Supplier Selection for a Biomethane Liquefaction Installation

(1)

Supplier Selection for a Biomethane Liquefaction Installation

Based on the Analytic Hierarchy Process

(2)

Preface

This report is the representation of my master thesis, the final part of the master degree program of

Technology Management at the University of Groningen. I conducted this research at Cirmac International in Apeldoorn, where I started my internship at the start of July 2010. During an internship of six months, I carried out an investigation into the area of supplier selection using the Analytic Hierarchy Process.

First of all I would like to thank everybody at Cirmac International for their help during the research. The helpfulness of everybody made the gathering of information a lot easier. A special word of thanks goes to my supervisors Michel Knoef and Geurt Aalderink for providing me with the necessary information, guidance and feedback on my thesis.

Also I want to thank my first supervisor Dr. L. Zhang, for the initial directions and the given feedback that helped me to develop this report. Also, many thanks to dr. X. Zhu for reading this report and the constructive feedback that was given.

This Master’s thesis marks the end of my MSc Technology Management at the University of Groningen. However this thesis could never have been completed without the everlasting support and feedback of my friends and family during these long and turbulent months. For everybody that has been waiting for so long I would like to promise that seven and a half year of studying is enough. Now the time has come to do something useful with all this knowledge.

Niels Rop

(3)

Abstract

Cirmac International is a world-wide operating company in the field of gas production and gas treatment systems for the petrochemical and chemical industry, refineries and other industries. It offers installations for nitrogen generation and biogas upgrading. Recently Cirmac noticed an emerging trend for the liquefaction of biomethane and wishes to offer an installation for the upgrading and liquefaction of biogas to their customers. However Cirmac has no experience with the liquefaction of gasses and is in search of a competitive liquefaction process that matches the characteristics and production size of a typical biogas installation. This research explores the main suppliers for these liquefaction installations and develop a procedure to make the optimal choice of supplier.

To be able to reach this goal a procedure for the comparison of the different suppliers has to be developed. The selection of a supplier is determined by both quantitative as qualitative factors as well as a trade-off between costs and benefits. To structure the problem and select the optimal alternative the Analytic Hierarchy Process (AHP) will be presented and adapted for use in this case. First the literature behind AHP will be presented and a method for a separate cost calculation will be presented. Next this method will be adapted and applied to the case of Cirmac International.

The adapted procedure will consist of three parts. In the first part the benefits of the alternatives are compared using the Analytic Hierarchy Process (AHP). The second part consists of a cost comparison of the production costs using the different alternatives. In the final part these scores will be synthesized to produce a final score and advice on the optimal supplier for a biomethane liquefaction process. Finally, to determine whether the selected processes are competitive a comparison of the selected processes with those of the competitors will be made.

(4)

List of abbreviations

AHP Analytic Hierarchy Process. MADM Multiple Attribute Decision Making

Biogas A combination of methane, CO2 and trace gases released during anaerobic digestion of vegetable waste.

Biomethane Biogas upgraded to at least 98% methane.

LBM Liquefied biomethane: biomethane liquefied by cooling to about -160 oC. Other names are LBG (liquefied biogas), Bio-LNG, PNG (Pseudo Natural Gas) and RNG (Renewable Natural Gas). It this document only the term LBM will be used. LNG Liquefied Natural gas: Natural gas liquefied by cooling to about -160 oC. CNG Compressed Natural Gas: Natural gas compressed to 200 bar.

CBM Compressed Biomethane: Biomethane compressed to 200 bar.

(7)

1. Introduction

In subsection 1.1 the background of the company and its mission will be presented. Subsection 1.2 explains what biogas and biomethane is. Subsection 1.3 will present a general description of the problem and the possible impact on the company. And subsection 1.4 will present an overview of the report.

1.1 Company background

Cirmac International is a world-wide operating company in the field of gas production and gas treatment systems for the petrochemical and chemical industry, refineries, renewable energy companies and other industries.

Cirmac was founded in 1979, when it acquired its first license to manufacture and sell nitrogen generating systems from air based on the Pressure Swing Adsorption (PSA) principle, using CMS (Carbon Molecular Sieve) and a few years later extended its product range with Nitrogen Generators using Membranes. First situated in Gouda, Cirmac expanded to the company it is today, situated at Apeldoorn . In 2006 the company was acquired by the Rosscor Group. Recently Cirmac has been taken over by the Atlas Copco Group, offering the company new possibilities for collaboration and growth.

Cirmac has always been playing an important role in searching for new technologies and applications. As a result, a wide range of new and promising solutions for economical gas treatment was created.

When new environmental rules became effective within Europe to reduce CO2 and CH4 emissions even more,

operators of landfill deposits and biomass digesters were looking for the most economical way to treat and use their biogas sources, in which development Cirmac has taken an important lead. This resulted in the

development of three biogas upgrading technologies and the largest landfill biogas upgrading plant to PNG (Pseudo Natural Gas) in Europe, engineered and constructed by Cirmac. This plant has been successfully in operation since 1990.

1.2 Biogas upgrading

Biogas is a gas produced by anaerobic treatment of wastewater sludge, organic solids, manure and specially grown energy crops. When upgraded to biomethane, it is a sustainable replacement of natural gas for gas grid injection, electricity and heat production (CHP) and vehicle fuel.

Cirmac has been specializing in biogas upgrading systems for over twenty years. A key step in the processing of biomethane is the separation of carbon dioxide (CO2) present in the gas. They supply complete turnkey systems

based on three out of five existing technologies for CO2 removal. This offers a benefit over the competition, as

they’re able to select the best technology for the customer based on the type of biogas sources and biomethane application for the project. Cirmac designs gas upgrading systems based on the following three CO2 removal technologies: Membrane Separation, Vacuum Pressure Swing Adsorption (VPSA) and Chemical

Absorption. The other two technologies available today are Water Scrubbing and Cryogenic Upgrading. The three technologies for CO2 removal offered by Cirmac are explained in Appendix A.

1.3 Problem description

(8)

biomethane has already made a big contribution to meeting the target of 20% renewable energy by 2020 in gas and electricity production. But it also has a potential to contribute to the goal of 10% of the vehicle fuels to be renewable.

Lately natural gas has received more attention as a vehicle fuel, either as Compressed Natural Gas (CNG) or as Liquefied Natural Gas (LNG). Although natural gas is superior to gasoline or diesel in terms of environmental friendliness (Sarkar, 2005) it is still a fossil fuel and will still contribute to global warming. Since biomethane has the same physical properties as natural gas it can easily offer a renewable, carbon neutral alternative, thereby contributing to the goal of 10% of renewable fuel.

