• No results found

Evaluating a Power Supply System for a Small-Scale Cocoa Processing Plant


Academic year: 2021

Share "Evaluating a Power Supply System for a Small-Scale Cocoa Processing Plant"


Loading.... (view fulltext now)

Full text


Evaluating a Power Supply System for a Small-Scale Cocoa Processing Plant

A Multi-Criteria Decision Analysis Approach

Alexander Rothoff


Student thesis, Master degree (one year), 15 HE Decision, Risk and Policy Analysis

Master Programme in Decision, Risk and Policy Analysis Supervisor: Fredrik Bökman

Examiner: Ulla Ahonen-Jonnarth


Evaluating a Power Supply System for a Small-Scale Cocoa Processing Plant A Multi-Criteria Decision Analysis Approach


Alexander Rothoff

Faculty of Engineering and Sustainable Development University of Gävle

S-801 76 Gävle, Sweden




The global supply and demand of energy is facing different challenges. On one hand an increasing energy demand, foremost in developing countries, and an increasing pressure on reaching climate goals changes the requirements on the design of power supply systems. This may be particularly relevant in terms of decentralized energy solutions and hybrid systems that incorporates renewable energy sources. This study exemplifies the use of multi-criteria decision analysis (MCDA) to evaluate different power supply alternatives for a small-scale cocoa processing plant (SSCPP), placed in Côte d´Ivoire. MCDA is an analytical approach to evaluate decision alternatives according to certain criteria, with the aim to find a preferred alternative. The function of a cocoa processing plant depends highly on its power supply. This study has been performed by first analysing the energy needs of the processing plant, which includes electricity, heating and cooling. Based on those specific energy needs different power supply alternatives have been created. In a following evaluation it has been exemplified how MCDA can be used within the energy sector to evaluate different alternatives. Within this exemplifying evaluation, following power supply solutions have been considered: power grid, power grid with back-up generator, power grid with LPG-heating (Liquified Petroleum Gas), power grid with solar energy, off-grid solar system with back-up generator and an off-grid generator with heat exchanger.

The evaluation of alternatives has been made by using the three evaluation attributes:

levelized cost of energy (LCOE), loss of load hours (LOLH) and carbon footprint of energy (CFOE).



Abbreviations and Terminology ... 1

1 Introduction ... 2

1.1 The Aim ... 3

1.2 Content and Structure ... 4

1.3 Terms and Expressions ... 4

1.4 Conditions ... 5

1.5 Assumptions and estimations ... 5

1.6 The Author ... 6

2 The Cocoa Industry ... 7

2.1 The Cocoa Value Chain ... 7

2.2 Small Scale Cocoa Processing Plant... 8

3 Methods ... 10

3.1 Multi-Criteria Decision Analysis ... 10

3.1.1 MCDA Approaches ... 10

3.1.2 Evaluation of Consequences ... 13

3.1.3 How MCDA has been used in this thesis ... 16

3.2 Energy Analysis ... 17

3.2.1 LCOE - Levelized Cost of Energy ... 17

3.2.2 CFOE - Carbon Footprint of Energy... 19

3.2.3 LOLH – Loss of Load Hours ... 20

3.2.4 Solar power input ... 22

3.2.5 Energy and power demand ... 24

3.3 Collection of information ... 25

4 Power Supply Requirements ... 27

4.1 Target Requirements ... 27

4.2 Energy Requirements ... 28

4.2.1 Drying ... 29

4.2.2 De-hulling ... 29

4.2.3 Roasting ... 30

4.2.4 Grinding... 30

4.2.5 Tank storage and piping ... 31

4.2.6 Tempering, blocking and storage ... 31

4.2.7 Summary of energy needs ... 31

5 Objectives and Criteria ... 33

5.1 Cost... 35

5.1.1 Investment cost... 35

5.1.2 Annual costs ... 35

5.1.3 Lifetime ... 35

5.2 Environmental Impact ... 36

5.3 Reliability ... 37

6 Power Supply Alternatives ... 39

6.1 Alternative 1 ... 40

6.2 Alternative 1b ... 41

6.3 Alternative 2 ... 42

6.4 Alternative 3 ... 43

6.5 Alternative 4 ... 45

6.6 Alternative 5 ... 46

7 Consequences ... 48

7.1 Levelized Cost of Energy (LCOE) ... 48

7.1.1 Investment Cost ... 48


7.1.2 Annual Cost ... 49

7.1.3 System Lifetime ... 50

7.1.4 LCOE values ... 51

7.2 Carbon Footprint of Energy (CFOE) ... 51

7.3 Loss of Load Hours (LOLH) ... 53

7.4 Summary of Consequences ... 54

8 Evaluation ... 55

8.1 Utility values ... 56

8.2 Weight assessment... 58

8.2.1 Swing weighting ... 58

8.3 Overall utility ... 60

9 Production scenarios... 62

9.1 Scenario evaluation ... 62

9.2 Production scenario results ... 63

10 Sources of uncertainty ... 66

10.1Internal uncertainties ... 66

10.2External uncertainties ... 66

11 Discussion ... 69

Acknowledgements ... 71

References ... 72

Appendix I - Fact sheet “Cocoa liquor mini-plant” ... 76

Appendix II - Energy needs of machinery ... 77

Appendix III - Hypothetical decision maker ... 79

Appendix IV - Power supply and energy cost by alternatives ... 82

Appendix V - Investment and annual costs of alternatives ... 84

Appendix VI - Carbon Footprint of energy sources ... 86

Appendix VII - Solar energy yield ... 88

Appendix VIII - PV simulation ... 92

Appendix IX - Production scenarios ... 94



Abbreviations and Terminology

CFOE = Carbon Dioxide Footprint of Energy1 CO2eqv. = Carbon Dioxide Equivalents GHG = Greenhouse Gases

HES = Hybrid Energy System LCA = Life Cycle Assessment LCOE = Levelized Cost Of Energy2 Li-ion = Lithium Ion

LOLE = Loss of load events LOLH = Loss of load hours LPG = Liquefied Petroleum Gas

MCDA = Multi-Criteria Decision Analysis MAUT = Multi-Attribute Utility Theory

Nibs = Cocoa bean kernels after having removed the shell

Off-grid = Electric power supply without connection to a local power grid On-grid = Electric power supply with connection to a local power grid OCC = Opportunity cost of capital

PV = Photovoltaics

SSCPP = Small Scale Cocoa Processing Plant

1 Also referred to as Carbon Dioxide Footprint of Electricity. Since the application here also relies on the generation of process heat, the expression energy has been used instead.

