The facility location problem in an assemble-to-order supply chain

The facility location

problem in an

assemble-to-order supply chain

Case study performed at:

Master thesis Technology Management G.J. Boezelman

Participants

University of Groningen

Faculty of economics and business Technology Management Landleven 5 9747 AD Groningen The Netherlands Wavin Group Stationsplein 3 P.O. Box 173 8000 AD Zwolle Student G.J. Boezelman Doornsteeg 14 8148 PL Lemele The Netherlands Tel. 0614581065 Student number s1653008

Company supervisor R. Hartman MSc. Coordinator department supply chain and

operational excellence

First supervisor university of Groningen

Prof. Dr. ir. G.J.C. Gaalman Department of Operations - Faculty of economics and business

Second supervisor university of Groningen

Preface

This report is written in the form of a master thesis, part of my study Technology Management at the University of Groningen. The thesis is based on my research conducted for the Wavin Group. Although it is an individual project, I could not have written this thesis without the help of others.

I would like to thank Ruurd Hartman, my company supervisor, for providing me with the opportunity to work on a challenging and interesting project, and for providing me with the opportunity to experience the daily affairs at the headquarters of Europe’s leading provider of plastic pipe systems.

I would like to thank all other people at Wavin, who in one way or another supported me during my research.

I would like to thank H. van der Meulen for becoming my second supervisor on a very short notice.

Management summary

This report contains a research executed to determine to which candidate sites market demand for Wavin’s climate control systems, also know as Tempower solutions, should be allocated. This in order to optimise performance of the supply chain for the next three years. Performance is measured based on the criteria costs, delivery time and emission levels. Costs are defined as the total financial input to create and deliver Tempower solutions. Delivery time is defined as the time required by Wavin, from receiving an order till delivering an order and emission levels are defined as the total level of CO2 emission throughout the entire life cycle of a Tempower solution.

To reach the research objective different analyses are performed. First the competitive strategy and customer needs are evaluated in order to weight the performance criteria. Second market demand is analysed to get an understanding of the external conditions Wavin has to deal with for the next three years and finally is analysed how performance of the Tempower supply chain will be influenced by allocation decisions. Based on these analyses is determined what implications allocation decisions have on overall supply chain performance.

From these analyses is concluded that the influence of allocation decisions on the total delivery time and on the total emission levels can be neglected when dealing with allocation decisions. The influence of allocation decisions on the total costs is however significant. The transportation distances from sites to customers are leading when dealing with allocation decisions. The scale advantages of one large site located in a geographic area facing low component cost, low fixed cost and low labour costs, are overruled by the high increases in transportation costs when increasing transportation distances.

Table of contents

1

Introduction ... 7

2

Research outline ... 12

2.2 Preliminary literature review ...12

2.3 Conceptual model...14

2.4 Research questions ...18

2.5 Report structure ...19

3

Competitive strategy... 21

3.1 Competitive strategy Wavin ...22

3.2 Customer needs Tempower ...22

3.3 Implications competitive strategy on allocation decisions ...23

4

Market demand... 24

4.1 Categorising Tempower systems ...24

4.2 Identifying leads...25

5

Allocation decisions and number of sites... 27

5.1 Economies of scale...27

5.1.1 Number of sites and short-term scale considerations ...27

5.1.2 Number of sites and intermediate-term scale considerations ...29

5.1.3 Number of sites and long-term scale considerations ...30

5.2 Diseconomies of scale...31

5.2.1 Span of control ...31

5.2.2 Complexity...32

5.2.3 Vulnerability ...32

5.3 Inventory pooling ...32

5.3.1 Safety stock levels Tempower ...33

5.3.2 Inventory costs ...34

6

Allocation decisions and site location ... 35

6.1 Delivery time ...35

6.1.1 Factors determining delivery time ...35

6.1.2 Contribution of transportation to delivery time ...36

6.2 Environmental impact ...37

6.2.1 Life cycle assessment...37

6.2.2 Contribution of transportation to emission levels ...37

6.3 Costs ...38

6.3.1 Component costs ...38

6.3.2 Facility and Labour costs ...38

6.3.3 Transportation costs ...39

6.3.4 Taxes ...40

6.4 Implication site location on allocation decisions...40

7

Allocation alternatives ... 41

7.1 Allocation model ...41

7.2 Alternatives based on competitive strategy ...41

7.2.1 Alternative 1 – Low cost approach ...42

7.2.2 Alternative 2 – Current approach...42

7.3 Evaluation of allocation alternatives ...43

7.4 Robustness of allocation alternatives ...44

7.4.1 Sensitivity towards transportation costs...44

8

Conclusion and recommendations... 46

1 Introduction

This report presents an allocation problem in an assemble-to-order (ATO) supply chain, in which customers have to be allocated to a pre-determined number of candidate facilities. In contradiction to make-to-stock (MTS) supply chains, in which finished goods are kept on stock, ATO supply chains keep components on stock (Akay et al., 2004). The required components will be collected, from inventory, and assembled into the final product after receiving an order. In ATO supply chains, components are not committed to a product, till an order comes in. This improves the ability of organisations to deliver customised products with short delivery times against low costs (Lee and Tang, 1997).

The allocation problem studied in this research is performed at Wavin, a Dutch organisation providing plastic pipe systems. The report is written in form of a master thesis, part of the master Technology Management at the University of Groningen. This chapter provides an introduction about Wavin, its products and the research motivation.

1.1 Company introduction

Both Royal Dutch Shell and the WMO sold its shares in Wavin in respectively 1999 and 2005, which resulted in Wavin’s stock listing in 2006. Currently Wavin has offices in 28 European countries, with manufacturing plants in 16 of those. Through the Wavin Overseas network of agents and partners, Wavin products are also available in Asia, Africa, Latin America, the Middle East and North America. 1

1.2 Organisational structure

Wavin manages its activities spread out over Europe by a multidivisional structure. Wavin is organised around six independent divisions which are self-contained. Each of the divisions focuses on a specific geographic area, those are: UK and Ireland, North West Europe, Nordic Europe, Central and Eastern Europe, South East Europe and South West Europe. In figure 1 is mapped which European countries belong to the different divisions.

