De clin in g Peak Demand

(1)

Peak Demand Management at GKN Sintermetals Inc.

CURE

IS BETTER THAN

PREVENTION

A Research in Peak Demand Management at GKN Sintermetals Inc., Cape Town, S.A.

Tom Witsenboer Groningen, 4^th July 2005

De clin in g Peak Demand

(2)

CURE

IS BETTER THAN

PREVENTION

A Research in Peak Demand Management at GKN Sintermetals Inc. CapeTown, S.A.

Thomas P. Witsenboer 4^th July 2005

Beco ISP

Principal B.Kothuis

Supervisor L.Coetzee

GKN Sintermetals

Supervisor E.Aggenbach

State University of Groningen

First supervisor dr. W.M.C. van Wezel Second supervisor dr.ir. D.J. van der Zee

University of Groningen

Faculty Management and Organization Landleven 5

NL-9700 AV Groningen

De clin in g Peak Demand

(3)

PREFACE

Energy provides our planet Earth with the capacity to work. Energy itself is inexhaustible, but most of the time useless for our energy demands. That is why we use energy sources which are easy obtainable, conveyable and useable. These energy sources were essential in the past but becoming even more essential in future. Unfortunately, energy sources, are just nót inexhaustible! Therefore it is necessary to use this sources in an efficient way. One of the manners to accomplish efficient usage of the energy sources is to minimize the usage of energy, by resisting waste of energy.

The Government of South-Africa makes it one’s aim to cut back this waste of energy in both households as well as in business. In the Western Cape, in cooperation with the consultants organization BECO-ISB, so called ‘waste-minimization-clubs’ (WMCs) are founded. These clubs, consisting of similar businesses (size, location, branch etc.), are doing research, both individual as together, in the possibilities of waste prevention. During periodically hold club meetings, results and experiences are shared with the other members.

By order of BECO-ISB and four members of the WMC Sacks-Circle Bellville-South I had to set up a research into the opportunities to find possible savings in energy use and energy costs at the members’

factories. My special interest is in the possibility to save costs by equalizing electricity demand during production time. Because of an policy against peaks in electricity requiries at the electricity utilities, companies are charged for their peak demand. This charge fills quickly half of the total electricity costs. The research itself took place at GKN Sintermetals in Cape Town under supervision of L.Coetzee (BECO ISP) and E.Aggenbach (GKN Sintermetals)

This final thesis is accomplished as being the final part of the study Technology Management at the Faculty Management and Organization of the University of Groningen. My supervisors from the University are W.M.C. van Wezel and D.J. van der Zee

Besides the four supervisors, I would especially thank Joost, Joerek, Judith, Saskia, Dagmar and Teun for having a great time living with them at Ocean View Drive in Cape Town, my parents and grand mom for financial support and Myrte for many other reasons.

(4)

ABSTRACT

Motivation

The last decade, utilities which offer electricity to their customers, pay a lot of attention to their peak demands and are constantly trying to get this peak demands under control. One of the possibilities to get peak demand under control is to transfer the accountability to the end-users, both business market and consumer market. One approach is to make use of time-of-use (TOU)-tariffs. Another approach is to discourage a dissimilar usage over time. This policy lead to an electricity account with an included charge over the largest demand of the past billing period. This largest demand is called peak demand, and is measured during intervals of fifteen minutes. A very common electricity account quickly consists for almost a half of this charge. This implies that an interval of fifteen minutes determines a significant part of the electricity costs.

This final thesis is based upon attempting to create a more energy-efficient production process, by trying to reach a more, even distribution of electricity use. Energy efficient in that way, that you save in the costs of energy, while using the same amount of energy. The aim thus, is to shift with demand, called peak shaving. There are different ways to carry out this peak shaving. In this thesis different peak demand management (also called as demand-side management and peak load management) methods are introduced into the production process. By means of a simulation study the desired results can be found. Simulation modeling is a quantitative tool of research, which tries to find solution by imitating the real situation. The tool is useful when the situation is too complex and uncertain to be figured out with real life simulation or mathematic modeling.

Research problem

The motivation result in a research problem:

Objective:

By order of BECO-Institute for Sustainable Business to set up a study, for their four clients and member of the WMC Sacks Circle Bellville-South: GKN Sintermetals, Falke Eurosocks, Nettex and Usabco, into the opportunities of energy efficient production by making use of peak load management decisions into their production planning and production process control.

Problem definition:

Which load management method or combination of methods show(s) the best results with regard to savings in the monthly peak demand charges and reach(es) a more energy-efficient production process?

Analysis

After sifting the production process at GKN Sintermetals several important characteristics came clear.

Firstly the machines could be divided into four different groups. This division depend on the production hours and the occupation rates of the machines. Only group three and group four, with a corresponding occupation rate below hundred percent can be used in the methods. Secondly the usages

(5)

of the machines are an important factor. Four machines are significantly using more electricity than the other machines. This four machines (method-machines) are used in the management methods.

Possibilities for improvements

Possible improvements can be made after introducing two management methods into the existing production process. First method is a planning method (method I). Purpose of this method is to avoid the existence of a peak demand. The method consists of a restriction that only a certain number of the method-machines are allowed to run simultaneously. This implies that you put in the production plan, before production, which machines are running and which not. Second method is a control method (method II). Purpose of this method is to handle accurately when a certain adjusted maximum level of peak demand is exceeded. If this happens, one or more of the method-machines are switched off to bring about a decrease in peak demand below this maximum level. This is a control method, and not a planning method, because you control a not planned situation: a too large peak demand. The two methods are introduced individually, method I is introduced individually (scenario1) and method II is introduced individually (scenario2), but also together (scenario3). Expectation is that the outcomes of each scenario have a better score than the previous scenario:

∆S((MI+MII)=ON) > ∆^S(MII=ON∩ MI=OFF) > ∆^S(MI=ON∩ MII=OFF) > ∆S((MI+MII)=OFF)

Results of the simulation

The results, generated with the simulation model, of the three scenarios are compared with the present situation (scenario0). The outcomes indicated that method I can be rejected and method II, on the other hand, can be accepted. Therefore the hypothesis can be rewritten into:

∆S((MI+MII)=ON) = ∆^S(MII=ON∩ MI=OFF) > ∆^S(MI=ON∩ MII=OFF) = ∆S((MI+MII)=OFF)

Conclusions

After the rejection of method I and the acceptation of method II, one of the conclusion is that to cure eventual occurring peaks has got a better outcome than to prevent peaks by changing the production plan. This conclusion strengthen the decision to purchase a peak alarm installation. The simulation study pointed out that a peak of 450 kVA was the best scoring maximum peak level. The savings in peak demand were larger than the costs of the interventions, and the length of the interventions were less than the available extra production time at the end of the month.