2 See remark above: Electricity vs. Energy.



1 Introduction

Two central issues in world politics are the efficient supply of energy and the worldwide distribution of wealth. This thesis tangents both these issues in that it deals with selecting the most suitable power supply out of a range of alternatives for a small, profit- bringing, processing plant in a developing country.

The future supply of energy is likely to change and already today it meets many challenges due to various requirements such as energy resources and climate impact. In Germany, the decision has been made to abandon the nuclear technology for the generation of electricity.3 Parallel with this, the automobile industry and the consumers are encouraged to increase the share of electric cars on the German roads.4 The closing down of nuclear power plants with a simultaneous increase of electric mobility put new requirements on electric grids and the overall supply of electricity. Globally international initiatives like the Kyoto-protocol stipulates the reduction of Carbon- dioxide emissions in order to reduce the global warming. At the same time the world is facing an increasing demand of energy in the years to come. In Africa, the energy demand of the sub-Saharan countries is predicted to raise by about 80 % until 2040.5 In India the increase in energy demand is forecasted to double within the same period and also China is expected to reach an energy demand peak in 2040.6,7 All in all the requirements on the designing and dimensioning of a power supply system change and face new challenges. Within this thesis an analytical evaluation process according to multi-criteria decision analysis has been used for choosing a power supply system. The use of multi-criteria decision analysis assists in selecting the best solution out of a range of alternatives according to defined objectives.

The processing of cocoa beans is typically made at an industrial level with high throughputs of 2 t/h of processed beans or more. While the cocoa tree is grown in a narrow area around the equator the refining process of the cocoa beans is mainly taking place in Europe and in the USA even if this trend has started to change. For the 2014/15 grinding season, the number one processing country was predicted to be Côte d´Ivoire before the Netherlands,8 indicating that local processing has been growing in importance even though this trend was interrupted in the following season 2015/16.9 Despite of this the cocoa industry is suffering from financial sustainability problems and a poorly balanced value chain. An increased refining in origin countries helps to improve local economies, but especially the farmers still suffer from low income and a very low share of the value chain of chocolate.10 To improve this situation one possibility could be to increase the value added on the cocoa farms or cooperatives. To be able to do this the cocoa beans need to be processed to a higher state of refinement, which requires processing equipment. This equipment would have to be adapted to the existing conditions and the amount of beans at such an installation site. Thus, it would require the use of a small-scale cocoa processing plant11 (SSCPP). Such a plant could either be

3 German Federal Ministry for Economic Affairs and Energy, 2017b

4 German Federal Ministry for Economic Affairs and Energy, 2017a

5 International Energy Agency, 2014, p.76

6 International Energy Agency, 2015, p.56

7 Xu et.al., 2017

8 International Cocoa Organisation, 2015, p.12

9 International Cocoa Organisation, 2017

10 Cocoa Barometer, 2015

11 The term "Small-Scale Cocoa Processing Plant" is by no means a standardized processing unit, which can be bought off the shelf. In this case the term refers to a processing unit with an output


3 installed directly on a cocoa farm that is big enough,12 or in smallholder farmer cooperatives.

Farmer cooperatives exist in most cocoa producing countries such as Côte d´Ivoire.13 This report will have a look at such a processing scenario in terms of the energy supply of the plant. The question that has been examined is how different power supply alternatives for such a processing plant can be evaluated in order to associate each alternative with a score according to chosen criteria and preferences. Thereby different possible solutions need to be evaluated and compared to each other. As approximately 72 %14 of the total produced cocoa comes from Africa, with the biggest share coming from Côte d´Ivoire, this has been chosen as the reference installation country for this report.

1.1 The Aim

The aim of this thesis is to exemplify how different power supply alternatives for a small-scale cocoa processing plant can be evaluated by using MCDA. The use of a SSCPP would imply a decentralized cocoa processing structure with benefits and challenges. One important topic for decentralized energy systems, especially in developing countries, is the adequate power supply.15 Secondly, an installation like a small-scale processing plant may have a lower “on-site” energy efficiency than bigger plants, which puts requirements on a well elaborated power supply to keep key-factors such as production cost and environmental impact low. Therefore, another aim has been to demonstrate how renewable energy solutions can be compared with traditional power systems in this kind of evaluation.

Similar MCDA-studies that deal with the evaluation of power supply systems already exist. However, most such studies deal with the electrification of rural villages with an optimization target while including renewable energy.16,17,18 This study deals with the evaluation of a set of power supply alternatives. The evaluation has been done according to a number of objectives: to minimize the overall cost, to minimize the emission of GHG and to maximize the power supply reliability. The chosen objectives are commonly used for the evaluation of energy systems, which allow this study to be related, if not directly compared, to similar studies. To enable such a comparison the aim has been to also use common attributes to evaluate the performance of the alternatives. The resulting selection of attributes are: LCOE (Levelized Cost of Energy), CFOE (Carbon Footprint of Energy) and LOLH (Loss of Load Hours).

Six relatively common power supply alternatives have been evaluated, with the idea to create an exemplifying study. The alternatives are mainly hybrid energy systems (HES), and consist of: power grid connection, power grid and LPG-heating (Liquified Petroleum Gas), power grid and solar power, power grid with back-up generator and two off-grid solutions: solar power with back-up generator and Diesel generator with heat recovery.

of 200 kg/h or less of grinded cocoa liquor. For a more detailed description of the plant see Section 2.2.

12 As in South America (Interview: Nengerman H.).

13 The size of a Côte d´Ivoire-cooperative varies considerably and may include between 200 – 2.000 members (Interview: Dr. Anga, J.-M.)

14 International Cocoa Organisation, 2015, p.6

15 In developing countries challenges such as regularly power losses or even non-existing power grids need to be dealt with.

