COST Control

COSTING SUPPORT AND COST CONTROL IN MANUFACTURING

A COST ESTIMATION TOOL APPLIED IN THE SHEET METAL DOMAIN

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit Twente, op gezag van de rector magnificus, m agnificus, prof.dr. F.A. van Vught, volgens besluit van het College voor Promoties in het openbaar te verdedigen op vrijdag 3 mei 2002 te 15.00 uur.

door

Erik ten Brinke geboren op 15 maart 1973 te Hardenberg

Dit proefschrift is goedgekeurd door: de promotor prof.dr.ir. H.J.J. Kals.

Dit proefschrift is goedgekeurd door: de promotor prof.dr.ir. H.J.J. Kals.

Costing support and cost control in manufacturing manufacturing A cost estimation tool applied in the sheet metal domain

Erik ten Brinke

ISBN 90-365-1726-5 © Erik ten Brinke, 2002 Printed by PrintPartners Ipskamp, Enschede, The Netherlands

Preface This thesis is the result of five years of research in the field of costing support and cost control in manufacturing. The research has been performed in the framework of a research program focussed on sheet metal manufacturing as part of the IOP-research program supported by the Dutch Ministry of Economic Affairs. The research has been supervised by Prof. H.J.J. Kals, former chairman of the laboratory of Design, Production and Management at the University of Twente. The research has been a continuation of two previous research projects at the laboratory mentioned above. The result of the project ”An architecture for cost control in manufacturing: the use of cost information in order-related decisions” by Arthur Liebers has been a starting point for this research. In addition, the results of the project “Manufacturing integration based on information management” by Eric Lutters have been employed in this research. I could not have accomplished this research without the help of others. First, I want to thank Prof. Kals and Ton Streppel for their support. Though our discussions weren’t that frequent and easy, they helped me to find a proper path in my research. Further, I want to thank Arthur Liebers for passing on the results of his research including his library of cost literature. Also, I want to thank Eric Lutters for his, sometimes philosophical and sometimes repeatedly, explanations of some of the principles of his Information Management approach. Furthermore, I want to thank all of my colleagues at the laboratory. I have appreciated their companionship at work, at activities outside work and at activities outside work at work. Within this research, two students finished their master assignment, resulting in valuable input for my research. Therefore, I want to thank René Veltman and Alex Huttinga for accomplishing their, in their eyes abstract, assignments. In spite of their lack of interest and lack of experience in programming, they made a valuable contribution to the prototype implementation. Finally, I want to thank my parents and sister for their support and patience.

Enschede, 24 February 2002

Erik ten Brinke

i

ii

Summary In the product development cycle several engineering tasks like design, process planning and production planning have to be executed. The execution of these tasks mainly involves information processing and decision-making. Because costs is an important factor in manufacturing, adequate information about costs is extremely valuable for all engineering tasks. Therefore, a cost estimation system for the generation of cost information and for cost control, integrated in the product development cycle, is required. The integration of engineering tasks in the product development cycle has been a major research topic in the last decades. Because engineering tasks can be seen as information processing tasks, information is the proper base for integration. The Manufacturing Engineering Reference Model developed at the Laboratory of Design, Production and Management, is based on the use of a central information management kernel that facilitates both the availability and the accessibility of meaningful representations of the evolving manufacturing information. Because this reference model is designed especially for the integration of engineering tasks, it is used for the development of the cost estimation system. Based on a literature review on cost control and cost estimation in manufacturing and the Manufacturing Engineering Reference Model, a generic cost estimation architecture has been developed. The architecture consists of six functional modules arranged around the information kernel of the Manufacturing Engineering Reference Model. The separate modules are: Cost Models, Cost Determination, Cost Reports, Risk Analysis, Data Analysis and Data Tuning. The Cost Model module is used for the definition and the management of cost models. Multiple cost models can be defined in order to support all engineering tasks and to be able to compare cost models. Based on a cost model, the Cost Determination module calculates the costs. A cost model can be selected based on a specific cost model, the required accuracy or the available information. Cost reports can be created with the Cost Report module. Information about the quality, the accuracy and the sensitivity of the costs has to be provided by the Risk Analysis module. The analysis of (historic) data is a task for the Data Analysis module, while the Data Tuning module has to tune data. A new method for variant based cost estimation is proposed and positioned in the architecture. The cost models are defined based on the cost structure. With the aid of the cost structure, costs can be defined for any object causing costs. Because cost structures can be attached to the information structures related to the Manufacturing Engineering Reference Model, the costs can be calculated for any object at any aggregation level. Additionally, the cost structure enables the differentiated storage of cost information. Based on the information structures and the cost structures, cost views can be constructed. These cost views visualise the differentiated costs for the user. The cost estimation architecture and the cost structure enable the use of four cost control loops: the engineering and planning feedback loop, the order acceptance feedback loop, the production feedback loop and the accounting feedback loop. Some parts of the cost estimation architecture have been implemented in a prototype system. The prototype system is demonstrated by means of an example from the product development cycle in the sheet metal manufacturing domain. Generative and variant based cost estimation are used to demonstrate cost support and cost control. For generative cost estimation two distinct cost models, direct costing and activity based costing, are used.

iii

iv

Samenvatting In de ontwikkelingscyclus van een product moeten verscheidene engineeringstaken worden uitgevoerd. De uitvoering van deze taken bestaat voornamelijk uit informatie verwerken en beslissingen nemen. Omdat binnen het hele voortbrengingsproces van een product de kosten een belangrijke rol spelen, is geschikte informatie over de kosten voor de engineeringstaken zeer waardevol. Daarom is een kostprijsschattingssysteem voor de generatie van kosten informatie en voor de beheersing van de kosten, geïntegreerd in de ontwikkelingscyclus van een product, noodzakelijk. In de laatste decennia was de integratie van engineeringstaken in de ontwikkelingscyclus van een product een belangrijk onderzoeksthema. Omdat engineeringstaken als informatieverwerkingstaken gezien kunnen worden, is informatie de juiste basis voor integratie. Het binnen het laboratorium Ontwerp, Productie en Management ontwikkelde “Manufacturing Engineering Reference Model” is gebaseerd op een centrale “information management kernel”, die de beschikbaarheid en de toegankelijkheid van betekenisvolle representaties van de zich ontwikkelende informatie mogelijk maakt. Omdat dit referentie model speciaal voor de integratie van engineeringstaken is ontwikkeld, is het gebruikt voor de ontwikkeling van het kostprijsvoorcaclulatiesysteem. Gebaseerd op een literatuurstudie naar het schatten en beheersen van kosten en het “Manufacturing Engineering Reference Model” is een generieke kostprijsschattingsarchitectuur ontwikkeld. De architectuur bestaat uit zes functionele modules, die rond de “information management kernel” zijn geplaatst. De afzonderlijke modules zijn: “Cost Models”, “Cost Determination”, “Cost Reports”, “Risk Analysis”, “Data Analysis” en “Data Tuning”. De “Cost Model” module wordt gebruikt voor het definiëren en beheren van kostenmodellen. Om alle engineeringstaken te kunnen ondersteunen en om kostenmodellen met elkaar te kunnen vergelijken, is het mogelijk om meerder kostenmodellen te definiëren. Gebaseerd op een kostenmodel berekent de “Cost Determination” module de kostprijs. Een kostenmodel kan worden geselecteerd op basis van een specifiek kostenmodel, een geëiste nauwkeurigheid of de beschikbare informatie. Kostenrapporten kunnen met de “Cost Report” module worden gegenereerd. Informatie over de kwaliteit, nauwkeurigheid en gevoeligheid van de kosten moet door de “Risk Analysis” module worden geleverd. De analyse van (historische) data is de taak van de “Data Analysis” module en de “Data Tuning” module moet data op elkaar afstemmen. Een nieuwe methode voor variant gebaseerd schatten van de kostprijs wordt voorgesteld en in de kostpijsvoorcalculatiearchitectuur gepositioneerd. De kostenmodellen worden gedefinieerd op basis van de kostenstructuur. Met behulp van de kostenstructuur kunnen de kosten voor ieder object dat kosten veroorzaakt worden gedefinieerd. Omdat kostenstructuren aan de informatiestructuren, gerelateerd aan het “Manufacturing Engineering Reference Model”, kunnen worden verbonden, is het mogelijk voor ieder object en op ieder aggregatieniveau de kosten te berekenen. Bovendien maakt de kostenstructuur de gedifferentieerde opslag van de kosten mogelijk. Op basis van de informatiestructuren en de kostenstructuren kunnen kosten “views” worden geconstrueerd. Deze kosten “views” visualiseren de gedifferentieerde kosten voor de gebruiker. De kostprijsvoorcalculatiearchitectuur en de kostenstructuur maken het gebruik van vier feedbackloops voor de beheersing van de kosten mogelijk: de engineering en planning feedbackloop, de orderacceptatie feedbackloop, de productie feedbackloop en de accounting feedbackloop.

v

Sommige delen van de kostprijsvoorcalculatiearchitectuur zijn geïmplementeerd in een protot ype kostprijsvoorcalculatiesysteem. Het prototype systeem wordt aan de hand van een voorbeeld uit de ontwikkelingscyclus van een product in het plaatwerk domein gedemonstreerd. Voor de demonstratie van kostenondersteuning en kostenbeheersing wordt gebruik gemaakt van generatief en variant gebaseerd schatten van de kostprijs. Voor het generatief schatten van de kosten wordt gebruik gemaakt van twee verschillende kosten modellen, nl. “direct costing” and “acitivity based costing”.

vi

Table of contents Preface .................................................................................................................................................. i Summary.............................................................................................................................................iii Samenvatting ....................................................................................................................................... v

Part I: The framework 1

Introduction................................................................................................................................... 1 1.1 1.2 1.3 1.4

2

Literature review on cost estimation and cost control in manufacturing...................................... 3 2.1

2.2

2.3

2.4

2.5

2.6

2.7 3

Cost estimation in manufacturing ..................................................................................... 1 Sheet metal manufacturing ............................................................................................... 1 Problem definition ............................................................................................................ 2 Scope of the thesis ............................................................................................................ 2

Cost ................................................................................................................................... 3 2.1.1 Cost types............................................................................................................. 5 2.1.2 Cost allocation ..................................................................................................... 7 The use of cost information in manufacturing.................................................................. 8 2.2.1 Task oriented cost information ............................................................................ 8 2.2.2 The need for information management.............................................................. 10 Cost control..................................................................................................................... 10 2.3.1 The manufacturing planning and control reference model................................ 10 2.3.2 The (cost) control architecture........................................................................... 13 Cost estimation................................................................................................................ 14 2.4.1 Generative cost estimation................................................................................. 14 2.4.2 Variant based cost estimation ............................................................................ 16 2.4.3 Hybrid cost estimation ....................................................................................... 18 Cost modelling................................................................................................................ 18 2.5.1 Determination of the scope ................................................................................ 19 2.5.2 Determination of the allocation base ................................................................. 19 2.5.3 Determination of the cost functions................................................................... 20 Sheet metal manufacturing ............................................................................................. 22 2.6.1 General............................................................................................................... 22 2.6.2 Mechanical cutting............................................................................................. 23 2.6.3 Non-mechanical cutting..................................................................................... 24 2.6.4 Joining operations.............................................................................................. 26 2.6.5 Bending operations ............................................................................................ 27 Restatement of the problem definition............................................................................ 27

Information Management ........................................................................................................... 29 3.1 3.2

The Manufacturing Engineering Reference Model ........................................................ 29 Information structuring ................................................................................................... 30 vii

3.3 3.4 3.5

The information structures.............................................................................................. 32 Ontologies ....................................................................................................................... 35 Process architectures related to Information Management............................................. 36

Part II: A generic cost estimation architecture 4

Design of a generic cost estimation architecture ........................................................................ 39 4.1 4.2 4.3

4.4 5

Functional specifications ................................................................................................ 39 Cost information and the information structures ............................................................ 40 The cost estimation architecture ..................................................................................... 43 4.3.1 Cost Models ....................................................................................................... 45 4.3.2 Cost Determination ............................................................................................ 47 4.3.3 Data Analysis..................................................................................................... 47 4.3.4 Risk Analysis ..................................................................................................... 48 4.3.5 Data Tuning ....................................................................................................... 48 4.3.6 Cost Reports....................................................................................................... 48 Variant based cost estimation and its position in the architecture .................................. 49

Employment of the architecture.................................................................................................. 53 5.1 5.2 5.3 5.4

Cost control..................................................................................................................... 53 Cost modelling................................................................................................................ 57 Costing support ............................................................................................................... 59 Concluding remarks ........................................................................................................ 59

Part III: A prototype cost estimation system 6

Implementation of a prototype cost estimation system .............................................................. 63 6.1 6.2

6.3

6.4 6.5 7

Application of the prototype cost estimation system in the sheet metal domain........................ 83 7.1 7.2

7.3

7.4

viii

System specifications...................................................................................................... 63 Cooperating systems ....................................................................................................... 63 6.2.1 Creation of databases......................................................................................... 63 6.2.2 Design ................................................................................................................ 66 6.2.3 Process planning ................................................................................................ 66 6.2.4 Production planning........................................................................................... 69 The cost estimation system ............................................................................................. 72 6.3.1 Cost Models module .......................................................................................... 72 6.3.2 Cost Determination module............................................................................... 76 6.3.3 Cost Reports module.......................................................................................... 77 Variant based cost estimation ......................................................................................... 79 Concluding remarks ........................................................................................................ 82

Example product and context information...................................................................... 83 Example cost models ...................................................................................................... 86 7.2.1 Direct Costing.................................................................................................... 86 7.2.2 Activity Based Costing ...................................................................................... 87 Example cost calculations............................................................................................... 88 7.3.1 Generative cost estimation................................................................................. 89 7.3.2 Variant based cost estimation ............................................................................ 90 7.3.3 Hybrid cost estimation ....................................................................................... 91 Costing support and cost control..................................................................................... 91

ïð

8

Conclusions and recommendations ............................................................................................ 93 8.1 8.2

Conclusions..................................................................................................................... 93 Recommendations........................................................................................................... 94 ïð

References.......................................................................................................................................... 95 Terminology....................................................................................................................................... 99

Appendices A

Time and cost functions from literature.................................................................................... 103 A.1 A.2 A.3 A.4 A.5 A.6 A.7

B

Punching ....................................................................................................................... 103 Nibbling ........................................................................................................................ 103 Cutting........................................................................................................................... 105 Water jet cutting............................................................................................................ 107 Laser cutting.................................................................................................................. 107 Laser welding................................................................................................................ 108 Bending......................................................................................................................... 109

Example cost structures ............................................................................................................ 111 B.1 B.2 B.3

Resource information structure..................................................................................... 111 Product information structure ....................................................................................... 112 Order Information Structure ......................................................................................... 113

C

Regression analysis................................................................................................................... 115

D

Neural networks........................................................................................................................ 119

E

Example product and context information ............................................................................... 121 E.1 E.2 E.3

Components .................................................................................................................. 121 Information structures................................................................................................... 128 Cost structures............................................................................................................... 130

ix

x

Part I The framework

1

Introduction

1.1 Cost estimation in manufacturing Within this thesis, the focus is on cost estimation in manufacturing. Manufacturing refers to the series of interrelated activities and operations involving the design, the materials selection, the planning, the production and the quality assurance of the products (Chisholm, 1990). The product development cycle consists of a combination of manufacturing activities resulting in a product. Production is only a part of manufacturing and the product development cycle. Production is the act or process (or the connected series of acts or processes) of actually physically making a product from its material constituents. Production can consist of fabrication and assembly operations (Chisholm, 1990). Fabrication addresses those operations applied during production that are not assembly operations. The preparation of the actual production of a product is dealt with by engineering. Therefore, engineering includes activities as design, process planning and production planning. During the product development cycle, the execution of engineering tasks includes many decisions to be taken. The decisions are concerned with the product, the production of the product and the disposal/recycling of the product. The decisions are based on several criteria, e.g. technical constraints, but costs are also an important criterion. In order to be able to use costs as a decision criterion, the costs of all aspects of the product have to be known. Because the costs are not known in advance, a cost estimation system is required to generate the required cost information. The cost estimates have to be based on the product information, which is available at a certain stage of the product development cycle. Because the available information is different in amount and detail in different stages of the product development cycle, it is difficult to support all engineering tasks. Besides the use of cost estimation for decision-making, it can also be used t o control costs. When the costs can be controlled, it is possible to propose specific product changes reducing the costs. In order to reduce the time span of the product development cycle concurrent engineering is used. In concurrent engineering, the engineering tasks are partially performed simultaneously. Concurrent engineering requires the integration of the engineering tasks in the product development cycle. In order to support the engineering tasks with cost information, cost estimation has to be integrated in the product development cycle as well.

1.2 Sheet metal manufacturing The investigation reported in this thesis has been performed in the framework of a research program focussed on sheet metal manufacturing as part of the IOP-research program supported by the Dutch Ministry of Economic Affairs. Trends in small batch manufacturing in the sheet metal industry are a further decrease of batch sizes, shorter delivery times and lower prices. Therefore, in the sheet metal industry it is crucial to be able to generate cost estimates and to control the costs in order to make fair profit. Many sheet metal companies in The Netherlands are supply companies. Especially these companies have to be able to generate cost estimates quickly and accurately. The cost estimates have to be generated quickly because quotations have to be offered to potential customers in a short time period. The cost

1

Chapter 1

estimates have to be generated accurately because the margins between cost price and sales price are small due to (inter-) national competition.

1.3 Problem definition The success of a company largely depends on the profit that it can realise. The profit is determined by the costs that are made and the extent in which these costs are recovered. Therefore, it is essential for a company to know the (future) costs and being able to control them. When the (future) costs are known throughout the entire product development cycle, the engineers can make use of cost information during the decision-making processes. Therefore, it is necessary to integrate the cost estimation activities in the product development cycle. For this a system is required that can support engineers and other systems in cost estimation and cost control. The objective of this research is to develop a prototype cost estimation system that can provide cost information throughout the whole product development cycle and that can be used for cost control. In order to be independent of the manufacturing environment, the cost estimation system has to be generic. An implementation of the system has to be applied in the sheet metal domain.

1.4 Scope of the thesis This thesis is composed of three parts: the framework, a generic cost estimation architecture and a prototype cost estimation system. Part I: The framework In the chapter “Literature review on cost estimation and cost control in manufacturing”, the terminology used in literature and used in this thesis is explained. Furthermore, some methods in the field of cost control and cost estimation are discussed. First, it is explained what costs are, how they can be characterised and how they can be assigned to products. The use of cost information in manufacturing is discussed. For different engineering tasks different cost information is used in a different manner. Furthermore, the cost information produced by different engineering tasks has to be managed properly in order to be able to use it effectively. Second, one method for cost control is discussed. This method is positioned in manufacturing and the distinguished functions are explained. The two most important functions: cost estimation and cost modelling are clarified based on literature. Third, common sheet metal processes are described. Furthermore, time and cost functions from literature are given per process. Finally, an information management approach is described that is very suitable for the integration of engineering tasks in the product development cycle based on information. The related information structures and process architectures are explained. Part II: A generic cost estimation architecture In this part, the design of a generic cost estimation architecture is explained. Based on information from the literature review, functional specifications and information structures, a cost estimation system is designed. The functional modules are described and the function and the use of the architecture are clarified. Part III: A prototype cost estimation system The cost estimation architecture was partially implemented in a prototype cost estimation system. The prototype cost estimation system is demonstrated by means of an example taken from the sheet metal domain.

The thesis is concluded with final conclusions and recommendations. 2

2

Literature review on cost estimation and cost control in manufacturing

This chapter provides a literature overview on cost estimation and cost control in manufacturing. First, the notion cost will be discussed in general in section 2.1. Next, the use of cost information in the product development cycle is discussed in section 2.2. The use of cost information by different engineering tasks and the need for an information management system is described. Besides the ability to estimate costs, it is necessary to be able to control the costs. Section 2.3 positions cost control in manufacturing and will discusses the sub functions of cost control. The two most important functions: cost estimation and cost modelling are discussed in section 2.4 and 2.5. Section 2.6 deals with the most common sheet metal processes and the related cost and time functions. Finally, the problem definition is restated in section 2.7.

2.1 Cost The broad definition of costs is related to the economic resources (manpower, equipment, real facilities, supplies and all other resources) necessary to accomplish work activities or to produce work outputs (Stewart, 1995a). Usually, costs are expressed in terms of units of currency. Therefore, costs are the amount of money representing the resources spent for the production of output. A resource is a physical entity that is required to be able to execute a certain operation. Resources can be e.g. machine tools, tools and fixtures, but also operators and materials. Output can be products and services. During the product development cycle, engineering tasks cause and fix costs. Figure 2.1 shows the influence of several company departments on the product costs as taken from a German research project in machine design (Wierda, 1990). The engineering tasks cause costs because of their contribution to the development of a product. At the start of the product development cycle, no costs are fixed yet (Figure 2.2). The consecutive engineering tasks fix the costs because of the decisions taken. The decisions taken during an engineering task at the beginning of the product development cycle can significantly influence the costs caused by engineering tasks later in the engineering cycle because the solution space for the engineering tasks is reduced by it. Figure 2.1 shows that design itself takes only about 10% of the product costs, whereas it fixes about 70% of the product costs. It has been argued that the latter percentage is misleading because the product specifications already imply some minimal costs. According to Ehrlenspiel, design is responsible for 20 to 30% of the total product costs (Wierda, 1990). The way in which engineering tasks contribute to the product costs depends on the production environment. Figure 2.3 shows a comparison between high-tech production and classic mass production (Thompson, 19??) . The figure shows that for high-tech production the costs caused before production are about 5 times higher than for mass production, while the costs of production are about one fifth. For mass production, it is required to be able to estimate the costs of production more accurately than the costs of design and engineering. For high-tech production, the opposite applies.

3

Chapter 2

Figure 2.1 Experimentally determined influence of the main departments of a company on the product costs (Wierda, 1990).

Figure 2.2 Decreasing costs not fixed and increasing costs caused during the product development cycle (Wierda, 1990).

Figure 2.3 Product life-cycle costs (Thompson, 19??).

It is easier to estimate costs accurately when more detailed information is available. Since design fixes about 70% of the product costs, it is required to make accurate cost estimates during design. However, during the design process the product information is not yet available in full detail, so it is

4


difficult to make accurate estimates. This phenomenon is known as the cost estimation paradox, see Figure 2.4.

Figure 2.4 The cost estimation paradox (Bode,1998a).

2.1.1 Cost types Total product costs are composed of several different cost items. Possible breakdowns of the product costs are the cost breakdown structures of Fabrycky & Blanchard (Asiedu, 1998) and Liebers (Liebers, 1998) (Table 2.1 and Table 2.2). These two examples show that multiple breakdown structures are possible. A general cost break down structure seems hardly possible. Two important criteria for a good cost breakdown are: all costs must be covered and no costs must be counted twice.

-

Research and development cost Product management Product planning Product research Design documentation Product software Product test and evaluation

-

-

-

Total product cost Production and Operations and Retirement and construction cost maintenance cost disposal cost Manufacturing/ - Operations/ - Disposal of nonconstruction maintenance repairable management management - Product retirement Industrial engineering - Product operation - Documentation and operations - Product distribution analysis - Product maintenance Manufacturing - Inventory Construction - Operator and Quality control maintenance training Initial logistic - Technical data support - Product modification

Table 2.1 The cost breakdown structure of Fabrycky & Blanchard (Asiedu, 1998).

The costs from a cost breakdown structure are caused by different resources and the way these costs are related to those resources can be different. In order to get a better perception of costs, costs are classified in different types according to their cause and the relation to their cause. Therefore, it is advantageous to know the costs per cost type Two general cost classifications are on the one hand direct versus indirect costs and on the other hand variable versus fixed costs. Direct costs are costs that can be identified specifically and consistently with an end objective (such as a product, service, software, function, or project), while indirect costs cannot be identified specifically and consistently with an end objective (Shuford, 1995). This means that direct costs can be allocated directly, i.e. the allocation base is known, whereas for the allocation of indirect costs an allocation base has to be defined (Cooper, 1991). The allocation of costs will be treated in more detail i n the next section. 5

Chapter 2

Variable costs are costs that change with the rate of production or the performance of services (Stewart, 1995a). Fixed costs are costs that do not vary with the volume of business (Stewart, 1995a). Furthermore, semi variable costs and step-fixed costs can be distinguished. Semi variable costs are costs that vary somewhat in relation to volume, but their percentage of change is not the same as the percentage of change in volume (Shuford, 1995). Step-fixed costs are fixed costs that alter their behaviour as the activity level moves from one relevant range to another (Shuford, 1995).

-

-

Product cost Costs of executing a production plan Costs of generating a Costs of successfully production plan Costs of re-planning executing a production plan Marketing and - Having production - Repair, rework and promotion resources scrap Company - Using production - Resource repair, management, resources including down time including control - Materials included in - Late delivery Sales and order intake the products Resource, process and - Waste product design Process planning Production planning

Externally imposed burdens - Contingency allowances - Cost increasing taxes (as opposed to profit reducing taxes)

Table 2.2 The cost breakdown structure of Liebers (Liebers, 1998).

The distinction between recurring & non-recurring and relevant & irrelevant costs is also often used. Recurring costs are repetitive costs that vary with the quantity being produced (Stewart, 1995b) . Non-recurring costs are elements of development and investment costs that generally occur only once in the life cycle of a work activity or work output (Stewart, 1995b). Relevant costs are costs that are present in one of several alternatives but are absent, either in whole or in part, in other alternatives (also called differential costs) (Shuford, 1995). These costs play a role in specific decision-making processes, whereas all other costs are irrelevant costs (Liebers, 1998). Other cost types that are frequently distinguished are listed here to illustrate the diversity of cost types: Acquisition costs: Total expenditures estimated or incurred for the development, manufacture, construction and installation of an item of physical or intangible property, or the total acquisition costs of a group of such items ( Stewart, 1995a). Conversion costs: A grouping of direct labour and manufacturing overhead into a single summary cost element (Shuford, 1995). Development costs: Costs of a system up to the point where decision is made to procure an initial increment of the production units or the operational system (Stewart, 1995a). Disposal costs: The costs of disposing of a facility, property item, equipment item, scrap, byproducts or excess material (Stewart, 1995a). Life-cycle costs: All costs incurred during the projected life of the system, subsystem or component (research, development, test, evaluation, production, maintenance and disposal) (Stewart, 1995a). Opportunity costs: Loss of income due to not selecting the optimum alternative from a financial point of view (Liebers, 1998)(Blommaert, 1998). Prime costs: Costs of direct material and direct labour (Shuford, 1995). Removal costs: The costs of dismantling a unit of property owing to retirement from service (Stewart, 1995a). Sunk costs: The total of all past expenditures or irrevocably committed funds related to a program/project (Shuford, 1995). 6


2.1.2 Cost allocation The allocation of cost is important for a correct interpretation of costs. Cost allocation is a method or combination of methods that results in a reasonable distribution of costs (Stewart, 1995a). For direct costs, this allocation is straightforward. The costs can be calculated from: C = P × Q

(2.1)

with: C : the costs P: the price variable Q: the allocation base

In the case of direct labour costs, C is the direct labour costs, P is the hourly rate and Q the number of hours. In the case of indirect costs, the allocation base has to be defined. After this, the price variable can be calculated from: P=

C Q

(2.2)

with: P: the price variable, usually called rate C : the indirect quantity Q: the allocation base

The values of the indirect quantity and the allocation base can be obtained from historic information or from prognoses or a combination of both. If, for example, the total direct costs are chosen as the allocation base for the total indirect costs, P is the direct cost burden rate, C the total indirect costs and Q the total direct costs. When the rate is known, the costs can be calculated with equation 2.1. This example illustrates the way in which indirect costs are calculated with the traditional costing method. With this method, the overhead is allocated to products using volume based allocation bases e.g. labour hours, machine hours. When the allocation base is chosen incorrectly, incorrect conclusions can be drawn from the indirect costs. When the indirect costs are calculated with the direct cost burden rate, it would mean that every product with high direct costs also has high indirect costs, which is not the case. Therefore, this can lead to wrong conclusion about the cause of costs. The ratio between direct costs and indirect costs has changed drastically over the past decades because of increasing automation, in both machinery and computers, as illustrated by Figure 2.5 (Thompson, 19??). In the 1950’s, the indirect costs were only a small part of the total product costs while direct labour constituted the biggest part of the total product costs. Therefore, it was not necessary to estimate the indirect costs in a very detailed and accurate manner and traditional costing was an adequate way to calculate the overhead.

