Electric Power System Design

PDHengineer.com Course

№

E-5005

Electrical Electrical Powe r System Design for Industrial Facilities Facilities

To receive credit for this course This document is the course text. You may review this this material at your leisure either before or after you purchase the the course. To purchase this course, click on the course overview page: http://www.pdhengineer.com/pages/E-5005.htm or type the link into your browser. browser. Next, click on the Take Quiz button at the bottom of the course overview page. If you already already have an account, log in to purchase the course. If you do not have a PDHengineer.com account, click the New User Sign Up link to create your account. After logging in and purchasing the course, you can take the online quiz immediately or you can wait until another day if you have not yet reviewed the course text. When you complete complete the online quiz, your score will automatically automatically be calculated. calculated. If you receive a passing score, you may instantly instantly download your certificate of completion. If you do not pass on your first try, you can retake the quiz as many times as needed by simply logging into your PDHengineer.com account and Completed. clicking on the link Courses Purchased But Not Completed. If you have any questions, please call us toll-free at 877 500-7145.

PDHengineer.com 5870 Highway 6 North, Suite 310 Houston, TX 77084 Toll Free: 877 500-7145 [email protected]

Principles of Electric Power System Design for Industrial Facilities Louie J. Powell, PE Saratoga Springs, NY

Introduction Industrial facilities utilize electric power in the process of manufacturing an industrial product. The requirements of the industrial manufacturing process drive the specific requirements that the facility must satisfy. Different industrial manufacturing manufacturing processes have vastly different technical requirements for electric power, but those requirements always come down to decision points regarding a relatively small number of key considerations: workplaces are inhabited by people. people. In the early days of industrialization, industrialization, Safety – all industrial workplaces debilitating accidents and even fatalities were often accepted as part of the cost of doing business. While tragedies for the immediate families of those affected, these events were viewed as the unavoidable consequence consequence of applying technology to activities that involved involved inherent risk. Today, however, safety is considered a “prime directive” in industry, and facilities have to be designed to achieve the highest level of of protection for employees. employees. Occasional accidents still still occur, but injuries and fatalities are almost always investigated thoroughly to determine a root cause, and corrective actions are undertaken undertaken to assure that similar similar events don’t occur a second time. time. And governmental agencies such as OSHA in the US are empowered to impose severe penalties on employers who fail to properly protect employees from workplace injury.  Reliability – A critical requirement in in industry is reliability. reliability. Industrial facilities facilities are almost always designed to achieve a target production rate – X automobiles per day, Y bottles of beer per month, Z tons of finished finished paper per week, etc. Failure of the power system will cause a shortfall in production, with direct economic consequences on the owner.  Maintainability – industrial facilities facilities also have to be maintained. maintained. Maintenance is not an objective objective unto itself, but rather is required to assure that the safety and reliability designed into the system at the time of its initial construction will continue continue to be there many years into the future. And maintenance must be possible without without compromising compromising either safety or reliability. As a result, it is necessary to anticipate maintenance practices at the time that the facility is first designed.  Economy – the industrial workplace exists to produce a product, and in the vast majority of cases, those products are sold in a competitive marketplace that focuses management attention on the cost of production. Accountants have cleverly invented invented processes to amortize the cost of the facility over the expected future production, and impose intense pressure on reducing per-product cost in order to increase operating operating margins. As a consequence, every design design decision is subject to review on the basis of its impact of final cost, and only those decisions that can be supported in terms of achieving reasonable safety or operating reliability are likely to be approved by management. In this context, electric power, power, and the infrastructure required required to supply electric power, are viewed as simply elements in the cost of doing business, and in this context, it is sometime helpful to identify that there are four elements of the cost of electric power: 1. The cost of the power itself, either in terms of the price that must be paid to a commercial power supplier, or the equivalent in fuel used to generate the power within the industrial facility. 2. The amortized cost of the infrastructure required to deliver electric power to the process. 3. The cost of maintenance. © 2008 Louie J. Powell, PE

Page 1 Rev 1

4. The cost of adverse events associated with power. The cost of reliability will be discussed in more detail further in this course. But there is also a cost associated with safety – the direct costs of “lost time” accidents, the legal and other indirect costs associated with personal injury, etc. And like any other component of cost in a business, there will be pressures to find ways to reduce these costs. The Basic Requirements of the Distribution System

One of the key considerations in the basic architecture, or structure, of the power system for an industrial facility is differentiation between distribution and utilization of electric power. Distribution is the process of moving electric power from its source to the point where it is consumed. The distribution system is a utility structure within the industrial facility, just like potable water, cooling water, natural gas, lubricants, raw material supply, byproduct handling, and waste treatment are utilities. A target production rate is always specified when new plants are initially designed. That production rate typically will take into account the expected operating cycle of the plant. In continuous process industries (“stuffmakers”), the usual expectation is that the plant will operate 24/7. Paper mills and oil refineries typically operate continuously, shutting down only for major, thorough rehabilitation once every couple of years. Discrete manufacturing (“thingmakers”) typically assume a daily operating cycle expressed in terms of the number of shifts per day, and that cycle may include an inherent maintenance period. For example, in the automotive industry, a new facility might be intended to operate in normal production mode for two 10-hour shifts per day, with a 4-hour maintenance window overnight, for five days per week, to produce the target number of “jobs”, or finished automobiles, for which the plant is designed. One of the challenges faced by the designer of an industrial power system is that while a target production rate is specified at the time of initial design, the expected rate of production typically will increase over time. Part of that increase is expected to come out of productivity. That is, the expected electrical energy consumption per unit of product is expected to diminish over time as process engineers find more efficient ways to produce the product. But another part of the increase in production will inevitably come at the prices of an increased total plant electrical demand. That means that the designer of the facility has to anticipate that the electrical demand will grow over time. But the economy objective also limits how much inherent expandability the power system engineer can afford to build into the initial design to allow for process growth. And eventually, that growth in production capacity presents a need for creativity on the part of the power system engineer who will be challenged to find ways to increase the electrical capacity of  the infrastructure at minimum costs – avoiding the need to replace otherwise functional equipment, while retaining safety, reliability and maintainability. The target production rate translates directly to a capacity figure for each of the utilities that support the plant. For example, the electric power requirement in kWHr for a cement plant can be related directly to the expected output of the plant in tons/day of final product. The conversion of  required infrastructure capacity to the product output target varies from one industry to another – aluminum smelting is very energy-intensive, paper and petrochemical slightly less so, while discrete manufacturing tends to require less energy per unit of final product, etc. In addition, the conversion will vary between product lines and between manufacturers. In fact, the rate of  energy conversion, and the techniques used to achieve that rate, are typically considered proprietary information in that they directly relate to how individual companies achieve competitiveness in their respective markets. As a result, that actual target energy requirement for a proposed new facility must be obtained from the engineers who are responsible for designing © 2008 Louie J. Powell, PE

Page 2 Rev 1

the process to be embedded in that facility. However, it may be possible to use benchmark  numbers from similar, older facilities for preliminary planning purposes. One of the most important steps in designing the power distribution system for a new industrial facility is to clearly define the performance that system is expected to deliver. If the system is described first in terms of performance, then the process of design can focus on arriving at costeffective solutions to deliver that performance. That will avoid questions about “why” certain features are (or are not) included in the design. It’s not difficult to design a system that has inherent capacity for growth. The real problem faced by the power system engineer is more likely to be either justifying capacity for expansion in the initial design, or justifying the costs of modifying an existing system to accommodate load growth. These challenges are often more political than technical. When the time comes to address the issue of expansion, having a performance specification that defines the parameters that were expected for the system at the time it was originally designed will help keep the discussion about what should be done about expansion focused on performance, and help avoid pointless arguments about “what should have been known” when the plant was first designed. Fortunately for electric power engineers, the equipment used to actually construct the electrical distribution infrastructure in an industrial plant has been highly commoditized to the point where it is available in standard unit ratings. Specifically, engineers have to make rating choices with respect to frequency, voltage, thermal capacity, and short circuit level. Of these, frequency is the easiest choice to make because frequency is almost always a given based on location. For example, for industrial facilities in North and South America, the frequency is almost certainly going to be 60Hz (although a few older facilities operating at 25Hz th th and 40Hz, dating back to the late 18 and early 19 centuries, still exist), while in Europe, Africa and most of Asia, the standard frequency is 50Hz. The selection of frequency also then drives the range and selection of voltage, thermal and short circuit capacities that are available for equipment. The other component ratings must be selected as part of the design of the facility. And a major effort in that design is to make decisions about the basic arrangement, or architecture of the system. A key consideration, however, is that these component ratings and the so-called “oneline arrangement”, or architecture, are not “independent variables” that have inherent significance on their own, but rather are “dependent variables” that are driven by four key decisions pertaining to expected system performance: 1. 2. 3. 4.