16 Bortolini et.al., 2015

17 Gharavi et.al., 2015

18 Moharil & Kulkarni, 2010



1.2 Content and Structure

The described energy supply alternatives are not intended to represent an exhaustive range of alternatives, but rather as an exemplifying list of different on- and off-grid solutions to demonstrate how these may be evaluated. The result is intended as a reference for similar comparisons and, on one hand as a discussion basis for the decision-making process in the layout of small hybrid energy systems and small power producing units. On the other hand, it is meant as a support for possible future discussions around small scale cocoa production in origin countries.

On the following pages, first the background of the cocoa value chain problem and the SSCPP will be described more in detail. In Section 3 the used methods are described, followed by a closer look at the actual energy needs of the exemplified SSCPP and chosen evaluation objectives in Section 4 and 5 respectively. In Section 6 the evaluated power supply alternatives are described. In the following sections, 7 and 8, the evaluation takes off with a description of consequences and the following analysis of the overall utility. A visual presentation of the structure can be seen in Figure 1 below.

The different shades of blue in the figure are explained in Section 8.

Figure 1. Schematic representation of the structure of the thesis.

1.3 Terms and Expressions

There are different terms for naming a small power producing unit. For single small power producing systems the term distributed generation might be used or, for the producer, small power producer. However, also expressions such as power supply system or energy supply system are frequently used. Within this thesis, both of the later mentioned expressions have been used. The reason for using both expressions is context depending. The term energy supply has been used when it comes to the amount of energy which is supplied to the SSCPP. The term power supply is used in a more general context to express the need and supply of power. When it comes to provide energy, the term generation has been used. The term power generation is widespread and has therefore been considered adequate also for this thesis. For power supply alternatives with combined energy sources the term hybrid energy system (HES) has been used and for non-power-grid connected alternatives the expression off-grid has been adopted even though the term autonomous can also be found in the literature.

During the multi-criteria evaluation, the expressions, objectives, criteria and attribute have been used to describe and evaluate characteristics of the power supply alternatives.

Hereby the term objective represents what the decision maker or committee considers important for the decision outcome. The criterion expresses a characteristic according to which an objective can be evaluated, whereas the attribute represents a measurable index or unit for a certain criterion.

Section 4 Section 5 Section 6 Section 7 Section 8


need Objectives Alternatives Consequences


Normalized Utility Values

Overall Utility

Weights Utility Curve



1.4 Conditions

This study has been made under certain conditions, which exert a significant influence on the outcome of the multi-criteria evaluation process. To begin with the definition of the small cocoa production unit, as described closer in Section 2.2, set the overall energy need. Furthermore, this evaluation is based on a 10 h/day production rate with the night hours in standby mode. This production rate could favour solar alternatives because of the time of the energy need corresponding to the time of solar irradiation (daytime).

Industrial sized cocoa processing plants normally operate around the clock. To verify this possible advantage, two further production scenarios have been analysed in Section 9. The annual production has been associated with a 5-day week or 260 productive days/year. The remaining days of the year have been considered as standby time with a corresponding standby energy need. The fictive installation site has been chosen to be located in southern Côte d´Ivoire. Apart from local prices for energy and fuels, this also has an impact on the daily, monthly and annual solar energy input.

1.5 Assumptions and estimations

This thesis includes different assumptions and estimations. In some cases, exact data have not been available and an extended research for this thesis would blow up the content too much. For example, this concerns the values of Carbon Dioxide emissions during the system lifetime. The used values have been obtained from different sources and in some cases averages or estimations based on these sources have been used. In the case of installation costs of equipment, estimations based on European conditions have been used for all alternatives. This has been done since reliable information about local installations is not available. In regard to the use of different heat sources for some processing equipment, the eventual necessary adaption of the machinery has been assessed as cost neutral. On one hand, it is difficult to estimate such a cost without diving into the machinery construction and on the other hand this cost has been assumed to be unimportant in the complete financial context.

The assessed lifetime of the power supply equipment is based on available approximate figures. These figures have different origins such as reports or statements from suppliers. When it comes to the reliability and energy yield of solar power an approximate daily and hourly output has been calculated based on a solar yield simulation (see Appendix VIII). The result of the simulation is a monthly energy yield from one PV-panel. Based on this average, daily energy deficits have been calculated and transformed into a generator running time or an amount of energy from a backup- system. By this the influence of temporary shadow has only been considered in the long run average irradiation level. Due to the use of a battery energy storage or a permanent power grid connection the impact of temporarily shadow has been neglected.

The utility curve, used for the assessment of utility values of consequences, represents the preferences of the decision maker. Within this analysis all attributes have been assigned a linear utility curve, which is motivated in Section 8.1. This assumption of a decision maker´s preferences comes along with other assumptions that are related to the lacking decision maker during the evaluation process. Within the different sections any made estimations have been explained specifically.



1.6 The Author

I have been active within the cocoa processing industry for more than 10 years and have been able to collect very interesting and diverse experiences during this time. As an engineering manager, I have been responsible for single machinery developments as well as for the planning and execution of complete processing plant projects.

Within most projects, the issue about an adequate energy supply in its various forms has played an important role, but still a wider perspective with an analytic approach for selecting a proper energy supply system has been missing. Without correct cooling the large grinding machines either get too hot and burn the cocoa liquor or get too cold and get blocked. Without proper heating of product piping and storage tanks the product builds up and eventually blocks the product line.

In most cases, traditional solutions are being preferred without evaluating hybrid solutions with renewable alternatives. As many projects have taken place in origin countries, mainly West-Africa with local investors, I have had the opportunity to get to know the conditions that exist at such an installation site. This experience has enabled me to write about this topic in the first place and many notes and remarks have their origin in personal experiences. This background has also enabled me to get an idea of the possibilities and problems associated with a small local processing unit. One of the major issues concern the energy supply system due to its share of the processing costs.