Figure 1 Organizational structure Wavin

The activities performed by the divisions are overseen by Wavin’s corporate headquarters, located in Zwolle the Netherlands. The staff at corporate headquarters is responsible for overseeing and coordinating activities of the divisions. Wavin’s corporate staff deals with different functional areas. The main functional areas are

marketing and sales (SBU’s) and research and development (Technology and Innovation) see figure 1.

1.3 Products

In general Wavin’s focus is on the two distinct market segments of “building and installation” and “civil and infrastructure”. Wavin provides above ground plastic pipe systems for the building and installation market and below ground plastic pipe systems for the civil and infrastructure market.

Figure 2 Above and below ground pipe systems

Figure 2 shows that above ground pipe systems can be installed in buildings to serve different purposes. Like transporting hot or cold water, collecting soil and waste water and heating or cooling buildings. Further figure 2 shows a below ground pipe system, consisting of pipes, inspection chambers, and manholes. Below ground pipe systems are responsible for the collection and removal of soil and waste water.

1.4 Research motive

wholesaler or merchant and build a pipe system out of it. This is different for Wavin’s sales of climate control systems, also known as Tempower solutions. With the sales of Tempower solutions, Wavin deals directly with individual orders placed by plumbers. After receiving an order, required components are collected, pre-assembled and send as one package to the right building site. A schematic overview of the supply chain responsible for these activities is given in figure 2.

Wavin site Wavin production facilities Assembly facility(s) DC Component suppliers Components Pre-assemblies Plumber

Figure 3 Schematic overview Tempower supply chain

The schematic overview shows that both, Wavin production facilities and external component suppliers, are responsible for the production of components. Produced components are shipped to a distribution centre (DC), where components are stored. Required components for the “pre-assemblies” are being moved from the DC to the assembly facility. When the “pre-assemblies” are finished they will be moved back to the DC and will be packed together with the remaining components of the same order. The packed order will be shipped in one or more trucks to the customers building site. The DC and assembly facility are located at the same location, which will be referred to as Tempower site.

Figure 4 Market areas and site locations

2 Research outline

The different aspects dealt with in this research outline are partly gained from the “blokkendoos model” of De leeuw (2003). The “blokkendoos” provides aspects which should be considered when executing a research.

2.1 Research objective

Based on the research motive given in previous chapter, the following research objective is formulated:

Determining how Wavin can optimise performance of the Tempower supply chain, for the next three years, by allocating customers to candidate sites.

2.2 Preliminary literature review

In the field of operations and supply chain management an extensive body of research exists on mathematical models dealing with the problem of allocating customers to a pre-determined number of candidate facilities. Melo et al. (2009) and Ozsen et all. (2009), conducted studies in which they give a clear overview about the existing problem situations in which such allocation models are used. The information presented in this preliminary literature review is partly based on these studies.

facility (single-sourcing). Models dealing with the CFLP under the conditions of multi-sourcing use continuous assignment variables and models dealing with the CFLP under the conditions of single-sourcing use integer assignment variables.

The models described above have some commonalities, such as the deterministic parameters (like costs, demand, distances etc.), the single time period, single product, the same facility types and the allocation decision of demand to pre-determined candidate facilities (Melo et al., 2009). In practice these models are not applicable to all problem settings, therefore a broad range of extensions to the models given above can be found in literature. An important extension to the models is the introduction of time considerations and uncertainty. Parameters like customer demand and costs are uncertain and will vary through time. Therefore the models are extended with stochastic parameters. Another method to deal with variations trough time is the use of multiple-period allocation models. In these models a time line is divided into a few time periods. Besides the extensions described, through time, different parameters are introduced into the allocation models. In general these parameters can be divided into three categories; these are scale effects, pooling effects and site characteristics. Scale effects or economies of scale refer to the cost advantages gained when organisations increase their output what will result in a reduction of unit costs. Pooling effects refer to the cost advantages obtained due to a reduction of inventory locations and site characteristics refer to parameters like wages, taxes, rental costs, transportation distances etc., related to each individual facility. More detailed information about these parameters is given in paragraph 2.3 of this research.

Factors influencing choice of facility location Cost Labor characteris-tics Infra-structure Markets Macro environment • Labour costs • Transportati on cost • Tax incentives and tax structures • Financial incentives • Handling costs • Skilled labor • Availability of labor force • Existence of modes of transportati on • Telecommu nication systems • Quality and reliability of modes of transportati on • Proximity to customers • Proximity to suppliers • Lead times and responsiven ess • Policies of government • Industrial regulations laws • Zoning and constructio n plan

Figure 5 Parameters influencing site selection (Dermirel et al., 2009).

The parameters given in figure 4 are identified by Dermirel et al. (2009) and based on a review of studies taking site selection parameters into account2.

2.3 Conceptual model

The problem examined in this research can be defined as an uncapacitated facility location allocation problem over a single time period, in which a single product has to be allocated to six candidate facilities. To give a clear overview of the UFLP under research the problem situation is visualised by making use of a conceptual model, see figure 6.

2

Market demand

Facility location problem

Performance criteria

Higher total costs

Longer lead time Demand allocation decisions

More sites Location sites Site characteristics Higher: • Facility costs • Equipment costs • Labour costs • Inventory costs • Material costs • Taxes Higher distance to customers Competitive strategy Demand volume Lower pooling advantages Emission levels Lower scale advantages Higher emission rate Lower travel speed

Higher transportation

costs Processes and

resources Supply and demand

uncertainty + + + + + + + + + + + Demand Location

Figure 6 Conceptual model

The elements in the conceptual model and their relations are partly identified based on findings of the preliminary literature review given in paragraph 2.2. The elements and their relations are defined and discussed in the sub-paragraphs below.