Figure 2.5 The components of product costs in the United States (Thompson, 19??).

7

Chapter 2

Nowadays, the opposite situation is true. Overhead constitutes the biggest part of the total product costs while the direct labour costs are only a small part. The part of the material costs in the total costs has hardly changed. Because of this change, it has become necessary to calculate the overhead more accurately and more detailed. The allocation of the overhead requires the use of other allocation bases that allocate the overhead in a more realistic way. Other methods for the allocation of overhead costs will be discussed in section 2.5.

2.2 The use of cost information in manufacturing Different engineering tasks have different information available for generating relevant cost information. In addition, the tasks use different kinds of cost information for different purposes. In section 2.2.1, an overview will be given of the use of cost information for the next tasks: design, process planning, production planning and management. This overview indicates a considerable interrelation between the different tasks and the information that is used. Therefore, the importance of an information management system will be indicated in section 2.2.2

2.2.1 Task oriented cost information In the embodiment design phase, decisions about materials, surface roughness, tolerances, shape, dimensions, production methods, etc. have to be made. Because all decisions are mutually dependent, a decision about one aspect can lower the costs for t hat aspect while increasing the costs of another aspect. For instance, the selection of a cheap material can lead to extra operation steps in order to achieve a certain surface tolerance, which increases the production costs. Several of these dependencies dependencies are present. From an analysis of the product design process and its decision making it can be concluded that the costs fixed f ixed during product design are caused by the following interrelated cost drivers: geometry, material, production processes and production planning (Weustink, 2000). Next, the interrelation of the cost drivers will be illustrated. The interrelation of the cost drivers causes the interrelation between the engineering tasks, which influences the use of cost information by these engineering tasks. Interrelated cost drivers The geometry includes the shape, dimensions, accuracy, etc. and it determines the material quantity and the production processes that are required. The influence of the geometry on the product costs is obvious. For instance, a higher level of accuracy requires more accurate resources (e.g. machine tools, equipment), which can result in higher production costs. The shape can also cause increasing production costs; a pocket with straight corners requires generally a more expensive machining process (e.g. electrical discharge machining) than a pocket with rounded corners (made by e.g. milling). Material is one of the most obvious cost driving product characteristics, because material costs constitute a large part of the total product costs (see for instance Figure 2.5). Production processes are required to transform (raw) material into a component or to assemble components and/or assemblies into higher-level assemblies. Resources (e.g. operators, machine tools, tool sets, fixtures) are needed to perform the required production operations. The resources have a certain capability, which means that resources have technical restrictions, concerning the power of the machine tool, accuracy, maximum dimensions of the workpiece, etc. The limited capability of resources restricts the execution of certain operations. This has to be taken into account in the planning phases of the product development cycle. The type of production method selected has a significant influence on the t he production costs. Besides the technical restrictions, resources have also logistic restrictions, which means that a resource will not always be available. With the knowledge of these restrictions, operations can be allocated to the available resources. It is necessary to ensure the due date, because otherwise a fine could be incurred. These time, and therefore, cost consequences must be regarded in the planning phases. 8


The goal of production planning with regard to costs is to minimise the variable costs that are a result of production planning decisions (Giebels, 2000). The variable costs to be regarded are: extra payments for overtime work, the price for subcontracting minus the variable costs for in-house production, inventory costs due to excessive Work In Process (WIP), lateness cost (in particular penalty costs) and earliness costs for short time delivery of blank materials. A prototype of a decision support system for integrated order planning, which incorporates these variable costs, will be discussed in more detail in section 6.2.4. The cost drivers and the engineering tasks In the design phase, the primary decisions are concerned with geometry and material. For proper decision-making, it is advantageous to have information from other engineering tasks, usually performed later in the product development cycle, like process planning and production planning. Initially the designer can get benefit out of cost information related to geometry and material. When cost information about production processes and production planning would be available, this would be of great help to the designer. Because geometry does not cause costs directly, the designer cannot calculate costs for geometry. A way to solve this problem is to use cost information from products that have been manufactured in the past. The assumption is made that geometrically equal products (of the same material) will cost the same. This method will be discussed in section 2.4.2 in detail. This method can give cost information quickly because no other engineering tasks, like process planning, are required to generate more information. The designer can estimate the material costs relatively easy when he has access to a material database, which also contains cost information. Another way to solve the problem is to perform other engineering tasks, like process planning, and to use the generated information for cost estimation. The designer must frequently choose between alternative solutions. In the choice between alternatives, the overhead costs are usually less important because they are made anyway. Therefore, only the direct costs are considered in choices between alternatives. In the process-planning phase, the primary decisions are concerned with production processes. A cost based decision between production processes is difficult because cost information about a process largely depends on the resources that are used. Similarity, based on production processes, with products from the past could be used in this case. When resources are selected, the extent of use can be estimated and costs can be calculated with the appropriate cost rates. The choices about the production processes are made based on the technical constraints and t he technical ability of the resources. If the availability of resources would be incorporated in the decision making of the process planner, the choices made would probably be more adequate. The unavailability of a resource can result in the use of a more expensive resource than chosen by the process planner, consequently leading to higher costs. In the production-planning phase, the primary primar y decisions are concerned with production planning. Usually, production planning is one of the last phases in the product development cycle before production starts. Production planning determines which resources are used and when they are used. The final decision about which resource will be used for a product, not only influences the costs of that product but also the costs of other products that have to be produced. When a resource is selected for one product, this resource cannot be chosen for another product, which could mean that a more expensive resource has to be chosen for the other product. For the production planner, besides the product costs, the costs of all products in the manufacturing cycle can be important. The primary interest of management is information about the costs and revenues at the end of a certain period, rather than the costs and revenues of a specific product. For the analysis of the costs, it is advantageous to have the costs split up in the different types. Furthermore, based on the analysis of the resource utilisation rates, management can adapt the resources rates. The cost information about products and resources can be used when considering buying new resources.

9

Chapter 2

2.2.2 The need for information management From the previous section, it can be concluded that decisions of different engineering tasks influence each other. Furthermore, engineering tasks use information from and/or generate information for other engineering tasks. In order to tune the need for and the availability of information from different engineering tasks, the structuring of information has to be unified and communication between the engineering tasks has to be made possible. When concurrent engineering is applied in the product development cycle, the need for communication increases even more. Concurrent engineering, i.e. the simultaneous execution of shared tasks by separate departments and the control of cooperative decision-making, requires additional tuning of engineering. The possibility of communication between the engineering tasks is based on both the availability and accessibility of coherent information (Lutters, 1997a). In the automation of the product development cycle, engineering databases play a key role (Billo, 1987). An engineering database can contain geometric, physical, technological and other properties of “technical” objects and the relations between these properties (Billo, 1987). Usually, the engineering tasks require information from multiple databases. Therefore, for the integration of engineering tasks in the product development cycle an information management system is indispensable. In chapter 3, an information management model based on the accessibility of transparent information will be discussed.

2.3 Cost control The function of cost control is twofold. On the one hand, it has to detect cost values and sources of these costs. It can initiate a well-founded product modification in order to keep costs within a predetermined range or to cut costs in general. On the other hand, it must be possible to compare cost estimates with the actual costs. In this way, cost models can be improved. The feedback of cost information is an essential part of cost control. A way of describing and understanding a complex system such as the manufacturing system is the decomposition of the system. A possible representation of decomposition is a reference model. A reference model represents a system as an organisation in terms of its structure of relatively independent, interacting components, and in terms of the globally defined tasks of these components (Biemans, 1989). Many reference models for the manufacturing system have been developed in the last decades, see for example (Lutters, 2001). In section 2.3.1 the manufacturing planning and control reference model of Liebers will be discussed. This reference model was developed especially to clarify the relation between cost control and manufacturing. When the position of cost control in the manufacturing system is known, the cost control component can also be decomposed. Another possible representation of decomposition is an architecture. An architecture is a framework which defines the functions, which are required to perform the task of a system, with their input and output (Arentsen, 1995). The cost control architecture that was developed in the context of the manufacturing planning and control reference model of Liebers will be discussed in section 2.3.2.

2.3.1 The manufacturing planning and control reference model The manufacturing planning and control reference model of Liebers depicted in Figure 2.6 consists of three main components, namely planning, execution and control. The three main components are subdivided into several sub-components. In the reference model, hierarchical planning is assumed. The four planning & control levels that are discerned are the strategic level, the tactical level, the operational level and production. However, for the implementation of the components other planning and control levels can be used.

10


Figure 2.6 The manufacturing planning and control reference model of Liebers (Liebers, 1998). Planning The main component planning consists of the sub-components strategic planning, technological planning and logistic planning. Technological planning is again subdivided into process planning & resource engineering, product engineering and process planning. Logistic planning is subdivided into resource location planning and order planning. The task of planning is the generation of a valid production plan. Strategic planning decides on the goals of the organization and the strategies for attaining these goals. Closely related to strategic planning is the selection of types of products and the amounts of products to be manufactured. The policy of how to deal with (potential) customers, vendors, makebuy decisions, hourly resource rates etc. are determined by strategic planning. Technological planning generates all required technological information, i.e. the process plan. Process and resource engineering deals with the development of processes and resources as well as the planning of the technical aspects of maintenance, service and repair. Product engineering and process planning comprises the generation of product information and the selection of processes and resources required for the production of the product. The final allocation of work to resources is done by logistic planning, it determines the time frame for the technological applicable resources, i.e. the production plan. Resource location planning sees to it that the right resources, including materials, are in the right place at the right time. Order planning selects the final resources from the set of technically applicable resources determined by process planning. Furthermore, it assigns work to time periods and it decomposes work into smaller units.

11

Chapter 2

Execution Plan execution creates the physical product according to the production plan. During production several deviations between the required product and the realised product occur. This can lead to rework that also requires additional planning. Control The main component control consists of the sub-components data collection & processing , strategic control, (functional) quality control, (delivery) time control and cost control. In order to be able to detect deviations between the planned product and the resulting product, it is necessary to collect data by monitoring the plan execution. The data can refer to the product or the processes. After processing the collected data, models for the planning activities can be created or adjusted. Strategic control determines the targets for the performance indicators and the initial solution space for planning activities. The three control components for the performance indicators (functional) quality, (delivery) time and cost have to evaluate the results of a planning and execution activity for each indicator. For the evaluation, the consequences of the results of an activity on the performance indicators have to be compared with the targets set for the performance indicators.

For the development of a valid production plan, frequent communication between the components is necessary. Communication with the company environment is dealt with by strategic control. The targets for the performance indicators are passed from strategic control to the other control components. Because the performance indicators affect each other, communication between the control components is essential. The input for the planning components comes from strategic planning. Communication between the components of technological planning and between the components of logistic planning is required. In addition, communication between technological and logistic planning is required. Although a reference model does not contain information flows, the information flow in the reference model is divided into a feedforward and feedback part. In the feedforward part, the planning and execution components generate and process information for the execution of other components. In the feedback part, the control components process and generate information for the improvement of the execution of other components. The main advantages and disadvantage of the reference model of Liebers (Figure 2.6) are listed in Table 2.3 (see also Liebers, 1998).

Advantages

Disadvantage

The reference model incorporates both manufacturing planning and manufacturing control. Cost control is part of a generic manufacturing planning and control reference model with generic planning and control components. Overall control is incorporated by the separate control components for the performance indicators (quality, time and costs) and the different levels of planning and control (strategic, tactical, operational, and production). Communication on all planning and control levels is possible. The planning & control levels are independent of the implementation of the components of the reference model. The reference model can be linked to reference models in business economics, incorporating cost control, thus enabling to incorporate financial control. The reference model does not have a desired single information structure. The three performance indicators limit the utilisation of the r eference model.

Table 2.3 Advantages and disadvantages of the manufacturing planning and control reference model of Liebers

12


2.3.2 The (cost) control architecture According to Liebers the cost control component can be decomposed into four functions: cost estimation, production monitoring, cost calculation & evaluation and cost modelling (Liebers, 1998). The functions of cost control and their input and output are depicted in Figure 2.7.

Figure 2.7 The generic cost control architecture of Liebers (Liebers, 1998).

The cost estimation function generates cost estimates based on the specification of a solution by a decision maker and a cost model with cost rates. The production monitoring function has to collect all actual relevant information from the execution of the production plan. The actual manufacturing data is passed on to cost calculation & evaluation and cost accounting. Based on the actual manufacturing information cost calculation & evaluation generates the actual costs. These actual costs are compared with the cost estimates with the underlying assumptions. Based on this comparison, the actual costs and actual manufacturing information is passed on to cost modelling. Cost accounting generates cost rates based on the actual manufacturing information. Based on the actual costs and actual manufacturing information, cost modelling can improve the cost models. Four feedback loops can be distinguished in the architecture: the engineering and planning loop, the order acceptance loop, the production loop and the accounting loop. The engineering and planning loop provides decision makers with cost information about different alternative solutions. In this case, qualitative information would be enough in order to make a choice between alternatives, but quantitative information can be used as well. The decisions made for one product obviously influence the total costs in a company. Therefore, the order acceptance loop provides cost information to the decision maker about the cost consequences for all products. In this case, the cost information has to be quantitative cost estimates. In the production loop, information from the actual production of a product is fed back in order to compare the cost estimate for the product with the actual costs for that product. Based on this comparison, the cost model can be improved. In the accounting loop, information from production over a certain period is fed back. In this case, the estimated costs of the period are compared to the actual of the period. Based on this comparison, rates can be improved. The advantages of the cost control architecture are listed in Table 2.4.

Advantages Disadvantage

The architecture is a generic control architecture. The individual functions are generic functions. Short term and long term control loops are incorporated. The architecture does not have a desired single information structure. Table 2.4 Advantages of the cost control architecture of Liebers.

13

Chapter 2

2.4 Cost estimation One of the four (sub-) functions of cost control is cost estimation. Within manufacturing, cost estimation is the procedure of approximating the cost of manufacturing a product before all stages of the product development cycle have been executed, based on the information available or that can be collected at the stage of the product development cycle. Two basic approaches for cost estimation can be distinguished: generative cost estimation and variant based cost estimation. The principle of generative cost estimation is the composition of the costs from its constituents while variant based cost estimation uses similar products manufactured in the past to determine the costs. The difference in applicability of both approaches is illustrated by means of the cost estimation paradox depicted in Figure 2.8. Because variant based cost estimation uses information from products manufactured in the past, more information is available in the beginning of the product development cycle than is the case with generative cost estimation. Furthermore, it is possible to combine the two approaches in a hybrid cost estimation method. The next three sections will elaborate on these three approaches.

Figure 2.8 The cost estimation paradox for generative and variant based cost estimation (Bode, 1998a).

2.4.1 Generative cost estimation When cost estimation is based on a decomposition of the expected production processes, the approach is called generative cost estimation. The costs caused per production process, direct or indirect, have to be calculated. Consequently, generative cost estimation will largely depend on process planning information. An important issue in generative cost estimation is the aggregation level of the manufacturing activities on which the costs occur and have to be determined. Figure 2.9 depicts an aggregated cost division (modified ABC Hierarchical Model (Cooper, 1991)), which illustrates where the most common costs occur. The most common lowest level is the (form) feature level. A form f eature is a group of faces of a product that together have an engineering meaning. Different types of form features are distinguished, the most common being design features and manufacturing features. These features have a specific engineering meaning for designers and process planners respectively and do not necessarily consist of the same faces. Two reasons to use design features to assign costs are (Wierda, 1991): 1. Cost functions are derived for classes of similar objects. These objects can serve as the building blocks for global cost estimation. Features are very useful objects, because they have an engineering meaning and they are used commonly. 2. A designer is interested in the causes of costs. If the costs can directly be linked to design features, the designer is able to influence the costs directly. In fact, these reasons are valid for any kind of feature. Because of these considerations feature based cost estimation is often aimed for. 14


Figure 2.9 Aggregated costs division (Cooper, 1991).

From an analysis of different types of costs, three categories of costs related to design features, can be distinguished (Wierda, 1991): 1. Costs that can be assigned directly to individual design features. 2. Costs that are incurred for a collection of design features. 3. Costs that cannot be assigned reasonably to design features. These three categories are valid for any class of similar objects to which costs are related and illustrate the difference between the allocation of direct and indirect costs. The costs that can be assigned directly to design features are costs on feature and assembly level. The costs incurred for a collection of design features are costs on component and batch level. The costs that cannot be assigned reasonably to design feature are costs on order and facility level. When no feature interrelationships are considered, resource selection can be done for every feature separately. Therefore, production times and consequently costs are easily calculated per feature. In (Geiger, 1996) specific time formulas are developed for each standard (design) feature. The time formulas use feature parameters and machine rates. It is also possible to assign machining operations to surfaces and to calculate the costs (Kiritsis, 1996). Some consider a rough process plan to be sufficient for manufacturability and cost analysis (Schaal, 1993). A rough process plan consists of process plans for every feature. A feature process plan only takes into account information about features and information from one higher aggregation level, which could mean the component level. The separate process plans have to be unified based on manufacturing rules. When used during design, for instance the sequence of operation steps does not matter for cost calculation since at this stage the accuracy will be low anyhow because only less detailed information is available, so a cost estimate can be generated based on a rough process plan. When more detailed production information becomes available, cost estimates with higher accuracy can be obtained by using more detailed process plans. The production of a component usually starts with a blank, to which the material costs can be related directly. The material costs can be related directly to the blank, which is usually considered 15

Chapter 2

a design feature. However, additional design features usually affect the material costs, which have to be accounted for. In (Wierda, 1991) a possible method is described: • If a (design) feature implies a positive volume, material costs can be directly assigned to it. • If a (design) feature implies a negative volume, negative material costs (waste revenue) could be directly assigned to it. However, a negative volume could also be created directly by casting or injection moulding for instance, which would not imply any removal of material. Complications are possible when positive and negative volumes of different features overlap. Furthermore, often parts of the blank that lie outside the final product are not described by design features. Therefore, the material costs of these parts cannot be assigned to design features directly. Further complications as a result from intersecting features as may arise in prismatic parts do not occur frequently in the sheet metal domain. Most operations are carried out for groups of features in one set-up. The costs involved with these tasks have to be assigned reasonably to the features involved. However, the use of such assignments can be confusing and misleading (Wierda, 1991). When for instance a feature is removed, the costs will not drop with the costs of that feature since the costs are incurred anyway resulting in a cost increase of the other features involved. A related problem is that the assignment of component costs to individual features in order to enable pure feature based costing will result in an attempt to minimise the costs for the individual features, while the overall costs will not necessarily be minimised. To overcome these problems (Wierda, 1991) suggests the use of highlevel features. Considering the remarks about component costs, these high-level features have to include the component and the assembly. It is therefore sufficient to calculate the costs on the product levels on which they occur without assigning them to the feature level or any other lower level. For assembly, batch, order and facility level the same remarks about the costs (except the material costs) on component level apply. For the costs on assembly, batch and order level usually a direct relation exists with one of the product levels and therefore these costs can be allocated to a product level relatively easy. The relation between the costs on facility level and product level is usually not straightforward and has to be defined explicitly, see section 2.5.2.

2.4.2 Variant based cost estimation For manufacturing companies, the increasing diversity of customer’s demands has led to high product variety. If high product variety is not taken into account in the product development cycle, it will result in inefficient manufacturing. Examples of inefficiency are large set-up times, high tooling costs, large work-in-progress, large throughput times (Srikantappa, 1994) and complex production planning. A manufacturing concept that overcomes the problems related to high product variety is group technology (GT). Mitrofanov, one of the first authors to write about GT, defined GT as a technique for manufacturing small to medium lot size batches of parts of similar process, of somewhat dissimilar materials, geometry and size, which are produced in a committed small cell of machines which have been grouped together physically, specifically tooled, and scheduled as a unit (Srikantappa, 1994). More recent broad definitions state GT as the technique of applying the same solution to similar problems. Therefore, GT can eliminate duplication and redundancy (Baer, 1985). The essence of GT is the comparison of a new product with products that have been manufactured before. This comparison requires the determination of similarity criteria, which is very important for the proper use of GT. Application of GT in cost estimation, i.e. variant based cost estimation, means that the cost estimate for a product under consideration is based on the actual costs of similar products manufactured before. In this case, the similarity criteria have to be based on cost driving product characteristics. The automation of GT in an integrated product development cycle using engineering databases is mainly determined by the layout of the engineering databases. It is possible to create an interface between the GT system/database and the other systems/databases or it is possible to maintain duplicate databases (Billo, 1987). From an information management point of view these solutions 16


are unsuitable, see chapter 3. A better solution would be to map the GT onto the general structure of the engineering database. A general-purpose, relational database is the best alternative in the case of GT (Bill, 1987). In GT, different approaches can be distinguished: classification, clustering (Kusiak, 1990) and production flow analysis (Agarwal). Classification is the process of grouping parts into families of similar parts; similarity is based on some set of rules and principles (Agarwal, 1994). The families of similar parts are identified by a code, representing similarity criteria of interest. Therefore, coding is the arbitrary assignment of one or more symbols to a part, which, when deciphered, communicates specific meaning or intelligence (Mosier, 1993). Three types of code structures exist (Agarwal, 1994): Hierarchical (monocode) structure

Chain (attribute, or polycode) structure

Hybrid structure

Principle

Each character is a further expansion of the previous character

The meaning of each character is independent of any other character

A combination of the monocode and polycode structure

Advantages

Information can be represented with a relatively small number of digits

Simple to implement

The advantages of the monocode and polycode structure

Disadvantages

Complicated and very difficult to implement

A large number of digits may be required to represent information

-

Examples

DCLASS (Brigham Young University)

MICLASS (TNO)

OPITZ (Opitz), KAMKODE (Kamrani)

Table 2.5 Code structures (Agarwal, 1994).

In literature, it is generally assumed that no universal method exists for classifying and coding parts (Bear, 1985), (Lewis, 1987), (Agarwal, 1994). Therefore, most research is concentrated on the development of various classification systems to support specific engineering tasks, specific production processes and product types. Classification can be based on geometrical information. This type of classification is particularly suitable for the support of the designer. In the context of cost estimation, the assumption is made that equal geometry is produced in the same way. This implies that equal geometry will cost the same. This assumption can be refined by considering cost factors like tolerances (Molengraaf, 1993). In the case of equal geometry and equal cost factors, it is more likely that the probable production processes and therefore the costs will be the same. In order to include manufacturability aspects, technological information is incorporated in classification (Lewis, 1987), (Peklenik, 1980), (Wu, 1992), (Luong, 1989). This type of classification is suitable for both process planners and designers. Even for some specific tasks of process planning, classification algorithms can be developed like for fixturing (Nee, 1992). Examples of classification for specific processes and product types are: sheet metal, die-casting, forming, machining, joining/welding, heat treatment and finishing (Lewis, 1987), (Agarwal, 1994), (Greska, 1995). It is possible to combine multiple processes in one system ( Agarwal, 1994). Advantages of classification related to cost estimation are (Schuttert, 1995): • Quick retrieval of historic data; • Improvement and consistency in cost estimation; • Rationalization of cost estimation; 17

Chapter 2

•

Support in quotation procedures and acquisition of new machine tools. However, classification systems have several disadvantages (Vliegen, 1993): • Insufficient retrieval chance; • Dependency on human interpretation; • Excessive coding effort; • Lack of flexibility, feedback of the performance and integration with other systems; Inflexibility is caused by the fact that only a limited set of characteristics is used. These geometrical and technological characteristics, of a selected part range, are chosen empirically and evaluated statistically (Peklenik, 1980). • Discussions about how to classify and who should classify; • High investments. Besides the primary function of classification systems, being basic retrieval of information, classification systems have to be usable for all interested parties within the firm (Mosier, 1993). Coding should reflect the patterns that have emerged in analysis and be compatible with the company’s other systems (Bear, 1985). Clustering is the process of grouping similar objects based on a similarity coefficient (Agarwal, 1994). A similarity coefficient represents the similarity between two objects and usually ranges form 0 to 1. Different characteristics can be used for the calculation of a similarity coefficient. Four types of similarity coefficients are usually distinguished: distance coefficients, association coefficients, correlation coefficients and probabilistic coefficients. Production flow analysis is the process of grouping products based on the sequence of operations. It consists of four steps: factory flow analysis, group analysis, line analysis and tooling analysis (Agarwal, 1994). From the description of clustering and production flow analysis, it can be concluded that both methods are very similar to classification. Clustering can be seen as temporary classification with only one part family. Only the objects with a sufficiently high similarity coefficient constitute the part family for the object under consideration. This means that the part family to be used is created every time the part family is needed. Therefore clustering is more adaptable and flexible than classification. Production flow analysis can be seen as classification, in which the creation of part families is based on the sequence of operations.

2.4.3 Hybrid cost estimation Both variant based and generative cost estimation can be applied at the same time for one product resulting in a hybrid cost estimation. In the development cycle of products, it can occur that different parts of a product will be in a different phase of the product development cycle. Therefore, the available information of different parts of the product will be different. When the costs of different parts of a product are calculated in a different way, the total product costs can be calculated by summing the costs for the different parts. When different cost models are used, a prerequisite is that the calculation of the overhead costs is carried out in the same way. If the overhead costs are calculated in a different way, it can occur that some overhead costs are counted more than once or that some overhead costs are excluded. In order to ensure a consequent calculation of the overhead costs, an aggregated product information structure and cost structure is required. Only in that case, it is possible to store the way the overhead costs are calculated and on which aggregation level the overhead costs are calculated.

2.5 Cost modelling Cost modelling represents the determination of the data, ground rules, assumptions and equations that permits the translation of resources or characteristics into costs (Stewart, 1995a). The most common steps in cost modelling are: 1. Determination of the scope, i.e. the costs subdivided in different types, which have to be modelled. 2. Determination of the allocation base for the (overhead) costs. 18


3. Determination of the cost functions, i.e. the relation between the product parameters and the costs. The main difference between different cost models occurs in step two. Furthermore, cost models are distinguished based on the scope they cover, determined in step one. For step three, different techniques are available.