2 The Cocoa Industry

Even if cocoa has been grown in large scale since the sixteenth century it was not until the twentieth century that it spread to Africa and the industry, as we know it today, started to take shape. The today’s largest cocoa bean producer, Côte d´Ivoire, is really a newcomer seen in a historical perspective and today it supplies about 40 % of the world’s cocoa.19,20

2.1 The Cocoa Value Chain

The value chain of cocoa is dealing with different issues related to terms as sustainability, traceability and fair-trade.21,22 These issues have been growing more important over the last decade due to increased environmental awareness and the financial situation of cocoa farmers. On the consumer-side a growing interest in this topic can be observed by the steadily increasing demand for certified cocoa products.23,24,25 On the producer’s side, we find different international projects and reports by international organizations, national cocoa societies and governments that deal with these issues. The focus of most such projects is to ensure, stabilize and increase the income of the cocoa farmers.

Of the total cocoa production, about 72 %26 comes from Africa with a typical cocoa farm being a small-scale family run farm with a size of 2-5 hectares and an annual yield of about 350 – 650 kg/ha.27,28,29,30 With the cocoa farmers getting older combined with a low income, not only the economy of millions of families is threatened, but also the cocoa supply as such.31,32 Figure 2 below shows the value share of each step within the chocolate value chain as a percentage of the total value added. With this in mind, the importance of improving the value chain of chocolate in favour of the cocoa farmers becomes more obvious.

19 International Cocoa Organisation, 2017

20 Encyclopedia, 2008

21 International Cocoa Organisation, 2007

22 Capelle, 2008

23 Felperlaan et.al., 2010, p.14

24 International Cocoa Organisation, 2006

25 Cocoa Barometer, 2015, p.19

26 International Cocoa Organisation, 2015, p.6

27 International Cocoa Organisation, 2007, p.2

28 World Cocoa Foundation, 2014, p.2

29 United Nations, 2008, p.16

30 The World Bank, 2012, p.4

31 Cocoa Barometer, 2015, p.3

32 There is an estimated no. of 14 million cocoa producers worldwide with some 90% of the world cocoa coming from family run smallholdings. Source: United Nations, 2008, p.16.


8 Figure 2. The cocoa value chain with the value share of each step noted as a percentage

of the total (Cacao Barometer 2015).

One possibility of adding more value to a certain geographical point in a value chain is to further refine the product at that point. In means of cocoa, the state of the product at the cocoa farm is a fermented and pre-dried cocoa bean. By applying the idea of further refining for adding value,33 this would imply one or more additional processing steps. For cocoa that could include, drying, de-hulling, alkalizing, roasting, de- bacterizing, grinding, pressing and powder-milling.34 A possibility to realise this is the use of an SSCPP, owned and operated by farmer cooperatives. This idea is the trigger of this investigation and as a result the starting point for this analysis is a hypothetical SSCPP for a farmer cooperative in Côte d´Ivoire. The refining end-stage for this unit is the producing of cocoa liquor, filled and blocked in 25 kg blocks. This means that out of the previously mentioned eight processing steps, five are included in the SSCPP:

drying, de-hulling, roasting, de-bacterizing and grinding.

By the approach of SSCPP a partly decentralised cocoa refining operation could help to add more value at the point where it is needed the most. A successful implementation could lead to an increased popularity of the cocoa farming, increased incomes for farmers and an improved social situation35 within farming areas.

2.2 Small Scale Cocoa Processing Plant

The term “Small Scale Cocoa Processing Plant” in this case refers to a small unit for the processing of pre-dried cocoa beans into industrial cocoa liquor. The cocoa liquor36 is the product, which is obtained when de-hulled and roasted cocoa beans are being fine- grinded. The processed cocoa liquor is the first semi-refined product of the fermented and pre-dried cocoa beans that is traded to a larger extent.37 This is the background of the argument for using a SSCPP to increase the value share at the geographic origin of the cocoa value chain. The processing steps of the exemplified small-scale unit contains the same steps as a full-scale plant, even if the machinery may be slightly different and easier to handle.38 The different processing steps are visualized in Figure 3 below, where grinding is the final refining stage and blocking, preceded by tempering, is necessary for handling, packaging and transporting reasons. The definition “blocking” means to

33 The value chain of cocoa is a complex economic structure that depends on politics, trading market, yield etc. In this study the intention has not been to analyse the cocoa value chain. Instead the assumption has been made, that a processing of a good increase the value added at the point where this processing takes place.

34 This list does not claim to be a complete processing description but is intended to give an impression of the different refining stages. For example, a primary cleaning stage is missing since this assumedly can be carried out by hand at this throughput rate.

35 By this we refer to the access of electricity and enhanced infrastructure as a result of the local processing plant.

36 Cocoa liquor has nothing to do with alcohol but is rather the liquid state of cocoa which occurs when the high-fat cocoa kernels are being grinded.

37 Other tradeable semi-finished product may be roasted nibs (cocoa kernels).

38 Based on personal experience.

Farming Transport Processing Manufacturing Retail

6.6% 6.3% 7.6% 35.2% 44.2%


9 fill up and let the cocoa liquor solidify in cardboard boxes with plastic inlays. In Section 4.2 the different processing steps and their respective energy needs have been described more in detail.

There is neither a common standardized size for an SSCPP nor is there a widespread used installation type in cocoa growing countries. Furthermore, as mentioned in the introduction, it is an experimental approach to possibly change the imbalance of the cocoa value chain. Nevertheless, the expression Small-Scale Cocoa Processing Plant is justified and known due to different trends within the cocoa and chocolate industry. Due to an increased trend of artisanal and small-scale chocolate manufacturing many machinery producers offer different small-scale production alternatives.39 However, the capacity and machinery range varies considerably between different manufacturers, which is reflected in price as well as in energy consumption. This makes the definition of the exemplified SSCPP important, and in this case the defined capacity is an output of 100 kg/h of liquid cocoa. Due to weight losses40 during the process this reflects a bean input capacity of approx. 120 kg/h. The liquid cocoa can then be traded either as cocoa blocks of 25 kg or as liquid mass. In Table 1 below the other defining characteristics of the SSCPP are listed. In Appendix I a fact sheet of one type of a small- scale cocoa processing plant is attached.

Figure 3. The processing steps from cocoa bean to cocoa liquor (Fincke et.al. 1965)41. Table 1. Specifications of the small-scale cocoa processing plant.