Reliability Steady-state performance Performance under expected dynamic conditions Protection

The discussion that follows presents some criteria in each of these areas that have proven to be helpful in making design decisions. It must be emphasized that there is no formula that automatically guarantees a perfect industrial system design. In fact, there may be no such thing as a perfect design. Instead, design is a collection of compromises, and the criteria discussed here may be helpful in sorting through the competing objectives to arrive a acceptable solutions for individual cases. Also, while there are a number of references available to the design engineer, there is no true “standard” for the design of an industrial power system. The reference that comes closest to being a standard is IEEE 141, Recommended Practice for Industrial Power Distribution for © 2008 Louie J. Powell, PE

Page 3 Rev 1

Industrial Plants, (the “Red Book”), but in reality that reference suggests alternatives rather than one ideal solution to serving the needs of an industrial process. It also should be stated that the criteria present here are all based on fundamental electrical physics. However, in the interest of time it may not be possible to fully develop the underlying basis for each criterion.

Reliability Considerations Every component in an electrical system is a candidate for failure. There is a branch of electric power engineering that specializes in the study of component failures, assessing the statistical probability of failure together with the expected time required to repair components. Maintenance is a major consideration in this analysis, both because projected expectations of  component reliability are predicated on the presumption that those components are maintained regularly, and also because it is necessary to take equipment out of service to perform maintenance, and those maintenance outages may have an impact on the overall availability of  the electrical system. Methods are available to use this information to make informed decisions about alternate system arrangements based on achieving the highest statistical expectation of  electrical system availability. Using the analytical approach, it is possible to achieve a very high overall system availability. However, that high availability comes at the cost of significant capital investment and an aggressive maintenance program. Whether those costs are justifiable depends on the nature of the industrial process that the electrical system must support. And in particular, the key issue is how the cost of an electrical service interruption impacts on the cost of doing business. For example, in discrete manufacturing, the direct impact of an electrical interruption typically is that products cannot be manufactured during the outage. The accounting cost of the interruption is primarily the margin on product NOT produced during the outage. There is also an unapplied labor cost (employees who cannot actually do anything productive but who are still collecting wages). The impact of an electrical service interruption on a continuous process application may be more severe. Product in process may be spoiled and have to be discarded, and the cost of the raw materials, labor and electrical energy invested in that unsalable product adds to the cost of the interruption. There may also be a need to retune the manufacturing process when production resumes, resulting in even more waste product. A more insidious risk is that the electrical interruption could damage the process in some way – a good example might be a plastics extrusion process in which material freezes during the interruption, requiring that a portion of the process be rebuilt before production can resume. Finally, there is the risk that the electrical interruption could cascade into a catastrophic failure. Some chemical processes have inherent instabilities during startup and shutdown, and an uncontrolled hard stop due to a power interruption could lead to an explosion. Or, if electric power is required in the process to maintain environmental controls, an unexpected interruption could result in release of hazardous materials with both legal and economic consequences. While an analytical approach to assessing reliability is intellectually interesting, it may not be  justifiable in every instance. In fact, most industrial facilities can be adequately designed using the n-1 concept. The n-1 concept. © 2008 Louie J. Powell, PE

Page 4 Rev 1

n-1 is the notion that every component of an electrical system is a candidate for failure, then the system must be designed in a way that allows it to deliver power to all important loads assuming the failure of every component. Obviously, the likelihood that multiple components will fail simultaneously is remote, so practical designs employ redundancy in a way that allows individual components, or perhaps logical groups of components to fail, and still provide power to important loads.

Fig 1 – Simple single-ended substation Fig 1 depicts a single-ended substation consisting of a primary switch, a transformer, a secondary breaker, and a load bus. Failure of the primary switch, transformer or breaker results in an inability to supply power to loads served from the bus. A simple, n-1 solution to enhancing the reliability of this design is shown in figure 2.

Fig 2 – Simple Double-Ended Substation In this arrangement, any switch, transformer or breaker could fail, and power could still be supplied to loads on the critical bus. There are three important assumptions embedded in this simple example. One has to do with the system that supplies power upstream of the primary switch. Ideally, the concept of redundancy would extend as far back into that system as practical. That suggests that the double-ended arrangement should also be applied to whatever primary system supplies this substation, and in fact most industrial system do just that. Obviously, however, there is a limit to how far that redundancy can go. For example, it would be very unusual for circumstances to provide two completely independent sources of power in one geographic location. The second assumption is that while each component has its own inherent failure characteristic, practical designs group components. So while the transformer, the connection between the primary switch and the transformer, and the connection between the transformer and the secondary breaker are actually independent, for design purposes they are considered to be a group. Failure of any one is addressed through redundancy of the entire group. © 2008 Louie J. Powell, PE

Page 5 Rev 1

Third, it is assumed that the load bus itself is not subject to failure, and of course that’s not a realistic assumption. Hence, a more practical design might be the arrangement shown in fig 3.

Fig 3 – Double-Ended Substation with a Tie B reaker In this instance, installation of a tie breaker in the load bus allows a failure to occur on one half of  the load bus without affecting the other half. Of course, that then raises two additional questions. First, electrical failures involve arcing, and arcs generate a lot of combustible and potentially conductive gas. Therefore, it is possible that a failure on one half of the load bus might actually  jeopardize the other half in spite of the fact that the breaker is present. Second, consideration also has to be give to the possibility that the tie breaker itself could fail. Addressing these two concerns leads to the arrangement of figure 4.

Fig 4. Double-Ended Substation with Dual Tie Breakers in Separate Vaults In this design, the single time breaker has been replaced with two breakers, and the two halves of  the double-ended substation have been placed in separate rooms, or vaults (represented by the dashed line), so that an ordinary electrical failure in one cannot communicate to the other. This example show how an intuitive approach to reliability, considering the n-1 principle, can evolve into an increasingly complex, and therefore increasingly costly, system design. As a practical matter, the arrangements in figures 1 and 2 are relatively typical of electrical system designs supporting commercial applications – offices and shopping, for example. Figure 3 is typical of what might be found in a great many discrete manufacturing and even continuous process industrial applications, while figure 4 might be reserved for a “mission critical facility” such as a data center or a hospital surgical suite.