2.5.1 Determination of the scope A cost model must cover all the costs so that all expenses of a company are included. The costs for different stages of the product development cycle can be assigned to different groups for analysis purposes, see for example Table 2.6. The position of some costs can be argued, but the table gives a fair idea about the costs companies should include in cost models.

-

Design

Production

Usage

Disposal/recycling

Company cost Market recognition Development Materials Energy Facilities Wages, salaries etc. Transportation Storage Waste Breakage Warranty service

Users cost

Society cost

- Waste - Pollution - Health damages -

Transportation Storage Energy Materials Maintenance

- Disposal/recycling dues

-

Packaging Waste Pollution Health damages

-

Waste Disposal Pollution Health damages

Table 2.6 Life-cycle stages and costs from Alting (Asiedu, 1998)

Lately, environmental aspects have influenced production; it has been recognised that the product development cycle and the product life cycle have to be extended with a disposal and/or recycling phase. Some costs from these phases have to be incorporated in the cost model. Although usually the perception exists that environmental aspects have to be treated apart, the costs concerned with the disposal and recycling phase can be treated in the same way as any other phase that has been discerned.

2.5.2 Determination of the allocation base In section 2.1.2 the principle of allocation was explained by means of the direct costing method. Because of the changing ratio between direct costs and overhead costs, other, more accurate methods have been developed. A costing concept that allocates overhead in a more realistic way is Activity Based Costing (ABC). The basic assumptions of ABC are (Cooper, 1991), (Shuford, 1995): • Costs are caused by activities; • Products consume activities. ABC can be implemented by following the next steps (see also Figure 2.10): 1. Determination of activity centres; Activity centres, sometimes called cost centres or responsibility centres, are parts of the product development cycle for which the costs are registered separately on request of management (Blommaert, 1998). These administrative units, which have managerial responsibility, are basic

19

Chapter 2

units of control in cost accounting (Stewart, 1995a). For i nstance, the department distribution can be an activity centre. 2. Determination of activity pools; Activity pools, also called cost pools, are sets of activities practised by a certain function (Blommaert, 1998). An activity pool can consist of only one activity. In the department distribution loading, unloading and transport could be identified as activities. In this case, the cost pools could be ‘loading and unloading’ and ‘transport’. 3. Determination of the allocation base per activity pool; In the context of ABC, the allocation base is often called cost driver. An allocation base is a measure that is directly related to the amount of an activity used. 4. Determination of overhead costs per activity pool; Usually, the overhead costs per activity pool are based on the overhead costs in the previous year and a forecast for the next year. 5. Calculation of overhead cost per cost driver (rate). The overhead costs per activity pool are divided by the budgeted quantity of the allocation base.

Figure 2.10 ABC implementation outline.

The overhead costs of a specific product can be calculated by multiplying the rate with the quantity of the allocation base that is used for the product.

2.5.3 Determination of the cost functions Cost functions define the relationship between costs and manufacturing characteristics, i.e. cost parameters. These cost functions of a cost model can be obtained in different ways. A simple solution is to use cost functions from literature. These will most likely indicate a proper dependency of cost (types) on cost parameters, but the degree of dependency will probably be context specific. Therefore, cost functions from literature have to be adapted for the new situation in which they are going to be used. Another straightforward solution is to use common sense to derive cost functions (Vin, 1995). Several mathematical techniques to determine cost functions are available. Examples of techniques, which use historical information to deduce cost functions, are: regression analysis and neural networks. Both techniques try to find a relation between cost parameters and costs by examination of sets of data containing the values of cost parameters and their corresponding costs. The basic principle of regression analysis is that it approximates the relation between the costs and the cost parameters with a parametric function, based on a best-fit of historical data. In terms of regression analysis, the costs are the dependent variable and the cost parameters are the independent variables. One of the simplest parametric functions is a linear function with one independent variable (equation 2.1). This simple regression analysis can be extended to multiple regression analysis by adding more independent variables to the equation.

20


D = α 0 I + α 1

(2.1)

with: D: the dependent variable I : the independent variable i: constants

The values of the constants can be found by fitting the function to historic data, i.e. sets of cost values with the corresponding values of the cost parameters. It is also possible to use non-linear functions in order to obtain a better fit. Several criteria for a good fit can be used, the most common criteria is the criterion of least squares. A more detailed description of regression analysis can be found in appendix C. In contrast to regression analysis, neural networks do not assume a parametric function on beforehand. However, the configuration of a neural network has to be chosen in advance. It is difficult to find the proper network and it is not possible to verify that a network is the best network, therefore the configuration of a network is a trial and error process (Zhang, 1996). However, there are neural network toolboxes with optimisation algorithms, which can help to form a proper network (Geiger, 1997). A neural network is a simplified model of the anatomy of the human brain; it consists of neurons (brain cells) and synapses (nerves). A neuron maps one or several input signals to an output signal by means of a mapping function. The signals are t ransmitted between the neurons through the synapses. The neurons are organized in layers. A simple mapping function is a threshold function i.e. if the accumulated input signals are above a t hreshold value, the output signal equals one or else the output signal equals zero. A neural network is trained by supplying it with historic data, i.e. sets of cost values with the corresponding values of the cost parameters. It will fit the data by using a learning function. The most prominent learning function is the error backpropagation algorithm (Bode, 1998a). This training will result in the functional relationship between the costs and the cost parameters. A neural network, like regression analysis, also uses statistical curve fitting procedures, but it applies them recursively to each neuron. A more detailed description of neural networks can be found in appendix D. The advantages and disadvantages of regression analysis are listed in Table 2.7 and the criteria for using regression analysis and neural networks according to (Bode, 1998b) are listed in Table 2.8.

• Neural networks

• • • •

Regression analysis

•

Advantages Training of the network can be done in a way that the Mean Squared Error decreases (see appendix D). It is possible to detect hidden relationships (Bode, 1988b). Any continuous function can be approximated (Bode, 1988b). A measure for the quality of fit is available (see appendix C). Conditions for the quality of the Least squares method exist (see appendix C). It is possible to check whether the independent variables are significant parameters (see appendix C).

• •

•

Disadvantages Configuration is difficult to control. The training process is difficult to control (Bode, 1998b).

A relationship between the costs and cost parameters has to be assumed on beforehand.

Table 2.7 Advantages and disadvantages of the use of neural networks and regression analysis (Bode, 1998b).

21

Chapter 2

Number of similar cases The attributes, which have a cost effect. Number of cost drivers The way the drivers influence the cost

Neural networks Quite a few. Quite certain. Few. Unknown.

Regression analysis Quite a few. Known precisely. Few. Quite certain.

Table 2.8 Criteria for a selection of methods for the determination of cost functions (Bode, 1998b).

When the cost parameters are unknown and the type of function is unknown, neural networks are suitable to deduce cost functions. However, it is easier to quantify the quality of the results of regression analysis, which allows a better quantification of the quality of the cost estimates calculated with the resulting cost functions. Furthermore, because in the case of regression analysis the parameters have to be chosen in advance, the relation with the costs will be clear. In the case of neural networks, this will not always be the case.

2.6 Sheet metal manufacturing For the production of sheet metal components, a number of different production processes are available. The choice between processes is largely determined by the technical constraints of the products to be produced. However, the choice must also be based on costs. In order to get more insight in the cost factors of sheet metal production, the most common production processes will be discussed. For most of these processes, cost functions and time functions have been derived. Furthermore, general cost factors of sheet metal manufacturing will be discussed. Typical sheet metal production operations are: cutting, joining bending and finishing (see Figure 2.11). In this section, the operations cutting, joining and bending will be discussed.

Figure 2.11 Overview of the most common sheet metal production processes (Veltman, 2000).

2.6.1 General A typical characteristic of sheet metal manufacturing is nesting. Nesting is the procedure that determines which parts of products or blanks are put together in sheets. Furthermore, it generates a layout of the parts and blanks for each sheet. Nesting has a large impact on the execution of many engineering tasks, such as process planning, production planning and cost estimation. Usually, 22


client orders are split up in production orders, containing one product. Next, the production orders are divided into a number of batches containing a number of products with the same (internal) due date. The nesting procedure can cause that one batch is distributed over several sets of differently nested sheets. Therefore, one sheet can contain different parts of different products. The selection of optimal tools for a sheet can result in the use of non-optimal tools for a batch in this sheet. Four different situations can occur in nesting (Vin, 1995): • 1 batch of products nested in 1 sheet • 1 batch of products nested over m sheets • n batches of products nested in 1 sheet • n batches of products nested in m sheets For the first two situations, the costs for a batch equals the costs for the sheet(s). In case of the last two situations, the costs incurred for the sheet(s) have to be split up for the different batches. Once again, an allocation base has to be determined for the proportional distribution of the costs. In small batch part manufacturing, the third situation is common practice (Vin, 1995).

2.6.2 Mechanical cutting Shearing Usually, blanks of suitable dimensions are cut from larger sheets or a coil by means of shearing. In the shearing process, the sheet is subjected to shear stresses developed by means of an upper and lower blade. The shearing process is seldom mentioned in literature about cost and time functions. Punching In punching, shear stresses are developed by means of a punch and die. Punching removes holes and notches in contours of products with a single punch stroke or with a number of partially overlapping punch strokes.

In (Vin, 1995) five types of cost centres are distinguished for punching: machine time, labour, material, storage and transport. Only the direct manufacturing costs are considered. The processing costs are the sum of machine running costs, labour costs, machine set-up costs and tooling costs. The total processing time (Appendix A, equation A.1) is the summation of the processing time (A.2), tool change time (A.3) and traversing time (A.5). In case a sheet consists of parts from several batches, the number of tool changes has to be divided over the batches. Equation A.4 splits the number of tool changes based on the number of different parts. This estimated sharing ratio is also used to divide set-up and tool load times. The original equation of the traversing time was modified after a time study for one example revealed that it overestimated the traversing time. Now, equation (A.5) is based on a non-optimal zigzag-traversing pattern. The deduction of the appropriate rate(s) is not dealt with. Nibbling Nibbling also removes material with a punch and die. In contrast with punching the shape of the material removed is not only determined by the punch but also b y the trajectory of the tool. Material is removed with a relative small punch by making a number of subsequent, partially overlapping punch strokes.

In (Vin, 1995) the processing time for nibbling is derived in the same way as it is done for punching. The equations for punching for tool change time (A.3), number of tool changes (A.4) and traversing time (A.5) are the same for nibbling. The total processing time and nibbling time are calculated with equation (A.6) and (A.7). In (Nollet, 1993) time and cost functions are derived for nibbling. The total processing time is the sum of the set-up and work piece time (equation A.8). The work piece time is the combination of the main time and the auxiliary time raised by 5% for personal care. The main time includes time to 23

Chapter 2

go to the start position, punching times, traversing times and tool change times. A distinction is made between punching of holes, nibbling straight contours and curved contours. The auxiliary time includes the fixing and unfixing time and the time for scrap removal. The costs are split up in fixed costs, variable costs and technology dependent costs (A.18). The fixed costs rate includes depreciation of the machine and the space (building) that the machine occupies (A.19). The rate for the variable costs contains the costs for energy, labour, maintenance and programming (A.20). The tool costs are considered to be technology dependent costs (A.21). The material costs are not considered because the total nibbling costs are compared to total laser cutting costs, which would be the same in both cases. Special tool punching Special tool punching is also quite similar to punching. However, in this case no material is removed but locally deformed. In literature, no specific time and cost functions are derived for special tool punching. Because of the similarity with punching, time and cost functions for punching can be used.

2.6.3 Non-mechanical cutting Non-mechanical cutting in general Most time and cost functions discussed in this section are mainly cutting process independent. Only small parts of the functions are cutting process dependent but they were derived specifically for one specific cutting process. In (Cuesta, 1998) time and cost functions are derived for ‘beam cutting processes’, i.e. oxygen-cut, plasma-arc, water jet and laser. The process time for a batch is composed of preparation times, cutting times, auxiliary times and programming times (A.23). The auxiliary times are times for unexpected matters, maintenance, damages and inspection, operator idle times, etc. The auxiliary times are not quantified in detail. The programming time covers the time for preparation (drawing, nesting, etc.) to the final post processing of the NC programs. The programming times are considered independently here for a number of reasons: programming times are relatively large compared to e.g. preparation times, programming times are often very variable and the labour rate for a programmer differs considerably from the labour rate for a machine operator. The costs that are considered for the total cost per batch are: machine-tool costs, material costs, tooling costs and programming costs (A.26). Abrasive water jet cutting Water jet cutting uses a high-pressure jet of water, possibly with an abrasive grain, to cut sheet metal. The abrasive water jet cutting process the has following characteristics (Kalpakjian, 2001): • Cuts can be started at any location without the need for pr edrilled holes; • No heat is produced; • No deflection of the rest of the work piece takes place; • Little wetting of the work piece takes place; • The produced burr is minimal; • Environmentally safe process. Some typical figures for water jet cutting are: variable value Diameter of water jet 0.03 – 1 mm Pressure about 400 MPa (up to 4000 MPa and even higher is possible) Cutting speed up to 7.5m/min for reinforced plastics, but much lower for metals Sheet thickness up to 25 mm depending on the materal

In (Zheng, 1996) an hourly rate is determined based on the costs of equipment, consumables, service and maintenance (A.43). For the equipment costs, linear depreciation based on 2000 working hours per year is used. The consumables are accounted for by rates for abrasive, water and electricity. The maintenance costs are captured by one rate. 24


Plasma cutting Plasma cutting uses a plasma beam to cut sheets. The plasma beam is generated by a electric potential difference, between a cathode and anode, and an igniter. With direct plasma cutting, the work piece is the anode and the nozzle is the cathode. With indirect plasma cutting, the nozzle contains both the anode and cathode. The plasma cutting process has following characteristics (Kalpakjian, 2001): • Fast process; • Small kerf width; • Good surface finish; • Good reproducibility; • Highly automated. Some typical figures for plasma cutting are: variable value Temperature 9400 ºC (in the torch for oxygen as plasma gas) Sheet thickness up to 150 mm

For cost and time functions, for plasma cutting see the paragraph about non-mechanical cutting in general. Laser cutting Laser cutting uses a focused beam of light to cut sheet metal. By stimulating a laser medium continuously with light energy or flashlight, an emission of radiation is generated. The stimulation at a constant power level result in a continuous wave laser beam (CW), while a pulsating power level (PW) results in a pulsed wave laser beam. The laser beam is guided to the work piece by means of mirrors and it is focussed by a lens. The highly focused, high-density energy melts and evaporates portions of the work piece in a controlled manner. In industry two laser media are used most: CO2 and Nd:YAG. The laser cutting process has following characteristics (Vries, 1996), (Zheng, 1996): • It is one of the faster cutting processes; • Non-contact machining (no tool wear); • Suitable for cutting complex, irregular shapes; • It can be easily automated with good prospects for adaptive control; For CO2 laser cutting applies: • Relative high efficiency; • Relative large wavelength (poor absorption of the beam by the metal, larger spot size, lower accuracy, lower efficiency); • Good energy distribution; • Good focusability; • Smaller kerf width; • High power (large thickness at relatively high speed); For Nd:YAG laser cutting applies: • Relative low efficiency; • Low power (low cutting speeds, and small maximum thicknes); • Relative small wavelength (good absorption of the beam by the metal, good focusability); • Less favourable energy distribution; • Better accuracy;

In (Zheng, 1996) an hourly rate is determined based on the cost of equipment, consumables, service and maintenance (A.58). For the equipment costs linear depreciation based on 2000 working hours per year is used. The consumables are accounted for by rates for electricity, laser gases (rates for

25

Chapter 2

He, CO2 and N2), assistant gasses (rates for O 2 and N2). For the service and maintenance rates for different power levels and assistance gasses are derived. In (Nollet, 1993) time and cost functions for laser cutting are derived in the same way as done for nibbling. The total processing time is the sum of the set-up and work piece time (equation A.44). The work piece time is the combination of the main time (A.45) and the auxiliary time raised by 5% for personal care (A.46). The main time includes time to go to the start position, time for cutting holes and time for cutting contours. The auxiliary time includes the fixing and unfixi ng time and the time for scrap removal. The costs are split up in fixed costs, variable costs and technology dependent costs (A.52). The fixed costs rate includes depreciation of the machine and the space (building) that the machine occupies (A.53). The rate for the variable costs contains the costs for energy, labour, maintenance and programming (A.54). The cutting and laser gas costs are considered technology dependent costs (A.55). The material costs are not considered because the total nibbling costs are compared to total laser cutting costs, which would be the same in both cases.

2.6.4 Joining operations Laser Welding In laser welding a high power laser beam is used as the source of heat to produce a fusion weld (Kalpakjian, 2001). For laser welding also CO 2 and Nd:YAG lasers are used, both in CW and PW mode. Inert gas is used to obtain cleaner welds. Laser welding has several advantages over conventional welding (Vries, 1996): • Excellent quality welds (finishing operations are rarely required); • Large depth-width ratio welds; • Small heat-affected zone; • A variety of metals can be welded; • Ease of automation. Some prerequisites for laser welding are (Vries, 1996): • Good alignment of the work pieces; • The surfaces to be welded must be clean and free of burrs; • High accuracy of both the laser beam and the work pieces. TIG welding In the case of TIG welding, the heat required for welding is obtained from electrical energy (Kalpakjian, 2001). An arc is produced between the tip of an electrode and the work piece. A consumable bare wire is fed automatically through a nozzle into the weld arc. The welding area is shielded by an inert atmosphere of a gas mixture. In order to prevent oxidation, the electrode metal usually contains deoxidisers. Because the heat generated is relatively low TIG welding can be used to weld sheets with a thickness of less than 6 mm. MIG welding In the case of MIG welding, the heat required for welding is also obtained from electrical energy (Kalpakjian, 2001). An arc is produced between the tip of an electrode and the work piece. In this case, the electrode is not consumed as in TIG welding. Instead, a filler metal supplied form filler wire is used. Because the electrode is not consumed, a constant arc gap can be maintained. In MIG welding shielding gasses are used as well.

In (Maree, 1997) the equation for MIG welding from the Design for Assembly software module of Boothroyd and Dewhorst is adapted based on a time study of the fabrication of two different frames. The costs incorporated are: the capital and maintenance costs of the machines, material costs and worker’s wages (A.59). The operating costs consist of the capital investment,

26


depreciation, maintenance and energy consumption. From the operating costs, a cost per time unit is calculated. The exact derivation of the rates is not explained in detail.

2.6.5 Bending operations Bending Several types of bending can be discerned, the most important are bottoming, air bending, wiping and swivel bending. In the bottoming process, the bend is fully determined by the punch and die. Therefore, every sheet thickness, every radius and every material type requires a different combination of punch and die. In air bending, the bend is hardly determined by the tool geometry. With one combination of punch and die, it is possible to bend sheets of different thickness and type and different bend angles can be achieved. Air bending also requires a lower press force. This makes the air bending process a very flexible bending process. Wiping and swivel bending are mainly used to create flanges. A problem with bending is springback, i.e. elastic recovery of the material after unloading. Therefore, the resulting bend angle is larger after springback. In order to overcome this phenomenon springback has to be compensated by overbending. Overbending results from using a punch displacement, which is larger than necessary. The determination of the correct punch displacement can be calculated from angle and force measurements during bending (Lutters, 1997b).

In (Somatech, 1998) actual bending, shoving, turning in the X-plane, turning in the Y-plane, turning in the Z-plane are considered as the main operations for bending (A.60). In (Maree, 1997) time functions for automatic (A.61) and manual (A.62) bending are derived based on a time study of the fabrication of two different frames. The costs are accounted for in the same way as done for MIG-welding.

2.7 Restatement of the problem definition The objective of this research is to develop a prototype cost estimation system that can provide cost information throughout the whole product development cycle and that can be used for cost control. In section 2.2, it is mentioned that information management is a good base for the integration of engineering tasks. The use of information as an integrator of engineering tasks requires a suitable information management system. In the next chapter, an information management approach will be discussed that allows the integration of engineering tasks based on information. This approach will be used in the development of the cost estimation system, as described in the next part of this thesis.

27

Chapter 2

28

3

Information Management

In section 2.2, the need for a proper information management system is mentioned. Such a system enables integrated cost estimation in the product development process. The integration of cost estimation is only a small aspect of the concurrent engineering philosophy, which has been developed in the past decades. Concurrent Engineering emphasises the simultaneous execution of shared tasks by separate departments and the control of cooperative decision-making (Sohlenius, 1992). The employment of Concurrent Engineering requires a good understanding of the interaction and communication between the manufacturing processes. Since the main input and output of most manufacturing processes is information, the main function of these processes can be considered to be information processing. Therefore, a prerequisite for communication is the availability and accessibility of coherent information. The access to meaningful representations of the existing information, reflecting the current state of affairs, is preferred over merely exchanging data (Kals, 1998). The cost control architecture of Liebers (section 2.3.2) is a suitable framework for the development of a cost estimation system. A drawback of the architecture is the fact that it is based on the manufacturing planning and control reference model of Liebers, which lacks a component for the access of information. The Manufacturing Engineering Reference Model of Lutters (Lutters, 2001) is based on the accessibility of information and offers a better base for the development of an integrated cost estimation system. Because the manufacturing engineering reference model already incorporates information management, it will be used for t he design of an integrated cost estimation system. The cost control architecture is still applicable in the Manufacturing Engineering Reference Model of Lutters. This chapter will discuss the principles of the reference model that are useful for integrated cost estimation.

3.1 The Manufacturing Engineering Reference Model The Manufacturing Engineering Reference Model as depicted in Figure 3.1 emphasises the equivalent importance of products, orders and resources in the manufacturing cycle (Kals, 1998).

Figure 3.1 The Manufacturing Engineering Reference Model of Lutters (Lutters, 2001).

29

Chapter 3

Product Engineering refers to all engineering activities related to the product life cycle of a specific type of product. It is concerned with the design and development of a product and its variants, starting from functional requirements up to final recycling/disposal. Resource Engineering refers to all the life cycle aspects of the resources, which are required for the execution of the production activities. It therefore includes the specification, design, development, acquisition, preparation, use and maintenance of the resources of a company. Order Engineering addresses those activities that relate a customer order to a specific (variant of a) product. It is the task of Order Engineering to compose production orders and to decide when given batches of products must be processed and with which resources. The objective of Order Engineering is the in-time execution of the production orders. Company Management is concerned with the control of customer orders. It is responsible for the strategic decisions concerning the range of products, which will be produced, and the processes and resources, which are required to this end. Production is concerned with the actual execution of the plans generated by the engineering tasks. From production, information is fed back to these engineering tasks. Information Management is discerned as the kernel of the manufacturing engineering reference model. It is responsible for the availability and accessibility of meaningful representations of the existing information, reflecting the current state of affairs.

3.2 Information structuring In concordance with the three piles of the reference model three information structures are distinguished: the product information structure (PRIS), the resource information structure (RIS) and the order information structure (OIS) (Figure 3.2). All manufacturing environments can be covered by these three information structures because they can evolve independently while their mutual relation remains the same. In an engineering-to-order environment the resource information structure is relatively static while the order and product information structures evolve almost simultaneous after an order has been accepted. In a mass production environment the product and resource information structures are developed almost simultaneous and remain relatively static henceforth while the order information evolves when orders for certain product variants become known.

Figure 3.2 The three information structures (Lutters, 1997a).

The function of the PRIS is the storage and management of all relevant product information in an evolving product development process, i.e. anything that is a consideration or result of a design or manufacturing decision. The RIS is used to define all the resources, i.e. the abilities, occupation, condition of the resources, etc. The OIS holds all information that provides the basis for , and results from, decision-making throughout the life cycle of the orders. Every information structure consists of a number of separate aspect systems, referred to as domains. So, a domain is the representation of one aspect system in an information structure. Domains are independent of each other, i.e. domains of one information structure can be 30


instantiated independently of each other. The number of domains per information structure is not known on beforehand and can be different in different situations. Different interpretations of a domain are possible. For this purpose, a view on a certain domain can be created. A view is a focused and partial representation of the information in one domain of an information structure. Views can be mutually dependent, i.e. a change in one view can force a change in another view. Because it is not always necessary to view all information in a domain, information can be filtered. A filter selects all information relevant to a user. It is generally accepted that an information structure reflects its content by means of the elements and relationships it consists of (Lutters, 1997a). Additional information can be related to the elements and relations by means of attributes (Figure 3.3). The element-relation representation has two major advantages. Firstly, there is no exclusive reference to geometric entities, i.e. an element can be part of any aspect system. The transition from one aspect system to another becomes more natural in this way. Secondly, there is no explicit hierarchy, i.e. part-of relations are only one of the possible relations.

Figure 3.3 Description of information by means of elements, relations and attributes (Lutters, 2001).

Figure 3.4 The fundamental structure (Lutters, 2001).

In practice, an attribute corresponds to an element with a relation. Therefore, the fundamental building block for information structures has no attributes (Figure 3.4). The fundamental structure allows the construction of any information structure, e.g. a product information structure, in the form of a conceptual graph. When the extension of a graph with instantiated fundamental structures is not controlled, it can result in meaningless structures. In order to prevent a random extension of an information structure, the structure must be constructed according to a certain blueprint. This blueprint is a description of an information structure at a higher level of aggregation and can be indicated by the term ontology. Therefore, an ontology describes information according to the type of elements and the type of relations the structure is constructed of. For the implementation of elements and relations by means of a computer language, object oriented programming is very suitable. The objects element and relation can be recorded in two separate classes. The description of a class is done by means of class members. In order to realise the characteristics of information structures described above, both classes need class members for the domain and the view it belongs to and for the type of information it represents. A unique ID number is also necessary and a name is very useful. The class for a relation also requires class members for the ID’s of the elements it relates in order to indicate the position of a relation in an information structure. The minimum number of class members per class is represented in the fundamental structure in Figure 3.5.

Figure 3.5 Class definition of elements and relations (Lutters, 2001). 31

Chapter 3

3.3 The information structures The three information structures discerned in the Manufacturing Engineering Reference Model will be elaborated on in this section. The Product Information Structure Although the domains are unknown on beforehand, some generally applicable domains can be discerned. For the PRIS these are: the objective domain, physical product definition domain and control domain (Lutters, 1999). The objective domain contains all information that fixes the specifications and requirements of the product. The physical product definition domain records all information about the physical elements of the products. The control domain holds all information describing the adaptable behaviour of the product, e.g. software. Often, the designer and process planner use different representations of the product. An example of a different representation is given in Figure 3.6. The definition of parallelism between two faces by the designer will be converted to a certain bending accuracy by the process planner. The parallel tolerance and bending accuracy are defined using the faces of the core model of the product. Therefore, useful views on the physical product definition domain are for example a design view and a manufacturing view. In order to monitor the costs of a product, a cost view on the physical product definition domain is very suitable.

Figure 3.6 The multiple views problem (Kals, 2000).

