for voltage dips (and short interruptions). From the results, the following Table A-2 shows the specific voltage dip costs: Table A-2 Voltage dip costs

Household Agriculture Commerce Industry

Specific voltage dips costs kp €/kW 0,29 0,35 (1,53) 2,47 5,49

Some other results from research in other countries are as follows. A Norwegian research project [108, 112] was undertaken during 2002, aiming besides others to evaluate customers’ costs associated with voltage dips and short supply interruptions. The results showed the following: Approximate numbers of occurrence per delivery point: Short supply interruptions, distribution network

13

Voltage dips, distribution network

13

Voltage dips, regional network

63

Overall customers’ costs: Associated with voltage dips of 23 – 44 M€/year, considering only business customers and voltage dips with voltage levels reduced by 50% for a maximum of 1 s. Associated with short supply interruptions of 80 M€/year, considering only business customers.

124

The average sum of both being in the order of magnitude of the overall customers’ costs associated with long supply interruptions: (> 3 min) of ~ 113 M€/year Average specific interruption costs: • •

For voltage dips: € 0,67/kW For short supply interruptions: € 0,93/kW

Investigations by EdF [113] resulted in specific costs of short supply interruptions < 1 min of 0,76 €/kW, while for such shorter supply interruption the Norwegian study resulted in specific costs of 0,85 – 1,25 €/kW [121]. The following example describes the economic effect of losses of supply in the case of a manufacturer in textile (threads): Consequences: • • • • • • •

Increased piling up of unsalable bobbins, due to malfunction of the manufacturing process. Need for 8 hours time until the entire manufacturing system is regularly working again. Costs for production loss: € 6.300,--, to be completed by staff costs resulting from staff not being able to do any work (manufacturing process, telephone, PCs), in the considered case to be calculated with € 2.000,--/h. Other considerations on remedial measures and cost aspects [114]. On principle, remedial measures can be taken on different supply network voltage levels, within the customer’s installation as well as on equipment design stage (considering modifications of the supply network (HV, MV) design). On an LV level, within the customer’s installation to reduce the effects of voltage dips or short supply interruptions, targeted at particularly important devices or processes. At design and construction of the (more or less) susceptible device

Voltage dips generally imply a solution that provides some means of supporting voltage, while interruptions usually require a source of energy to replace the lacking one from the electricity supply network. Economic considerations play an important part in balancing the cost of the remedial measures with the gravity of the possible disturbance arising from voltage dips or short supply interruptions. In HV-/MV-network oriented, voltage dips and short supply interruptions are mainly caused by events in the MV networks, the related percentage being reported by more than 80%. On the network side, the most effective method for reduction of ARCs appears to be increased realization of MV lines as buried cables. According to some scientific investigations, 100% cabling would reduce the number of dips by 67%, but due to longer durations of loss of supply, the end costs would be reduced only by 1%. Huge financial efforts for reaching comprehensively cabled MV networks would be facing a quite modest success related to the supply interruption costs. Another option is given by dividing given networks. Dividing a network into two halves results in a reduction of voltage dips by 50% each. This measure is followed by a decrease of redundancy, a freedom of switching of network parts, as well as a freedom of power plant use and therefore of security of supply [113]. Based on the measure of dividing a given network, a calculation example may highlight the cost relation for statistically avoiding one voltage dip. The example is based on an MV network with a length of 4.539 km, this one supplying 616.000 customers. Measurements conducted in the 30 substations of this network show a mean value of 21,2 voltage dips per substation and year to be expected and a statistical occurrence of 0,14 disturbances per km. Two-busbar operation enables a reduction of occurring voltage dips to half, i.e., 10,1/substation, year. For enabling this kind of operation, costs of around 15,5 M€ are to be afforded, for 30 Peterson coils and 8 transformers. By this measure, over the entire MV network, statistically 10,1 x 30 = 318 voltage dips per year are avoided.

125

Starting from this point and considering the overall effort of 15,5 M€, the costs for avoiding one voltage dip/year for every customer in the considered MV network would statistically amount to around 15.500.000 : 318 = ~ 49.000,-- €. Without considering the costs for the transformers, this cost would amount to around € 25.000,--. Taking into account the overall costs of 15,5 M€ for the statistical avoidance of the reduction of the expectable number of voltage dips to half, these costs were equal to e.g. providing flywheel energy storage (1 MW, € 50.000,--/piece) for 320 enterprises or of flywheel energy storage for 160 enterprises (8 M€) + 1.500 smaller UPS (à € 300,--) during 50 years. Further options are given by the application of voltage stabilizers (dynamic voltage restorers, DVRs), with or without using energy storage units. These devices, normally expected to support the load for a short period, using heavy-duty batteries, super capacitors, or other forms of energy storage such as highspeed flywheels, generate the missing part of the supply [105]. For shallow dips, where there is considerable retained voltage, so that energy is still available, but at too low of a voltage to be useful to the load, there are several established automatic voltage regulator technologies (automatic voltage stabilizers) [105]. They rely on generating full voltage from the energy still available at reduced voltage during a dip. Because there is no need for any stored energy mechanism, these devices can be used for long-duration events. Attention is to be drawn to the selection of an automatic voltage stabilizer in a way to solve the particular problem without creating additional ones. For example, an inappropriate connection of a ferroresonant stabilizer to the output of an inferior generator aiming at a reduction of voltage variations would result in the affection by frequency fluctuations of the inferior generator, which would produce an AC voltage change of 1.5% for each 1% change of frequency. Opposite to the options for voltage stabilizing at MV network level, a well-proven measure to reduce the effect of voltage dips or short supply interruptions is the application of a UPS to retain the supply of the considered system at least for a time to arrange an orderly shutdown, thereby protecting the data, and therefore enabling the immediate restart of the process after return of the supply. Such UPSs cover a range from very small home user systems to huge systems for the protection of industrial processes. UPSs for the home or small commercial applications (for example, for the protection of PCs or smaller servers) cost about € 300,--/piece. When considering cost aspects related to the protection of PCs, smaller EDP systems, for 2003, a number of PC shipments for Western Europe of 31,076,639 units is reported, and for 2004, a number of such shipments of 33,720,772 units was forecasted [105]. For a much larger 230/400 VAC UPS used in a customer’s installation, the following data may serve as an example for the features/costs: • • • •

Max. storable energy: 2,7 MJ Bridgeable time at 100% loss of supply: 1.4 s at 1.400 kVA, 2,8 s at 800 kVA Purchase costs 1999: ~ 840.000 USD, operating costs per year: ~ 50.000,-- € Effect: 126 successful carryovers within 55 months

The effect of application of this UPS is to be evaluated by comparing the overall costs for purchasing and operation with the costs to be expected due losses resulting from 126 events over the time, i. e.: Expected costs = # Events x $/Event Alternatively, a buffering unit with electrolytic capacitor can be applied. Compared with a UPS, buffering units are smaller, don’t need any maintenance, and are cheaper. Operating without any battery, buffering units cause fewer problems at operation and disposal in general. DC-UPS or buffering unit should ensure • •

Bridging of a voltage dip/supply interruption by delivery of Ersatz-current. Notifying of the occurrence of a voltage dip/supply interruption to the supplied control unit for starting data storage for the case that the reserve energy would be used up.

126

•

Device oriented.

It is possible to cope with voltage dips/short supply interruptions by enhancing the immunity of electrical equipment and systems from these phenomena, an option that most economically were to be used at designing equipment/systems. Sometimes the replacement of older existing equipment, systems or control units, may be the most economic solution. See Brauner study, typical supply dip characteristic and ITIC curve [115].

Fig. A-2 Typical supply dip characteristic and ITIC curve As conclusions on PQ solutions cost, it appears as very costly to try to cope with the phenomena of voltage dips and short supply interruptions by improving the network performance. Through this type of solution, elimination of dips may be recognized as probably impossible. In some special cases, where the need justifies the expense, it may be possible to arrange for dual supplies that are derived from sufficiently separated parts of the grid as to be considered independent. Considering the different options for coping with voltage dips and short supply interruptions, Fig. A-3 [105] shows the given cost situations, being characterized by a steady increase of costs when moving from equipment through the plant to the infrastructure. The cheapest solution appears to be consideration of dips and short supply interruptions at equipment design, with small effort per piece to make it resilient to the effects of dips and short interruptions. In most cases, some form of mitigation equipment is applied within the customer’s installation, due to the kind of equipment/system to be protected.

