Anticipation on seasonal demand

(1)

Master thesis

Technology Management

Anticipation on seasonal demand

Ensuring on time delivery of paint products

By Daan Hoevers

Zwolle, August 2010 Student ID: s1389076

E-mail: daan_hoevers@hotmail.com

AkzoNobel Wapenveld Supervisor: Ir. B. Dogger Supervisor: H.J. Töpfer University of Groningen

(2)

Management summary

The production and warehouse site Wapenveld is part of the Business Unit Decorative Paints Continental Europe (Deco CE), one of the business areas of AkzoNobel. This research project focuses on the paint production unit of Wapenveld. The management of Wapenveld is concerned about the low service level performance of the location, including the paint production unit. The reliability of on time delivery of paint products is for some successive years under the target service level of 98%. In the fall of 2009, it was decided to response to this negative performance by building inventory in 2010, i.e. pursue a level strategy, to anticipate on the seasonal demand. However, a lot of uncertainty existed with regard to this decision.

The remainder of the research projects focuses on the paint production unit, which produces products for two groups of customers: Decorative Paints and other Business Units. The products for the former are replenishments of the inventory in the Distribution Centers (DCs), Make-to-Stock (MTS) products, while the orders of the latter are only executed on arrival, Make-to-Order (MTO) products. The production process of the paint production unit consists basically of three steps: production, where the paint is made, the quality control, and the filling of the packaging materials. The steering is done by a production planning and control function consisting of a yearly master production plan and a short term control.

Diagnosis of the problem

The problem indicated by the management of Wapenveld are the low service levels, besides, the decision making ability with regard to the level strategy requires an examination. It is established that during the seasonal demand peaks of particularly the MTS demand, the capacity of the paint production is not sufficient to meet demand. This results in replenishments that arrive later than planned in the DCs, products run out of stock and the service level performance for those products decreases. Also, the delivery time for MTO orders becomes longer than agreed and consequently, the service level for these orders decreases as well. Moreover, the available labor is not constant during a year, which causes a decrease of capacity in certain periods. In order to reduce the scope of this research project, only the most critical operation, the bottleneck, of the paint production unit was taken into account: a filling machine, named the ETL. The ETL was not able to meet demand in 2009, not even when three instead of two shifts were employed. The performance of the ETL is monitored by measuring the Overall Equipment Effectiveness (OEE) and the annual OEE of 2009 was 27%, far below the target of 35%, mainly caused by time lost due to setups.

(3)

costs, thus the lower both are the better. Within the current formal production planning control structure, there is no tool that enables a thorough analysis of the level strategy.

Design of a capacity planning model

Although the low efficiency of the ETL, it is decided to focus on the design of a capacity planning model that is able to provide insights in the behavior between demand, MTS inventory and available capacity of the ETL. This model can assists the decision making how to anticipate on seasonal demand fluctuations by a level strategy. The outcome can be analyzed by performance indicators, estimates of the real performance indictors used in Wapenveld.

The demand of 2009, expressed in hours ETL capacity, is used to illustrate the use of the capacity planning model. The planning horizon is chosen such that the important seasonal fluctuations are incorporated and set at a year. The planning horizon is divided in weekly periods. The aim of the model is predict on the one hand how much capacity is needed to deliver the MTO demand on time and on the other to see how much capacity is left for production of MTS products. The latter capacity is used together with the known inventory level and MTS demand to calculate the development of the MTS inventory level. The available capacity is restricted by the hours generated by the number of shifts employed.

Conclusion

The capacity model proved to be able to decide how many shifts must be employed to satisfy the MTO and MTS demand. Besides, it is able to decide how much anticipation inventory of MTS products must be build such that the target service levels of 98% are achieved. In order to keep the average on hand inventory, and consequently the inventory costs as low as possible the capacity planning model can be used to determine when to build inventory. Furthermore, the effect of the parameters on the outcome is tested by means of a sensitivity analysis. The decision to explicitly set the planning horizon at a year proved to be necessary to see the important seasonal fluctuations and to anticipate on them. It must be stressed that the aim of the model is to capture the essence of the behavior of demand and available capacity, so the outcome is an approximation of reality. However, the model incorporates enough detail to assist the decision making.

(4)

Preface

This report is the representation of my master thesis, the final part of the master degree program of Technology Management at the University of Groningen. I conducted this research at AkzoNobel in Wapenveld, where I started my internship at the end of January 2010. Now, 7 months later, I graduate and can start my professional career.

The internship at AkzoNobel gave me a great opportunity to apply my theoretical knowledge to practical problems. The advantage of doing an internship at a production site such as Wapenveld is that it enabled me to see the production process at close quarters. On the other hand, I was introduced in the world of the multinational AkzoNobel.

During my research project I was very well supported by my supervisor at AkzoNobel, Bart Dogger. I would like to thank him for giving me the opportunity to start this research project, his great knowledge about production planning, and advice about the process of writing a thesis. My second supervisor at AkzoNobel, Harm Jan Töpfer, was also of great importance, since he was always available to answer my questions about the production site Wapenveld. Furthermore, I would like to thank the site manager Tamme Bartels for offering master students, like me, the opportunity to conduct a research project at Wapenveld. Finally, I want to express my gratitude to all my colleagues at AkzoNobel Wapenveld, especially my co-intern, Daniël Krant.

At the University of Groningen, I received great support from Dr. N.D. van Foreest. He especially helped me with the process of constructing a quantitative model. His guidance and coaching was essential for delivering this thesis. I would like to thank my co-assessor Dr. J. Riezebos for reviewing my thesis in an early stage, his feedback was very useful.

(5)

List of abbreviations

BU Business Unit MRP Materials Requirements Planning

CLT Customer Lead Time MTO Make To Order

CONWIP CONstant Work In Process MTS Make To Stock

DC Distribution Centre MTT Manufacturing Throughput Time

DDC Diagnosis, Design, and Change OEE Overall Equipment Effectiveness Deco CE Decorative Paints Continental Europe OoS Out of Stock

DTT Delivery Throughput Time OWC Operating Working Capital ETL Etiketteren en Tappen in een lijn (Dutch) PPC Production Planning and Control IMI Intermediate Material Inventory PTT Production Throughput Time

KPI Key Performance Indicator SL Service Level

LS Low on Stock QC Quality Control

(7)

1 Introduction

AkzoNobel is a leading, multinational company active in three main business areas: Decorative Paints, Performance Coatings, and Specialty Chemicals. Each business area consists of several Business Units (BUs) in multiple countries worldwide. The location Wapenveld (WPV), where this research is conducted, is a production and warehouse site, part of the BU Decorative Paints Continental Europe (Deco CE). However, WPV is a multi BU site and produces not only products for Deco CE but also for other (sub) BUs, like Car Refinishes, Marine & Protective Coatings, and others from the BU Performance Coatings.