At this moment CNG is the most used form of natural gas for fuel. When fuelling passenger cars or light or medium duty vehicles CNG will provide a good alternative. However because of the relative low energy density of CNG it is not efficient if the gas has to be transported with trucks or trains, and it doesn't provide a sufficient range for heavy duty transport. LNG overcomes these problems of CNG because it is 2.4 times more dense than CNG meaning that a LNG vehicle with the same tank size will travel 2.4 times the distance a CNG vehicle will. Also transporting the gas by road or rail will be much more efficient because of the higher energy density and easier handling of liquefied gas opposed to compressed gas (Sarkar, 2005).

Natural gas liquefaction is traditionally done at very large scales near the source of the gas to be able to liquefy at very low costs. Recently more research and development has been directed towards the so-called Small-Scale Liquefaction plants (SSL). These SSL’s provide to possibility to liquefy biomethane at a reasonable cost and production volume to produce Liquefied BioMethane (LBM). LBM can directly substitute LNG in any application and thus provide a renewable alternative.

Although Cirmac is a leading company in the field of biogas upgrading, it has no experience with the liquefaction of gasses. Because of the emerging market for the production of LBM Cirmac wants to find a liquefaction process that matches the characteristics and production size of a typical biogas installation to be able to offer a complete LBM production installation in the future. If Cirmac would not be able to offer such an installation, they would lose potential projects and their leading role in the biogas industry. This would hinder the goal of Cirmac to remain an innovative and leading company in the gas treatment industry. So the goal of the research is to help Cirmac develop a competitive biogas liquefaction installation.

1.4 Report overview

(9)

2. Research design

This section defines the goal and objective of this research with respect to Cirmac International and the research question and sub questions necessary to reach the stated objective. Subsection 2.3 presents and explains the conceptual model.

Cirmac International wants to offer a biogas liquefaction installation to their customers. This research explores the main suppliers for these liquefaction units and develops a procedure to make the optimal choice.

2.1 Research goal & objective

The goal of this project is to help Cirmac develop a competitive biogas liquefaction installation so as to attract their customers.

To achieve the above goal, the research objective is to design a procedure to determine the optimal choice of a liquefaction system for Cirmac.

2.2 Research question

In line with the research objective, the main research question is formulated as follows:

Which liquefaction system is best suited to liquefy biomethane for Cirmac?

To be able to answer the main research question, several sub questions are identified:

 What is the optimal methodology to choose between the different liquefaction systems?

 Who are the available suppliers of biomethane liquefaction systems?

 Which criteria are of importance when selecting a liquefaction system?

 What is the optimal liquefaction system for Cirmac?

 How competitive is the liquefaction of biogas using the chosen liquefaction system?

2.3 Conceptual model

In accordance with the major influencing factors, the conceptual model is developed, as shown in Figure 1 below. It involves the elements that contribute to competitiveness of a biogas liquefaction installation. These elements will now be further explained.

Competitiveness biogas liquefaction installation Technical characteristics Vendor specific charateristics Operational characteristics Costs

(10)

The choice of the elements is based on a research performed by Tam and Tummala (2001), who proposed a division of criteria for supplier selection into four groups. These are used here as elements in the conceptual model. The groups are explained below, and the relevant criteria per group will be determined in section 4.

 Costs: The group of cost factors constitute any financial factor connected to an alternative but also factors like energy usage and the use of spare parts.

 Technical characteristics: The technical characteristics relate to any factor that influences the performance of a biogas liquefaction system. For instance, the system reliability and capacity or the quality of the produced LBM.

 Operational characteristics: Operational characteristics relate to the factors influencing the way the system needs to be operated when in normal production. This might be the ability to run stand alone, or the ease of maintainability.

 Vendor specific characteristics: The vendor specific characteristics include the factors that don’t directly relate to the offered system, but related to the company offering the system. This are i.e. their experience in related products, the quality of support systems or the delivery lead time.

(11)

3. Relevance of framework

This section presents the relevance of the framework of the research. First the group of methods called Multiple Attribute Decision Making (MADM) is discussed. After which the choice for the Analytic Hierarchy Process (AHP) is explained and the method is further explained. Subsection 3.3 and 3.4 will show how AHP can be adapted when costs are a major factor in the decision making process.

3.1 Multiple Attribute Decision Making

The problem described earlier is a problem facing a limited number of alternatives which have to be prioritized on several (possibly conflicting) attributes. For these kinds of problems Multiple Attribute Decision Making (MADM) provides a number of ways to compare the alternatives and find the best one. MADM arises in many real world situations but always aims to rank the given alternatives based on preference information provided by the decision maker. Although MADM covers a large amount of methods and is capable of solving a diverse set of problems, all the problems considered here share the following common characteristics:

 Alternatives: A finite number of alternatives, from several to thousands, are screened, prioritized, selected, and/or ranked. In this case the alternatives are the installations capable of liquefying biomethane in the required amounts.

 Multiple attributes: Each problem has multiple attributes. A decision maker must generate relevant attributes for each problem setting. The number of attributes depends on the nature of the problem. In this case, attributes may be the capital investment, system availability or delivery time. The term ‘attributes’ may also be referred to as ‘goals’ or ‘criteria’.

 Incommensurable Units: Each attribute has different units of measurement. Using the above mentioned attributes the capital investment will be measured in dollars or euro’s, the availability in days per year or a percentage and the delivery time in months.

 Attribute Weights: Almost all MADM methods require information regarding the relative importance of each attribute, which is usually supplied by an ordinal or cardinal scale. Weights can be assigned directly by the decision maker or, as in this case, may be computed when executing the chosen methodology. (Yoon & Hwang, 1995)

(12)

Multiple Attribute Decision Making No information Information on Environment Information of Attribute Dominance Pessimistic Optimistic Standard level Ordinal Cardinal Maximin maximax Conjuctive method Disjunctive method Lexicographic Method Elimination by Aspect Simple Additive Weighting Weighted Product TOPSIS ELECTRE Median Ranking Method

AHP Type of information

available

Salient Feature of

Information Major Class of Method

Figure 2: Taxonomy of MADM methodologies(Yoon & Hwang, 1995)

In this case the information is obtained on the attributes and is of a cardinal scale. Also, the problem deals with multiple criteria or attributes and considers a small set of alternatives. This classification leads to the six methods shown in the lower right box of Figure 2. Next these six methods will be briefly discussed and it will be shown why AHP is the best fitting method and why the other five are not applicable for the presented case. Simple Additive Weighting (SAW) works by simply multiplying the alternative value with the criterion weight and then adding over the criteria. The Weighted Product Method (WPM) works in a similar way but the attributes are connected by multiplication. The alternative values are raised to the power of the criterion weight and then multiplied to give the alternative scores. However both SAW as WPM only assist in calculating the final scores of alternatives and presume that the criteria weights and alternative values are known (Yoon & Hwang, 1995). In the presented case the criteria weights and alternative values are unknown and have to be determined by the chosen method.

