Decision-making in the allocation process

Allocating products to warehouses in process industry

University of Groningen Faculty of Economics and Business

Decision-making in the allocation process

J. A. Lambert 1730541

University of Groningen Faculty of Economics and Business

January 2010

Supervisors University: Supervisors AVEBE:

PREFACE

For my master thesis I was interested in an assignment in where theory meets practice. The opportunity for this was given to me by AVEBE. I gladly accepted that chance, which was very nice because it gave me the opportunity to see things in a more realistic way. The practical experience gained during this period was achieved by working with a tool that is used within the decision support system. In this case the tool used was a linear programming model. This model was constructed in Microsoft Excel with the solver add-in called Premium Solver 7.1.

Now that I am finishing my master program thanks goes in first place to my supervisors that assisted me in successfully concluding this assignment. I want to thank all of them for their time, support and feedback which helped me in a great way for finishing my master assignment. Without their support this would not have been possible.

Thanks goes also to the workers of AVEBE for allowing me to be part of their team during this final assignment for my master thesis. I also want to thank them for their time and answers to my questions.

I also want to thank Danny Dirksz for his help with the formulation of the model. Because of my less technical orientation he was of great help with the mathematical equations for the linear programming model. During this period we both once more confirmed that: Mathematics = fun!

Alex Lambert

ABSTRACT

Purpose – The purpose of this paper is to seek to investigate the way how make-to-stock products at a production process company can be allocated and stored.

Design/methodology/approach – Unstructured interviews were conducted in order to reveal factors that influence decision-making within the process. These factors were then modelled in a linear formulation manner and made use of in a linear programming model. Findings – Allocating the make-to-stock products based on their turnover ratio can provide much savings. Decision-making in this manner is facilitated and simplified because the turnover ratio is proved to be the most important influence factor.

TABLE OF CONTENTS

PREFACE ... III ABSTRACT... IV

INTRODUCTION ... 3

1 SUPPLY CHAIN ... 5

1.1: AVEBE’S SUPPLY CHAIN... 5

1.2: FOCUS OF THE PROJECT... 6

2 PROBLEM STATEMENT... 8

2.1 RESEARCH QUESTION &SUB-QUESTIONS... 10

2.2 RESEARCH SETTING &BOUNDARIES... 11

3 METHODOLOGY... 12

4 CURRENT SITUATION ... 14

4.1: THE COST DRIVER &COST GROUPS IN THE CURRENT SITUATION... 14

4.2: INITIAL SITUATION... 16

4.2.1: Allocation of the SKUs ... 17

4.3: ASSESSMENT OF THE CURRENT SITUATION... 21

4.3.1: Reviewing the current situation ... 21

4.3.2: Assessment for the LP-model ... 21

5 MODELLING THE SITUATION & DYNAMICS ... 23

5.1: THE OBJECTIVE FUNCTION... 27

5.2: THE COST GROUPS... 27

5.2.1: Handling cost... 28

5.2.2: Pallet space cost... 28

5.2.3: Transportation cost ... 29

5.2.4: VAL cost... 29

5.3: CONSTRAINTS... 30

5.3.1: Basic constraints... 30

5.3.2: Service related constraints ... 31

5.3.3: Technical constraints ... 31

5.3.4: Constraints related to practicality ... 32

5.3.5: Compatibility issues constraints ... 33

5.4: MODEL VALIDATION... 33

5.4.1: Set up in MS Excel ... 33

6 RESULTS ... 36

6.1: OUTCOMES OF THE RESEARCH... 36

7 RECOMMENDATIONS ... 46

7.1: IMPLEMENTATION... 46

7.2: CONTROL SYSTEM... 48

7.3: MONITORING... 51

7.4: UPDATE FREQUENCY... 52

7.5: GROUPING,ARRANGING AND ALLOCATING SKUS... 52

8 DISCUSSION & CONCLUSION ... 54

REFERENCES... 56

APPENDIX ... 1

APPENDIX 1: ONLY NON-FOOD BULK IN TAK... 1

APPENDIX 2: FOOD BULK IN TAK... 2

INTRODUCTION

In this introductory chapter a brief history of the company under study is given. This is followed by an outline of the thesis.

History of AVEBE

“AVEBE, founded in 1919, was at first a marketing and sales organisation for several independent Dutch potato starch co-operatives. The word AVEBE is an abbreviation of Aardappelmeel Verkoop Bureau (Potato Starch Sales Office).

For several decades AVEBE focused on sales of potato starch. However, in 40’s and 50’s of the twentieth century the still independent factories, connected to AVEBE, began

innovating by developing and producing special products (derivatives) based on potato starch. After that AVEBE expanded her capabilities in manufacturing modified starches.

In 1971 AVEBE finally was no longer only a sales office, but the factories decided to integrate their production sites into this sales office. From that time on AVEBE was formally a production and sales organisation with several thousands of farmers attached to it. The main focus was on producing potato starch, but the research and development of modified starches became more and more important.

Currently, AVEBE operates wholly owned starch processing facilities, mainly in Europe, and is selling high quality starch and starch specialties worldwide.”1

“The cooperative’s strategy, which was framed at the beginning of 2006 in the AVEBE policy plan and is validated annually, is based on three pillars:

1. Focusing on the core business: processing starch potatoes into starch and proteins. 2. Achieving cost-price leadership in the potato starch market.

3. Innovation aimed at adding value to starch and other components of the starch potato.”2

This thesis assignment is related to the second point.

1_{Source: http://www.avebe.com/}

Thesis outline

This master thesis concerns the development of a decision tool that is used by AVEBE regarding the allocation of its end products. This decision tool is used within the decision support system (DSS) at another location.

In the next chapter, the supply chain of AVEBE is described. This facilitates the

clarification of the use of the tool. It also makes clear where this assignment fits within the entire supply chain.

In chapter 2, the problem statement for this assignment is discussed. As mentioned before, AVEBE seeks to become the cost-price leader. That is why managing and minimizing costs are important matters. By strategically allocating their end products this goal can be achieved. Therefore, it is important to investigate the factors that influence the decision-making within the allocation process. In Chapter 2, the research question and

sub-questions are formulated and discussed.

The methodology for dealing with the research question and sub-questions, resulting from the problem statement, is described in chapter 3. The method followed consists of 4 steps. The first was to look at the current situation and analyse the problem. Once this was clear it was important to formulate the problem in a way that can be used within the (DSS). After this has taken place, making use of the tools becomes possible in the third step. With the help of the tool a feasible allocation plan was created. The final step was creating an implementation plan for the (feasible) allocation plan.