Performance criteria

The main focus in this research is on the performance criteria cost, delivery time and the emission levels. These are defined as:

Costs:

Delivery time:

The time required by Wavin, from receiving an order till delivering an order at a customer’s building site.

Emission levels:

The total level of CO2 emission during the life cycle of a Tempower

solution.

Competitive strategy

The competitive strategy obtained by Wavin determines what objectives have to be reached while dealing with allocation decisions. One of the most common frameworks used by organisations to express their competitive strategy is “Porters’ framework of generic strategies”. The framework categorises competitive strategies along two dimensions. One deals with the “strategic scope” and defines the external market served and one deals with the “strategic strength” and defines the strengths of an organisation. Chapter 3 of this research presents more information about competitive strategies and discusses the way in which Wavin’s competitive strategy influences the allocation decisions to make.

Market demand

Besides the competitive strategy, market demand has influence on allocation decisions. Organisations facing predictable demand with low variations through time deal with different problems than organisations facing variable and unpredictable demand. In order to make allocation decisions for a longer time period, demand forecasts have to be made. The accuracy in which demand can be forecasted is highly influenced by the type of products organisations sell (Fisher, 1997; Huang et al., 2002; Wang et al., 2004). Chapter 4 of this research presents more information about product types in combination with demand uncertainty and gives an analysis of market demand Wavin will face during the next three years.

Allocation decisions and number of sites

Economies of scale and diseconomies of scale

The concept called “economies of scale” refers to the cost advantages gained when facilities increase their output. On the other hand “diseconomies of scale” refers to the cost increases when facilities increase their output. Economies of scale and diseconomies of scale can result out of different reasons (Hayes et al., 2005). Chapter 5 of this research gives an overview of the different reasons which can cause economies- and diseconomies of scale. Further is analysed which reasons cause scale effects at Tempower sites and is discussed what implications scale effects have on the allocation decisions to make.

Inventory pooling

Inventory considerations are adopted into allocation models, while reducing the number of inventory points is a common used way by organisations to reduce inventory cost. The basic idea behind this approach is that the standard deviation of either demand or supply decreases, when pooling inventory, which will result in a reduction of the overall level of safety stock. According to Zinn et al. (1989), pooling inventory will result in cost savings, which can be calculated by the “square-root-law”. The square-root-law states that the inventory costs will reduce proportional with the square root of the difference between the old number of inventory locations with the new number of inventory locations. Chapter 5 of this research provides more information about inventory pooling. The pooling effects in the Tempower supply chain are analysed and the implication of pooling effects for allocation decisions are discussed.

Allocation decisions and site location

Site locations have influence on each of the performance criteria dealt with in this research.

Delivery time

Emission levels

The distance between a site and its customers has influence on the emission levels made to deliver an order. Longer travel distances cause higher emission levels. In chapter 6 is discussed what influence travel distances have on the total environmental impact of a Tempower solution during its life cycle. Further is discusses what role emission levels should have when dealing with allocation decisions.

Costs

Costs can be divided into different parameters, examples are facility costs (rental costs, energy costs, depreciation, etc.), component costs, transportation costs, labour costs etc. The value of cost parameters can vary per candidate site. Chapter 6 of this research provides more information about the cost parameters which has to be considered when dealing with allocation decisions.

2.4 Research questions

In order to reach the main objective of this research the following research question will be answered:

How can Wavin optimise performance of the Tempower supply chain, for the next three years, by allocating customers to candidate sites?

To find an answer on the main research question, different sub-questions are asked. The sub-questions deal with four different research areas these are the competitive strategy, market demand, number of sites and finally site locations:

Competitive strategy:

1. How does the competitive strategy for Tempower solutions influence allocation decisions?

Market demand:

2. What is the estimated market demand for the next three years? 3. What influence does market demand have on allocation decisions?

4. What scale effects and pooling effects can be recognised when reducing the number of Tempower sites?

5. What influence do economies of scale and inventory pooling have on allocation decisions?

Allocation decisions and site location:

6. What is the influence of travel distances, on Wavin´s ability to meet required delivery times?

7. What contribution do transportation distances have to the total emission levels, during the life cycle of a Tempower system?

8. What role should delivery times and emissions levels have when dealing with allocation decisions in the Tempower supply chain?

9. Which cost parameters have to be considered when dealing with allocation decisions in the Tempower supply chain?

2.5 Report structure

Report structure Diagnose: Design: Change: Chapter 3 Competitive strategy Chapter 4 Market demand Chapter 6

Allocation decisions and site locations

Chapter 5

Allocation decisions and number of sites

Chapter 7

Allocation alternatives

Chapter 8

Conclusion and Recomendations

Chapter 1

Introduction

Chapter 2

Research outline

Figure 7 Report structure

3 Competitive strategy

In business literature a broad base of literature exist about strategy formation. The framework developed by Porter (1980) is one of the common used frameworks, by organisations, to develop a competitive strategy. In his framework of “generic strategies” Porter defined three strategies: differentiation strategy, cost leadership strategy and segmentation strategy. The strategies are developed by combining a market based view with a resource based view. Organisations adopting a differentiation strategy and a cost leadership strategy have a broad market scope, whether the first focus on unique competences and the latter on low costs. In contradiction to the broad market scope of these strategies, organisations adopting a segmentation strategy have a narrow market scope, combined with a focus on unique competences or low costs.

3.1 Competitive strategy Wavin

For most of its businesses Wavin follows a cost leadership strategy, with a focus on the entire building and installation or civil and infrastructure market. Due to developments in the market which put pressures on margins, like globalisation and competition of local competitors which are more and more able to provide customised products against competitive prices, Wavin started providing Tempower solutions. With Tempower solutions Wavin has a more narrow market scope, which shifted from the entire building and installation market to large building projects like hospitals, residential projects, office buildings and factories.