(13)

The basic concept of TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) is that the selected alternative should have the shortest distance from the ideal solution and the farthest distance from the ideal solution in a geometric sense. TOPSIS assumes that it is easy to locate an ideal and negative-ideal solution to compare alternatives to (Triantaphyllou et al., 1998). However in this case locating an negative-ideal and negative-ideal solution is not possible so TOPSIS is not usable for this case.

The Median Ranking Method (MRM) is explicitly able to handle qualitative data, where the former methods require qualitative data to be quantified before the analysis. MRM uses rank data to rank the alternatives with respect to the criteria. However, MRM suffers from the same setbacks as SAW and WPM and is therefore not applicable for this study.

AHP uses pairwise comparisons to determine the relative weights of the criteria and to determine the alternative values for each criterion. AHP thereby offers the possibility to calculate the criteria weight, alternative values and synthesize this information into alternative scores. By applying pairwise comparisons AHP is also able to integrate both quantitative as qualitative information into the decision making process. With this information taken into account, AHP is chosen as the method to be used in this research. The next subsection will further explain AHP and present some additions to the method for using it in a group decision environment and to cope with an high importance of cost factors.

3.2 Analytic Hierarchy Process

(14)

Problem 1. Collection of alternatives 2. Determination of minimal demands 3. Determine collection of alternatives 4. Identification of criteria 5. Construction of decision hierarchy 6. Pairwise comparison of alternatives 7. Pairwise comparison of criteria 8. Calculation of priorities of alternatives 9. Sensitivity analysis Advice best alternative Relative preference of alternatives Relative preference of criteria

Figure 3: The nine steps of AHP (Huizingh & Vrolijk, 1993)

1. Collection of alternatives

In the first step a collection of alternatives is comprised.

2. Determination of minimal demands

At the same time the minimal demands for the alternatives have to be determined. Alternatives that don’t meet these demands will be excluded from the process. If necessary the Lexicographic

Semiorder methodology (LS) as proposed by Yoon and Hwang (1995) can be applied. With LS the minimal demand is based upon the best alternative for an important criterion. All the alternatives with a score within a certain range are included while the alternatives outside of the range a excluded from the process.

3. Determine collection of acceptable alternatives

(15)

4. Identification of criteria

After determining the acceptable alternatives the criteria need to be determined on which the alternatives will be compared. These criteria can be on one level or a second (or third or fourth) level can be added to account for sub-criteria.

5. Construction of decision hierarchy

When the acceptable alternatives and the relevant criteria are known the decision hierarchy can be constructed. This hierarchy consist of (at least) three levels, namely the goal, the criteria (and possibly the sub-criteria) and the alternatives. These elements are expressed in a tree-structure. The hierarchy shows the structure of the decision problem and is the basis for the calculations in the next phases.

6. Pairwise comparison of alternatives

Based on the decision hierarchy the decision maker must compare the alternatives pairwise. The alternatives are hereby compared to each other to one of the criteria. The decision maker has to decide which alternative performs better on the criterion at hand, and to what magnitude. The other criteria and alternatives are left out of the decision. Every combination of criteria and pairs of alternatives need to be compared. The way these comparisons are made and the underlying mathematics which are used to calculate the priority ratings will be explained in section 4.

7. Pairwise comparison of criteria

To be able to compare the alternatives in the right way the relative importance of the criteria also has to be calculated. These are calculated in the same way as the relative scores of the alternatives for a criteria. The criteria are compared pairwise and the decision maker has to decide which criterion is more important and to what magnitude.

8. Calculation of priorities of alternatives

After obtaining both the relative preference of the alternatives for each criteria and the relative preference of the criteria to each other, the priorities of the alternatives can be calculated. This is done by a simple linear additive function where the relative preference of an alternative is multiplied by the importance of the criterion and summed over all the criteria.

9. Sensitivity analysis

AHP also offers the possibility of executing a sensitivity analysis. With this analysis the influence of changes in criteria priorities on the alternative priorities can be calculated. Smaller changes in priorities mean more stable and more reliable results.

The process explained above is an intuitive and relatively easy method for formulating and analyzing decisions. The AHP is based on three major concepts, being analytic, hierarchy and process. These concepts will now be further explained.

 Analytic: Simply put, the AHP uses numbers. In holistic decision making no numbers are needed to arrive at a decision, simply choose the alternative that is most desired. However, there are very good reasons why the use of mathematics to understand and/or describe a choice to others is preferred. In this sense of the word, all methods which seek to describe a decision are analytic since they must use mathematical/logical reasoning.

(16)

compare 7 ± 2 items at a time (Miller, 1956). Thus it is vital to use a hierarchy when dealing with complex situations.

 Process: Decisions which are truly important cannot be made in a single meeting, one cannot expect the AHP to counteract this basic human tendency. People need time to think about a decision, gather new information, negotiate if it is a group decision, etc. Thus, any real decision problem involves a process of learning, debating, and revising one’s priorities. AHP is meant to be used to aid and shorten this process through the insights which this method can generate ; it will never replace the overall decision making process. The AHP points to where more information is needed, where major points of disagreement lie, etc. Also, when one goes through the structured process, the final result may not agree with a “gut feelings”. At this point, a decision maker must return to the hierarchy in order to see if true feelings have been misrepresented, or it may be that intuitive feelings will change after considering the problem in detail. This process is unavoidable and is in fact quite healthy; the AHP is meant to aid and not destroy this natural process of decision making.

Therefore, the overall philosophy of the AHP is to provide a solid, scientific method to aid in the creative, artistic formulation and analysis of a decision problem (Harker, 1989).

3.3 Dealing with the importance of costs

In some AHP situations costs can constitute a large proportion of the total weight of the criteria in one single criterion. These costs can be expressed in capital expenditure, total cost of ownership or costs per unit produced. These costs may be calculated using multiple figures but they sum up to one, very important, criterion. Dealing with such an important criterion in the regular AHP method may compromise the results of the method. There are a number of solutions to dealing with costs in the AHP method:

 The original AHP method, according to which the hierarchy is initially constructed so that the criteria include both qualitative and cost factors. In disciplines in which costs traditionally constitute the major key factor, this mixture of factors appears to be problematic, mainly due to the difficulties expected in pairwise comparisons between cost and noncost factors, since costs are likely to be assigned relative weights that are excessively high (Fong and Choi, 2000).