But because of the complex and uncertainty of the study, the experience gathered in the future, when the peak alarm installation is used, will be the most important reason for success.

This study grounds the possibilities of peak demand savings with regard to reaching a more energy- efficient production process. Efficient in that way that savings in electricity cost are obtained. And apparently the usage at the factory remains equal, the peak demand at the utility will be flatten. And that is what the whole charge of peak demand is about.

(6)

PREFACE 3

ABSTRACT 4

1. INTRODUCTION 9

§ 1.1 BECO-INSTITUTE FOR SUSTAINABLE BUSINESS 9

§ 1.2 WASTE MINIMIZATION CLUB (WMC) SACKS CIRCLE 9

§ 1.2.1 GKN Sintermetals Inc. 9

§ 1.2.2 Falke Eurosocks Inc. 9

§ 1.2.3 Nettex Inc. 9

§ 1.2.4 Usabco Inc. 9

§ 1.4 REASON TO DO THE RESEARCH 10

§ 1.5 RESEARCH PLAN 11

2. JUSTIFICATION OF THE RESEARCH 13

§ 2.1 INTRODUCTION 13

§ 2.2 ENERGY MANAGEMENT IN SOUTH AFRICA 13

§ 2.3 LOAD MANAGEMENT 14

§ 2.4 THE ROLE OF ELECTRIC LOAD ANALYSIS 15

§ 2.5 ELECTRIC LOAD ANALYSIS AT THE COMPANIES 15

§ 2.6 SIMULATION 16

§ 2.7 CONSEQUENCES FOR THE RESEARCH 17

3. RESEARCH PROBLEM 20

§ 3.2 PROBLEM ANALYSIS 20

§ 3.2.1 The Problem Definition 20

§ 3.2.2 The Research Restrictions 20

§ 3.2.3 Type of Research 21

(7)

4. ANALYSIS 23

§ 4.2 THE PRODUCTION 23

§ 4.3 THE EXPERIMENTAL FACTORS 26

§ 4.3.1 First Experimental Factor: The Maximum Adjusted 26 Monthly Peak Demand

§ 4.3.2 Second Experimental Factor: The Planning and Control Methods 26

§ 4.3.2.1 The opportunities at GKN Sintermetals Inc. 26

§ 4.3.2.2 The possibilities to improve the load factor 27

§ 4.3.2.3 The planning method and the control method 29

§ 4.3.3 Third Experimental Factor: The Maximum Allowed Interventions 36

§ 4.3.4 Fourth Experimental Factor: The Production Quantities 36

§ 4.4 THE PRODUCTROUTINGS PASSING THE METHOD MACHINES 38

§ 4.5 THE RESTRICTIONS AND HYPOTHESES 39

§ 4.5.1 The Restrictions 39

§ 4.5.2 The Hypotheses 39

§ 4.6 USEFULNESS OF SIMULATION 42

§ 4.7 SUMMARY 42

5. SIMULATION MODEL 44

§ 5.2 SCOPE AND LEVEL OF THE MODEL 44

§ 5.3 SUMMARY OF THE MODEL 44

§ 5.3.1 The Stochastic Input 45

§ 5.3.2 The Decision Variables 45

§ 5.3.3 The Parameters 45

§ 5.3.4 Stochastic Output 45

§ 5.3.5 The Presumptions of the Black Box Model 45

§ 5.4 NUMERICAL SPECIFICATION 46

§ 5.4.1 Parameters 46

§ 5.4.2 Validation Values 47

§ 5.4.3 Verification vs. Validation 47

§ 5.5 SCENARIOS 48

§ 5.5.1 Scenario 0 48

§ 5.5.2 Scenario I 48

§ 5.5.3 Scenario II 48

(8)

§ 5.6 THE EXPERIMENTS 49

§ 5.6.1 The Simulation Models 49

§ 5.6.2 The Experiments 49

§ 5.7 SUMMARY 51

6. RESULTS 53

§ 6.2 RESULTS OF SCENARIO1 53

§ 6.5 THE RESULTS AND THE HYPOTHESIS 61

§ 6.6 SUMMARY 64

7. CONCLUSIONS AND RECOMMENDATIONS 66

§ 7.2 CONCLUSIONS FROM THE RESEARCH PROBLEM 66

§ 7.2.1 Conclusion of the Analysis 66

§ 7.2.2 Conclusions of the Results from the Simulation Model 67

§ 7.3 RECOMMENDATIONS 68

§ 7.3.1 Recommendations with regard to the Production Process 68

§ 7.3.2 Recommendations with regard to Potential Further Research 68

BIBLIOGRAPHY 69

APPENDICES 71

A.I THE DIAGNOSIS 71

A.II DIFFERENCE BETWEEN KVA AND KWH: THE POWERFACTOR 72

A.III DATA NEEDED TO THE SIMULATION MODEL 73

III.1 The Priority Rules which Machine to Stop 73 III.2 Products, Setup Time, Production rates and Master Plan 73 III.3 TypeNRs, Usage, Operation time and Occupation Rates 74 III.4 Production Quantities and Available Intervention Time 75

A.IV COMPLETE RESULTS OF SCENARIO2 78

A.V COMPLETE RESULTS OF SCENARIO3 82

(9)

1. INTRODUCTION

1.1 BECO-Institute for Sustainable Business

Based in The Netherlands and started in 1990 as BECO Milieumanagement en advies BV the company was first focused on just the prevention of waste and emissions, energy savings, water savings and the pushback of transport movements. This beforehand unknown approach of the environmental load seems also to have economical benefits looking to the economical efficiency.