A general template for the physical product definition domain can be defined (Figure 3.7). The core model of the product is created with the object elements depicted in this f igure. For the construction of views, additional elements e.g. module and feature have been added. Tolerances can be added to this figure in a similar way. Every view will always include the object elements; the subject elements will be different for different views. The domains and views allow the definition of a framework for the product information structure (Figure 3.8). This framework is extremely useful in defining tasks, co-ordinating processes and keeping abreast with the current state of affairs. Furthermore, the framework allows the access of meaningful representations of the existing product information, reflecting the current state of affairs. The Resource Information Structure The generally applicable domains for the RIS are: the method domain and the physical resource domain (Kals, 1998). The method domain comprehends the ways in which (combinations of) available resources can be applied to perform production tasks. The method domain contains the description of fabrication and assembly operations including the materials it can be performed on, the suitable type of machines, the suitable types of tools and competent operators. The method domain does not contain any resource specific instructions or constraints. The physical resource 32


domain contains a description of all resources present on the shop floor, including the technical capabilities, instructions and logistic characteristics.

Figure 3.7 A template for the physical product definition domain (Lutters, 2001).

Figure 3.8 The framework for a Product Information Structure (Lutters, 2001).

Two useful views on the physical resource domain are the capability view and the capacity view. The capability view shows technical capabilities of the resources. The capacity view shows information about the time schedules of the resources. It shows when a resource is allocated to which job and for how long. From the allocation history of machines the utilisation rates and performance of resources can be deduced. With cost views on both domains cost information about methods and resources can be viewed. The framework for a Resource Information Structure is depicted in Figure 3.9.

33

Chapter 3

Figure 3.9 The framework for a Resource Information Structure (Lutters, 2001). The Order Information Structure The generally applicable domains for the OIS are: the external order life cycle domain, the activity domain (Giebels, 2000) and the performance domain (Lutters, 2001). The external order life cycle domain contains all the information about stakeholders outside the company, which influence the policy and decisions of the company. The most important stakeholders are the clients, the suppliers and the subcontractors. Other stakeholders can be (temporary) employees, shareholders, trade unions and government authorities. The information about former orders is also stored in the OIS and can be used to analyse the performance of the company, to forecast, to make investment plans, etc. The activity domain comprehends the information about all the manufacturing activities,

Figure 3.10 The framework for an Order Information Structure (Lutters, 2001).

34


including supporting activities, performed in a company. It also includes the logistic information (e.g. time, place, required resources and batch sizes) of the activities. Usually, activities are planned on different aggregation levels. The OIS also contains the planning hierarchy used. The elements of the performance have yet to be determined. The purpose of the views on this domain is to have direct access to e.g. performance indicators (Lutters, 2001). These views will not be elaborated on. Useful views on the external order life cycle domain can be a subcontracting view and a sales view. Views on the activity domain could be a capacity planning view and a resource requirement view. With a cost view on the activity domain, the balance between optimising production plans and the costs of planning can be controlled. The framework for the OIS is depicted in Figure 3.10.

3.4 Ontologies In section 3.2, ontology was introduced as the type definition of information. The type characterisation of elements and relations positions them in proportion to each other. This ontology is referred to as semantic ontology, i.e. the definition of elements and relations between these relations. The domain and view characterisation of elements and relations positions them in the entire information structure. This ontology is referred to as symbolic ontology, i.e. the definition of the structure the semantic ontology adheres to (Lutters, 2001). Both the semantic and symbolic ontology are extended whenever a new ‘type’, i.e. elements, relations, domains or views, is added to the corresponding information structure. The ontologies ensue from the information structure. An important characteristic of the ontology used here is the fact that it needs not to be predefined but that its construction can follow the evolving information structure as part of the manufacturing process. Therefore, an ontology constitutes a convex-hull of the information content of an information structure. Whenever an ontology is present, it can be used to instantiate an information structure in a consistent way. An ontology can be constructed with the same fundamental structure as is used for the construction of information structures. An ontology, as introduced above, records the static characterisation of an information structure. This characterisation of an ontology is indicated as the ontology of state, i.e. the description of the stationary dependency of elements and relations. The knowledge about how to go from one state of an information structure to another state is an important tool to control the design and engineering processes based on information. For example, the ontology of state can hold that a component has a process plan and a production plan. The fact that a production plan requires a process plan can be recorded by adding a relation, for instance ‘needs’, between the production plan and process plan (Figure 3.11). When the production plan of a component has to be created and no process plan exists yet, the process planner can be triggered to create a process plan based on the extra relation. This property allows the construction of a so-called task-chain, i.e. a possible sequence of tasks that will eventually generate the required information. In contrast to prescriptions of process paths like scenarios, task-chains are very flexible because they capture all possible process paths. A task-chain can be carried out more or less automatic. The establishment of information dependency is indicated as the ontology of transition, i.e. the description of the process-related dependency of elements.

Figure 3.11 Ontology of state and transition.

35

Chapter 3

3.5 Process architectures related to Information Management The Information Management architecture incorporates all the functions concerned with information processing. Information Management also offers the control basis for governing the engineering processes in order to initiate, accompany, control and evaluate development cycles of products, orders and resources. In the execution of its functions, the architecture treats the types of information, i.e. products, resources and orders, the same. The architecture does not incorporate a central database. The architecture redirects information from and to databases owned by different users. The information about which databases are relevant for an order is stored in a metabase. A detailed description of the information management architecture can be found in (Lutters, 2001). In order to fully exploit all advantages of Information Management in a manufacturing system, the design and engineering processes have to be rearranged. The design and engineering applications have to be split into functional modules and arranged around the Information Management kernel. In this way, all modules can use the information management functionality without having to incorporate information management functionality themselves. Figure 3.12 depicts the co-operating architectures in a manufacturing system, with the architecture in the front representing a process planning architecture. In order to fully exploit the characteristics of information management as described in this chapter, the architecture for cost estimation has to fit in this constitution of a manufacturing system as well.

Figure 3.12 Co-operating architectures in a manufacturing system (Lutters, 2001).

36

Part II A generic cost estimation architecture

4

Design of a generic cost estimation architecture

In this chapter a generic architecture for cost estimation will be developed. First, the functional specifications will be formulated in section 4.1. Because the architecture will be based on the information structures related to the Engineering Manufacturing Reference Model, the relation between cost information and the information structures will be clarified in section 4.2. The actual cost estimation architecture and its functional modules will be described in section 4.3. The use of the information structures makes it possible to redefine variant based cost estimation. The new variant based cost estimation method and its position in the cost estimation architecture will be discussed in section 4.4.

4.1 Functional specifications From the literature review in chapter 2, the functional specifications of an integrated cost estimation system can be derived. The system has to 1. be application domain independent; Different application domains use partially different cost calculation instructions, but when these instructions are arranged in an identical way, they can be executed in the same way. A template for the definition of cost calculation instructions will guarantee a uniform execution of the instructions. In the case of Information Management such a template can be stored as an ontology. For such an ontology it is necessary to identify the type of information in cost calculation instructions and to identify their relations. 2. be cost estimation type independent; Different situations require different cost estimation types. Therefore, the system has to be able to deal with fundamentally different cost estimation approaches, like generative and variant based cost estimation, and to use them together in a hybrid mode. 3. enable cost control; The most important aspect about costs is the control of costs. Cost information is the basis for cost control. Therefore, the cost estimation system has to be able to generate cost information, which is useful for cost control. 4. support engineering and management tasks; Engineering tasks can base their decisions on cost information. Different engineering tasks use different cost information and have different information available for cost estimation. Engineering tasks that have to be supported are, for example: quotation, design, process planning, en production planning. Management should be supported as well. The Manufacturing Engineering Reference Model offers an ideal basis for the integration of cost estimation in the product development cycle. Therefore, the system has to use the information management concepts and information structures developed in the context of the Manufacturing Engineering Reference Model 5. be able to deal with different cost models; In order to compare and to improve cost models, it would be practical to be able to use different cost models side-by-side. 6. be able to handle different cost types; In cost control, cost types are very suitable for the analysis of the costs. Although certain general 39

Chapter 4

cost types can be distinguished, the definition of individual cost types must still be possible. 7. be able to calculate costs on different aggregation levels; In different situations the amount and the extent of detail of information is different. It must be possible to calculate and represent cost information on different levels of aggregation. 8. be able to apply different analysis tools for related information; As stated before, the analysis of cost information is an important aspect of cost control. Different data analysis software is available and could be used for the purpose of cost analysis. 9. be able to generate reports for engineering tasks; Despite the fact that most companies nowadays use computers for the performance of engineering tasks, results of these tasks are often still recorded on paper for archive and exchange purposes. 10. be modular; The advantages of a modular system are the possibility of integration with other software packages, the possibility of extending the system and an appropriate environment for maintenance (Leung, 1996). 11. be transparent; The use and behaviour of the system should be easy to understand. 12. be highly automatic. When a system can operate in an automatic mode, it easier for a user to use it. Nevertheless, the user must have the possibility to intervene the cost calculation at any time.

4.2 Cost information and the information structures Besides the functional specification, the input and output of a cost estimation system are important. In case of the Engineering Manufacturing Reference Model, the input comes from the information structures (product, resource and order information) and the output is cost information, which is related to these information structures. Most of the information structures are constituted by aspect systems other than cost estimation. Therefore, the input of the cost estimation system cannot be determined by the cost estimation system. The output can completely be determined by the cost estimation system. However, the way in which the cost information is used by other aspect systems is an important factor for the formulation of the cost output. Therefore, the way cost information can be attached to the three information structures needs to be clarified. Product Information Structure Finally, all costs are allocated to a product. In order to get a differentiated view of the costs, it must be possible to record the costs for every product element. This can be realised by relating an element of type ‘total costs’ to a product element by a relation of type ‘has’. Additional information about the costs, like currency and date of calculation, can be related to a costs element in a similar way. When generative cost estimation is applied, the costs are calculated at the lowest product level possible. The costs for an element on a higher level can be calculated by adding the costs of the elements it consists of and the costs incurred for the element self. This principle is illustrated by a cost view on the physical product definition domain in Figure 4.1. The costs are represented in this figure by attributes of the elements. In this case, the features are the lowest level on which the costs are calculated. The costs of a feature will usually only imply manufacturing costs, therefore the total costs for elements at this level are the same as the costs incurred on this level. The costs for the components are composed of the costs of the features plus the costs incurred for the components, e.g. material costs. The costs for the assembly are the costs of the components plus the costs incurred at this level, e.g. assembly costs. From this, it becomes clear that the costs for a relation are accounted for on the highest level that is part of the relation. Usually, the costs incurred at a certain level are of a certain type, e.g. material costs. Because different cost types can be incurred at a certain level, more cost attributes of different types can be defined for an element.

40


Figure 4.1 Cost view on the physical product definition domain.

The extension of the product structure with more cost attributes can be illustrated by the use of cost information by the designer. From an analysis of the product design process and its decision making it can be concluded that the costs fixed during product design are caused by the following cost driving product characteristics, i.e. cost drivers: geometry , material, production processes and production planning (Weustink, 2000). Figure 4.2 illustrates a cost view based on the cost drivers relevant for the design process.

Figure 4.2 Cost view on the physical product definition domain based on the cost drivers relevant for design.

When the product information contains alternative solutions, the costs per alternative have to be calculated. For the calculation of the total product costs, one specific alternative has to be indicated. The product in Figure 4.3 has two possibilities for one component. The total costs per component can be compared easily. But for a good comparison, the total costs for the assembly have to be calculated and therefore a specific configuration has to be chosen and the total costs can be calculated. Hereafter, another configuration can be chosen and be calculated also. At first sight, component B in Figure 4.3 is cheaper than component A, but because of assembly costs the total assembly costs are lower when component A is used. 41

Chapter 4

Figure 4.3 Illustration of the use of cost information in the case of alternatives.

The cost views presented above demonstrate the effectiveness of differentiated cost information for cost control. For every element on every aggregation level the costs can easily be made visible. Based on the differentiated cost information well-founded decisions can be taken. Besides the choice between alternatives, it can be used to detect irrational designs (Ou-Yang, 1997), e.g. a designer can notice an overrated requirement for the precision of a feature because of its high costs. In order to use cost information in the context of the product information structure as described above, the cost estimation system has to store the cost information it generates in the way as described above. Resource Information Structure Cost related information can be attached to the resource information structure in the same way as to the Product Information Structure. Relevant cost information for the Resource Information Structure is for example cost price of resources, rates and production times (Figure 4.4). Suitable elements for these attributes are for instance machines, tools, materials and operators but also production methods. Based on the related cost information, cost related views can be created. The cost related information from the Resource Information Structure is usually used to calculate the costs in the Product Information Structure. Information like rates can be calculated from information, e.g. utilisation rates, from the Order Information Structure. Order Information Structure For the Order Information the same counts as for the Resource Information Structure. Relevant cost information consists of, for instance, overall costs of orders, clients and suppliers (Figure 4.5). The cost related attributes are usually based on the information from the Product Information Structure, e.g. product costs, and the Resource Information Structure, e.g. production times.

42


Figure 4.4 Illustration of the use of relevant cost Figure 4.5 Illustration of the use of relevant cost information in the Order Information Structure. information in the Resource Information Structure.

4.3 The cost estimation architecture The design of the integrated cost estimation system is based on: • The functional specification of the cost control architecture of Liebers (section 2.3.2) • The Manufacturing Engineering Reference Model of Lutters (chapter 3) • The literature review on cost types (section 2.1.1) • The literature review on cost allocation (section 2.1.2) • The literature review on cost information in manufacturing (section 2.2) • The literature review on cost estimation (section 2.4) • The literature review on cost modelling (section 2.5) • The functional specifications of a cost estimation system (section 4.1) • The use of cost information in information structures (section 4.2) The integration of a cost estimation system in the product development cycle can be achieved by the creation of a cost estimation architecture, which uses the information management architecture. This implies the construction of functional modules around the information management architecture. In this way, the functional modules do not require to have any information management functionalities. The division of the system in functional modules will enhance the transparency of the system. Further transparency has to be achieved in the implementation of the modules. When the information structures related to the Manufacturing Engineering Reference Model are used and cost information is related to these structures, cost control can be performed based on the information structures. The main function of a cost estimation system obviously is cost estimation. In order to support the engineering tasks with cost information, one has to be able to use the cost estimation function in relation with the engineering tasks. Therefore, the cost estimation architecture has to contain a module that executes the cost estimation function. The cost estimation function comprises the execution of cost calculation with estimated values for the given parameters. The resulting costs have to be attached to the information structures. In the cost control architecture (see Figure 2.7), a function for cost calculation & evaluation is discerned. From a point of view of cost control, these two aspects can be combined because evaluation of the cost estimates is very important and cannot be carried out without cost calculation. However, these two aspects are clearly distinct and have to be dealt with separately. The cost calculation function is the same as the cost estimation function, but for the execution of cost calculation the actual values for the parameters are used. Both functions have to available to support the engineering tasks. The resulting costs, estimated and actual costs, have to be related to the information structures. The cost estimation and cost calculation function can easily be combined in one module, i.e. the cost determination module. The evaluation function analyses estimated costs versus actual costs. For this, a data analysis module is required. Both the cost estimation function and the cost calculation function require cost calculation instructions. The cost modelling function contains the formulation of these instructions. The 43

Chapter 4

formulation itself is a task of a cost engineer. Therefore, the cost modelling function is located in a separate cost model module. The actual deduction of cost functions is based on the analysis of historic data, for which different methods exist. This functionality also belongs to the data analysis module. Besides the cost functions, information about the quality, accuracy and sensitivity of cost estimates is very important. For the determination of the quality, accuracy and sensitivity of cost estimates different other analysis methods exist. As a result, these functionalities can be grouped in a separate module, the risk analysis module. When application domain specific information is used in a cost calculation this will occur in the cost calculation instructions. Therefore, the cost model module has to be generic. Also this module has to be generic to be able to define different cost types. In order to support different stages in the product development cycle it has to be possible to define multiple cost models. When the cost model module is generic and multiple cost models can exist, the use of different types of cost estimation is easier as well. The use of different cost models requires that the cost model module has management functionalities. The generation of cost reports is a separate functionality and has to be available for every engineering task; therefore, a separate cost reports module is desired. In the cost control architecture, production monitoring is discerned as a separate function. This function is required for other processes like process planning and production planning as well. Therefore, the cost estimation architecture is not the most logic location for the production monitoring function. However, because of the modular layout of the cost estimation architecture, a production-monitoring module could be added anyway. In the cost control architecture (see Figure 2.7), the cost accounting function is positioned, though it is not a function of the cost control architecture itself. Since the cost accounting function is usually discerned as a separate process in manufacturing, a separate process architecture has to be developed. The interaction between the cost estimation architecture and a cost accounting architecture can be dealt with by the information management architecture. Often, before data can be used it has to be tuned. For instance currencies, inflation or the correct timeframe has to be accounted for. This function has to be available for the other modules and can exist as a separate data tuning module. The modules that are derived above are depicted in Figure 4.6 and will be discussed in more detail in the next sections.

Figure 4.6 The cost estimation architecture.

44


4.3.1 Cost Models The main function of the cost models module is the definition of cost models. The module has to be generic to enable the definition of different cost models and the use of different cost types and formulas. Therefore, a generic template for the definition of cost models has been developed (Figure 4.7). In order to achieve synthesis between the structuring of cost information and cost calculation the format of the information structures related to the Manufacturing Engineering Reference Model is used for the definition the template. Consequently, the storage and manipulation of the cost models can be dealt with by the information management system in the same way as all the other information. The template consists of the following elements: cost structure, cost type, cost function and cost parameter. The cost structure has relations with cost actuators, cost models and cost drivers.

Figure 4.7 The cost structure.

The purpose of a cost structure is to record the costs for all the objects that cause costs. Therefore, the cost structure template has a relation with a cost actuator, i.e. an object that causes costs. In using the Manufacturing Engineering Reference Model, the cost actuators can be any object from the three information structures PRIS, RIS, OIS or the ontology. For one cost actuator multiple cost structures can be defined and one cost structure can be related to multiple cost actuators. The attributes of the cost structure element (not shown in Figure 4.7) are depicted in Table 4.1. A hierarchy of cost actuators is possible, e.g. a production method can be executed by a machine. For both the production method and the machine a cost structure can exist. Attribute Name

Type Method Accuracy Description Value

Meaning

A unique identifier. The type of cost calculation. The method of cost calculation. The achievable accuracy of the result of the cost structure. A short description of the cost structure. The costs represented by the cost structure.

Example Generative machine cost structure. Generative cost estimation. Activity based cost estimation. -

-

Table 4.1 The attributes of the cost structure element.

Because a cost structure is defined for a cost actuator, a cost structure cannot cover a complete cost model. So a cost model consists of a collection of cost structures for which the method of cost 45

Chapter 4

calculation is the same and which covers the total manufacturing costs. The method of cost calculation has to be the same for the different cost structures in order to guarantee consistency of cost calculations. Within one cost model the type of cost calculation can be different in order to enable hybrid cost estimation. It is possible to define different cost models by defining different cost structures for the cost actuators and use them simultaneous. The benefits of this option are: 1. Cost models can be compared easily. When a cost model can be selected from a collection of different cost models, cost calculations with the different cost models can be performed. Based on the results, e.g. completeness, extent of detail, a well considered choice about a suitable cost model can be made. 2. Cost models can easily be improved. Two versions of the same cost model can exist at the same time. For instance, one of these cost models can be the released cost model that is used to estimate the costs. The other version can be used by a cost engineer to improve or change the cost model. When an improved model is accepted, it can be released and used to estimate the costs. 3. Cost models can be defined for different aggregation levels. In order to support different engineering tasks, it is necessary to have cost models based on different levels of detail of information. For a designer a global cost model can be defined while for a process planner a more detailed cost model can be made available. In the case of generic cost structures, i.e. cost structures, which do not contain any cost estimation method specific information, it is possible that a cost structure belongs to different cost models. For every cost structure, different cost types can be defined; the attributes of a cost t ype are listed in Table 4.2. Attribute Name Type Description Value

Meaning

A unique identifier. The type of costs. A short description of the cost type. The costs represented by the cost type.

Example Material costs. Direct costs. Material costs. -

Table 4.2 The attributes of the cost type element.

It is possible to define multiple cost functions for every cost type. By means of the option point one of the cost functions is indicated as the active function. In this way different cost functions can be compared. The attributes of a cost function are listed in Table 4.3. Attribute Name Description Value

Meaning

A unique identifier. A short description of the cost function. The costs calculated by the cost function.

Example

Cmat. Material cost rate. -

Table 4.3 The attributes of the element cost function.

In the context of the cost structure, the parameters from the cost functions are indicated as cost parameters. Because a cost parameter can be present in several cost functions, the cost parameters are directly related to the cost structure element. In this way, it is easy to determine whether the values of all cost parameters necessary for the cost calculation of every cost structure are available. The attributes of the cost parameter are listed in Table 4.4. The cost drivers introduced in the section 4.2, are very suitable for cost analysis on a higher level of aggregation than the cost parameters. For the construction of the cost dri vers, the cost parameters are needed, therefore the cost parameters are related to the cost drivers. The cost drivers can be related almost directly to engineering processes. This can be used to create task chains. When the value of a cost parameter that is required for a calculation is not known, the cost drivers to which the cost parameter belongs can be found. The engineering processes related to these cost drivers can 46


be triggered to generate a value for the cost parameter. Attribute Name Type

Description Value

Meaning

A unique identifier. Variable (dependent on the cost actuator, e.g. a dimension) or pseudo-constant (fixed for a certain time period, e.g. an overhead factor) or constant (fixed, e.g. initial costs of a machine). A short description of the cost parameter. The value of the cost parameter.

Example

Tsetup. -

Set-up time. -

Table 4.4 The attributes of the cost parameter element.

Examples of cost structures are listed in appendix B and the use of cost structures will be explained in the next chapter.

4.3.2 Cost Determination The main function of the cost determination module is the determination of the costs, whether it is a cost estimation or a calculation of the actual costs. This function can be split into two sub-functions: 1. Selection of suitable cost structures. 2. Calculation of the costs. Selection of suitable structures From the set of available cost structures, a sub-set of suitable cost structures can be generated. The generation of this sub-set can be based on three criteria: 1. Cost model; The sub-set of cost structures contains the cost structures of one specific cost model. 2. Accuracy; The sub-set of cost structures contains the cost structures of one cost model that can calculate the costs with a certain accuracy. The user can request a cost estimate with a certain accuracy. The cost model that is able to generate a cost estimate with an accuracy closest to the requested accuracy, i.e. an equal or higher accuracy, is selected for the cost calculation. 3. Information content; The sub-set of cost structures contains the cost structures of one cost model, which can calculate the costs with the information available. Based on the sub-set of cost structures, the product information structure of an instantiated product can be searched for cost actuators. When a cost actuator is found, the accompanying cost structure can be copied and related to the product information structure. Calculation of the costs After the cost actuators in a product information structure are extended with cost structures, it has to be checked whether the values of the cost parameters are available. When all the necessary values are available the cost functions can be calculated. Next, for every cost type, the results of the cost functions can be summed. After this, the costs per cost type can be summed for every cost structure and the costs can be attributed to the cost actuator. From here on, the costs can be summed for every aggregation level in the product information structure as described in section 4.2.

4.3.3 Data Analysis Much information that is required for cost calculations is based on historic data, e.g. cost rates, utilisation rates etc. For a correct use of this data, the historic information has to be analysed, e.g. averages, variances and trends are required. The analysis of data is not a specific cost aspect, therefore a common data analysis software package could be used for this purpose. The data to be analysed can be passed to an external package and the results can be imported in an information 47

Chapter 4

structure. This module must also compare the estimated costs with the actual costs in order to be able to improve the cost models. The cost estimation and the calculation of the actual costs can be set sideby-side and subsequently be compared. Based on this comparison, (parts of) the cost model can be updated. For one specific product it can be checked if all the production activities have taken place according to the process plan upon which the cost estimation has taken place. The cause of cost increase or decrease can be analysed. Over a certain time period it can be checked whether all the costs per cost type are accounted for. When the costs per cost type are not accounted for, the allocation base can be changed or the rates can be recalculated, based on new assumptions or more recent historic data.

4.3.4 Risk Analysis An important aspect of cost estimates is an indication of the accuracy of the estimates. The sensitivity of the costs to changes of the cost parameter values in the cost model must also be known (Luong, 1995). This module must take care of all the aspects concerning quality, accuracy and sensitivity of cost estimates. The data of these aspects can be related to the elements for which they apply, e.g. cost functions. Risk analysis is not a specific cost aspect either and therefore common software for risk analysis can be used for this purpose in the same way as for data analysis.

4.3.5 Data Tuning In many cases, data has to be transformed to be able to use it. Some examples are the tuning of currencies and the correction for inflation. Another aspect is the arrangement of data for a certain selection e.g. for a certain time interval, product family, customer, etc. This is an important support function for the data and risk analysis modules and for the cost reports module.

4.3.6 Cost Reports The element-relation structures with cost information are very suitable to construct cost reports of any kind. Important elements that a cost report has to contain are: the objects from the information structures, the objects from the cost structure, the arrangement of the objects and the type of costs (estimation or calculation). Furthermore, information about the layout has to be known. Figure 4.8 represents a possible template for the definition of cost reports.

Figure 4.8 A template for the definition of cost reports.

The use of this template is explained for the three information structures in Table 4.5. Information about the layout of the cost report can be included in the same way. The template is available for all engineering tasks, so after the instantiation of the template, uniform cost reports will be generated. 48


Element Cost report Subject list

Subject Information aggregation level Cost structure aggregation level Range Type

PRIS RIS OIS Detailed product costs. Rates of resources. Cost per order. Products of a certain Certain types of Orders of a certain time time period, products resources. period, orders of a of a certain customer. certain customer. Product. Resource. Order. Assembly, component, Method, resource. Orders, sub-orders, feature. activities. Cost actuator, cost structure, cost t ype, cost type, cost parameter.

Arranged per subject, arranged per subject list. Cost estimation information, cost calculation information or both.

Table 4.5 Use of the cost report template for the three information structures.

4.4

Variant based cost estimation and its position in the architecture

Usually, a variant based cost estimation algorithm is implemented and used as a separate cost estimation system. But in fact, a variant based cost estimation algorithm can be considered a cost function. It uses information from the product development process and adds cost information to the product information just like a cost function. Therefore, variant based cost estimation is incorporated in the cost estimation architecture as a cost function. The element-relation based information structure and the ontology of the product information are an ideal basis for comparing products, which is required for variant based cost estimation. All product characteristics directly related to costs are recorded in the elements and can be used for comparison. Therefore, the comparison algorithm can be mapped onto t he information structures. The use of the information structures for the comparison algorithm makes it relative easy to integrate variant based cost estimation with other engineering tasks. Besides, it is not necessary to maintain duplicate databases. Because of the generic character of the information structures, the comparison algorithm is not limited to one specific product type. The use of the information structures will also reduce the dependency of comparing products on human interpretation and reduce the effort required for coding. The use of the information structures will also increase the flexibility of the comparison algorithm because all information stored in an information structure can be used for comparison. In order to increase the flexibility of the comparison algorithm, the comparison criteria can be grouped in categories and every category can be valued differently. In this way, the comparison algorithm is usable for any task in the product development cycle. Also, the comparison results can be represented more differentiated. The variant based cost estimation method based on the information structures related to the Manufacturing Engineering Reference Model is split up in two phases. Phase one is the initialisation phase, performed by a cost engineer (but sometimes also by a ‘regular’ user). Phase two is the execution phase, which is carried out by a ‘regular’ user. A ‘regular’ user can be a human, e.g. a designer, or another software system, e.g. a process planning system. Initialisation phase The cost engineer has to define characteristics relevant to variant based cost estimation. A characteristic consists of a category, a category item and a determinative quantity (Table 4.6, grey columns). The determinative quantity should be directly related to the costs of the category item.