 Benefit-to-cost ratios (Hastak and Halpin, 2000): Cost estimates and evaluation of benefits are performed separately, and benefit-to-cost ratios are then computed for each alternative (the higher the quotient, the better the alternative). The drawback of this method is that it has no capacity for distinguishing between alternatives that exhibit the same benefit-to-cost ratio but vary in the difference between costs and benefits. Hence this method is suitable only for cases in which cost differences between alternatives are relative small.

 Highest level criteria are costs and benefits (Skibniewski and Chao, 1992): This method is similar to the first method, but the highest level of criteria in the hierarchy contains only two groups of selection criteria: costs and benefits. This method suffers from what appears to be a problem that is inherent in AHP for sets containing only two attributes that are to be compared (i.e., only one comparison is required). The major limitation here is that the relative weighting of the costs and benefits cannot be reaffirmed by additional pairwise comparisons with other attributes in the set. (Consistency Ratio computation is not applicable for a two-attribute set).

(17)

For this research the method proposed by Shapira and Goldenburg (2005a) will be used because the first three options all have big disadvantages and a case-study of Goldenburg and Shapira (2005b) showed promising results with their method. In the next subsection this method will be further explained.

3.4 Cost versus benefits

The evaluation of the alternatives is performed by weighting its cost score and its benefits score. The cost score is determined by an extensive cost analysis and the benefit score is determine through the AHP method. The total score representing the weighting process is obtained by a technique similar to AHP-based computation of priority vectors, i.e., it is the sum of two products: the cost weight multiplied by the cost score; and the benefit weight multiplied by the benefit score. Since the total score comprises two components, costs and benefits, the sum of their respective weights is always 1.0. Note that unlike with benefit scores, the higher the cost of an alternative, the lower the cost score. The determination of cost and benefit weights is guided by the following principles.

 If the costs of the alternatives are similar, the decision will be based solely on the benefits. In this case, the weight of the benefits will be 1.0, and that of the costs 0.0. The alternative with the highest benefit score will be the one selected.

 If the cost of the alternative differ significantly from each other (with significance margins defined by the decision maker so that no matter how high the benefits are, they cannot shift the balance towards a higher cost alternative), the decision will be based solely on the costs. In this case, the weight of the benefits will be 0.0, and that of the costs 1.0. The alternative with the lowest cost will be the one selected.

 For any other situation, i.e., the cost difference between the alternatives is neither negligible nor decisively great, the weights will be determined in linear proportion to the difference. In other words, the greater the difference, the bigger the weight of the costs and the smaller the weight of the benefits.

(18)

Benefit score for

alternative i Maximum cost difference permitted

Cost estimate for alternative i

Calculation of weights of costs and soft

factors

Calculation of cost scores for each

alternative

Calculation of total score for each

alternative

AHP Evaluation User input Cost evaluation

Final selection

(19)

4. Supplier selection based on AHP

Cirmac International wants to select a preferred supplier for biomethane liquefaction installations. Currently Cirmac only offers the possibility to upgrade biogas to biomethane, and they aim to offer the complete process of biogas to LBM in the future. The selected biomethane liquefaction process needs to be cost competitive or offer certain benefits to overcome the cost difference. The presented procedure first calculates benefit scores using AHP, then calculates the cost prices of the alternatives after which the scores will be combined using the method proposed by Shapira and Goldenburg (2005a).

4.1 Supplier selection procedure

This procedure for the selection of a supplier for a liquefaction installation is based on an APH analysis and a cost analysis to make the trade-off between costs and benefits. Figure 5 shows the used procedure in more detail. Alternative selection Total score calculation Sensitivity analysis Optimal choice Benefit score calculation using

AHP

- Indentification of the criteria - Construction of the decision hierarchy

- Priority establisment - Calculation of benefit scores

Cost score calculation

- Determine basis for calculation - Determine cost factors - Calculation of costs - Calculation of cost scores

Figure 5: Installation selection procedure

(20)

4.2 Selection of alternatives

The first step in finding the right supplier for a gas liquefaction system is compiling a list of the possible suppliers of gas liquefaction systems in the field. The list is compiled based on a report by the Holland

Innovation Team (Van der Gaag, 2008) and an internet search, and consists of ten suppliers. Of these suppliers four are already known by the company and six are new to the company. The suppliers are then contacted and asked to send quotations for gas liquefaction plants of 400 Nm3/h and 900 Nm3/h upgraded biogas. The numbers of 400 Nm3/h and 900 Nm3/h are chosen to represent two representative cases based on the expectations of Cirmac. The use of two cases offers the possibility to determine whether the size of an installation might influence the choice of supplier. The first case, represented by the 400 Nm3/h installation describes a relative small installation when the supply of biomethane may be limited. The second case, of 900 Nm3/h , represents a large installation. This will in general lead to lower liquefaction costs, but the supply of enough biomethane and demand for LBM has to be secured. The cost comparisons and final comparisons will be done for both the small and large installation. The calculation of the benefit scores will not be done for each case, since the benefits of an installation don’t differ with the size.

Of the contacted suppliers five sent quotations, and five did not respond or are not able to supply the right product. The received quotations are then reviewed to check whether the proposed systems fit the gas properties the adapted LP Cooab system is able to produce, and if not, the suppliers are contacted whether their systems could be adapted to create this match.

To limit the amount of work associated with the pairwise comparisons later, a pre-selection of the alternatives is executed. Table 1 shows the relationship between the number of alternatives that are compared and the number of pairwise comparisons that need to be made for each criterion. When all five alternatives would be compared this would mean that ten comparisons have to be made for each criterion. This would mean a total of 200 comparisons assuming twenty criteria will be used. AHP is only practically usable if the number of alternatives and criteria is sufficiently low so that the number of pairwise comparisons performed by the decision makers remains below a reasonable threshold (Briand, 1998). Forman (1993) postulates that performing an excessively large number of comparisons costs a lot of time and has a negative effect on the consistency because of loss of concentration. Tam and Tummala (2001) also noted that shortlisting the number of alternatives is advisable to limit the amount of data collection and computational problems.

Table 1: Number of alternatives versus number of pairwise comparisons

Number of alternatives 2 3 4 5 6 7 8

Number of pairwise comparisons 1 3 6 10 15 21 28

The shortlisting is based on the Lexicographic Semiorder (LS) using estimated production costs as decisive factor (i.e. all alternatives with costs 15 percent higher than cheapest alternative are eliminated) (Yoon & Hwang (1995)). The threshold of 15 percent is the same as the maximal allowed cost difference used later to determine the weights of the cost and benefit scores, based on the input of the decision makers.

