Linköping Studies in Science and Technology Thesis No. 1654
Supply chain coordination using optimal
transfer pricing to balance co- and
by-product demand within a process industry
Martin Kylinger
2014
Division of Production Economics Department of Management and Engineering
© Martin Kylinger 2014
Linköping studies in science and technology, Thesis No. 1654
ISBN: 978-91-7519-358-8 ISSN: 0280-7971
Printed by: LiU-Tryck, Linköping
Distributed by: Linköping University
Department of Management and Engineering SE-581 83 Linköping, Sweden
Abstract
Process industries has for long been important for the development of Swedish industry and society. All industries face different conditions that affect how to best run their operations. This thesis aims to describe some of the conditions that characterize process industries compared to other industries. Further one of these characteristics has been studied more closely.
One of the traits of process industries is that they are positioned at the start of the transformation process close to the raw material mixing, separating or forming it into products often used for further transformation. Process industries hence become dependent on the properties of these materials. One of the most prominent characteristics inherent from the raw material properties is the divergent bill of material. The divergent bill of material originates from the fact that a given raw material is made up of different components that will yield several products with different characteristics when processed. When splitting the raw material into the desired products the yielded products from a certain raw material usually have different value to the producer, some more desired than others. These multiple products generated poses a challenge from a planning perspective raising questions like “How should we balance the supply and demand of all the products produced?”, “What shall we do with the excess products produced?”.
The first paper in this thesis describe the production planning in four Swedish process industries with the ultimate aim to connect their planning to the supply chain characteristics they face as process industries. The study concludes that the industry specific conditions mainly affect planning at short time ranges when planning becomes more detailed. In general the use of planning or decision support systems is low, stemming from a, warranted or not, belief that general decision support systems do not fit process industries. Another finding is that the case companies mainly operate in niche markets. This study also highlighted that the planning complexity arising from characteristic of co- and by-product generation in combination with the lack of decision support systems requires further studies.
The subsequent two papers focus on supply chain planning and coordination with a divergent bill of material. They present a mathematical model over the supply chain planning in a real case company in the specialty oils industry. The second paper investigates transfer pricing as a coordination tool by comparing decentralised supply chain planning with centralized planning in an integrated model. Transfer pricing is found to have a potential positive effect on supply chain planning while simultaneously creating problems in terms of uneven distribution of the contribution margin among supply chain partners.
Finally the third paper more closely investigates different ways to set transfer prices and comparing them to the optimal transfer prices. Setting optimal transfer prices with a divergent bill of material has proven to be less straightforward than the case with no
dependencies between products. Some optimal transfer prices could even be set lower than the marginal cost for producing them due to the dependency between the products in the divergent bill of material. This indicates that there is an opportunity cost for a product that is dependent on the demand of other products.
Sammanfattning
Processindustrier har länge varit viktiga för utvecklingen av svensk industri och det svenska samhället. Alla branscher möter olika förutsättningar som påverkar hur verksamheten bäst skall drivas. Avhandlingen syftar till att beskriva några av de villkor som kännetecknar processindustrin jämfört med andra branscher. Slutligen har en av dessa kännetecknande egenskaper studerats närmare.
En av de egenskaper som utmärker processindustrier är att de är positionerade vid början av omvandlingsprocessen nära råvaran och blandar, separerar eller formar den till produkter som ofta används för ytterligare förädling. Processindustrin blir därmed beroende av egenskaperna hos råmaterialet som används. En av de mest framträdande egenskaperna härrörande från dessa råmaterialegenskaper är det divergenta materialflödet. Det divergerande materialflödet har sitt ursprung i att ett visst råmaterial består av olika komponenter som ger flera produkter med olika egenskaper. När råmaterialet separeras i sina komponenter erhålls flera produkter som kan ha väldigt olika värde för producenten. Efterfrågan på de produkter som erhålls ur separeringen kan vara vitt skild från vad som erhålls ur produktionsprocessen. Denna obalans mellan efterfrågan och försörjning är en utmaning ur ett planeringsperspektiv och väcker frågor som " Hur ska efterfrågan och försörjning av de produkter som produceras balanseras?", "Hur skall överskottet på vissa produkter hanteras? ".
Den första artikeln i denna avhandling beskriver produktionsplaneringen i fyra svenska processindustriföretag inom områdena pappersmassa, specialkemi, specialoljor samt papper och pappersmassa. Målet med studien var att jämföra planeringen av försörjningskedjan med de förutsättningar som processindustrin ställs inför. I studien dras slutsatsen att de branschspecifika förhållandena främst påverkar den kortsiktiga planeringen med mer detaljer. I allmänhet verkar användningen av planerings- eller beslutsstödsystem vara låg, som härrör från en, befogat eller inte, tro att generella beslutsstödssystem inte passar processindustri. Fallföretagen förefaller också huvudsakligen vara verksamma på nischmarknader. Denna studie betonar också vikten av djupare studier av energiplanering samt den planeringskomplexitet som uppkommer till följd av divergerande materialflöden.
De två efterföljande artiklarna fokuserar på planering och koordinering av försörjningskedjan vid ett divergerande materialflöde. De presenterar en matematisk modell över försörjningskedjan i ett fallföretag som tillverkar specialoljor. Den andra artikeln undersöker effekter av att använda internprissättning som verktyg för koordinering av försörjningskedjor genom att jämföra decentraliserad försörjningskedjeplanering med en fullt integrerad centraliserad planering. Internprissättning visar sig ha potentiellt positiva effekter på planeringen av försörjningskedjan men skapar samtidigt problem i form av att täckningsbidraget kan fördelas väldigt ojämnt mellan partner i försörjningskedjan.
Slutligen behandlar den tredje artikeln olika sätt att bestämma internpriser och utvärderar deras koordineringseffekter på försörjningskedjan genom att jämföra med optimal satta internpriser. Att bestämma optimala internpriser med ett divergerande materialflöde har visat sig bli komplicerat i och med det beroende som finns mellan produkterna som kommer från samma råmaterial. Optimala internpriser har visat sig kunna vara både avsevärt högre och avsevärt lägre än marginalkostanden för att producera dem. Detta indikerar att alternativkostnaden för en produkt kan påverkas starkt av efterfrågan på andra produkter den har en koppling till via råmaterialet de båda produceras från.
Foreword
Working on this thesis has been a bumpy ride and I have often doubted the sanity of embarking on this journey. In spite of all obstacles, frustration, and disappointments I have learnt some valuable lessons about research and life itself. On my journey to finishing this thesis I have encountered many people that helped me on my way in one way or another.