Figure A-3 Cost mitigation increases as the power level of the load that must be protected increases

127

B - Hydro-Quebec-Ireq Report for Economical Aspect of Harmonics on Distribution and Transmission System In 1998, Hydro-Quebec finalized a technical and economical study on the impact of voltage and current harmonics on distribution network. Previous studies have been done earlier at a smaller scale covering only a few lines. The 1998 study was one of the first to cover an entire distribution network and the results were published in 1999 [165]. An interesting estimation of worldwide harmonic cost was then presented, based on a gross national product extrapolation of countries. Systematic methodologies were developed for the study considering the literature and know-how of that time. Formulas representing every aspect of harmonic losses were analyzed. Those formulas cover power losses (heating) and equipment loss of life. The study includes analysis of distribution and end user equipments. Three level of voltage harmonic were considered corresponding to 50%, 100% and 150% of the planning level fixed by IEC 61000-3-6 [130]. This approach gives a good figure of the projected economic losses in term of harmonic level and the results can be used to evaluate the impact of harmonic emission limit changes, which is mainly ruled by the network regulator.

B.1 Harmonics Power Losses Evaluation The evaluation of harmonic losses requires the knowledge of voltage and current harmonic level at PCC. For this reason, a typical Hydro-Québec distribution line was used to evaluate the harmonic current from the voltage harmonic planning level of the IEC 61000-3-6 [130]. The harmonics losses of this typical line were then used to extrapolate the results to Hydro-Québec overall distribution network.

B.2 Harmonics Losses Evaluation The following table gives the overall evaluation of power losses for the distribution network. The equipments and lines were considered as loaded at nominal capacity. Table B.1 - Distribution system power losses produced by harmonics at rated load (kW) Harmonics level of 50% Harmonics level of 100% Harmonics level of 150% IEC IEC IEC LV Line 9109 36424 81957 MV Line 6894 27575 62043 Transformer 2679 10713 24105 Capacitor 141 565 1271 Total 18823 75277 169376 A similar evaluation was done for the industrial load, which is presented in the following table. Table B.2 - Industrial power losses produced by harmonics at rated power (kW) Harmonics level of 50% Harmonics level of 100% IEC IEC Motors 18286 73145 Capacitors 111 443

Harmonics level of 150% IEC 164575 997

In the following tables, factors were applied to present a more realistic figure for the lines and the equipments loading (Lines = 33.8% ; Transformers = 36.4% ; Capacitors = 97% ; Motors = 36.7%). Those factors were determined by Direction Distribution, Hydro-Québec based upon the 1998 operating costs and use factor as a function of the annual fluctuation of network load. Table B.3 - Estimated distribution system power losses produced by harmonics (kW) Harmonics level of 50% Harmonics level of 100% Harmonics level of 150% IEC IEC IEC

128

LV Line MV Line Transformer Capacitor Total

3078 2330 975 137 6491

12311 9320 3899 548 26078

Table B.4- Estimated industrial power losses produced by harmonics (kW) Harmonics level of 50% Harmonics level of 100% IEC IEC Motors 6711 26844 Capacitors 108 430

27701 20970 8774 1233 58678 Harmonics level of 150% IEC 60399 967

B.3 Harmonic Losses Cost Evaluation A cost of 0.085$US for each kWh of energy sold was used to produce the annual cost shown in the next table. This rate is based upon the projected production cost of a kWh in a thermal plan for year 2000. Table B.5- Estimated annual cost for distribution system power losses produced by harmonics (k$US) Harmonics level of 50% Harmonics level of 100% Harmonics level of 150% IEC IEC IEC LV Line 2290 9167 20626 MV Line 1733 6940 15614 Transformer 726 2903 6533 Capacitor 102 408 918 Total 4833 19418 43692 Table B.6- Estimated annual cost for industrial power losses produced by harmonics (k$US) Harmonics level of 50% Harmonics level of 100% Harmonics level of 150% IEC IEC IEC Motors 4997 19988 44973 Capacitors 80 320 720

B.4 Conclusion Considering that in 2000 the total annual Hydro-Québec distribution electrical consumption was 150TWh, the following table presents the percentage of energy losses caused by harmonics. Table B.7- Estimated distribution system power losses produced by harmonics in percent of total energy used Harmonics level of 50% Harmonics level of 100% Harmonics level of 150% IEC IEC IEC LV Line 0.018 0.072 0.162 MV Line 0.014 0.054 0.122 Transformer 0.006 0.023 0.051 Capacitor 0.001 0.003 0.007 Total 0.038 0.152 0.343 At the time of the study, the harmonics level of Hydro-Quebec’s distribution network was evaluated at near the 50% IEC planning level. This represents a 0.038 percent of losses caused by harmonics. For comparison, a similar study [166], done over the Greek MV and LV distribution network, gives values in the range of 0.15% to 0.20%, which is more like a harmonics level of 100% IEC level of the HydroQuébec study. In Greek publication, simplifications and assumptions were made in order to obtain an over estimation of the economical impact of harmonics in the distribution network. The aim was to conclude that it is useless to put effort in network harmonics improvement. Both results could not be directly compared in term of network configuration, equipment characteristics, harmonics level, technique used and assumptions made, but they are quite similar.

129

(Losses) = (Harmonics)2

P = R x I2 losses % harm

50%

100% IEC threshold

150%

130

APPENDIX 4 A

Structuring the Data Collection Process

For the purpose of structuring the data collection process, a proper taxonomy can be useful; in the following the most important items are recalled and described. Critical sectors They are in general PQ critical and demonstrate similar PQ sector sensitivity, for which the methodology can be potentially targeted. CS1. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. CS2. 1. 2.

3. CS3

Industrial sectors type uni-product / uni-process (continuous manufacturing) Food / Beverage (production processing and preserving, NACE25 15) Glass, ceramics, cement, lime and stone (NACE 26) Metallurgy (NACE 27) Pharmaceutical ( NACE 24.4) Plastic and rubber (NACE 25) Publishing, printing and reproduction of recorded media (NACE 22) Pulp and Paper industry ( NACE 21) Refineries, chemical industry (NACE 23.2, 24 except 24.4) Semiconductor industry (NACE 32.1) Textile (particularly preparation, spinning and manufacture NACE 17.1 and 17.5) Wood and wood products (particularly production of sheets, boards and panels NACE 20.2) Industrial sectors type multi-product / multi-process Automotive industry (NACE 34) Continuous or highly automated or precision manufacturing – not defined in other sectors metal products (NACE: 28 fabricate metal products - except structure work 28.1 , 30 - office equipment, 31- electrical equipment, 32 – except 32.1 – RTV and telephony electronics, 33 – medical equipment) Manufacture of machinery (NACE 29 and 31) Services sectors

1. 2. 3. 4. 5. 6. 7.

Air transport (NACE2 62) Database activities e.g. hosting services (NACE 72.4) Financial intermediation (particularly central banking, but also other general transactions, section J; NACE 65-67) Hospitals (NACE 85.1) Hotels (NACE 55.1) Railways (NACE 60.1) Telecommunications (NACE 64.2)

Cost categories 1. 2. 3. 4. 5.

Process interruptions Process slowdown Equipment damage Reduced lifetime and mis-operation (postponed costs) Reduced energy efficiency - increased energy loss

25

The NACE Code is a pan-European classification system which groups organisations according to their business activities; http://ec.europa.eu/comm/competition/mergers/cases/index/nace_all.html

131

6. 7. 8.

Product quality Worker productivity Other indirect costs

Cost types - Operating consequences 1. 2. 3. 4.

5. 6. 7.

WIP loss, often referred to as production loss or production damage. This category includes this part of labor and material costs which has been inevitably lost. This category has two major components - labor and material cost. Working capacity loss – basically quantifies efforts to make up this part of production which can still be repaired or reused – WIP recovery Labor cost resulting from production outage – lost or extra paid Other related costs when quantification using above mentioned categories is not easily possible. These could be a process slow down when it cannot reach its nominal efficiency including process restart cost and additional maintenance costs. These include process and process restart cost. Equipment related costs, including equipment damage and replacement costs, hire of temporary equipment, and running costs of back up equipment Indirect costs, e.g. consequences of late delivery such as penalties to clients, extra compensation to personnel, cost of personnel or equipment evacuation, extra insurance cost Savings from unused resources (labor, energy, material).

PQ phenomena 1. 2. 3. 4. 5. 6.

Voltage dips and short interruptions Harmonics (current and voltage) Surges and transients Flicker Unbalance Earthing and EMC

Equipment: as a PQ source and affected by PQ 1. 2. 3. 4. 5. 6. 7. 8. 9.