The chapters following chapter 3 gives detailed information about the steps described in chapter 3. This holds in that chapter 4 illustrates the current situation and complications. Chapter 5 translates the factors that play a role within the allocation process into

mathematical equations that are used within the DSS. Chapter 6 discusses the output of the tool and also the analyses carried out. In chapter 7 the implementation plan is discussed. Steps towards implementation of the feasible plan are discussed. There is also recommendation provided in chapter 7 for controlling and maintaining a correct allocation of the end products.

1

SUPPLY CHAIN

The first section of this chapter describes the supply chain of AVEBE. The tool that is constructed for this assignment focuses on a specific part of the supply chain. This specific part of the supply chain is the focus of the second section of this chapter.

1.1: AVEBE’s supply chain

The figure below illustrates the entire supply chain of AVEBE.

Figure 1 The supply chain of AVEBE

In the figure there is also a shaded area. This area indicates the focus of this project. This is further explained in the next section. First, the supply chain of AVEBE is described here. The process industry has many characteristics. For example, it contains a divergent

process is in general a continuous production. This makes it difficult to stop a production once it is started. There are many other characteristics of the process industry; however, not all will be discussed since this is not the purpose of this paper. Some other important characteristics that play a role at AVEBE are:

 Products are produced in high volume (large quantities).  There is batch-continuous production.

 Products have low value density.

The supply chain of AVEBE (see figure 1) consists of the supply of the raw material, which is potato that is processed into starch. This (wet) starch can either be dried or processed in its current form. When dried (now called native starch), it is stored in silos. There are three routes for native starch. One option is to send it directly to the customer, a make-to-order (MTO) product. A second option is to send it first to the packaging process and then store it in a warehouse. This is usually a make-to-stock (MTS) production

environment. The third option is to further process the native starch, thereby obtaining derivative products. This may seem as a preventable step. However, the reason for this “extra” step has to do with the fact that AVEBE produces in campaign cycles, meaning that AVEBE produces only during certain time in the year (thus not the whole year through). For the derivative products, there are also three possibilities from here on. (1) The derivative products can be sent directly to the customers (MTO). (2) They can also be put into larger silos and be kept as stock (MTS); bulk products. (3) Alternatively, they can be packaged and then stored in warehouses, which are also MTS products.

1.2: Focus of the project

The shaded area in figure 1 concerns the packaging and storage of the derivative products, which relates to the MTS products. Because of the low-density value of the products, the logistics costs (e.g. transportation, inventory, warehousing) costs are dominant. Reducing these costs takes AVEBE a step closer in achieving its goal in becoming the cost-price leader in the potato starch market.

The MTS products are kept in big bags, small bags or in silos. Once these finished products are put into bags and set on pallets, they are stored in the warehouses. These are the stockkeeping units (SKUs). These SKUs can be stored at the location, internal, or allocated to an external warehouse. Once stored, these are ready to be sold. Some clients request additional operations on purchased products. An additional operation can be for instance putting a special label on the bags or placing the bags on another type of pallet. These extra operations are called value added logistics (VAL). The VAL differs per bag type (big bag and small bag). That is why these processes are separated in figure 2. For example, there is the possibility that a big bag could be poured back into a silo. This depends on what is sold and thus promised to the customers.

Figure 2 Focus of the assignment within the supply chain

Deciding whether a SKU should be allocated internally or externally is an important feature within the process. This is where the use of the constructed tool plays a role in the

decision process. The process of allocating the SKUs to the different warehouses is called the allocation process.

The following chapter deals with the problem statement, meanwhile chapter 3 explains the way that this assignment was carried out.

2

PROBLEM STATEMENT

This chapter presents the problem statement for this research. The problem statement consists of the problem background, a business goal and a main research question with sub-questions. Together the answers of the sub-questions give answer to the main research question. The research setting and boundaries are also pointed out in this chapter. First, the origin of the problem is discussed.

Background

As mentioned previously, AVEBE wants to become the cost-price leader in the potato starch market. In order to do that it is important to manage the costs. One of the cost elements concerns the allocation of their SKUs. Previously, decisions regarding the allocation of the SKUs were mainly based on intuition and experience. No well-designed decision tool for the allocation of SKUs to the warehouses was available. This was somewhat troublesome for the management because decision-making in this way is very subjective. This means that no standard was used when making decisions.

To improve the decisions making process it was desired to make use of a support tool. Therefore, a research was conducted (at the location in Foxhol) which resulted in a well-defined decision support system (DSS). This DSS makes use of a linear programming model (LP-model) as a tool. The result of that research is given in table 1.

Table 1 The cost driver and cost groups of the previous research

Component Description

Cost driver Fast and slow-movers Turnover ratio of the SKUs.

Cost group

Combined shipments SKUs that are shipped together in the same container or truck.

Handling Handling of SKUs in a warehouse

location.

Transportation Delivery of the SKUs from the production facility to the warehouse.

The table shows that the main cost driver is the turnover ratio of a SKU. This is so because the turnover ratio gives an indication of how long space (in the warehouses) needs to be reserved for storing the SKU. Alternatively, it indicates the frequency an SKU is

transported to external warehouses.

The table also shows that the total allocation cost was divided into four cost groups. The research showed that if the SKUs are allocated to a correct warehouse location, the total cost will be reduced. A correct warehouse location (in this sense) holds in that the chosen warehouse location for the SKU, considers the different factors that do influence the total cost, with the purpose of reducing this cost.

Assigning the different SKUs to a warehouse location is called an allocation plan. By (correctly) influencing the decision for allocating a SKU to a warehouse location, the total cost can be reduced.

With the help of the LP-model an (optimal) allocation plan can be created. However, impractical situations may arise for implementing the allocation plan in the real world. These impractical situations can complicate an efficient use of the available resources. The research also showed (as expected) that the feasibility of the plan must also be considered.

The introduction of this DSS at the location in Foxhol proved to be successful. Due to the success, implementation of the DSS at another location was the following step. However, this was soon interrupted. The reason for this was that the tool did not include all factors that play a role at the other location. For example the location in Foxhol has storage space for food products. Meanwhile at the other location there are also non-food products involved. This and other issues were not included in the model.

Business goal

This research was set up in order to reconstruct and update the tool (the LP-model) so that it represents the situation at the location in Ter Apelkanaal (TAK). In this way, it is

2.1 Research question & sub-questions

The goal of the research is to reduce the allocation cost while satisfying the requirements set at the other location, hence the following research question.

