Master Thesis Technology Management Throughput times at Mallinckrodt Baker

(1)

Master Thesis Technology Management

Throughput times at Mallinckrodt Baker

Student: J.T. Smal

Student number: 1389157

Supervisor: H.L. Faber

Second Supervisor: D.P. Van Donk

Company: Mallinckrodt Baker, Deventer

(2)

(3)

Preface

This report represents the research I performed at Mallinckrodt Baker Deventer in the period from February 2009 till July 2009. I would like to thank the company for giving me the opportunity to do this.

Everybody at Mallinckrodt Baker was willing to help me and without that help I couldn’t perform this research. I want to thank everybody who helped in any way, especially Mike Te Dorsthorst, John Mittendorf, for good and useful feedback and guidance and Marco Voskamp for information and knowledge about the quality control process.

Also I like to thank my supervisors from the Rijksuniversiteit Groningen. Henk Faber, for his support and feedback during the whole process and Dirk-Pieter van Donk for his critical and useful feedback in the last phase of the research.

Sjoerd Smal

(4)

(5)

Summary

This research was initiated by the management of Mallinckrodt Baker, a producer of specialty chemicals. There are two problems that they faced. First, the customers experience long lead times and second, the stock levels are too high. A common factor in both problems is the throughput time. It was assumed that a decrease of the throughput time would led to a decrease of the lead times and a decrease of the stock levels. That is why the main research question was formulated as: How to decrease the throughput times of Mallinckrodt Baker?

In the first phase of the research, the assumption of the effect of the decrease of the throughput time was confirmed. After that, three main processes were selected for investigation, Incoming materials, Production and Quality Control. It was found that of these three, Quality Control had the longest throughput times, and therefore it was assumed to hold the most potential for improvement.

A detailed analysis was needed in order to investigate the Quality Control processes. First to confirm the assumption of the potential for improvement, and second to research the improvement possibilities.

Inside the Quality Control department, the major problem that occurred was the variation between the input and the output. This is the source of a significant waiting time of the samples before analysis. And this waiting time confirms the assumption that there is potential for improvement. With a throughput time of an average of 25 hours, almost 22 hours are spend waiting

Eight major areas of attention could be indentified that are an influence on the throughput time and could be adjusted in the scope of this research. For every area of attention there was a solution proposed to improve the situation of that area.

(6)

The basis for the proposed alternative is an adjustment of the allowed throughput times, this changes the constraints without increasing the workload. Other solutions aim to optimize the product en information flow in and outbound of the Quality Department.

(7)

1. Introduction

This is the master thesis about the research that was performed in 2009 at Mallinckrodt Baker in Deventer, a producer and distributor of specialist chemicals. This company faced several problems with regard to responsiveness and stock size.

A common factor in these problems is the throughput time. That is why the goal of this research is to find a way to decrease the throughput time.

In this report, Mallinckrodt Baker is introduced as first, in order to provide a bit of background. A more detailed description of the company can be found in the appendix II. This is followed by a elaboration and a theoretical background of the management problem in chapter 2.

The management problem leads to the research objective and the research questions as presented in the first part of chapter 3, directly followed by a description of the research methodology.

In chapter 4 the theoretical background of this research is discussed. This is followed by the presentation of the results in chapters 5, 6 and 7. These results will be the input for the improvement suggestions, that can be found in chapter 8, followed by a recommendation in chapter 9. Finally, a overall conclusion is presented in chapter 10.

(10)

1.1 Mallinckrodt Baker

1.1.1 Profile

Mallinckrodt Baker International (MBI) is a producer and distributor of specialist chemicals. Worldwide there are five production locations, three of them are located in North America, one in Deventer, the Netherlands, serving the European market and one in Malaysia for the Asian market. MBI has a catalogue of about 4,500 different products.

MBI is a subsidiary of Covidien, which is a large multinational healthcare company. The turn-over in 2008 was nearly $10 billion. Until 2007 Covidien was part of the conglomerate Tyco International and it was known as Tyco Healthcare.

Covidien has more than 41,000 employees working in 59 countries. Covidien produces in 65 manufacturing facilities, located in 16 countries, in addition, there are more than 4,000 sales representatives in more than 50 countries.

This research regards the only production location of Mallinckrodt Baker International (MBI) in Europe, which is located in Deventer. The production location is situated in Deventer since 1963, located near the river IJssel. About 200 people are employed here.