Capacitors Contacts and relays Electric motors Electronic equipment Lighting equipment Processing equipment UPS uninterruptible power supplies VSD and other static converters Welding and smelting equipment

PQ consequences 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

Circuit breakers (including protective devices) nuisance tripping Capacitor damage Capacitors – dielectric loss Computer lock up Computers / other electronics damaged Data loss Electric shock Lights flicker or dim Loss of synchronization of processing equipment Motors / process equipment - malfunction or damage Motors overheating – energy losses Noise interference to telecom lines Relays /contactors nuisance tripping

132

14. Transformers / cables overheating with related energy losses 15. Premature ageing and loss of reliability of electrical equipment 16. Overheating of neutral conductor in lines and transformers and related problems (e.g. transient overvoltage, tripping of RCDs, losses) Solutions 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

B

Equipment immunity Backup generator Dynamic voltage restorers Harmonic filter Isolation transformers Line conditioners or active filters Multiple independent feeder Oversizing equipment Shielding and grounding Site generation capable of substituting supply Static transfer switches Static VAR compensator Surge protectors on key pieces of equipment Uninterruptible power supply (UPS) devices Voltage stabilizers

Executing Data Collection Process – End User Perspective

With reference to the Model A presented in Chapter 4, the following step-by-step procedure can be recommended to estimate the process interruption cost, PIC: •

Step 1: Based on the assumptions mentioned above, evaluate the total number of product variants, the total number of process activities at any given instant and the maximum number of potential failures among all process activities.

•

Step 2: For each product variant, determine the associated progressive cost components from A1 to A726 for each process activity. Note: If for a particular process activity the maximum number of failure scenarios is less than the maximum number of failure scenarios in all process activities, then the cost components A2, A3, A4 and A5 associated with failure scenarios assumes zero value. Establish the cost related to component A6, particularly employees tolerance for each failure instance of process activity. Establish customer satisfaction and reputation retained level for instance of nondelivery of a product variant in time. Finally calculate savings A7 due to failure scenarios.

•

Step 3: Prepare a work schedule highlighting the active process activities for a typical day for which process interruption cost profile has to be established. This work schedule should include process activities for various product variants and their simultaneous.

The proposed specification and division of sectors by Taxonomy A may help an end user to focus economic data collection on certain aspects. The proposed methodology suits the collection of cost data in ‘industrial - uni-process’ sectors. Most of processes are organized in a series topology as indicated in the Figure B.1. With reference to Fig.B.1, the stream I is the simple extreme. In such a case, close attention should be paid to the calculation of process interruption costs as all the processes are closely interdependent. In a ‘Just in time’ scenario, where there are no buffers in the process, one process failure may stop the whole production line. These sectors are particularly vulnerable to voltage dips and related process interruptions For sectors classified as ‘Industrial - multi-process’, the production process is less often performed in continuous stream – Figure B.1 – stream VI, as an extreme. In such case one process interruption may not 26

Chapter 4 clause 4.3.2

133

necessarily stop other processes. The consequences are limited to the critical processes which are needed to make up for lost lead time of the final product. The focus is therefore on extra cost (e.g. bonus extra time labor cost) to recover lost or partly lost WIP (A127). These sectors are vulnerable to voltage dips but also other phenomena like harmonics and unbalance, transient and surges.

I II III IV

VI

V

Figure B.1: Six typical configurations considered for industrial processes In the ‘Services’ sectors it can be difficult to distinguish the root cause of process interruption, particularly in a commercial environment where software, hardware or a PQ issue maybe responsible. Once the root cause has been attributed to PQ, the consequences could be: • • • •

Loss of transactions in progress requiring data recovery, reprocessing and repeated transmission. The standard data collection process described here should be modified either using a mix of A1, A2 and A5 cost components. Process restart cost using A4 cost calculation Other costs, particularly lost revenues (missed opportunities) as result of customer dissatisfaction and loss of reputation, but also such consequences as penalties and other elements of A5 cost component. Potential savings from A7 are usually negligible.

The alternative to the procedure described above is to use A3, the process slow down method, to simply calculate business slow down rate. In addition, due to lack of clear differentiation whether a process was interrupted or not, all phenomena related cost components should be used to check that nothing has been omitted. It also should be checked whether any items have been double counted, particularly A5 (equipment damage due to process interruption) and equipment damage due to occurrence of PQ disturbance. Service sectors are relatively more vulnerable to the consequences of long interruptions but PQ may still be the root cause of substantial economic losses.

C

Conclusions

Deregulation and industry restructuring are placing utilities under increasing pressure to both improve customer reliability and decrease cost. To remain competitive, it is critical to prioritize maintenance tasks so that the best possible reliability is achieved with increasingly constrained maintenance budgets.

27

Chapter 4, clause 4.3.2

134

The purpose of maintenance is to extend equipment lifetime and/or reduce the probability of failure. Corrective maintenance replaces or repairs failed components, while preventive maintenance is a proactive effort to improve the condition of an unfailed component that may be deteriorated to some degree. Power quality can be surveyed for three major purposes: 1. 2. 3.

Technical reasons (which are more important in industrial centers and to facility managers) Economic reasons (including all sectors linked with the electrical system) Social reasons (in which the governmental system is bound to offer desirable services)

Appropriate quality of electrical energy can greatly reduce expenses arising from losses or system disturbances. Improvement of power quality can overcome these problems as well as increase equipment longevity and system reliability. In poor power quality, financial damages imposing upon residential, industrial, and trade consumption would be very different. Therefore, it must not be neglected that some huge portion of electrical energy is consumed in residential utilization. Losses of power, decline in useful life of power system equipment, as well as non-purchased energies due to power quality deficit are counted as parts of financial damages imposed upon facility managers. Furthermore, governments fulfill community satisfaction and demands with legislation for facility managers. All the above-mentioned instances appear as positive pressure in promotion of power quality in a power system. The scientific response is that they are only due to economic limitations, which of course the alternative implication may be suffering from poor initiatives in establishment of state laws in this respect and also lacking practical scientific capability necessitated in supplying those wants. In this report, attempts have been made toward implicating economic damages resulting from quality problems encountered with shape of consumption. Requirement investment for power quality promotion counts as main criteria for comparison.

135

APPENDIX 5 A

Illustrative Case Study

This section illustrates the application of the NPV approach to a hypothetical high-tech facility based on an actual facility. The example considers a semiconductor wafer-fabrication factory located in the United States. Wafer fabrication requires a high level of power quality and reliability due to the sensitivity of the equipment and process controls and therefore is a strong candidate for applying the NPV analysis.

A.1

Base Case: Facility Data and Base Case Calculations

For purposes of this case study, each process interruption resulting from an unmitigated voltage dip is assumed to cause an overall business loss of $500,000 from various factors, including lost production, extra labor, and scrap. Losses due to unmitigated protracted power interruptions are expressed as a constant $/hr rate of $750,000/hour. The actual measured rates for PQ and reliability phenomena for this case study, shown in Table A.1, were obtained from power quality monitoring and/or statistical analysis of historical data. Although this facility was subjected to 73 voltage dips during the course of the year, only 12 dips were sufficiently low to cause a process interruption (see Figure A.1). Only those voltage dips resulting in process interruptions (i.e. costs to the facility) are useful for the NPV analysis. This data, including assumptions for other sources of PQ-related costs, is shown in Table A.1. A combined discount and inflation rate of 5% was assumed for this analysis. Table A.1 Assumed Rates and Costs of PQ and Reliability Phenomena Failure Type Failure Repair Time Costs Rate Long-term utility interruption 2/year 4 $750K per hour (feeder) hours/interruption Voltage dips (producing 12/year 1 hour/dip $500K per event interruptions) Transformer and local equipment 0.1/year 3 hours/interruption $750K per hour failure RMS Variation Magnitude-Duration Scatter Plot Sensitivity of Facility Equipment to Voltage Sags

2.00

Total Events: 73 Events Violating ITIC Lower Curve: 12 Events Violating ITIC Upper Curve: 0

Voltage Magnitude (pu)

1.75 1.50 1.25 1.00 0.75 0.50 0.25 0.00 10-3

10-2

10-1

100 Seconds

101

102

Electrotek/EPRI

103 PQView®

Figure A.1. Process Susceptibility to Voltage Dips at the Fabrication Facility Having identified the cost of unmitigated PQ events and a discount/inflation rate, we now turn our attention to the impact that different mitigation approaches will have. The analysis below employs different scenarios, or cases, to allow a comparison of facility benefits/costs with different combinations of mitigation.