Technical Specifications –Small-Scale Cocoa Processing Plant

Mode of operation: 10 h/day; 260 days/a

Output: 100 kg/h, Cocoa Liquor

Energy consumption: 40-50 kWh/100 kg42

Necessary external sources: electric power / fresh water

39 Foxwell, 2015

40 The loss of weight occurs due to different reasons in different processing steps: the removal of stones and other impurities within the cleaning step, the removal of shells in the de-hulling and the removal of moisture during drying and roasting. The caused weight difference can be approximated to 20%.

41 The displayed order between de-hulling and roasting may be reversed depending on if a beans roasting or a nibs roasting process is used.

42 Average energy consumption. The consumption varies slightly between the different alternatives depending on the energy sources. For example, the use of LPG for heating applications has a different energy efficiency than electrical heaters. See also Section 4.2.

Drying De-hulling Roasting/ Grinding Storage Blocking Storage Debact.



3 Methods

In this section, the used methods will be named and briefly explained. Made assumptions, for example in regard to power supply characteristics, are also mentioned even though a more detailed description and explanation will follow under the section specifically dealing with the concerned power supply alternative or evaluation criterion.

In those cases, there are references to the sections where this information is available.

When it comes to the energy analysis, the “background” data can be found in appendices, also referred to at the relevant place. Section 3.3 describes how and where information for the thesis has been collected.

3.1 Multi-Criteria Decision Analysis

The importance of the use of MCDA (Multi-Criteria Decision Analysis) has increased since its introduction and is today used in many areas to help solving decision problems.

Over the years many different MCDA methods have been developed each with its respective typical features, advantages and disadvantages.43 Important developments for the solving of multi-criteria problems are the works by Keeney and Raiffa in the mid 1970´s.44

MCDA is, as the name indicates, used to support the solving of decision problems45 containing multiple criteria. If a problem contains at least two, if not more, alternative solutions, these are evaluated according to each criterion and provided with a consequence (i.e. score, value, ranking or monetary amount) according to how a criterion is met. The consequence of an alternative solution according to a specific criterion may be of a normative or a descriptive nature depending on the criterion characteristics. In the following section, we will have a look at how a multi-criteria decision problem can be solved.

3.1.1 MCDA Approaches

Within this thesis the structure has been inspired by the “PrOACT”46 (Problem, Objectives, Alternatives, Consequences, Trade-offs) method and for the evaluation of alternatives, with its utility assessment, elements of “MAUT” 47 (multi-attribute utility theory) have been used. A common systematic of most MCDA-approaches is the means of “divide and conquer”, meaning to split up the decision problem in smaller portions.

In most cases a modelling process can help to identify key elements that are helpful, or even essential, to find a satisfying solution of a given decision problem. In PrOACT the initial modelling steps include to carefully consider the decision problem (is the problem correctly formulated), to formulate the objectives that are to be met, to create alternatives and to describe the consequences of the different alternatives. In Figure 4 below, these initial steps are presented in a schematic way. The subsequent text describes how MAUT and the additive utility function have been used within this study.48

43 To read more about MCDA, see, e.g., Greco et.al., 2016

44 Keeney & Raiffa, 1976

45 The “problem” could also be named “opportunity”, however the term problem is undoubtedly the most common expression in this context. Hammond et.al., 1999, p.17

46 Hammond et.al., 1999

47 Clemen & Reilly, 2014, p.717

48 Clemen & Reilly, 2014, p.720


11 Figure 4. Schematic representation of the four first modelling steps as used within


To formulate the problem may seem trivial, but in some cases the initial formulation of the problem might be wrong, focusing only on a limited part of the problem or on a resulting symptom. To get this right it can sometimes be helpful to know one’s objectives before dealing with the formulation of the problem, which is why the above displayed order of the analysing steps is controversial.50 To some extent or maybe even exclusively this will depend on the decision context. In the current study, with a choice of a power supply system, we assume the problem to be known as an external requirement of energy. Based on this, the order in Figure 4 will be kept as shown and the formulation of objectives will follow that of the problem. The objectives should describe what the decision maker wants to achieve by making the decision in the first place. Consequently, the formulation of objectives is a subjective act that depends on personal fundamental values or the values of an enterprise. The expression value in this context refers to the fundamental goals and wishes of a person, organization, enterprise etc. (Later on, the term value will also be used for naming a numerical value).

In the literature, the objectives are divided into fundamental and mean objectives.51 For a decision analysis, the fundamental objectives are the ones that should be kept in focus. However, in some cases an easy to measure mean objective can replace a fundamental objective.52 To distinguish fundamental from mean objectives is not always easy and sometimes it could be helpful to use a series of questions in order to reach the core of an issue.53 By asking oneself why a certain objective is important, the answer will vary from, helping to achieve something else (means objectives) or, for fundamental objectives: it is important just because it is important. While fundamental objectives are organized in hierarchies, the mean objectives are organized in networks.

Figure 5 shows an example of an objectives hierarchy for evaluating different power supply systems. Depending on the circumstances and the decision maker such a hierarchy could be composed and appear differently. For example, an objective

“minimizing risk of accidents” could play an important role for a big power plant.

49 Hammond et.al., 1999, p. 5

50 Clemen & Reilly, 2014, p.8

51 Clemen & Reilly, 2014, p.49

52 Clemen & Reilly, 2014, p.51

53 Clemen & Reilly, 2014, p.51

Alternatives Consequences Problem Objectives


12 Figure 5. An objectives hierarchy for a power supply system.

Once the objectives are known, different alternatives to reach the mentioned objectives are formulated. The list of alternatives should be created with an open mind in order not to limit the possibilities and leave out potential solutions. On the other hand, too many alternatives might make the subsequent evaluation unnecessarily complex.

Instead, in accordance with an iterative view on the decision analysis, missed out alternatives can be added in a second review. With the help from the previously defined objectives, suitable criteria need to be found according to which the alternatives can be expressed in terms of consequences. The nature of the consequence depends on the chosen attribute to represent a certain criterion. Depending on the objective and ultimately on the attribute, the resulting consequence might be a numeric value, a judgement or a description.