(11)

The department Quality Assurance & Control operates within the Operations Department, but it is in fact a hierarchically part of MBI.

1.1.2 Process overview

In figure 1.2 the process flow of MB can be seen, which starts with the customer in the upper right corner and will eventually also end there.

Figure 1.2, process overview

(12)

(13)

2. Problem description

2.1 The context of the research

Just as any other company, MB has formulated several goals and objectives that should help MB to maintain its profitability now and in the future. Two of these goals are increasing the customer responsiveness and decreasing the stock size.

Responsiveness

One of the goals of MB is to increase the customer responsiveness. In scientific literature many different definitions of (customer) responsiveness can be found. As Bernades and Hanna (2009) argued, the concept of responsiveness is often confused and interchangeably used with the concepts flexibility and agility. Based on their literature review, Bernades and Hanna (2009) distinguished the scopes and definitions of these concepts. That lead to the following definition of customer responsiveness:

A firm’s propensity to act on market knowledge to anticipate and/or rapidly address modifications in customer’s expectations (Bernades and Hanna, 2009)

That means in short; how fast can an organization react on actions in the environment. In most cases the environment is the customer and responsiveness is thus how fast a company reacts to the wishes of the customer. With a higher responsiveness, the wishes and demands of the customer can be fulfilled in a satisfying manner. That does not only embrace the delivery of the right product at the right time (logistics), but also developing new products (R&D), capturing the whishes and demands in the market (sales/marketing) and the ability to produce what is needed as quickly as possible (production), thus almost every part of the business is involved.

Stock Size

(14)

experience, just as is the case at MB. Defining the correct costs for holding stock would require a siginificant effort, therefore it is chosen to use the percentage of 25%, which is used throughout the company for matters, such as cost accounting.

MB has a yearly turnover of € 35 Mln and had in February 2009 a stock with the value of € 9 Mln, the goal is to decrease that to € 8,2 Mln at end of the fiscal year 09. The fiscal year runs from September till August. This goal was set by the management of Mallinckrodt Baker International.

Management Problem

The problem of the management of MB is that there are two goals that at first sight are conflicting. For a high level of customer responsiveness a high stock level for every product is very helpful, because then the customers can be supplied immediately, regardless of the product that is ordered. Low stock levels can results in longer customer order lead times, which decreases parts the responsiveness (the production and logistics parts), because the chance that a product cannot be delivered depends on the level of the inventory (Bertrand et al. 1998).

The management of MB aims to solve this problem within the operational departments. It was assumed by the management that lower throughput times would result in both lower stock levels and better responsiveness.

The question from the management of MB was thus:

Is it possible to decrease the throughput times of Mallinckrodt Baker?

(15)

2.2 Conceptual model

In order to understand the assumption that the throughput times influences both the responsiveness and the inventory a model was developed that incorporates the relationships between the elements that matter is this case. Which is shown in figure 2.1

Figure 2.1 Conceptual model of Responsiveness, inventory and Throughput time at Mallinckrodt Baker

Make-to-Stock (MTS) and Make-to-Order (MTO) items have different relationships with responsiveness and inventory.

- Responsiveness and Throughput Time only have a relationship for the MTO items, since the throughput time for these products is directly affecting the customer lead time. This relationship is a negative one. Higher throughput times result in a lower responsiveness. - The inventory can be divided in three parts, the raw materials, the work-in-progress and

the finished goods.

(16)

based on the planned usage of the materials. If the throughput time decreases, the raw materials are refilled more often, but the level remains equal.

o _{The throughput time has a relation with the work-in-progress and the finished} goods. For both MTO and MTS products the throughput time influences the work-in-progress, according to Little’s law (1961). This is a positive relation, if the throughput time increase, so does the work-in-progress.

o _{Obviously, only the throughput time of the MTS products does influence the} finished goods inventory, because only the MTS products are stored as finished goods. In order to calculate the required level of a certain product, it is needed to know the demand during the lead time (Bertrand et al., 1998), which has to be satisfied. Therefore, there is also a positive relation between the throughput time and the level of the finished goods. If the throughput time of a MTS product decreases, there is less demand to satisfy, so the level of finished goods can also be decreased.

(17)

3. Research outline

In this part the methodology of the research is explained, this is done to make clear how the research is executed and what methods are used. First it is determined what questions should be answered and second it is defined what path has to followed in order to be able to answer these questions.