136

The base case NPV analysis is summarized in Table A.2. Note that the facility loses US$12.2 M/year due to the cost of unmitigated power quality events when no mitigation is employed. Table A.2: Base Case: NPV for Existing Facility and Conditions

A.2

Case 1: Redundancy in the Utility Supply

One method to mitigate a power supply interruption is to have an alternate feeder with a fast switch. The alternate feeder is assumed to have the same reliability and PQ characteristics as the primary feeder (see Table A.1). The costs associated with the alternate feeder are shown in Table A.3. Failure Mode and Effect Analysis (FMEA) for this case is given in Table A.4. Table A.3 Feeder Cost Information Initial cost of building the $ 525,000 feeder + mechanical switch Installation cost 10% of initial cost O&M cost of the feeder + 5% of initial mechanical switch cost Useful life 10 years Table A.4 FEMA Analysis: Case 2 Failure Mode for Effect Interruption Supply would switch on to Feeder 1 failure feeder 2. The critical load won’t be affected. Supply would switch on to Feeder 2 failure feeder 1. The critical load won’t be affected. Feeder 1 and 2 both fail Critical load is interrupted. Transformer or cable Critical load is interrupted. failure Voltage dips from the Critical load is interrupted. utility side

137

The two-feeder system introduces redundancy in the power supply. A reliability block diagram (RBD) for interruption failure is shown in Figure A.2. For the reliability calculations, a common-mode factor of 0.05 is assumed. The QRA analysis results are shown in Table A.5. For the cost analysis, an equipment life of ten years is assumed.

Feeder 1 Local switchgear Feeder 2

Figure A.2. Adding a Redundant Utility Feeder Table A.5 Ten-year NPV results for a Redundant Feeder

Adding a redundant feeder to this facility does nothing to mitigate voltage dips, but significantly reduces the impact of protracted interruptions, thereby improving the NPV of unmitigated PQ by approximately $40M.

A.3

Case 2: Applying a Battery UPS

Case 1 examined the benefit of reducing the impact of long outages. However, the cost impact of voltage dips is considerably greater, making it likely that the most beneficial mitigation option will likely include a dip-mitigation solution. This case considers a battery UPS for dip ride-through. Details of the UPS configuration and cost are given in Table A.6. Table A.6 UPS Information Failure rate of each unit Repair time per unit Redundancy Common-mode factor for 2-out-of-3 and 3-out-of-3 failure mode Initial cost of three units Installation cost O&M cost/yr UPS life span

1 failure/yr 6 hours/yr 2 out of 3 0.01 $1,000,000 $100,000 $100,000 10 years

138

FEMA analysis for this case is shown in Table A.7. N-1 redundancy is assumed. Therefore, the UPS system would protect the load even if one unit (out of three) fails. A common-mode failure factor of 0.01 is assumed for the redundancy calculations. The RBD for the UPS configuration is shown in Figure A.3. Table A.7 FEMA Analysis: Case 3 Failure Mode for Voltage Effect Dips 1 out of 3 UPS units fails. Critical load will still be protected. 2 out of 3 units fail. Critical load would be exposed to dips. 3 out of 3 units fail. Critical load would be exposed to dips. By-pass switch

UPS 1

UPS 2

Critical Load

UPS 3 2 out of 3 UPS system

Fig. A.3 NPV Calculation for Adding Redundant UPS As shown in Table A.8, adding redundant UPSs to protect critical loads reduces the average number of unmitigated voltage dips from 12 to 0.0135 per year, resulting in substantial savings and a positive net present value. Table A.8 10-year NPV Results: Redundant UPS

Because of its impact on voltage dips, a redundant UPS has a profound impact on reduction of unmitigated PQ—all but eliminating these costs while adding capital and ongoing maintenance costs. The overall improvement in NPV with this option is over $90M (i.e., -$2M – (-$95M) = $93M).

139

A.4

Case 4: Using Distributed Energy Resources (DER)

From cases 2 and 3, it is evident that the optimal QRA solution should include mitigation of most voltage dips as well as interruptions. This case is built upon case 3, except that distributed energy resources (DER) (in this case, on-site generation) is considered for protecting against long interruptions. The QRA analysis for dips is the same as in Case 3. DER information is given in Table A.9. Table A.9 DER Information 2/year DER failure rate (λ) DER repair time 6 hours Redundancy 1-out-of-2 Initial cost of DER $1,000,000 Installation cost $100,000 Fuel cost/yr $95,278 O&M cost/yr $50,000 DER life span 10 years The facility is assumed to have two DER units with N-1 redundancy. FMEA analysis for this case is shown in Table A.10. NPV for long interruptions is shown in Figure A.4. The block of “Local DGs” represents the equivalent parallel combination of two DER units in parallel at the facility. Table A.10 FEMA Analysis: Case 4 Failure Mode for Interruption Effect The utility feeders fail. Backup generator should come online. Critical load won’t be affected. Utility feeders fail and the DERs Critical load will be interrupted. fail to start.

Utility feeder Local switchgear Local DGs

Fig.A.4 Configuration of local distributed generation with utility feed The results of the QRA analysis are shown in Table A.11. For the cost analysis, it is assumed that DER provides ancillary benefits such as CHP and peak shaving. The savings due to these are assumed to be 5% of the total initial cost. Positive NPV indicates that DER can be an economical option for the semiconductor-fabrication process. Note that DER will not provide any protection against voltage dips. The NPV in Table A.11 does not include the losses due to twelve voltage dips per year. If losses due to voltage dips are considered, NPV will come out to be 0.88M$. Therefore, installing DER without any dip ride-through technology is not an effective QRA solution. Table A.11 QRA Results: Case 4

140

Distributed generation is able to mitigate the duration of power outages; however, it’s lack of impact on voltage dips is a strong disadvantage in this particular analysis. At $44M, the overall improvement in NPV of this option is roughly comparable to that of a redundant power feed (-$51M – (-$95M) = $44M).

B

Case Comparison and Sensitivity

The results of the Base Case and three alternative cases are illustrated in Figure B.1. Although a myriad of other options can be considered, among the three cases, Case 2: Battery UPS offers the best 10-year NPV and should be seriously considered, along with other solutions that mitigate the impact of voltage dips. It is also important to note, however, that the other options may offer benefits not taken into account in this analysis, such as improved safety, environmental issues, or options for combined heat and power (CHP) or cogeneration. These and many other issues can be considered in the NPV analysis depending on the sophistication of the application and those designing it. The key is that all cases considered be treated equivalently. Comparison of NPV Cases $0 Base Case

Case 1: Redundant Feeder

Case 2: Battery UPS

Case 3: On-Site Generation

NPV (10-yr)

-$20,000,000

-$40,000,000

-$60,000,000

-$80,000,000

-$100,000,000 Cases

Figure B.1 NPV Values for Mitigation Cases It is also important to consider the sensitivity of such an analysis to variations in the input parameters used—in particular, the number of voltage dips assumed per year can have a profound impact on the results of the NPV calculations. For example, if the true cost of voltage dips to this facility were actually

141

$250,000 per event rather than $500,000, the economic analysis would be impacted as illustrated in Figure B.2 below. Although the relative rankings of the various solutions considered here are not changed, the disparity in their impact on 10-year NPV is considerably reduced. For PQ solutions that address only voltage dips (such as dynamic voltage restorers, etc.), reassessment of the cost of individual dips would likely have a profound impact on economic performance. Comparison of NPV Cases $0 Base Case

Case 1: Redundant Feeder

Case 2: Battery UPS

Case 3: On-Site Generation

NPV (10-yr)