49

Chapter 4

Category

Percentage of Category item desired similarity Contour Round hole Rectangular Manufacturing 75 % hole feature Obround hole Polygon hole Bending line Punching Production 60 % Laser cutting method Bending Production Order size 60 % planning

Number of Determinative the category quantity items 1 Length 10 Diameter 10 Height/width

Value of determinative quantity 1200 mm 5 mm 3

10 2 2 1 1 1 1

3 2 40 mm 100 mm 300

Height/width Area Bending length Area Length Length Product/order

Table 4.6 Example of product characteristic definitions (grey: cost engineer, white: user) Execution phase The product information structure of a new product can be searched for categories, category items and determinative quantities. This search will result in the number of category items and the value of the determinative quantities (see Table 4.6). When this is carried out also for the products that have been manufactured in the past, a similarity coefficient per historic product for every category can be calculated with equation 4.1. This equation represents the average ratio of the number of category items and the value of the determinative quantities between a new product and a historic product. The denominators in this equation are the highest numbers accounted for in order to keep the similarity percentage between 0 and 100%. In order to reduce the calculation time, the number of historic products for which the similarity has to be calculated can be reduced by first checking whether the aggregation level and the type of the new product and a historic product are equal. N  t v  ∑  i ⋅ i  N i = 1 d1 d 2   n, if n > n j t ij , if t ij > t i N= d1 =  n j , if n < n j t i , if t ij < t i

P j = with:

1

(4.1)

v ij , if v ij > v i v i , if v ij < v i

d2 = 

P j: similarity percentage between the new product and historic product j. n: number of category items of the new product. n j: number of category items of historic product j. ti: number of category item i of the new product.

tij: number of category item i of the historic product j. vi: value of the determinative quantity of category i of the new product. vij: value of the determinative quantity of category i of historic product j.

The user has to indicate the required extent of similarity per category in a percentage. The percentages of desired similarity can be a personal preference or a default set of percentages suitable for a specific engineering task. For a user or engineering task, the metabase containing references to the characteristics and historic products database can be personalised. The characteristics database containing the personalised set of percentages of desired similarity can be stored separately for a person or engineering task. In the metabase, the reference to the characteristics database has to be changed for the person or engineering task. The use of default sets of similarity will increase the consistency of the cost estimates and will simplify the application of this method in an automatic mode. The costs of a new product equals the average costs of the historic products that satisfy the set of desired similarity.

50


The possibility to change the percentages of desired similarity enables the use of this method in different engineering tasks. For instance, the designer will probably require a high similarity in geometry while the process planner requires a high similarity in production methods and a lower similarity in geometry. The similarity search can be controlled in several ways. The number of similar products found and the set of desired similarity percentages used are an indication for the accuracy of the cost estimate. The percentages can be adjusted in order to get a larger or smaller number of similar products. Another way to increase the number of similar products is to ‘modify’ a historic product temporarily. Since all the costs are known for every element of the historic product, it is possible to add or delete elements that cause dissimilarity. The average costs of similar elements can be added to or subtracted from the costs of a historic product. The costs for a relation between an element and the product have to be altered as well. The altered historic product will have better similarity coefficients and can be used as a reference to calculate a cost estimate.

51

Chapter 4

52

5

Employment of the architecture

For a proper explanation of the employment of the architecture, the process architectures related to Information Management as depicted in Figure 3.12 are essential. For reasons of clarity, a different view of this figure is depicted in Figure 5.1. The process architectures of the main processes of the product development cycle are depicted without specifying their functional modules (Figure 5.1 A). The cost estimation architecture is depicted with its functional modules (Figure 5.1 - B). The information structures are depicted separately in this figure (Figure 5.1 - C). The Information Management architecture (Figure 5.1 - D) is the central pivot in the exchange of information. In the cost control architecture of Liebers (section 2.3.2) four feedback loops are distinguished: the engineering and planning loop, the order acceptance loop, the production loop and the accounting loop. These four feedback loops can be identified in Figure 5.1 as well, they are described in the context of the cost estimation architecture in section 5.1. The generation of cost information required for the feedback loops depends on the cost models defined. Modelling of cost models by means of the Cost Models module is explained in section 5.2. Based on the cost models defined, all engineering tasks are supported by cost information in the decision making process as is clarified in section 5.3. In section 5.4, some concluding remarks about the employment of the cost estimation architecture are given.

5.1

Cost control

The feedback loops of information as described in section 2.3.2 are a prerequisite for cost control. Besides these feedback loops four functions, which have to enable the feedback loops, are distinguished in the cost control architecture of Liebers (section 2.3.2). These functions: cost modelling, cost estimation, cost calculation & evaluation and production monitoring can to a certain extent be distinguished in the cost estimation architecture. They are only organised differently as described in section 4.3. Therefore, the four feedback loops can be used in the context of the cost estimation architecture as well, thus enabling cost control. The engineering and planning feedback loop Every engineering task uses information, including cost information, from the three information structures for decision-making. The cost information is added through the cost estimation architecture. The results of the decisions of an engineering task will result in a modification or extension of the information. In terms of Figure 5.1 this can be identified by A – D – B and vice versa. The processes involved are cost estimation and any of the other processes depicted in Figure 5.1. The information used is usually limited to (the variants) of one product. For a proper use of the information in the information structures, it is important that the information is related to the information structures at the right aggregation level. The use of the information depends on the engineering task. Because different engineering tasks are involved in the product development cycle, different aggregation levels have to be considered. This can be achieved in three ways: by defining cost models on different aggregation levels, by defining information on different aggregation levels and by using information from other aggregation levels. The definition of cost models on different aggregation levels, by means of t he Cost Models module, and the use of these models is discussed in the next section.

53

Chapter 5

Figure 5.1 A ‘different’ view on process architectures related to Information Management.

On different aggregation levels, different objects are meaningful. Based on the cost attributes of these meaningful objects, a cost view can be constructed. In this way, the appropriate cost information becomes available at every aggregation level. The meaningful objects, which are the base of a cost view, will be determined by the requirements of a specific engineering task. These meaningful objects are part of one of the three information structures, which can be accessed through the information management kernel (Figure 5.1). An engineer performing an engineering task can use the cost view based on relevant objects with their cost attributes and he can use filters in order to focus on a certain piece of (cost) information. In order for an engineering task to use information from another engineering task, it must be possible to "translate" the information from one aggregation level to another aggregation level. If for instance process planning and quotation are taken as an example, the information is used on different aggregation levels and the cost information is generated on different aggregation levels as well (see Figure 5.2). During process planning, information about the resources and the extent of use that is required is generated. Based on this information, cost information can be generated at 54


feature, component and assembly level (Figure 5.2). During quotation, less detailed information is available or can be generated. Based on this information it can for instance only be possible to generate cost information on assembly level because detailed information on component and feature level is not available yet (Figure 5.2). In both cases, the cost information generated during each task can be used to change the product by that task in order to create an alternative better or cheaper solution.

Figure 5.2 Cost information on different aggregation levels

It can be convenient for quotation to use cost information generated by process planning. For instance, the actual costs of previously manufactured products can be useful. In this case, only the actual costs on assembly level, which are generated based on process planning information, is required (Figure 5.3). For process planning, it can be useful as well because during process planning it can be checked on assembly level whether the costs generated by quotation are still met. If the costs generated during process planning exceed the costs given by quotation, a product modification is advisable because the price given by quotation usually cannot be changed.

Figure 5.3 Use of cost information on a higher aggregation level.

It is even possible for process planning to use the cost information by quotation on lower aggregation levels. The cost information generated by quotation on assembly level can be split into component and feature level cost information based on the equations used (Figure 5.4). This allows process planning to compare the costs on lower aggregation levels with the costs of quotation. This situation is less preferable because the costs generated by quotation on assembly level are less accurate because more global parameters are used. By splitting the costs on assembly level to component and feature level based on these global parameters, the cost figures at the lower levels will normally be less accurate.

55

Chapter 5

Figure 5.4 Use of cost information at a lower level of aggregation. The order acceptance feedback loop The order acceptance feedback loop is almost the same feedback loop as the engineering and planning feedback loop. In terms of Figure 5.1 it can also be identified as A – D – B and vice versa. In this case, the processes involved are quotation and cost estimation and the information used is not limited to (the variants of) one product. In the example used to explain the engineering and planning feedback loop, quotation generated cost information for one product. The acceptance of an order can influence the costs of all accepted orders. For instance, the newly accepted order can require specific resources that were allocated to other orders. For these orders other resources have to be selected, which can cause an increase of the costs of these products. For order acceptance, it would be convenient to have cost information about all accepted orders and the orders under consideration at order level. This is only useful when the effect of accepting an order under consideration on the costs of the already accepted orders is immediately calculated. Because the accepted orders will not be in the same stage of the product development cycle, the costs of the accepted orders will be calculated at different aggregation levels. However, the results of these calculations will be made visible in a view at order level. The total costs of accepted orders can be examined in for example a view on facility level by summing up the costs of the individual orders (Figure 5.1). When the effects on the costs of the accepted orders is unacceptable, it can be considered to assign the extra costs to the order under consideration instead of dividing the costs over all accepted orders, as described above. The production feedback loop The production monitoring function is not part of the cost estimation architecture as described in section 4.3. When the production monitoring function is considered as a more general function that supports other engineering processes, it can be considered as a separate process in Figure 5.1. In that case, the production feedback loop can be identified as A – D – B as well. The processes involved are production monitoring and cost estimation. The information concerned is mainly product and resource related. During production, performance information about the production processes has to be stored in order to be able to check the correctness of the estimated process information. The information from the production processes can be related to the instantiated product information structure of the products. Based on this information the actual costs can be calculated and be related to the product information structure (Figure 5.1). Deviations in the estimated and actual costs can occur because the estimated values of cost parameters are wrong. It is also possible that the production of a product is different than used in the cost estimation process. In that case, other, more or fewer production processes are involved, resulting in different costs than estimated.

56


The accounting feedback loop In the context of Figure 5.1, the accounting feedback loop can also be described as A – D – B and vice versa. The processes involved are accounting and cost estimation. The information concerned will mainly be resource and order related. In the cost models usually many different rates are used, e.g. machine cost rates. These rates will be constant for a certain period for which the values are considered to be valid. The determination of rates is normally based on historic information and prognoses. After a certain period, new historic information is available and new prognoses are made. The new historic information can come from the calculation of the actual costs based on a cost model on process planning level. In the case of a machine cost rate, one cause of a different rate can be a different capacity utilisation or different overhead costs allocated to the machine. In order to update the rate, historic information of the capacity utilisation is required or information about the new overhead costs allocated to the machine is required. The new capacity utilisation can be calculated using information from instantiated product information structures from historic periods. Based on the product information of the products that have used a particular machine, the time that the machine has produced in a certain time period can be calculated. Based on this information the new capacity utilisation is calculated and related to the machine in the resource information structure. The new overhead costs have to be determined by the cost engineer or cost accounting and will probably come from the resource information structure. In the context of the Manufacturing Engineering Reference Model, accounting is a separate process with its own architecture. Accounting can obtain the information it requires through the information management architecture and it can store new and/or updated information through the information management architecture (Figure 5.1).

5.2 Cost modelling The feedback of cost information required for cost control as discussed in the previous section has indicated the importance of cost information on different aggregation levels. Therefore, it is important to be able to define cost models on different aggregation levels. In this way, relevant cost information can be generated for different engineering tasks. For the definition of cost models it is, therefore, important to understand the requirements of the engineering tasks. The engineering tasks have to indicate the objects relevant for the task to which the costs can be related. These objects will represent different aggregation levels. In the context of the cost estimation architecture, this means that the cost actuators in the cost structure (figure 4.5) are determined by the engineering tasks (see Table 5.1). For example, the designer can indicate a component and material as cost actuators, whereas the process planner can indicate machines and material as cost actuators (see appendix B, figure B.4, B.5 and B.2). When the cost actuators are known, cost structures can be defined for each of them. All resulting cost structures can be stored in a single database. The cost types will depend on the cost model chosen, e.g. Activity Based Costing, but also on which types of costs are considered important for cost control and cost reports. The cost model chosen will be determined by the cost engineer and will ideally be the cost model best suited for the specific case. The cost types for cost control and cost reports will largely be determined by the tasks on the higher aggregation levels like management and accounting. For example, the selection of a variant based cost model can limit the cost types to only total costs. Whereas the selection of a generative costs allows the definition of more cost types. The cost types are also used to define the cost views. The cost types as discussed in section 2.2.1 can be chosen for control purposes by for instance management. In the cost structures for component, material and machine in appendix B, direct costs, overhead costs and total costs have been discerned. The cost parameters will largely be determined by the cost actuators and the cost model chosen. Characteristics of the cost actuators will likely be candidates for cost parameters. The cost model chosen will largely determine the way costs are allocated and therefore it determines the allocation base, which is a cost parameter. In the cost structures for component, material and machine in 57

Chapter 5

appendix B, geometrical parameters and production time are selected as characteristics of a component and machine and an overhead rate is selected based on the cost model.

Engineer Task

designer

Product development cycle process planner

conceptual design concepts

embodiment design geometry

conceptual model variant based

embodiment model variant based

Accuracy

conceptual accuracy

embodiment accuracy

Cost actuators

concepts

Results Cost models Method

Cost Parameters

geometry, material concept dimensions, characteristics tolerances

production planner macro process micro process shop floor planning planning planning suitable suitable resources processes resources macro model micro model micro model

ABC and/or variant based macro accuracy ABC and/or micro accuracy variant based material, processes process characteristics

ABC and/or variant based micro accuracy ABC and/or micro accuracy variant based

ABC and/or variant based micro accuracy ABC and/or micro accuracy variant based

resources

resources

production times, resource rates, overhead rates

production times, resource rates, overhead rates

Table 5.1 Example of some engineering tasks and cost models in the product development cycle.

The cost functions can be deduced based on the cost parameters chosen and historic data about the cost parameters and the accompanying costs. The data sets of cost parameters and accompanying costs can be send to the data analysis module or to an external analysis program. The results of the analysis can be recorded in the cost structure. Based on the results of the Risk Analysis module or an external analysis program, information about the accuracy of the cost functions can be added to the cost structures as well. During cost modelling, the cost models can be controlled by means of a cost modelling view on the cost actuators. The cost modelling view can help to check the completeness and consistency of the cost structures and therefore of the cost models. Updating and improving cost models Cost models have to be maintained in order to keep them up to date and they need to be improved when possible. For this purpose, the feedback loops are essential. The accounting feedback loop is essential to keep the rates updated. The engineering tasks can indicate the practicability of the cost models based on the engineering and planning loop. By comparing the expenses made in a certain period with both the estimated and the actual costs made in the same period, the completeness of a cost model can be checked. When the calculated costs are less than the expenses made in the period, some costs have to be added to the cost model. When the calculated costs exceed the expenses made in the period, it is possible that some costs have been accounted twice in the cost model. When a cost model is changed, it is important to update the production feedback loop. A change in a cost model may imply that other information has to be fed back from production.

58


5.3 Costing support The Cost Determination module, which calculates estimates or actual costs, can be addressed at any point in the product development cycle. The user, requiring cost information, can activate a cost calculation by selecting one of the three criteria discussed in paragraph 4.3.2. The user can select the cost model he wants to use. He can indicate the accuracy he requires, after which the system can select an appropriate cost model. Or he can only indicate the available information so the system can select an appropriate cost model. In the first two cases, the available information has to be indicated as well. Whenever a cost estimate is required for which more information is required, task chains (section 4.3.1) will be activated to obtain the required information. Usually, a user or a group of users will always use the same method to calculate the costs. These preferences can be stored for a user or a user group and be used automatically when a cost calculation is required. If for instance a process planner requests a cost estimate based on an Activity Based Cost model, the cost structures with method ABC will be looked up (Table 5.1, micro model, ABC). If the accompanying cost actuators are present in the product information structure under consideration, the cost structures will be copied to this information structure. If a process planner requests a cost estimate with a relatively low accuracy, a variant based cost estimation model can be selected (Table 5.1, macro model, variant based). The cost structures with type variant based cost estimation will be looked up and if the accompanying cost actuators are present in the product information structure under consideration, the cost structure will be copied to this information structure. A request of a cost estimate with a higher accuracy will result, for instance, in the selection of a cost model of type ‘generative cost estimation’ (Table 5.1, micro model, ABC). If a designer requests a cost estimate and supplies only the product information structure as the only constraint, the cost model, which can produce a cost estimate based on the available information, is selected. In this case, this will probably be a cost model of the type ‘variant based cost estimation’ (Table 5.1, embodiment model, variant based). When the value of a cost parameter is unknown but required, a task chain can be activated. For instance, when a designer requests a cost estimate with a high accuracy, this can mean that a generative cost estimation model is selected. This model could for instance contain production times as cost parameters. The production time is related to the cost driver production processes, which in turn is related to process planning. Based on this knowledge, ‘process planning’ can be activated to determine the value of the requested production time. A cost calculation can be started event driven. This means for instance that the cost calculation starts automatically whenever a change is made to the information. Another option is that the calculation starts only whenever the user explicitly requests a cost calculation. All in all, the user will hardly notice the execution of the cost calculation and the use of the architecture is the s ame for every engineering task. After the cost calculation, the user can immediately use the result of the cost calculation because it is directly related to the information structures. The result can be made visible to him in his view (see figure 4.1 to 4.3). Normally, the results will contain the cost value and the information about the quality of the cost value. The way the cost value has been determined is not important for the use of the cost value and therefore it needs not to be shown.

5.4

Concluding remarks

From chapter four and five (in particular Figure 5.1), it becomes clear that the employment of the cost estimation architecture in combination with Information Management and its related information structures has several advantages: • Cost estimation can be integrated in the product development cycle. All cost information is exchanged through the centrally accessible information management kernel, so every engineering task can obtain and use cost information. 59

Chapter 5

•

Cost models, suited for different aggregation levels, can be defined based on the requirements of the engineering tasks. Every engineering task can indicate the suitable level of aggregation for cost support through the information structures. This means that every engineering task can be supported with cost information in an appropriate manner. • Cost views can be constructed on different aggregation levels, so that the cost information can be viewed by every engineering task. Because of the synthesis of the information structures and the cost calculations by means of the cost structures, the costs of a product are better accessible and more transparent. • Cost control can be performed by means of the four cost control loops. The cost control loops allow the support of the engineering tasks with cost information and they allow the maintenance of cost models. In addition, the development of the cost estimation architecture in the context of the Manufacturing Engineering Reference Model of Lutters has gained insight into and knowledge about the application of Information Management.

60

Part III A prototype cost estimation system

6

Implementation of a prototype cost estimation system

The generic cost estimation architecture is partially implemented in a prototype cost estimation system. In section 6.1, the development environment is described. In order to properly test the prototype system, other systems are required, mainly for the creation of input for the cost estimation system and for viewing the results. In section 6.2 implementations of systems for databases, design, process planning and production planning are described. The two main modules of the cost estimation system that have been implemented, Cost Models and Cost Determination, (see section 4.3.1 and 4.3.2) are described in section 6.3.1 and 6.3.2. In section 6.3.3 the start of the implementation of the Cost Report module is given. Section 6.4 describes the implementation of the variant based cost estimation algorithm (see section 4.4).

6.1 System specifications For the implementation of the prototype cost estimation system Microsoft Visual C++ has been used. The system consists of one executable and separate dynamic link libraries (DLLs) for every module. The executable can call all the dynamic link libraries and is mainly intended for use by a cost engineer who has to maintain the system. Other engineering tasks can call the dynamic link libraries directly, whenever they need the execution of a specific task. Parts of a realised prototype system for Information Management as described in chapter 3 are used (Lutters, 2001). The functionalities of this prototype used are graph manipulation and database access, both implemented as dynamic link libraries as well. For the storage of information in databases, Microsoft Access databases are used. The calculation of cost functions is performed by MatLab, by means of a so-called engine; so any function can be dealt with by the system. For the creation of geometry, SolidWorks is used. SolidWorks has been extended to be able to cooperate with the Information Management software and to perform some simple process planning functions (see next section).

6.2

Cooperating systems

6.2.1 Creation of databases A separate system was implemented to fill and edit the three information structures: the product information structure (PRIS), the resource information structure (RIS) and the order information structure (OIS). For reasons of simplicity, the system stores the contents of each of the information structures in one database. Theoretically multiple databases can be used, which are dealt with by the information management kernel. Every database consists of four tables for the storage of elements and relations for both the information structure and the corresponding ontology. A metabase is also created, which refers to the databases for the three information structures. The contents of the metabase can be characterised with a name (Figure 6.1).

63

Chapter 6

Figure 6.1 The database manager.

An information structure can be filled with elements; for the elements attributes can be defined. Figure 6.2 and Figure 6.3 show the user interface for the manipulation of elements and attributes. The figures show an example of a Resource Information Structure and the corresponding ontology.

Figure 6.2 A resource information structure.

64


Figure 6.3 The corresponding ontology of the resource information structure.

For debugging reasons, it is possible to change and delete elements and attributes from the ontology and the information structure changes accordingly. It is possible to add elements directly to an ontology in order to be able to create types of any aggregation level. If for instance an Amada FBD3512E is of the type “Press brake” and Press brake is of the type “Machine”. The element “Press brake” of the type “Machine” has to be defined in the ontology. The remainder of the ontology arises from the information structure. In the information structure, the elements are defined by a name, t ype and domain; see Figure 6.4 and Figure 6.6 for an example in the RIS. For reasons of simplicity, the view to which an element belongs is left out.

Figure 6.4 The definition of an element.

An attribute is defined by giving a name, unit, type and description (Figure 6.5 and Figure 6.6). It is also possible to assign a value; this is useful when the value of an attribute is more or less fixed. In the information structure an element is created of the type indicated by the name and with the value indicated by the number. The domain is the same as that of the element where the attribute belongs to. A relation between the element and the attribute is created as well. In the ontology an element of type parameter is created (if it does not yet exist) with the type of attribute as the name. For the description, the type and the unit of the attribute three elements are created in the ontology with the corresponding values. When an attribute is created for an element, this attribute belongs to elements of the same type as the element for which the attribute was defined.

65

Chapter 6

Figure 6.5 The definition of an attribute.

The examples are from the RIS but the same can be done for the PRIS and the OIS. In the PRIS it is for instance possible to define frequently used features with their characteristics and in the OIS it is possible to define for instance orders with their characteristics. The three information structures are used to create an instantiated product information structure, which can be used to test the cost estimation system.

Figure 6.6 Graph representation of elements and attributes.

6.2.2 Design In the manufacturing of physical products, the geometrical representations of the products are the basis for communication. Therefore it is be convenient to have a design system that can store geometric information in the element-relation format used in the context of the Information Management kernel. For this purpose, SolidWorks has been extended based on its Application Programming Interface (API). The feature tree of the design view of SolidWorks is traversed and information from this tree is transformed into elements and relations and stored in a database. The information that is stored includes: the assembly, the component, the features and the faces. No attributes, like dimensions, are stored yet.

6.2.3 Process planning The geometry collected from SolidWorks serves, amongst others, as the basis for process planning. Because of the fact that SolidWorks offers a nice user interface and because it can be extended based on its API, a simple manual process-planning module has been added to SolidWorks. Process planning requires at least information about the resources. So, in the process planning module, the metabase as described in section 6.2.1 has to be selected. Then the databases of the PRIS, the RIS and the OIS, which are referenced in the metabase, can be imported. From now on, all the information of the three information structures is available in the system. In principle, the features used in SolidWorks are design features. For process planning manufacturing features are required. Because feature mapping is a research area in itself and because it is not the focus of this research, feature mapping is performed manually and on a one-toone basis. In the case of sheet metal features, the one-to-one mapping of design features on 66


manufacturing can do the job in most cases. When the (design) feature tree of SolidWorks is traversed, as described in the previous section, the user is prompted to select the corresponding manufacturing feature. The user can select the manufacturing features defined in the product information structure (Figure 6.7). In contrast to the design features, the manufacturing features do have attributes, like dimensions. When the traversal of the (design) feature tree is finished, a new tree with manufacturing features is created (Figure 6.8). Process planning is performed, based on this (manufacturing) feature tree.

Figure 6.7 Manual one-to-one feature mapping.

Figure 6.8 The design (left) and manufacturing (right) feature tree.

Process planning is performed separately for each feature. For every feature, fabrication methods have to be selected (Figure 6.9), for every fabrication method machines have to be selected (Figure 67

Chapter 6

6.10) and for every machine tools and operators have to be selected. The information about fabrication methods, machines, tools and operators is stored in the resource information structure. When a selection is made, it is possible to quantify the value of the attributes, like machining times (Figure 6.11). It is not possible to quantify attribute values with a fixed value, like cost rates (Figure 6.12). The quantification of the attribute values is not based on technological information and calculation.

Figure 6.9 Production method selection.

Figure 6.10 Machine selection.

Figure 6.11 Definition of attribute values. 68


Figure 6.12 Fixed attribute values.

The selected fabrication methods, machines, tools and operators are added to the feature tree (Figure 6.13). The feature tree and the attributes can still be edited. The new feature tree and all the attributes can be saved to a database and used by the cost estimation system. In the database, a reference to the SolidWorks file is saved as well. When in SolidWorks a database is opened, the SolidWorks file is loaded and the manufacturing feature tree and other information is added to the SolidWorks user interface.

Figure 6.13 The manufacturing feature tree and the selected fabrication methods and machines.

6.2.4 Production planning As part of the research on a concept for concurrent manufacturing planning and control, recently a prototype of a decision support system for integrated order planning (EtoPlan) has been developed (Giebels, 2000). The prototype has also been developed based on the information structures related to the Manufacturing Engineering Model, though the prototype system for Information Management was not used in the implementation of the prototype of EtoPlan. The EtoPlan concept is based on the construction of holarchies, i.e. flexible temporary 69

Chapter 6

hierarchies of manufacturing resources. The holarchies are constructed by dynamically grouping the resources acccording to the requirements of individual orders. Within the EtoPlan concept the holarachies are referred to as Applicability Groups. A resource is considered applicable if: • The resource is considered capable of meeting the already known technological requirements in the roughly defined process plans for executing (a part of) the order • The resource is considered to be available during a time period that is roughly planned for executing (a part of) the order. Therefore, an Applicability Group (AG) is a virtual group of all the resources that are considered applicable for the (partial) execution of a given order. During the product development cycle an AG will constantly be adapted by the results from executed tasks like process planning and production planning. Because AGs consist of resources, they are stored in the Resource Information Structure. An applicability view can be constructed by combining the capability and capacity view for managing the AGs. Besides AGs, it also possible to construct Method Applicability groups (MAGs). A MAG consists of a virtual group of all production methods, which are considered applicable for the (partial) execution of a given order. In order to be able to support, for example, macro process planning, the planning system has to be able to deal with incomplete and uncertain information. Therefore, EtoPlan models uncertainty in its planning concept. The lead times and processing times of orders are modelled by a Betadistribution, fitted on the minimum, maximum and most likely value estimation. The start times are modelled by a normal distribution. A generic order profile based on these assumptions is depicted in Figure 6.14.