© 2008 Louie J. Powell, PE

Page 6 Rev 1

Fig 5 – A Double-Ended Medium Voltage bus feeding a double-ended low-voltage substation The double-ended substation shown in figures 2-4 is very typical of low voltage1 substations in industrial or commercial applications. Double-ended arrangements are also very commonly seen 2 in medium voltage systems. Figure 5 shows a typical double-ended medium voltage industrial substation that also feeds multiple double-ended low voltage substations, and illustrates how the concept of double-ending and redundancy can be extended upstream from the load toward the source of electric power. The designs shown here call for only one layer of redundancy. This is not a casual decision, and in fact is based on an important principle that can be found through a more analytical approach to reliability. That principle is that the amount of improvement caused by redundancy decreases dramatically as the number of layers of redundancy increases. Redundancy 10000000 1000000 100000

     F      T      T      M

10000 1000 100 10 1 1

2

3

4

Number of Transformer s

Fig 6 – Improvement in Reliability due to Multiple Transformers in Parallel Figure 6 shows the calculated mean-time to failure for a hypothetical system served by one, two, three or four transformers, and shows that adding a second transformer makes a dramatic improvement in calculated mean-time-to-failure, but the improvement associated with a third or fourth transformer is inconsequential. 1

Under ANSI standards, “low voltage” refers to applications operating at 1000 volts or below. Typical low voltage industrial applications are at 480v. 2 Under ANSI standards, “medium voltage” refers to applications operating at voltages greater than 1000 volts but lower than 72kV. Typical industrial medium voltage applications include 4.16kV, 13.8kV, and (less frequently) 34.5kV. © 2008 Louie J. Powell, PE

Page 7 Rev 1

Most industrial applications are therefore served using designs that call for only one layer of  redundancy. Only in very usual cases (data centers, ethylene plants etc) is there a justification for additional redundancy, and in those instances that additional redundancy is most commonly in the form of local emergency generation that provides a source of energy that is independent of a commercial utility supplier. There are two special arrangements that need to be mentioned here for completeness. One is the low voltage sparing arrangement. While the majority of industrial applications utilize simple double-ended low-voltage substations, a special concern exists in applications where the concentration of load is heavy enough to require multiple substations to meet thermal requirements. In those situations, a design that originated in the automotive industry may have a more attractive initial cost.

Fig 7 – Low voltage sparing arrangement In figure 7, one single-ended low voltage substation (shown in blue) has been set up to spare four other single-ended substations. In theory, this arrangement could be extended to any number of  other substations if there is sufficient load in the vicinity to justify additional capacity. In this arrangement, the breakers connecting the sparing bus to the individual load buses are normally open in order to keep the short circuit levels within reason. In some industries, most notably pulp and paper but also petrochemical, there is a need to integrate local generation. This generation is typically integrated into the thermal cycle of the process (it is classic “cogeneration” in that steam is produced that is used to both generate electricity and to perform some function in the process), so there is a desire for the electrical output of the generator(s) to be utilized at a point in the process that is close to where the steam is utilized; in that way, the impact of a maintenance shutdown is localized to the one area immediately associated with both the steam and electrical loads on that boiler. The problem that often occurs with generators is that their presence c auses a dramatic increase in the short circuit current available on the system, so an arrangement is require that permits the generator(s) to be integrated while providing control over short circuit duties. The answer to this dilemma is the medium voltage “synchronizing bus” design.

© 2008 Louie J. Powell, PE

Page 8 Rev 1

G

Fig 8 – Medium Voltage Synchronizing Bus Arrangement The synchronizing bus arrangement requires more careful consideration during the design phase, and in particular more attention to concerns relating to steady-state loading, motor starting, short circuit levels and system protection. But is can also be a very powerful tool in addressing design challenges in system with cogeneration integrated with the process.

Steady State Performance Considerations The power system in an industrial facility is largely passive, so it is important that careful attention be given to several aspects of steady state operation. The most important is that the thermal ratings of the components must be adequate to handle the load that will be imposed on the electrical system by the process. And here the challenge of  predicting the future is important. It’s not at all unusual for a facility to be commissioned one year, and then have receive a management request to operate at 115% of design capacity one year later. And that’s just the beginning. All electrical components have a fundamental thermal rating, usually expressed in terms of  continuous amperes or kVA. That rating describes how much power can flow through the component without exceeding a design temperature rise in an environment where the ambient temperature falls within the constraints defined by the standards against which the component was designed. It is important to recognize that the thermal stress on electrical system components is related to the current flow through the electrical conductors making up those components. For convenience, thermal ratings are expressed either in amperes or kVA; in the case of kVA rating, there is an implied assumption of rated voltage. Occasionally, one will see a rating expressed in kilowatts (or some multiple of kW). Most often this is associated with generators as is based on the mechanical output of the prime mover. One should take care to use the actual kVA rating of  the generator rather than the kW rating of the prime mover. In most industrial applications, loads are close enough to constant that one need not be concerned with intermittent loading, thermal cycling and related issues. If those issues do arise, guidance can often be found in standards. In extreme cases, it may be necessary to define the thermal duty as part of the specification that the component manufacturer is expected to meet, and rely on the manufacturer’s understanding of the fundamental physics of his equipment to offer assurances that their proposed design is adequate.

© 2008 Louie J. Powell, PE

Page 9 Rev 1

The fundamental principle is that loading should never exceed the rated thermal capacity of  individual system components under steady state conditions. One does not normally have to be concerned about exceeding continuous thermal ratings under abnormal conditions such as motor starting or other impact loading, or short circuits, unless these conditions occur frequently enough to be considered a cycling load. Some electrical components, most notably transformers, may be equipped with supplemental cooling provisions that increase the continuous thermal ratings of the basic component carcass. For example, outdoor oil filled transformers typically have a fan-cooled rating that is 30 higher than the base rating. Other transformers have other force-cooled ratings, and it is appropriate to consult both manufacturer’s specifications and standards to determine what the actual forcecooled ratings of a specific transformer will be. Earlier, the double-ended substation (figure 2) was suggested as a reliability enhancement over the single-ended substation (figure 1). If all of the components have the same rating, then that double ended enhancement would have a cost of approximately twice that of the single-ended design, and an initial assessment is that the incremental reliability comes at a price premium of  100%. Fortunately, the ability to manipulate thermal ratings significantly changes that assessment, and also offers the power system engineer a number of design options. Consider the case of a low voltage unit substation that is designed to carry a projected initial load of 800 kVA. There are several potential ways to serve this load: o

Solution A: Single-ended substation, 1000 kVA transformer rating This solution provides 20% margin for future growth, but with a singlecontingency failure risk.

o

Solution B: Double-ended substation, 750kVA transformer ratings The total capacity of the two transformers is 1500kVA. Therefore, not only does the design provide enhanced (n-1) reliability, it also has the ability to accommodate 87% load growth. Since cost is approximately proportional to the installed kVA rating, it will cost about 50% more than A.

o

Solution C: Double-ended substation, 500kVA transformers with 50% force-cooled capability This design offers identically the same loading and reliability performance as B, but it will cost only slightly more than A. The premium will be associated with the addition of forced cooling, and the incremental installation cost of the second unit.

o

Solution D: Double-ended substation, 750kVA transformers with provision for future 50% forced-cooled capability. This design offers enhanced reliability, an initial ability to accommodate 87% load growth and at a cost that is only slightly more than solution B. But it also offers the ability to expand the load capability to 281% of the initial requirement with the addition of relatively inexpensive accessories and with no additional footprint requirements in the future.

© 2008 Louie J. Powell, PE

Page 10 Rev 1

In this example, it was assumed that the transformer(s) would be the most costly components of  any design, and that ratings could be chosen for other components (switches, low voltage breakers, cables, etc) that could accommodate any of the rating options available with transformers. In an actual example, of course, it would be appropriate to test these assumptions. Voltage spread

The most common load served by the power system in an industrial facility is motors. Motors typically have voltage ratings that are slightly lower than the nominal ratings of the corresponding voltages in a system. For example, motors designed for application on a 480 volt system commonly have a 460 volt rating. The torque produced by an induction motor, the most common variety used in industry, is proportional to the square of applied voltage. So if a 460 volt motor is applied on a system where the applied voltage is 480v, the motor will produce about 109% rated torque. Or conversely, the motor will be able to produce rated torque with up to 4% voltage drop in the feeder cables. It is not necessary that motors always produce 100% rated torque. But it is necessary that motors produce more torque than is required by the loads that they are driving. Fortunately, most motors are oversized for their mechanical loads. But it is still incumbent on the power system engineer to design a distribution system that will deliver a reasonable voltage to the terminals of motors during all steady state operating conditions. A design “rule of thumb” that has worked reasonably well is that systems generally can withstand sustained voltages down to about 95% of rated during “normal” operations. In this context, “normal” means either the normally intended mode of operation, or one of several modes that might have been envisioned as a possible mode of operation at the time the system was designed. Thus, for example, in the example above, solution C involves a potential future load of 1500kVA on two paralleled transformers, each with a base rating of 500kVA. Therefore, it would be necessary that the system deliver at least 95% rated voltage at the terminals of motors for this heavy load condition for the design to be considered adequate. Figure 9 shows this condition in a power flow solution. In this instance, the load was represented as one 100 horsepower motor plus an aggregation of other loads totaling 1393kVA at 93% power factor. The voltage at the 480v bus with rated voltage on the. © 2008 Louie J. Powell, PE