-$20,000,000

-$40,000,000

-$60,000,000

-$80,000,000 Cases

Figure B.2 NPV Values for Mitigation Cases with reduced impact of voltage dips

142

REFERENCES [1] Bollen, M., Verde, P., 2008. A framework for regulation of rms voltage and short-duration under and overvoltages. IEEE Transactions on Power Delivery. Vol. 23, Issue 4, Oct. 2008, pages: 2105 – 2112. [2] NRS, 2004. Electricity Supply - Quality of supply - Part 2: Voltage characteristics, compatibility levels, limits, and assessment methods. NRS 048-2:2003. NRS, Pretoria, South Africa. [3] ERGEG, 2007. Towards voltage quality regulation in Europe. Conclusion paper E07-EQS-1503. Available from: [www.ergeg.org]. [4] CEER, 2005. Third Benchmarking Report on Quality of Electricity Supply. Available from www.energy-regulators.eu. [5] CEER, 2008. Fourth Benchmarking Report on Quality of Electricity Supply. Available from www.energy-regulators.eu. [6] ERD, 2000. Contrat d'accès au réseau public de distribution d'électricité pour un site consommateur éligible raccordé en moyenne tension. Available (in French) from: [www.edfdistribution.fr]. [7] Seljeseth, H., Sand, K., Samdal, K., 2005. Quality of supply regulation in Norway: going beyond EN 50160. In: CIRED 2005, International conference on electricity distribution, Turin, Italy]. [8] P. Heine, P. Pohjanheimo, M. Lehtonen, and E. Lakervi, “A method for estimating the frequency and cost of voltage dips,” IEEE Transactions on Power Systems, vol. 17, pp. 290 – 296, 2002. [9] P. Heine, P. Pohjanheimo, M. Lehtonen, and E. Lakervi, “Estimating the annual frequency and cost of voltage dips for customers of five Finnish distribution companies,” presented at Electricity Distribution, 2001. Part 1: Contributions. CIRED. 16th International Conference and Exhibition on (IEE Conf. Publ No. 482), 2001. [10] S. Quaia and F. Tosato, “Interruption costs caused by supply voltage dips and outages in small industrial plants: A case study and survey results,” in The IEEE Region 8 EUROCON. Computer as a Tool, vol. 2, 2003, pp. 258 – 262, vol.2. [11] M. M. A. Aziz, G. A. A. Salam, and S. M. Kozman, “Cost and mitigation of voltage dip for industrial plants,” presented at Electrical, Electronic and Computer Engineering, 2004. ICEEC ’04. 2004 International Conference on. [12] H. Arango, E. G. Domingues, J. P. G. Abreu, and D. M. Camposilvan, “Applying real options methodology to value electrical PQ projects,” presented at Harmonics and Quality of Power, 2002. 10th International Conference on, 2002. [13] S. Quaia and F. Tosato, “A method for the computation of the interruption costs caused by supply voltage dips and outages in small industrial plants,” presented at EUROCON 2003. Computer as a Tool. The IEEE Region 8, 2003. [14] R. Lamedica, A. Patrizio, A. Prudenzi, E. Tironi, and D. Zaninelli, “PQ costs and upgrading solutions: The energy centre,” presented at Harmonics and Quality of Power, 2000. Proceedings. Ninth International Conference on, 2000. [15] J. V. Milanovic and C. P. Gupta, “Probabilistic assessment of financial losses due to interruptions and voltage dips – Part I: The methodology,” IEEE Transactions on Power Delivery, vol. 21, pp. 918 – 924, 2006. [16] J. V. Milanovic and C. P. Gupta, “Probabilistic assessment of financial losses due to interruptions and voltage dips – Part II: Practical implementation,” IEEE Transactions on Power Delivery, vol. 21, pp. 925 – 932, 2006. [17] M. McGranaghan and B. Roettger, “Economic evaluation of PQ,” IEEE Power Engineering Review, vol. 22, pp. 8 – 12, 2002. [18] G. J. Lee, M. M. Albu, and G. T. Heydt, “A PQ index based on equipment sensitivity, cost, and network vulnerability,” IEEE Transactions on Power Delivery, vol. 19, pp. 1504 – 1510, 2004. [19] IEEE recommended practice for evaluating electric power system compatibility with electronic process equipment. IEEE Standard 1346-1998, 1998. [20] J. Y. Chan, J. V. Milanovic, and A. Delahunty, “Generic failure risk assessment of industrial processes due to voltage dips,” IEEE Transactions on Power Delivery, to be published. [21] P. Pohjanheimo and M. Lehtonen, “Introducing Prob-A-Dip - A probabilistic method for voltage dip management,” in 11th International Conference on Harmonics and Quality of Power, 2004, pp. 250254.

143

[22] J. Y. Chan and J. V. Milanovic, “Methodology for assessment of financial losses due to voltage dips and short interruptions,” in 9th International Conference on Electrical PQ and Utilisation. Barcelona, 2007. [23] J. Y. Chan and J. V. Milanovic, “Severity indices for assessment of equipment sensitivity to voltage dips and short interruptions,” in IEEE PES General Meeting 2007. Tampa, Florida, 2007. [24] J. Wang, S. Chen, and T. T. Lie, “Estimating economic impact of voltage dips,” vol. 1, 2004 International Conference on Power System Technology, POWERCON 2004. Singapore: Institute of Electrical and Electronics Engineers Inc., New York, NY 10016 – 5997, United States, 2004, pp. 350 – 355. [25] S. Y. Yun and J. C. Kim, “An evaluation method of voltage dip using a risk assessment model in power distribution systems,” International Journal of Electrical Power and Energy Systems, vol. 25, pp. 829 – 839, 2003. [26] R. C. Leborgne, J. M. C. Filho, J. P. G. d. Abreu, T. C. Oliveira, A. A. Postal, and L. H. Zaparoli, “Alternative methodology for characterization of industrial process sensitivity to voltage dips,” in IEEE Bologna PowerTech Conference Proceedings, vol. 3. Bologna, Italy, 2003, pp. 6. [27] F.Tosato and S.Quaia, “A method for the computation of the interruption costs caused by supply voltage dips and outages in small industrial plants,” in The IEEE Region 8 EUROCON 2003. Computer as a Tool, vol. 2, 2003, pp. 249 – 253. [28] C. P. Gupta, J. V. Milanovic, and M. T. Aung, “The influence of process equipment composition on financial losses due to voltage dips,” in 11th International Conference on Harmonics and Quality of Power, Lake Placid, NY, United States, 2004, pp. 28 – 34. [29] M. Delfanti, E. Fumagalli, P. Garrone, L. Grilli, and L. L. Schiavo, “Towards voltage quality regulation in Italy”. Submitted to IEEE Transactions on Power Delivery. [30] J. T. Crozier and W. N. Wisdom, “A PQ and reliability index based on customer interruption costs,” IEEE Power Engineering Review, vol. 19, pp. 59 – 61, 1999. [31] R. Ghandehari and A. Jalilian, “Economical impacts of PQ in power systems,” Bristol, United Kingdom, 2004. [32] Y. Shih-An, S. Chun-Lien, and C. Rung-Fang, “Assessment of PQ cost for high-tech industry,” presented at Power India Conference, 2006 IEEE, 2006. [33] M. F. McGranaghan and W. C. Roettger, “The economics of custom power,” presented at Transmission and Distribution Conference and Exposition, 2003 IEEE PES, 2003. [34] M. Lehtonen and B. Lemstrom, “Comparison of the methods for assessing the customers outage costs,” in International Conference on Energy Management and Power Delivery, vol. 1, 1995, pp. 1 – 6. [35] IEEE recommended practice for monitoring electric PQ. IEEE Standard 1159-1995, 1995. [36] B. Lee, G. K. Stefopoulos, and A. P. S. Meliopoulos, “Unified reliability and PQ index,” in 12th International Conference on Harmonics and PQ, Portugal, 2006. [37] J. V. Milanovic and N. Jenkins, “PQ cost for manufacturing industries,” Paris, 2003. [38] F. Carlsson and P. Martinsson, “Does it matter when a power outage occurs? – A choice experiment study on the willingness to pay to avoid power outages,” Department of Economics, Gothengurg University, 2004. [39] F. Carlsson and P. Martinsson, “Willingness to pay among Swedish households to avoid power outages – A random parameter Tobit model approach,” Department of Economics, Gothenburg University, 2004. [40] R. Targosz and J. Manson, “Pan European LPQI PQ survey,” in CIRED 19th International Conference on Electricity Distribution, Vienna, 2007. [41] K. Samdal, G. Kjølle, B. Singh, and F. Trengereid,. Customers’ interruption costs: What’s the problem? In: CIRED 2003, International Conference on Electricity Distribution, Barcelona, Spain. [42] K. K. Kariuki and R. N. Allan, “Evaluation of reliability worth and value of lost load,” IEE Proceedings: Generation, Transmission and Distribution, vol. 143, pp. 171 – 180, 1996. [43] T. Andersson and D. Nilsson, “Test and evaluation of voltage dip immunity,” STRI AB and Vatenfall AB, 27 Nov. 2002. [44] D. Chapman, “The cost of poor PQ,” Copper Development Association, November 2001. [45] M. J. Sullivan, T. Vardell, and M. Johnson, “Power interruption costs to industrial and commercial consumers of electricity,” IEEE Transactions on Industry Applications, vol. 33, pp. 1448 – 1457, 1997. [46] K. H. LaCommare and J. H. Eto, “Understanding the cost of power interruption to U.S. electricity consumers,” Ernest Orlondo Lawrence Berkeley National Laboratory, LBNL-55718, 2004. [47] Primen, “The cost of power disturbances to industrial & digital economy companies,” EPRI CEIDS, June 2001.