Figure 6.14 A generic order profile.

By summing the resource profiles of all resources belonging to an AG, an AG availability view is constructed (Figure 6.15). The solid line represents the estimated loading profile of the group of resources (AG) that corresponds to the specific order for which the order profile is shown in the figure. The dotted lines show the loading profile if the estimations of the maximum lead times are considered instead of the Beta-distribution of the lead time. When the order shown in Figure 6.15 has to be planned, it can be moved along the time axis. Besides the effect on the loading profile, the resulting variable costs that are a result of planning decisions can be considered. The costs can be made visible by a cost view based on the AG availability view. The costs that are relevant are (Huttinga, 2000): • Extra payments for overtime work; • The price for subcontracting minus the variable costs for in-house production; • Inventory costs due to excessive Work In Process (WIP); • Lateness costs, in particular penalty costs; • Earliness costs for short time delivery of blank materials. An example cost structure for these variable costs is depicted in appendix B (figure B.7).

70


Figure 6.15 An AG availability view.

Figure 6.16 A cost diagnostic view. 71

Chapter 6

6.3

The cost estimation system

6.3.1 Cost Models module The main purpose of the Cost Models module is the definition and management of cost structures. All defined cost structures are stored in a separate database, the cost structure database. First, the cost actuators have to be selected. Therefore, all the objects in the three information structures and their corresponding ontologies can be browsed (Figure 6.17). Cost structures can be defined for each selected cost actuator. The characteristics of a cost structure have to be defined (Figure 6.18) or an existing cost structure can be selected (Figure 6.19). The definition of the achievable accuracy might be difficult at this point because it depends on the derivation of the cost functions. However, all characteristics, including the accuracy, can be changed at any time.

Figure 6.17 The selection of cost actuators from the information structures.

Figure 6.18 The definition of a cost structure.

72


Figure 6.19 The selection of existing cost structures.

Cost types can be defined for every cost structure. The characteristics of a cost type have to be defined (Figure 6.20) or an existing cost type can be selected (Figure 6.21).

Figure 6.20 The definition of a cost type.

Figure 6.21 The selection of existing cost types.

Cost functions can be defined for every cost type. The characteristics of a cost function have to be defined (Figure 6.22) or an existing cost function can be selected (Figure 6.23). For every cost type, one cost function has to be set as the active cost function. This cost function will be used in cost calculations. The cost parameters can be selected from the parameters from the three information structures (Figure 6.24). Before a cost function is saved, it is send to a MatLab engine, which checks the syntax of the function. When a function is not a valid function the probable cause is presented to the user, so the cost function can be adjusted accordingly. 73

Chapter 6

Figure 6.22 The definition of cost functions.

Figure 6.23 The selection of existing cost functions.

Figure 6.24 The selection of cost parameters.

The cost structures defined, can be controlled with a cost structure view (Figure 6.25). Normally, it shows all cost structures that are defined. For a more focussed view it is possible to use filters (Figure 6.26). It is possible to filter on almost all objects and object types present in the cost structure database. After a filter is applied, all the cost structures that contain the selected objects are shown.

74


Figure 6.25 The cost structure view.

75

Chapter 6

Figure 6.26 The definition of filters on cost structure view.

6.3.2 Cost Determination module From the cost estimation system, it is possible to start a cost calculation. It is necessary to set a number of parameters to start and control the cost calculation process (Figure 6.27). The user has to indicate the product for which the calculation has to be performed by selecting the product information structure containing all the available product information. It has to be indicated which kind of calculation has to be performed, i.e. a cost estimation or a cost calculation. Based on this the system knows which data has to be used, i.e. estimated data or actual production data. The selection of the appropriate cost structures can be controlled by either setting the method and/or type or by setting the accuracy. By setting a certain method, e.g. ABC, only cost structures of that method are used. It is possible to select a certain type, e.g. generative or variant based, which further limits the number of appropriate cost structures. When different types of cost calculations are allowed, this can be indicated as well. When the accuracy option is set, only the cost structures that can produce the required accuracy are used.

76


Figure 6.27 The possible parameters for the cost determination module.

Normally, it is not practical that a user uses this user interface. Instead, the user can call a function in the Cost Determination DLL. The function call contains the parameters as described above. Additionally the database containing the cost structures has to be indicated. After the calculation has been performed the product information database is extended with cost structures and cost values. The user can refer to this database in order to view t he calculated costs. With the current prototype it is only possible to do a cost calculation by setting a costing method and one type of cost calculation (Veltman, 2000). After the calculation is started, the cost structures of the specified method and type are collected from the database with all cost structures. With this reduced set of cost structures, the product information structure is searched for cost actuators. When a cost actuator is found, the accompanying cost structure is copied to the product information structure. The value of the cost parameters is searched for and if found, it is copied to the cost parameter elements. When all cost structures have been added to the product information structure and all cost parameters have a value, the cost functions are calculated. The cost parameters in the cost function are replaced by their values and the function is send to a MatLab engine. The result returned by the MatLab engine is stored in the value field of the cost function element. Next, the values of the cost functions are added per cost type, the values of the cost types are added per cost structure and the values of the cost structures are added per cost actuator. A MatLab engine also performs these summations and the results are stored in the value field of the appropriate elements. Next, the total assembly or component costs are calculated by summing the costs of every cost actuator. The result is stored in a new cost element, which is related to the component element. The resulting costs can be viewed in the manufacturing feature tree of SolidWorks. The calculated costs per cost actuator and component can be made visible.

6.3.3 Cost Reports module The Cost Reports module has been implemented partially. It is possible to select products in a subject list for which a cost report has to be generated (Figure 6.28 and Figure 6.29). For every subject list, multiple cost reports can be defined (Figure 6.28). For every cost report, all objects from the three information structures and the cost structures can be selected (Figure 6.30). The information in a cost report can be arranged per subject or per subject list. A cost report can contain information about cost estimates and/or information about the actual costs.

77

Chapter 6

Figure 6.28 Global definition of cost reports.

Figure 6.29 Definition of the subject list for a cost report.

78


Figure 6.30 Definition of the information of the subject list to be displayed in a cost report.

6.4

Variant based cost estimation

The variant based cost estimation algorithm was implemented as a DLL as well, so it can be used by any engineering task. Also, a separate executable for the variant based cost algorithm was made to allow for the manual use of the algorithm. The algorithm uses three databases: one metabase, which refers to the databases that contain the characteristics and historic product information to be used in the similarity calculation. In the configuration overview (Figure 6.31), historic products can be selected and categories for the similarity calculation can be defined. For every category it can be indicated whether the category has to be used in an automatic similarity calculation (Figure 6.32). If a category has to be used in an automatic calculation, the desired percentage of similarity has to be indicated as well. For every category, the category items and their determinative quantities can be defined (Figure 6.33).

79

Chapter 6

Figure 6.31 The configuration of the variant based cost estimation algorithm.

Figure 6.32 Definition of categories.

The user interface for manual use is shown in Figure 6.34. Here, by means of defined categories and category items, a product can be described. The categories and category items can be selected from the defined categories and category items (Figure 6.35). Furthermore, the number of category items, the value of the determinative items can be set and the desired percentage of similarity can be indicated. The type of product has to be indicated as well. Based on the input defined, a percentage of similarity is calculated for every product in the “historic database”. When a product meets the desired percentages of similarity, the product is shown in the output field together with its costs and the percentage of similarity per category (Figure 6.34). The estimated cost equals the average costs of all the products shown in the output field.

80


Figure 6.33 Definition of the determinative quantity per category item.

Figure 6.34 Overview of input and output of the variant based cost estimation algorithm.

81

Chapter 6

Figure 6.35 The definition of a characteristic.

In case of an automatic similarity calculation, the product information structure for which a cost estimate has to be generated has to be indicated. From this product structure, the categories and category items, which have to participate in the automatic calculation, are collected. The number of category items and the value of the determinative items are collected as well. The same procedure is carried out for every product in the “historic database”. For every historic product, the percentage of similarity is calculated. When the percentage of similarity of a historic product meets the percentages of similarity set for an automatic calculation, the product is used to calculate a cost estimate.

6.5

Concluding remarks

A test of the prototype implementation of the cost estimation system is reported in the next chapter by means of an example. Because not all modules of the cost estimation architecture have been implemented and because the implemented modules have not been implemented completely, not all features of the cost estimation architecture could be tested. The fact that the modules Data Analysis, Risk Analysis and Data Tuning have not been implemented, causes that many features of the cost estimation architecture are not available in the prototype implementation. Because the Data Analysis module is not present, it is not possible to deduce cost functions from historic data within the system. This limitation can be resolved by using a separate analysis program. However, because of the lack of historic data no cost functions have been derived. It is also not possible to compare cost estimates with actual costs. The absence of the Risk Analysis module makes it impossible to generate accuracy information about the cost functions and to use this information in cost calculations. Because the Data Tuning module is not available, it is not possible to consider different currencies or inflation. The Cost Model module allows the definition of multiple cost models. The Cost Determination module can only generate a cost estimate for a component when one specific cost model is selected. Therefore, a hybrid cost calculation is not yet possible. It is also not possible to generate a cost estimate based on a pre-set accuracy level or based on the information available. Because the module Cost Reports has only been implemented partially, it is not possible to generate cost reports yet. Only the described systems for design and process planning are capable of using the information structures. This means that only cost support of the designer and process planner can be considered. Cost control can only be tested partially because the viewing capabilities of the results of a cost calculation are limited. The implementation of the variant based cost estimation algorithm has proven to be very slow. The search algorithms used have to be optimised in order to be able to use the variant based cost algorithm in a real environment. 82

7

Application of the prototype cost estimation system in the sheet metal domain

In this chapter, the prototype cost estimation system will be demonstrated by means of an example. The example represents the manufacturing of the bottom part of a sheet metal frame of a copier. Cost estimates are generated in a generative, variant based and hybrid way. Generative cost estimation is demonstrated by means of two cost models: a direct costing model and an activity based cost model. The way costing support and cost control is performed by means of the prototype cost estimation system will be elaborated.

7.1 Example product and context information The product used in this example is the bottom part of a sheet metal frame of a copier, from here on referred to as bottom-frame. The bottom-frame is an assembly, consisting of nine components (see Figure 7.1). The components are depicted in more detail in appendix E.1. The material of the bottom-frame is steel sheet of 2 mm thickness, all bend radii are 3 mm and the components are welded together.

Figure 7.1 Bottom part of a sheet metal frame of a copier (the bottom-frame).

In the manufacturing company, which manufactures this bottom-frame, six departments can be distinguished, see Table 7.1. Based on historic information of the product and prognoses for the coming year, a number of assumptions about production and the costs have been derived for the coming year, see Table 7.2 and Table 7.3.

83

Chapter 7

Department Assembly Components Engineering Logistics Inspection General Painting

Description Welding Laser-cutting, punching, bending, painting Design, process planning, production planning Purchasing, stock, transport Inspection of components and assembly Management, administration, sales Painting of components Table 7.1 The departments.

Number of copiers per year Batch size Life span tools Number of orders per year Number of subassemblies per product Number of components per product Production time department Components per product Production time department Assembly per product

14,600 (-) 2000 (-) 150,000 (pieces) 7.3 (-) 6 (-) 9 (-) 20.13 (min) 13.82 (min)

Table 7.2 Assumptions about the production.

Total direct costs per year

Overhead costs

1,776,868.00 Assembly Components Engineering Logistics Inspection General

180,000.00 840,000.00 120,000.00 400,000.00 160,000.00 960,000.00

Total overhead costs per year

2,660,000.00

Table 7.3 Expected direct and overhead costs per year.

For every component a process plan is generated by the process planning department. The resources used are listed in appendix E.2. The methods and resources are selected according to the overview in Table 7.4. The process planning feature tree for component 8 is depicted in Figure 7.2 as an example, only the methods and resources related to the manufacturing features are depicted in this figure. Manufacturing feature Bend Component Contour Depression Hole irregular Hole rectangular round Hole rectangular sharp Hole round

Methods/resources Bending on the Amada FBD3512E Assembly, general, inspection, logistic and painting Lasercutting on the Trumpf 2507 Punching on Amada Vipros-357 with a universal tool Added to the contour Added to the contour Punching on Amada Vipros-357 with a universal tool Punching on Amada Vipros-357 with a universal tool

Table 7.4 Overview of method and resource selection.

84


Figure 7.2 A detail of the process planning tree of component 8.

A summary of the process plans is listed in Table 7.5. This information will be used for the cost calculations.

Comp. 1 2 3 4 5 6 7 8 9 Total

Material A 2 (mm ) 350463 694610 105423 239988 3932 95477 334399 21909 17009 1863210

Lasercutting L t (mm) (min) 5782 2.9 5306 2.7 1927 1.0 3008 1.5 319 0.2 1644 0.8 5625 2.8 853 0.4 932 0.5 25395 12.7

Bending # t (-) (min) 10 2.0 10 2.0 2 0.4 4 0.8 3 0.6 8 1.6 7 1.4 1 0.2 3 0.6 48 9.6

Nibbling Painting Assem. # t t t (-) (min) (min) (min) 37 0.2 0.9 2.6 37 0.2 1.9 5.2 8 0.0 0.3 0.8 67 0.3 0.6 1.8 0 0.0 0.0 0.0 2 0.0 0.3 0.7 45 0.2 0.9 2.5 5 0.0 0.1 0.2 0 0.0 0.1 0.1 199 1.0 5.0 13.8

Table 7.5 Process planning information.

85

Chapter 7

7.2 Example cost models The cost models that are discussed here, representing direct costing and Activity Based Costing, only differ in the way the overhead costs are determined. Therefore, the costs of the bottom-frame are split into direct costs and overhead costs (equation 7.1). TPC = DC + OC

(7.1)

with: TPC : total product costs DC : direct costs OC : overhead costs

Because the overhead costs are different for the different cost models, they are discussed separately in the next two sections. The direct costs are composed of costs for material, tools, component production, assembly and painting (equation 7.2). DC = MC + TC + CPC + AC +PC

(7.2)

with: MC : material costs TC : tool costs CPC : component production costs AC : assembly costs PC : painting costs

The way in which the components of the direct costs are calculated is depicted in equation 7.3 to 7.7 inclusive. MC = MA ⋅ MR TC =

(7.3)

(TTC / LST )

TTU CPC = LPT ⋅ LMMR + PUPT ⋅ PUMMR + PRPT ⋅ PRMMR AC = ASSPT ⋅ ASSMMR PC = PPT ⋅ PMMR 2

with: MA: material area (mm ) 2 MR: material rate (costs/mm ) (dependent on material and thickness) TTC : total tool costs (costs) LST : life span of the tools (number of products) LPT : laser production time (min) TTU : Total number of tool usage per product (-) PUPT : punch press production time (min)

(7.4) (7.5) (7.6) (7.7)

PRPT : press brake production time (min) LMMR: laser man-machine rate (costs/min) PUMMR: punch press man-machine rate (costs/min) PRMMR: press brake man-machine rate (costs/min) ASSPT : assembly production time (min) ASSMMR: assembly man-machine rate (costs/min) PPT : painting production time (min) PMMR: painting man-machine rate (costs/min)

The cost actuators to which these costs are related are: Material, Tool, Laser, Punch press, Press brake, Assembly and Painting. The cost structures defined for these cost actuators are depicted in appendix E.3.

7.2.1 Direct Costing As stated in section 2.1.2, direct costing uses the direct costs as allocation base for the determination of the overhead costs. The overhead costs can be determined with equation 7.8. The required overhead rate required, is determined with equation 7.9.

86


OC = DOR ⋅ DC DOR =

(7.8)

OCY

(7.9)

DCY

with: DOR: overhead rate OCY : overhead costs per year DCY : direct costs per year

With the values from section 7.1, the overhead rate can be determined (equation 7.10) and the overhead costs can determined with equation 7.11. DOR =

2,660,000.00

1,776,868.00 DOC =1.4970 ⋅ DC

=1.4970

(7.10) (7.11)

The cost actuator to which these costs are related is: Component. The cost structure defined for this cost actuator is depicted in appendix E.3.

7.2.2 Activity Based Costing For the definition of the Activity Based Costing model, the five implementation steps described section 2.5.2 have been used. The results of the first four steps are depicted in Table 7.6, they are based on the values given in section 7.1. Activity centre General Inspection Logistics Component production Assembly

Activity pool General Inspection Logistic Component production Assembly

Allocation base Number of orders (-) Number of components (-) Number of subassemblies (-) Production time (min) Production time (min)

Costs 960,000.00 160,000.00 400.000.00 840,000.00 180,000.00

Table 7.6 The Activity Based Costing arrangement.

Based on this Activity Based Costing arrangement, the overhead costs can be described by equation 7.12. Step five of the implementation of Activity Based Costing means the determination of the rates, see equation 7.13 to 7.17 inclusive. OC = GR ⋅ NOO + IR ⋅ NOC + LR ⋅ NOP + LPT ⋅ ORLM + PRPT ⋅ ORPR + PUPT ⋅ ORPY + ASSR ⋅ ASSPT GR = IR =

GY

(7.13)

NOOY ⋅ NOPRY IY

(7.14)

NOCY ⋅ NOPRY LY

LR =

(7.15)

NOPY ⋅ NOPRY

ORLM = ORPR = ORPU = ASSR =

ARY PTY A ⋅ NOPRY

(7.12)

CPY PTY CP ⋅ NOPRY

(7.16) (7.17)

87

Chapter 7 with: GR: general rate NOO: number of orders IR: inspection rate NOC : number of components LR: logistics rate NOP: number of subassemblies ORLM : laser rate ORPR: press brake rate ORPU : punch press rate PT CP: production time component production ASSPT : production time assembly ASSR: assembly rate

GY : general overhead costs per year NOOY : number of orders per year IY : inspection overhead costs per year NOCY : number of components per product per year LY : logistics overhead costs per year NOPY : number of subassemblies per product per year CPY : component production costs per product per year PTY CP: production time components per product per year PTYA: production time assembly per product per year ARY : assembly overhead costs per year NOPRY : number of products per year

Using the data from section 7.1, the rates are (equation 7.18 to 7.22 inclusive): GR = IR =

960,000.00

7.3 ⋅ 14,600 160,000.00

(7.18)

=1.83

6.0 ⋅ 14,600 400,000.00

LR =

= 9.01

9.0 ⋅ 14,600

(7.19)

= 3.04

(7.20)

ORLM = ORPR = ORPU = ASSR =

180,000.00 13.82 ⋅ 14,600

840,000.00 20.13 ⋅ 14,600

= 2.86

(7.21)

= 0.89

(7.22)

In terms of Activity Based Costing the rates are the costs per allocation base (Table 7.7). Activity centre

General Inspection Logistics Component production Assembly

Activity pool

General Inspection Logistic Component production Assembly

Allocation base

Number of orders (-) Number of components (-) Number of subassemblies (-) Production time (min) Production time (min)

Costs per allocation base 9.01 1.83 3.04 2.86 0.89

Table 7.7 The finalised Activity Based Costing arrangement.

Now, the overhead costs can be calculated with equation 7.23. OC = 9.01⋅ NOO + 1.83 ⋅ NOC + 3.04 ⋅ NOP + 2.86 ⋅ LPT + 2.86 ⋅ PRPT +

2.86 ⋅ PUPT + 0.89 ⋅ ASSPT

(7.23)

The cost actuators to which these costs are related are: General, Inspection, Logistics, Laser, Press brake, Punch press and Assembly. The cost structures defined for these cost actuators are depicted in appendix E.3.

7.3 Example cost calculations The cost actuators that have been distinguished in the previous sections are: Assembly, General, Inspection, Laser, Logistics, Material, Painting, Press brake, Punch press and Tool. For these cost actuators, cost structures have been defined as part of the Direct Costing and Activity Based Costing models, see appendix E.3. 88


7.3.1 Generative cost estimation The costs of the components are calculated using both cost models. The calculations are started by selecting the cost model and the method of calculation (generative). The results of the calculations are listed in Table 7.8. C C T

3 6 . 1 6

3 7 . 9 8

0 2 . 8 1

5 8 3 8 8 5 3 8 . . 5 . 9 . 4 . 0 . 8 . 3 0 3 9 8 6 6 5 0 4 1 5 3

C O

8 9 . 6 3

4 8 . 3 5

2 9 . 0 1

1 5 6 9 5 1 1 5 . . 1 . 9 . 0 . 6 . 1 . 2 4 2 1 5 3 4 3 8 2 1 3 1

7 8 . 2 5

6 6 . 5 6

2 5 . 3 2

2 7 . 6 3

0 2 . 5 1

0 1 . 7 2

6 6 . 9 4

8 9 . 5 1

7 4 . 7 1

8 1 . 4 0 3

C L

4 4 4 4 4 4 4 4 4 6 0 . 0 . 0 . 0 . 0 . 0 . 0 . 0 . 0 . 3 . 3 3 3 3 3 3 3 3 3 7 2

C I

3 3 3 3 3 3 3 3 3 7 2 . 2 . 2 . 2 . 2 . 2 . 2 . 2 . 2 . 0 . 1 1 1 1 1 1 1 1 1 1 1

C G

0 0 0 0 0 0 0 0 0 0 3 . 3 . 3 . 3 . 3 . 3 . 3 . 3 . 3 . 7 . 7 7 7 7 7 7 7 7 7 5 6

C P

6 0 0 5 1 8 3 4 3 0 6 . 3 . 2 . 4 . 0 . 1 . 6 . 0 . 0 . 5 . 0 1 0 0 0 0 0 0 0 3

1 8 9 8 3 3 1 4 2 9 3 . 5 . 6 . 5 . 0 . 6 . 2 . 1 . 1 . 2 . 2 4 0 1 0 0 2 0 0 2 1

C A O C A D C T C P P O C P P D

2 1 5 5 2 0 4 1 9 9 8 . 6 . 5 . 2 . 0 . 5 . 7 . 1 . 0 . 6 . 1 3 0 1 0 0 1 0 0 9

g n i t s 1 9 4 5 0 6 8 5 0 8 o 1 . 9 . 2 . 9 . 0 . 0 . 3 . 1 . 0 . 8 . C 1 0 0 1 0 0 1 0 0 5 t c e r i D

7 3 8 5 0 2 6 5 0 6 3 . 3 . 0 . 6 . 0 . 0 . 4 . 0 . 0 . 9 . 0 0 0 0 0 0 0 0 0 1

g 2 . n 8 i 1 t s o C d 1 1 e . s 1 a B y t i 7 v . i t 3 c 0 A

1 5 5 2 0 4 1 9 9 6 . 5 . 2 . 0 . 5 . 7 . 1 . 0 . 6 . 3 0 1 0 0 1 0 0 9 9 4 5 0 6 8 5 0 8 9 . 2 . 9 . 0 . 0 . 3 . 1 . 0 . 8 . 0 0 1 0 0 1 0 0 5 3 8 5 0 2 6 5 0 6 3 . 0 . 6 . 0 . 0 . 4 . 0 . 0 . 9 . 0 0 0 0 0 0 0 0 1

7 3 8 5 0 2 6 5 0 6 3 . 3 . 0 . 6 . 0 . 0 . 4 . 0 . 0 . 9 . 0 0 0 0 0 0 0 0 0 1 0 0 4 8 1 6 9 7 1 6 7 . 7 . 1 . 2 . 7 . 5 . 9 . 5 . 7 . 3 . 5 5 1 2 1 4 3 0 1 7 2

C B P O C B P D

6 0 0 5 1 8 3 4 3 0 6 . 3 . 2 . 4 . 0 . 1 . 6 . 0 . 0 . 5 . 0 1 0 0 0 0 0 0 0 3

0 0 6 2 4 4 6 8 4 4 8 . 8 . 5 . 1 . 8 . 2 . 9 . 2 . 8 . 4 . 2 2 0 1 0 2 1 0 0 3 1

0 0 6 2 4 4 6 8 4 4 8 . 8 . 5 . 1 . 8 . 2 . 9 . 2 . 8 . 4 . 2 2 0 1 0 2 1 0 0 3 1 7 9 6 0 6 5 4 2 3 2 2 . 5 . 7 . 3 . 4 . 3 . 0 . 2 . 3 . 3 . 8 7 2 4 0 2 8 1 1 6 3

C L O C L D

9 3 0 1 5 0 7 9 0 4 0 . 4 . 7 . 2 . 4 . 3 . 8 . 1 . 3 . 5 . 8 7 2 4 0 2 7 1 1 5 3

9 3 0 1 5 0 7 9 0 4 0 . 4 . 7 . 2 . 4 . 3 . 8 . 1 . 3 . 5 . 8 7 2 4 0 2 7 1 1 5 3

C M

0 3 5 1 1 7 5 1 8 1 8 . 4 . 9 . 7 . 1 . 6 . 3 . 6 . 4 . 1 . 9 9 2 6 0 2 9 0 0 2 1 5

0 3 5 1 1 7 5 1 8 1 8 . 4 . 9 . 7 . 1 . 6 . 3 . 6 . 4 . 1 . 9 9 2 6 0 2 9 0 0 2 1 5

#

1

2

3

4

5

6

7

8

. t 9 o t

1

2

3

4

5

6

7

8

. t 9 o t

, s t s l o a c t o y t l b = m C e s s C a T , t s c t e s r o i , d c s t s = d a o C e c h e A r k D e a , v o r s b t = s s s o C c e r l O p o s o , t t c t s e r = o c i d C c i T = , t s s i C t g s B o l o P c = D s , s C e s t s r p L , o h s c c t s r n o e s u c a p n l o i d d t a a e e h c e h p r r s e e n v v i o o = = = C I C C , L P P t s O O s , , o s s c t s t l o s a r c o c r s e n e e s s e g a l r t p = c h e C r c i n d u G , s p t = t s o C c e r c L i D d g n , = i t s t n s C i o P a c l P p a D = i r s , e t C t P a s , o s m c t s = e o k c C a r y b l M , s b r m e s r e b e p s s m d a u a d s n e a t t h e s n r h o e e r c n v e t o o v n p o e m = = n o C o c B C p = P A m o # O O c

Table 7.8 Results of generative cost models. 89

Chapter 7

The example demonstrates the possibility to use different cost models at the same time. Based on the results, analysis possibilities and accuracy, of both cost models, a choice between the two models can be made.

7.3.2 Variant based cost estimation Because the variant based cost estimation program is inefficient, the cost calculation for large products like the bottom-frame is very time consuming. Therefore, the variant based cost estimation program will be demonstrated by means of component 3, 5, 6, 9 and variants of component 8. In Table 7.9 some historic information about these components is listed. In all components the contour is cut by laser, the processes listed in Table 7.9 refer to the holes in the components.