A very warm thank you goes to Mario Guajardo for the fantastic experience of a great collaboration resulting in the two main articles of this thesis. Thank you for all your efforts, interesting discussions, and good ideas. Additional thanks goes to the co-authors of one of the papers Johan Johansson and Martin Waldemarsson, thank you for the good cooperation, all the fun and all the support you have provided.
My supervisors Mikael Rönnqvist and Mathias Henningsson also deserve lots of thanks for their support in me choosing which path to go down. You have brought me back on track when I have made a wrong turn and have helped me find meaning in what I do. I also thank all those people at the companies that have invited me and shared their precious knowledge and time.
Every day my colleagues at Productions Economics have made coming to work a pleasure. I have especially enjoyed my fellow PhD students, the ones still at Production Economics as well as those that have moved on to other challenges. I thank you for all the support and sharing of your experiences in research as well as in life.
My most precious thanks go to all my family, both those present and those not yet here. During the work with this thesis my parents Laila and Max have been a great support both mentally as well as helping out with everyday life and various emergencies. I am especially grateful for my wife Malin and my son Jesper for showing me what life is all about and what is important. You have suffered from my frustrations stemming from this work but you have always been my highest priority and will always be.
Thesis outline
This thesis entitled Supply Chain Coordination Using Optimal Transfer Pricing to Balance
co- and by-product Demand Within a Process Industry is a licentiate thesis in Production
Economics at Linköping University. The research is performed within the Process Industry Centre (PIC) supported by the Swedish Foundation for Strategic Research (SSF).
The thesis is comprised by two parts, where the first part is an introduction to the setting wherein the research has been conducted and is relevant while the second part consists of three appended papers. The first part aims to give a background to the studied area as well as an introduction to the theoretical foundation the papers are built upon. Scope and research questions to be studied are followed by a presentation of the case company. A summary of the appended papers and concluding remarks with suggested further research concludes the first part. The second part consists of the three appended papers listed below.
Paper 1
JOHANSSON, J., KYLINGER, M. & WALDEMARSSON M. 2012. Production planning in process industries. Working Paper: LIU-IEI-WP-12/0002, Department of Management and Engineering, Linköping University, Sweden.
An earlier version of this paper was presented at NOFOMA 2011 Conference, 9-10 June, 2011, Harstad, Norway.
Paper 2
GUAJARDO, M., KYLINGER, M. & RÖNNQVIST, M. 2013b. Speciality oils supply chain optimization: From a decoupled to an integrated planning approach. European Journal of
Operational Research, 229, 540-551.
An earlier version of this paper was presented at INFORMS Annual Meeting 2011, November 13 – 16, 2011, Charlotte, North Carolina, USA.
Paper 3
GUAJARDO, M., KYLINGER, M. & RÖNNQVIST, M. 2013a. Joint optimization of pricing and planning decisions in divergent supply chain. International Transactions in Operational
Content
1 Introduction ... 1
1.1 Background and Context ... 1
1.1.1 Importance of Swedish Process Industry and Relevance of Study ... 1
1.1.2 Process Industry Characteristics... 3
1.1.3 Production and Raw Material Characteristics ... 6
1.1.4 Transfer Pricing and Coordination of Decoupled Supply Chains ... 8
1.1.5 Operations Research in Supply Chain Planning ... 8
1.2 Scope and Objective ... 8
2 Frame of Reference ... 9
2.1 Divergent Bill of Material ... 9
2.1.1 Raw Material Properties and Divergent Bill of Material ... 9
2.1.2 Fixed Proportions Between co- and by-products in Petroleum Industry .... 10
2.2 Supply Chain Management ... 11
2.2.1 Strategic / Long-term Planning ... 13
2.2.2 Tactical / Mid-term Planning ... 14
2.2.3 Operational / Short-term Planning ... 14
2.2.4 Managing Divergent Supply Chains ... 15
2.3 Operations Research in Petroleum Supply Chain Management ... 16
2.3.1 Procurement and Production Planning Tasks ... 17
2.3.2 Distribution ... 18
2.3.3 Sales and Optimal Pricing ... 19
2.4 Transfer Price Coordination and Collaboration in Supply Chains ... 20
3 Research Challenge ... 23
3.1 Scope ... 23
3.2 Research Design ... 23
3.2.1 Research Objectives and Research Questions ... 24
3.2.2 Choice of Methodology ... 24
4 Description of Case Industry ... 27
4.1 General Description ... 27
4.2 Case Company Supply Chain ... 27
4.4 Experienced Difficulties at Case Company ... 29
5 Summary of Papers ... 31
5.1 Paper 1 ... 31
5.2 Paper 2 ... 32
5.3 Paper 3 ... 33
6 Concluding Remarks and Further Research ... 35
1
1 Introduction
Coordinating activities along the supply chain is essential for the profitability for all companies. Supply chains often span over several parties both within the same organisation as well as between different organizations. When supply chain entities are partially or completely independent from each other planning in the different entities has to be coordinated in order to achieve as good overall performance as possible. Coordination such a decoupled supply chain involves one or several mechanisms aligning decisions in two or more directly interacting entities with the aim to achieve a common goal. Transfer pricing is such a mechanism that provides an interface between two different supply chain partners connecting decisions between them. Hierarchical planning where long term plans set frames for shorter time periods is another way of coordinating supply chain entities.
Challenges faced by a company when planning their supply chain vary from company to company depending industry specific as well as company specific conditions. Process industries are fundamentally different from assembly and manufacturing industries and are often found in the beginning of the supply chain processing raw material and are therefore affected by the properties inherent in the raw material. Not all process industries are similar but share several characteristics that are found less frequently in other industries.
1.1 Background and Context
To succeed in your industry alignment between your business environment and your operation strategy is a key ingredient. (Safizadeh et al., 1996; Jonsson and Mattson, 2003; Platts et al.,1998). Process industries, like all other industries, face conditions that are common or even unique for their specific industry. Process industries are predominantly found in the beginning of the supply chain processing raw material into products often used for further processing. The Association for Operations Management (APICS) defines process industries as companies that: “produce products by mixing,
separating, forming and/or performing chemical reactions.”(Blackstone et al., 2005).
Process industries are often divided into batch and continuous flow industries Fransoo and Rutten (1994) point out differences between these two groups. What separates them from discrete manufacturing though is that the unit of measure is continuous and not discrete. Even if the main part of the production process is continuous most process industries however have some point where their products become discrete as acknowledged by for example Pool et al.(2011).