Up to the point of the description of consequences, many methods for solving multi- criteria decision problems follow the same approach. At this point the intermediate outcome is a so-called consequences table.54 The consequences table includes all alternatives, the evaluation criteria and the consequences of the considered alternatives related to criteria. In Table 2 below a consequence table of a simple example of selecting a power supply system is shown. At this place, it is worth to mention that the criteria that are supposed to reflect the objectives should be chosen in a way that attributes can be found, that can be measured or otherwise validated in a way that makes sense for the decision. In this context, the attribute is the explaining scale according to which a consequence is expressed. In the consequence table below the attributes are: cost measured in Euro [€], reliability, as the availability of power over time, expressed in percent [%] and pollutions, as unwanted summarized emissions, measured in g per generated kWh [g/kWh] in a lifecycle approach. The consequences below represent descriptive values.

54 Hammond et.al., 1999

Minimize emission

of GHG

Minimize environmental


Minimize toxic emission

Minimize emission of CO2

Minimize emission of other


Maximize reliability

Minimize power outages

Minimize outage


Minimize overall cost

Minimize investment

Minimize annual


Maximize lifetime

Find the best power supply alternative


13 Table 2. Consequence table for a decision problem on the choice of an energy


Alternative Price

[€] Reliability

[% availability] Pollutions [g/kWh]

Diesel generator 25,000.- 95 1,000

Wind turbine 50,000.- 65 150

Solar panels 150,000.- 75 100

In this example, we have kept it simple and assumed that the criteria can be clearly measured for all alternatives. In regard to “Pollutions” for example, a clearly defined list of substances that belong in this category are presumed to exist.55 Within this thesis slightly different criteria have been used as is explained in Section 3.2.

3.1.2 Evaluation of Consequences

As the consequences of different decision alternatives have been described, the alternatives can be evaluated related to each other. As mentioned earlier, this is where many MCDA-approaches differ. The PrOACT-method begins the evaluation of consequences by looking for dominated alternatives, where an alternative A is dominated when another alternative B is better than A in at least one consequence and at least as good as A in the other. In the example of Table 2 there is no alternative being dominated by another. Instead we are facing three different alternatives that have different advantages and disadvantages. The PrOACT method deals with this by making Trade-offs, which is done by applying the approach of even swaps.56 Within this paper, elements of MAUT and more specifically an approach with an additive utility function will be used to evaluate the alternatives and their consequences.

By recalling the consequences in Table 2, we will have a look at the example of a power supply system choice by using an additive utility function. For an evaluation based on an additive utility function we start by assessing the different consequences as utility values. A utility value should reflect how well a consequence meets an objective in the eyes of the decision maker. This step could therefore be called utility assessment as the consequences are being reformulated onto utility scales. Mostly, a utility scale is ranging from 0 – 1. How this assessment is done depends on the decision maker´s utility function for the different consequences. The utility function describes the preferences of a decision maker and could be seen as a translating tool to turn a descriptive consequence into a normative utility value according to a specific decision maker and decision context.57 The utility value denotes how “good” a consequence is, or how well a consequence meets the preferred decision target. The shape of the utility curve, therefore (linear, exponential etc.) reflects how the utility value changes as the consequence changes. In the case of a linear utility curve the worst consequence is represented by a utility value of zero and the best consequence by a utility value of 1.

In between these end points the utility value is increasing linear from 0 to 1.58

55 Such a listing would also need to include a definition of how hazardous the different substances are. Typically, this could be defined by a recommended critical value or concentration.

56 Hammond et.al., 1998

57 Apart from descriptive numerical consequences also other types of consequences, such as colours or a describing text can be turned into utility values.

58 This does not always have to be the case. In the case of pollution, for example, it could be that a change from 100 to 490 g/kWh is relatively uncritical for the decision maker, whereas a regulation law makes the step from 490 to 500 g/kWh highly relevant. In that case the range from 100 to 490 could be linear, followed by a great drop in utility value as the 500 mark is


14 To return to the example in Table 2 we assume that the decision maker has considered all attribute values, with the given range of consequences, to have a linear utility function. In such a case, with numerical values, the consequences can be assessed as utility values by using Equation 1 below.

𝑈, =(( , , )

, , ) Eq. 159

X1,A is the numeric consequence for alternative A and criteria 1 and U1,A is the resulting utility value. This will result in a utility scale ranging from 0 to 1, where 0 is the worst utility value and 1 the best. After applying this formula, the consequences in Table 2 can be expressed with the utility values as shown in Table 3 below.

Table 3. Utility values of the consequences caused by different power supply alternatives.

Alternative Price Reliability Pollutions

Diesel generator 1 1 0

Wind turbine 0.8 0 0.94

Solar panels 0 0.33 1

At this point the consequences are all expressed on utility scales with a range from 0 to 1. The next step to evaluate the alternatives is to create an overall utility score for each decision alternative. As this analysis uses the additive utility function to obtain the overall utility, this overall score is obtained by summarizing the individual utility values. This is done by applying weight coefficients for the different utility functions according to the preferences of the decision maker. For a utility scale ranging from 0 to 1 the weight coefficients are often assessed in a way that they sum up to one. The assessed weight coefficients are incorporated in the overall utility according to Equation 2 below.

𝑈 = 𝑘 ∗ 𝑈, + 𝑘 ∗ 𝑈 , + 𝑘 ∗ 𝑈 , Eq. 260 where k1, k2 and k3 represent the weight coefficients for the utility functions of the different attributes. Ux,A is the normalized utility value of consequence x for alternative A.

As mentioned, the assigning of weight coefficients is a subjective process done by the decision maker according to his or her preferences. Concluding, the weight assessment may very well differ significantly between different decision makers.

Furthermore, when assessing the weights, it is important to not only focus on the criteria and their attributes, but to also consider the actual consequence values and their respective ranges between the worst and the best consequence. This is important since a small difference between the worst and the best consequence of a specific attribute may make this irrelevant for the further evaluation. At the same time a seemingly irrelevant attribute should not be removed from the evaluation, since possible later inclusions of alternatives may drastically change consequence ranges and thus their importance. The weight assessment is a tricky procedure and there are different methods

exceeded due to the consequences (penalty payments or similar) of exceeding a regulating limit.