How and where should the different SKUs be allocated in order to reduce the current allocation cost?

The answer to this question shows the way and the place where the SKUs ought to be allocated in order to reduce the allocation cost in comparison to the current allocation cost.

The figure below illustrates the conceptual model setup for this assignment.

Figure 3 Conceptual model

The sub-questions for this research are given below.

1. What requirements are of importance for AVEBE regarding the allocation plan? These requirements refer to the ease of implementing the allocation plan and, in this case, updating the LP-model.

2. How are the SKUs allocated at the moment?

The allocation process needs to become clear, especially discovering the constraints that influence the decision-making must become apparent. In this way, it is then possible to look at the factors that play a role within the allocation process.

Allocation plan Allocation cost

Cost drivers

3. How can the important factors in the current method be modelled in a linear programming formulation in order to incorporate these factors in the model? As mentioned in the problem statement, AVEBE makes use of a linear programming model to assist in the allocation decisions. Therefore, the result of this

sub-question is a set of mathematical equations that describe the factors mentioned in the previous sub-question. With the mathematical model, it is then possible to generate an allocation plan for the SKUs.

4. How can the model be used to find a solution for the allocation problem that is feasible for implementation?

The model is used as a tool, which means that the final allocation decisions remain at management. This is so because of the specific management preferences at a certain moment in time, but also issues that cannot be modelled in the LP-model (e.g. distribution patterns of several SKUs). The purpose of this sub-question is to describe the way in which a solution was found for the current allocation problem (with the help of the model), that can be implemented in the real world. In other words: creating a feasible allocation plan.

2.2 Research setting & boundaries

 This research concerns the production location at Ter Apelkanaal (TAK), Groningen.  The cost driver (found in the previous research) ought to be taken into account.  The solution ought to be feasible. In other words, it should be possible to

implement the results in the real world and significantly reduce the allocation costs compared to the current situation.

 The focus should be on optimizing the use of AVEBE’s own resources. Investment plans and decisions are out of the scope of this assignment.

 The initial LP-model is constructed in Microsoft Excel (MS Excel). Further development and construction of the “new” LP-model ought to happen in this program as well. In this way, it is assumed that the loss of knowledge can be prevented. This would be the case if a specific (or specialized) program was used for constructing the LP-model.

3

METHODOLOGY

In the first step, the current situation of the problem was examined. In general, there are different stakeholders involved in a problem. The allocation decision problem was rather well defined; it was not a vague set of symptoms as is usually the case in other industry, such as service industry. The ones involved in this problem were the supply chain manager and the warehouse manager at Ter Aplelkanaal (TAK). The problem here was that the current supporting tool (the LP-model) did not include the necessary requirements for the location in TAK. This resulted in the rejection of the LP-model. The consequence was that decisions remained very subjective, based on intuition and experience. Therefore, the purpose of this project was to make the necessary improvements to update and correct the LP-model. This would make the utilization of the DSS possible in order to make allocation decisions based on data rather than only intuition and / or experience.

Illustrating the current way of working is important. In this way, it is then possible to gain insight in the key factors that needed to be taken into consideration when constructing the new LP-model. For this part, the warehouse manager from TAK was interviewed. The purpose of these interviews was mainly to gather information regarding the factors that were important in the decision-making. Especially the requirements set by the

management at TAK when implementing an allocation plan in the real world were of importance during this first step.

Step 2

Modelling the problem, as was described in the previous step, was the following step. In this case, mathematical equations based on the characteristics of the products and warehouses and the requirements set were created.

The previous LP-model for Foxhol was constructed in Microsoft Excel (MS Excel). The “new” model was also constructed in the same program. MS Excel is extensively used within AVEBE, therefore it is expected that by using this program:

The model was also validated in this step. This validation was done by verifying that the model was constructed right. This was as follows:

 The approach used to build the model was consistent with the previous research conducted at Foxhol.

 The financial figures gave satisfactory results (total cost of previous year is roughly the same in the model).

 Trial results indicated that the model is behaving satisfactory. This was done with the help of the designer of the LP-model from Foxhol.

Moreover, the input data used for the model regards the sales of the previous year (baseline). This data was used as (expected) demand data during the model run. Step 3

In this step the model was used to create a feasible allocation plan. There was much interaction here with the warehouse manager in order to create an acceptable plan. This is so because not everything can be modelled in the LP-model and specific preferences of the manager were also taken into consideration.

Sensitivity analyses were also carried out in this step. This was done by setting up several scenarios and assessing the results. Detailed information regarding these issues are provided in chapter 6.

Step 4

Once an acceptable allocation plan was created, the translation of this plan into an implementation plan was the following step. In this implementation plan the allocation and the priority of the different SKUs were given.

4

CURRENT SITUATION

In section 4.1 the cost driver and cost groups mentioned in table 1 are explained with regard to the current situation at Ter Apelkanaal (TAK). Section 4.2 deals with the current situation, at the beginning of the research, at TAK. In that section the lines and their different allocation is clarified. The current way of allocating the different SKUs is made clear in that section. An assessment with regard to the LP-model is given in the final section of this chapter.

4.1: The cost driver & cost groups in the current situation

First, to keep things clear, the 5 different warehouse locations that are used by the production location in TAK are shown in the table below.

Table 2 The 5 warehouse locations

Internal warehouse External warehouses

TAK CDC NAVO Teuben Ter Apel

The names of the warehouses hold no further meaning. Cost driver: Fast and slow-movers

The focus of the decision-making is mainly on the fast-movers. The warehouse manager states that fast-movers are SKUs that have a high turnover ratio. The turnover ratio is the number of times a company's inventory is replaced during a given time period (usually a year). The slow-movers are the SKUs that are not sold very often in a year, thus having a low turnover ratio. These are usually stored in NAVO because in this way the other space available can be used for the other SKUs without worrying much about these SKUs. Cost group: Handling

not differ. This means that in general SKUs that are sold by half pallet are more expensive to handle.

Cost group: Combined shipments

These are SKUs that are shipped together in the same container or truck. Therefore it is important to have certain SKUs allocated at the same location in order to prevent these costs. This issue can be dealt with by either allocating certain amount of SKUs at several locations or by allocation many different SKUs at the same location. The warehouse manager usually makes use of the second option. The second option is usually used also because of simplicity.

Cost group: Transportation This cost group is divided into:

 Pre-trip transportation cost. These SKUs are transported to Rotterdam for further delivery. The costs from the different warehouse location to Rotterdam differ per location.