Fig 9 – Power flow solution for a fully-loaded double ended substation. 3

3

See appendix I for an explanation of the information in this figure. Page 11 Rev 1

transformer primaries were depressed to 95% of rated, the voltage delivered to the secondary would be 89.2% of rated, well below the threshold of acceptability. This illustrates a critical point. While it may be possible to perform much of the analysis required to determine the thermal ratings of system components using a simple spreadsheet, it is not possible to gain complete insight into the performance of a power system without using a true power flow model. That is especially true if the design involves significant lengths of feeder cables or a medium-voltage synchronizing bus design. The reason for this is that motors and other actual power system loads are non-linear. Motors have a constant kVA characteristic in response to changes in bus voltage. As a result, increases in load tend to amplify voltage drop. The only accurate way to assess these considerations is by computer simulation that can deal with the nonlinear equations required to describe the performance of those loads One solution to the possible problem presented in this example is to take advantage of the voltage taps that are usually present on transformers. Moving the transformer taps to the 2.5% boost position would cause the secondary voltage in this example to be a completely acceptable 97.3% with rated voltage on the primary. With the primary voltage depressed 5%, the 2.5% boost tap would elevate the secondary voltage to 91.8% of the nominal 480 volt value. However, that would equate with 95.7% of rated motor voltage leading to about 92% rated torque. It might be possible to arrive at a conclusion that if the combination of reduced primary voltage and heavy loading is not likely to occur frequently enough to worry about this being unacceptable performance. . Voltage spread is primarily a function of  reactive power flow through system inductive   0   0 .   @ reactance. That can be demonstrated quite   0 BUS-1   0  0   1 . dramatically by adding a capacitor bank to the system simulated in figure 9. This simulation shows that a 500kVAR capacitor bank causes the voltage at the 480v. bus to be higher. The reactive demand of the 480v loads is unchanged; 13.800 13.800 because the capacitor bank supplies some of that 0.480 0.480 reactive locally (at the 480 v bus), it is not necessary for the entire reactive demand to be supplied from the primary, thereby reducing the    5  . amount of voltage drop experienced in the  -  4   @    7  4 BUS-2 transformers  .  9         7         1         5   .         7         0         4         1

        )         5         4         8   .         1         8         1         (

        9         5         7   .         3         0         7

        )         3         2         9   .         0         9         (

        9         5         7   .         3         0         7

        )         3         2         9   .         0         9         (

        3         4         5   .         0         9         6

        )         3         4         5   .         4         3         (

        3         4         5   .         0         9         6

        )         3         4         5   .         4         3         (

  0

It has been said that “vars don’t travel well”. That’s a profound consideration. Designing the power system in a way that provides local reactive sources helps improve the overall voltage profile of a system.

1300.000 (500.000)

0.000 (474.680)

81.087 (43.766)

Fig 10. Power flow solution with the addition of a 500kVAR capacitor bank 

© 2008 Louie J. Powell, PE

Page 12 Rev 1

If inductance and vars determine voltage drop and the magnitude of voltage, then resistance and real power determine the angle of the voltage phasors. Voltage angle is normally not a concern in the design of industrial power systems. Steady-state supply and management of reactive power is also an economic consideration. Must electric utilities impose a penalty on industrial customers with low power factor. That penalty may be expressed directly in terms of power factor, or it may be formulated in other ways. One of the more subtle approaches is to define demand limit in terms of total apparent power (MVA), while setting the power consumption measurement on real power component of consumed energy (kilowatthours). Regardless of how it is done, it is important that the process of designing the industrial system anticipate the need to manage reactive power demand on the host utility. Ordinary capacitor banks are a relatively inexpensive way to supply and manage steady-state reactive power, and with many typical reactive penalty arrangements, the economic payback is less than one year. The concern that is usually most vexing is designing the system to deliver at least 95% of rated voltage everywhere and at all times. However, the designer also must be sensitive to the fact that the sustained voltage cannot be too high. It’s not unusual for system to operate with elevated voltages, with a practical upper limit usually considered to be about 103% of rated. That is, the acceptable spread of steady-state voltage is commonly 95% to 103%. Incidentally, a question that is often debated is what is the optimum location of power factor capacitors. The answers are: If the power factor capacitor is needed ONLY to reduce the reactive demand at the point of metering, then the optimum solution is to install one capacitor bank on the load-side of  the revenue meter. If power factor capacitors are needed to address steady state voltage spread and to relieve thermal loading in cables and transformers, then the optimum solution is to apply capacitors at the terminals of loads and arrange for them to be switched on and off with their associated loads. It should be noted, however, that there are limitations in the amount of capacitance that may be switched in combination with motors, and those limitations may necessitate the addition of other capacitor banks in order to achieve the overall power factor improvement target for the system. If there are sources of harmonic currents in the system such that there is a need for the capacitors to be configured as harmonic filters, then the optimum solutions will usually be to provide the smallest number of capacitor banks possible. And there are exceptions to each of these general rules. 







In recent years, industry has become much more concerned about efficiency and losses, and operating economy clearly needs to be taken into account in designing industrial power distribution systems. That said, there may not be a lot that can be done with the power distribution system itself to enhance operating efficiency. The fact is that the components that make up the power system (mainly switchgear, transformer and cables) are quite efficient and are responsible for only a very small fraction of the losses incurred by typical industrial systems. Most losses are incurred in motors, and the greatest opportunity for improving overall operating efficiency is in the way that motors are selected to meet individual process requirements. Certainly, the opportunity exists to select between traditional motor designs and those with reduced losses. In the late 1970’s, when manufacturers started offering higher efficiency motors, many industrials were tempted by the promise of economic savings. But when they did the analysis, they found that the actual savings that would result from switching from existing © 2008 Louie J. Powell, PE

Page 13 Rev 1

standard motors to higher efficiency motors of the same rating would not pay for the cost of  retiring the older machines and buying new motors. However, the use of higher efficiency motors does appear to make economic sense when new systems are being built. A more important consideration is that the motor must be sized properly for the load. Motors are most efficient when they are operated at a point fairly close to their electrical ratings, and motors that are oversized relative to the actual mechanical requirements of the load they drive will incur higher energy losses. The principle that seems to be involved here is that the mechanical engineer responsible for the driven load adds a margin when specifying the brake-horsepower requirement of the load. Then, the application is handed of to an electrical engineer who adds an additional margin to determine the theoretical rating of the motor. But because motors come in standard frame sizes, the motor that is actually chosen will be larger than the theoretical rating.. The result is a process that almost certainly will lead to oversized motors. Obviously, a nonparochial design process in which there is a single assessment of margin and economy would help avoid this kind of problem.