According Porter’s framework of generic strategies Wavin’s competitive strategy for Tempower solutions can be categorised as a segmentation strategy and according to the value disciplines developed by Treacy and Wiersema as a customer intimacy approach. The Tempower solutions are customised to the needs of individual customers. In order to increase market share, the focus is not only on plumbers but on “influencers” as well. Influencers are architects, engineers and investors. Investors initiate building projects, and architects and engineers are hired by investors to shape their wishes into a building plan. With focus on influencers, Wavin’s product managers assume that influencers will “push” plumbers to use Wavin’s systems.

3.2 Customer needs Tempower

To determine how products and services can deliver value to plumbers and their influencers, Wavin’s product managers identified their needs. Information is gained by local sales teams, responsible for the sales of Tempower solutions. For this research only the needs regarding delivery time, emission levels and costs are presented.

Delivery time

Environmental impact

From the three parties, investors are most interested in low emission levels. On this subject architects, engineers and plumbers will work according to the wishes of the investor. A trend in the market is recognised that investors in building projects focus more and more on environmental friendly buildings. Even though not all investors are willing to pay higher prices for “environmental friendly” products, they are pushed in a certain way by legislation.

Costs

Costs aspects are important for investors and plumbers, but the nature of these wishes are different. In most situations plumbers get a fixed price for the projects they execute, therefore they search for systems which require low efforts to install and low costs. Some years ago this was no problem for investors, as long as the systems confirmed to a certain level of quality. A trend in the market is seen however that investors are willing to pay more for a system if the additional costs they have to pay for a system will be earned back when the system is in use.

3.3 Implications competitive strategy on allocation decisions

From the analysis in previous paragraph can be concluded that low lead times and low emission levels become more important for customers. It should however been noticed that customers value environmental aspects, but are not willing to pay every price for it, without being pushes by legislation. From this analyses can be concluded that low emission rates and short lead times are becoming more and more important, but costs are still an important issue. The best way to comply with these customer needs is by focussing on projects which will both:

− reduce the negative influences on the environment and

4 Market demand

The accuracy, in which demand can be forecasted, highly depends on the products organisations sell (Fisher, 1997; Huang et al., 2002; Wang et al., 2004). A common used framework for the categorisation of products on basis of their demand patterns is the framework developed by Fisher (1997). He groups products in two categories; these are functional products and innovative products. Functional products satisfy basic needs, have a predictable demand and have long life cycles. In contrary innovative products give customers an additional reason to buy the product. This brings with it that demand for innovative products is unpredictable and that they have short life cycles. More recent Huang et al. (2002) extended the framework developed by Fisher (1997) and added the category, “hybrid products”, to the framework. Hybrid products exist out of functional components, which are combined into the final product, customized to the wishes of a customer. The final product can be seen as an innovative product with uncertain demand, but the components are functional products from which demand is predictable. In this chapter is discussed in which of the categories Tempower solutions fit and on what demand levels allocation decisions should be made for the next three years.

4.1 Categorising Tempower systems

Tempower systems can be categorised in the group of hybrid products. Tempower systems are customized to the wishes of each individual customer, but components used to assemble the systems are standardized. Due to these characteristics planners are not able to forecast demand of finished systems and manage uncertainty by holding finished items on stock. On the other hand it is possible to forecast demand patterns of required components.

Wavin provides four different types of Tempower systems, see figure 8. The different systems are developed to provide architects and engineers with design freedom and plumbers with ease of instalment. The components, from which the systems are assembled, differ per system type. Due to the difference in components demand forecasts have to be made per system type.

4.2 Identifying leads

Currently demand forecasts are based on leads, see figure 9. Leads are planned building projects for which a chance exist that Tempower solutions will be used as climate control system. Based on these leads can be estimated how many components are needed, while leads are expressed in square meters, and every square meter of a system contains the same number and type of components.

Figure 9 Market areas with leads

least 10000 m2 of each system type will be sold during the next three years. Figure 10 below contains on overview of the current leads with their sizes.

Figure 10 Current leads with their sizes

From figure 10 is determined that order sizes are highly fluctuating. Currently more than half of the leads exist out of orders less than 500 m2, but ass a result of the focus on large building projects is assumed that future orders are at least 5000 m2 or more.

Tem pow er order sizes

5 Allocation decisions and number of sites

The number of sites used by organisations to fulfil demand of a single product or product group has influence on the total costs organisations face. In this chapter an analysis is given about the role the number of sites should have when allocating demand in the Tempower supply chain. The influence of the number of sites on allocation decisions is discussed by dealing with the topics economies of scale, diseconomies of scale and inventory pooling.

5.1 Economies of scale

Organisations allocating more products or services to one facility or organisations expanding the size of facilities often use economies of scale as an argument (Hayes et al., 2005). Economies of scale can occur as a result of different reasons. Related to these different reasons the decreasing costs per unit, when increasing output, can be the result of a large variety of cost parameters. Therefore the reasons which cause economies of scale should be examined carefully, in order to determine on forehand what effects the allocation of Tempower demand to candidate sites will have. In this chapter the different reasons which will cause economies of scale at Tempower sites are examined by focussing on different time periods. According to Hayes et al. (2005), this is “the most logical way to sort out the various causal factors that underlie such economies”. Throughout this chapter the framework developed by Hayes et al. (2005) will be used in order to identify economies of scale in short-term, intermediate-term and long-term time periods.

5.1.1 Number of sites and short-term scale considerations

Short term economies of scale as a result of the disproportional increase of fixed costs are not discussed further in this paragraph, while this research deals with allocation decisions over a time period of three years. The effect of “spreading overhead costs” can however by identified on the long term as well. More information about this subject is provided in paragraph 5.2.2.