Three alternatives are within the 15 percent threshold compared to the lowest cost alternative and make the shortlist. For every alternative a general description of the used technology is presented, as is a comparison on the costs and technical, operational and vendor specific characteristics.

Supplier A

General description: The alternative offered by Supplier A is based on a Mixed Refrigerant cooling system. A

(21)

biomethane as good as possible. The installation includes an additional drying step upstream, that will also reduce the level of CO2 to about 10 ppm. This offers the possibility to supply the biomethane with a CO2

concentration of 100 ppm, instead of 50 ppm required by the other alternatives. The installation is offered as a skid mounted system, making it easier to transport and install the installation on site.

Costs: Supplier A offers the cheapest installation in terms of initial investment. However the power

consumption is higher than Supplier B, and Supplier A is the only supplier that needs to flash part of the biomethane, resulting in a loss of biomethane. The flashing is done to lower the temperature of the product, and to remove some of the nitrogen.

Technical: The installation offered by Supplier A is able to work on 30 percent of the maximum capacity for a

limited amount of time, providing some flexibility with respect to the supply of biomethane. Supplier A guarantees an availability of 95 percent taking into account both scheduled maintenance en possible breakdowns.

Operational: The installation is capable to run unmanned during normal operation continuously. It requires 2

days for service three times a year. Supplier A offers a service agreement to take care of this.

Vendor specific: Supplier A is a Thai company previously unknown to Cirmac. They specialize in cryogenic

equipment primarily for the small scale liquefaction of LNG. However, so far they have only commissioned one installation near their factory. Because of the large distance between the customers in Europe and Supplier A in Thailand the response time in case of breakdowns will be long and repairs might take long.

Supplier B

General description: Supplier B also offers a system based on a Mixed Refrigerant cycle. However Supplier B

already has 30 years of experience with the liquefaction of different gasses and over ten years in the LNG business. The system is designed to use standard equipment and minimize the loss of biomethane.

Cost: The installation offered by Supplier B is the most expensive of the three alternatives with respect to the

capital investment. However, they offer the lowest energy consumption and will convert all the biomethane to LBM.

Technical: The Supplier B alternative can only be turned down to 50 percent of its maximum capacity. However

Supplier B guarantees an availability of 97 percent and is able to supply the LBM at any pressure between 0 and 20 barg.

Operational: The installation is designed to run unmanned and continuously for as long as a year before any

service is needed. Service is mainly necessary to unfreeze the system to remove any CO2 that may have built up

in the system. Supplier B has a service department that offers service agreements to take care of this.

Vendor specific: Supplier B is a company active in the liquid gasses business for over 30 years, and has a lot of

experience with the liquefaction of LNG, both on large and small scale. They have developed a small-scale liquefaction system for LNG and biomethane, but have not sold this system yet. However, the company has offices all over the world, and the Oil and Gas department is situated in Norway. Because of the use of standard components they can guarantee short service and repair times.

Supplier C Cryogenics

(22)

built system with any number of standard modules necessary to reach the production volume. Both the modular as the standard aspects offer many advantages over the other systems.

Costs: The initial investment for a Supplier C liquefaction system is linearly dependent on the number of

modules that are needed. With the chosen volumes the investment costs are between those of Supplier A and Supplier B. However Supplier C has the highest energy consumption, and both the investments as the energy consumption are very dependent on the required pressure for the LBM.

Technical: Every single module can be switched on and off whenever necessary, meaning that the installation

can be turned down to less than 2% off the full capacity without any problems. Also because of the modular concept and some redundancy in production capacity, the availability of the installation is more than 98%. If one module is not working because of service or repairs, the other module will take over the work.

Operational: The installation is fully automated and programmed to take the service intervals of the different

modules into account when deciding which modules to switch on or off. This is to make sure that there will never be two or more modules out for service at the same time. Every module needs about one day of service every year, which Supplier C is able to do. However, because of the high numbers of modules they advice to train client personnel to perform the service themselves.

Vendor specific: Supplier C used to be part of Philips, and was split of in 1990. They have a lot of experience

with cooling and liquefying gasses using the Stirling principle. Because the simple and modular design the experience Supplier C has with the technology they can repair a single module in at most five days without interrupting the production of the other modules.

4.3 Calculation of benefit scores

The calculation of the benefit scores is essentially the execution of the AHP methodology, without taking any cost related factors into consideration. The following steps have to be taken to calculate the benefit scores:

 Identification of the criteria

 Construction of the decision hierarchy

 Priority establishment

 Calculation of benefit scores

These steps will now be further discussed in the following subsections. The benefits scores will not be determined for the small and large installations separately since the benefits of the installations don’t differ between different sizes, only the costs will be different.

4.3.1 Identification of the criteria

Dickson (1966) indentified 23 different criteria for vendor selection including quality, performance history and price. Tam and Tummala (2001) grouped these criteria into three groups, being financial, technical and operational aspects, and added 14 vendor-specific criteria. To limit the number of criteria in the hierarchy, a survey is conducted. The survey has the goal to make sure that only the important criteria are included in the hierarchy. However because every situation is unique, a number of criteria are removed from the list, and some others are added to reflect the problem of this research in the best way possible. The most important change is the choice to treat the cost factors separately from the other criteria as proposed by Shapira and Goldenburg (2005a) to deal with the high importance of costs.

(23)

received. The resulting questionnaire was then distributed to the relevant people within the company, including people from the sales, engineering and service department.

In order to identify the relevant criteria, the respondents were asked to rate each factor using the three-point scale of “less important”, “important” and “very important”. The results of the survey are summarized in appendix A, where the mean value is calculated using the values of 1, 2 and 3 for “less important”, “important” and “very important”, respectively. The criteria are arranged in descending order of their mean values (Tam & Tummala, 2001).

To limit the number of criteria used in the AHP model, and thereby the number of necessary pairwise comparisons, a cutoff value of 2.0 is used. This value is chosen because it is the average of the highest and lowest average value of the criteria. All factors with a mean value higher than this cutoff value are identified as relevant criteria and will be used in the AHP model. All factors with a mean value lower than the cutoff value are viewed as either not important enough to take into consideration or as part of a more important factor and will not be used in the AHP model. Thus, the following criteria are identified for selecting the right liquefaction system.

Technical:

 System performance

 System reliability

 Ease of adapting upgrading process to gas inlet specifications

 Outlet specifications of LBM Operational:  Ease of operations  Maintainability Vendor specific:  Technical expertise

 Problem solving capabilities

 Delivery lead time

The above success factors are now considered as the relevant criteria to formulate an appropriate AHP model for the selection of a preferred supplier of a biomethane liquefaction system. Theoretically, all the success factors used in the survey can be included in the AHP model, as the AHP methodology is able to compare and prioritize any number of criteria and alternatives. However, it is not practical to include all factors as they increase the number of pairwise comparisons and the related computational effort. It is also possible that assessment biases may occur in obtaining the pairwise comparison judgments from evaluators. Furthermore, some of the factors that are not selected can be grouped into other selected criteria (Tam & Tummala, 2001). Therefore a cutoff value of 2.0 was used and nine criteria were selected for formulating the AHP model.