Comp. 3 5 6

8

9

Variant 1 1 1 1 2 3 4 5 6 7 8 9 10 11 12 1

Material steel steel steel steel steel zincor zincor steel steel zincor zincor steel steel zincor zincor steel

Hole rectangular sharp

Hole round 5 0 2 5 5 5 5 4 4 4 4 3 3 3 3 0

3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Processes punching punching lasercutting punching lasercutting punching lasercutting punching lasercutting punching lasercutting punching lasercutting punching -

Costs 23.52 15.20 27.10 15.92 15.98 15.96 16.02 15.89 15.93 15.93 15.97 15.85 15.88 15.89 15.92 17.47

Table 7.9 Variants of component 8 of the bottom-frame.

The similarity of component 8 variant 5 with the other components is calculated, the results are listed in Table 7.10. The similarity is calculated for the material, geometry (features) and macroprocess planning (production methods). When for instance a designer requires 100 % similarity in material and 90 % similarity in geometry, component 8 variant 1, 2, 6, 9 and 10 are suitable for the cost calculation. In this case the cost estimate is 15.91. A process planner could require 100 % similarity in material, 50 % similarit y in geometry and 90 % similarity in macro-process planning. In this case, component 8 variant 1 and 9 are suitable and the cost estimate becomes 15.89.

90


Comp. 3 5 6

Variant 1 1 1 1 2 3 4 5 6 7 8 9 10 11 12 1

8

9

Material 100.00 100.00 100.00 100.00 100.00 0.00 0.00 100.00 0.00 0.00 100.00 100.00 0.00 0.00 100.00

Similarity (%) Geometry 62.50 50.00 54.17 93.33 93.33 93.33 93.33 100.00 100.00 100.00 91.67 91.67 91.67 91.67 66.67

Macro-PP 50.00 75.00 37.50 97.92 77.50 97.92 77.50 77.50 100.00 77.50 97.14 77.50 97.14 77.50 66.67

Table 7.10 Similarity results of the variant based cost calculation.

7.3.3 Hybrid cost estimation A hybrid cost estimation can occur when the method (generative/variant) for the cost calculation is not specified. When for instance the process plans for all components except component 8 have been determined, the cost estimates (generative) of components 1 to 7 and 9 can be combined with the cost estimate (variant) of component 8, see Table 7.11. Component 1 2 3 4 5 6 7 9 8 Total

Method

Generative, Activity Based Cost estimation

Variant Based Cost estimation Hybrid cost estimation

Costs Design Macro-PP 52.87 52.87 65.66 65.66 23.52 23.52 36.72 36.72 15.20 15.20 27.10 27.10 49.66 49.66 17.47 17.47 15.91 15.89 304.11 304.09

Table 7.11 Example of hybrid cost estimation.

7.4 Costing support and cost control At any point in the product development cycle, a cost calculation can be started. For every object to which costs are related, the costs can be made visible (see Figure 7.3). Based on the cost structures, the costs can be arranged into different types and the cause of the costs can be deduced. Based on this cost information, well-founded decisions can be taken.

91

Chapter 7

Figure 7.3 Costing support during process planning (component 2).

When the process planner makes changes to the process plan, the costs can immediately be recalculated and the new costs can be made visible in the same way. Based on the cost information, he can create the final process plan.

92

8

Conclusions and recommendations

The objective of the research described in this thesis was the development of a generic architecture for a cost estimation system. The system had to be integrated in the product development cycle and had to enable costing support and cost control. A generic cost estimation architecture was aimed for, which had to be applied in the sheet metal domain. In chapter two, a literature review on relevant cost aspects has been given. The review has been structured according to an existing cost control architecture. The literature review has been complemented with information about time and cost functions of sheet metal manufacturing processes. The Manufacturing Engineering Reference Model has been introduced in chapter three. This reference model and its associated information structures have been developed specifically for the integration of the processes in the whole product development cycle. It is also an ideal base for the control and support of engineering tasks with cost information. Based on the literature review and the Manufacturing Engineering Reference Model, a generic cost estimation architecture has been developed in chapter four. The use of this architecture has been explained in chapter five. A partial prototype of the cost estimation architecture has been implemented and applied in the sheet metal domain by means of an example as described in chapters six and seven. This chapter gives the conclusions about the use of the cost estimation architecture. In addition some recommendations, concerning the prototype implementation and future research will be given.

8.1 Conclusions The cost estimation system is based on a generic cost estimation architecture. The prototype implementation is not based on specific manufacturing domain characteristics. Specific domain knowledge about costs is added to the system by means of the cost models defined with the system. Therefore, the system can be applied in different manufacturing domains. The cost estimation architecture is based on the so-called Manufacturing Engineering Reference Model. This reference model has a central information management kernel that facilitates both the availability and the accessibility of meaningful representations of the evolving manufacturing information. As such, the cost estimation architecture fits in the system of co-operating architectures that can be arranged around the information management kernel. This means that the cost estimation function is integrated in the manufacturing process based on the information management kernel. In the Cost Models module of the cost estimation system, the cost models are defined based on the cost structure. The cost structure incorporates cost types, cost functions and cost parameters. The flexibility of the cost structure and the relation to any object that causes costs allows the definition of different cost models. Because different cost models can be defined, it is possible to set them side-by-side and compare the results. Moreover, it is possible to define different cost models for different stages of the product development cycle, i.e. for different levels of aggregation. This means that the complete product development cycle can be supported with cost information. As a result, for any engineering task cost information can be made available for decision making. The cost structures can be attached to the information structures, used in the Manufacturing Engineering Reference Model. As a result, the costs can be calculated for any object at any aggregation level in an information structure. Additionally, the costs can be stored in a differentiated way. 93

Chapter 8

Through the use of the information management kernel, historic information can be accessed, which can be used for the deduction of cost functions or for variant based cost estimation. The Data Analysis module and the Data Tuning module of the cost estimation architecture can assure a proper use of the historic information. The use of the centrally accessible information structures facilitates cost control. A cost view can show the position of costs in an information structure. The cause of these costs can be deduced based on the information structures and the cost structures. All the cost control functions determined by Liebers are present in the cost estimation architecture. So, it is possible to apply the four cost feedback loops: the engineering and planning feedback loop, the order acceptance feedback loop, the production feedback loop and the accounting feedback loop. The major advantages of the proposed variant based cost estimation method are the flexibility in defining comparison criteria and in valuing the comparison criteria. Because of this flexibility, the method can be used in the whole product development cycle. Based on the use of the information structures, hybrid cost estimation can be dealt with more easily. The application of the (partial) prototype cost estimation system has proven the suitability of the Cost Models module and the Cost Determination module. The definition of different cost models is possible. Although the Cost Determination module was not completely implemented, the use of the cost structures for the cost calculations has proven the generic nature of the cost calculation method. Based on this experience, the extension of the Cost Determination module will be relatively simple.

8.2 Recommendations For the demonstration of the cost estimation architecture, a prototype system was implemented and tested by means of an example in the sheet metal domain. The implementation did not cover the complete architecture; only the most essential modules were implemented partially. In order to further test the cost estimation architecture, all the modules have to be implemented fully. Currently, the Cost Determination module can only calculate the costs based on a selected cost model. The module has to be extended in order to allow cost calculations based on a pre-specified accuracy and based on the information available. This extension will enlarge the flexibility of the cost estimation system. Because the Cost Determination module is the cost calculation “engine”, this extension is very important. Especially the Risk Analysis module has to be incorporated in the cost estimation system, because the accuracy of cost estimates is important information for a proper use of cost information. As described before, an existing analysis program can be used for this purpose. Concerning the variant based cost estimation algorithm, it has to be noticed that the search and similarity calculation algorithms have to be optimised. Furthermore, the prototype cost estimation system needs to be tested in practice. A test in combination with other systems based on the Manufacturing Engineering Reference Model would be ideal, though this is not necessary. The system can be enhanced by creating “standard” costs structures for various cost models and manufacturing domains. The “standard” cost structures can speed up the implementation of the cost estimation system in practice.

94

References Agarwal, 1994

M. Agarwal, A. Kamrani, H.R. Parsaei, ‘An automated coding and classification system with supporting database for effective design of manufacturing systems’, Journal of Intelligent Manufacturing 5, pp. 235-249.

Arentsen, 1995

A.L. Arentsen, A generic architecture for factory activity control , Ph.D. thesis, University of Twente, Enschede, The Netherlands.

Asiedu, 1998

Y. Asiedu, P. Gu, ‘Product life cycle cost analysis: state of the art review’, International Journal of Production Research 36/4, pp. 883908.

Bear, 1985

T. Bear, ‘With group technology, no one reinvents the wheel’, Mechanical Engineering 60/November, pp. 60-67.

Billo, 1987

R.E. Billo, R. Rucker, D.L. Shunck, ‘Integration of a group technology classification and coding system with an engineering database’, Journal of Manufacturing Systems 6/1, pp. 37-45.

Blommaert, 1998

A.A.M. Blommaert, J.M.J. Blommaert, Bedrijfseconomische analyses, rd 3 ed., Houten, Educatieve Partners Nederland BV. (in Dutch)

Bode, 1998a

J. Bode, ‘Decision support with neural networks in the management of research and development: concepts and application to cost estimation’, Information & Management 34, pp. 33-40.

Bode, 1998b

J. Bode, ‘Neural networks for cost estimation’, Cost Engineering 40/1, pp. 25-30.

Chisholm, 1990

A.W.J. Chisholm, ‘Nomenclature and Definitions for Manufacturing Systems’, Annals of the CIRP 39/2, pp. 735-742.

Cooper, 1991

R. Cooper, R.S. Kaplan, The design of cost management systems; text, cases and readings , New Jersey, Prentice-Hall Inc.

Cuesta, 1998

E. Cuesta, J.C. Rico, S. Mateos, C.M. Suarez, ‘Times and costs analysis for sheet-metal cutting processes in an integrated CAD/CAM system’, International Journal of Production Research 36/6, pp. 1733-1747.

Geiger, 1996

T.S. Geiger, D.M. Dilts, ‘Automated design-to-cost: integrating costing into the design decision’, Computer-Aided Design 28/6-7, pp. 423-438.

Geiger, 1997

M. Geiger, J. Knoblach, ‘Cost estimation of sheet metal parts with neural networks’, Proceedings of the 5th International Conference on Sheet Metal: SheMet’97 , Belfast, pp. 69-78.

95

Giebels, 2000

M.M.T. Giebels, EtoPlan, a concept for concurrent manufacturing planning and control, Ph.D. thesis, University of Twente, Enschede, The Netherlands.

Greska, 1995

W. Greska, Wissenbasierte Analyse und Klassifizierung von Blechteilen, Ph.D. thesis, Universität Erlangen-Nürnberg, Erlangen, Germany. (in German)

Horngren, 1994

C.T. Horngren, G. Foster, S.M. Datar, Cost accounting, a managerial th emphasis, 10 ed., London, Prentice Hall.

Huttinga, 2000

A.P. Huttinga, Cost analysis in capacity loading – representation of cost information to support cost estimation and minimisation , M.Sc. thesis, University of Twente, Enschede, The Netherlands.

Kalpakjian, 2001

S. Kalpakjian, S.R. Schmid, Manufacturing engineering and th technology, 4 ed., Prentice Hall, inc., New Jersey.

Kals, 1998

H.J.J. Kals, D. Lutters, ‘The role of information management in intelligent manufacturing’, Proceedings of the CIRP International Seminar on Intelligent Computation in Manufacturing Engineering , Capri, pp. 21-28.

Kals, 2000

H.J.J. Kals, D. Lutters, A.H. Streppel, E. ten Brinke, ‘Information th management in manufacturing’, Proceedings of the 8 International Conference on Sheet Metal: SheMet’98, Birmingham , pp. 19-40.

Kiritsis, 1996

D. Kiritsis, P. Xirouchakis, ‘A software prototype for cost estimation of process plans of machined parts’, Proceedings of the International Symposium on Automotive Technology and Automation (29th ISATA) , Florence, pp. 19-26.

Kusiak, 1990

A. Kusiak, Intelligent Manufacturing Systems, Englewood Cliffs NJ, Prentice Hall.

Leung, 1996

A.C.K. Leung, C.E.R. Wainwright, R. Leonard, ‘The development of an integrated cost estimation system’, International Journal of Computer Integrated Manufacturing 9/3, pp. 190-204.

Lewis, 1987

R.L. Lewis, Y-L. Chou, ‘Group technology coding system for die casting’, Proceedings of the SDCE 14th international die casting congress and exposition , pp. 1-10.

Liebers, 1998

A. Liebers, An architecture for cost control, the use of cost information in order-related decisions , Ph.D. thesis, University of Twente, Enschede, The Netherlands.

Luong, 1989

L.H.S. Luong, ‘Process planning via computer-assisted classification and coding’, International Journal of Advanced Manufacturing Technology 4, pp. 311-320.

Luong, 1995

L.H.S. Luong, T. Spedding, ‘An integrated system for process planning and cost estimation in hole making’, International Journal of Advanced Manufacturing Technology 10/6, pp. 411-415.

Lutters, 1997a

D. Lutters, A.H. Streppel, H.J.J. Kals, ‘The role of information structures in design and engineering processes’, Proceedings of the third WDK workshop on product structuring , Delft, pp. 125-136.

96

References

Lutters, 1997b

D. Lutters, A.H. Streppel, B. Kroeze, H.J.J. Kals, ‘Adaptive press brake control in air bending’, Proceedings of the 5th International Conference on Sheet Metal: SheMet’97 , Belfast, pp. 471-480.

Lutters, 1999

D. Lutters, H.J.J. Kals, ‘Control of design and manufacturing processes based on information content’, Proceedings of the 1999 CIRP International Design Seminar; Integration of Process knowledge into Design Support Systems, Enschede, pp. 363-372.

Lutters, 2001

D. Lutters, Manufacturing integration based on information management , Ph.D. thesis, University of Twente, Enschede, The Netherlands.

Maree, 1997

W.G. Maree, A.H. Basson, ‘Design for fabrication early cost estimation for small production volume’, Journal of Manufacturing Systems 26/2, pp. 95-100.

Molengraaf, 1993

J.C.M. van den Molengraaf, H.H. van Mal, J. Wijnia, ‘Selection procedures for manufacturing processes for design engineers’, Robotics & Computer-Integrated Manufacturing 10, No. 1/2, pp. 5764.

Nee, 1992

A.Y.C. Nee, A.S. Kumar, S. Prombanpong, K.Y. Puah, ‘A featurebased classification scheme for fixtures’, Annals of the CIRP 41/1, pp. 189-192.

Nollet, 1993

H. Nollet, R. Aerens, A. Witters, Lasersnijmachines of ponsmachines in de plaatbewerking:een kostenvergelijking , WTCM Report MC 103, Heverlee, Belgium.

Ou-Yang, 1997

C. Ou-Yang, T.S. Lin, ‘Developing an integrated framework for feature-based early manufacturing cost estimation’, International Journal of Advanced Manufacturing Technology 13/9, pp. 618-629.

Peklenik, 1980

J. Peklenik, J. Grum, ‘Investigation of the computer aided classification of parts’, Annals of the CIRP 29/1, pp. 319-323.

Schaal, 1993

S. Schaal, K. Ehrlenspiel, ‘Design concurrent calculation: a CAD- and data-integrated approach’, Journal of Engineering Design 4/2, pp. 7589.

Schuttert, 1995

M.A. Schuttert, Design of a feature-based coding and classification system, M.Sc. thesis, University of Twente, Enschede, The Netherlands.

Sen, 1990

A.K. Sen, M.S. Srivastava, Regression analysis: theory, methods and applications, New York, Springer-Verlag.

Shuford, 1995

R.H. Shuford Jr., ‘Activity-based costing and traditional cost allocation structures’, In: R.D. Stewart, R.M. Wyskida, J.D. Johannes nd (eds.), Cost estimator’s reference manual , 2 ed., New York, Johan Wiley & sons Inc, pp. 41-94.

Sohlenius, 1992

G. Sohlenius, ‘Concurrent Engineering’, Annals of the CIRP 41/2, pp. 645-655.

Somatech, 1998

Somatech CAD/CAM applicaties B.V., Brochure Veenendaal, The Netherlands. (in Dutch)

SomaCalc,

97

Srikantappa, 1994

A.B. Srikantappa, R.H. Crawford, ‘Automatic part coding based on interfeature relationships’, In: J.J. Shah, M. Mäntylä, D.S. Nau (eds.), Advances in Feature Based Manufacturing, Elsevier Science B.V., pp. 215-237.

Stewart, 1995a

R.D. Stewart, R.M. Wyskida, J.D. Johannes (eds.), Cost estimator’s nd reference manual , 2 ed., Johan Wiley & sons, inc, New York.

Stewart, 1995b

R.D. Stewart, ‘Detailed cost estimating’, In: R.D. Stewart, R.M. nd Wyskida, J.D. Johannes (eds.), Cost estimator’s reference manual , 2 ed., New York, Johan Wiley & sons Inc, pp. 193-231.

Thompson, 19??

F. Thompson, ‘Cost Analysis’, Unpublished paper , Fellow Willamette University, http://www.willamette.edu/~fthompso/.

Veelenturf, 1997

L.P.J. Veelenturf, Neurale netwerken: een inleiding voor de praktijk: cursushandboek , Den Haag, The Netherlands, ten Hagen & Stam. (in Dutch)

Veltman, 2000

R.J. Veltman, Design and application of a generic cost determination module and cost functions for sheet metal , M.Sc. thesis, University of Twente, Enschede, The Netherlands.

Vin, 1995

L.J. de Vin, U.P. Singh, W. Urquhart, ‘Cost pre-calculation as an aid to design and manufacture’, Proceedings of the 3rd International Conference on Sheet Metal: SheMet’95 , Birmingham, pp. 125-134.

Vliegen, 1993

H.J.W. Vliegen, Classification systems in manufacturing, managerial control of process knowledge , Ph.D. thesis, University of Twente, Enschede, The Netherlands.

Weustink, 2000

I.F. Weustink, E. ten Brinke, A.H. Streppel, H.J.J. Kals, ‘A generic framework for cost estimation and cost control in product design’, Journal of Materials Processing Technology 103, pp. 141-148.

Wierda, 1990

L.S. Wierda, Cost information tools for designers, a survey of problems and possibilities with an emphasis on mass produced sheetmetal parts, Ph.D. thesis, University of Delft, Delft, The Netherlands.

Wierda, 1991

L.S. Wierda, ‘Linking design, process planning and cost information by feature-based modelling’, Journal of Engineering Design 2/1, pp. 3-19.

Wu, 1992

X. Wu, H. Huang, ‘Group technology and its relation with CIM’, Computers in Industry 19, pp. 143-149.

Zhang, 1996

Y.F. Zhang, J.Y.H. Fuh, W.T. Chan, ‘Feature-based cost estimation for packaging products using neural networks’, Computers in Industry 32, pp. 95-113.

Zheng, 1996

H.Y. Zheng, Z.Z. Han, Z.D. Chen, W.L. Chen, S. Yeo, ‘Quality and cost comparisons between laser and waterjet cutting’, Journal of Materials Processing Technology 62, pp. 294-298.

98

Terminology Acquisition costs

Total expenditures estimated or incurred for the development, manufacture, construction and installation of an item of physical or intangible property, or the total acquisition costs of a group of such items.

Activity center

Part of the product development cycle for which the costs are registered separately on request of management.

Activity pool

Set of activities practised by a certain function.

Allocation base

A measure that is directly related the amount of an activity used.

Architecture

An architecture is a framework which defines the functions, which are required to perform the task of a system, with their input and output.

Assembly

An assembly is the result of physically combining components and/or other assemblies, according to their mutual relations as specified in the product data structure.. An assembly is the result of physically combining components and/or other assemblies, according to their mutual relations as specified in the product information structure.

Classification

The process of grouping parts into families of similar parts; similarity is based on some set of rules and principles.

Clustering

The process of grouping similar objects based on a similarity coefficient.

Coding

The arbitrary assignment of one or more symbols to a part, which when deciphered communicates specific meaning or intelligence.

Company Management

Company Management is concerned with the control of the customer orders. It is responsible for the strategic decisions concerning the range of products which will be produced and the processes and resources which are required to this end.

Component

A component is an atomic constituent of a product.

Concurrent Engineering

The simultaneous execution of shared tasks by separate departments and the control of cooperative decision-making.

Conversion costs

A grouping of direct labour and manufacturing overhead into a single summary cost element.

Cost actuator

An object that causes costs.

Cost allocation

A method or combination of methods that results in a reasonable distribution of costs.

Cost driver

A cost driving product characteristic.

99

Cost function

A relationship between costs and manufacturing characteristics, i.e. cost parameters.

Cost model

The collection of cost structures for which the method of cost calculation is the same and which covers the total manufacturing costs.

Cost modelling

The determination of the data, ground rules, assumptions and equations that permits the translation of resources or characteristics into costs.

Cost structure

A template to record the costs for all cost actuators.

Costs

Costs is the amount of money of resources spend for the production of output.

Decomposition

Decomposition of a system represented as a black box with a globally defined task, is the replacement of this representation with a configuration of system components, each with its own glabally defined subtask, that interact to realise the globally defined task of the black box.

Development costs

Costs of a system up to the point where decision is made to procure an initial increment of the production units or the operational system.

Direct costs

Direct costs are costs that can be identified specifically and consistently with an end objective (such as a product, service, program, function, or project).

Domain

A domain is the representation of one aspect system in an information structure.

Engineering

Engineering is that part of manufacturing that is concerned with the preparation of the actual production of a product. Therefore, engineering includes activities as design, process planning and production planning.

Fabrication

Fabrication addresses those operations applied during production that are not assembly operations.

Face

A face is the element of a component that separates that component from its surroundings.

Filter

A filter can exclude information from a certain view.

Fixed costs

Fixed costs are costs that do not vary with the volume of business.

Form feature

A form feature is a group of faces that together have an engineering meaning.

Group Technology

The technique of applying the same solution to similar problems.

Indirect costs

Indirect costs are costs that cannot be identified specifically and consistently with an end objective (such as a product, service, program, function, or project).


Information management includes the functionality related to the initialisation, use, analysis and maintenance of information concerning orders, products and resources. Based on this, information

100

Terminology

management can initiate processes in one of the engineering sections of the reference model. Irrelevant costs

Costs that do not play a role in a specific decision making process.

Life-cycle costs

All costs incurred during the projected life of the system, subsystem or component (research, development, test, evaluation, production, maintenance and disposal).

Manufacturing

Manufacturing refers to the series of interrelated activities and operations involving the design, the materials selection, the planning, the production and the quality assurance of the products. Manufacturing covers the complete product development cycle, of which production is only a part.

Module

A module is a reference to a sub-assembly or a group of components, which -for reasons of convenience- is considered to be a distinct section of the geometric domain of the product data structure.

Non-recurring costs

Costs that are elements of development and investment costs that generally occur only once in the life cycle of a work activity or work output.

Ontology

An ontology is an explicit specification of some topic. For our purposes, it is a formal and declarative representation which includes the vocabulary (or names) for referring to the terms in that subject area and the logical statements that describe what the terms are, how they are related to each other, and how they can or cannot be related to each other. Ontology’s therefore provide a vocabulary for representing and communicating knowledge about some topic and a set of relationships that hold among the terms in that vocabulary.

Ontology of state

The description of the stationary dependency of elements and relations.

Ontology of transition

The description of the process-related dependency of elements.

Opportunity costs

Loss of income due to not selecting the optimum alternative from a financial point of view.

Order Engineering

Order engineering addresses those activities that connect a customer order to a specific (variant of a) product and subsequently are responsible for the decisions on when a batch of products must be processed with which resources. The objective of Order engineering is the in-time execution of the production orders.

Prime costs

Cost of direct material and direct labour.

Product development cycle A combination of manufacturing activities resulting in a product. Product Engineering

Product engineering refers to all engineering activities related to the product life cycle of a specific type of product. Therefore product engineering covers aspects from functional specification to final recycling/disposal.

Production

Production is the act or process (or the connected series of acts or processes) of actually physically making a product from its material

101

constituents. Production can consist of fabrication and assembly operations. Recurring costs

Costs that are repetitive costs that vary with the quantity being produced.

Reference Model

A reference model represents a system as an organisation in terms of a structure of relatively independent, interacting components, and in terms of the globally defined tasks of these components.

Relevant costs

Costs that are present under one of several alternatives but are absent, either in whole or in part, under other alternatives (also called differential costs).

Removal costs

The cost of dismantling a unit of property owing to retirement from service.

Resource

A resource is a physical entity that is required to be able to execute a certain operation. Resources can be e.g. machine tools, tools an fixtures, but also operators and materials.

Resource Engineering

Resource engineering refers to all life cycle aspects of all resources which are required for the execution of the production activities. Resource engineering therefore includes the specification, design, development, preparation, use and maintenance of the resources of a company.

Semantic ontology

The definition of elements and relations between these relations.

Semi variable costs

Semi variable costs are costs that vary somewhat in relation to volume, but their percentage of change is not the same as the percentage of change in volume.

Step-fixed costs

Fixed cost are costs that alter their behaviour as the activity level moves from one relevant range to another.

Sunk costs

The total of all past expenditures or irrevocably committed funds related to a program/project.

Symbolic ontology

The definition of the structure the semantic ontology adheres to.

Traditional costing

Overhead is allocated to products based on volume based allocation bases e.g. labour hours, machine hours.

Variable costs

Variable costs are costs that change with the rate of production or the performance of services.

View

A view is a focused and partial representation of the information in one domain of an information structure. Views can be mutually dependent, i.e. a change in one view can force a change in another view.