The location WPV consists of three production units: paint, fillers, and tinting. The paint production unit produces products with a relatively low viscosity while the products with a higher viscosity are produced in the fillers production unit. The tinting production unit adds color to paints. The total production of 2009 (in pieces) consisted of 57% paint products, 42% fillers products, and 1% tinting products. Moreover, the products are considered specialties, i.e. products that cannot be produced easily by other production sites without major investments. It concerns mostly small orders of products in the beginning or end of their life-cycle.

The strategy of WPV has to ensure that the business activities of WPV contribute to the overall business targets of AkzoNobel Corporate1. It is used as a guideline when making business decisions and is made explicit by the key performance indicators (KPI). WPV has four main KPI: (1) service levels & customer satisfaction, (2) operating working capital (OWC), (3) quality, and (4) health, safety, and environment. The development of the first KPI, service levels, is a concern of the management of WPV.

1

(8)

2 Research approach

In order to conduct a well-structured academic research, a research method is introduced in this chapter. Thereafter, initiated by the research motivation, a preliminary problem statement will be formulated as prescribed by the research methodology. To conclude, the structure of the thesis is explained.

2.1 Research methodology

The Diagnose, Design, and Change methodology (DDC) of the De Leeuw (2002) is used to structure this research project because it is particularly suitable for solving problems in (business) organizations. The point of the departure of this method is a problem indicated by the management of an organization, is this case WPV. Subsequently, a thorough diagnosis is made in order to gain useful insights in the problem and its context. Thereafter, (re-) designs are suggested that should solve the problem. During the change phase, the (re-) designs are implemented.

Each step of the methodology requires information. The information will be gathered using multiple techniques like interviews, attending meetings, observation, quantitative data gathering using software, and literature.

2.2 Problem introduction

In this section the problem as determined by the management of WPV is introduced; it is the motivation of this research project.

One of the KPIs of WPV is the service level, an indicator of the reliability of delivery, i.e. it illustrates the ability of WPV to meet the requirements of the customer with regard to fulfilling the right amount of demand in time (Bertrand et al., 1998). The BUs for which WPV produces products all report a service level and WPV aims for a target service level of 98%.

However, WPV is often not able to reach the target service level of 98%, see Table 2-1, in which the annual service level performance for Deco CE and the average annual service level for the other BUs of 2008 and 2009 are depicted. In section 3.1, the difference between Deco CE and the other BUs will be further elaborated, while in appendix A the different calculations are explained.

Table 2-1 – Annual Deco CE service levels and average annual service levels other BUs

Deco CE Other BUs

2008 93% 93%

(9)

The developments of the monthly service levels over a year are displayed in Figure 2-1. It shows a large variation between the monthly service levels, with a minimum monthly level below the 90%. According to the management of WPV and the research project of Dogger (2008), these low figures are not uncommon in the years before 2008. The low service levels are a major concern of the management of WPV because higher management judges the performance of the location by it. Moreover, it is announced that service levels are the KPI with the highest priority since a high service level prevents lost sales and, subsequently, losing customers.

Figure 2-1 – Development of monthly and average monthly service levels over 2008 and 2009

82% 84% 86% 88% 90% 92% 94% 96% 98% 100% 01 02 03 04 05 06 07 08 09 10 11 12 01 02 03 04 05 06 07 08 09 10 11 12 2008 2009 Deco CE Average other BUs

In the fall of 2009, it was decided to pursue a level strategy in order to achieve the desired service levels. The reasoning of the management of WPV was that building inventory before the peak in demand occurs, i.e. a level strategy (Buxey, 2005), would enable a reliable on time delivery to the customers. However, there was such an urgent need for action that the decision was taken without a thorough contemplation of the implications and alternatives.

Summarizing, the problem that initiates this research is the low service level performance of WPV. Moreover, WPV was not able to take a well founded decision on how to respond to the low service level performance.

2.3 Preliminary problem statement

(10)

Preliminary research question:

How can WPV respond to its low service level performance such that the target service level of 98% can be achieved?

Preliminary research goal:

From the preliminary research question and the research cause, a preliminary research goal is formulated.

Explain why WPV is not able to achieve the target service level such that possible responses to the problem can be determined and in doing so explicitly review the decision making abilities with regard to the level strategy decision.

Responses can be defined as possible solutions which, when implemented, should result in the achievement of the target service levels.

Preliminary scope:

o The three production units of WPV have different characteristics and due to the limited time for this research project, it is decided to concentrate on the paint production unit only. An advantage of this choice is that several research projects already have been conducted in the paint production unit, e.g. Dogger (2008) and Neijenhuis (2010). Their results will provide a solid input for this research project.

o The start of this research was in January 2010; therefore, this research will particularly focus on the events before that date.

o The last phase of the DDC method, the change phase, will be excluded from this research.

2.4 Structure diagnosis phase

(11)

3 Paint production unit

In this chapter a detailed account of the paint production unit is given, explicating the context in which the research is done. The account will subsequently discuss the customers and demand, the production process, and the production planning.

3.1 Customers and demand

As for the whole site, the paint production unit produces products for Deco CE and other BUs. In 2009, 89% of the pieces were produced for Deco CE and 11% for the other BUs. The products for Deco CE are distributed to Distribution Centers (DCs) in Europe where the customers purchase the paint. The products are replenishments of the inventory levels in the DCs; accordingly these are called Make-to-Stock (MTS) products. In contrast, the other BUs are seen as the customer entity by WPV, to who the products are delivered. WPV awaits the orders of these customers and executes them on arrival; therefore, these are Make-to-Order (MTO) products.

The demand for both customers, in pieces per month in 2009 is depicted in Figure 3-1. It must be stressed that the Deco CE demand represents the customer demand at a DC, while the other BUs demand is the arrival of orders at WPV. The graph shows the difference of the amount of demand of Deco CE and the other BUs. Also, a pattern in the demand is visible, primarily caused by the Deco CE demand. The demand starts relatively low, and then grows to a peak in April and another in June. Only in December, the demand decreases again. This pattern is not only typical for the year 2009, in interviews with the management of WPV came to the fore that every year approximately the same pattern occurs. Due to its repetitive character, depending on the time of the year, the fluctuations are seasonal (Silver et al., 1998).