Dynamic Performance Considerations The third major set of considerations for the designer of an industrial electrical distribution system relate to system performance during or in response to changes in dynamic changes on the system. For typical industrial systems there are two general sets of concerns. One relates to motor starting or other impact loads. The second is the pervasive concern for short circuits. Impact Loading

Impact loading is a generic term for the concerns associated with sudden changes in electrical load. The most common example of impact loading is motor starting, but there are other fairly common examples in industry. In the mining industry, draglines have a dramatically cyclical characteristic. Real o power peaks as the shovel is digging into the bank, or “crowding” the bucket back  toward the cab, but then drops and may actually reverse direction as the bucket is emptied. Reactive power also goes through a cycle that may be a bit less extreme, but that may actually be more significant because of the effect on voltage. In the metal rolling industry, the power input to a rolling mill increases sharply when o the raw stock enters the mill. Again, the reactive swing may be of far greater concern than the real power impact. In one rather notorious instance, the reactive demand of  a slab mill increased 70MVAR in a few seconds. The load associated with dynomometers (in the automotive industry) can swing o between net-motoring and net-generating. o Even some continuous loads have cyclical nature. Reciprocating compressors are notorious for applying a pulsating demand on the electrical system. The major concern with impact loading is for how the load affects voltage, and of course the load that is of interest is the impact load. One way to quickly approximate the effect of a change in reactive demand is equation [1].

Voltage change (%) ≈

© 2008 Louie J. Powell, PE

∆ MVAR

 MVAshort circuit

× 100

[1]

Page 14 Rev 1

In this equation, MVAshort circuit is the short circuit “stiffness” at the point where the impact load is applied, and the voltage change is relative to the steady-state voltage that existed prior to application of the impact. So, for example, a step change of 10MVAR would lead to a 2% 4 voltage change on a system with a 500MVA short circuit stiffness . If a more exact numerical assessment is required, then it is again necessary to employ a system power flow model to simulate the conditions that exist before and after the application of an impact load. More commonly, the concern is for starting of motors. The power factor of motor starting current tends to be in the range of 15-20%, so the initial demand of the motor is predominantly reactive. And the magnitude of the starting current is several times the rated full load current of the motor. Typical, traditional induction motors have starting currents that are about six times normal full load current, while the starting current of higher-efficiency motors can be as high as 8.5 times full load. Synchronous motors tend to have lower starting currents – perhaps four to five times full load. Special “soft start” motors can be specified that have starting currents as low as three times full load. And, of course, motors that are served through electronic power converters (typically, but not always, for variable speed applications) have significantly lower starting impacts. Using equation [1], it is possible to arrive at a quick estimate of the voltage dip that will accompany starting of an induction motor. For the sake of consistency, consider the system depicted in the power flow solution in figure 9 consisting of two 500kVA transformers supplying a 100 horsepower motor. The short circuit stiffness can be approximated by dividing the transformer kVA by the transformer impedance. Standard low voltage unit substation transformers have an impedance of  5.75% on their own base, so

Stiffness =

2 × 500 0.0575

= 17,391 kVA = 17.4

MVA

[2]

When starting, a typical 100 horsepower motor will impose an impact load on the system of  ∆Load =

6 × rated kVA

=

600 kvar

=

0.6 MVAR

[3]

Therefore, from [1], the voltage change will be

Voltage change (%) ≈

0.6 17.4

× 100 ≈

3.44%

[4]

Another way to arrive at an estimate of the voltage impact from motor starting is by performing two successive power flow simulations, one without the motor, and the other with the motor represented by its starting impedance.

4

In the past, it was routine to express short circuit levels in terms of MVA. Today, circuit breaker standards define breaker short circuit ratings in amperes, and it is necessary to calculate short circuit levels in compatible terms. However, it remains a convenience for system planners to think in terms of the equivalent short circuit MVA level, where MVAshort  circuit  = kAshort  circuit  × kV  × © 2008 Louie J. Powell, PE

3. Page 15 Rev 1

Before start         8         9         1   .         8         2         3         1

BUS-1

After start

        )         0         0         3   .         0         2         6         (

  0   0 .   @   0  0   0  .   1

  0   0 .   @   9   9  9   0 .

BUS-1

        9         9         0   .         4         6         6

        )         0         5         1   .         0         1         3         (

        9         9         0   .         4         6         6

        )         0         5         1   .         0         1         3         (

        9         3         9   .         9         9         6

        )         0         7         6   .         7         9         5         (

        9         3         9   .         9         9         6

        )         0         7         6   .         7         9         5         (

        0         0         0   .         0         5         6

        )         0         0         0   .         0         5         2         (

        0         0         0   .         0         5         6

        )         0         0         0   .         0         5         2         (

        6         7         6   .         7         7         6

        )         1         9         6   .         2         0         5         (

        6         7         6   .         7         7         6

        )         1         9         6   .         2         0         5         (

BUS-2

  0  .  -  4   @    5  0   9  .   0

BUS-2

  9  .  -  3   @   6   9  1   0 .

1219.420 (469.008)

1300.000 (500.000)

135.932 (536.374) MSHp = 100 MSPF = 12%

Fig 11 – Steady-state power flow simulations of a system before and after starting a motor Note that the simulations show that starting the motor will cause the voltage to dip from 95% of  rated to 91.6% of rated.- a 3.58% dip. Note also that this answer is very close to that calculated by the quick approximation formula [1]. The problem of motor starting is a special case of the more general problem of impact loading. Generally, the major issue involved in impact loading is the increase in reactive loading, although there are instances in which changes in real power loading can be just as disruptive. It was noted in the discussion of steady state loading that addition of power factor capacitors can be an effective means to address voltage spread problems. That leads to the question of whether power factor capacitors might also help with impact loading. The answer, in general, is that capacitors are not helpful by themselves. To illustrate that point, consider the example presented in figure 11, but with a 500kVAR capacitor bank applied on the 480v bus.

© 2008 Louie J. Powell, PE

Page 16 Rev 1

Before starting         3         6         1   .         3         2         3         1

BUS-1

After starting

        )         9         7         9   .         9         1         1         (

  0   0 .   @   0  0   0  .   1

        1         8         5   .         1         6         6

        )         9         8         9   .         9         5         (

        1         8         5   .         1         6         6

        )         9         8         9   .         9         5         (

        0         0         0   .         0         5         6

        )         0         8         5   .         0         1         (

        0         0         0   .         0         5         6

        )         0         8         5   .         0         1         (

BUS-2

1300.000 (500.000)

  2  .  -  4   @    7  9   9   0 .

  0   0 .   @   9   9  9   0 .

BUS-1         8         6         3   .         4         9         6

        )         6         0         5   .         3         6         3         (

        8         6         3   .         4         9         6

        )         6         0         5   .         3         6         3         (

        3         2         2   .         8         7         6

        )         7         2         6   .         4         9         2         (

        3         2         2   .         8         7         6

        )         7         2         6   .         4         9         2         (

BUS-2

1217.121 (468.123)

0.000 (478.838)

  1  .  -  4   @   3   9  4   0 .

0.000 (444.417)

139.324 (565.548) MSHp = 100 MSPF = 12%

Fig 12 - Steady-state power flow simulations of a system before and after starting a motor with a capacitor bank supporting system voltage This example shows that the starting the motor will cause the voltage to drop from 97.9% of rated to 94.3%, or a change of 3.68%. There are two important points to note here. First, the presence of the capacitor elevates both the pre- and post-impact voltage by about the same amount compared with the example of figure 11. That is, the effect of the capacitor bank is to change the baseline from which the dip occurs, More significantly, the presence of capacitors appears to actually amplify the magnitude of the voltage dip – from 3.58% to 3.68%. This is not a numerical oddity, but rather is an actual consequence of the physics of the electrical circuit. A power factor capacitor is passive shunt device –an impedance. The reactive flow in the capacitor is directly proportional to the square of  the voltage applied across the capacitor – and as the bus voltage is depressed by the impact load of the starting motor, the capacitor will be less able to provide reactive support of voltage. Hence, while a capacitor bank is most helpful in addressing steady-state voltage profile issues, it may actually aggravate impact-related voltage issues. Instead, dynamic source of reactive support is required to address voltage dips associated with impact loading. A dynamic reactive source is one that provides progressively greater reactive support as the bus voltage becomes lower. The simplest example would be to associate a © 2008 Louie J. Powell, PE