After the 500 g/kWh the curve might return to linearity up to the next regulation value depending on the regulation circumstances.

59 Clemen & Reilly, 2014, p. 722

60 Clemen & Reilly, 2014, p.721


15 available in order to assist the decision maker to do this.61 Different methods have different qualities when it comes to biases and to incorporate the consequence range.

This means that also the method, which is being used to assess the weight coefficients, has an impact on the result. In this study the weight coefficients will be assessed by applying the swing weighting method.62 Swing weighting method

To describe the swing weighting method, an imaginary decision maker is used for the example in Table 2 for choosing a power supply alternative. The first step of the swing weighting method is to create an “all-worst” benchmark, which is a hypothetical worst- case alternative, incorporating the worst consequences for all criteria. This will be the reference point from which the consequences of one attribute at a time “swings” from the worst to the best by considering hypothetical alternatives that reflect the different best consequences. Table 4 below shows the resulting compilation of swung attributes in the first column. The next step is to rank the hypothetical swings from the best to the worst, according to the personal preferences of the decision maker. In the example, we assume the reliability to be valued as the most important, followed by pollution and price. As mentioned previously, it is important that the priorities are not set only by valuing the criteria as such, but also to consider the range of the consequences for the different criteria, i.e. the swing amplitude. Having filled out the ranking numbers, the rates of the best and the worst swings are set to 100 and 0 respectively.

Table 4. Swing-weight assessment table for choosing a power supply system.

Swing attribute Consequences Rank Rate Weight

Benchmark 150,000,- € / 65 % / 1,000 g/kWh 4 0

Price 25,000,- € / 65 % / 1,000 g/kWh 3

Reliability 150,000,- € / 95 % / 1,000 g/kWh 1 100

Pollution 150,000,- € / 65 % / 100 g/kWh 2

To assess the other rates, we use swings. Principally this is done by relating the imaginary satisfaction of swinging a consequence from the worst to the best related to the satisfaction of swinging the top-ranked attribute from the worst to the best. In the example above we could compare to swing the price from 150,000.- to 25,000.- with the swing of the reliability from 65 % to 95 %. Hereby the task is to value how much less satisfaction the swinging of price would yield compared to the swinging of the reliability. Or, how many percent of the satisfaction achieved by swinging the reliability from worst to best would the swinging of price achieve? In the example we assume the imaginary decision maker to rate the swinging of price to be worth 40 % and the swinging of the pollution level to be worth 80 % of the swinging of reliability. Even if the percentage rating is presented very abruptly here, it is important to point out that it should not be an intuitive rating. In an actual example with a power supply system, surrounding conditions, such as legal regulations or cost factors, influence how the decision maker value these swings. A legal pollution limit may prohibit production if this is to be exceeded, or a power outage might result in lowered profitability due to production losses. Such known data can be helpful for the swing weighting process as

61 Such as: pricing out, lottery weights and swing weighting method. Clemen & Reilly, 2014, p.730 ff.

62 Clemen & Reilly, 2014, p.731


16 it can support the valuation of consequences against each other in an analytical way. To continue with the example, the assessed percentage rates for price and pollutions of 40

% respectively 80 % are filled out as shown in Table 5 below. To obtain the corresponding weights for the consequences, these are calculated by using Equation 3, where kx is the weight factor, rx is the rating value for attribute x and ri is the rating value of attribute i with a total of n attributes. In Table 5 the resulting weights have been noted.

𝑘𝑥= 𝑟𝑥( ) Eq. 3

Table 5. Completed swing-weight assessment table for the choice of a power supply system.

Swung attribute Consequences Rank Rate Weight

Benchmark 150,000.- € / 65 % / 1,000 g/kWh 4 0 0

Price 25,000.- € / 65 % / 1,000 g/kWh 3 40 0.18

Reliability 150,000.- € / 95 % / 1,000 g/kWh 1 100 0.46

Pollution 150,000.- € / 65 % / 100 g/kWh 2 80 0.36

With the weights set, the overall utility for the different alternatives can be calculated according to Equation 2. The resulting overall utility of the alternatives are as listed in Table 6. As can be seen, the assessment of weight coefficients according to the preferences of the hypothetical decision maker results in the diesel generator being the recommended first choice.

Table 6. Overall utility for energy systems.

Alternative Overall utility

Diesel generator 0.63

Wind turbine 0.48

Solar panels 0.51

3.1.3 How MCDA has been used in this thesis

In this paper, MCDA has been used to demonstrate how different power supply alternatives can be evaluated according to a number of objectives. When assessing the overall utility, the additive utility function has been used. To be able to perform the evaluation with utility values, the preferences of a hypothetical decision maker have been created. With help from these preferences the utility functions of the attributes and the weight coefficients have been assessed. The weight coefficients have been assessed by using the swing weighting method. To simplify the reasoning around priorities and assumptions made by the hypothetical decision maker, the preferences have been given a strict financial character. The economic reasoning by the hypothetical decision maker during the weight assessment can be found in Appendix III. The resulting overall utility may subsequently be analysed by means of a sensitivity analysis.

Even if there are different computer-based tools available for the evaluation of decision alternatives, this part has been done manually within this study. The reason for this is to obtain a high level of transparency and to provide an easy access to the different stages of the evaluation process for possible further analyses.



3.2 Energy Analysis

The following section describes how the energy analysis has been performed in terms of the energy demand of the SSCPP as well as the energy supply alternatives and the description of their consequences. It begins by describing how the consequences, according to certain attributes, have been assessed. For this, the initial sub-sections describe the chosen attributes and how they have been calculated for the different power supply alternatives. Section 3.2.4 explains how the energy yield of solar based systems has been assessed and Section 3.2.5, deals with the energy demand of the SSCPP. The attributes, that have been used are:

Levelized cost of energy (LCOE)

Carbon footprint of energy production (CFOE) Loss of load hours (LOLH)

More of the background for choosing these attributes is described in the following subsections and in Section 5. For common, single modular power supply systems, such as the power grid option or the stand-alone generator, information on the above- mentioned indicators have either been available directly or have easily been possible to calculate based on available information. For combined, hybrid systems and especially for the solar alternatives these values have had to be calculated, partly based on assumptions.63 For all alternatives the energy need of the SSCPP has been considered instead of the energy generated where this applies. The reason for using the energy need of the SSCPP instead of the energy generated is discussed further in Section 3.2.1, but principally it is based on the approach of considering the useful energy. For the index LCOE, where a lifetime is included, the lifetime of the power supply system has been used. The used approach is described in the following sub-sections.