If demand will be allocated to multiple local sites instead of one or a few central sites, the chance increases that the total under utilisation of capacity is higher. Capacity of Tempower assembly facilities can be increased with batches of 80m2. This while Tempower systems, are pre-assembled in parallel workstations, see figure 11. Each workstation is operated by one worker performing all needed assembly steps, ranging from cutting pipes till pre-assembling the final system.

Workstation

1

2

3

1. Component buffer 2. Worktable

3. Finished systems buffer Figure 11 Workstation

Figure 12 Capacity increases

“Capacity (logistic)” in figure 12 refers to the workers responsible for providing the workstations with components and “capacity (workstations)” refers to the parallel workstations.

5.1.2 Number of sites and intermediate-term scale considerations

In contrast to economies of scale on short-term, economies of scale on intermediate-term are a result of the way in which processes are managed. Organisations can increase the output of processes by “increasing the size of batches being processed, thereby reducing the number of changeovers required to satisfy a given volume of sales” (Hayes et al., 2005). One way to reduce the number of changeovers is by dedicating workstations or facilities to a specific product or product group. In their work Hayes et al. (2005) identified three factors which influence the scale advantages which can be gained by reducing the number of changeovers:

− Costs when changing from one product to another

− Time related to changeovers can result in lower outputs. This can result in costs of “lost sales” or additional costs for overtime (Hayes et al., 2005).

− From the start of a batch and at the end of a batch costs can be higher as a result of “higher-than-normal errors or defects and reduced labour efficiency” (Hayes et al., 2005).

systems. It is not required to prepare workers at the workstations every time when a switch will be made to another system type. Second there is no time related to changeovers at all. Workstations can keep running, if the right components are in place. Third no differences can be seen in the number of mistakes operators make during the cycle time of one batch. In contrast a slight difference in labour efficiency exists during the assembly of a batch. The efficiency at the start of a new batch differs only slightly from the overall labour efficiency during a batch, as a result of low complexity of the assembly process. Further each order contains only one system type and orders are large, what results in large batches with a minimum size of at least 150m2. The exact difference in efficiency is not explicitly measured, but it is assumed that the difference can be neglected due to the large batch sizes and slight differences in labour efficiency. From this analysis is concluded that no economies of scale will be reached when dedicating workstations to specific system types and intermediate-term economies of scale, will not influence allocation decisions.

5.1.3 Number of sites and long-term scale considerations

Hayes et al. (2005) divide long-term economies of scale in “static economies of scale” and “dynamic economies of scale”. Static economies of scale are not a result of skills and experience which change over time, but are a result of the characteristics of equipment used. Consider for example a machine responsible for packing products. The initial investment of a machine with high capacity is higher than the initial investment in a machine with low capacity, but as a result of the higher capacity the total costs per unit of output are lower. This makes that the machine with high capacity will deliver scale advantages. Investing in other equipment or tools, in order to reduce the costs per unit of output, is currently no issue for Wavin. No machines or tools are currently available which will increase capacity of Tempower workstations. Capacity only can be increased by adding workstations or increasing operating time of workstations in use.

member facilities. Even though it is not possible to share knowledge about all small changes like small equipment adjustments or small adjustments to working procedures, it is assumed that the same learning effects will occur when making use of multi sourcing or single sourcing.

5.2 Diseconomies of scale

Besides the costs reductions which can be gained by increasing the scale of a facility, cost increases can be recognised as a result of scale increases. These negative scale effects are also known as “diseconomies of scale” (Hayes et al., 2005). In this paragraph an analysis is given about the different types of diseconomies of scale which are considered. The types of diseconomies of scale dealt with in this research are based on the work of Hayes et al. (2005).

5.2.1 Span of control

levels are required at candidate sites, no matter what allocation decisions will be made for the next three years.

5.2.2 Complexity

Even though the allocation of Tempower activities to candidate sites will have no influence on the number of management levels, managing activities can become more complex. Hayes et al. (2005) state that a difference should be made between “scale” and “size”. With scale they refer to increasing a facilities output with the same units of output, and with size they refer to increasing a facilities output with multiple dissimilar products. Wavin produces and assembles different products at the candidate sites and adding Tempower activities to these sites will result in size increases instead of scale increases. When increasing the size of a candidate site by allocating Tempower activities to it, can be assumed that scale advantages by spreading overhead, will not be gained. This while overhead costs will increase, as a result of the need for more complex process control systems, more planning activities etc. (Hayes et al., 2005).

5.2.3 Vulnerability

Organisations allocating demand to a single facility are entirely dependent on this facility and face higher risks as a result of fire, strikes etc., than organisations dividing demand over multiple facilities.

5.3 Inventory pooling

Figure 13 Safety stock and re-order point

The concept of inventory pooling is widely used to reduce the overall safety stock levels of organisations (Cai et al., 2009). Inventory pooling is the shift from a large number of decentralised inventory locations to one (or a few) centralised inventory location(s). The average demand level faced by the reduced number of inventory locations is the sum of demand faced by the individual decentralised inventory locations. The difference is that the variability of demand and material supply, measured in standard deviation, for the centralised inventory locations is much smaller than the sum of the variability’s for the decentralised inventory locations. According to Cai et al. (2009), “a high demand from one market tends to get balanced out by a low demand from another. This reduction in variability allows a decrease in safety stock and average inventory, and therefore reducing expected system cost.”

5.3.1 Safety stock levels Tempower

Components required for Tempower solutions can be categorised into two groups. For one group of components Wavin carries no safety stock, while replenishment lead times are short and variability of replenishment lead times are low. For the other group of components Wavin carries safety stock, while replenishment lead times are long and highly variable. Not carrying safety stock of these components will have a negative influence on Wavin’s ability to meet required delivery times.

safety stock levels, for Tempower components, based on experience and intuition. No standard levels of safety stock are set, but planners determine reorder levels. The reorder level is not fixed and varies trough time. It is hard to determine the actual safety stock levels for Tempower components, while the difference between safety stock and overall inventory is not explicitly made. For this research is determined that safety stock covers 10% of the average component inventory, dedicated to Tempower systems.