Page 17 Rev 1

switched capacitor with the starting motor – a capacitor that is applied at the instant that the motor contactor is closed, and that is subsequently removed from the system after the motor accelerates to rated speed. Synchronous motors are also effective as dynamic sources of reactive support. Alternatively, it is also possible to purchase solid-state reactive compensators that can react very quickly to changes in reactive loading. Transient voltage dips are an unavoidable consequence of step loading and motor starting. Before opting for a solution to transient voltage disturbances, it is i mportant to know whether there is really a problem. There are really two separate issues involved in answering that question. Like steady-state voltage spread, transient depressions of system voltage cause motors to produce less mechanical torque. Whether that is actually a problem really depends on how long the transient voltage depression lasts. In most instances, a voltage dip that lasts only a few cycles of  time, or perhaps even a few seconds, will not cause any problems with operating motors. However, there is one very notable exception to that generalization. The reduction in torque can cause some slow-speed mechanical systems to slow down sufficiently that there is a significant difference between the phasor angle of the residual magnetic field in the motor and the system voltage, and this can result in significant torque transients when the voltage dip is corrected. The most notable examples of this problem are very slow speed applications such as autogenous grinders in the mining industry and cement kiln drives. The other issue is that the instantaneous magnitude of voltage drops low enough to cause some kind of magnetically-latched control device to drop out. The most traditional example of this is that motor starter contactors will open if the applied bus voltage drops too low. The exact threshold is a variable, but the traditional rule of thumb is that limiting the instantaneous magnitude of voltage to no less than about 75% of rated should avoid contactor dropout. On that basis, it is common to limit the lowest voltage under motor starting conditions to be no lower than 80% of rated system voltage. Computers and PLCs also have an undervoltage drop-out characteristic, but they may be more sensitive to voltage dips than magnetically latched contactors and may require special provisions to assure ridethrough. Short Circuits

Short circuits (faults) are relatively common occurrences on electric power systems. The currents that flow into a short circuit are much higher than normal load currents. As a result, a fault can be a destructive event that results in a significant energy release at the point of the fault and doing considerable damage to the components that make up the power system. The primary means of  protective against short circuits is by means of protection devices – sensors and relays in combination with circuit breakers. As the magnitude of the potential short circuit currents that a system can produce increases, it is necessary to resort to larger (and therefore more expensive) circuit breakers that have the rated capability to interrupt those higher currents. Every power system must be equipped with fault interrupting devices that are intended to automatically open in the event of short circuits. And under the National Electric Code, those fault interrupting devices must be rated to interrupt the short circuit currents that the system is capable of producing.

© 2008 Louie J. Powell, PE

Page 18 Rev 1

In addition, in recent years engineers have become much more aware of the potential hazard to individuals in the vicinity of a short circuit. Not only does the arc associate with a short circuit throw off a significant amount of radiant heat, but there are also concerns with the concussive impact of a fault, the audible noise that accompanies the short circuit, and the possibility of  shrapnel being projected away from the fault point. As noted earlier, process load requirements dictate the size of motors and other cyclic loads, and these in turn determine the minimum short circuit stiffness that the system must have to be able to cope with those loads. But the desire to choose lower-rated (and therefore less costly) switchgear, combined with the desire to minimize arc flash hazards, are factors that limit the maximum short circuit stiffness. So designing a system to address short circuits is a challenge in balancing these competing objectives. There are a number of factors that have a bearing on the available short circuit current – the presence and ratings of motors and generators and the voltage are significant, but the ratings of  transformers is probably the most important. Ultimately, the magnitude of available short circuit current comes down to the Thevenin equivalent driving point impedance at each point in the system – and lower impedances mean higher short circuit currents. Since the actual impedance of  a transformer is primarily determined by its rating, then the ratings of the transformers supplying a system are critical. But transformer rating is also closely tied to the load to be served by the system. That leads to an important but seldom documented observation: in an ideal system, the available short circuit will be in the neighborhood of 25 times the maximum load to be served by that system. While this ratio can be expressed in terms of current (amperes), practiced system designers more typically talk in terms of MVA in order to eliminate system voltage from consideration. So, for example, if the load that must be served is 10MVA, one should design the system for a maximum available short circuit level of about 250MVA. There are two important implications from this observation. The first is that the magnitude of  load drives the selection of transformer ratings, and the available short circuit level is a consequence of that selection. That is, one generally does not design a system to achieve a desired short circuit level. Instead, one designs a system to serve identified loads, and then has to deal with the short circuit level that is necessary to support those loads That observation in turn leads to several conclusions about how loads should be served. First, with regard to unit substations, there is an advantage to design for a larger number of smaller substations instead of combining all loads on a small number of larger substations. That is because substations built around transformers with lower ratings have lower available short circuit currents. Obviously, there are both economic and real estate penalties in using a larger number of substations, so the designer is faced with a tradeoff. Second, as discussed earlier, the ability to start motors is a major factor driving the required short circuit level (and hence, the required rating of substation transformers). While the size of  individual motors is driven by process non-electrical process considerations and generally has to be accepted as a given, the power engineer may be able to influence the choice of starting means for those larger motors. Motors applied through power converters (as adjustable speed drives) have a significantly lower impact upon starting that do single-speed motors with across-the-line start. In addition, in many instances the ability to throttle back the speed of the drive translates

© 2008 Louie J. Powell, PE

Page 19 Rev 1

into savings in energy consumption. Therefore, applying larger motors through adjustable speed power converters may be an attractive option. There are also other options for reducing the starting impact of larger motors. Motor starting reactors and autotransformers are an old solution that still works quite well. On very large motors (typically 10,000 HP or larger), it is not uncommon to see the motor applied on a dedicated “captive transformer”. That arrangement transfers the starting impact of the motor to a higher voltage bus that almost certainly will have a higher available short circuit level, while also reducing the magnitude of the impact. Obviously, these solutions require capital investment and real estate, so there selection has to be part of an overall trade-off evaluation. The second major observation derived from the “25:1” rule is that because the available short circuit level is a consequence of the load that must be served, the ideal system voltage is also determined by that load. Under ANSI standards, the availability of short circuit ratings in medium voltage switchgear is tied to voltage level (Table 1), and it may be necessary to resort to a higher voltage in order to select commercially available switchgear for a desired fault level. Table 1 – Typical Short Circuit Ratings of Medium Voltage Circuit Breakers under ANSI Standards Short Circuit Rating

5kV Class (4.16kV)

250 MVA 350 MVA 500 MVA 750 MVA 1000 MVA 1500 MVA

Voltage Rating 7.2kV Class (6.9kV)

15kV Class (13.8kV)

X X X

X X X X

So, for example, if the load to be served is 10MVA, then the optimum fault level is 250MVA, and 4.16kV might be a good voltage choice. But if the load to be served is 15MVA, the optimum fault level will be 375MVA and one could choose between 6.9kV and 13.8kV. There are a few special voltage considerations worth mentioning: 2.4kV used to be a relatively common system voltage, especially in the pulp and paper o industry. Today, that voltage is generally viewed as obsolete, and it is difficult to purchase switchgear designed specifically for 2.4kV. However, 2.4kV may still be an economically appealing voltage for motors, and could be a practical motor voltage when the motor is applied through a captive transformer served from a higher-voltage bus. 4.8kV is occasionally seen as a distribution voltage. Under ANSI standards, 4.16kV o switchgear has a maximum voltage rating of 4.76kV and may not be applied at 4.8kV. Therefore, 4.8kV requires at least 7.2kV class switchgear; most typically, 15kV switchgear is used. It is generally not an ideal voltage for industrial distribution. The technical and economic tradeoffs involved in motor design result in the situation o where the optimum machine voltages tend to range from 2.4kV through 6.9kV although higher voltage ratings are seen on very large machines. While 6.9kV might appear to be an ideal voltage for a system serving a large population of motors, that voltage has not been well supported by switchgear manufacturers. 12.47kV and 13.2 kV are voltages used by distribution utilities in some areas. Typically, o these voltages are only used in industrial systems when those systems are being fed © 2008 Louie J. Powell, PE

Page 20 Rev 1

directly from the distribution utility at those voltages and without intervening transformers.