 Internal transportation cost. These are the costs for delivering the SKUs from the production location TAK to an external warehouse location.

By allocating the fast-movers at the warehouse location in TAK, high transportation costs can be prevented. If they are stored in another warehouse location this would result in much internal transportation, increasing the cost group transportation.

Cost group: Value added logistics

The assumption is that if SKUs are allocated externally, especially the fast-movers, the value added logistics costs are likely to increase because of the operations that need to be carried out for the sold SKUs. Therefore, these SKUs are stored internally.

Feasibility for implementing an allocation plan

manager at TAK states that an allocation plan is feasible when the SKUs are grouped in a logical way that keeps the allocation plan simple. This can be achieved when:

 An entire product line is as intact as possible. This becomes possible when SKUs within the same line are allocated at the same location. In other words, 1 line (with all its corresponding SKUs) at 1 location.

 If separation of SKUs is required, it is preferred to allocate the same bag types (of the same line) together.

In this way the allocation plan remains simple and easy to manage. However, the previous research showed that by separating the SKUs from the same line (instead of 1 line at 1 location), much savings can be made. This conflict is a vital issue in the current situation because this drastically affects the simplicity within the allocation process.

The next section describes the initial situation with regard to the above-mentioned topics.

4.2: Initial situation

Table 2 gives the allocation of the different lines with the market they serve. The

allocation of the lines is given at an aggregated level. This holds in that most SKUs of a line are allocated to that location.

Table 3 Allocation of the lines in the current situation

Line Most SKUs located in Market

A Teuben Food

BC Teuben Food

GUM Teuben Food

V Teuben Food

PN1 TAK Food

PN-2 TAK Non-food

4.2.1: Allocation of the SKUs

The different lines and the allocation of the SKUs within each line are discussed here. This is done for the current situation, which is at the beginning of the research. Restrictions (when developing the LP-model) that need to be taken into account are also mentioned. Line A

Every SKUs within this line are allocated to the warehouse location Teuben. The warehouse manager sees no special reason to allocate the SKUs within this line differently. In other words, there are no restrictions within this line.

Line BC

This line can be split into 3 groups.

 SKUs that are sold in bulk. These SKUs are sent (in big bags) to the warehouse location Ter Apel because that is the location where the big bags can be pored back into silos and be sold in bulk. Furthermore, this process can only be done at this location (in Ter Apel) for the food products. In this way this is a restriction that needs to be taken into account.

 SKUs that are put on other type of pallets. Certain customers request for other types of pallets to be set under their products. These pallets have other

measurements compared to the standard pallets used by AVEBE. This usually results in difficulty when loading these products onto trucks or containers, which result in broken bags, and thus unsatisfied customers. At the moment this issue is handle by allocating these SKUs to the internal warehouse. This makes it possible for the warehouse manager to better control these operations. When this is carried out by the other locations, much customer complaint is received. The remaining SKUs that belong to this group are also allocated to the internal warehouse. This is done for simplicity reasons, this group of SKUs remain together.

 The remaining SKUs. These are sent to the warehouse location Teuben because there is not enough space for every SKU to be allocated internally.

Line GUM

Line V

For this line, there are also SKUs that are sold in bulk. These SKUs are also sent to the external warehouse location Ter Apel (just as the SKUs in line BC) in order for them to be pored back into silos and sold in bulk.

The pallet issues which hold for line BC also play a role here (service related). Therefore, the SKUs to be set on other pallets are also allocated internally. All the other SKUs are sent to the warehouse location Teuben.

Line PN1

Many of the SKUs within this line are considered fast-mover. Therefore, choice was made to allocate the largest part of this line to the internal warehouse TAK. Nonetheless, there are also SKUs that remain internally because of the pallet issue (service related).

The SKUs that are allocated to an external warehouse are:

 Slow-movers. These are sent to the warehouse location NAVO.

 Semi-finished products. These are also allocated to the warehouse location NAVO, because not only they are perceived as slow-movers, but also because they are then allocated closest to the location where they will be sent to; in Germany. They are sent there for drying and returned back for further processing.

Line PN-2

This is the non-food line. Because of the non-food character, the warehouses for these SKUs have other quality standards compared to the food lines. The warehouses that are assigned to these SKUs are the warehouse location NAVO and the warehouse location TAK. Furthermore, the SKUs that are sold in bulk for this line are allocated in the warehouse location TAK. This is so because here it is possible to carry out this task for the non-food SKUs. Moreover, this line is also perceived as a fast-mover, which results in allocating the most of the line internally.

Figure 4 The current allocation of the different SKUs

The main concerns in the current situation are whether the judgements are sound (e.g. fast and slow movers) and the decisions taken are correct (such as remaining SKUs sent to Ter Apel), since they are mostly based on experience and intuition. Quantifying these decisions would provide much support, comfort and most likely more profit, because better decisions can be made.

Current strategy

Generally, the warehouse manager attempts to keep SKUs that are packed in small bags internally and allocate the big bags externally. The reason for this is that small bags can be stacked higher compared to big bags. This means that more products can be stored

internally if their packaging is a small bag. The result of this is that the use of the available space can be optimized.

Furthermore, there is a difference in the preference for allocating the different SKUs. The warehouse manager favours (as mentioned in the beginning of this chapter) keeping the entire line as intact as possible. The reason for this is that the allocation process remains uncomplicated and comprehensible. This keeps it manageable and simple to control with regard to the FIFO principle. However, the previous research conducted at Foxhol, showed that allocating the SKUs separately (called the SKU-level), rather than every SKU within the same line together (called line-level), more savings could be made. Nevertheless, this (allocating at a SKU-level) can complicate the planning & control for the warehouse manager because this would result in planning & controlling 131 SKUs instead of 6 lines.

Non-food SKUs + Slow movers NAVO Fast movers +

Non-food bulk SKUs +

Service related SKUs

TAK

Remaining food SKUs

Teuben

Food bulk SKUs + Remaining food

SKUs

Ter Apel

Additionally, the warehouse manager states that maintaining a smooth or constant flow is also important. This is so because in this way it is possible to make efficient use of the available resources (e.g. workers), which contributes to the cost-price leadership strategy. This smooth / constant flow can be achieved by allocating the products based on a line-level. This method (based on line-level) also facilitates the FIFO principle.

Summation

Since there is not enough space to store every SKU internally, some SKUs need to be stored externally. The main issue is to determine which SKUs should be allocated internally and which ones externally. Consequently, quantifying these decisions (to either allocate a SKU internally or externally) is the major concern in the current situation. In other words, what trade-offs are made when deciding where to allocate a SKU? Moreover, are these trade-offs correct / acceptable, keeping in mind that the allocation plan ought to reduce the

allocation cost while maintaining a practical (workable) situation.