5.3.2 Inventory costs

According to Zinn et al. (1989), pooling inventory will result in cost savings, which can be calculated by the “square-root-law”. The square-root-law states that the inventory costs related to safety stock will reduce proportional with the square root of the difference between the old number of inventory locations with the new number of inventory locations (Zinn et al., 1989). Croxton and Zinn (2005) deal with the inventory considerations as well. In their study they proved that including inventory considerations in network design will result in a network with less inventory locations. Further they implicate that this effect is larger with a high number of inventory locations in a network. In a more recent study Cai et al. (2009) revised the square root law. They state that cost related to one centralised inventory location are not always lower than when dealing with more decentralised inventory locations. While no exact information is found to determine the precise advantages which will be gained when reducing the number of inventory locations in the Tempower supply chain, different scenarios are evaluated. Ranging from no cost advantages at all, till costs advantages which are calculated by the square root law.

5.4 Implications number of sites on allocation decisions

6 Allocation decisions and site location

The geographic areas in which sites are located have influence on the performance criteria delivery time, emissions levels and costs. In this chapter an analysis is given about the role site locations should have when allocating demand in the Tempower supply chain.

6.1 Delivery time

This paragraph provides insight in the factors which determine the time required by Wavin to fulfil an order from start to finish. The information is used to determine the role of transportation time in meeting “customer lead time”. Customer lead time can be defined as “the amount of time allowed to fulfil a customer order from start to finish” (Hopp and Spearman, 2000). Customer lead time determines how much time is available in order to process an order and deliver it at a customers building site. If the delivery time required by Wavin is less than or equal to the customer lead time, orders can be fulfilled on time.

6.1.1 Factors determining delivery time

The delivery time for Tempower solutions is made up out of the following factors: − Time for picking components

− Time waiting in batch, before components move from DC to assembly facility

− Time for moving components from DC to workstations at assembly facility

− Queue time before workstation − Process time at workstation

− Time waiting in a batch, before pre-assemblies move back from the assembly facility to the DC

− Time for moving pre-assemblies from the assembly facility to the DC − Queue time before packing

− Time for packing and loading

If Wavin would have infinite capacity and would face no variability in demand, determining the total delivery time on forehand would be simple. This is however not true. The actual time for picking, moving, assembling, packing, loading and transportation is fixed with only slight variations. Unfortunately these factors do not determine delivery time alone. The total delivery time is mainly determined by the queue time before assembly stations. The length of queues is highly dependent on the fluctuations in demand and available assembly capacity.

6.1.2 Contribution of transportation to delivery time

While only limited information is available yet about the arrival time of orders and their size, it is hard to determine the actual queue time and the contribution of transportation time to delivery time. Possibilities exist however to manage uncertain queue times as a result of demand fluctuations. These possibilities are related to average customer lead time and characteristics of the assembly process. It is estimated that customer lead times for building projects have an average of five months, with deviations from two weeks up to a year. These relative “long” customer lead times, combined with high flexible workstations towards capacity, provide Wavin with time to adapt capacity to demand in a short timeframe. By increasing or decreasing the total capacity of processes, it is possible to manage the time Wavin needs to fulfil orders from start to finish. The high flexibility of assembly capacity is mainly caused by two reasons. The first is that capacity can be increased in a short time frame by adding workstations. Workstations contain a worktable and standardized tools for which low investments are required and which can be ordered with lead times less than a week. Further Wavin is able to attract flexible labour in a short timeframe, at each of the candidate sites. The second way in which capacity can be increased is by increasing operating time of workstations. A common used way by organisations to increase operating time is covering day and night by different shifts.

6.2 Environmental impact

During the life cycle of a product from extraction of raw material till disposal and reuse, the environment is influenced in different ways. Think for example of CO2

emissions which contribute to global warming. In this paragraph is discussed which contribution the transportation of Tempower systems from Tempower sites to customers has on the total environmental impact of a Tempower system during its life cycle.

6.2.1 Life cycle assessment

Life cycle assessments (LCA) are common used tools by organisations to identify the environmental impact of a product. By executing an LCA the environmental impact of a product in each phase of its life cycle from raw materials till disposal are identified. Figure 14 below gives a schematic overview of the lifecycle of a Tempower system.

Figure 14 Stages in the life cycle of Tempower solutions

6.2.2 Contribution of transportation to emission levels

research performed by Jorgensen et al. (1996) is based on the LCA’s of four products: coffee milk packaging (glass and cardboard), TV’s, glass fiber mats and flax fleece. Jorgensen et al. (1996) conclude that, in general, transportation is the source of 5% to 10% of the total environmental impact during the life cycle of a product.

Based on the study performed by Jorgensen et al. (1996) is assumed that transportation will be the source of 5% to 10% of the total environmental impact of a Tempower system during its life cycle. These percentages are considered as low. Therefore transportation distances will be neglected when allocating customers to candidate sites.

6.3 Costs

For this research the total costs of Tempower systems are divided into the cost factors component cost, facility costs, labour costs transportation costs, inventory costs and taxes. In the sub-paragraphs below these costs factors are evaluated and discussed in more detail.

6.3.1 Component costs

As mentioned in the introduction of this research Wavin deals with two groups of component suppliers. One group of suppliers exists out of Wavin’s own production facilities and one group exists out of external component suppliers. It is for Wavin not possible to deliver internal produced components, at the same price, to each of the candidate sites. One of the reasons of these price differences is the varying distance between production facilities and candidate sites. Therefore Wavin developed a pricelist for inter-company trade. The costs of components bought from external components vary throughout Europe as well. Therefore component costs are considered when making allocation decisions.