Figure 3-1 – Monthly demand Deco CE and other BUs

(12)

3.2 Production process

The products in the paint production unit are manufactured according to the process depicted in Figure 3-2. The first step, production, is the actual conversion of raw materials to paints; where after the intermediate products are stored, intermediate material inventory (IMI). Before the next processing step, Quality Control (QC) must ensure that the product attains the quality standard; otherwise it is corrected. When approved, the packaging material is filled. The production is a batch process using several tanks, while the filling is a discrete process, using three automatic and several half automatic and manual filling machines. The filled packages, pieces, can be separated in three distribution flows. Flow (1) contains the items distributed to DCs of Deco CE in Europe. The other flows, (2) and (3), contains products for the other BUs. Strategic stock is held of a small proportion of those products, flow (3), at the DC in WPV (WPVD).

Figure 3-2 - Production process; including fraction of flows of total demand in pieces of 2009

3.3 Production planning and control

The production planning and control (PPC) hierarchy that is used to steer the production process at WPV consist formally of two main elements: a yearly planning, the master production plan (MPP) and a short term planning considering weeks, Figure 3-3.

Figure 3-3 - WPV planning and control hierarchy

(13)

Formally, the MPP should be used by short term planning, however it does not contain useful information for planning purposes; only a planned production volume in kilograms per month is given, see appendix B for an example. Consequently, the MPP is mostly neglected. The short term planning uses actual production orders. There is a difference between orders from MTO and MTS customers. As aforementioned, the orders of MTO customers arrive at WPV, are converted to a production order by the short term planning, and are then executed. The MTS orders are created by the Material Requirements Planning (MRP). This software package determines, according to parameters controlled by the DC, how much and when a certain product must be replenished, resulting in a planned order. The task of the short term planning is to convert this order to a process order, i.e. a production order, if WPV is able to produce that order. Simultaneously, the availability of the expected raw and packaging materials requirements are checked and reordered if necessary.

(14)

4 Analysis demand and production

In this chapter the low service levels performance of the paint production unit will be clarified by analyzing the relation between service levels and demand, and the production. The following topics will be reviewed: MTS demand, MTO demand, and achieved production of the paint production unit.

4.1 MTS demand

In the previous chapter it was concluded that the MTS demand accounts for the largest part of the demand, i.e. 89%, and it is the main cause of the total seasonal demand pattern. In Figure 4-1, the demand of Deco CE and its monthly service levels of products produced in the paint production unit are depicted. It shows that when the demand increases in the beginning of the year, the service levels decrease dramatically, only to recover at the end of the year when the demand decreases. The correlation between demand and the service levels is -0,66 see appendix C. This is interpreted as a fairly negative linear relation since a correlation of -1 is a total negative linear relation (Keller and Warrack, 2003). Thus, the seasonality of the demand has a significant negative effect on the reliability of the on time delivery of MTS products.

Figure 4-1 – Service levels and demand of Deco CE

0 100.000 200.000 300.000 400.000 500.000 1 2 3 4 5 6 7 8 9 10 11 12 Months 2009 P ie ce s 86% 88% 90% 92% 94% 96% 98% 100% Demand Deco CE Service level Deco CE

(15)

o When there is a significant under forecast relative to actual sales while WPV is producing the product, the stock runs OoS before the planned replenishment arrives at a DC. The DC is attributable to this failure.

o A product can also run OoS due to an increased replenishment throughput time, the average time it takes to traverse all production activities, such that is greater than the planned delivery lead time. The replenishment arrives too late at a DC. This failure is attributable to WPV.

In both cases, the MRP system will react by generating more replenishment orders since the inventory levels must be replenished till a certain safety level in order to satisfy future demand. Thus, if the customer demand increases, either the forecasts will be adjusted by the DCs or the stock level decreases, resulting in an increase of replenishment orders; WPV has to produce more. Finally, if WPV is not capable to produce the requested replenishments, the number of products that run OoS increases. The graph in Figure 4-2 depicts the weekly number of products which are low on stock (LS), i.e. under its safety stock level, and OoS in 2009. The two peaks of OoS items, in May/June and in November clearly resemble the peaks in demand, Figure 3-1, so WPV is not sufficiently able to replenish the stock levels in those periods.

Figure 4-2 – Number of products with a LS and OoS status (of the first weeks of 2009 is no data available)

0 10 20 30 40 50 60 70 80 90 1 2 3 4 5 6 7 8 9 10 11 12 Months 2009 N u m b e r o f p ro d u c ts Low Stock Out of Stock 4.2 MTO demand

(16)

customer lead time, determining the due date of an order, is one of the specifications of delivery which is recorded in service level agreements between WPV and customers.

4.3 Achieved production

In Figure 4-3, the monthly achieved production in pieces in 2009 of the paint production unit is depicted. This graph illustrates the seasonal pattern since the production unit’s production attempts to follow the demand pattern. However, according to the previous sections, it cannot keep up with it. During interviews with the management of WPV it came to the fore that the dip in the production volume in April 2009 can be explained by the public holidays, thereby decreasing the number of available working days. The dip in July and August can be attributed to the summer period in which many (experienced) operators are not available. Thus, the available labor is not constant during a year.

Figure 4-3 – Monthly achieved production paint production unit in 2009

0 50.000 100.000 150.000 200.000 250.000 300.000 1 2 3 4 5 6 7 8 9 10 11 12 Months 2009 P ie ce s Total production Deco CE Other BUs 4.4 Conclusion

(17)

5 Bottleneck identification

The capacity of the paint production unit is composed of the capacity of the equipment in the paint production unit. However, taking into account all the equipment would require too much time, which is not available for this research project. Therefore, it was decided to focus on the performance of the bottleneck operation first since it usually dominates the behavior of a production system (Hopp and Spearman, 2008) The notion of bottlenecks is widely applied since the work of Goldratt and Cox (1984), who state: “to increase the capacity of the plant is to increase the capacity of the bottleneck” (p. 152). Thus, focusing on the bottleneck is a legitimate demarcation to bring down the problem area to more manageable proportions while ensuring that the performance of the whole paint production unit will improve.