As mentioned, the decisions made at the moment are mainly based on intuition and experience. Therefore, there is the idea that the space available can be better used. The different aspects that play a role in the current situation are summed up here.

 SKUs sold in bulk. These ought to be sent to the location where this task is carried out. There is a distinction between the food bulk SKUs and non-food bulk SKUs.  Service related SKUs. SKUs that require another type of pallet (other

measurements), on request of the customers. Transporting these SKUs causes a lot of trouble (e.g. broken bags). To prevent customer complaints, these are stored internally at the moment to better control this task. This eventually is a service related issue.

 Semi-finished products that need to be sent to Germany for drying and then returned to TAK for further processing. These SKUs are also to be stored in the warehouses, thus making use of the available space.

 Allocating the products on a line-level (most SKUs of the same line together) is preferred opposed to allocating the products on a SKU-level (separately allocating SKUs from the same line).

4.3: Assessment of the current situation

At the moment there is certain doubt about the decisions taken. This is so because the decisions are very subjective. As a consequence, everyone makes decisions based on their own experience and intuition. To make these decisions more objective, AVEBE wants to implement the DSS.

4.3.1: Reviewing the current situation

When comparing Foxhol with TAK, the following can be observed.

 At the location in Foxhol only non-food SKUs are stored. In TAK, however, there is storage space for both, food and non-food SKUs.

 At TAK some SKUs for certain clients require a special type of pallet. This often leads to complaints. In Foxhol this is not an issue.

These points are to be taken into account when constructing the LP-model. The way this done is further clarified in the next section.

4.3.2: Assessment for the LP-model

Based on the assessment of the current situation and relating this to the previous model, it is vital to consider the a number of points (i.e. constraints) for the “new” LP-model. These are grouped as follows:

Service related

 Due to complaints, certain SKUs are (temporarily) allocated to the internal warehouse. These SKUs need better supervision when loading them for transport. Therefore, allocating these to the production location TAK is desirable.

Compatibility issue

Technical reasons

 The bulk SKUs. Allocating these SKUs to the right location is important because not every location has the required resource (machines) to perform this task.

 The actual available space needs to be checked. This is so because changes have taken place during the past period.

Practicality

 Allocating semi-finished products at a location closer to Germany must become possible. There is where they are eventually sent for further processing.

The way that these changes are to be implemented in the new model is as follows: Service related, Technical reasons and Practicality

 The possibility for restricting certain SKUs to certain locations needs to be added. This is of importance for the SKUs related to the pallet issues (service related), the bulk SKUs, and the semi-finished SKUs that need to be sent to Germany.

 The available space in the internal warehouse ought to be calculated and the amount is assigned in the model as a restriction. This is done for food and non-food products. This gives insight in the number of pallet space available for both

markets. Compatibility issue

 Because of the food and non-food issue, getting insight on these SKUs must also become possible. In other words, knowing which SKUs belong to the food market and which SKUs to the non-food market must become visible. This is done by assigning the line to which a SKU belongs to.

Moreover, the combined shipments cost group must not be forgotten. This had no

significant influence at Foxhol and therefore was omitted. This is still the question for the location TAK, does this cost group have a significant influence on the total cost?

5

MODELLING THE SITUATION & DYNAMICS

There are many matters to consider when making an allocation decision. For this reason, the mathematical formulation of the problem has vital importance. Therefore, in this chapter, the mathematical model of the allocation problem is presented. This consists of the objective function, the decision variables, and the set of constraints.

Before proceeding to the explanation of the linear programming model (LP-model), two important points are discussed. First, the combined shipment issue is discussed here with the help of figure 5. This is followed by an explanation of the supply network of the SKUs within the allocation process. This is done with the help of figure 6.

Previous research conducted at the location in Foxhol showed that the combined shipment did not play a significant role there. This also holds for the production location in TAK. The chart below illustrates this.

Handling 14% Transport Truck 13% Transport Sea 40% VAL 11% Pallet Space 20% Comb. Shipm. 2%

Figure 5 The share of the cost components in the previous year

The chart clearly shows that the influence of the combined shipment is at a minimum. Therefore, it is safe to state that the impact of this cost driver is insignificant. That is why the mathematical equations of the LP-model will not consider this cost group.

Figure 6 on the next page illustrates the supply network of the SKUs within the allocation process. This is further explained in its subsequent page.

24 Figure 6 Supply network of the SKUs within the allocation process

Destined to truck or sea

Handle in

- Pallet space

- Value added logistics

Handle out External warehouses Production location TAK Delivery by truck or destined for sea

(Rotterdam)

Decision-making: Send SKU extern or

keep SKU intern?

Extern

Intern

Delivery by truck or destined for sea

The model starts at the production location TAK. The main decision to be made is whether a SKU should be allocated to the internal warehouse (so it remains in TAK) or whether the SKU should be allocated to an external warehouse? Here is where the model gives

assistance in this decision-making.

When SKUs are allocated internally, no costs will have to be paid to a third party. If the decision is made to allocate the SKUs to external warehouses, then SKUs are transported to the different warehouses. Since the model makes use of historical data, it is possible to distinguish sales destined to be delivered by truck and sales delivered by container, thus destined for sea (these are send to Rotterdam). At the warehouses, the SKUs are taken (handle-in) to their storage location (pallet space) until sold to a customer. Once soled, the SKUs are taken out of their storage location (handle-out) and made ready for delivery. Customers often request for other type of pallets or a special label on their products, or another type of activity needs to be carried out. These extra activities are called value added logistics (VAL).