3.1 Research objective

The objective of this research is based on the management problem and the question that is derived from it. This question is: How to decrease the throughput times of Mallinckrodt Baker?

The objective of this research is: To decrease the throughput time of the products at Mallinckrodt Baker In order to reach the objective, several research questions need to be answered;

1. What is throughput time?

The objective of this question is to provide a theoretical definition of the concept throughput time.

2. What are the current throughput times at Mallinckrodt Baker?

The objective of this question is to uncover the throughput times at MB via an analysis of the primary process.

3. Which part of the process has the most potential for improvement?

Based on the results of question 2 one part of the primary process will be defined for further research.

4. What are the current throughput times of this specific part of the process?

This question has the objective to provide an in-depth analysis of the process part that was defined in question 3.

5. How can the throughput times be reduced?

The objective of this question is to provide several ideas that can help to reduce the throughput times.

6. How can the solutions be implemented at Mallinckrodt Baker Deventer and what are the expected effects?

(18)

3.3 Research Approach

In this section it is explained how the research is designed and how it is executed. This will be done by subsequently elaborate all the necessary steps needed for answering the six research question from the previous section.

3.3.1 First steps

In the first phase of the research, the scope embraces the entire operations processes of MB, only the shipping of finished goods is excluded. Thus the research area is defined by all the activities from the moment of receiving the raw materials until the moment a product is ready for shipping. At MB, that is a complicated research area. MB offers more than 4,500 different products, of which two-third is produced on site and all these products do have different routings.

For the research, it is needed to narrow down the scope and isolate a smaller part of the whole, because the time and the manpower to perform the research is limited. It is not possible to investigate the whole area of research as it was initially defined. That is why is chosen for a funnel approach with three basic steps:

− A general analysis of the situation of the whole scope − A detailed analysis of a subset of the scope.

− Recommendations, applicable to the chosen subset.

(19)

Figure 3.1Research steps

The first question, How can throughput time be defined? is necessary for a theoretical background of the subject. A literature review will be executed in order to determine a solid definition of the concept of throughput time.

3.3.2 General Analysis

The second question, “What are the current throughput times at MB?” represents the first part of the general analysis. In order to answer this question, there are two different options how to do that. Both options can be used to execute this second part of the research and thus to narrow down the scope.

− Method 1: take a few representative products or product groups and follow these through the whole process. For this approach it is needed to map the flow from raw material to finished product. In order to determine the throughput times, it is also needed to connect batches of raw materials to batches of finished products.

− Method 2: isolate a part of the process and analyze that for all the products and this way determining the throughput times for each process steps.

(20)

that a good choice of the investigated products is made, because the diversity is very high at MB, in terms of products, volume and production methods.

A drawback of method 1 is that is not possible to do it with the current data recorded in the system of MB. Some steps are not measured in the system. A possibility is to connect raw materials and finished goods and measure the throughput time manually, which is a time-consuming method.

The advantage of method 2 is that all products can be taken into account, because every product goes through at least one of the steps that are investigated.

The drawback of method 2 is that the connection between the steps is lost. The data needed to determine these throughput times is relative easy to retrieve. The data is recorded in the system or it is possible to derive it.

It is chosen to use method 2 and divide the process into several steps ofwhich the throughput times are determined. This method is chosen because the most data for this option is available, so it is possible to derive reliable data in a short period of time. This data can be derived from Prism, the information system that Mallinckrodt Baker uses. Furthermore, several semi-structured interviews will be held in order to gain insight in what the data means. These interviews will be held with stakeholders who work in the selected production steps.

(21)

maybe because of technical reasons. If that is the case, another process part has to be selected and researched.

3.3.3 Detailed Analysis

In order to verify the choice made after the general analysis and to create a good starting point for improvement, it is needed that the chosen process part is analyzed in detail. In this section, first a analyzing tool suited for this situation and purpose is chosen. Second, the chosen analyzing tool is explained in more detail.

3.3.3.1 Analyzing tools

By answering the third question, also the subject of the reminder of the research is chosen. The fourth question, “What are the current throughput times of this specific part of the process?” digs deeper in the chosen process part. The goal of this question is to make it clear what this process part is, what it does and what the performance of the process part is. The process has to be modeled and analyzed.