144

[48] EPRI, “Incentives and penalties for electrical PQ under performance based regulations,” CEA Technologies Inc. (CEATI) T044700-5124, November 2005. [49] B. H. Chowdhury, “PQ,” in IEEE Potentials: The Magazine for Up-and-Coming Engineers, 2001, pp. 5 – 11. [50] J. D. Lamoree, “Cost of Outages,” Enernex Corporation, October 2004. [51] Y. Shih-An, L. Chan-Nan, E. Liu, H. Yu-Chang, and H. Chinug-Yi, “Assessment of interruption cost to high tech industry in Taiwan,” presented at Transmission and Distribution Conference and Exposition, 2001 IEEE/PES, 2001. [52] S. B. Choi, K. Y. Nam, D. K. Kim, S. H. Jeong, H. S. Ryoo, J. D. Lee, “Evaluation of interruption costs for industrial customers in Korea,” presented at Power System Conference and Exposition, 2006. PSCE’06. [53] G. Linhofer, P. Maibach, and F. Wong, “PQ devices for short term and continuous voltage compensation,” in International PQ Conference 2002, Singapore, 2002. [54] E. Emanuel, M. Yang, and D. J. Pileggi, “The engineering economics of power system harmonics in subdistribution feeders: A preliminary study”, IEEE Transactions on Power Systems, Vol. 6, No. 3, August 1991, pp. 1092 – 1098. [55] A. E. Emanuel, J. Pileggi, T. J. Gentile, E. M. Gulachenski, D. Sorensen, and J. Janczak, “Distribution feeders with nonlinear loads in the Northeast U.S.A.: Part I – Voltage distortion forecast”, IEEE/PES Winter Meeting, Jan. 1994, New York (USA), 94 WM 089-3 PWRD. [56] J. Pileggi, T. J. Gentile, A. E. Emanuel, E. M. Gulachenski, M. Breen, D. Sorensen, and J. Janczak, “Distribution feeders with nonlinear loads in the Northeast U.S.A.: Part II – Economic evaluation of harmonic effects”, IEEE/PES Winter Meeting, Jan. 1994, New York (USA), 94 WM 090-1 PWRD. [57] P. Caramia, G. Carpinelli, E. Di Vito, A. Losi, and P. Verde, “Probabilistic evaluation of the economical damage due to harmonic losses in industrial energy systems”, IEEE Transactions on Power Delivery, vol. 11, no. 2, April 1996, pp. 1021 – 1031. [58] P. Caramia, G. Carpinelli, A. Losi, A. Russo, and P. Verde, “A simplified method for the probabilistic evaluation of the economical damage due to harmonic losses”, 8th Int. Conference on Harmonics and Quality of Power, October 1998, Athens (GR), pp. 767 – 776. [59] G. Mazzanti, A. Cavallini, G.C. Montanari, P. Caramia, G. Carpinelli, and P. Verde, “An approach to life estimation of electrical plant components in the presence of harmonic distortion,” presented at ICHQP 2004, Lake Placid (NY), USA. [60] P. Verde, “Cost of harmonic effects as meaning of standard limits,” Proceedings. Ninth International Conference on Harmonics and Quality of Power, 2000, Volume 1, 1 – 4 Oct. 2000 pp. 257 – 259, vol.1. [61] P. Caramia and P. Verde, “Cost-related harmonic limits,” Power Engineering Society Winter Meeting, 2000. IEEE Volume 4, 23 – 27 Jan. 2000, pp. 2846 – 2851. [62] A. Cavallini, D. Fabiani, G. Mazzanti, G.C. Montanari, and A. Contin, “Voltage endurance of electrical components supplied by distorted voltage waveforms,” Proc. 2000 International Symposium on Electrical Insulation, pp. 73 – 76, Anaheim, California (USA), April 2000. [63] D.Gallo, R. Langella, and A. Testa, “Predicting voltage stress effects on MV/LV components,” 2003 IEEE Bologna PowerTech Conference, June 23 – 26, Bologna, Italy. Volume 2. [64] IEEE Task Force, “The effects of power system harmonics on power system equipment and loads,” IEEE Transactions Power Apparatus and Systems, Vol. PAS-104, Sept 1985, pp. 2555 – 2563. [65] IEEE Task Force, “Effects of harmonics on equipment,” IEEE Transactions on Power Delivery, Vol. 8, April 1993, pp. 672 – 680. [66] M. A. Miner, “Cumulative damage in fatigue,” Journal of Applied Mechanics, September 1945, pp. A159 – A163. [67] G. J. Anders, Rating of electric power cables, Mc Graw hill, IEEEE Press, Piscataway, NJ, 1997. [68] L. Simoni, “Fundamentals of endurance of electrical insulating materials,” CLUEB Publ., 1st issue 1985, 2nd issue 1993. [69] Z. Kowalski, Unbalance in Power Systems [in Polish], PWN 1987. [70] Y. Baghzouz et al., “Time-varying harmonics. II. Harmonic summation and propagation,” IEEE Transactions on Power Delivery, Volume 17, Issue 1, Jan. 2002, pp. 279 – 285. [71] R. Billinton and R. N. Allan, Reliability Evaluation of Power Systems, Plenum Press, New York, 1994. [72] Math H. J. Bollen, Under PQ problems: Voltage dips and interruptions, IEEE Power Engineering Series, 2000. [73] L. E. Conrad and M. H. J. Bollen, “Voltage dip coordination for reliable plant operation,” IEEE Transactions on Industry Applications, Vol. 33, No. 6, Nov./Dec. 1997, pp. 1459 – 1463.

145

[74] M. F. McGranaghan, D. R. Mueller, and M. J. Samotyj, “Voltage dips in industrial systems,” IEEE Transactions on Industry Applications, Vol. 29, No. 2, March/April 1993, pp. 397 – 402. [75] S. Z. Djokic, K. Stockman, J.V. Milanovic, J. J. M. Desmet, and R. Belmans, “Sensitivity of AC adjustable speed drives to voltage dips, short interruptions and under-voltage transients,” IEEE Transactions on Power Delivery (in press). (Manuscript no. TPWRD- 00380-2003 accepted for publication in November 2003). [76] EPRI Brief Report No. 46, “Performance of a hold-in device for relays, contactors and motorstarters,” October 1998. (http://www.portlandgeneral.com/business/products/business power problems/PQ school/brief46.pdf). [77] E. Collins and M. Bridgwood, “The impact of power system disturbances on AC-coil contactors,” Proc. IEEE Textile, Fiber and Film Industry Technical Annual Conference, Greenville, USA, 6-8 May 1997. [78] S. Z. Djokic, J. V. Milanovic, and D. S. Kirschen, “Sensitivity of AC coil contactors to voltage dips, short interruptions and under-voltage transients,” IEEE Transactions on Power Delivery, Vol 19, No. 3, 2004, pp. 1299 – 1307. [79] R. Caldon, M. Fauri, and L. Fellin, “Voltage dip effects on continuous industrial processes: Desensitizing study for textile manufacture,” Proc. of Second Int. Conf. on PQ: End-Use Applications and Perspectives, Atlanta, USA, Sept. 28 – 20, 1992, pp. D – 13: 1-6. [80] S. Z. Djokic, J. Desmet, G. Vanalme, J.V. Milanovic, and K. Stockman, “Sensitivity of personal computers to voltage dips, short interruptions,” IEEE Transactions on Power Delivery (in press). (Manuscript no. TPWRD- 00159-2003 accepted for publication in November 2003 ). [81] J. Lamoree, D. Mueller, P. Vinett, W. Jones, and M. Samotyj, “Voltage dip analysis case studies,” IEEE Transactions on Industry Applications, Vol. 30, No. 4, July/August 1994, pp. 1083 – 1089. [82] R. G. Brown and P. Y. C. Hwang, Introduction to random signals and applied Kalman filtering, John Wiley and Sons, New York, 1997. [83] R. Billinton and R. N. Allan, Reliability evaluation of engineering systems, Plenum Press, New York, 1992. [84] M. T. Aung, J. V. Milanović, and P. A. Simmons, “Automated comprehensive assessment and visualization of voltage dip performance,” CD Rom of the 11th IEEE International Conference on Harmonics and Quality of Power, ICHQP 2004, Lake Placid, NY, USA, September 12 – 15, 2004. [85] G. Olguin and M. H. J. Bollen, “The method of fault positions for stochastic prediction of voltage dips: A case study,” CD-ROM of the International Conference on Probabilistic Methods Applied to Power Systems, Naples, Italy, September 22 – 26, 2002. [86] G. Mazzanti, G. Passarelli, A. Russo, and P. Verde, “The effects of voltage waveform factors on cable life estimation using measured distorted voltages,” Power Engineering Society General Meeting, 2006. IEEE, 18 – 22 June 2006, p. 8. [87] P. H. Allen, and A. Tustin, “The aging process in electrical insulation: A tutorial summary,” IEEE Transactions on Electr. Insul., vol. EI-7, 1972, pp. 153 – 161. [88] A. Cavallini, G. Mazzanti, G. C. Montanari, and D. Fabiani, “The effect of power system harmonics on cable endurance. A critical review to IEEE std. 519 voltage distortion limits,” Proc. IAS Annual Meeting, Roma (Italy), October 2000. [89] G. C. Montanari and G. Pattini, “Thermal endurance of insulating materials,” IEEE Transactions on Electrical Insulation, Vol. 21, No. 1, February 1986, pp. 66 – 75. [90] J. Manson, Business model for investing in PQ solutions, PQ application guide background notes, Chapter 2 – “Costs,” www.lpqi.org. [91] A. H. Eltom and N. S. Moharari, “Motor temperature estimation incorporating dynamic rotor impedance,” IEEE Transactions on Energy Conversion, vol. 6, 1991, pp. 107 – 113. [92] J. Aubin, R. Bergeron, and R. Morin, “Distribution transformer overloading capability under cold-load pickup conditions,” IEEE Transactions on Power Delivery, vol. 4, 1990, pp. 1883 – 1890. [93] IEC Standard, Calculation of the continuous current rating of cables. Part I: 100% load factor, Pub. n. 287, 1982. [94] R. Gretsch and G. Krost, “Moglichkeiten zur bestimmung der frequenzabhangigen netzimpedanzen,” EtzArchiv, 1979, H2, pp. 45 – 52. [95] R. Gretsch and G. Krost, “Messung der oberschwingungs-impedanzen von verbrauchern,” EtzArchiv, 1981, H1, pp. 13 – 22. [96] V. Isastia, M. Perez de Vera, and M. Scarano, “Distribuzione delle perdite nel rame nelle gabbie dei motori asincroni a barra alta alimentati con tensioni deformate,” Italian Research Council, n. 49, Naples (Italy), 1983.