5.1 ETL filling machine

According to the findings of Dogger (2008), the bottleneck operation of the paint production unit is the ETL2 filling machine. In Figure 5-1, the demand of products processed at the ETL and the achieved production of 2009 is depicted. It shows that the ETL is never able to meet the demand. The peak production in June 2009 is mainly due to the expansion of machine hours by the employment of three shifts instead of two.

Figure 5-1 – Demand and production of the ETL in 2009

0 40.000 80.000 120.000 160.000 1 2 3 4 5 6 7 8 9 10 11 12 Months 2009 P ie c e s Total demand Achieved production

The production situation did not change much compared with the situation analyzed by Dogger (2008). Still, the same configuration of equipment is used and most of the orders, i.e. 44%, are processed by the ETL. Those orders represent a major part of the total pieces produced and the highest variety of different packages in 2009; see appendix E for more details. Besides, the ETL is perceived as the bottleneck of 2009 by the management and operators. The implementation of

2

(18)

the results of the research projects of Dogger (2008) and Neijenhuis (2010) did improve the performance of the paint production unit but it did not change the location of the bottleneck.

5.2 Performance of the ETL

The performance of the ETL is monitored by measuring the Overall Equipment Effectiveness (OEE). This performance measurement incorporates availability, quality, and speed, indicating how long a good product with maximum speed is produced. Thus, a higher OEE of the ETL is related to a higher capacity of the ETL. The exact formula of the OEE is given in appendix F. The target OEE is 35%, however in 2009 the annual OEE was 27%. This figure is chiefly affected by the availability factor since the quality factor of the ETL is (disputably) set at 100% and the speed factor is 91%. In Table 5-1, the decomposition of the availability factor is summarized. It shows the fraction of net time the ETL was actually filling, i.e. the availability factor, and the various activities due to which time was lost. Improving the availability can be attained by reducing the time for one of those activities. Reducing the total time lost as a result of setups should have priority over the others since those improvements probably would have the biggest impact. Moreover, the second largest activity, miscellaneous, is caused by more than 100 reasons so it is a very heterogeneous factor. However, methods for setup reduction are not further discussed here; various textbooks (e.g. Bertrand et al., 1998 and Hopp and Spearman, 2008) can be consulted.

Table 5-1 – Availability factor decomposed, related to 2009

Activity % Net time

Filling = Availability factor 29,1%

Setups 29,0%

Miscellaneous 19,6%

Breaks 11,5%

Small interruptions 5,8%

Missing packaging 5,0%

Total net time 100%

5.3 Conclusion

(19)

6 Production strategies

As mentioned in the problem introduction, WPV made a not so well considered decision in the fall of 2009 to pursue a level strategy in 2010. The level strategy and its opposite alternative, the chase strategy, are so called production strategies; used to anticipate on seasonal demand patterns (Buxey, 2005). A level strategy prescribes maintaining a relatively constant production rate with a constant workforce, building anticipation inventory prior to peak demand which is depleted when the demand rate exceeds the production rate. Alternatively, a chase strategy dictates adjusting the workforce such that the production rate can track the demand rate. Due to case-specific constraints it is not always possible or desirable to pursue a ‘pure’ strategy, instead a modified production strategy is developed (Buxey, 2003). The decision which production strategy to pursue should be based on both quantitative factors, e.g. forecasts of demand, inventory levels, available labor, production rates (Buxey, 2005), and qualitative factors, e.g. labor morale and relations (Silver et al., 1998).

6.1 Level or chase strategy

In this section, the possibilities of a level and chase strategies are evaluated using the demand of 2009. It is chosen to do this evaluation with past demand because it is the most recent annual demand which is believed to be important, since the seasonal pattern must be visible. Moreover, the purpose is to show the fundamental principles of a level and chase strategy.

Since the ETL is the bottleneck operation, the demand of products in 2009 processed by the ETL is expressed in hours of capacity needed from the ETL, see appendix G. This enables a straightforward comparison with the hours the ETL is able to produce, i.e. when operators are present in the production unit: the available capacity. Also, the restrictions on the available capacity, partly mentioned in section 4.3, are incorporated, see appendix H. These restrictions cause a decrease of available capacity in certain periods.

(20)

Figure 6-1 – Weekly demand in hours ETL 0 20 40 60 80 100 120 140 160 180 200 1 6 11 16 21 26 31 36 41 46 51 Weeks 2009 H o u rs Total demand 2 shifts 3 shifts

Table 6-1 – Summary of hours generated or required in 2009

Hours 2009

Total demand 5546

2 shifts 4032

3 shifts 5688

Highest weekly demand peak 188

In 23 weeks of 2009, the demand exceeds the available capacity when employing three shifts, Figure 6-1. The highest peak in weekly demand is 188 hours, Table 6-1. These figures immediately rule out the ‘pure’ chase strategy as an option, since there will not be enough hours available to track demand in those weeks; three shifts generate only 120 hours.

(21)

the current formal PPC framework, WPV is not able to provide robust quantitative arguments to support this decision.

6.2 Production planning and control framework

Several textbooks (e.g. Bertrand et al., 1998, Hopp and Spearman, 2008) provide examples of PPC frameworks, which are all based on hierarchical principles; separating planning and control along two dimensions: planning horizon and level of detail, essential for coping with the complexity of planning problems (Hopp and Spearman, 2008). Despite small differences they normally account for three to four hierarchical levels, from aggregate to detailed, representing long term, intermediate term, and short term planning horizons. However, a PPC used by a firm should be adjusted in order to meet the needs of the specific production situation (Bertrand et al., 1998). Longer term planning activities for MTS demand, i.e. mainly forecasting and demand management, are the responsibility of the DCs, hence out of control of WPV. Instead, WPV only executes a short term planning and control function which is used to process the production orders, i.e. MTS replenishments orders and MTO orders. Basically, with regard to the MTS orders, there is only a flow of orders from the DCs to WPV, without giving feedback about the capacity of the paint production unit. WPV does not execute any proper formal medium or long term planning functions for its own resources which have to manage both MTS and MTO orders. Thus, there is no feedback whether the paint production unit has enough capacity to the meet the forecasted MTS demand and additional MTO orders.

A well performed capacity planning, taking into account the right planning horizon and level of detail could provide WPV with the quantitative information to make a reasoned decision whether and how to pursue the level strategy. Moreover, the study of Wacker and Sheu (2006) states that conducting a capacity planning enhances the overall competitiveness of a firm, especially the on-time delivery, which relates to the reliability of delivery, and consequently service levels.