Protection Considerations The final aspect of system design is protection. The ultimate test of a design is whether it can be easily and economically protected using simple, commercially available protection equipment. There are four fundamental objectives that must be met by the protection system: 1. Economy – protection must be achieved at the lowest practical cost. A rule of thumb is that the protection system should contribute about 10% of the total cost of an industrial electrical distribution system (and the electrical distribution system itself should cost about 10% of the total capital investment in the industrial facility). 2. Reliability – the protection system must work as expected, every time, even though the system may go for years with no events that require action by the protection system. 3. Selectivity – while the protection system must isolate portions of the system that have failed, but the portion is isolated must be as small as practical. 4. Speed – the protection system should be as fast as possible while achieving the objectives of reliability and selectivity. Of these, economy, reliability and selectivity can generally be achieved by applying traditional, well-tested protection solutions. These solutions are well documented in the technical literature, and references such as IEEE 241 – Recommended Practice for Protection and Coordination of  Industrial and Commercial Power Systems (the “Buff Book”) are highly recommended. Speed is another matter. Speed of the protection system is primarily a consequence of the design of the system in combination with economy, reliability and selectivity. Hence, the speed of  the protection system itself is a measure of the quality of system design. The most common form of protection in an industrial system involves time overcurrent devices, and the traditional rule is that it should be possible to detect and clear a fault at any location in one second or less.

    e     m      i      t

D C B A

It’s important to recognize that this is a traditional measure, and in today’s world there may be a desire for faster fault clearing times to manage the energy release associated with arc flash. That desire may lead to the application of  faster protective functions that, in order to achieve selectivity, could also distort traditional economics. Even so, the one-second criterion is still a valid measure of system performance.

0.3 sec 0.3 sec

0.3 sec

current

Fig 13 – Selective Coordination of  5 Overcurrent Protection Devices .

5

By tradition, time-current curves are always log-log plots. © 2008 Louie J. Powell, PE

Page 21 Rev 1

Figure 13 illustrates an array of four protective devices, A, B, C and D, that are arranged selectively. Device A is termed a “primary protective device” and includes an instantaneous overcurrent element, while devices B, C and D are time-overcurrent devices that provide backup protection in the event A fails to operate in addition to perhaps providing primary protection to other areas in the system.. Note that at the maximum fault level, the time differential between the devices is 0.3 seconds, so that the slowest device (device D) will operate in 0.9 seconds or less for the maximum fault. This time-current curve can then be used to translate the one-second criterion into a system design criterion – if this curve depicts the limit of acceptable system operation, then the limit of acceptable system design is one that can be adequately protected by this array of protective devices. Therefore, if the one-second rule is to be met, the system must be designed to require no more than three layers of timeovercurrent backup protection at any voltage level. Figure 13 could also be applied to a medium voltage system with fundamentally the same conclusion – there can be no more than three levels of time overcurrent backup protection in a well-designed system. However, in that instance the interpretation of the conclusion might differ because of differences in the kinds of equipment used to provide protection. Fig 14 – Low Voltage Substation that might Fig 14 depicts a typical double-ended low correspond to the protection in Fig 13. voltage unit substation that might have protection characteristics such as those in figure 13. Here, device A is in a motor control center, B is the substation feeder supporting the motor control center, C is the tie breaker, and D is the secondary main breaker. Note that it would not be possible to insert any additional time-overcurrent protective devices into this system without causing the time delay associated with device D to exceed the one second criterion. That consideration therefore discourages insertion of sub-buses or other complications into this system. Finally, there is one other criterion in the world of protection that provides insight into the quality of design of a system. In Figure 15, the time current curves have been redrawn with devices A and B having the same pickup sensitivity, and with devices C and D having different sensitivities. Note that it is still possible to achieve the maximum one second clearing time but only if devices C and D have a higher pickup sensitivity that devices A and B. The pickup sensitivity of a protective device is related to the full load that it is designed to carry – for obvious reasons, the pickup sensitivity must be great than full load current. Therefore, the © 2008 Louie J. Powell, PE

Page 22 Rev 1

message here is that in a well designed system, the backup breaker must have a significantly higher full load rating than the largest feeder breaker. As a practical rule of thumb, the largest feeder breaker should be no larger than one-third the rating of the main breaker in a substation, and loads that would otherwise violate this rule should be viewed as candidates for special treatment, eg, a captive transformer connection for larger motors.

    e     m      i      t

D C B A

0.3 sec 0.3 sec

0.3 sec

current

Fig 15 – Feeders should have higher current ratings than mains.

The System Performance Specification The “performance specification” does not have to be a long document, but it does need to explicitly define the expected range of parameters that may either be calculated as part of the design process, or that may be measured on the physical system after it is constructed. Critical attributes of performance include: 

 

 

Reliability – “n-0”, “n-1”, or for systems demanding a more rigorous approach to reliability, a quantitative measure of reliability such as mean-time-to-failure (MTTF). Note that statements of reliability in “nines” is vague and not at all meaningful when designing a system. Note also that it may necessary to establish different reliability targets for different components of the load. Load to be served – load should be identified by process elements. Power factor – power factor is important in system performance, but the power factor of  loads will be what it will be, and power factor improvement will mainly be needed to address other system performance issues (eg, steady state voltage spread). However, if  the host utility imposes a power factor penalty at the revenue meter, the system specification should identify the target power factor that the owner wishes to achieve for the purpose of managing that penalty. Expected voltage spread – including minimum and maximum steady-state voltage levels. Thermal rating limits – It would normally be assumed that power system components would require continuous thermal ratings that exceed their expected continuous loading.

© 2008 Louie J. Powell, PE

Page 23 Rev 1

















However, it may be appropriate to specify a minimum margin between loading and component limits. Specifying a minimum margin establishes the principle that there should be some margin, but more importantly, defining the margin at the outset eliminates second guessing about what should have been anticipated about provisions for future expansion. Note that the National Electrical Code may impose minimum required margins between electrical loading and ratings, and the specification should clearly delineate between margins mandated by the Code and additional margins that the owner/operator of the facility may be able to utilize for future growth. It is also helpful for the specification to define the minimum voltage level required for motors of various horsepower ratings. That is, it is helpful to define in the specification the size of the largest motors that will expect service at, for example, 480v versus 4.16kV. Limits should be established for the minimum voltage that can be tolerated under expected impact load conditions. While a limit on the minimum impact load voltage ultimately does provide guidance on how large motors should be started, it is sometimes helpful to go one step further in the system specification to delineate horsepower thresholds that at least meet, and possibly exceed, the voltage dip specification. The National Electrical Code requires that circuit breakers be rated to interrupt the maximum short circuit current that they will encounter in their application. That means that the minimum margin legally required between short circuit current and short circuit rating is only slightly greater than zero. Given the critical safety issues involved with short circuits, most industrial system designers prefer to specify a finite minimum design margin target (such as 10%). In recent years, operators of industrial electrical systems have become acutely aware of  the concerns associated with arc flash. The system specification should clearly state the expected maximum radiated arc flash thermal energy rating for each voltage level in the system. Obviously, this thermal energy level must be coordinated with the maintenance procedures expected to be followed in the facility. Various protection parameters need to be specified, including the maximum fault clearing time, the target minimum coordinating time margin between fault protective devices, whether protection will be applied on a per-phase basis (as opposed to an application in which one protective device monitors multiple phases), etc. This material has not addressed the issue of system grounding, but it is important that the system specification clearly define the expected modes of neutral grounding at each voltage level in the system.

© 2008 Louie J. Powell, PE

Page 24 Rev 1

Appendix I Several power flow diagrams have been included as figures in this material. The following defines the conventions used in preparing these diagrams.