4.3.2 Construction of the decision hierarchy

(24)

consistency. For this problem three main criteria are selected, being technical, operational and vendor specific. The sub criteria chosen before will be assigned to these criteria.

The decision hierarchy consists of four levels, being the goal, the main criteria, the sub-criteria and the

alternatives. The first level of the hierarchy is the goal of the decision making process. In this case the goal is to assign benefit scores to the alternatives. To achieve this three main criteria were selected which form the second level of the hierarchy. The main criteria are technical, operational and vendor specific criteria. The third level consists of the sub-criteria selected in subsection 4.3.1. The sub-criteria are allocated to the main criteria based on the categorization made in subsection 4.3.1. The fourth and last level of the hierarchy consists of the acceptable alternatives that are to be evaluated. Alternatives must be connected to all of the leaf attributes, or the lowest criteria in the hierarchy, potentially affecting their evaluation (Shapira & Goldenburg, 2005a). The constructed hierarchy is shown in Figure 6.

Determination of the benefit scores

Technical criteria Operational criteria Vendor specific criteria Ease of operations Maintainability System performance Technical expertise System reliability Ease of adapting upgrading proces Outlet specifications Problem solving capabilities

Delivery lead time

Level 1: Goal Level 2: Criteria Level 3: Sub-criteria Level 4:

Alternatives Cryothai Hamworthy Stirling

Figure 6: Benefit score decision hierarchy

The decision hierarchy presented in Figure 6 is the basis for the following steps. In the next step the priorities of the criteria will be established and the alternatives will be rating on the criteria, after which the benefit scores can be calculated.

4.3.3 Priority establishment

(25)

has to allocate time for this and have a minimal knowledge of the subject. Furthermore Mitchell and Wasil (1989) showed that, in general, smaller groups are favorable. For obtaining the pairwise comparisons it is attempted to reach consensus with every comparison in accordance with Dyer and Forman (1992).

When establishing the priorities the decision makers have to compare the criteria or alternatives pairwise, and decide which one is more important and to what magnitude. These values are then converted to numerical values and used to calculate the priority ratings. The possible magnitudes of importance and their numerical values are shown in Table 2(Saaty, 1987).

Table 2: AHP magnitudes of importance (Saaty, 1987)

Verbal Judgment of Importance Numerical Rating

Extreme difference in importance 9

Very large difference in importance 7

Large difference in importance 5

Moderate difference in importance 3

Equally Important 1

Intermediate values between two adjacent values 2,4,6,8

Elements very close in importance 1.1 through 1.9

Criteria weights

First, the comparisons for the calculation of the relative weights of the criteria are performed. The criteria and sub-criteria are assessed using pairwise comparisons of elements in each level with respect to every parent element located one level above. These pairwise comparisons are then placed in a reciprocal matrix like Table 3 below for the main criteria. A reciprocal matrix means that for any entry Aij=1/Aji.

To derive the weights, or priority vector, from this matrix three steps have to be taken. First the matrix has to be normalized. This means that every element is divided by the sum of its column. Next, the rows of the resulting matrix have to be summed, and then divided by the number of attributes (i.e. criteria or alternatives). In mathematical terms this process can be explained with equation 1 below (Shapira & Goldenburg, 2005a).



_

 



n j n k kj ij i

a

n

PV

1 1

*

1

(1) In equation 1 PVi stands for the priority vector, or weight of the attribute; n stands for the number of attributes

in the matrix; and Aij stands for the relative importance of attribute i over attribute j.

Table 3 shows the results of the pairwise comparisons for the main criteria and the results of the first step. The matrix derived after the second step and the resulting priority vectors are shown in Table 4.

Table 3: Comparison matrix for main criteria

Technical Operational Vendor

specific

Technical 1 3.00 2.00

Operational 1/3 1 0.50

Vendor specific 1/2 1/0.50 1

(26)

Table 4: Normalized matrix for main criteria

Technical Operational Vendor

specific Sum Priority vector Technical 0.545 0.500 0.571 1.617 53.9% Operational 0.182 0.167 0.143 0.491 16.4% Vendor specific 0.273 0.333 0.286 0.892 29.7%

The same calculations are executed for every group of sub criteria to obtain their local weights. A set of global priority weights can then be determined for each of the sub criteria by multiplying the local weights of the sub criteria with the weights of the parent node above it.These weights are summarized in Table 5.

Table 5: Local and global weights of the criteria

Criteria Sub criteria Local weight Global weight

Technical 0.539

System reliability 0.558 0.301

System performance 0.259 0.140

Outlet specifications of LBM 0.112 0.061

Ease of adapting upgrading process to gas inlet specifications 0.071 0.038 Operational 0.164 Ease of operations 0.750 0.123 Maintainability 0.250 0.041 Vendor specific 0.297

Problem solving capability 0.656 0.195

Technical expertise 0.265 0.079

Delivery lead time 0.080 0.024

Alternative scores

Similar to the criteria, the comparison of the different alternatives can be made using pairwise comparisons. After receiving the quotations and selecting the three offers within the given cost range, documents were prepared to clearly show all the necessary information for the pairwise comparisons in an understandable and concise way. These documents were used to inform the participants of the comparison sessions without overwhelming them with information, and as a basis for the discussions when assigning the pairwise

(27)

Table 6: Alternative score per criterion

Criteria Sub criteria Scores

Supplier A Supplier B Supplier C Technical System reliability 0.07 0.17 0.76 System performance 0.06 0.28 0.66 Outlet specifications of LBM 0.14 0.43 0.43

Ease of adapting upgrading process to gas inlet specifications 0.07 0.22 0.71 Operational Ease of operations 0.14 0.43 0.43 Maintainability 0.09 0.22 0.69 Vendor specific

Problem solving capability 0.08 0.26 0.66

Technical expertise 0.06 0.35 0.58

Delivery lead time 0.57 0.10 0.33

4.3.4 Calculation of benefit scores

The last phase of the AHP-based calculation of the benefit scores (BS) is the aggregation of the criteria weights (PV) and alternative scores (AS). Every alternative score is multiplied with the criterion weight and then summed for all criteria to give the alternative benefit score (equation 2).

j n j ij i

AS

PV

BS

*

1





(2)

(28)