1.1.1 Importance of Swedish Process Industry and Relevance of Study
Process industries have for long been important for the industrial development of Sweden and is often said to have built Swedish welfare. Many provincial towns were also built around process industries. (IVA, 2006; Skogen och kemin, 2006; Ministry of Enterprise Energy and Communications, 2001; ITPS, 2006)
2
In a study of Swedish process industry the Royal Swedish Academy of Engineering Sciences establish the importance of Swedish process industry for the Swedish society and industry. Process industries account for 30% of Swedish exports and 60% of the net export while directly employing 320 000 people. Including equipment suppliers, service providers and other indirect industries the total number of people employed has been estimated to be as high as 640 000. Process industries also accounts for 85% of transported volumes in Sweden. (IVA, 2006)
In line with the identified importance of process industry for Swedish industrial development and competitiveness the Swedish Foundation for Strategic Research initiated two research centres connected to process industry. More recently Process industry IT and automation has been identified as a strategically important area in an extensive effort to strengthen Swedish growth and competitiveness.
Conditions for the development of competitive process industries in different regions vary from industry to industry but many process industries have been and still are dependent on proximity to raw material and suitable energy sources. Raw materials used for several process industries are accessible in Sweden such as forests and many important minerals. What Sweden lack though is a sufficient supply of crude oil but still harbour crude oil refineries. Historically Sweden has had access to suitable and cheap energy sources such as hydro power, which helped make Swedish process industries competitive. Recent studies conflict in their view but indicate that the energy cost for Swedish process industries are in line with European competitors while being higher than for example for Canadian and U.S. competitors (Skogen och kemin, 2006, IVA, 2006). Sweden has a relatively well developed road network even in sparsely populated areas, aiding the raw material dependent industry.
According to a report released by the Government Offices of Sweden (Communications, 2001) the situation for Swedish process industries differs between industries but some common characteristics can be seen for the investigated process industries. For some process industries Sweden has a higher added value per employee but it is not a general conclusion for all process industries when comparing to the strongest competitors. The degree of automation in Swedish process industry is though relatively high. The study also concludes that Swedish process industries are prominent when it comes to environmental thinking and development of technology to reduce environmental impact.
Swedish process industries are characterised by a high degree of automation and technologically advanced facilities with relatively high productivity compared to worldwide competitors. Another trait of Swedish process industry is a high degree of niche products made possible by good access to well educated staff and strong research organizations. Larger than average proportion of niche products is also supported by a well developed cooperation with suppliers and customers. The close relationship with equipment suppliers is one of the key factors to the high level of technology in the
3
facilities and has contributed to building a strong base of equipment manufacturers for Swedish industry in general.
1.1.2 Process Industry Characteristics
Most attempts in literature to describe what characterises process industries identify what differentiates process industries from discrete manufacturing firms. These differences will be the starting point for describing the conditions process industries operate under and affects their planning.
The two most commonly used dimensions to differentiate process industries in literature are marketing environment and manufacturing environment. Marketing environment is also denoted product structure and manufacturing environment denoted process structure. Henceforth marketing environment and manufacturing environment will be the concepts used. Hayes and Wheelwright (1979) use the same dimensions when suggesting a product-process matrix describing the usually preferred relationship between product structure (market environment) and process structure (manufacturing environment). Market environment range from low volume, one of a kind, to high volume commodities while the dimension manufacturing environment range from job shop with multipurpose equipment to flow shop with highly specialized equipment. Taylor et al. (1981) have mapped different industries into this matrix as seen in Figure 1, indicating that process industries in general tends to end up towards the flow shop and commodity product end of the spectrum.
Figure 1. Product-process matrix influenced by Taylor et al.(1981).
Most mapped industries end up on or close to the diagonal of the matrix indicating that there is a direct connection between market environment and the preferred choice of process. Safizadeh et al. (1996) concludes that best performance is achieved if process
Custom Low Volume Differentiated High Volume Differentiated Pr oces s St ru ct ur e / M an uf ac tu rin g En vi ro nme nt Aerospace Industrial Machinery Machine Tools Drugs Automobile Tire and Rubber
Steel Products Paper Containers Brewers Oil Forest Products
Product Structure / Market Environment
Flow Shop Job Shop
Commodity
Electrical and Electronics
Oil Specialty Chemicals
Steel Major Chemicals
4
choice is aligned with market environment according to Hayes and Wheelwright (1979) product-process matrix. Some profitable process industry companies found off the diagonal are niche companies where process and equipment are similar to the commodity producers while their market environment has many traits similar to custom products. However stepping away from the diagonal must be a conscious choice made with knowledge about the implications of such a shift.
Process industries are fundamentally different from fabrication and assembly industries, operating in different marketing, manufacturing and financial environments (Taylor et al., 1981). Seeing process industries as a homogenous group is though a doubtful simplification that often is too simple. This can also be seen in that more resent literature do not deal with process industry as a group but rather as separate industries, albeit still with many common traits.
When trying to distinguish process industries from fabrication and assembly industries Taylor et al. (1981) use the dimensions found in the product-process matrix as a base and above that add the financial environment. Taylor et al. (1981) study these dimensions one at a time to synthesize what characterizes process industries. Market environment characteristics they contrast for commodity products compared to custom made products are listed in Table 1.
Table 1. Commodity products market environment according to Taylor et al.(1981).
Marketing Environment for Commodity Products
• Marketing Emphasis on Product availability and Price • Few Products
• Few Product design changes • Derived Demand
• High Sales Volume • Low Unit Value
• Relatively High Transportation Costs
Manufacturing facilities are often classified according to the choice of production process, ranging from job shops with multipurpose equipment at one end to flow shops with highly specialized equipment in the other end. Process industries tend to be toward the flow shop end of the scale. Taylor et al. (1981) contrasts job shops and flow shops environments on a number of areas such as fixed or variable routings, layout, flexibility of equipment, volume, lead time to increase capacity. Further they highlight a number of characteristics from both job shop and flow shop that are frequently found in process industries. They classify process industries in general as flow shops producing commodity products and what they think are the most important manufacturing environment characteristics for process industries are summarized in Table 2.
5
Table 2. Characteristics of process industry manufacturing environment according to Taylor et al. (1981).
Process Industry Manufacturing Environment
• Variability in Raw material Quality • Variability in Bill of Material • Product Yields May Vary Widely
• May Have Large Demands for Intermediate Products • Co-product Demand Must be Balanced
• Products or Raw Materials May Have a Shelf Life
Fransoo and Rutten (1994) have condensed process industry characteristics and give examples of specific process industries where these characteristics are present. Some of these are listed in Table 3.
Table 3. Process industry characteristics, adapted from Fransso and Rutten (1994).