© 2008 Louie J. Powell, PE

Page 25 Rev 1

Review: Here are the major points you should remember from this course: o

o

o

o

o

o

o

o o

o

o

o

o

o

The power distribution system in an industrial facility is the responsibility of a power systems engineer. However, the role of that power system is to deliver electric power to the process, and the power system engineer usually has little influence over how the process consumes electric power. In industry, electric power is a COST of doing business, and like any other cost of  business, management will demand that the cost be managed and potentially reduced. The four major objectives in the design of an industrial power distribution system are 1. Safety 2. Reliability 3. Maintainability 4. Economy Industry constantly strives for growth, and the power distribution system for an industrial manufacturing system will almost inevitably be called upon to support increases in the production at that facility. Power distribution systems are built up using standard components, and the challenge to the power system engineer is to make informed decisions about frequency, voltage, thermal capacity and short circuit capacity in selecting those components. The overall architecture, or one-line arrangement of the electrical system must be driven by the requirements of the manufacturing process it serves. There are significant differences in requirements between discrete manufacturing businesses and continuous process businesses. The risk of failure can be minimized but never eliminated, and every component of the system is a candidate for eventual failure. Ultimately, what matters is whether the failure of a component results in failure of the electrical system to deliver power to the process. The key to reliability is redundancy. Single-point failure modes should be identified. Whether every single-point failure mode must be eliminated really depends on the requirements of the business, and in particular, on the impact of an interruption in the supply of electric power to the process. Ultimately, it comes down to whether the cost of addressing a single-point failure mode exceeds the cost of living with that single-point failure if and when it does occur. This is referred to as the “n-1” principle. While assessing reliability numerically is interesting, the main advantage is to add insight into the elements contributing to reliability. Except for special instances in missioncritical facilities, it rarely contributes anything of value to the design process. In designing a system to address reliability, elements are often grouped in a way that failure of one element requires that there be redundancy for the entire group. In most instances, one layer of redundancy is sufficient. Adding one layer of redundancy greatly increases overall system reliability. Adding additional layers will add additional theoretical reliability, but the incremental performance improvement generally does not  justify the added cost. The expression of that principle is that the most common “building block” in designing an industrial power system is the double-ended unit substation depicted in Figure 3. Predicting steady-state loading requirements while allowing for the uncertainties of load growth is a major challenge. All electrical system components have thermal ratings that define the maximum loading they can withstand without overheating. While these ratings may ultimately be related to

© 2008 Louie J. Powell, PE

Page 26 Rev 1

o

o

o

o

o

o

o o

o

o

o

current flow in the conducting elements making up those components, it is typical for the ratings to be expressed in terms of the total power (kVA or MVA) equivalent. Generator ratings are often expressed in terms of real power (kW or MW). Power engineers need to recognize that these ratings may be the mechanical output ratings of the generator prime mover (turbine, engine, etc), and need to also ascertain the actual MVA rating of the generator itself. Some loads have cyclic characteristics. In those instances, it is usually necessary to consult either the applicable industry standards, or perhaps the manufacturer, to determine how cyclic loading relates to the continuous thermal rating of power distribution system components. Some components (eg, transformers) may have forced cooling capabilities that increase the inherent continuous thermal capacity beyond that of the basic component carcass. Careful exploitation of these forced cooled ratings can result in considerable savings in capital investment, and the ability to add forced cooling may be an option to address future load growth. In most cases, a system can tolerate steady-state voltages between 95% and 103% of  rated. That criterion is based on the fact that the voltage rating of motors is typically less that the voltage rating of the system that supplies those motors. For example, the motors that would normally be applied on a 480v. system are rated 460v. System voltage drop is a result of the interaction of the inductive reactance of the system components, and the flow of reactive power to loads. The torque produced by a motor, in per unit of rated torque, is approximately proportional to the square of applied per unit terminal voltage. Most motor applications include design margins that mean that the motor is not actually required to produce rated torque. Unfortunately, those design margins are generally not known to the power system engineer. Transformer voltage taps are helpful in reducing steady-state voltage spread. The addition of power factor capacitors is another cost-effective way to address steadystate voltage spread. The optimum location of those capacitors depends on the objective for adding them to the system design Capacitors intended mainly to reduce reactive demand at the point of revenue billing should be installed in one bank at a point adjacent to the point of metering. Capacitors intended mainly to remediate system voltage drop should be distributed across the system adjacent to loads, and in a fashion such that capacitors are switched on and off at the same time that their associated loads are switched on and off. The only accurate way to predict voltage spread in advance of commissioning of a system is by means of a steady-state power flow simulation that can correctly address the nonlinear relationship between load and voltage. The real-power energy losses in an industrial electric power distribution system are generally negligible. There are, however, often opportunities to improve efficiency at the points of electric power consumption. In particular, the use of higher efficiency motors, and taking care to more carefully match motor ratings with the actual mechanical torque requirements of their driven loads can result in an overall improvement in energy efficiency in a manufacturing process. The voltage dip associated with an impact load can be estimated from the formula: ∆ MVAR Voltage change (%) ≈ × 100  MVAshort circuit In the specific case of motor starting, ∆MVAR can be estimated to be six times the motor horsepower rating.

© 2008 Louie J. Powell, PE

Page 27 Rev 1

o

o

o

o

o

o

o

o

o

o

The addition of power factor capacitors will change the baseline for the voltage dip associate with starting motors or other impact loads, but in general will make the magnitude of dip greater than would be the case without capacitors. Reactive solutions to motor starting or other impact loads must be dynamic reactive sources that increase their reactive output as the impact forces a decrease in bus voltage. Magnetically latched contactors drop out when the instantaneous voltage falls below about 75%. A common rule-of-thumb is that the voltage dip associated with an impact load should not force the instantaneous voltage below 80-85% of rated system voltage based on avoiding contactor dropout. Microprocessor devices may have a higher undervoltage dropout level, and may require special provisions to ride through transient voltage dips. Starting reactors and autotransformers may be useful in reducing the starting impact of  larger motors. Adjustable-speed power converters can significantly reduce the voltage impact associated with starting a motor, and the ability of these system to allow the speed of the motor to be adjusted to match the requirements of the load may also result in enhanced energy efficiency. Consideration should be given to applying very large motors (typically 10,000 HP are above) in captive-transformer arrangements. The captive-transformer design minimizes motor starting voltage drop, and also transfers the application of the associated large motor to a higher voltage bus that is likely to have greater short circuit stiffness such that the impact is less significant. The National Electric Code mandates that the short circuit rating of fault interrupting devices be sufficient for the available short circuit currents that the system can produce. The loads to be served will determine the short circuit levels that the system must have to support those loads. An undocumented but effective criterion is that the available short circuit stiffness should be about 25 times the magnitude of peak stead-state loading. In general, a modular design involving a larger number of individually-smaller unit substations is superior to a design that relies on a smaller number of larger substations. Application of the “25:1” rule is helpful in deciding the optimum voltage of a mediumvoltage distribution bus in an industrial system. While the designer of a system has the freedom to choose any voltage that makes sense, there are a few voltage choices that are not recommended: 2.4kV is obsolete as a distribution voltage although it may make sense as a machine voltage in a captive transformer application. 4.8kV is a poor choice of voltage for industrial distribution. 12.47kV and 13.2kV are utility distribution voltages and make sense for industrial distribution only if the industrial system is to be fed directly from the utility distribution system. The maximum fault clearing time using time overcurrent protection should be kept to one second or less. In general, the system design should not require more than three layers of timeovercurrent backup protection at any given voltage level In general, the system design should result in a minimum 3:1 ratio between the pickup threshold of a backup overcurrent device and the pickup threshold of the largest feeder overcurrent device. Consideration must be given to the thermal energy released by faults and the resulting requirements for personnel protection. That consideration may result in the desire to reduce the available short circuit level, to reduce the fault clearing time, or both. 

 

o

o

o

o

© 2008 Louie J. Powell, PE

Page 28 Rev 1