After five years of focusing on the ‘planet’ and ‘profit’ also the aspect ‘people’ is emerging. BECO is involved in projects all with regard to durability where the aspect people is on the same level as the other two. During the following years thinking in production processes changed into product-related environmental control. Advice in prevention led to ICT-solutions and co-operations between different companies.

Since respectively 1998 and 1999 BECO is also based in Belgium and South Africa. The company in South Africa is an alliance between BECO and the Institute for Sustainable Business, which gave the alliance the name: BECO-ISB. BECO-ISB is a research and consultancy institute specialised in strategic environmental management issues. Its philosophy is based on the principles of “prevention is better than cure”, and its experiences are based on taking a system view of sustainability. BECO-ISB has got offices in Johannesburg and Cape Town and employs 7 consultants (BECO in total more than 50). BECO-ISB is experienced in cleaner production and products:

• Prevention of wastes & emissions

• Environmental management

• Monitoring environmental performance

• Préventive environmental Chain management

• Training and workshops for authorities and companies

• Improvement of energy, raw material and water efficiency

• Management of, assistance for Waste Minimization Clubs

The last two notes are the most important services BECO-ISB offers. BECO is managing and assisting a lot of Waste Minimization Clubs in South Africa like the WMC for Large companies (like British American Tobacco’s, Caltex, BMW, Nissan etc.), the WMC for the food sector (like Appletizer, Clover, Nestle, Parmalat etc.) and the WMC for companies located in the industrial area, called Sacks Circle, in Bellville, near Cape Town.

1.2 Waste Minimization Club (WMC) Sacks Circle

The Waste Minimization Clubs are founded and subsidized by the South African Government in cooperation with BECO-ISP to endeavor companies in South Africa to strive for industry with an eye for

(10)

The WMC founded by BECO in co-operation with the Western Province of South Africa for companies located in the Sacks Circle industrial area Bellville South is the division of BECO-ISP where the final thesis is based upon. The four members of this WMC are:

- GKN Sintermetals - Falke Eurosocks - Nettex

- Usabco

1.2.1 GKN Sintermetals

As the name supposes GKN Sintermetals is occupied with the production of metal parts for the automotive industry. Started in 1963 with introducing the Powder Metallurgy (P/M)-Technology into South Africa, they are nowadays producing more than 12 million parts, mostly used in the motor industry, a year. The parts are normally designed and created by the customers so GKN role is to produce the parts the clients are asking for. This means that price, quality, and speed of delivery are important values for GKN.

1.2.2 Falke Eurosocks

The South-African division of the German Company Falke fabricates socks and pants for the African and American market. They are operating as a independent divisions which means that the whole company is located around the factory, with own departments as marketing, design, CRM, acquisition and sale. Most important values for Falke are quality and innovation.

1.2.3 Nettex

Nettex is an home furnishing fabric manufacturer which produces jacquards, linings and lace for the African and Britain market. Nettex’ strength is the large range of fabrics they offer for a relatively low price. They are, just as the other companies do, aiming for sustainable solutions with regard to the environment.

1.2.4 Usabco

As the largest producer of household products in South Africa Usabco offers a large range of products from mobs to plastic office products. They are making use of the molding techniques to transform fluid plastics into the most thinkable shapes.

1.3 Purpose of the WMC Sacks Circle - Bellville South

The purpose of the WMC is to provide medium-sized companies in the Western Province with the opportunity to collect and share the existing knowledge about environmental issues and to create new

(11)

knowledge by doing studies and surveys with the help of the consultants of BECO-ISP, but also by encouraging foreign students to follow their internship, or do their final thesis, at BECO-ISP.

1.4 Reason to do the Research

The occasion of this research is strengthened because of some influences. The encouraging role of BECO-ISP towards the members of the WMC to do a study with regard to the energy usage, consumption and costs of the company on itself is stimulated by new circumstances. BECO-ISP couldn’t get the clients encouraged if the following circumstances where not rising:

• In the past the price of energy in South Africa was relatively low, because the primary fuels, needed as or to generate the energy were available in South Africa, no import from foreign countries was necessary. But prices rise, and not that slowly, so the importance of energy efficient production is becoming an big issue.

• The cultural change in the way of thinking in South Africa (Cornelissen, 2004) is changing from short term thinking towards long term thinking. This change is noticeable because of the growing South-African economy and the international role of this grow.

Western companies pay a lot of attention to strategic decisions where in South Africa this strategic management is a new kind of management. This new view also pays attention to investments with longer payback times like environmental decisions.

• The whole way of thinking about energy and energy savings is the last new issue. Because people are now known with the coming lack of energy they are easier to encourage to pay attention to the energy-issue.

This way of thinking about energy resources, usage en costs was the main reason for BECO-ISP to start a research with as purpose to search for possible savings of energy at the members’ factories.

Appendix I shows the outcome of searching for energy savings in a factory (Dr. R. Muller, State University of New Jersey: Rutgers, a self-assessment workbook). This search results in two different kinds of interesting problems. The first group of problems are direct savings by making improvements of the equipment in the factory. The second group is savings with a management and control character; the main problem was the understanding of peak demand. Where the first group of savings are savings in energy usage, the savings with peak demand are savings in energy costs. This thesis pays attention to peak demand management and peak demand control as being a part of the coordinated energy management philosophy.