Characteristic Example of Industry
Variable yield Chemical industry
Variable quality Oil, forest products
Variable quantity/availability Coffee, agricultural industry
Variable recipe Oil, (animal) food industry
Divergent BOM/by-products Oil, forestry Unit of measures/batch problem Fine chemicals, drugs
Safizadeh et al. (1996) list the extreme demand characteristics, their required competitive priorities and process type attributes typical for job shops and continuous flow shops. Ashayeri et al. (1996) makes a similar comparison, divided into four categories; Relationship with the market, The production process, Quality and Planning &
Control.
Finch and Cox (1988) explore planning and control in six process industry companies with the goal of describing how planning and control functions should be accomplished in process industry firms. They identify factors that influence planning and control functions and how they are related to planning and control. Among other they identify type of bill of material as important for planning system requirements. The divergent bill of material is usually used for basic producers and converters in which category most process industries end up in. The divergent bill of material refers to the increasing numbers of products produced from a single raw material, often described as a V shape. Characteristics frequently emphasized in literature as present in process industries have been divided into three categories; production and raw material, sales and distribution and financial. These are summarized in Table 4.
6
Table 4. Process industry characteristics.
Category Process Industry Characteristics
Production and raw material • Divergent BOM/co- and by-products • Variable yield from raw material • Variable quality
• Low flexibility • Product layout • Specialized equipment • Fixed routings
• Variable yield from process • Variable recipe
• Long setup times / High change-over cost • Sequence dependent setup times • Energy intensive
• High degree of automation • Capacity constrained
Distribution and sales • Commodity products
• Few products • High volumes • Long life cycles
• Co-/by-product demand must be balanced • Low variance in demand
• Relatively high transport costs
• Market emphasis on product availability and price • Efficiency as competitive priority
• Consistent quality as competitive priority • Low unit cost as competitive priority • Timely delivery as competitive priority
• May have large demands for intermediate products
Financial environment • Capital intensive
• High fixed cost
• Long lead times for capacity increases
Several of these characteristics are commonly, but not exclusively, found in process industries and the characteristics themselves constitute an interesting area to study. Even if different sets of characteristics are found in different process industries they still set conditions that are common for many process industries.
1.1.3 Production and Raw Material Characteristics
Several of the characteristics for process industries are inherent from raw material properties and also affect the design of the production process. Some of these will be discussed in this chapter as well as what effect they have on planning of production and distribution.
One of the traits of process industries is that they are positioned at the start of the transformation process close to the raw material mixing, separating or forming it into products often used for further transformation. Process industries hence become dependent on the properties of these materials in varying degree. Since the raw material is acquired from natural sources, such as mining operations, its content can vary very
7
much depending on where it originates from. For instance crude oils from different oil fields have very different proportions of hydrocarbons and content of impurities such as sulphur. Variations in amount of desirable product yielded from a specific production process create uncertainties that have to be considered in the planning stage. Variable quality of the raw material is a similar source of uncertainty that has to be taken into account. Variations in moisture, acidity, colour, viscosity or concentration of the active ingredient are all examples of quality variations affecting the production process and planning. (Rutten and Bertrand, 1998; Taylor et al, 1981)
In many process industries recipes can be varied to obtain different proportions of output or to alter the properties of the output to fit current product quality requirements. Recipes can also be used to minimize material cost by considering changes in raw material prices or variations of raw material availability (among others Akkerman et al 2010; Floudas and Lin, 2005; Fumero et al, 2012 and Rutten and Bertrand, 1998).
Another characteristic that certainly separates planning in process industries from manufacturing or assembly industry is the divergent bill of material or the existence of co- or by-products resulting from production processes separating raw material into several products based on raw material content. Co-products are products simultaneously generated from a production process with similar value to the producer and all both be the trigger of production. By-products are products generated as a side effect from producing the product that triggers production but can still bring a significant value. Apart from co- and by-products waste products with very low or even no value can be generated. These intermediate products are sometimes further mixed or processed to create a larger variety of final products with specific properties.
A divergent supply chain could be characterised by both an increasing number of physical locations as well as an increasing number of products downstream. Most supply chains contain a distribution network resulting in an increasing number of locations. In some cases slight variations of the same product also create an increasing number of products. What will be referred to as divergent henceforth is the increasing number of products derived from splitting a single raw material into several products and more specifically when raw material content dictate what co- and by-products can be obtained from a certain raw material. This will be further described in Chapter 2.1.
From a planning perspective it is important to balance the demand of the final products with the availability of intermediate products. Variation of recipes is one way of dealing with this problem. When the possibility of varying the recipe is limited the creation of co- and by-products is highly dependent on the content of the raw material. Fixed proportions between co- and by-products can create an unbalance between demand and supply of several products (e.g. Taylor et al. 1981, Duncan, 1983, and Bertrand and Rutten, 1998). Co-product demand can sometimes be balanced through affecting demand by pricing products to balance supply and demand. (eg. Gallego and Ryzin, 1997 and Elmagrhraby and Keskinocak, 2003)
8
1.1.4 Transfer Pricing and Coordination of Decoupled Supply Chains
Transfer price is the price paid by one entity of a firm to another entity of the same firm when exchanging goods or services. Transfer prices are used when separate entities of a company are responsible for their own financial results or when operations cover several taxation zones as well as most of the execution of their operations.
Herein a decoupled supply chain will denote a supply chain whose entities are partly planned and operated independently from each other. Even if the entities of the supply chain belong to the same company and have some common planning and control at mainly longer time horizons they are themselves responsible for being profitable and running the operations as efficient as possible given certain frames. Typically independence increase the shorter time frame the decision considers. When exchanging goods or services between two entities in a decoupled supply chain they to some degree act as supplier and customer towards each other.
Between two more or less independent entities in a supply chain a transfer price can increase transparency of costs and customer demand in a supply chain. Transfer prices can also work as a control mechanism to induce certain behaviour from supply chain partners. In either way the decisions of both upstream and downstream supply chain partners will be affected by the transfer prices they face.
1.1.5 Operations Research in Supply Chain Planning
Operations research has supported supply chain planning for a long time. Advanced Planning Systems (APS) have in the past years become a supplement to enterprise resource planning system, coordinating flows, exploiting bottlenecks and keeping due dates. APS benefit widely from using operations research tools to support planning in the supply chain and are in fact vital part of an APS. (Stadtler and Kilger, 2008)
Operations research is still also used for supporting specific planning problems even if not part of an integrated system as APS. Even though not specifically developed for it many tools within operations research can be used to support planning with a divergent bill of material. These tools can be used to solve real world problems as well as gain theoretical insights about relationships within the supply chain.
1.2 Scope and Objective
The scope of this thesis covers supply chain planning at tactical or mid-term level in a process industry with a divergent bill of material. More specifically the thesis will be focused on supply chain coordination using transfer pricing to balance co- and by-product demand.