6.3.2 Facility and Labour costs

area as well. Figure 15 is added to give an impression of the differences in labour cost of low skilled workers per geographic area and the trends of the past years. The data about labour cost per geographic area is gained from euro-statistics.

Figure 15 Labour costs for low skilled workers (€/hour)

6.3.3 Transportation costs

The total transportation costs which will be made when delivering a square meter of Tempower system to a customers building site are influenced by three factors. These are the amount of Tempower systems which are put into one truck, the travel distance and the costs per km. Figure 16 contains a list of the travel distances from market areas to the candidate sites.

Figure 16 Travel distances from Tempower sites to market areas

The distances in the table are calculated by taking the average distance from the candidate sites to the leads per market area. Transportation cost can be calculated with the formula:

Tra vel distance from Te mpow e r site to marke t area (km)

( cost (€/km) x travel distance (km)) / number of units per truck ( m2_{) = Transportation}

costs (€/m2)

The transportation cost varies per geographic area, as a result of varying fuel costs and labour costs of drivers. Further, in most situations, not all trucks carrying Tempower systems towards a building project are fully loaded. In order to calculate transportation costs for this research is assumed that the average utilisation of trucks, as a result of varying order sizes, will be 95%.

6.3.4 Taxes

Each of the candidate sites is located in the Value Added Tax Area (VATA) of the European Union. Countries within this area follow the tax harmonization rules collectively set by the members of the European Union and no taxes exist if products will be imported or exported within the VATA. Only products leaving the VATA are considered as in- or exported. From the Tempower solutions sold only export taxes have to be paid over products sold to Croatia, because Croatia does not belong to the VATA. The export taxes related to the exportation of products to Croatia will not influence allocation decisions, while export taxes are the same whatever candidate site is used3.

6.4 Implication site location on allocation decisions

From the analyses presented in this chapter is determined that delivery time and emission levels can be neglected when dealing with allocation decisions. Further differences in costs are identified between the different candidate sites. Chapter seven of this research will provide more insight in the implications these costs differences have on the allocation decisions to make.

7 Allocation alternatives

When considering possible allocation alternatives, without taking inventory pooling and economies of scale into account, it would be sufficient to allocate demand per market area. Each of the six market areas can be allocated to one of the six candidate sites, what would result in forty two allocation alternatives. Based on the analyses in chapter 5 is however determined that pooling and scale effects have to be considered when dealing with allocation decisions in the Tempower supply chain. This increases the number of alternatives to choose from, while the number of sites in use has influence on performance as well. From the different analyses performed in previous chapters is concluded that costs should be leading, when dealing with demand allocation decisions in the Tempower supply chain. Both emission levels and time related to travel distances will not have an active role, when dealing with allocation decisions.

7.1 Allocation model

From the different analyses performed in previous chapters is determined that multiple cost parameters have to be considered when dealing with allocation decisions. A spreadsheet model is developed; in Microsoft excel, to evaluate the different allocation alternatives based on costs. While multiple alternatives are possible, the solver in Microsoft excel is used to evaluate each of the possible alternatives, this in order to find the one with lowest costs. By running the solver the cost function containing the following cost parameters will minimised:

Z =

The different parameters are discussed earlier in this report, see chapter 5 and 6. The model developed in excel is deterministic, but it is possible to incorporate stochastic behaviour of parameters by developing different scenario’s.

7.2 Alternatives based on competitive strategy

based on a low cost approach and the second alternative represents Wavin’s current situation. The alternative representing the current situation is added to provide decision makers with insight in the advantages which can be gained by reconsidering allocation decisions.

7.2.1 Alternative 1 – Low cost approach

Figure 17 visualised the allocation alternative representing the low cost approach, by making use of binary assignment variables. This alternative is found by minimising the objective function.

Figure 17 Alternative 1 – Low cost approach

The vertical axe in the table presented in figure 17 contains the site locations and the horizontal axe the market areas. If the binary variable “1” is filled out on the intersection of a market area with a site location, it states that the site representing that specific row will be responsible for the market area representing that specific column. Based on this information can be conclude that the site in The Netherlands will be responsible for the Dutch market, the site in France for the French market, the site in Germany for the German market, the site in Hungary for the Hungarian and Croatian market and the site in Slovakia for the Slovakian market.

7.2.2 Alternative 2 – Current approach

Figure 18 Alternative 2 – Current approach

7.3 Evaluation of allocation alternatives

The concepts are evaluated based on an average market demand of 40000 m2 of Tempower systems, per market area, for the next three years. Figure 19 gives an overview of both alternatives, evaluated against costs.

Concept cost evaluation : Demand 40000 per market area

0 1000000 2000000 3000000 4000000 5000000 6000000

Alternative 1 - low cost approach Alternative 2 - current approach Concepts T o ta l c o s ts ( € ) Inventory costs Equipment costs Fixed costs Component costs Labour costs Transportation costs

Figure 19 Total costs per allocation alternative

From figure 19 can be concluded that, the costs related to alternative 1 are approximately six hundred thousand euros lower, over a timeframe of three years, than the costs of alternative 2. It is interesting to see that the total cost of alternative 1

are lower, even though labour costs, equipment costs, fixed costs and inventory costs are higher compared to alternative 2. The lower labour cost of alternative 2 can be explained by the relative low labour costs of workers in Hungaria, compared to labour costs of workers in other countries where other sites are located. The low equipment costs, fixed costs and inventory costs of alternative 2 are mainly the result of a combination of different reasons causing economies of scale and pooling advantages.

The higher total costs of alternative 2 can be explained as a result of the higher distances from Tempower sites to customers. Costs related to the transportation of Tempower systems are relative high. Increasing distances between customers and sites results in fast increasing transportation costs. Besides the lower transportation costs of alternative 1, component costs of alternative 1 are lower as well. This can be explained as a result of the lower travel distances from component suppliers to Tempower sites. The difference in component costs between both alternatives is however small. This can be explained as a result of the relative low costs required for the transportation of components. Components require less truck capacity than pre-assembled Tempower systems.