6.3 Conclusion

(22)

7 Conclusion diagnosis phase

The aim of the diagnosis phase was to gain a better insight in the problem, introduced in section 2.2, and how it is related to the problem context. In this section, the conclusion of this phase will be drawn which provides the input of the next phase, the design phase. The conclusion is structured according to the difference between functional and instrumental problem, explained below.

According to De Leeuw (2002), business problems can be functional or instrumental. The former is a characteristic of the output of a system, while the latter is a characteristic of the system itself. A functional problem is caused by an instrumental problem and an instrumental problem leads to a functional problem. De Leeuw (2002) urges to use a functional formulated problem as point of departure when improving organizations. Considering instrumental formulated problems does not necessarily result in achieving the desired result since they only concern means to an end. However, it must be established which instrumental problems lead to the functional problem in order to determine the scope for the design phase.

Considering the products as the output of the paint production unit of WPV, i.e. the system, the low service levels are definitely a functional problem. The remainder of this conclusion will discuss the instrumental problems that lead to the functional problem and the achievement of the preliminary research goal:

Explain why WPV is not able to achieve the target service level such that possible responses to the problem can be determined and in doing so explicitly review the decision making abilities with regard to the level strategy decision.

(23)

sufficient to provide a satisfying answer how to anticipate on the seasonal demand fluctuations by using a level strategy.

In Figure 7-1, the conclusion of the diagnosis phase is summarized. The under forecasts that can cause a product to run OoS and unrealistic due date quotes in the service level agreements are not taken into account since it concerns topics that are out of control of WPV. In accordance with the management of WPV it is decided to exclude improving the efficiency of the scope of this research project since at this point it is more desirable to improve the decision making abilities with regard to the level production strategy. Furthermore, the chase strategy is not able to anticipate on the seasonal demand, as already was established.

Therefore, it is decided to focus on the design of a capacity planning model for the bottleneck resource, i.e. the ETL, in order to provide quantitative arguments to support the decision making in the future with regard to the level strategy. This should result in achieving the target service level of 98% for both kinds of customers.

(24)

8 Design phase

In the diagnosis phase it was concluded that the seasonal demand pattern negatively affects the service levels. Moreover, current planning and production planning practice primarily concerns short-term planning and there are no formal structures that can provide feedback about the availability of capacity at longer terms. Therefore, the design phase will suggest a solution that fills the gap such that a more founded decision on how to execute the level production strategy can be taken.

8.1 Problem statement design phase

The design phase is preceded by a problem statement, which consists of a main research questions, divided in sub questions, a research goal, and a research scope (De Leeuw, 2002), see also section 2.3.

Main research question:

How can insights be obtained in the behavior of demand, available capacity, and MTS inventory level such that a level production strategy can be evaluated by performance indicators?

The performance indicators will be estimations of the reliability of delivery and the inventory levels.

Sub questions:

In order to answer the main research question, two sub questions are formulated:

1. How can the production situation be modeled such that with a given demand, available capacity, and MTS inventory level, the performance indicators can be estimated?

2. How should the model be used in order to support decision making with regard to production strategies?

Research goal:

Construct a capacity planning model such that insights in the behavior of demand, available capacity, and MTS inventory level can be obtained which are useful for the management of WPV because they can support future decisions with regard to the level production strategy.

Scope:

o As was determined in the diagnosis phase, the ETL is the bottleneck operation; therefore the focus of the design phase is on this operation.

(25)

However, costs will be used as an analytical argument when evaluating the level production strategy.

o The distribution time to the DCs in Europe is not taken into account. Also, it is assumed that there are always enough raw and packaging materials available.

8.2 Research structure design phase

The design phase is structured according to steps from the methodology of Mitroff and Kilmann (1977), see Figure 8-1. Due to its emphasis on quantitative models, it is better suitable for this research than the methodology for the Design phase of De Leeuw (2002). The Mitroff and Kilmann’s (1977) model provides a methodological path which support the validity of the results obtained (Bertrand and Fransoo, 2002).

Figure 8-1 – Mitroff and Kilmann’s (1977) methodology

Conceptual model Scientific model Solution Problem situation Conce ptualiz ation Mode l solvin g Imp_lem enta_tion Mod_elin g

Within the design phase, the modeling step, resulting in a scientific model and the model solving step, which generate a solution, are executed. Those steps connect very well with the diagnosis phase and the causal model will be the starting point of the modeling step. The last step, the implementation resembles the change phase of the DDC method and is not executed as already was decided in section 2.3.

(26)

9 Capacity planning model

In this chapter, a model of a production line in the paint factory with the ETL as filling machine will be introduced. The rate of production of the production line is determined by the ETL since it is the bottleneck operation (Hopp and Spearman, 2008). The model must capture the relations between demand, MTS inventory level, and the available capacity of the ETL such that it enables an evaluation of the level production strategy by performance indicators. Those indicators estimate the reliability of delivery and the inventory levels. The target of the former is a service level of 98% while the latter is related to costs, i.e. operating working capital (OWC), thus its level must be as low as possible. This chapter will first discuss the unit of measurement used in the model; next, the variables used in the model are explained, third, the remarks and assumptions are summarized.

9.1 Units

In order to conduct an understandable analysis, all the variables are expressed in one unit of measurement, hours. The advantage of using hours as the unit of measurement is that it enables a straightforward comparison of the available capacity and the demand. In Appendix G is explained how the demand is converted from pieces to hours.

The model will divide the planning horizon into periods of time instead of continuous time. Only the start and end state of each period will be known, the time between is not. Since the model is intended to evaluate the planning strategies, the important seasonal fluctuations must be incorporated (Silver et al., 1998). Therefore, the planning horizon is set at a year. A period will correspond with a week because it shows enough detail in demand fluctuations, Figure 6-1, compared to months, Figure 5-1; while using days would show too much detail with regard to a planning horizon of a year. Moreover, the shifts, i.e. the available capacity, can be assigned on a weekly basis at WPV.

9.2 Model

The variables used to build the model can be divided into decision variables, model parameters, and performance indicators. These variables are used for controlling the outcome of the model, to set up the recursions and to evaluate the outcome, respectively. There are some differences between the variables related to the two kinds of customers, MTO and MTS, see section 3.1. These differences are explained first.

(27)

immediately. The ETL is the last production activity before the product can be distributed to DCs and since the time for distribution is not incorporated in this research, the output of the ETL is assumed to be a direct replenishment of the total inventory level in the DCs. The different subsystems are depicted in Figure 9-1.