This study will be performed in a set of Swedish process industry companies with the overall objective to investigate supply chain planning in process industries in general and more specifically transfer pricing coordination in a setting with a divergent bill of material.
9
2 Frame of Reference
The aim of this chapter is to explain the context in which the thesis is relevant as well as give a basic literature background for the papers it is based on. First the origin of divergent bill of material and unbalanced co- and by-product demand is explained. The second part will be devoted to different aspects of supply chain management. One view of supply chain management and its planning tasks will be the base for further distinguishing what characterizes supply chain management with a divergent bill of material. Operations research tools frequently used in supply chain management will be explained and last supply chain coordination and transfer pricing will be discussed.
2.1 Divergent Bill of Material
Several of the characteristics for process industries are inherent from raw material properties and also affect the design of the production process.
2.1.1 Raw Material Properties and Divergent Bill of Material
One of the traits of process industries is that they are positioned at the start of the transformation process close to the raw material mixing, separating or forming it into products often used for further transformation. Process industries hence become dependent on the properties of these materials in varying degree. Since the raw material is acquired from natural sources, such as mining operations, its content can vary very much depending on where it originates from or when it is acquired. For instance crude oils from different oil fields have very different proportions of hydrocarbons and content of impurities such as sulphur. Variations of amount of desirable product yielded from a specific production process create uncertainties that have to be considered in the planning stage. Variable quality of the raw material is a similar source of uncertainty that has to be taken into account. Variations in moisture, acidity, colour, viscosity or concentration of the active ingredient are all examples of quality variations affecting the production process and planning. (Taylor et al., 1981; Rutten and Bertrand, 1998) In many process industries recipes can be varied to obtain different proportions of output or to alter the properties of the output to fit current product quality requirements. Recipes can also be used to minimize material cost by considering changes in raw material prices or variations of raw material availability. (e.g. Rutten and Bertrand, 1998; Floudas and Lin, 2005; Akkerman et al. 2010; Fumero et al., 2012) Another characteristic that certainly separates planning in process industries from manufacturing or assembly industry is the divergent bill of material or the existence of co- or by-products resulting from production processes separating raw material into several products based on raw material content. Co-products are products simultaneously generated from a production process with similar value to the producer and can both be the trigger of production. By-products are products generated as a side effect from producing the product that triggers production but can still bring a significant value. Apart from co-and byproducts waste products with very low or even
10
no value can be generated. These intermediate products are sometimes further mixed or processed to create a larger variety of final products with specific properties.
From a planning perspective it is important to balance the demand of the final products with the availability of intermediate products. Variation of recipes is one way of dealing with this problem. When the possibility of varying the recipe is limited the creation of co- and by-products is highly dependent on the content of the raw material. Fixed proportions between co- and by-products can create an unbalance between demand and supply of several products (e.g. Taylor et al. 1981; Duncan, 1983 and Bertrand and Rutten, 1998). This characteristic is often referred to as balancing co- and by-product demand. Co- and by-product demand can sometimes be balanced through affecting demand by pricing products to balance supply and demand (e.g. Gallego and Ryzin, 1997 and Elmagrhraby and Keskinocak, 2003).
This phenomenon is frequently occurring in process industries as for example steel-, forestry-, chemical- and oil industry (Lasschuit and Thijssen, 2004, Viswanadham and Raghavan, 2000, D'Amours et al., 2008, Paiva and Morabito, 2009). A steel slab can be rolled into plates of different thicknesses which then can be cut into different lengths and widths. Additional processing to give the plates a variety of characteristics further increases the number of products. Timber can be cut into boards of different dimensions and quality while other parts of the tree are used for pulp or energy production. Crude oil is made up of a variety of hydrocarbons that is specific for every oil field. Distillation of crude oil into desired fractions is therefore affected by the content of the crude oil and several different fractions are always obtained, with wanted or unwanted properties. An example from the specialty oils industry is given in chapter 2.1.2.
2.1.2 Fixed Proportions Between co- and by-products in Petroleum Industry
In the case of steel industry you are rather free to decide what intermediate and final products you want to produce from a slab. Of course there are some restrictions given by the production process, and raw material input but within physical limits you are free to choose thickness and combination of dimensions to be obtained from a certain slab. When distilling crude oil you are highly restricted by the content of hydrocarbons therein. The raw material to a large extent dictates combinations of fractions and proportions of them that are obtainable. Heavy hydrocarbons can be cracked into lighter hydrocarbons in a subsequent hydrocracking process. The opposite is however not reasonable to any larger extent.Figure 2 illustrates the divergent bill of material with fixed proportions through an example with distillation of crude oil into different fractions. There is some flexibility on what fractions to extract from a crude oil. Altering which crude oil is used for distillation will change the proportion between different fractions obtained. By changing the cut between fractions it is also possible to change what fractions are obtained but for a certain set of fractions the proportions is always the same for a given crude oil. If more of one fraction is needed more of the other fractions will also be produced. If the proportions of demand of the different fractions is different from the proportions
11
obtained from the distillation process it creates shortages of some fractions while creating excess of others.
Figure 2. Divergent Bill of Material with fixed proportions.
An example of different fractions obtained from distillation of crude oils given by Bengtsson et al. (2013) can be seen in Table 5.
Table 5. Yield of fractions from different crude oils. From Bengtsson et al. (2013).
Output Crude oil 1 Crude oil 2
Refinery gas 0.0010 0.0020 LPG 0.0401 0.0056 Light naphtha 0.1389 0.0287 Heavy naphtha 0.3182 0.1330 Kerosene 0.1263 0.0920 Gas oil 1 0.2683 — Gas oil 2 — 0.3540 Vacuum gas oil 0.0926 0.2810 Vacuum residue 1 0.0146 — Vacuum residue 2 — 0.1037
2.2 Supply Chain Management
Many views exist on what a supply chain is and what supply chain management includes. This chapter aims at describing one view of what supply chain and supply chain management is with focus on processes and their planning. Additionally the special situation with a divergent supply chain is described.