Table 4 Notation for the model development Notation Remark Index A I L T Input parameters Bi Ci Di Eit Fl Gilt Hil Jilt Ki Mial Nl Ol Plt Ql Rilt Sia Uia Zi Decision variable xilt

Set of value added logistics (VAL) activity {1, 2, …, A} Set of SKUs {1, 2, …, I}

Set of warehouse locations {1, 2, …, L}

Transportation type, t = 1 = truck or t = 0 = sea;

Average inventory in kilograms (KG) of SKU i (Metric Tonne-MT) Average sales in KG of SKU i (MT)

Number of stacking height for storage of SKU i

Number of stacking height for transportation type t of SKU i Maximum pallet spaces per month at warehouse location l The number of pallets sold of SKU i at warehouse location l for transportation type t

The number of pallet spaces for SKU i at warehouse location l The number of pallet spaces of SKU i to be transported by type t to warehouse location l

KG per pallet of SKU i (kg/pallet)

The number of pallets sold of SKU i at warehouse location l by VAL a Rate charged for handling-in plus handling-out at warehouse location l (€/pallet)

Rate charged for pallet space at warehouse location l (€/pallet space) Rate charged for transportation type t to warehouse location l (€/pallet space)

Rate charged for VAL at warehouse location l (€/pallet)

Sales of SKU i at warehouse location l of transportation type t (KG MT) Sales of SKU i for value added logistics a (KG MT)

Sales percentage of SKU i for value added logistics a Turnover ratio of SKU i

5.1: The objective function

In figure 6 the network of the model is visualized with the related costs. The objective function is to minimize the total cost, which is computed as the sum of the following: {handling cost + pallet space cost + transportation cost + value added logistics cost} The decision variable is given as:

xilt= the fraction of SKU i at warehouse location l of transportation type t In mathematical form the objective function can be written as:









The way the different components of the objective function are calculated, are clarified in the next section.

5.2: The cost groups

In this section, the different components of the objective function are clarified. First the handling cost is explained, then the pallet space cost, followed by an explanation of the transportation cost, and finally the value added logistics cost is explained.

5.2.1: Handling cost

The handling cost is expressed by:

Handling cost =



The number of pallets sold of SKU i at warehouse location l for transportation type t is obtained with the following calculation:

Gilt= Rilt

x 1,000 Ki

As mentioned before, the model is based on historical sales data. Considering this fact, allows grouping of the handling-in and handling-out rates. This results in the formulation of the handling rate (consisting of handling-in plus handling-out) charged at warehouse location l, given as Nl.

5.2.2: Pallet space cost

The pallet space cost is expressed by:

Pallet space cost =



The number of pallet spaces for SKU i at warehouse location l is calculated by:

Hil=



The turnover ratio is calculated as follows:

Zi= Bi Ci

With the Zicalculation of the number of months a pallet space needs to be reserved becomes possible.

5.2.3: Transportation cost

The transportation cost is expressed by:

Transportation cost =



With the help of the notation Gilt, the calculation of the number of pallet spaces for SKU i for transportation type t to warehouse location l, is simplified. This is given as:

Jilt= Gilt

Eit

5.2.4: VAL cost

The VAL cost is expressed by:

VAL cost =



To simplify the calculation of the VAL component the following notation is used:

Mial= (





T

t 1

The above notation on its turn consists of: Uia= Sia



This gives the sales percentage of SKU i for value added logistics a. With the help of this percentage it become possible to calculate the number of pallets sold of SKU i at warehouse location l for VAL a.

5.3: Constraints

The model should take into account several restrictions. The restrictions are grouped as service related, technical reasons, practicality, and compatibility issues. These are clarified here. First, the basic constraints are presented.

5.3.1: Basic constraints

The following restriction makes sure that the decision variable is not a negative number. 0 ≤ xilt≤ 1 (6)

The following constraint indicates that the summation of all the warehouse locations l for SKU i and transportation type t, should equal 1, making sure that the total sales of SKU i is stored in a warehouse location(s).



5.3.2: Service related constraints

Due to complaints, certain SKUs are (temporarily) allocated to the internal warehouse. This restriction is given as:

(8) t (i), I all for       





l = α = the chosen warehouse location

The restriction is given in general form. In this way, it can be used for other situations that also require this type of restriction. In this situation, the chosen warehouse location is the internal warehouse location TAK.

5.3.3: Technical constraints

This group can be divided into two sub-groups. These are 1) the bulk SKUs (for both market food and non-food) and 2) the available pallet space at a warehouse location. Regarding the pallet space the choice was made for making calculations per month because, for certain warehouses, it is possible to terminate contracts on a monthly basis.

1. Bulk SKUs for the food market and bulk SKUs for the non-food market.

For the food market this activity can only be carried out at a certain warehouse location. This is so because the resources for performing this task are only available at that location.

The restriction for this condition (for the food market) is the same as restriction number 8. Only difference is that the chosen warehouse location in this case is the warehouse location Ter Apel.

For the non-food market SKUs the restriction is given as: xiαt + xiβt= 1 for all i and for all t (9)

2. Available pallet spaces.

Flgives the maximum capacity of available pallet spaces per month at warehouse location l, and the constraint is given by:



5.3.4: Constraints related to practicality

 Related SKUs regarding rework.

After certain quality control, several SKUs are reworked into other group of SKU because they do not meet the required standards. Allocating these group of SKUs at the same warehouse location can prevent unnecessary extra work (such as

reallocation).

 Related SKUs regarding same product group.

Many SKUs belong to the same product group, but for example due to the standard pallet they are delivered on, they receive different number. Once more, in order to prevent unnecessary extra work, it is preferred to allocate these SKUs together. The constraint for both of these types of issues is given as:

xilt = xi’lt for all i ',i



This means that the combination of SKUs that are related will change, but the restriction is the same for all.

 Semi-finished SKUs.

Some SKUs are destined to be delivered to Germany for further processing. Allocating these SKUs closest to Germany can provide much advantage, such as short transportation distance, which increases the reaction time.

5.3.5: Compatibility issues constraints

The SKUs for the food market cannot be in the same area as the SKUs for the non-food market because compatibility issues. At certain warehouse locations, the SKUs are handled with the correct precaution. Therefore, to prevent unnecessary (and unwanted) extra work, restricting the SKUs of the non-food market to certain warehouse locations is preferred. The restriction is given as:

xiαt+ xiβt+ xiγt = 1 for all i,t (12) Where: l = α = warehouse location TAK

l = β = warehouse location CDC l = γ = warehouse location NAVO

5.4: Model validation

The model validation was done in the following way:

 The approach used to build the model was consistent with the previous research conducted at Foxhol.

 Experimentations (by changing values and amounts) with several parameters gave satisfactory and expected results.

 Trial results indicated that the model is behaving satisfactory. This was done with the help of the designer of the LP-model for Foxhol.

5.4.1: Set up in MS Excel

34 Figure 7 Overview of the sheets of the LP-model in MS Excel

The design in the above figure illustrates the different worksheets used in MS Excel for calculating the total cost. The illustration can is divided into four groups. These are explained here.

1. Input parameters.

The different parameters and notations of table 4 is found in this group. These are the input for the model. Here 4 worksheets are used. With the help of the INPUT worksheet, pivot tables are made to group and facilitate the input data. These worksheets consist of the characteristics of the SKUs (e.g. kg/pallet, stacking height, sales, etc.), and the warehouse locations (e.g. prices charged for VAL, available spaces, etc.).