(12)

1.5 Research Plan

Before defining the research problem, the peak demand problems should get analyzed for the four companies one by one to detect the need to start the research. This pre-research will encourage the attention to the problem and sign the need to do the research. The remaining of this thesis will include the following

Chapter 2 : Justification of the research Chapter 3 : Research Problem

Chapter 4 : Analysis

Chapter 5 : Simulation Model Chapter 6 : Results

Chapter 7: Conclusions and Recommendations

(13)

2. JUSTIFICATION OF THE RESEARCH

2.1 Introduction

“would you tell me, please, which way I ought to go from here?”

“that depends a good deal on where you want to get to,” said the cat.

Lewis Carroll, 1865

Alice’s Adventures in Wonderland

Energy is essential to life and survival. Reduced to bare essentials, stripped of thermodynamics, economics, and politics. Energy may well be the item for which historians remember the last half of the twentieth century. For several centuries mankind has grown lazy, lulled into complacency by the ease with which multitudes could be fed, housed, and transported using the abundant supplies of low- cost energy which were readily available. Then in less than a decade (1973-1981) the bubble which had taken 114 years to swell (since Drake’s first well in 1859) finally burst. Long unheeded warnings took on a prophetic aspect as fuel shortages and rising costs nearly paralyzed the industrial economies and literally shocked the world into an inflationary period which is not yet ended.

Tragically, the finiteness of energy resources may also be moving the world closer to war. Resources of all types are essential to war, and in themselves can be causes for war and for the rise and fall of nations. Twenty-four centuries ago, Greece denuded its forests building ships to continue the Peloponnesian Wars, in 1940, Germany seized the Rumanian oil field at Ploesti when it could no longer import petroleum due to the British blockade; the Israelis captured the Egyptian fields in Sinai.

Even nowadays wars are existing because of energy resources, like opponents of American President Georg W. Bush, call it the motive of the war in Iraq.

Efficient energy use therefore not only increases one’s independence of external energy supplies and decreases the relative costs of energy, but also helps defuse a potential unstable international situation.

The whole type of management related to efficient energy use is called energy management.

2.2 Energy management in South Africa

Also in South Africa, around fifteen years ago, the attention rises towards the opportunities to produce more energy-efficient. Before that change in thinking about efficient energy usage, the attitude towards energy is the same as nowadays the negotiations of talking about the slinking amount of the energy resources: “It’s cheap, we’ve got enough, so why paying attention to it?” This approach of thinking about the energy use no longer exists and therefore energy management issues are rising in the South African Government, the industry and also the households. After the usage of fuels, the mostly used energy is electricity. Electrical energy management is one of the biggest issues in the Energy Management. One of the reasons for the rapid growth of electricity use is convenience and

(14)

ease of control. The first of the ten energy principles for electrical load is therefore to optimize the control of the use. All the principles are (Smith, 1991):

1. Optimize controls 2. Optimize capacity 3. Reduce the (peak) load 4. More efficient processes 5. More efficient equipment

6. Employ special techniques to reduce losses 7. Energy containment

8. Cascade of energy uses 9. Energy conversion 10. Energy storage

2.3 Load management

Principle three of the energy management principles (2.2), reduce the (peak) load, is an existing and located problem at the companies this thesis is based upon. So reducing the peak load, called as both demand side management (DSM) and peak load management, is a basic energy management principle and can therefore not be ignored.

Firstly Demand Side Management was only recognized as an important type of control at the electrical utilities. Nowadays utilities are putting a part of the responsibility upon the shoulders of the factories by making use of different types of price-policies. One type is by offering time-of-use (TOU) tariffs that influences the cost-effectiveness of load shifting during a day. The electricity user is encouraged to use electricity during off-peak times (by night and during the weekend). Another policy, also making use of the principle of load shifting is to charge peak electrical demands. Peak electrical demands occur when the largest or greatest number of electrical loads are ‘on’ simultaneously. There are two ‘time’ components of peak demand data: the integrating period of the meter, which is in the Western Province (WP) fifteen minutes, and the period over which the demand is assessed, normally around the four weeks. The most important characteristics of peak demand charges can be quite different: (Piette, 1991)

• The size of the charge (in WP, SA: R43 per kVA)

• The daily schedule of the demand period – the peak demand might be measured during any hour of the day, or during on-peak hours only (in WP, SA: during any hour of the day)

• The seasonal variation – demand charges may be imposed every month, or only at certain seasons of the year (in WP, SA: every month and always at same rate, unless price changes)

• Demand Retchets – the ‘billed’ demand may be based partly or totally on the highest monthly peak demand from past months (in WP, SA: billed demand is independent from past months)

(15)

2.4 The role of electric load analysis

First step to indicate the importance of peak demand management at the companies in the WMC is to do electric load analysis, an extremely useful method that can be implemented with the existing electric meter. A daily load curve can be constructed by reading the meter every 15 minutes. To construct the daily load curve it is necessary to plot these data against time. The information contained in such graphs can then be analyzed to determine the following electrical load characteristics:

• time of occurrence of peak demand

• energy use during lunch breaks

• ratio of maximum to least demand

• percentage of total energy use occurring during off shift hours

• percentage of total energy use occurring on weekends

In an ideal situation, the load curve would be in the shape of rectangle. It would be zero during off- shift hours, rise instantly to a maximum value when the working day starts, remain at a constant value until the end of the working day, and the decrease to zero again. To get insight about these characteristics there are some parameters for electric load analysis: (Smith, 1979)

• The diversity factor; which is the ratio of the sum of maximum demand of the total individual loads at the plant and the maximum peak demand of the plant.

(UsageMachine 1 + Usagemachine 2 + ….. + Usagemachine N)

Peak demand at month(n)

• The load factor; which is the ratio between the energy used during period p and the maximum demand during period p times the period p

Total electricity used at month (n) 1 *

Peak demand at month (n) hours in month (n)

The diversity factor points out the difference between the real peak demand and the maximum possible demand (total load of the factory). So a small factor implies a large use of the total load during peak demand. The load factor on the other hand points out the difference between the real peak demand and the mean demand during the billing period. A large difference between peak demand and mean demand implies opportunities to decrease peak demand, whereas a small diversity factor strengthens the opportunities because of better possibilities to shave peaks.