Figure 9-1 – MTO and MTS demand

9.2.1 Decision variables

The decision variables are the variables that can be controlled by the decision maker. The first decision variable is how many shifts per period must be employed which represents the available capacity: Ctav. Secondly, it must be decided how much capacity is needed to satisfy the MTO

demand, this decision variable cannot be greater than the available capacity in a period, equation (1): tav

mto t C

c ≤ . The MTO demand must be satisfied first, since it is bound by the arrival time of the demand and the due date. In contrast, the replenishments of the MTS demand can be varied. Therefore, the third variable depends on the capacity left after the capacity allocated to MTO demand is subtracted from the available capacity, equation (2): tmto

av t mts

t C c

c ₌ ₋ _{. Strictly, it is}

not a decision variable, but it is directly related to decision variables.

9.2.2 Model parameters & recursions

In this section is explained how the decision variables are incorporated in recursions. These recursions are used to describe the behavior of the demand and available capacity.

MTO model element

(28)

Figure 9-2 – MTO model element Production & QC MTO queue ETL t + CLT t t + PTT ct mto delivery time customer

It is assumed that immediately after arrival at WPV, the MTO demand (in hours) will be released into the production system (t), Figure 9-2. This assumption does not necessarily violate the CONWIP control mechanism principles because MTO demand will pass MTS demand in the release list; it is assumed they have priority. Also the queue can be justified with regard to CONWIP because it can be used as a buffer to prevent idle time of the ETL. The average time to traverse the production and quality control operations, i.e. to reach the queue, is the production throughput time (PTT), which is relatively constant since it concerns non-bottleneck operations and it is controlled by CONWIP (Hopp and Spearman, 2008). The queue contains MTO demand of previous periods, waiting to be processed by the ETL. The ETL serves the queue every period a certain amount of time, i.e. the allocated capacity (ctmto), which determines the deliveries in that period.

The demand arriving at the queue at the beginning of a period, atmto = dt+PTTmto, should be completed at the end of the period before the due date, (CLT – PTT - 1). Then, the demand is finished within the CLT. The aim is to allocate a minimal amount of capacity to each period such that the total allocated capacity of the periods [t, t + CLT-PTT- 1] is greater than or equal the amount of work in process present in the queue at the end of the previous period plus the demand arriving at the queue in the beginning of that period. It is expressed by equation (2), t = 1,…,T, where T represents the planning horizon.

mto t t mto t t t CLT PTT W a c [ , + − −1]≥ ₋₁ + Eq. (3) where: 0

W

initial amount of work in process in the queue,

W

_T

≤

W

₀so the same amount of work in process or less is transferred to the beginning of the next year

mto t

a MTO demand arriving at the queue before the ETL in period t, assumed to

be known at the beginning of the period = dtmto₊PTT t

W

amount of work in process in the queue at the end of a period =

(29)

mto t

c

capacity allocated manually to process MTO demand during a period,

av t mto t C

c ≤ , see equation (1)

It must be emphasized that allocating the capacity manually does not generate an optimal solution, rather it generates a solution. The decision maker has to make the choice when to allocate how much capacity. This should be done for all the periods within the planning horizon, and then by means of an iterative process a solution can be found. The aim must be to allocate as little as possible capacity to satisfy demand and not to exceed the available capacity. On the other hand, the targets of the performance indicators discussed below should be achieved. MTS model element

As aforementioned, the task of the ETL with regard to MTS customers is to replenish the inventory level in the DCs. Therefore, the aim of the model is to capture the balance between the replenishments and the customer demand; both affect the MTS inventory.

The MTS inventory level is represented by the net inventory, a positive net inventory indicates that there is physical stock; a negative net inventory level means that there are backorders (Hopp and Spearman, 2008). Backorders are the demand not immediately filled from stock and it is assumed that those backorders has to be produced by WPV in future periods additional to the regular demand. In practice, this is what WPV does when they produce products for items which are OoS. The following recursion is used to calculate the net inventory level of MTS demand, t = 1,…, T.

The net inventory level is

mts t mts t t t

I

c

d

I

₌

₊

₋

−1 Eq. (4) where: 0

I _{initial total net inventory level in the DCs}

mts t

c capacity allocated for production of MTS demand,

mto t av t mts t C c c = − , Eq. (2) mts t

d arrival of the MTS customer demand in period t at the DCs, assumed to be

known at the beginning of a period

t

OH

_{the on hand inventory =}

_max(

_;

₀

₎

t

I

t

B

_{the backorders =}

_max(

_;

₀

₎

t

I

(30)

9.2.3 Performance indicators

The outcome of the model should be an allocation of as little capacity as possible while achieving the targets of the performance indicators. Three performance indicators are defined: reliability of delivery for both customers, i.e. the service levels, and the average on hand inventory.

The reliability of delivery is estimated differently for both customers. For the MTO demand it is represented by the fraction of demand that is delivered within the CLT, while for the MTS demand the fraction of demand that is immediately filled from stock is determined. Given that the planning is done for a year, the fractions are calculated annually. In doing so, it gives a good estimation of the service level calculations for MTO demand and the On Time and In Full calculations for MTS demand, i.e. the real measure of delivery performance of WPV, see Appendix A . The service level estimations explained below for MTO and MTS demand are named MTO service level and MTS fill rate, respectively.

For the MTO service level, the demand delivered on time in each period is calculated, which is the amount of demand that is delivered within the CLT. There are three possibilities with regard to on time delivery of demand: the total demand, a part of the demand, none of the demand is delivered on time. The total demand is delivered on time if ctmto [t, t + CLT-PTT- 1] is greater than or equal to Wt -1 plus atmto, i.e. the capacity allocated for processing MTO demand is large enough to finish the demand in the queue and the demand arriving at the queue on time. When ctmto [t, t + CLT-PTT- 1] is smaller than Wt -1 plus atmto, but bigger than Wt -1, only a part of the demand will be delivered on time: ctmto [t, t + CLT-PTT- 1] minus Wt -1. None of the demand will be delivered on time when the demand in the queue Wt -1 is greater than or equal to the capacity allocated for processing MTO demand: ctmto [t, t + CLT-PTT- 1].