2. Calculations.

Here the different calculations are done. It is important to mention that for the transportation costs though, this is divided into pre-trip transportation and internal transportation. This keeps the distinction between the truck and sea transportation. As mentioned before, the combined shipment component is not taken into account when searching for an optimal solution. Nonetheless, the cost for this issue in the different scenarios is (automatically) calculated.

3. Output.

With the help of the solver the objective function, variables and constraints are given in MS Excel. In this worksheet the model run is performed.

4. Base of comparison.

Here the result of the model run is translated into another worksheet, called Ton. The worksheet baseline contains data of the sales of the previous year.

Nonetheless, this can be changed to other data, making comparison of different scenarios possible.

6

RESULTS

This chapter consists of two sections. The first section deals with the output of the model after generating several allocation plans. More specifically three main situations are discussed. These are the baseline (allocation based on the data of the previous year), the LP-solution (the model generates a plan based on these data), and the feasible plan (by rearranging the output until an acceptable plan is created).

The second section deals with the different sensitivity analyses carried out.

6.1: Outcomes of the research

Figure 8 illustrates the idea behind the use of the LP-model when searching for a feasible plan. In the figure the coloured area called Baseline is the situation based on the data of the previous year. The LP-model makes use of this data in order to generate a solution. That solution is called the LP-solution.

high

LP-solution

Baseline

low high

Where lays the feasible & financially most attractive allocation plan?

Complexity

Sa

vi

ng

s

Figure 8 Finding the best allocation plan

When looking for the feasible plan one most consider the complexity (for implementing the allocation plan) opposed to the savings that can be made.

37 Number of SKUs in the locations for Baseline

0 10 20 30 40 50 60

CDC NAVO TAK Ter Apel Teuben

Location # o f S K U s

Number of SKUs in the locations for LP-solution

0 10 20 30 40 50 60

CDC NAVO TAK Ter Apel Teuben

Location # o f S K U s

Number of SKUs in the locations for Feasible plan

0 10 20 30 40 50 60

CDC NAVO TAK Ter Apel Teuben

Location # o f S K U s Legend:

The colour of the lines is given here.

A BC GUM V PN1 PN-2

The differences between the three situations are given in the table below.

Table 5 The differences between Baseline, LP-solution and feasible plan Situation Differences

Baseline vs. LP-solution

 In the baseline, the location TAK (the internal warehouse) was generally used for the lines PN1 and PN-2. The LP-solution makes use of the available space in another manner. Mainly the fast moving SKUs are allocated in the location TAK until there is no more space. The others, medium and slow, are allocated to an external warehouse. Regarding the other lines, in the baseline they are kept as intact as possible. Whilst in the LP-solution they are somewhat spread through the locations.

 In the baseline other SKUs than bulk, were also sent to the location Ter Apel. However, this is not the case in the LP-solution due to the reduction in available space. Only bulk SKUs can be sent there.

Baseline vs. feasible plan

 In the feasible plan, due to certain restrictions, SKUs of lines BC and V are allocated to the location TAK (internally). Several fast movers of the lines BC and GUM are also allocated to the internal warehouses. In the baseline however, only lines PN1 and PN-2 were allocated internally.

 In the feasible plan, the available space in the location Ter Apel (due to other matters) has been reduced in comparison to the baseline.

 Furthermore, the location NAVO is mainly used for the lines PN1 and PN-2 in the feasible plan. In the baseline, this location was used for other lines also.

Feasible plan vs. LP-solution

 The LP-model makes a clear distinction between the different turnover ratios of the SKUs. This is clearly visible in the above figure. It can be said that fast movers are allocated in TAK and the others, medium and slow movers, are allocated to the other locations. In the feasible plan, however, this is kept at a minimum. Only for the lines GUM and BC this is done.

 Even though, not visible in figure 9 a distinction between the truck and sea sales of the same SKUs is made by the LP-model. For the feasible plan this is not the case, these sales are allocated together at the same location.

 The difference between bag types (small and big bags) is made in the feasible plan. This eases the grouping of many SKUs (within the same line).

 In the feasible plan, a line is kept as intact as possible. This is done by allocating as many SKUs as possible of the same line together.

39

Comparisson of the different situations

0 100 200 300 400 500 x € 1 ,0 0 0 p e r y e a r Costs Baseline 194 171 540 141 265 31 Costs Feasible plan 152 143 540 114 273 39 Costs LP-solution 133 129 540 107 274 36

Handling Transport

Truck Transport Sea VAL Pallet Space CombShipm

Figure 10 A comparison of the costs for the different situations

Total cost

In table 5 the differences between the three situations was explained. Figure 10 illustrates the differences between the three situations once more, but in financial figures. Here the differences between the costs / savings of the different cost groups are explained.

Savings LP vs. Savings Plan

Handling and Internal transportation

It is expected that the savings between the different cost groups would show a (more or less) linear result. However, when comparing the cost groups handling and internal transportation with the other cost groups, the results are somewhat unexpected at first. However, when assessing these cost groups it becomes apparent that this unexpected result comes from the SKUs that are sold by half pallets. As a result, handling cost (which is charged per pallet), and the internal transportation cost (these SKUs are stacked 2 high) can be affected considerably by the decision of either allocating these SKUs internally or externally.

Pre-trip transportation

It was expected that the location CDC might provide some savings with regard to SKUs that are sent to Rotterdam. This is so because the charges of this cost group, at the location CDC, is less compared to the other locations. Nonetheless, based on the data, no savings can be made regarding this cost group. The reason for this is that (most of) the other cost groups (handling, VAL, etc.) are more expensive at the location CDC when compared to the other locations, which reduces its advantage regarding the pre-trip transportation cost. Value added logistics

With regard to the cost group VAL, saving can be made when the SKUs are allocated differently compared to the baseline, no great surprises here.

Pallet space

The cost group pallet space gives a negative result. The reason for this is that:

1) It is expected that the gap between the periods of high and low sales will increase because of allocating more fast movers internally.

Consequently, there would be periods when the available space is not utilized efficiently. Therefore, by reducing the available space, when calculating the optimal solution with the model, this gap can be kept at a minimum.

The baseline showed that more than 4,600 pallet spaces were used according to historical data. However, the calculations were done based on 4,500 pallet spaces, thus resulting in a loss.

Combined shipment

As mentioned before, the impact of this cost groups is not significant. Nonetheless, the difference between the LP-solution and feasible plan is made visible.