A supply chain consists of all parties involved in fulfilling customers’ requests as well as all functions within those parties needed to fulfil the requests. The supply chain not only includes the flow of physical products but also the flow of information and monetary funds. (Chopra and Meindl, 2007)
Chopra and Meindl (2007) and Fleischmann et al. (2008) among others divide supply chain decisions into three phases depending on within which timeframe the decisions have impact. Different names for the phases are used but the content and time frame of
12
the phases are similar. These three phases are often referred to as strategic, tactical and operational and are described below with respect to timeframe and type of decisions made within that phase. The phases are often tied together in a hierarchical structure where the planning from the longer timeframe sets the limits for the shorter timeframe. Planning a supply chain is an extensive task on all hierarchical levels not to mention when to make them work together. Advanced planning systems (APS) is a late addition to the family of decision support systems that can aid in the planning of an entire supply chain, considering and coordinating decisions on different hierarchical levels. As is the case with supply chain management APS also do not have one clear definition but Cederborg (2010) condenses APIC’s definition to that an APS should:
1. Use advanced mathematics to perform optimization or simulation 2. Consider finite resources
3. Include at least one of the following components a. Demand planning
b. Production planning c. Production scheduling d. Distribution planning e. Transportation planning
The components listed under point number 3 above can be condensed into three “areas” to be managed, Production, Distribution and Demand. The Supply Chain Planning Matrix developed by Rhode et al. (2000) and reworked by Fleishmann et al. (2008) in Figure 3 also includes procurement as a task to be managed in the supply chain. The Supply Chain Matrix is used by Stadtler and Kilger (2008) when describing the modules in an APS. Long -Term Mid - Term Short-Term Distribution Sales • Materials program • Supplier selection • Cooperations • Plant location
• Production system • Physical distribution structure • Product program • Strategic sales
planning • Personnel planning • Material requirements planning • Contracts • Master production scheduling • Capacity planning
• Distribution planning • Mid-term sales
planning
• Personnel planning
• Ordering materials • Lot-sizing • Machine scheduling
• Shop floor control
• Warehouse replenishment • Transport planning • Short-term sales planning Procurement Production Information flows Flow of gods
13
Supply Chain tasks to be managed within a company and coordinated with external parts of the supply chain according to Rhode et al. (2000) are similar to those presented by Cederborg (2010). The tasks can be summarized as:
• Procurement • Production • Distribution
• Sales and demand management
2.2.1 Strategic / Long-term Planning
Long-term planning, often referred to as strategic planning, in supply chains typically concern decisions on design and structure of the supply chain. The objective of supply chain design is to configure the supply chain to enable the organization to maximize its economic performance over a considerable period of time (Goetschalcks and Fleischmann, 2008). The planning horizon at this level is typically three to five years or even longer. Supply chain design includes decisions such as location of manufacturing and distribution facilities, allocations of products to manufacturing and distribution facilities, choice of suppliers, transport mode along different routes, type of information system to be used etc. Related decisions affecting the supply chain design are product program, strategic sales planning, and strategic cooperation with suppliers and other supply chain partners. These decisions have impact on long term company performance and need to support company strategy. Since changes in the supply chain design are very slow and expensive companies must consider uncertainties in for example market conditions. (Goetschalcks and Fleischmann, 2008, Chopra and Meindl, 2007)
Stadtler and Kilger (2008) use the term Strategic Network Design (SND) for the APS module that supports planning over all tasks at this level. The strategic planning in an APS often utilizes other modules foremost designed for lower level decisions such as master planning modules. With aggregated data such modules provide important information to the strategic planning. A Strategic Network Design module of an APS can aid planning by incorporating optimization models when designing the structure of the supply chain. SND modules include a linear programming solver to find optimal flows in a supply chain. Some SND modules also supports mixed integer programming models, enabling them to open and close facilities within a supply chain to find the best configuration. The main functions of an SND module are according to Goetschalcks and Fleischmann (2008):
• Generating alternatives • Evaluating alternatives
• Administrating alternatives and scenarios • Reporting, visualizing and comparing results
14
2.2.2 Tactical / Mid-term Planning
Supply chain planning at mid-term level takes place within the boundaries set by the strategic planning. Time frame for decision at this level varies a lot depending on industry and company but typically ranges from three months to a year, taking seasonal variations into account. The goal of planning at this level usually is to maximize the financial contribution over the planning horizon.
Sales and operations planning determine an outline of the operations based on forecasts. Demand planning forecasting potential sales of product groups in specific regions is an important input to the master production scheduling and capacity planning performed at this stage. Inventory levels and policies are decided for manufacturing and distribution facilities along with plans for which market is supplied from which facility. Personnel planning is derived from the capacity plan to supply the appropriate workforce. Material requirements are derived from the master production to secure supply of material. Contracts can be established with key suppliers for key materials and components. Time frame is shorter than in the strategic planning and thereby uncertainties are lower and planning should utilize any flexibility from the strategic plan to optimize performance.
Modules in an APS supporting mid-term planning are according to Stadtler and Kilger (2008) mainly Demand planning and Master planning modules. One important task of demand planning is to supply information in order to make decisions on replenishment of product inventories and raw materials before actual customer orders are received.
Demand Planning modules can support this process by supplying tools for statistical
analysis of historical data. The APS also help keeping track of aggregation and disaggregation for time, geographical regions and products groups. Based on the forecasts from the Demand Planning the Master Planning has the main goal to synchronize flows in the supply chain by creating aggregated production and distribution plans. The Master Planning module supports decisions to efficiently utilize production and distribution resources as well as balancing supply and demand by considering capacities. Master Planning module can use modules designed for lover level planning such as modules for production or distribution planning to do adequate calculations when needed. To consider bottle necks in the supply chain Master Planning module often operates with aggregated data.
2.2.3 Operational / Short-term Planning
At this level all activities that are to be executed are planned and require high degree of detail. Time horizon for plans at this level is daily and up to three months and is conditioned by structure and mid-term plans. Since the level of detail at this level is very high actual tasks to plan varies very much depending on the conditions set in earlier stages. Sales or demand planning must be executed and differ depending on type of manufacturing environment but has to consider how to supply customers with their demanded products in a timely manner. How and when to replenish warehouses and raw material stock or deliver to customers on product and location level has to be
15
decided. Manufacturing planning and control activities such as lot sizing and machine scheduling has to be performed with respect to capacity constraints.
Actual execution of supply chain operations has been left out here since focus is on planning but it is sometimes included as for example in Chopra and Meindl (2007). Typical APS modules used are Purchasing and Material Requirements Planning,
Production Planning, Scheduling, Distribution Planning, Transport Planning, Demand Planning and Demand Fulfillment and Available To Promise. Several tools to tackle these
kinds of problems have since long been readily available within operations research. The difference now is that they are more integrated with the APS than has been the case with their predecessors.
2.2.4 Managing Divergent Supply Chains
As described earlier a divergent flow within a supply chain can arise from either an increasing number of products or packaging as well as an increasing number of physical locations. What will be referred to as divergent herein is the increasing number of products derived from splitting a single raw material into several products and more specifically when raw material content dictate what co-and by-products are obtained from a certain raw material.