102

A Time and cost functions from literature A.1 Punching Time functions for punching (Vin, 1995). T Ptot = T P

+ T C + T T

(A.1)

with: T Ptot Ptot : total processing time [sec] T P: punching time [sec] T C C : tool change time [sec] T T T: traversing time [sec]

T P

= t p N F

(A.2)

with: t p: time for a single punch stroke including lost ti me for deceleration and acceleration of the sheet [sec] N F F: number of punched features in batch [-]

T c

= N c t c

(A.3)

with: N c: number of tool changes [-] t c: time required for one tool change [sec]

N c

= N t ,min Rs

(A.4)

with: N t,min t,min: minimum number of tools [-] Rs: estimated sharing ratio [-]

( N − 1) T T

L NL W

=

−1

(A.5)

V t

with: N : number of features evenly spread over the area WxL [-] L: length of smallest rectangle enclosing all features [m] W : width of smallest rectangle enclosing all features [m] V t t : uniform traversing speed [m/sec]

A.2 Nibbling Time functions for nibbling (Vin, 1995). T Ptot = T N + T C + T T

(A.6)

with: T Ptot Ptot : total processing time [sec] T N : nibbling time [sec] T C C : tool change time [sec] T T T: traversing time [sec] N N

T N

= ∑ (T H ( p ) N H ) i

(A.7)

i =1

103

Appendix A with: N N : number of nibbled features in batch [-] T H (p): time for a single hit ( function of pitch p) [sec] N H : number of hits required to make feature i [-] i: feature i which has to be nibbled [-]

Time functions for nibbling (Nollet, 1993). T tot = T s + T w T s =

t vs

(A.8)

+ t gl N tool

(A.9)

N tot

T w = 1.05 t p1 + t p 2 + t p 3 + t p 4 + t p 5 + t p 6 + t p 7 t p1 =

t epl

(A.11)

N s

t p 2 = ( N p1 + N p 2 ) t p 3 =

a gem v pos

+ t kap

LCT  LT LT

 a gem  + t kap  +   v pos  v pos 

t p 4 = t nib N nib

t p 6 = t p 7

N s t iu

(A.13)

(A.15) (A.16)

N s

=

(A.12)

(A.14)

( N tool −1)t tw

t p 5 =

(A.10)

t af

(A.17)

N s

with: T tot tot : total processing time per workpiece [min] T s: set-up time for N tot tot [min] T w: processing time per workpiece [min] t p1: time for startposition [min] t p2: effective punching time [min] t p3: nibbling time of rectangular contours [min] t p4: nibbling time of curved contours [min] t p5: time for removing scrap [min] t p6 : fixing and unfixing time [min] t p7 : time for removing scrap [min] t vs vs: fixed set-up time [min] t gl gl: time for one tool setup [ min] N tool tool: number of different tools [-] N tot tot : total number of workpieces [-] t epl epl: time to move from zero to first cut [min]

N s: number of parts per sheet [-] N p1: number of punches [-] N p2: number of punches to complete a contour [-] agem: average distance between holes [mm] v pos: positioning speed [mm/min] t kap kap: time to for one punch [min] LCT : length of the rectangular contours [mm] LT : efficient length of the rectangular contours [mm] vlaser : laser cutting speed [mm/min] t nib nib: time of a punch for curved contours [min] N nib nib: number of punches for curved contours [-] t tw tw: tool change time [min] t iu iu: time for fixing and unfixing a sheet [min] t af af : time for removing scrap per sheet [min]

Cost functions for nibbling (Nollet, 1993). K tot = (K vast + K var )T tot + K tech ’

’

 PV   + Ropp O   60 hy  PV a  ’ = eu ⋅ pe + k w’ + k m’ + k p’ K var K tech = N p K 1 p + N t K 1t + N nib K 1nib N p = N p1 + N p 2 ’ = K vast

104

1 1

(A.18) (A.19) (A.20) (A.21) (A.22)

Time and cost functions from literature with: K tot tot : total costs per workpiece [costs] K’vast : fixed cost rate [costs/min] K’var : variable cost rate [costs/min] T tot tot : total processing time per workpiece [min] K tech tech: technology dependent costs [costs] hy: working hours per year [hour] PV : present value [costs] PV a: annuity factor [-] 2 Ropp: costs of surface area [costs/m ] 2 O: area [m ] eu: electricity usage [kWh] pe: price of electricity [costs/kWh]

k’w: labour rate [costs/min] k’m: maintenance rate [costs/min] k’ p: programming rate [costs/min] N p: total number of punches [-] K 1p 1p: tool costs of one punch [costs] N t t : number of punches for rectangular contours [-] K 1t 1t : tool costs of one punch in rectangular contour [costs] N nib nib: number of punches for curved controus [-] K 1nib 1nib: tool costs of one punch in curved contour [costs] N p1: holes in the workpiece [-] N p2: punches from the contours [-]

A.3 Cutting Time functions for cutting (Cuesta, 1998). T batch = T pre + T c + T aux

+ T pro

(A.23)

with: T batch batch: production time for a batch [ min] T pre: preperation time [min] T c: cutting time [min]

T pre

T aux aux: auxiliary time [min] T pro: programming time [min]

m

p

q

r

j =1

k =1

l =1

n =1

= ∑ (T NC 1 + T NC 2 + T NC 3 ) j + ∑ (T SH 1 + T SH 2 + T SH 3 ) k + ∑ (T Tool ) l + ∑ (T Sp ) n

with: m: number of NC programmes used per machine [-] p: number of sheets present in the batch [-] q: number of tools loaded per machine [-] r : number of specimens to make [-] T NC1: loading and selection time for NC program j [min] T NC2: time required for tool position (returns and set tool to the program origin) [min] T NC3: program start-up time [min]

T c =

n

∑ (T + T ci

nci

)

i =1

(A.24)

T SH1 SH1: time for loading sheet/part k [min] T SH2 SH2: time required for tool (e.g. positioning) and sheet/part preparation for sheet k (previous heat treatment, etc.) [min] T SH3 SH3: time unloading or changing sheet/part k [min] T Tool Tool: time loading tool l [min] T Sp Sp: time to produce one specimen (e.g. bending) [min] n

Pc

= ∑ N i (Pci + Pnci )

(A.25)

i =1

with: n: the numberof time-files (number of NC programs) [-] T c: effective cutting time (for all the NC p rograms) [min] Pc: effective cutting perimeter (for all the NC progams) [m] T ci ci: cutting time for the ith program [min] T nci nci: rapid (noncutting) time for the ith program[min] Pci: cutting perimeter for the ith program [m] Pnci: rapid (noncutting) perimeters for the ith program [m] N i: number of simultaneous tools for the ith pr ogram ( N N i = 1,2,4,6,...,12) [-]

Cost functions for cutting (Cuesta,1998). C batch = C mt + C mat + C t + C pro with: C batch batch: total costs per batch [costs] C mt mt : machine-tool costs [costs] C mat mat : material costs [costs]

C mt = x

T batch

60 x = C ms + C ml C ms

= N ft Ps

(A.26) C t t: tooling costs [costs] C pro: programming costs [costs]

(A.27)

+ C ma + C mp + C mt + C mi

(A.28) (A.29)

105

Appendix A

S t Pre

C ml

=

C ma

=

C mp

= Pkw E mt

Pmt

(A.31)

T a H mt r

(A.32)

100

K Pmt

C mt = C mi

(A.30)

H mt

(A.33)

100 H mt

= C ind

(A.34)

H mt

with: C ms: labour expenditure or operator income cost rate [costs/hour] C ml: machine-location cost rate, lighting, etc. [costs/hour] C ma: machine-tool amortisation cost rate [costs/hour] C mp: electrical power consumption [costs/hour] C mt : maintenance and reparations cost rate [costs/hour] C mi: machine-tool indirect costs rate [costs/hour] N ft : number of full-time operators per machine-tool (usually N ft = 1) [-] Ps: labour cost rate [costs/hour] Pmt : total machine-tool price [costs] T a: amortisation time or period [years]

C mat =

K   [10 ∑ n   1+ 100  m

a

−3

j

j =1

H mt : working machine-tool hours per year [hour/year] S t : total working area, including operator and 2 machine-tool areas [m ] 2 Pre: price of shop floor rent, lighting, etc. [costs/m year] K : maintenance and reparation supplement (usually K ≈ 3% of the total machine-tool price) Pkw: kWh price [costs] E mt : machine-tool power [kW] r : machine utilisation rate (%) C ind : indirect and annual cost which could be charged to the machine-tool [costs/year]

 

w j ρ j Sb j ]Psh − [10 −3 w j ρ j Sr j ]P j 

(A.35)

with: m: number of equal sheets and equal cutting way (the same areas and the same cutting perimeters) [-] n j: number of equal sheets (like the jth sheet) and equal cutting way (the same areas and the same cutting perimeters) [-] K a: generic supplement for loading, internal transport, etc [%] w j: jth sheet thickness [mm] 3 j: jth sheet density [kg/m ] 2 Sb j: original jth sheet area [m ] Psh: original sheet price [kost/kg] 2 Sr j: remaining jth sheet area [m ] P j: remaining material price in the jth sheet: P j = Psh when the remaining material (Sr j) is for offcut re-use and P j = Psc when (Sr j) is sold like scrap [costs/kg] S sc: material metal scrap sale price [costs/kg]

= C gas1 + C gas 2 + C tor + C tk + C bot C gas1 = 10−3 C g1 Pg1 Pc C gas 2 = 10 −3 C g 2 Pg 2 Pc C t

Ptor T c

C tor

=

C tk

= Ptk

106

(A.37) (A.38) (A.39)

H s 60 T batch

(A.40)

43200

C bot = ( N b Pbot + Pbt C b )

(A.36)

T batch

43200

(A.41)

Time and cost functions from literature with: C gas1 / Cg as2: gas 1 and gas 2 consumption [litre/m] 3 Pg1 /Pg2: gas 1 and gas 2 price [costs/m ] Pc : effective cutting perimeter [m] T c : effective processing time [min] T batch: production time [min/lot] N b: number of rented bottles [-] Ptk : price for tank-rent [costs/month] Pbot : price for bottle-rent [costs/month]

= P pro

C pro

C b: bottles consumption [bottles/month] Pbt : bottles transport price [costs/bottle] Ptor : tool price (torch, gauges, and all devices) [costs] H s: average tool life estimate [hour] C bot : total bottle cost [costs] C tk : total tank cost [costs] C tor : total tool cost [costs]

T pro

(A.42)

60

with: P pro: programmer cost rate [costs/hour]

T pro: programmer cost rate [costs/hour]

A.4 Water jet cutting Cost functions for water jet cutting (Zheng, 1996). ’ K tot = (K eq

’ + K co’ K ma )T tot

(A.43)

with: K tot : total cutting costs [costs] K’eq: equipment rate [costs/hour] K’co: consumables rate (electricity, water, abrasive) [costs/hour] K’ma: maintenance rate [costs/hour] T tot : total cutting time [hour]

A.5 Laser cutting Time functions for laser cutting (Nollet, 1993). T tot = T s + T w t vs

T s =

+ t ls

N tot

T w = 1.05(t l1 + t l 2 + t l 3 + t l 4 + t l 5 ) t l1 =

t epl N s

 a gem  1.15 Lg + t inzet  +  v pos  vlaser    a gem  1.05 Lc + t l 3 =  t inzet  +  v pos  vlaser   t l 2 = N g 

t l 4 = t l 5 =

t iu N s t af

N s

(A.44) (A.45) (A.46) (A.47) (A.48)

(A.49) (A.50) (A.51)

107

Appendix A with: T tot : total processing time per workpiece [min] T s: set-up time for N tot [min] T w: processing time per workpiece [min] t l1: time for startposition [min] t l2: effective cutting time for holes [min] t 13: effective cutting time for contours [min] t 14: fixing and unfixing time [min] t 15: time for removing scrap [min] t vs: fixed set-up time [min] t ls: time for laser parameter setup [min] N tot : total number of workpieces [-]

t epl: time to move from zero to first cut [min] N s: number of parts per sheet [-] N g: number of holes [-] agem: average distance between holes [mm] v pos: positioning speed [mm/min] t inzet : time to drill start hole [min] Lg: length of the contours of all holes [mm] vlaser : laser cutting speed [mm/min] Lc: length of the contour [mm] t iu: time for fixing and unfixing a sheet [min] t af : time for removing scrap per sheet [min]

Cost functions for laser cutting (Nollet,1993). K tot = (K vast + K var )T tot + K tech ’

’

(A.52)

 PV   + Ropp O   60 hy  PV a  ’ ’ ’ ’ K var = eu ⋅ pe + k w + k m + k p K tech = K sngas + K lagas  1.15 Lg +1.05 Lc  ’ + T inzet N g  K sngas =   K sngas v laser   ’ K lagas = (T slaser + T wlaser )K lagas ’ = K vast

1 1

with: K tot : total costs per workpiece [costs] K’vast : fixed cost rate [costs/min] K’var : variable cost rate [costs/min] T tot : total processing time per workpiece [min] K tech: technology dependent costs [costs] K’sngas: cutting gas costs [costs] K’lagas: laser gas costs [costs] hy: working hours per year [hour] PV : present value [costs] PV a: annuity factor [-] 2 Ropp: costs of surface area [costs/m ] 2 O: area [m ] eu: electricity usage [kWh]

(A.53) (A.54) (A.55) (A.56) (A.57) pe: price of electricity [costs/kWh] k’w: labour rate [costs/min] k’m: maintenance rate [costs/min] k’ p: programming rate [costs/min] Lg: length of the contours of all holes [mm] Lc: length of the contour [mm] vlaser : laser cutting speed mm/min] t inzet : time to drill start hole [min] N g: number of holes [-] K’sngas: cutting gas costs [costs] T slaser : set-up time for N tot [min] T w: processing time per workpiece [min] K’lagas: laser gas costs [costs]

Cost functions for laser cutting (Zheng, 1996). ’ K tot = (K eq

’ + K co’ K ma )T tot

(A.58)

with: K tot : total cutting costs [costs] K’eq: equipment rate [costs/hour] K’co: consumables rate (electricity, laser gases, assistant gases) [ costs/hour] K’ma: maintenance rate [costs/hour] T tot : total cutting time [hour]

A.6 Laser welding Time function for laser welding (Maree,1997). T = rp (1.66 + 0.212 NT + 0.042WL + 0.49(t )1.03 WL)60 with: T : time to complete process [sec] rp: repeats [-] NT : number of tacks [-]

108

WL: weld length [inches] t : thickness of material [inches]

(A.59)

Time and cost functions from literature

A.7 Bending Time function for bending (Somatech, 1998). T Ptot =

N p

∑ (n (n s

ab

T ab

+ nshpT shp + ntxT tx + ntyT ty + ntzT tz )) i

i =1 with: T ptot : production time for a batch [sec] N p: number of products in batch [-] ns: skill factor for worker ( ≥1) [-] nab: number of bends [-] nshp: number of shove movements [-] ntx: number of x turnings [-] nty: number of y turnings [-]

(A.60)

ntz: number of z-turnings [-] T ab: time to produce a single bend [sec] T shp: time to shove the part [sec] T tx: time to turn the part in the x-plane [sec] T ty: time to turn the part in the y-plane [sec] T tz: time to turn the part in the z-plane [sec]

Time functions for bending (Maree, 1997). T = s + ((m * p ) * b) * rp

(A.61)

with: T : time to complete the process [sec] s: set-up time = 45 for 90 ° bends, 105 for other angles [sec] m: measuring correct dimensions and marking it on the part = 20 [sec] p: process time = 8 for 90 ° bends, 16 for other angles [sec] b: number of bends per part [-] rp: repeats [-]

T = s + ( m +

A * ct * Ang

90

) * rp

(A.62)

with: T : time to complete the process [sec] s: set-up time = 20 [sec] m: measuring correct dimensions and marking it on the part = 15 [sec] 2 A: cross-section area of the part [mm ] ct : time factor for the process = 0.5 [-] Ang: angel to be bent [degrees] rp: repeats [-]

109

Appendix A

110

B Example cost structures B.1 Resource information structure

Figure B.1 A hierarchy of cost actuators.

Figure B.2 A cost structure for a machine.

111

Appendix B

Figure B.3 A cost structure for a specific machine.

B.2 Product information structure

Figure B.4 A cost structure for variant based cost estimation.

Figure B.5 A cost structure for material costs.

112

Example cost structures

B.3 Order Information Structure

Figure B.6 A cost structure design overhead costs for an order.

Figure B.7 A cost structure for the costs resulting from planning an order. 113

Appendix B

114

C Regression analysis This appendix gives some theory about regression analysis relevant for the determination of cost functions. For a more extensive and detailed overview, see for example (Sen, 1990). Single regression for n data sets can be described with the following n functions, i.e. the regression model: y1 = β 0 + β 1 x11 + ε 1 M

(C.1)

y n = β 0 + β 1 x n1 + ε n with: yi: observed dependent variables i: constants i: resduals x1i: observed independent variables i: 1, 2, ..., n with n the number of datapoints

The residuals ε i indicate the difference between the value of the observed dependent variables y i and the predicted dependent variables yˆ i . The smaller the residuals, the better the fit.

ε i = yi − yˆi

(C.2)

with: yˆ : predicted dependent variables i

In case of multiple regression, the regression model becomes: y i = β 0 + β 1 x11 + β 2 x12 + ... + β k x1k + ε 1 M

M

M

M

M

M

M

(C.3)

y n = β 0 + β 1 x n1 + β 2 x n 2 + ... + β k x nk + ε n with: k : number of independent variables

The regression model can be rewritten by considering equations C.4 and substituting them in equation C.3.

 y1   y   2 y =   M  y n 

 β 1   β   2 β =   M  β n 

ε 1  ε   2 ε =   M ε n 

1 1 X =  M 1 

x11

L x1k 

x 21

L x 2 k

M

M

x n1

L

  M   x nk 

(C.4)

As a result, the regression model can be written as: y = X β + ε

(C.5)

To obtain an equation that fits the data sets the best, the constants β j have to be determined so that the residuals ε i are minimised. A technique that is usually used is the least squares method. The least squares estimates are the β j which minimise:

115

Appendix C

S =

n

n

∑ ε = ∑ ( y − β x 2 i

i

i =1

1i

0

− ... − β k x1k )2

(C.6)

i =1

For reasons of simplicity and overview, equation C.6 is simplified to one independent variable (i.e. k=1). The partial derivatives of equation C.6 to the β j , in this case β 0 and β 1 , become: n ∂S = − 2∑ ( y i − β 0 − β 1 xi1 ) ∂ β 0 i =1 n ∂S = − 2∑ ( y i − β 0 − β 1 xi1 ) xi1 ∂ β 1 i =1

(C.7) (C.8)

The least squares estimates b 0 and b1 of β 0 and β 1 can be obtained by setting equation C.7 and C.8 to zero. After rearranging, this results in: b0 = y − b1 x1

(C.9)

n

∑ y x − n x y b = ∑ x − n x i

i1

i

i =1 n

1

(C.10)

2 1

i1

i =1

with:

y = n

−1

n

∑ y

x i

i

=n

−1

i =1

n

∑ x

i1

i =1

With the values of b 0 and b1, the estimated values of the dependent variable can be calculated with: yˆ i

= b0 + b1 xi1 = y + b1 ( xi1 − x1 )

(C.11)

The residuals are an indication of the fit of the least squares calculation, the smaller the residuals the better the fit. A formal way to measure the fit is: n

∑ ( y − yˆ ) =1 − ∑ ( y − y ) i

R 2

i =1 n

2

i

(C.12)

2

i

i =1

2

2

R ranges from 0 to 1, with R closer to one indicating a better fit. The least squares method gives good result when the Gauss-Markov conditions are met; the Gauss-Markov conditions are: var(ε i ) = E (ε i − E (ε i )) E (ε i ) = 0 for all i E ε i ε j

= 0 when i ≠ j

2

= E (ε i2 )= σ 2

(C.13) (C.14) (C.15)

The meaning of the first two Gauss-Markov conditions can be clarified graphically. The continuous line in Figure C.1 represents the least squares fit through the points, while a more desirable line is indicated by the dotted line. The continuous line is a relatively bad fit because the variance of the residuals increases with increasing x. When condition C.13 does not hold, this situation, called heteroscedasticity, occurs. The second condition is violated in the case depicted in Figure C.2. A single point or a small number of points can influence the least squares method extremely as illustrated in Figure C.3 and Figure C.4. An outliner results in a violation of C.14. When data points are related C.15 is violated, this occurs e.g. when data points are the same. 116

Regression analysis

Figure C.1 Violation of condition C.13.

Figure C.2 Violation of condition C.14.

Figure C.3 An influential point and outliner.

Figure C.4 An influential point, not an outliner.

It is possible to check whether the independent variables are significant parameters by means of the so-called t-value. If the t-value of an independent variable is low, the variable is not worth mentioning as cost parameter. The t-value is low when |t| < 2 ( Horngren, 1994).

117

Appendix C

118

D Neural networks This appendix gives some theory about neural networks relevant for the determination of cost functions. For a more extensive and detailed overview, see for example (Veelenturf, 1997). As stated in section 2.5.3, a neural network consists of neurons and synopsis. The working principle of a neural network will be explained with the binary neuron depicted in Figure D.1.

Figure D.1 A binary neuron.

The relation between the output signal Y and the input signals X i is given by: y = 1

if

w1 x1 + w2 x 2 + L + wm x m > D

=0

if

w1 x1 + w2 x 2 + L + wm x m ≤ D

y

(D.1)

with: D: a threshold value xi: input signals y: output signal wi: weights

The weights wi can be determined by training the neural network with a number of historic cases. The weights are determined so that the neural network best fits the historic cases, see equation D.2.

 i =m  y = S ( w1 x1 + w2 x 2 + L + wm x m ) = S  ∑ wi x i   i =0 

(D.2)

with: y: output historic case xi: input historic case wi: weights S(): linear threshold function

An algorithm that can be used to find the proper wi is: w’( k + 1) = w’(k ) +

∑ ε x + ∑ ε x i

i

with: x i

∈T +

j

j

(D.3)

x j ∈T −

In the case of a multi layer perceptron (Figure D.2), the minimisation of the Mean Squared Error can be used to train the network.

119

Appendix D

Figure D.2 A multi layer percepton. E =

1

∑ {t ( x ) − y( x )} ’ i

N xi ∈ L

w ’new = w ’old +

2

’ i

1

∑ ε {t ( x N

(D.4) ) − y ( x ’i )}

’ i

dy ds

x ’i

(D.5)

with: E = Mean Squared Error N = number of used cases L = learning set xi = extended input vector wi = extended weight vector t(xi) = target values y(xi) = output values ( y ( s )

=

1 1+ e

s

with s =

∑ w x ) i

i

ε = learning factor ( 0 < ε <1 ) The weight factors for neuron 1 and 2 become: dy ε {δ ( x )} x ∑ N ds dy 1 + ∑ ε {δ ( x )} x N ds

w ’new,1 = w ’old ,1 + w ’new, 2 = w ’old , 2 with:

1

1

’ i

2

’ i

1

’ i

2

(D.6) ’ i

δ 1 ( xi ) = t 1 ( xi ) − y1 ( xi ) = (t 3 ( xi ) − y 3 ( xi ))

(D.7) dy 3 ds 3

δ 2 ( xi ) = t 2 ( xi ) − y 2 ( xi ) = (t 3 ( xi ) − y 3 ( xi ))

dy 3 ds 3

w31 w32

The learning rule can be generalised for k neurons in j layers with weights w jk : w ’new = w ’old +

1

∑ ε {δ N

jk

( x ’i )}

xi ∈ L

with:

δ jk ( xi ) =

∑ m

120

δ j +1,m ( xi )

dy j +1, m ds j +1, m

dy jk ds jk

w jkm

z ’ jk ( xi )

(D.8)

E

Example product and context information

E.1 Components The components of the bottom-frame (figure 7.1) used in the example in chapter seven are depicted here in more detail, including their flat pattern.

Figure E.1 Component 1 of the bottom-frame.

121

Appendix E

Figure E.2 Flat pattern of component 1 of the bottom-frame.


122





123

Appendix E




124





125

Appendix E



126






127

Appendix E

E.2 Information structures The information about resources and products used for the example in chapter seven are listed here. The information is limited to the information relevant for cost estimation. Resource Information Structure - Physical Resource Domain Element Attribute Name Type Name Description Value Amada Press brake ORPR Overhead rate 2.86 FBD3512E press brake PRMMR Press brake 1.40 man-machine rate PRPT Press brake production time Amada Punch press ORPU Overhead rate 2.86 Vipros-357 punch press PUMMR Punch press 1.40 man-machine rate PUPT Punch press production time Trumpf 2507 Laser LMMR Laser man- 2.80 machine rate LPT Laser production time ORLM Overhead rate 2.86 laser machine My Univeral Universal tool LST Life span tool 150000 tool TTC Total tool 906200.00 costs TTU Total number 199 of tool usage per product Steel Material MA Material area MR Material rate 0.00002797 Zincor Material MA Material area MR Material rate 0.00002977

Unit Fl./min Fl./min

min

Fl./min Fl./min

min

Fl./min min

Fl./min Fl./component -

2

mm 2 Fl./mm 2 mm 2 Fl./mm

Table E.1 Relevant information in the Physical Resource Domain of the Resource Information Structure.

128


Resource information structure - Method Domain Element Attribute Name Type Name Description Value MyAssembly Assembly ASSMMR Assembly 0.70 man-machine rate ASSPT Assembly production time ASSR Assembly 0.89 rate. MyBending Bending MyGeneral General GR General rate 9.01 NOO Number of orders MyInspection Inspection IR Inspection 1.83 rate NOC Number of components MyLasercutting Lasercutting MyLogistic Logistic LR Logistic rate 3.04 NOP Number of parts MyPainting Painting PMMR Painting man- 0.70 machine rate PPT Painting production time MyPunching Punching -

Unit Fl./min

min

Fl./min Fl./order Fl./component Fl./part Fl./min min

-

Table E.2 Relevant information in the Method Domain of the Resource Information Structure.

129

Appendix E

Product Information Structure - Physical Product Definition Domain Element Attribute Name Type Name Description Value Bend Manufacturing feature Component DC Direct costs DOR Direct 1.5 overhead rate Contour Manufacturing feature Depression Manufacturing feature Hole irregular Manufacturing feature Hole Manufacturing rectangular feature round Hole Manufacturing rectangular feature sharp Hole round Manufacturing feature

Unit Fl. -

-

-

Table E.3 Relevant information in the Physical Product Definition Domain of the Product Information Structure.

E.3 Cost structures The cost structures used in the example in chapter seven are listed here. Cost actuator: element of type Assembly Cost structure: Activity Based Costing / generative Cost parameter: assembly man-machine rate [Fl./min] Cost parameter: assembly production time [min] Cost parameter: assembly production time [min] Cost parameter: assembly rate [Fl./min] Cost type: direct assembly costs of type direct costs Cost function Cost type: overhead assembly costs of t ype overhead costs Cost function Cost structure: Direct Costing / generative Cost parameter: assembly man-machine rate [Fl./min] Cost parameter: assembly production time [min] Cost type: direct assembly costs of type direct costs Cost function

130

Example product and context information Cost actuator: element of type Painting Cost structure: Activity Based Costing / generative Cost type: direct painting costs of type direct costs Cost function Cost parameter: painting man-machine rate [Fl./min] Cost parameter: painting production time [min] Cost structure: Direct Costing / generative Cost type: direct painting costs of type direct costs Cost function Cost parameter: painting man-machine rate [Fl./min] Cost parameter: painting production time [min] Cost actuator: element of type Component Cost structure: Direct Costing / generative Cost parameter: direct costs [Fl.] Cost parameter: direct costs rate [-] Cost type: overhead costs Cost function Cost actuator: element of type General Cost structure: Activity Based Costing / generative Cost parameter: general rate [Fl./order] Cost parameter: number of orders [-] Cost type: overhead general costs of type overhead costs Cost function Cost actuator: element of type Inspection Cost structure: Activity Based Costing / generative Cost parameter: inspection rate [Fl./co mponent] Cost parameter: number of components [-] Cost type: overhead inspection costs of type overhead costs Cost function Cost actuator: element of type Logistic Cost structure: Activity Based Costing / generative Cost parameter: logistic rate [Fl./subassembly] Cost parameter: number of subassemblies [-] Cost type: overhead logistic costs of type overhead costs Cost function Cost actuator: element of type Laser Cost structure: Activity Based Costing / generative Cost type: direct laser costs of type direct costs Cost function Cost parameter: laser man-machine rate [Fl./min] Cost parameter: laser production time [min] Cost parameter: laser production time [min] Cost parameter: overhead rate laser [Fl./min] Cost type: overhead laser costs of type overhead costs Cost function Cost structure: Direct costing / generative Cost type: direct laser costs of type direct costs Cost function Cost parameter: laser man-machine rate [Fl./min] Cost parameter: laser production time [min]

131

COST Control

Recommend Documents