2.5 Electric load analysis at the companies

For each of the four companies the load curves are collected during a period of around the forty days.

In the graphs the mean peak loads of an interval of 15 minutes are drafted for this period of forty days.

(16)

factor and the load factor, are shown. One assumption with regard to this parameters is that the peak loads during off-shifts are not taken along. Typical values of the diversity factor and the load factor for industrial plants are respectively 1.3-2.5 and 0.8-0.9. (Smith, 1979, Tennessee Valley Authority)

Company

Total Electricity use (kVAh)

Peak Demand (kVA)

Periods of 15

minutes Power (kW) Load Factor

Diversity Factor GKN 559590 510,3 1521 772 0,72 1,51 Falke 676792 692 1214 1124 0,81 1,62 Nettex 1867524 1610 1544 2262 0,75 1,40 Usabco 1067364 1388 1166 2080 0,66 1,50 Table 2.1 the diversity factor and load factor

Results of the calculations indicate that firstly the load factors are beyond typical value. There is also a difference noticeable between the more constant loaded companies as Falke and Nettex are in comparison with the variable load factories as GKN and Usabco are. Firstly the outcome of the load factors results in an collective need to proceed the research. Second notice are the diversity factors.

The inverse of these factors shows the percentage of total machines running during peak time. First implication seems to be that the lower the diversity factor the easier it is to improve the load factor.

Second indication is that the diversity factors are very satisfying so strengthen the proceeding of the research.

The load factor is a ratio, which means it can be improved by changing the parts of the ratio. So the load factor can be improved by enlarging the energy used during a period or by declining the maximum (peak) demand during period p. The first opportunity is a not-wanted, contrarily one, and indicates the weakness of making use of ratios: the opportunity to improve the ratio by using window- dressing-methods.

The only accepted way to improve the load factor is by declining the peak demand. This is named peak shaving.

The definition of peak shaving is to decline the amplitudes of a graph. This can be done by moving demands through time. The objective of this thesis (chapter 3) is to explore the consequences of implementing management decisions with the purpose to reach this movements of demand through time. This exploration will be accomplished by simulating the differences between the old situation and the new situation after implementing one or more possible management decisions.

2.6 Simulation

The word simulation is descend from the Latin verb ‘simulare’ which means something as ‘imitate’

and ‘copy’. Simulation is the imitation of a system based on knowledge or hypotheses concerning the conduct of parts from the whole system with the objective to obtain conception into the conduct of the studied entire system. (Hillen, 1993) simulation model is a model that mimics reality (Robinson) Applications of simulation and simulation models can be found in science and industry. The simulation of the peak demand management tools is categorized into the simulation of production and logistic systems. Simulation is an experimental method, so purpose is not to obtain the optimal

(17)

Description of the System

Conceptual Framework

Simulation Model

Results of Experiments

Conclusions solution to a problem, as with mathematical methods, but to find a correct, precise and useful solution.

In this research the management decisions has got to be useful.

A simulation study consists of an action plan with different iterative phases (Kelton, Law, 2000).

These phases are described shortly: (Figure 2.1)

1) Description of the reality. In this orientated phase the issues are surveyed and the research problem is drawn up

2) The translation from the reality into a conceptual simulation framework.

3) The conceptual framework converts into an actual simulation model.

Specified are the numeric data which has to be used. Adjacent to this the conceptual framework will be converted into programming language. The model will be checked on its behavior.

4) Implementing the management decisions into the model and obtain the results of the experiments. Then comparing the results with each other. Each possible change is called a scenario.

5) Last step is the translation from results in the model into results into the reality. What can be concluded about the results after translating them into the reality?

2.7 Consequences for the research

The exploring view of this chapter, on energy management and environmental management in general and peak load management as a part of this management principles, results in a large bases to proceed the research. The electric load analysis shows the micro-importance of this research where also macro- influences with regard to modern environmental thinking, as stated in the beginning of this chapter, subscribe the importance of this research.

This research will follow the action plan as stated above. The first phase, the description of the system will be reviewed in both chapter three (research problem) and chapter four (description of the reality).

Chapter five will account for phase two and three, chapter six for phase four and at chapter eight the simulation study finishes with phase five: the conclusions.

(18)

0 10 20 30 40 50 60

30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 1 2 3 4 5 6 7 8

January February March

Date

kVA peak (div 10)

0 10 20 30 40 50 60 70 80

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

May June

Date

kVA peak (div 10)

0 2 4 6 8 10 12 14 16 18

20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

May June

Date

kVA peak (div 100)

0 20 40 60 80 100 120 140 160

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

May June

Date

kVA peak (div 10)

Figure 2.2: The load graphs of respectively GKN, Falke, Nettex and Usabco

(19)

Description of the System

Conceptual Framework

Simulation Model

Results of Experiments

Conclusions

(20)

3. RESEARCH PROBLEM

After the justification of the research in the previous chapter the overall problem fields became clear.

The rising attention to energy efficiency took along the intention to become more familiar with energy management principles. One of these principles pointed out to be the (peak) demand management, also called as (peak) load management and demand-side management (DMS). This research deals with demand management as being one of the tools of the covering energy management style. This chapter starts with the problem analysis and the accompanying research questions. Subsequently the boundary conditions and the type of research are described.

3.2 Problem analysis

The objective of this research is by order of BECO-Institute for Sustainable Business to set up a study for their four clients.

By order of BECO-Institute of Sustainable Business to set up a study, for their four clients and members of the WMC Sacks Circle Bellville-South: GKN Sintermetals, FALKE Eurosocks, Nettex and Usabco, into the opportunities of energy-efficient production by making use of peak demand management decisions into their production planning and production process control.

The primary part of the objective is to improve the energy-efficiency of the production by making use of peak demand management decisions into the production planning. Second part are management decisions with regard to the control of the production process. The first part is the prevention of high peaks, the second part is to cure existing high peaks during the production..