7.4 Robustness of allocation alternatives

In this paragraph stochastic behaviour of transportation costs is introduced, in order to determine the robustness of the alternatives presented in previous paragraph. Only the sensitivity of the alternatives to transportation costs is measured, while transportation costs is the most dominant cost factor.

7.4.1 Sensitivity towards transportation costs

Sensitivity towards transportation costs 3900000 4200000 4500000 4800000 5100000 1 1,1 1,2 1,3 1,4 1,5 Transportation costs (€/km) T o ta l c o s ts ( € )

Alternative 1 - low costs approach

Alternative 2 - current approach

Figure 20 Sensitivity towards transportation costs

In figure 20 can be seen that the costs of alternative 1 are not decreasing linear. When transportation costs reach the level of 1.1 €/km the total costs of alternative 1 will decrease faster. This non linear behaviour is the result of the different allocation decisions which should be made. In this situation the allocation alternative with lowest costs is not the one presented in paragraph 7.2.1, in which customers are allocated to the nearest sites. Under these conditions the allocation alternative should be chosen in which the site in The Netherlands is responsible for the Dutch and France market, the site in Poland for the Polish and German market and the site in Slovakia for the Hungarian, Croatian and Slovakian market, see figure 21. It should however be noticed that it is highly unlikely that that transportation costs will get below the level of 1.1 €/km.

8 Conclusion and recommendations

Based on the different analyses performed is concluded that the influence of allocation decisions on the total delivery time and on the total emission levels can be neglected when dealing with allocation decisions in the Tempower supply chain. The influence of allocation decisions on costs is however significant. Further is determined, based on the evaluation of different allocation alternatives, that the scale and pooling advantages gained when reducing the number of sites to one large site located in a geographic area with low labour costs, are overruled by the increasing transportation costs. Based on these results is concluded that optimised performance of the Tempower supply chain, by allocating customers to candidate sites, can be reached by allocating customers to the nearest candidate sites.

When focussing on assemble to order supply chains in more general, in which products are assembled on order out of standardised components, it can be assumed that making use of the model developed in this research will result in the same allocation decisions, if is complied to the following conditions:

− equipment costs are low and processes are flexible − required safety stock levels are low

− the (pre-)assembled products require high truck capacity as a result of high product volumes

During the research some assumptions had to be made, while not all required data could be gained from Wavin during this research. Therefore it is advised to conduct further research in order to fill the allocation model with more reliable data.

9 References

Books:

Hayes, R., Pisano, G., Upton, D., Wheelwright, S., 2005. Operations, Strategy, and Technology: Pursuing the Competitive Edge. Wiley

Hillier, F. S., Hillier, M. S., 2003. Introduction to Management Science; a modelling and case studies approach with spreadsheets. Second edition. Mc Graw Hill

Hopp, W.J., Spearman, M.L., 2001. Factory Physics; Foundations of Manufacturing Management. Second edition. Mc Graw Hill.

Jones, G.R., 2007. Organizational Theory, Design, and Change. Fifth edition, Pearson Prentice Hall.

Slack, N., Lewis, M., 2008. Operations strategy. Prentice Hall. Great Britain.

WebPages:

Wavin’s public webpage: www.wavin.com

Eurostat European commission : http://ec.europa.eu

Articles:

Akcay, Y., Xu, S.H., 2004. Joint Inventory Replenishment and Component Allocation Optimization in an Assemble-to-Order System. Manangement Science 50 No. 1, p. 99-116

Cai, X., Du, D., 2009. On the effects of risk pooling in supply chain management: Review and Extensions. Acta Mathematica Applicatae Sinica 25 p. 709 – 722

Croxton, L., Zinn, W., 2005. Inventory considerations in network design. Journal of Business Logistics 26 p 149 - 168

Fischer, M.L., 1997. What is the right supply chain for your product? Harvard Business Review

Hale, T.S., Moberg, C.R., 2003. Location Science Research: A Review. Annals of Operations Research 123 p. 21 – 35

Huang, S.H., Uppal, M., Shi, J., 2002. A product driven approach to manufacturing supply chain selection. Supply chain management, an International Journal 7 p. 189 – 199

Jorgensen, A.M.M., Ywema, P.E., Frees, N., Exner, S., Bracke, R., Transportation in LCA; A Comparative Evaluation of the Importance of Transport in Four LCAs. International Journal of Lyfe cycle analyses. 1 p. 418 - 420

Klose, A., Drexl, A., 2005. Facility location models for distribution system design. European Journal of Operational Research 162 p. 4 –29

Lee, H.L., Tang, C.S., 1997. Modelling the costs and benefits of delayed product differentiation. Management Science 43 p 40 – 53

Melo, M.T., Nickel, S., Saldanha-da-Gama, F., 2009. Facility location and supply chain management – A review. European Journal of Operational Research 196 p. 401 – 412

Ozsen, L., Daskin, M.S., Coullard, C.R., 2009. Facility Location Modeling and Inventory Management with Multisourcing. Transportation Science 43 No. 4, p. 455-472

Persona, A., Battini, D., Manzini, R., Pareschi, A., 2007. Optimal safety stock levels of subassemblies and manufacturing components. International journal of production economics 110 p. 147 – 159

Treacy, M., Wiersema, F., 1993. Customer intimacy and other value disciplines. Harvard Business Review

Wang, G., Huang, S.H., Dismukes, J.P., 2004. Product-driven supply chain selection using integrated multi-criteria decision-making methodology. International journal of production economics 91 p. 1 – 15

Wanke, P.F., Saliby, E., 2009. Whether and how inventories should be pooled. Transportation Research 45 p. 678 – 692

Whybark, D.C., Williams, J.C., 1976. Material requirements planning under uncertainty. Decision Science 7 p. 595 – 606