146

[97] Verde Paola, Carpinelli Guido, and F. Gagliardi, “Probabilistic modelings for harmonic penetration studies in power systems,” 5th IEEE International Conference on Harmonics in Power Systems, Atlanta (USA), September 1992, pp. 35 – 40. [98] P. Caramia, G. Carpinelli, F. Rossi, and P. Verde, “Probabilistic iterative harmonic analysis of power systems,” Generation, Transmission and Distribution, IEE Proceedings, Volume 141, Issue 4, July 1994, pp. 329 – 338. [99] M. Munasinghe and M. Gellerson, "Economic criteria for optimizing power system reliability levels", The Bell Journal of Economics, vol. 10, pp. 353-365, 1979. [100] K. K. Kariuki and R. N. Allan, “Application of customer outage costs in system planning, design and operation,” IEE Proceedings on Generation, Transmission and Distribution, Vol. 143, No. 4, July 1996, pp. 305 – 312. [101] EN 50160:2007 Ed.3 "Voltage characteristics of electricity supplied by public distribution networks". [102] S..L Lau and M.R. Tamjis “The economics of power quality,” IET Power Engineer, pp. 38-41 December/January 2006/07. [103] Ming Su, Gengyin Li, Ming Zhou , and Jin Yang, “A power quality service pricing approach considering treatment expenses and stop loss insurance,” Power Tech 2007, Paper no 626. [104] D. M. Didden, “Voltage disturbances – considerations for choosing the appropriate dip mitigation device,” in Power Quality Application Guide, Copper Development Association, 2005. [105] European Information Technology Observatory (EITO), Press release 2004. [106] EURELECTRIC NE Standardisation / N-EMC & Harmonics. For PQ costs study. Considerations on voltage dips and short supply interruptions. Gerhard Bartak. [107] „Auftretenswahrscheinlichkeit von Spannungseinbrüchen verschiedener Restspannung und Dauer für Kundenanlagen in Mittelspannungsnetzen“, Diplomarbeit am Institut für Elektrische Anlagen der TU Graz, 2000. [108] A. Haber: „Analyse von Spannungseinbrüchen und mögliche Abhilfemassnahmen in einem Industriebetrieb“, Diploma thesis, TU Graz/Prof. Fickert, June 2001. [109] M. Schmidt: „Minimierung der Kundenausfallskosten unter Berücksichtigung der automatischen Wiedereinschaltung“, Diploma thesis, TU Graz/Prof. Fickert, April 2003. [111] Knut Samdal et al.: “Customers’ interruption costs – what’s the problem ?”, 17th CIRED Conference, Barcelona, 12 – 15 May 2003, Session 6, Paper No 31. [112] Norwegian water resources and energy directorate (NVE). Energy and regulation department. [113] Renner, H.; Fickert, L.: “Costs and Responsibility of Power Quality in the Deregulated Electricity Market”, TU Graz, Institut für Elektrische Anlagen und Hochspannungstechnik, Abteilung Elektrische Anlagen, 2000. [114] Copper Development Association: Power Quality Application Guide, March 2001. [115] G. Brauner, CH. Hennerbichler: “Empfindlichkeit von elektrischen Geräten gegenüber Spannungseinbrüchen und Kurzzeitunterbrechungen“, Study on behalf of VEÖ-EFG, 2001. [116] IEC 61000-4-7: Electromagnetic compatibility (EMC) - Part 4-7: Testing and measurement techniques - General guide on harmonics and interharmonics measurements and instrumentation, for power supply systems and equipment connected thereto. [117] IEC 61000-4-30: Electromagnetic compatibility (EMC) - Part 4-30: Testing and measurement techniques - Power quality measurement methods. [118] F., Porrino, A., Chiumeo, R., Malgarotti, S., 2007. The power quality monitoring of the MV network promoted by the Italian regulator. Objectives, organisation issues, 2006 statistics. In: CIRED 2007, International conference on electricity distribution, Vienna, Austria. [119] S. C. Vegunta and J. V. Milanović, Estimation of costs of Downtime of Industrial Process Due to Voltage Sags. [120] PQ measurement campaign in Austria, 2000 – 2004, VEÖ-Journal: Part I, H.11/2000; Part II, H.7/2001; Part III, H.8/2001; Part IV, H.8/2002; Part V, H.2-3/2003; Part VI, H.10/2003, Part VII, H.12/2004, Part VIII, H. 10/2004. [121] EURELECTRIC Report “Power Quality in European Electricity Supply Networks” – 2nd edition, November 2003. [122] Qualita’ del servizio di trasmissione - Rapporto annuale per l’anno 2006, 30 april 2007, www. Terna.it. [123] Power quality: Customer financial impact/risk assessment tool, BC Hydro, A04-587b, 2005. [124] IEEE recommended practice for evaluating electric power system compatibility with electronic process equipment, IEEE Std 1346-1998, 1998. [125] M. M. Izhwan, M. Norman, and M. A. M. Radzi, The effects of power quality to the industries,, 5th Student Conference on Research and Development, SCOReD, 2007.