3.2.1 The problem definition

With this objective the following problem definition can be drafted:

Which load management method or combination of methods show(s) the best results with regard to savings in the monthly peak demand charges and reach(es) a more energy-efficient production process?

3.2.2 The research restrictions

The restrictions belonging to this research are:

(21)

• The performance of the production process is more important than the energy efficiency, so may not suffer under outcomes of this research.

• The intention is to make the energy management principles accessible for the management of the companies, not to implement actually the management styles into the companies

• Potential energy savings must be even matching with the costs to create this saving.

• The study is accomplished at GKN Sintermetals and not at one of the other companies because the production process at GKN showed the most similarities with the other companies, has got a relatively low complexity and the data at GKN is easier collectable.

• The total applied time to complete this research is a maximum time of 8 months.

• The main part of the thesis is written in The Netherlands, so the research is restricted with the possibilities to get provided with the needed data and information.

3.2.3 Type of research

De Leeuw discerns five types of research:

• Pure scientific research

• Social relevant research

• Policy relevant research

• Policy supporting research

• Problem solving research

He depends this distinction in the growing degree of customer directed research. This makes it possible to classify this research in a mix of social relevant, policy relevant and policy supporting research. (see figure 3.1)

Figure 3.1: Classification of the research according to De Leeuw Social relevant

research

RESEARCH

Policy supporting research

Policy relevant research

(22)

Baarda and De Goede distinguish three types of research:

• Exploring research

• Empirical research

• Describing research

This research can be classified as a empirical research with both exploring and describing research influences.

(23)

4. ANALYSIS

In this chapter an analysis is done into the present situation at GKN Sintermetals, the opportunities to save in peak demand and reproduced experimental factors, the planning method, the control method and the hypothesis with their restrictions. In addition, the usefulness of a simulation study should be described.

4.2 The production

GKN Sintermetals is a company that manufactures sintered components, including sprockets, bearings and filters for automotive OEMs and the auto aftermarket business. They make use of the metal transformation tool sintering. Sintering means that metal parts are pressed from metal powders into their desired conditions and being heated subsequently to reach the desired material properties. After the sinter process secondary operations take care for the optimal needed conditions. Figure 4.1 shows the division of the factory:

Figure 4.1: The divisions of the factory

FACTORY

Offices Factory Floor

Red Green

Blue Orange Yellow

Toolroom

Other Furnaces

(24)

The different machines mostly all execute batch-operations and are divided into manufacturing cells.

Some machines in the furnace cell execute process-operations. The different operations consists of:

- Blending the powder of the raw materials and other ingredients together - The compacting (pressing) of the blended powders.

- Sinter process (the thermal treatment where the compacted parts are heated) - Repressing the parts to increase the density of sintered parts

- De-burring (removing edge protrusions (‘burrs’) - Other operations (CNC, drilling ed.)

- Steaming (This treatment improves corrosive strength, increases surface hardness and above all wear resistance)

- Impregnation (makes the parts self-lubricated components, decreases the porosity) - Shot blasting (removing edge protrusions on sensitive parts)

- Packing of the end products. (no machines needed)

The ten operations at the factory floor are divided into the seven existing manufacturing cells. They are grouped by both line manufacturing structures as well into functional set up. Because of two reasons the set up structure used in this thesis is the functional set up. First reason is that the needed product routings all pursue a functional lay-out, second reason is that a lean structure can be a part of the functional structure and not visa versa. Motives to choose a functional set up instead of a lean manufacturing set up are firstly the large variations in produced products. It is impossible to set up a separate lean for each possible routing, because of lack of space and too small occupation rates of the machines. A production process with a large variation of product routings while producing small quantities is job shop. This amplifies already the existence of both structures at GKN. When some products are produced in large quantities it is possible to set up a product-oriented lean manufacturing cell. Second motive to choose for a functional lay-out is the rise of flexibility and a third motive could be the ease to exchange function-specific knowledge when similar work places are grouped together.

Disadvantages of the functional set up are the large run times en the large stocks of work in process (WIP). Queue times are playing the biggest role in these large run times. Queue times exists because all the product routings have got different operation times. Queue times largely increases when the occupation rates of machines are high (80%-100%). Queue times also increase when a planned operation is cancelled, this affects a large part of the total production plan. These two characteristics of queue times are very important to the research and should return further on this thesis. Figure 4.2 (next page) shows the both functional layout of the factory floor at GKN with its machines grouped into the different manufacturing cells and the product routings necessary for the research (paragraph 4.4).

(25)

e 4.2The floor plan and product routings

Po wder Blending

Yellow Cell Red Cell Green Cell

Blue Orange Cell

FH D F urn ace

Crem er f urn ace

Elino S5

100t 200t 250 g 60t

500t400t Maldaner Impregnation

Steam Shotblast

R4 R4

R4 lathe Bar rell 6t Wire brush CNC 5035 CNC 5036

S5

A- pex CNC 5036

C60 dryBar .

280G 280G 50t C 60

Bar rell

oilmc

6ot 100t

C 110

drills

C60

OUT

IN

(26)

4.3 The experimental factors

4.3.1 First experimental factor: the maximum adjusted monthly peak demand

The first experimental factor in the simulation study is the item where the simulation study is based upon: the monthly peak demand of power use. In paragraph 2.4 and 2.5 it came clear what the peak demand is, why it exists and how it is used. Paragraph 2.6 gave insight into the opportunities to decline the monthly peak demand by introducing two factors, the load factor and the diversity factor.

In this paragraph the both factors at GKN are analyzed, what are the possibilities at GKN to decline the peak demand and how can this decline been achieved? This analysis results in the second experimental factor, namely the introduction of a planning and a control method. After drawing this methods a third and last experimental factor is needed to get the most useful results out of the simulation study.

4.3.2 Second experimental factor: the planning and control method

This paragraph describes the next experimental factor. The monthly peak demand is the experimental factor we want to reach, whereas this second experimental factor is the way we want to reach that first factor. Before the second experimental factor can be drawn, first it is needed to get clear the opportunities to decline the monthly peak demand and the possibilities of successfully making use of this opportunity.