The planning tasks to be managed in a supply chain listed in chapter 2.2 usually corresponds to an activity including processing, moving or packaging of products and thereby increasing the number of products. Each of these tasks can be said to make up points of divergence. A point of divergence is an activity that increases the number of products from a planning perspective and can hence include products with different properties as well as the same product located at different places or in different packaging. Figure 4 illustrates the increasing number of products through the supply chain, starting with one or a few raw materials processed in one or several steps, each step increasing the number of products.
The general planning task Purchasing and the purchased raw material itself does not constitute a point of divergence since the actual increase in number of products will not occur until the raw material is processed. However the raw material will be considered a point of divergence since the properties of the raw material will affect what products can be obtained through the production process. The possibility to use different raw materials as input also has an effect on what intermediate products are obtained. Therefore raw material is also included as a point of divergence in Figure 4. Production can have several stages represented by several processes each of them potentially a point of divergence. Distribution planning task involves internal distribution among production and storage facilities as well as distribution to customers in coordination with the sales planning task. At these two last stages the increasing number of products is inherent from mainly different physical locations and different packaging and sizes.
16
Figure 4. Increasing number of products with points of divergence with general planning tasks.
Kallrath (2002) give an overview of the current state-of-the-art planning and scheduling systems in process industries that have developed the last 20 years. These systems are to a large extent based on operations research tools to support planning. Whether the concept of material requirements planning (MRP) works in process industries with divergent bill of material and continuous flows or not have been debated with arguments on both sides. ERP (Enterprise Resource Planning) systems based on MRP or MRP II (Manufacturing Resource Planning) are still used in many companies to support planning (Stadtler and Kilger, 2008). Some literature (e.g. Nelson, 1983 and Duncan, 1983) argue that MRP structure do not fit process industries at all while for example a study of six process industries by Finch and Cox (1987) shows that MRP cannot be rejected as unusable for all process industries. The basic structure of MRP however does not have a perfect match with divergent bill of material as is one of the basic arguments Powell and Robertson (1982) bring forth when arguing for the benefits of a customised decision support system in process industries. Moreover MRP does not really fit with continuous flows that are present within many process industries.
2.3 Operations Research in Petroleum Supply Chain Management
Within operations research several tools with well developed solution methods already exist that can potentially be used to deal with balancing co- and by-product demand in a divergent supply chain. For the petroleum industry in particular there are some commonly used problem formulations that can be of use. Figure 5 is an adaption to petroleum industry of the more general activities presented in Figure 4. The refining activity could further be divided into several company specific sub activities. Each of these activities represents points of divergence but also present opportunities to balance co-and by-product demand if the preconditions are correct.
Raw materials Production process 1 Production process n Sales and distribution of final products Internal distribution Number of products
. . . . .
17
Operations research tools that can be used to address these problems will be described for each of the supply chain planning tasks. Several of these formulations are of cause already used for supply chain planning problems in process industries as well as in other industries. Operations research is frequently used in planning in the petroleum industry (e.g. Bodington and Baker 1990, Cooper and Charnes, 2002 and Iachan, 2009) and this chapter will focus on operations research tools frequently used in the petroleum industry and describe how they can be used to balance co- and by-product demand in divergent supply chains.
Some operations research planning formulations available range over more than one of the supply chain planning tasks but the tasks will be the starting point to finding operations research tools to balance co- and by-product demand with a divergent bill of material. It is only natural that the operations research formulations overlap between the different supply chain planning tasks since the planning of one affects the other. Therefore the tasks illustrated in Figure 5 will be condensed to three areas; Procurement
and production, Distribution and Sales.
2.3.1 Procurement and Production Planning Tasks
Procurement and production planning tasks are considered together as they are too closely connected to be managed separately from an operations research perspective. Procurement as a separate activity is not considered but rather the choice of raw material to use considering what is available.
Process Adjustments
In some cases the production process, for example chemical reactions or distillation process, can be altered to control what intermediate products are obtained. As an example a certain crude oil can be split in different ways depending on what intermediate products are currently desired as is illustrated in Figure 2. Simply put it is possible to control what hydrocarbons goes into which fraction. It is however not
Raw materials Refining Blending Sales and distribution of final products Internal distribution Number of products
18
possible to send a lighter hydrocarbon to a heavier fraction in this process but rather at which weight of the hydrocarbons to draw the line between fractions. In other cases different secondary raw materials can be used depending on what by-products are desired from the process.
Li et al. (2003) and Li and Hui (2007) employ marginal value analysis when deciding what fractions to produce. Both consider fixed demand and prices and no dynamic demand.
Raw Material and Intermediate Products
The yield of different intermediate products is different depending on the content of the raw material processed. The ability to use different raw materials in different proportions can hence help balancing the proportion of desired intermediate products obtained from the first stage in the productions process to better match their demand. Operations research formulations exist for this type of problem with different objectives depending on type of decision to support. Common objectives for this type of formulations are minimizing raw material cost or maximizing value of produced intermediate products.
This kind of problem formulations can advantageously be combined with blending problems because of their high dependency.
Blending Problem / Recipes
One of the most classical operations research planning problems within petroleum industries is the blending problem. Blending problems are closely connected to the formulations in previous section with very similar formulation and objectives. From a set of raw materials or intermediate products final products are to be blended with the objective of minimizing material cost or maximizing contribution or revenue. A final product can be blended using different recipes with different proportions of different raw materials or intermediate products (Audet et al., 2004). Common constraints for this kind of formulations are; properties of final products, demand of final products and limited availability of raw material or intermediate products.
These problems are often referred to as blending or pooling problem and can sometimes be solved using linear programming but sometimes require non-linear formulations. An interesting example of a pooling in the food industry is given by Akkerman et al. (2010). They show how a reduced set of intermediate products can be used to reduce inventory costs when blending final products using different recipes. Using different recipes to blend intermediate products to final products present an opportunity to potentially use this type of formulation to balance co-product demand.
2.3.2 Distribution
Distribution is a complicated planning task faced by most industries. For most process industries the challenges differ in the respect that transport cost often constitutes a relatively large part of the total product cost. From a divergent point of view the issue at
19
hand is a physical supply chain with an increasing number of locations to store products in the further downstream you come and is commonly found in most industries. Depending on planning level typical planning problems are, where to place storage facility, how much of each product to store at each facility and when and how to replenish those inventories.
Operations research can supply good tools for these kinds of problems and are usually implemented in APS modules for distribution planning.