In order to gain insight in the chosen process part, it is helpful to model it (De Leeuw, 2000). Sequentially, the model has to be analyzed. There are several approaches and methods available for that. A difference has to be made between tools that provide one part of the analysis and approaches that embrace all analysis steps.

Aguilar-Saven (2004) identified a whole range of business process modeling tools. The tools mentioned by Aguilar-Saven (2004) are qualitative tools that provide a picture of the process structure. The perspective and the symbols are depending on the tool of choice. Aguilar-Saven (2004) included tools like Flow-chart, Workflow, IDEF0, Role Interaction Diagram among others. Besides the modeling tools, there are also more complete approaches, which include also analyzing and improvement of the process. These are approaches like the Theory of Constraints (TOC) (Gupta and Boyd, 2008) and Value Stream Mapping (VSM) (Rother and Shook, 2003). In this research, it is chosen to use such a complete approach, because then the modeling and analyzing tools are certain to fit to each other.

(22)

the TOC, where the process has to be known in order to identify the constraint. The advantage of VSM above TOC is that VSM has a focus on the throughput times of the process steps. Also VSM shows the possibilities for improvement of the whole process, while the problem solving part of TOC only focuses on the identified constraint.

3.3.3.2 Value stream mapping

So it is chosen to use Value Stream Mapping to unveil the characteristics of the chosen process part. That will be done by a making a current state VSM (Rother & Shook, 2003). A current state VSM shows the process of all value adding activities at a company from door to door. It shows the product and the information flows, together with measures like people involved, processing time, que-time, waiting time and transporting time. It is a powerful tool to show just the value adding activities and the times involved. VSM is a technique that is originated in the lean philosophy. It is a tool to discover waste in the value adding process, waste that has to be eliminated, according to the Lean Principles (Womack and Jones, 1996).

The VSM has to be made in cooperation with stakeholders of the subject process or processes. This is done with those stakeholders, because they do have the most information about the actual processes. It is important to capture the actual process instead of the ‘theoretical’ process, which might probably exist in a company’s archive. The actual process is the process which has to be adjusted or improved, because the problem also occurs there.

In this research the stakeholders will be employees from the process part that will be selected after the general analysis. With a brainstorming session and several semi-structured interviews the current state map will be defined. The data accompanying the current state map will be retrieved from the companies information system, Prism, and eventual other sources available.

(23)

In sum, in order to perform a good VSM activity, the following steps need to be performed (Rother & Shook, 2003):

1. Defining the customers and the value provided to that customer. These are the starting points of the value stream map.

2. Defining all process steps which are necessary in order to deliver the value to the customer.

3. Subsequently, all the previous defined steps are connected to each other. A current state value stream will appear.

4. Assign all associated numbers like processing time, waiting time, transport time, buffer size, filling rate and other relevant numbers.

5. When the map is finished, problem area’s can be identified. The first improvements can already be devised through brainstorming, other improvements require more time and research.

The problem areas and the improvements will be the guide for the design of the future state value stream map. This new map shows how the process will look like if all the improvements are implemented.

3.3.4 Improvement

Once the current state is mapped, the future state has to be designed. A future state VSM is the VSM that should be obtained after all improvements are done. In order to reach the future state, several tools or techniques can to be used. These tools or techniques can be identified by a literature review and brain storming. This will result in an extensive list of ideas. Not all of the generated ideas can be transformed into a solution.

It is important for MB that it is possible to use a proposed solution. Therefore it was decided to formulate several criteria in cooperation with MB. In order for an idea to be further developed into a solution it has to meet these criteria.

- Executable.

(24)

- Significant.

The gain of the solution has to be significant with respect to the length of the throughput time.

- In proportion.

The investment, financial and time-wise, that has to be made to make the solution work, should be in proportion to the gains of the solution. The result should be at least break-even.

From the initial list of ideas, several solutions can be developed, in order to answer the fifth: question “What solutions can be used for improvement of the throughput times?”.

3.3.5 Implementation

The next step is to asses these developed solutions for the specific situation of MB. This can be done by answering the last question; “How can the ideas be implemented at Mallinckrodt Baker

Deventer and what are the expected effects?

The effects and implications of the solutions on the goals, the organization and the financial resources will be evaluated here. This will result in a recommendation of one or several of these solutions to MB.

The goal of this research is to decrease the throughput time. However, the solutions will not be evaluated solely on a reduction of throughput time. The underlying problems are responsiveness and stock level. The effects of the solutions on these areas is also incorporated with the evaluation.