4.3.2.1 The opportunities at GKN

The four companies in paragraph 2.6 wherefore the load factors and diversity factors are calculated showed a relatively small load factor for all the four companies. In that case the diversity factor indicates the opportunities to improve the load factor. Because of the high score on the diversity factor at all the factories, all far above the optimal factor one, these opportunities are relatively large. At GKN the mean load factor in the month February 2004 was 0,72, which means an increase of 0,08 must be easily reachable, whereas an increase of 0,18 is more desired . (Tennessee Valley Authority).

An increase of the load factor is attainable when there is a decline in monthly peak demand (or an increase in the monthly total use of power). The possibility to increase the load factor brings along an opportunity to decline the peak demand. The monthly load factors for GKN are shown in table 4.1.

This table includes the decline in monthly peak demand when the load factor is increased to both a reachable level of 0.8 and a desired level of 0,9. The load factor for both the months January needs to be neglected because of the Summer stop.

The table shows some interesting points. Firstly, the mean load factor for the years 2002 and 2003 was around 66%, so the calculated load factor for February 2004 was above average and indicated a better load factor than the load factors normally are. This information grounds the need to do the simulation study and enlarges the possibility to improve the mean load factor. Secondly, an increase of the mean load factor to 80%, resulted the last two years in an decline of peak demand around the 18%. This relatively large decline enlarge the possible benefits of this research. Third implication is the

(27)

possibilities to decline the peak demand. The smallest peak demand in table 4.1 is the peak demand of 348 kVA´s in September 2002. This value shows the possibility to produce a whole month without reaching a peak demand of 348 kVA. When this is possible in one month, why not in any other month?

2002

Load Factor*

Peak Demand

Peak at LF 80%^

Peak at

LF 90% 2003

Load Factor

Peak Demand

Peak at LF 80%

Peak at LF 90%

January N/a 431 n/a N/a January n/a 449 n/a n/a

February 61,0% 531 405 360 February 65,9% 453 373 332

March 61,6% 530 408 363 March 75,0% 396 371 330

April 62,4% 438 342 304 April 55,3% 451 312 277

May 50,3% 483 304 270 May 62,9% 468 370 327

June 74,3% 384 357 317 June 63,0% 394 310 276

July 56,7% 458 324 288 July 72,7% 461 419 373

August 60,0% 441 331 294 August 63,4% 534 423 376

September 80,0% 348 348 309 September 58,7% 472 346 308

October 70,4% 382 336 299 October 66,7% 516 430 383

November 64,1% 417 334 297 November 79,9% 518 517 460

December 76,7% 417 400 355 December 66,6% 531 442 393

Mean 65,2% 438 354 314 Mean 66,4% 470 392 349

* the load factor has got a correction of 13% because of the non-productive weekends.

^ calculated by filling in this load factor in the existing load factor calculations Table 4.1 The peak demands, load factors and opportunities at GKN

4.3.2.2 The possibilities to improve the load factor at GKN

The description of the production process in paragraph 4.2.1 showed the job-shop structure at the factory floor. This structure makes it possible to improve the load factor by shaving peak demands.

Shaving peak demands amplifies that the power use can be spread over a certain time span (the billing period). This distribution can be achieved by spreading running hours of certain machines during this period of normally around one month. Which machines are needed to optimize this spread depends on both the power consumption and the occupation rates of the machines. The bigger the machine’s use of power the more effective the distribution is, and the smaller the occupation rates the easier it is to spread this machine. In figure 4.3 the running hours and the power usage of the machines are drafted.

The occupation rate seems to be the same as the running hours, but is not. The type of running hours of the machines can be split up in four groups:

1) Machines running constantly during the whole production time, occupation rate = 100%

2) Machines running constantly during a certain part of the production time, like offices opened at daytime and furnaces operating densely one month a year.

3) Machines running variable during the whole production time, the occupation rate < 100%

4) Machines running variable during a certain part of the production time, occupation rate <

100%

(28)

0,0%

10,0%

20,0%

30,0%

40,0%

50,0%

60,0%

70,0%

80,0%

90,0%

100,0%

500t (3) cremer (1) fhd (1) compr. (1) 400t (3) 60t (3) degussa (2) 200t (3) 100t (3) CNC (3) CNC (3) CNC (3) CNC (3) CNC (3) TX (2) Calch. (4) Maho (4) APEX (3) 280G (3) Blohm (4) Elino (2) 280G (3) Cut wire (4) 280G (3) 60t (3)

Occupation rate (%)

0 10 20 30 40 50 60 70 80 90 100

Power usage (kW) Occupation

rate (%)

Power (kW)

Figure 4.3 The running hours and power usages for heaviest power users

A machine’s occupation rate of 100% makes it impossible to spread the production at this machine over time. So only type 3 and type 4 machines are useful into this research. The machines in figure 4.3 are categorized into these four groups. The three large power users grouped in type 1 are utilities (furnaces and compressor) and run every hour of the day. The figure shows another four distinctive large users, namely one machine of type 2 and three machines of type 3 (500-t press, 400-t press and 60-t press. These last machines make the difference with the other four large users and show their importance towards the possibility of peak shaving at GKN. The three type 3 machines add-on with two other type 3 machines, namely the 100t and 200t, are located in the same factory cell (yellow).

These five machines are further called as the ‘method-machines’. Reasons to not increase this group with other type-3 or type-4 machines, like the CNC’s in the Green factory cell are:

- The usage of the other machines

The other type-3 and type-4 machines don’t have got an significant share in the total usage of all the machines. The five now called method-machines (< 20% of total type-3and type-4 machines) pays for a large part (>80%) of total electricity usage. This 20/80-rule is a method to separate items in distinctive groups (Kaplan, 1998). If an machine has got a significant value of electricity usage there are two more reasons not adding it to the group of method-machines:

- The place in the production process

The type-3 and type-4 machines with an significant usage, like the CNC’s in the green fabrication cell, are used for operations to many different product. This machines are the core of the functional lay-out, which makes using them as a method machine in this simulation study too complex.