2.3.3 Sales and Optimal Pricing
An area that has been present in service industries such as airline and hotel for a long time (Bitran and Caldentey, 2003) and lately has rendered lots of attention in a wider spectrum of industries is revenue management. Revenue management aims at increasing revenues without necessarily increasing sales volumes by setting different prices for different customer segments according to their sensitivity to price. Apart from price differentiation between customer segments common tools in revenue management is bundling, overbooking and dynamic pricing based on time. In short revenue management aims at selling the right product to the right customer at the right time. Related to and part of revenue management is pricing of products to affect sales volumes. Important in both revenue management and pricing is the identification of customers’ sensitivity to price or their demand functions.
There is a difference between long term strategic pricing, aiming to position your product in a marketplace, and the short term opportunistic pricing aiming to utilize available capacity as efficiently as possible. Common for the two is tough to maximize profit by finding the optimal level of price and quantity. In pricing optimization the objective of the optimization model is to maximize contribution margin, i.e. revenue minus incremental cost.
Expressed in general terms the profit of a company is equal to its revenue minus costs, where revenues are a function of quantity and price while costs are a function of produced quantity. Assuming the demand function can be identified the profit function can be expressed as a function of the produced and sold quantity according to equation (1). (e.g. Samuelson and Marks, 2006 and Salvatore, 2007)
F(Q) = R(Q) – C(Q) (1)
Profit maximizing volume is reached when the derivative of (1) is zero expressing that the marginal revenue minus marginal cost is equal to zero. With the marginal revenue (MR) being the added revenue for selling one more unit at each quantity Q and marginal cost (MC) being the added cost for producing and selling one more unit at each quantity Q. Extending to several products with independent marginal cost, MC(Qi), the same relation as in must be valid for all products i giving (2).
20
2.4 Transfer Price Coordination and Collaboration in Supply Chains
Centralized and decentralized planning and control both have their advantages. Central planning for example is taking advantage of the overall picture while decentralized planning is more flexible to take advantage of local knowledge. Transfer pricing is one way of trying to coordinate a decoupled supply chain by setting transfer prices in a way to give all parties incentives that is aligned with the overall objective. Transfer price is the price for products or components sold by one semiautonomous division to another semiautonomous division of the same enterprise. Here transfer price will be used to denote a price between two closely collaborating supply chain partners whether they belong to the same organization or not.
In an early production-pricing model Whitin (1955) describe optimal pricing in combination with EOQ lot-sizing. Jeuland and Shugan (1983) describe joint production-pricing decision for coordination of a decoupled supply chain involving not only the marketing perspective but also production and inventory. They model a case consisting of a single manufacturer supplying a single product to a single retailer that face a deterministic and price sensitive demand. Jeuland and Shugan (1983) argue that a quantity discount to the retailer will give incentive for the retailer to price its products in a chain optimizing way. In a critical comment Moorthy (1987) disagree with the interpretation of the results and argue that this is only valid when marginal cost is declining. This concave cost function arises in cases with a quantity discount as for example the EOQ setting in Whitin (1955). Further he argues that the same coordination can be achieved with a simple two-part tariff consisting of a fixed payment and a transfer price equal to the marginal cost of the manufacturer. The fixed portion suggested by Moorthy (1987) is also referred to as franchise fee by for example Weng (1995).
Zhao and Wang (2002) consider a joint production-pricing problem in a case where the manufacturer has outsourced distribution and retailing using a convex cost function and as opposed to much of the earlier literature. Weng (1995) includes multiple identical retailers in his model while Chen (2010) extends this model with differentiated retailers. Both of the previous mentioned papers investigate supply chain coordination using concave cost functions and deterministic demand.
Samuels and Marks (2006) as well as Salvatore (2007) discuss transfer pricing between divisions within the same organization in a decentralized setting where the manufacturing and retailing divisions are not coordinated centrally. To achieve overall coordination and give incentives to the parties to act in a profit maximizing way for the entire company they distinguish between the cases when there is an external market for the product and when there is not. If there is an external market profit maximizing transfer price should be equal to current market price. If no external market exist transfer price should be set to marginal cost for the manufacturer such that total marginal cost for the retailing division becomes MC = MCmanfacture + MCretail and thereby price the products to the final customers as would be done in a centrally controlled
21
supply chain. Problems that arise in such a setting is how to distribute total contribution margin in a fair way since contribution margin received from the above setting can be lower for one of the parties than it would have been if they acted to maximize their own contribution margin and hence has to be compensated for that loss.
In the basic case where the marginal cost of producing one more unit of a product is independent of other products optimal transfer prices between two supply chain partners that maximize the joint contribution margin is the marginal cost. If each echelon in the supply chain set the transfer price to the marginal cost it will give incentive to the entire chain to price the final product to the final customer to maximize total contribution margin. Identifying marginal cost might however not be very straight forward and supply chain partners might set transfer prices differently for local gain, sacrificing overall contribution margin.
23
3 Research Challenge
Divergent bill off material is present in many process industries and balancing co-product demand has been identified as a common feature in process industries from literature as well as from a case in the petroleum industry. Due to the nature of the products produced in the specialty oils industry balancing co-product demand, as described in chapter 2.1.2, becomes an important issue. The problem arises since production of one product triggers production of other product as well enforcing a decision to either over produce some products or not supply all demand of others. Operations research offers several tools to support planning in supply chains in general as well as in the petroleum industry in particular. Even if the original idea of these tools were not to balance co-product demand some of them can still be applied to this particular problem. What separates this research from other transfer pricing research is the setting with divergent bill of material. Demand has been assumed to be deterministic and has mainly been modelled as fixed prices and volumes as well as linear demand functions. A minor addition with exponential demand functions has been investigated as well.
3.1 Scope
The scope of this thesis covers supply chain planning at tactical or mid-term level in process industries. More specifically the thesis will be limited to how operations research can support supply chain planning to balance co- and by-product demand with a divergent bill of material. The main focus will be on transfer pricing as a tool for coordination of decoupled supply chain planning.
3.2 Research Design
The background studies leading up to this research consist of a joint workshop between researchers and representatives from several process industry companies as well as an exploratory study at four process industry companies. Not all ideas and needs addressed at the workshop and the study at the companies are dealt with in this thesis but rather a selection of them. From the workshop, with the title Supply Chain Optimization, the need to investigate transfer prices as a supply chain coordination mechanism was brought forward as well as the planning difficulties that arise when faced by a divergent bill of material. The study at the four process industry companies also made it clear that the divergent bill of material is an area that requires further investigation.
Both the workshop and the study at the companies lead up to the formulation of the research objectives and research questions stated in chapter 3.2.1. The process of obtaining the objectives and research questions for this thesis as well as how the three papers are related to the questions and how they are used to answer them is described in Figure 6.