147

[126] M. F. Han and K. Yeon-hee, Samsung Elec says outage loss not to top $54 mln, in Reuters, 2007. [127] Samsung shares lower on concerns over power outage-production loss, in AsiaPulse News, 2007. [128] R. B. Chandler, M. J. Jellison, J. W. Skogsberg, and T. Moore, NACE 02056 advanced drill string metallurgy provides enabling technology for critical sour drilling, 2002. [129]Y. Levendel, “The manufacturization of the software quality improvement process, IEEE Region 10 International Conference on EC3-Energy, Computer, Communication and Control Systems (TENCON), 1991. [130] IEC/TR 61000-3-6 (2008). Electromagnetic compatibility (EMC) - Part 3-6: Limits - Assessment of emission limits for the connection of distorting installations to MV, HV and EHV power systems [131] Applying QRA Methodology in Power System Applications, EPRI, Palo Alto, CA, 2004. [132] Analysis of Extremely Reliable Power Delivery System, EPRI, Palo Alto, CA, 2002. [133] A. G. Maitra and T. Short, “Assessing financial feasibility of PQ improvement devices using monte carlo simulations,” presented at Probabilistic Methods Applied to Power Systems, 8th International Conference, September 2004. [134] L. Conrad, K. Little, and C. Grigg, “Predicting and preventing problems associated with remote fault-clearing voltage dips,” IEEE Transactions on Industry Applications, Vol. 27, No. 1, Jan./Feb. 1991, pp.167-172. [135] L. E. Conrad and M. H. J. Bollen, “Voltage dip coordination for reliable plant operation,” IEEE Transactions on Industry Applications, Vol. 33, No. 6, Nov./Dec. 1997, pp. 1459-1463. [136] J. Lamoree, D. Mueller, P. Vinett, W. Jones, and M. Samotyj, “Voltage dip analysis case studies,” IEEE Transactions on Industry Applications, Vol. 30, No. 4, July/August 1994, pp. 1083-1089. [137] IEEE recommended practice for emergency and standby power systems for industrial and commercial applications (IEEE orange book), New York, IEEE Std. 446-1995. [138] Information Technology Industry Council: “ITI (CBEMA) curve and application note,” Washington, 1998, revised 2000, (http://www.itic.org/technical/iticurv.pdf). [139] Semiconductor Equipment and Materials International: “SEMIF47-0200, specification for semiconductor processing equipment voltage dip immunity,” SanJose, USA, 2000. (http://www.semi.org/pubs/semipubs.nsf). [140] S. Z. Djokic, J. Desmet, G. Vanalme, J.V. Milanović, and K. Stockman, “Sensitivity of personal computers to voltage dips, short interruptions,” IEEE Transactions on Power Delivery, Vol 20, No. 1, 2005, pp. 375-383. [141] R. Caldon, M. Fauri, and L. Fellin, “Voltage dip effects on continuous industrial processes: Desensitizing study for textile manufacture,” Proc. of second Int. Conf. on power quality: end-use applications and perspectives, Atlanta, USA, Sept. 28-20, 1992, pp. D-13: 1-6. [142] S. Z. Djokic, K. Stockman, J.V. Milanović, J. J. M. Desmet, and R. Belmans, “Sensitivity of AC adjustable speed drives to voltage dips, short interruptions and under-voltage transients,” IEEE Transactions on Power Delivery , Vol 20, No. 1, 2005, pp. 494-505. [143] EPRI Brief Report No. 46, “Performance of a hold-in device for relays, contactors and motorstarters,” October 1998. (http://www.portlandgeneral.com/business/products/businesspower problems/power quality school/brief46.pdf). [144] E. Collins and M. Bridgwood, “The impact of power system disturbances on AC-coil contactors,” Proc. IEEE Textile, Fiber and Film Industry Technical Annual Conference, Greenville, USA, 6-8 May 1997. [145] S. Z. Djokic, J.V. Milanović, and D.S. Kirschen, “Sensitivity of AC coil contactors to voltage dips, short interruptions and under-voltage transients,” IEEE Transactions on Power Delivery, Vol 19, No. 3, 2004, pp. 1299-1307. [146] P. Pohjanheimo, A probabilistic method for comprehensive voltage dip management in power distribution systems, PhD Thesis, Helsinki University of Technology, Espoo, Finland, November 2003. [147] S. Z. Djokic, Generalized methodology for the assessment of voltage dip performance indices and equipment sensitivity, PhD Thesis, UMIST, Manchester, UK, April 2004. (http://www.umist.ac.uk/departments/mcee/research/publications.htm) [148] M. H. J. Bollen, Under power quality problems: voltage dips and interruptions, IEEE Power Engineering Series, 2000. [149] M. R. A. Qader, Stochastic assessments of voltage dips due to short circuits in electrical networks, PhD thesis, Department of Electrical Power Engineering, Manchester Centre for Electrical Energy, UMIST, Manchester, UK. 1997. [150] IEEE Recommended Practice for Evaluating Electric Power System Compatibility with Electronic Process Equipment, IEEE Std 1346-1998.

148

[151] M. R. A. Qader, M. H. J. Bollen, and R. N. Allen, “Stochastic prediction of voltage dips in a large transmission system,” IEEE Transactions on Industry Applications, January/February 1999. Vol. 35, No. 1: pp. 152-162. [152] G. Olguin and M. H. J. Bollen, “The method of fault positions for stochastic prediction of voltage dips: A case study,” Probabilistic Methods Applied to Power Systems Conference, Naples, Italy, September 22-26, 2002. [153] M. H. J. Bollen, M. R. Qader, and R. N. Allen, “Stochastic assessment of voltage dips in large transmission networks theory and case study,” International Conference on Large Electric Networks (CIGRE), August 1998. [154] R. Grover Brown and P. Y. C. Hwang, Introduction to random signals and applied Kalman filtering, John Wiley and Sons, New York, 1997. [155] R. Billinton and R. N. Allan, Reliability evaluation of engineering systems, Plenum Press, New York, 1992. [156] J.M.C. Filho, et al., “Analysis of power system performance under voltage dips,” Electrical Power System Research, December, 1999. Vol. 55, No. 3: pp. 211-218. [157] R. Grunbaum, R. Sharma, and J. P. Charpentier, “Improving the efficiency and quality of AC transmission systems (Draft 3),” Joint World Bank / ABB Power Systems Paper, 24 March, 2000. [158] D. M. Didden, “Voltage disturbances: Considerations for choosing the appropriate dip mitigation device,” in Power Quality Application Guide, Copper Development Association, 2005. [159] A. Baggini and F. Bua, “Investment analysis for PQ solutions,” in Power Quality Application Guide, Copper Development Association: online available: http://www.cda.org.uk/pqp/pqag.htm, May, 2007. [160] A. Anderton, ECONOMICS, 4th edition: Harlow: Causeway, 2000. [161] T. L. Minnie, E. G. Gaylon, and T. Phillip, Investment analysis for real estate decisions, Dearborn Real Estate Education, 2003. [162] H. D. Keulenaer, “Power quality self-assessment guide,” in Power Quality Application Guide, Copper Development Association: online available: http://www.cda.org.uk/pqp/pqag.htm, May, 2007. [163] A. A. Groppelli and E. Nikbakht, Finance, Barrons Educational Series, Inc., 2000. [164] H. Gregersen and A. Contreras. “Economic assessment of forestry project impacts,” World Bank United Nations Environment Programme Food and Agriculture Organization of the United Nations, 1995. [165] Oury BA, Roger Bergeron, « Harmonic cost on distribution power system », EPQU, September 1999, p5-12 [166] Stavros Papathanassiou, Nikos Drossos, Dimitris Stravaopoulos, « A practical evaluation of distribution network losses due to harmonics », CIRED, Vienna 2007 [167] ANSI C84.1-2006, American National Standard for Electric Power Systems and Equipment— Voltage Ratings (60 Hertz)

149

ACKNOWLEDGMENTS Members C4.107 - Economic Framework for Voltage Quality Jose GUTIERREZ IGLESIAS Detmar ARLT Gerhard BARTAK Math BOLLEN Dave BYRNE David CHAPMAN Alice DELAHUNTY Philippe EYROLLES Elena FUMAGALLI Mats HAGER Zbigniew HANZELKA Bill HOWE Rafaël JAHN Alex McEACHERN Ian McMICHAEL Jovica V. MILANOVIC Patxi PAZOS Roman TARGOSZ Mario TREMBLAY Jasper Van CASTEREN Mathieu VAN DEN BERGH Raghavan VENKATESH Paola VERDE

Spain Germany Austria Sweden Ireland UK UK France Italy Sweden Poland USA Belgium USA Australia UK Spain Poland Canada Netherlands USA India Italy

Utility Academia Utility Research Utility Institution Utility Utility Academia Industry Academia Research Research Manufacturer Research Academia Utility Institution Research Institution Manufacturer Manufacturer Academia

Chairman C4.107 & Coord. Ch. 3

Coordinator Chapter 1 Coordinator Chapter 3

Coordinator Chapter 5

Coordinator Chapter 2 Coordinator Chapter 4

Coordinator Chapter 2 and 4

Other former C4.107 members or that collaborate by correspondence: Herivelto de Souza BRONZEADO Brasil Jhan Y. CHAN UK Philippe GOOSSENS Belgium Kurt REYNDERS Belgium Angela RUSSO Italy Helge SELJESETH Norway Gregorio VARGAS Spain

Utility Academia Utility Utility Academia Research Utility

150