- The occupation rate of the machines

An other reason not to use this type-3 and type-4 machines, like the CNC’s with an significant electricity usage, is that the occupation rates at this machines are too large (>80%). Problem with interrupting this machine is that the occupation rates are rising towards the maximum of 100%. Queue lengths will also rise progressive when the occupation rates are increasing (see figure 4.4: arrival of orders uncertain (Poisson distribution)). (Land, 1999)

The above described boundaries in enlarging the group of method-machines under scribe the need to involve only the five yellow cell presses into the planning and control methods researched later on in

(29)

this thesis. Now it is clear with which machines peak shaving is possible, planning and control methods can be drawn up to achieve this peak shaving. The drafted methods are the second experimental factors into this simulation study.

Figure 4.4 The Occupation Rate related to the Mean Queue Lenght

4.3.2.3 The planning method and the control method

In this subparagraph the second experimental factor, the drafted methods, are framed. The purpose of these methods is to decline the monthly peak demand which results in a more equal power usage curve. The methods to use can be divided into two types of methods: planning methods and control methods. Possible planning methods could be:

• to avoid heavy powers users running simultaneously

• to avoid complete factory cells running simultaneously

Possible control methods are: when the peak demand reaches a certain level

• to switch off the most heavy power user

• to switch off the machine with the highest ratio usage/occupation rate

• to switch off the machines with the smallest occupation rates

• to switch off the machines with the smallest labor occupation

In this research it is preferred to use, for each type of method, only one of the methods above into the simulation study as the second experimental factor.

Method I: the planning method

The first method, a planning method, is drawn to avoid heavy power users to run simultaneously. In

Mean Queue Length

Occupation Rate

0% 25% 50% 75% 100%

(30)

all members of the yellow fabrication cell. The four type-3 machines used into the planning method should be the 500-ton press, the 400-ton press, the 200-ton press and the 60-ton press. The 100-t press situated in the yellow cell should be deleted because the power measurements monitors some difference with the 100-t press values, collected at this machine located in the Orange Cell. Method I can be described as:

Only 1, 2 or 3 machine(s) of the heavy users: the 500-ton press, the 400-ton press, the 200-ton press and/or the 60-ton press (all members of the yellow fabrication cell) are allowed to run together at the same time.

Method II: the control method

The second method, a control method, is drawn to act when the peak electricity demand reaches a certain level. This level was already called as the first experimental factor. When the peak reaches this level the control method is then used to keep the peak on or below this level. This is done by switching off one of the three heavy users, namely one of the machines also used in the planning method, method I. Method II can be described as:

When the peak demand reaches a certain adjusted level, the method-machine complying having the best results with regard to the decision values (see model III at page 35 and the explanation of model III at page 34) needs to be switched-off. This machine could be the 500- ton press, the 400-ton press, the 200-ton press and/or the 60-ton press. When the level is still above the level or reaching the level after this intervention again, the next machine needs to be switched-off and so on. When the peak is decreasing below the adjusted peak level a machine is allowed to run again when the difference between the peak level en de current peak is larger than the usage of that particular machine.

Figure 4.5 shows the decision models of the both methods. Model I is reproduction of the present situation, Model II of the present situation added with method I, Model III of the present situation added with method II and model IV a reproduction of the present situation added with both methods (I+II). The models show: when a method is called, which influence the method has got, and the priority rules of the method. The figure contains four models. All the models have got two static starting points: when an order is sent into the production process (the product itself is following the model route) and when an finished order is pushed out of the production process. This time the product is no longer part of the model. The model this time is just an decision model how to act after one machine or worker can starting a new order out of the order stream or out of one of the possible two (model II, III and IV) ‘waiting rooms’. Model III and IV have got a third, continuous starting point: the peak measuring. At this starting point the product and the machine which is producing that product needs to follow the route about what to do when peak demand is to large. The explanation of the four models: (in the explanation the used boxes are printed italic).

Model I:

Model I is the decision model of the present situation. An order is send (order flow) to the (empty machine in) production process. If the maximum number of machines is already running (Number

(31)

machines running => max number machines?) the product (and his machine) has to wait (waiting room). The second starting point of the model is when an finished order is leaving the production process. If there is still an product in the waiting room, this product can be sent to its machine in the production process. If not, the model stops running till an new order is sent to the production process or an order is finished.

Figure 4.5: Model I The present situation

Model II:

In this model method I is added to the model of the present situation. The situation now is than an order can only be send to a method-machine if not more than a certain amount of method-machines is already producing (and the total number of machines is not proceeding the maximum number of machines). The two starting points are the same as in model I: when an order is placed and when an order is finished. When an order is placed: First decision is to separate the products going to a

ORDER FLOW

PRODUCTION PROCESS

Waiting Room Occupied?

Number Machines Running => Max Number

Machines?

Waiting Room

Send first in

out

NO

YES

De clin in g Peak Demand

CURE

IS BETTER THAN

PREVENTION

A Research in Peak Demand Management at GKN Sintermetals Inc., Cape Town, S.A.

De clin in g Peak Demand

CURE

IS BETTER THAN

PREVENTION

De clin in g Peak Demand

PREFACE

ABSTRACT

TABLE OF CONTENTS

PREFACE 3

ABSTRACT 4

1. INTRODUCTION 9

2. JUSTIFICATION OF THE RESEARCH 13

3. RESEARCH PROBLEM 20

4. ANALYSIS 23

5. SIMULATION MODEL 44

6. RESULTS 53

7. CONCLUSIONS AND RECOMMENDATIONS 66

BIBLIOGRAPHY 69

APPENDICES 71

1. INTRODUCTION

2. JUSTIFICATION OF THE RESEARCH

Description of the System

Conceptual Framework

Simulation Model

Results of Experiments

Conclusions

3. RESEARCH PROBLEM

4. ANALYSIS