3.2.1 LCOE - Levelized Cost of Energy

The Levelized Cost of Energy (LCOE) is an economic index for power supply systems that considers the initial investment, the annual cost, the system lifetime and the generated energy. It also integrates the inflation rate and the opportunity cost of capital (OCC).64 The resulting unit of the LCOE is the cost per generated energy unit expressed in €/kWh. In this study, a small exception to the standard procedure for calculating the LCOE has been made. As already mentioned in Section 3.2, the amount of energy, used in the calculations, is based on the energy need of the consumer, in this case the SSCPP, instead of the energy generated by the power supply system. This approach has been used due to the solar alternatives that might generate more energy than can be used.

When it comes to power supply systems which include renewable energy it can cause a remarkable difference if the total generated energy or the total needed energy is considered... Due to power reducing effects such as dust and climate conditions, the off- grid solar plant is highly over-dimensioned when considering the peak power of the solar plant related to the peak power of the SSCPP. By that the solar plant can generate more power than the SSCPP can use during sunny weather. This is partly compensated for by the battery storage, but only to an extent which gains the energy need of the SSCPP. In this particular case there has been no consideration of a possible feed-in tariff or storage of surplus energy. Therefor any generated surplus energy is of no use. To make the LCOE index valid for this case, the annual energy need, instead of energy generated, has been used. In regard to the considered time frame for the index, the system lifetime of the power supply system has been applied. To calculate the LCOE index of a single power source the Equation 4 has been used, where Cinv is the initial

63 As for example the monthly average solar irradiation, which is based on climate statistics.

64 The expected return from an investment that could have been done instead of the current one.


18 investment, cj the annual cost, Et the average annual energy demand of the SSCPP, n the expected lifetime of the power supply system in years, g the inflation rate and OCC the opportunity cost of capital.

𝐿𝐶𝑂𝐸 = 𝐶 + ∑ ∗( )

( ) / ∑ ∗( )

( ) Eq. 465

For hybrid systems, with multiple energy sources, the combination of the single LCOE values has been made by applying their respective percentage rate of the total generated energy. See Equation 5, with a total of p energy sources and an energy supply share of fi percent.

𝐿𝐶𝑂𝐸 = ∑ (𝑓 ∗ 𝐿𝐶𝑂𝐸 ) Eq. 5

Whereas the investment cost for the different alternatives is relatively indisputable, the annual cost is not. This is due to the approach to only consider useful energy. Since no further trade of energy to external nets or consumers has been considered only the energy that can be used by the SSCPP is accounted for. Due to HES alternatives, solar irradiation differences, power failures and different backup systems the average energy need of the processing plant alone does not include enough information to define the annual cost. To make a comparison between LCOE values relevant, the annual cost must consider the expected split of energy on the different energy sub-systems.66 Consequently, the considered useful annual energy generation is based on results from the calculated hourly solar energy (for alternative 3 and 4) and the expected downtime of sub-systems. In line with this the generated useful energy, thereby reflecting the energy need of the SSCPP, of a sub-system is calculated according to Equation 6. For the back-up generator, the relevant amount of supplied energy is based on the LOLH of the main system as shown in Equation 7. Since the backup generator is not expected to fail when it is not operating, its own LOLH value has been extracted from the expected active running time. An exception concerns the alternative with an LPG-burner for heat applications. In this case, a system efficiency rate for the heating system has been used, with the result that the annual cost accounts for the gross amount of energy, necessary to deliver a useable net amount of energy according to Equation 8. Apart from this, the LPG-based energy supply has also been considered with the same downtime as the main power supply system. This is due to the fact that the processing plant cannot operate without electric power, which makes heating energy useless during this downtime.

𝐸 = 𝑃 ∗ 𝑇 − 𝐿𝑂𝐿𝐻 + 𝑃 ∗ 𝑇 − 𝐿𝑂𝐿𝐻 Eq. 6

In Equation 6, Esu is the annually generated useful energy, Psu the average generated useful power during production time, Tp the annual production time and LOLHp the annual power downtime during production time. The index stby denotes the responding standby specifications.

𝐸 = 𝑃 ∗ 𝐿𝑂𝐿𝐻 , + 𝑃 ∗ 𝐿𝑂𝐿𝐻 , − 𝑃 ∗ 𝐿𝑂𝐿𝐻 Eq. 7

65 Bortolini et.al., 2015, p.1029

66 Note: the installed percentage of different energy sources may differ from the annual average percentage. This is the case with the solar powered alternative with a back-up generator, where the estimated generator running time has been considered.


Related documents

46 Konkreta exempel skulle kunna vara främjandeinsatser för affärsänglar/affärsängelnätverk, skapa arenor där aktörer från utbuds- och efterfrågesidan kan mötas eller

Generally, a transition from primary raw materials to recycled materials, along with a change to renewable energy, are the most important actions to reduce greenhouse gas emissions

Both Brazil and Sweden have made bilateral cooperation in areas of technology and innovation a top priority. It has been formalized in a series of agreements and made explicit

The increasing availability of data and attention to services has increased the understanding of the contribution of services to innovation and productivity in

Generella styrmedel kan ha varit mindre verksamma än man har trott De generella styrmedlen, till skillnad från de specifika styrmedlen, har kommit att användas i större

Parallellmarknader innebär dock inte en drivkraft för en grön omställning Ökad andel direktförsäljning räddar många lokala producenter och kan tyckas utgöra en drivkraft

Närmare 90 procent av de statliga medlen (intäkter och utgifter) för näringslivets klimatomställning går till generella styrmedel, det vill säga styrmedel som påverkar

Swedenergy would like to underline the need of technology neutral methods for calculating the amount of renewable energy used for cooling and district cooling and to achieve an