Table 7: Calculation of Benefit Score

Criteria Sub criteria Weight Scores

Supplier A Supplier B Supplier C

Technical

System reliability 0.301 0.07 0.17 0.76

System performance 0.140 0.06 0.28 0.66

Outlet specifications of LBM 0.061 0.14 0.43 0.43

Ease of adapting upgrading process to gas inlet specifications 0.038 0.07 0.22 0.71 Operational Ease of operations 0.123 0.14 0.43 0.43 Maintainability 0.041 0.09 0.22 0.69 Vendor specific

Problem solving capability 0.195 0.08 0.26 0.66

Technical expertise 0.079 0.06 0.35 0.58

Delivery lead time 0.024 0.57 0.10 0.33

Benefit score 0.096 0.267 0.637

4.4 Calculation of cost scores

The basis for the calculation of the cost scores is the expected production cost per kilogram of LBM a customer would incur with one of the alternatives. The calculation takes as much information into account as possible, and is based on the way customers will calculate and compare different alternatives. The calculation is derived from the evaluation of an earlier tender of the Swedish government (Malmberg, 2001).

The calculation of the production cost per kilogram takes into account the depreciation and interest on the initial investment, the running cost and the maintenance cost. The calculation will also take into account the cost of the upgrading process and the cost of the biogas because the alternatives differ in amount of LBM produced from the same amount of biomethane. This means that these costs will be distributed on different production amounts.

(29)

Table 8: Cost calculation for a small installation (400 Nm3/h)

Cost (€/year)

Alternative Supplier A Supplier B Supplier C

Investment Liquefaction installation 1 200 000 3 100 000 1 900 000 Additional costs 550 000 250 000 500 000 LBM storage 250 000 Included 250 000 Depreciation 200 000 335 000 265 000 Interest 100 000 167 500 132 500 Electricity 214 433 135 955 330 939

Service and maintenance 23 000 57 500 61 000

Liquefaction cost (€/year) 537 433 695 955 789 439

Yearly LBM production 2 144 330 2 432 748 2 482 908

Liquefaction cost (€/kg LBM) 0.250 0.287 0.318

Biogas cost (€/kg LBM) 0.260 0.229 0.225

Upgrading cost (€/kg LBM) 0.129 0.113 0.111

Total cost (€/kg LBM) 0.639 0.629 0.654

Table 9: Cost calculation for a large installation (900 Nm3/h)

Cost (€/year)

Alternative Supplier A Supplier B Supplier C

Investment Liquefaction installation 1 800 000 3 600 000 2 818 000 Additional costs 550 000 300 000 600 000 LBM storage 250 000 Included 250 000 Depreciation 260 000 390 000 366 800 Interest 130 000 195 000 183 400 Electricity 434 227 306 579 735 420

Service and maintenance 32 000 57 500 130 000

Liquefaction cost (€/year) 856 227 949 079 1 415 620

Yearly LBM production 4 827 742 5 473 684 5 586 543

Liquefaction cost (€/kg LBM) 0.177 0.179 0.253

Biogas cost (€/kg LBM) 0.260 0.229 0.225

Upgrading cost (€/kg LBM) 0.129 0.113 0.111

Total cost (€/kg LBM) 0.566 0.521 0.589 However the cost per kilogram can’t directly be used as cost scores. Instead a cost score has to be calculated. The Cost Score (CS) is calculated using equation 3.To calculate the cost score the inverse of the cost of an alternative has to be calculated and then divided by a summation of the inverses of all the alternatives.

(30)

The scores resulting from these calculations are shown in Table 10 and can be used to compare with the benefit score and to calculate the total score of each alternative. This will be done in the next subsection.

Table 10: Calculation of cost scores

Alternative Small installation Large installation

Cost per kilogram Cost score Cost per kilogram Cost score

Supplier A 0.639 0.334 0.566 0.328

Supplier B 0.629 0.339 0.521 0.357

Supplier C 0.654 0.326 0.589 0.315

4.5 Determining the optimal alternative

Using the cost and benefits scores the optimal alternative can be determined. This is done separately for the small and large installation. Also the results are checked for changes in the maximal allowed cost difference (Cmax) and for changes in the cost price of the raw biogas.

4.5.1 Optimal alternative for a small installation

The first step in calculating the total scores of the alternatives is to select the lowest cost alternative to be the benchmark and to set its cost and benefit scores to 1.0. Next, the scores of the other two alternatives are adjusted proportionally. Next the cost and benefit weights have to be calculated. The cost weights (CW) are calculated using equation 4 and the benefit weights (BW) are calculated using equation 5 (Shapira &

Goldenburg, 2005a). The maximal allowed cost difference (Cmax) is determined by the decision maker to reflect

the largest difference in costs the customer is willing to pay for certain benefits. If the cost difference is larger than this number the lowest cost alternative will be selected. In this case the maximal allowed cost difference is determined to be 15 percent of the cost of the lowest cost alternative. This means for a small installation that the Cmax is 9.4 €cent. Any alternative with a cost 9.4 €cent higher than the lowest cost alternative will be

excluded from further analysis.

max

/ C

C

CW

_i





_i (4)

Where ΔCi denotes the cost difference between alternative i and the lowest cost alternative and Cmax is the

maximal allowed cost difference.

i

CW

BW



1 

(5)

The total scores are then calculated by multiplying the scores by their respective weights and then normalize these scores so their sum equals 1.0. The benefit and cost scores and weights and total scores are shown in Table 11.

Table 11: Total score calculation for a small installation

Alternative Cost Benefits Total score

(normalized)

Score Adjusted

score

Weight Score Adjusted

score

Weight

Supplier A 0.334 0.984 0.105 0.096 0.360 0.895 0.124

Supplier B 0.339 1.000 0.000 0.267 1.000 1.000 0.291

(31)

Supplier C is selected to be the preferred alternative for a small installation. However, this choice is dependent on a number of different variables that might influence the choice of supplier. The most important variables are Cmax and the cost price of the biogas. Figure 7 shows the relationships between these variables and the

total scores of the alternatives.

Figure 7: Relationship between Cmax and the total score for a small installation.

Figure 7 shows that Supplier C is the best option, and that it stays the best alternative as long as Cmax is larger

than the difference in costs between Supplier C and Supplier B. In case Cmax is below 2.5 €cent Supplier C will be

marked as too expensive, and Supplier B will be the optimal alternative as it outperforms Supplier A on both the costs and the benefits.

A change in biogas cost price has the biggest influence on the score of Supplier A. The scores of Supplier C and Supplier B remain the same opposed to each other because they both convert all of the biomethane into LBM whereas Supplier A has to flash part of the biomethane in the process. This means that in case the biogas cost price will be more than the € 0.10 used for the calculations the installation offered by Supplier A will be deemed too expensive and the choice will be made between Supplier B and Supplier C.