Equation (5) summarizes the three given possibilities into one equation. Next, equation (5) is used in equation (6) to calculate the annual MTO service level. To prevent taking into account capacity of periods beyond the planning horizon, the latest period that should be included in equation (6) is T- (CLT –PTT-1):

(

)

(

mto

)

t t mto t mto t

c

t

CLT

PTT

W

a

s

_min

_max

_[

_,

₁

_]

_;

₀

_;

1 −

−

+

=

Eq. (5)

The annual MTO service level is:

∑

− − − = − − − =

=

) 1 ( 1 ) 1 ( 1 PTT CLT T t mto t PTT CLT T t mto t mto

d

s

S

Eq. (6)

In order to estimate the MTS fill rate, for each period the quantity of demand that can immediately be filled from the on hand inventory level at the end of the previous period is calculated, equation (7). The annual MTS fill rate is calculated by equation (8).

(31)

)

;

min(

₋₁

=

t mts t mts t

d

OH

s

_{Eq. (7)}

The annual MTS fill rate is:

∑

= =

=

T t mts t T t mts t mts

d

s

S

1 1 Eq. (8)

The last performance indicator is the average on hand inventory of MTS products, i.e. how much inventory is held on average per week in the DCs. Although costs are not part of this research, it can be stated that an increase of the average on hand inventory will lead to an increase of the operating working capital related to inventory and vice versa.

The average on hand inventory is:

∑

=

T t t

OH

T

oh

T 1

1

Eq. (9) 9.3 Remarks

The model is a derivation of the reality. This section will summarize remarks and assumptions on which the model is based.

Comments performance indicators

All the performance indicators, MTO service level and MTS fill rate, and average on hand inventory level, are calculated by using hours. In reality, these figures are calculated by using number of pieces. Therefore, the performance indicators can deviate from reality because the demand not filled immediately during a year are, for example, only belong to product group 3, instead of an even representation. The performance indicators are to be interpreted as an estimation of the reality. Besides, it was decided that adding more detail and more complexity to the model, by converting the performance indicators into pieces, would decrease the utility of the model.

Assumptions

The model is based on several assumptions which should not be overlooked in order for the model to be useful; therefore, the main assumptions are summarized here:

o Time periods of a week are used, of which only the start and end state will be known, section 9.1.

o The output of the ETL is a direct replenishment of the total inventory level of the DCs, section 9.2.

o Unfilled demand is backordered, section 9.2.2.

o The MTO demand is immediately released into the production system, section 9.2.2. o The performance indicators are expressed in hours, not pieces, affecting the accuracy of

(32)

10 Using the capacity planning model

In this chapter the use of the capacity planning model, constructed in chapter 9, will be illustrated. It represents the model solving stage of Mitroff and Kilmann’s (1977) model. First, the capacity planning model is used to evaluate the production strategy. Secondly, a sensitivity analysis will be performed.

10.1 Production strategy evaluation

In the diagnosis phase it is concluded that given the demand of 2009, employing two shifts at the ETL does not generate enough hours to satisfy demand while the employment of three shifts generates too much hours. In this section, the capacity planning model will be used to find a level production strategy that is able to achieve the performance indicators, i.e. the target service level of 98% for both customers and the lowest possible average on hand inventory.

In Figure 10-1, the sequence of stages to solve the model and to find a suitable level production strategy, is given. First, the available capacity must be set per period within the planning horizon, i.e. decided how many shifts must be employed. Next, in each period capacity must be allocated to satisfy the MTO demand at a service level of 98%. Third, the capacity left for MTS production can be calculated, see equation (2). Fourth, the performance indicators are estimated. If the outcome is not satisfactory, redo step 2 till 5.

Figure 10-1 – Sequence of stages

10.1.1 Variables

(33)

10.1.2 Results production strategy evaluation

It is already determined in the diagnosis phase that the employment of three shifts is a feasible scenario but it generates too much capacity, and consequently has too much labor and inventory costs. However, this scenario, named scenario A, is solved here with the capacity planning model to create the starting point, from where it is possible to establish the most desirable solution; less shifts employed and less average on hand inventory.

In Figure 10-2, the scenario when employing three shifts is depicted. It shows the available capacity and its restrictions. The maximum number of periods in which three shifts can be employed is 44, due to the summer period of 9 weeks in which only two shifts can be employed. Furthermore, the net inventory position of MTS products and the total demand, MTS and MTO, are displayed. In Table 10-2, the numerical results are summarized; the average on hand inventory of this scenario is 716 hours.

Figure 10-2 – Scenario A: production strategy with maximum number of weeks with 3 shifts

0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 20 25 30 35 40 45 50 Weeks 2009 D e m a n d a n d a v a il a b le c a p a ci ty ( h o u rs ) 0 200 400 600 800 1.000 1.200 N e t in v e n to ry ( h o u rs ) Total demand Available capacity Net inventory position (MTS)

(34)

Table 10-1 – Alternatives to switch from 3 to 2 shifts per week

Alternative Weeks from 3 to 2 shifts tot T av T d C − (hrs)

oh

T (hrs) 1 1 – 4 6 584 2 24 – 26 22 650 3 36 – 38 22 677 4 50 – 53 22 709

The graph of the best alternative production strategy, named scenario B, with regard to satisfying the performance indicators and having the lowest average on hand inventory given the demand of 2009, is depicted in Figure 10-3. The three circles indicate the differences between this graph and Figure 10-2. Starting to build the anticipation inventory as late as possible in the beginning of the year, the left circle, causes the peak of inventory to be lower, the second circle. The decrease of inventory is the same in the weeks 12 till 35, so the dip in inventory is also lower than in Figure 10-2, the right circle. However, this does not result in lower service levels since the initial on hand inventory is high enough.

Figure 10-3 – Scenario B: production strategy with an adjusted number of weeks with 3 shifts

0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 20 25 30 35 40 45 50 Weeks 2009 D e m a n d a n d a v a il a b le c a p a ci ty ( h o u rs ) 0 200 400 600 800 1.000 1.200 N e t in v e n to ry ( h o u rs ) Total demand Available capacity Net inventory position (MTS)

Table 10-2 – Numerical results of scenario A and B

Scenario A Scenario B Total demand dTtot (hrs) 5.546 5.546

Total available capacityCTav (hrs) 5.688 5.552

Fraction 2 shifts/ 3 shifts 9/44 13/40

T

(35)

10.2 Sensitivity analysis

In the previous section, a solution was derived using a set of variables. However, these variables are subject to internal or external forces that can cause changes. Therefore, it is worthwhile to use the capacity planning model to execute a sensitivity analysis to see how the decision variables and performance indicators react on internal or external based modifications of the variables used (Ulrich and Eppinger, 2008). The model will be used in the same manner as it was used in the previous section 10.1.