6.2: Sensitivity analysis

Several sensitivity analyses were carried out to check the consequences of the assumptions made and the restrictions given. These analyses concern:

 Altering the available pallet spaces.  Altering the average pallet size.  Altering financial figures.  Only Non-food bulk in TAK.  Food bulk also in TAK.

The reason for performing these analyses and their results are explained and shown in the subsequent sections.

Available pallet spaces

Calculating the available space was a bit problematic during the research. This is so because the space in the warehouses was also used for other activities such as rework of rejected products. Also purchased products concerning other departments (sometimes) made use of the available space. This shows that in the old situation the other cost groups (such as transport, logistics production) played a too less important role in the decision making process. Therefore, this analysis was carried out in order to display the

Total cost when altering pallet spaces 0 200,000 400,000 600,000 800,000 1,000,000 1,200,000 1,400,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000

Amount of pallets spaces

T o ta l c o s t in € p e r y e a r

Figure 11 Sensitivity test altering the available pallet spaces

The graph shows that the pallet space does have a significant effect on the total cost. This happens until about 8.700 pallet spaces (given by the dashed line), which is approximately the total amount of pallet spaces needed for the SKUs to be allocated internally. Only the bulk SKUs for the food market are sent to another location where this activity can be carried out. The result also shows that the increase in pallet space is not linear to the costs for storing the SKUs. The straight green line gives this conclusion. This is so because of the differences between the SKUs and their characteristics. Especially the turnover ratio has an impact on this.

Another interesting matter after assessing the results of this analysis is that the SKUs that provide savings differ per scenario. The consequence of this is that opportunities for greater savings can be missed when the amount of pallet spaces is not accurate. Pallet size

changing the 2.7 m2_{of the average pallet size. The result of this analysis is shown in the} graph below.

Consequence when altering average pallet size

-40,000 -30,000 -20,000 -10,000 0 10,000 20,000 30,000 40,000 2.4 2.5 2.6 2.7 2.8 2.9 3

Average pallet size in m2

E x tr a s a v in g s /c o s ts i n € p e r y e a r

Figure 12 Sensitivity test altering the pallet size

The result of this analysis shows that if the average pallet size can be decreased, extra savings can be gained. The average pallet size is concerns 1) the storage space, 2) the preparation space (for loading into truck or container) and 3) the access strip(s) in the warehouses. Therefore, if these spaces can be utilized efficiently, extra savings can be gained easily.

Other options such as decreasing the m2_{of the bags and pallets were also explored. Due to} the standards that AVEBE works with, regarding the density of starch per kilogram, the consequences of these options would lead to increasing the height of the bags which would result in other difficulties. One of the main difficulties would be the stacking height issue. The stacking height on the pallets would change, in the warehouses and also for

transportation.

Altering financial figures

groups that are likely to change. These are the handling cost and the internal transportation cost.

Handling cost

The results showed that even after lowering the handling cost to the same price as the other warehouse locations, the warehouse location CDC does not become financially attractive. This is so because the other cost groups (internal transportation and VAL) remain more expensive.

Internal transportation cost

The same result took place here. The location in CDC does not become financially attractive after lowering the cost equal to the other locations.

After assessing these results, the following was visible:

Table 6 Comparison of the costs of the different external warehouse locations

Cost group CDC NAVO Ter Apel Teuben

Handling (per pallet) € 3.01 € 2.76 € 2.76 € 2.76

Internal transportation € 99.71 € 82.90 € 82.90 € 55.83 Pre-trip transportation € 288.50 € 317.00 € 317.00 € 317.00 Storage (per pallet space) € 5.40 € 4.54 € 4.19 € 4.59

VAL (average) € 39.08 € 31.52 € 31.52 € 31.52

Table 6 clearly shows that the pre-trip transportation cost group is the only attractive cost group. The prices charged for the other cost group are significantly higher at the

warehouse location CDC. This shows that the warehouse location CDC is expensive if mainly assigned to the production location TAK. In other words, it is preferred for TAK not to make (much) use of the location CDC.

Only Non-food bulk in TAK

looked at and clarified. In appendix 1, there is more information available about this analysis. The results are given below.

Scenario 1: Savings feasible plan € 81,000 Scenario 2: Savings only non-food bulk in TAK € 65,000

Loss € 16,000

-The result of this analysis shows that it is not a good idea to allocate only the bulk SKUs of the non-food market internally. Other SKUs of the non-food market also gain significant savings when allocated to the internal warehouse.

When assessing the results it became clear that there are several SKUs (mainly the fast movers) that do gain much savings when they are allocated internally.

Food bulk also internally

There are possibilities of making use of mobile resources regarding the bulk SKUs for the food market. These mobile resources make it possible to also carry out this task at the internal warehouse location TAK. This analysis considered that option. In appendix 2, more information can be found about this analysis. The results are given below.

Scenario 1: Savings feasible plan € 81,000 Scenario 2: Savings food bulk in TAK € 100,000 Extra savings € 19,000

7

RECOMMENDATIONS

This chapter presents some recommendations regarding the following points:  An implementation plan for the reallocation of the SKUs. (7.1)

 A control method in order to maintain a correct allocation of the SKUs. (7.2)  A way to monitor so that workers can have an insight on their performance. (7.3)  Advice regarding the update frequency of the LP-model. (7.4)

 Further advice is also provided with regard to the grouping and arranging of the different SKUs. (7.5)

7.1: Implementation

Figure 13 illustrates the idea behind the reallocation process. The starting point is the baseline. Based on the out come of the model (the feasible plan), decisions are made concerning the priority of reallocating certain SKUs. Once the SKUs are reallocated, the new allocation plan should be in place.

Figure 13 From baseline to new allocation plan

As the figure shows, the advised reallocation process consists of three steps. The figure on the next page illustrates the idea behind the steps.

Baseline New allocation

of SKUs Step 3

Step 2

Step 1

high

LP-solution

Baseline

low high

Where lays the feasible & financially most attractive allocation plan?

Complexity

Sa

vi

ng

s

Figure 14 The consequences of the reallocation process

The figure shows that during the process savings are made when compared to the past financial figures. Nonetheless, the figure also shows that higher savings can only be made at high complexity. This remains a discussible issue / assumption.

The different steps shown in figure 14 are clarified in the subsequent sections. Step 1: Much savings with little effort

The figure below is a screenshot of the first step in the implementation plan.

Figure 15 Screenshot of the first step in the implementation plan Step

1

Step 2