4.5.2 Optimal alternative for a large installation

In the same manner as with the small installation the scores for the large installation can be calculated. The relevant figures and total scores are shown in Table 12. Again the Cmax is set at 15 percent of the lowest cost

alternative. In this case this means a Cmax of 7.8 €cent.

Table 12: Total score calculation for a large installation

Alternative Cost Benefits Total score

(normalized)

Score Adjusted

score

Weight Score Adjusted

score

Weight

Supplier A 0.331 0.920 0.577 0.096 0.360 0.423 0.248

Supplier B 0.352 1.000 0.000 0.267 1.000 1.000 0.362

(32)

Supplier C is elected to be the preferred alternative for a large installation. Figure 8 shows that although Supplier C is the best option, when the costs become more important, or Cmax becomes lower 7.4 €cent then

Supplier B will become the better alternative. In case Cmax is below 6.8 €cent Supplier C will be too expensive,

and be deemed too expensive. Supplier B will then be the optimal alternative as it outperforms Supplier A on both the costs and the benefits.

Figure 8: Relationship between Cmax and the total score for a large installation.

(33)

5. Competitiveness of the combined processes

To be able to determine whether the developed Cirmac biogas upgrading and liquefaction system is competitive, a comparison with the competitors has to be made. However most of the companies offering upgrading or liquefaction technology won’t share information about their installations concerning the technology or prices. The information that is available is concerning the energy consumption and methane losses. This information is used to compare the different processes. The different processes are also compared on the positive and negative aspects of the processes.

Table 13 shows the energy consumption of the different processes calculated as kWh/kg LBM. A specified table with the original data is added in appendix F. As can be seen from the table the cryogenic technology offered by GtS is the most energy efficient of the cryogenic technologies. The Cooab system offered by Cirmac is the most energy efficient of the biogas upgrading technologies, but only when residual heat can be used in external processes. The combination of the Cirmac upgrading and Supplier B liquefaction is very competitive in terms of energy consumption when a large amount of the residual heat can be reused. The liquefaction of biogas using the Supplier C technology is more energy intensive, but offers certain advantages opposed to Supplier B or the other liquefiers. The advantages and disadvantages of the liquefaction systems will be discussed next.

Table 13: Energy consumption for biogas liquefaction (Adapted from Johansson (2008))

Company Energy consumption (kWh/kg LBM)

Electricity Heat Cryogenic technology

GtS 1.1

Acrion 2.0

Prometheus 2.2 Conventional upgrading and liquefaction

The Linde Group 2.0

Cirmac & Supplier B 1.05 0.14 (1.4)*

Cirmac & Supplier C 2.03 0.14 (1.4)*

* 1.4 kWh/kg is the gross heat consumption, but of these 1.26 kWh/kg can be reused for heating in external processes. This results in a net consumption of 0.14 kWh/kg in case all of the heat can be reused.

The advantages and disadvantages of the different liquefaction systems are summarized in Table 14. As can be seen from this table all the systems except the Linde Group have low CH4 losses and most are energy intensive

(34)

Table 14: (Dis)advantages of liquefaction systems

Company/process Positive Negative

Cryogenic technology

GtS Low CH4 losses No commercial experience

Extraction of liquid CO2

Experience from pilot plant Appropriate for landfill gas

Acrion Low CH4 losses Energy intensive process

Extraction of liquid CO2 No commercial experience

Experience from pilot plant Appropriate for landfill gas

Prometheus Low CH4 losses Energy intensive process

Experience from commercial and pilot plants

No extraction of liquid CO2

Appropriate for landfill gas Conventional upgrading and liquefaction

The Linde Group Simple and sturdy design Low CH4 recovery (85%)

Experience from commercial plants Energy intensive

Cirmac & Supplier B Low CH4 losses Use of chemicals

Energy efficient Large use of heat

Experience from pilot plant Complex for small-scale applications No commercial experience

Cirmac & Supplier C Low CH4 losses Use of chemicals

Modular design Large use of heat

Simple, reliable technology Energy intensive Cheap installations, especially for smaller

sizes

Technology not applied to biomethane yet

Possibility to vent excess nitrogen Requires a large amount of cooling Both Supplier B and Supplier C have different, but convincing reasons to be chosen as the supplier of a liquefaction installation. The choice between either one of these two should be made based on project

(35)

6. Conclusion

In this thesis an approach was presented for the selection of a preferred supplier for biomethane liquefaction installations. Recently Cirmac noticed an emerging trend for the liquefaction of biomethane and wishes to be able to offer an installation for the upgrading and liquefaction of biogas to their customers. However Cirmac has no experience with the liquefaction of gasses and is in search of a competitive liquefaction process that matches the characteristics and production size of a typical biogas installation.

The goal of this project was to help Cirmac develop a competitive biogas liquefaction installation so as to attract their customers. To achieve this goal, the research objective was to design a procedure to determine the optimal choice of a liquefaction system for Cirmac. The selection of a supplier is determined by both quantitative as qualitative factors as well as a trade-off between costs and benefits. To be able to structure the problem and select the optimal alternative the Analytic Hierarchy Process (AHP), part of Multiple Attribute Decision Making (MADM) framework, was presented and adapted for use in this case. First the literature behind MADM and AHP was presented and adapted for use in a group decision making environment and a method for a separate cost calculation was presented. Next this method was adapted and applied to the case of Cirmac International.

The adapted procedure consisted of three parts. In the first part the benefits of the alternatives were compared using the Analytic Hierarchy Process (AHP). The second part consisted of a cost comparison of the production costs using the different alternatives. In the final part these scores were synthesized to produce a final score and advice on the optimal supplier for a biomethane liquefaction process.

First, a set of alternatives had to be created. The alternatives were determined by requesting quotations of most of the companies offering gas liquefaction installations. The installations offered by Supplier A, Supplier B and Supplier C were selected to be included in the procedure. The three selected alternatives all had different pros and cons on which any of the three could be selected or discarded.

In the first part of the procedure AHP was used to compare the benefits of the alternatives and to produce a benefit score for the alternatives. The first step in AHP was determining the criteria on which to rate alternatives. These criteria were determined based on literature combined with a survey among the staff of Cirmac. Based on the survey, the following nine sub criteria were selected and divided over three main criteria. Technical:

 System performance

 System reliability

 Ease of adapting upgrading process to gas inlet specifications

 Outlet specifications of LBM Operational:  Ease of operations  Maintainability Vendor specific:  Technical expertise

 Problem solving capabilities

Supplier Selection for a Biomethane Liquefaction Installation