Not all parameters of the model are used in this sensitivity analysis. It was chosen to include the efficiency, the MTO and MTS demand, and the target MTO service level. The efficiency variable is the only internal variable, that is, it depends on the performance of the ETL, see section 5.2. Changing this parameter will reveal how the production strategy will perform when the performance of the ETL is altered. The MTS and MTO demand variable are external parameters, dependent on the customers. Last, the target MTO service level is dictated by service level agreements between WPV and the customers, i.e. other BUs. Therefore, it can only be changed by mutual agreement.

10.2.1 Values of variables

The values of the variables used in the sensitivity analysis are summarized in Table 10-3.The base case values represent the variables that are used to obtain the graph in Figure 10-3, i.e. scenario B. An efficiency of 27% was used to convert the demand from pieces to hours, see appendix G. The MTS and MTO demand of 2009 were set at 100%. And the target MTO service level was 98%. Next, each variable is changed to simulate a modification, steps lower are indicated by a (-) and higher by a (+).

The changed efficiency values are based on the achieved monthly efficiency values of the ETL in 2009; the lowest and highest value are represented by case (-1) and (+1). The (+2) case represents the target efficiency value of the equipment in paint production unit, see appendix F. The values of the MTS and MTO demand in the columns left and right from the base column, are chosen arbitrarily, but they represent a change of the demand of plus or minus 15 or 30% with respect to the 100% base case demand. Also a modification of the target MTO service level is incorporated, to see how the production strategy will perform when the target MTO service level is set higher or lower.

Table 10-3 – Values of the variables used in the sensitivity analysis

(36)

The model is used, as prescribed in Figure 10-1, to perform the sensitivity analysis. So, 14 new scenarios are obtained additional to the base case. Each scenario is obtained by changing one variable; the other variables are equal to the base case. For example, the (-1) efficiency scenario uses an efficiency value of 23,3% instead of 27% while the other variables have the same value as in the base case, 100%, 100%, and 98%. This enables an evaluation of the effect of the decrease of the efficiency factor on the production strategy decision.

10.2.2 Results sensitivity analysis

The numerical results of 14 new scenarios and base case are presented in Appendix I . In Figure 10-4 the percentage of weeks per year in which three shifts must be employed can be seen for each scenario and the base case. The maximum number of weeks in which three shifts can be employed is 44, due to the holiday period, and is set at 100%. The base case scenario, for example, needs 39 weeks with three shifts which results in 89%. Since extra shifts involves extra labor costs, it is more desirable to be able to meet demand with as less weeks in which three weeks must be employed as possible.

Figure 10-4 – Fraction of shifts per variable modification

0% 20% 40% 60% 80% 100% -2 -1 base +1 +2 % o f 3 s h if ts Efficiency MTS demand MTO demand Target MTO SL

The graph in Figure 10-4 shows that the (+1) and (+2) MTS demand, the (+1) and (+2) MTO demand and the (-1) efficiency scenarios reach the maximum of 44 weeks in which three shifts must be employed. Only the (+1) MTO demand scenario is feasible since the total available capacity generated exceeds the total demand in that case, and all the performance indicators are achieved, see Appendix I

(37)

the timing of capacity allocation to the MTO product group, resulting in almost no effects on the ctmts left for MTS production.

Modifications of the MTS demand clearly have the biggest impact on the production strategy. The (+2) MTS demand scenario, the MTS demand is 130% with respect to the base case demand, will lead to the maximum of 44 weeks in which three shifts must be employed, but still a MTS fill rate of only 52% is achieved. While the other extreme, (-2) MTS demand scenario only needs four weeks in which three shifts must be employed and in the other weeks of the year, two shifts will suffice. Moreover, the average on hand inventory is lower, 564 hours, than the average on hand inventory of the base case: 584 hours, see appendix I.

(38)

11 Conclusion design phase

In chapter 8, Mitroff and Kilmann’s (1977) method is introduced which is used to structure the design phase. Recall that the design phase consists of the two stages: modeling and model solving. In the modeling stage, a scientific model was developed and it is solved in the model solving stage to obtain a solution. The solution should be in line with the research goal:

Construct a capacity planning model such that insights in the behavior of demand, available capacity, and MTS inventory can be obtained which are useful for the management of WPV because they can support future decisions with regard to the level production strategy.

The solution obtained in the design phase is a capacity planning model that is able to provide insights in the behavior between MTO and MTS demand, available capacity, and MTS inventory which is in line with the research goal. The outcome of the capacity planning model can be analyzed by the performance indicators.

The working of the capacity planning model is illustrated in section 10.1 by evaluating the production strategy which was not possible before in WPV due to a lack of a formal longer term planning model. The evaluation shows it is now possible to decide how much shifts must be employed per week to meet the demand and build anticipation inventory. Also, it can be determined when to build inventory to keep the average on hand inventory as low as possible. Moreover, the sensitivity analysis, section 10.2, showed the possibilities of performing a what-if analysis of possible scenarios by modifying the variables used. The management of WPV can use these results to discuss the various scenarios and take appropriate and founded actions, if necessary.

(39)

12 General conclusion

This chapter will outline the conclusion that can be drawn according to the insights obtained in the diagnosis and design phase. Also, a brief discussion of the result is made. Next, recommendation for the implementation of the capacity planning model will be given and additional benefits are summed. Finally, some general remarks with regard to the recurring seasonal demand pattern are given.

12.1 Conclusion

The motivation of this research project was the problem of low service levels. Moreover, the management of WPV acknowledges that their decision to pursue the level strategy in order to respond to the low service levels was a not well founded decision.

In the diagnosis phase it was established that the low service levels are a functional problem given the paint production unit as the system and its products as output. The low service levels can be explained by the under capacity of the paint production unit during the seasonal demand peaks. The production cannot satisfy the demand during those periods, which results in too little replenishments of the DCs and too late deliveries of the MTO customers, so service levels deteriorate. Since the bottleneck, the ETL, is the most critical resource, the research project is further focused on that operation.

The ETL was, not able to satisfy the demand. It appears that the efficiency of the ETL is far below the target, mainly due to time lost caused by setups. On the other hand, the level production strategy, as suggested by the management of WPV, proved to be a better alternative to meet demand than its alternative, the chase strategy However, the current production planning and control structure is not able to evaluate the level strategy well which explains the uncertain decision making.

Anticipation on seasonal demand

Master thesis