When MB implements the selected solutions, then the designed future state will be reached. That then becomes the current state and the process of improvement normally would start again. The lean philosophy encourages continuous improvement, so this results in an iterative process. This research might also result in a methodology for improvement of other process parts.

(25)

Figure 3.1, Phases of the VSM process (Rother & Shook, 2003)

In sum, in order to research the throughput times at MB, the operational processes are broken down into three subsets. These three subsets are analyzed and based on the performance, one is chosen for further analysis. The analysis of this subset is guided by the principles of value stream mapping, which also provides for generation of solutions for improvement.

It is chosen to use historical data over a period of one recent year (2008), this period is long enough to cover possible seasonal trends. Also, it excludes obsolete products that may influence the results.

(26)

(27)

4. Theoretical background

In this chapter the theoretical background of this research is presented. First a literature review is presented of the concept of throughput time, thereby answering the first research question. Second, the concepts of stock size and responsiveness will be evaluated further after the brief review in the problem statement (page 12)

4.1 Throughput time

This paragraph aims to answer the first research question, how can throughput time be defined?. Throughput time is the core concept of this research, therefore it deserves a more elaborate discussion than simply providing a definition from the literature, as was done in the previous paragraph.

Throughput time is: “The longest time required to convert materials into finished products” (Psossl, 1988).

This definition implies that the throughput time is the time that a product can be qualified as Work-in-Progress (WIP). The scope of this definition can be adjusted to any situation where materials are converted into something else, the product, even if it is a virtual conversion.

The concept of throughput time and WIP gained attention when the lean manufacturing systems were emerging. Reduction of waste is an important lean principle. Unnecessary WIP can also be regarded as waste. Focusing on the throughput times is one of the methods to reduce the WIP.

Johnson (2003) identified several factors that determines and influences the throughput time. These factors are:

- Setup time - Processing time - Move time

(28)

Johnson (2003) also provided a framework on how to reduce the throughput time. The essence of that framework are three points:

- Reduction of batch sizes

- Reduction of work station utilization rate

- Review of earlier implemented procedures or policies

4.2 Stock size

The stock size can be defined as: The value of all products and materials that are stored in the

warehouses or are being processed.

Because of the term value in this definition, it implicates that there is chosen to use a financial measure. The stock level can be measured by several units of measurement, which can be units, kilos, liters, square meters etcetera. This gives problems when the properties of the products differ, which is the case at MB. Therefore the value of the stock is generally used to indicate the stock level. At MB the fully absorbed cost price(Atkinson, Kaplan and Young, 2004) is used to value the stock.

As was mentioned in section 2.1, the throughput time has an effect on the stock size. Via Little’s Law (1961) on the WIP and via the safety stock calculation on the stock size of the MTS products (Bertrand et al., 1998).

From section 2.1 it became clear that the problem of the management is not really with the stock size, but rather more with the costs associated with keeping stock (Atkinson et al., 2004). These stock costs are caused by several factors:

- Material handling activities; All activities that are associated with transporting and handling materials in, out and through a warehouse. The costs are personnel costs and costs for machines like forklifts.

(29)

- Others; There are also other costs associated with inventory, costs like insurance, taxes etcetera.

In order to measure the possible effects of throughput time decrease, it is necessary to predict the stock size after implementing the recommended solutions and calculate the stock costs.

4.3 Responsiveness

Responsiveness is defined as a firm’s propensity to act on market knowledge to anticipate and/or rapidly

address modifications in customer’s expectations (Bernades and Hanna, 2009)

In short, responsiveness is about reacting as quickly as possible to what customer wants. One of the determinants of responsiveness is lead time (Holweg, 2005) and thus the throughput time has an influence on the responsiveness via the lead time.

Holweg (2005) defines a set of factors that determine responsiveness. These factors can be quantified, but it is not clear yet how the responsiveness should be quantified. The determinants are all defined in different units and concepts, an overall measure of responsiveness is missing. Beamon (1999) defined for measures to capture flexibility. However, flexibility is not the same as responsiveness, but it is quite similar. Flexibility was defined by Beamon (1999) as the ability to respond to a changing environment. The concept of responsiveness is broader than the concept of flexibility, but the defined measures of Beamon (1999) address all three categories defined by Holweg (2005).

The best way capture the responsiveness is probably with a customer survey, because responsiveness is aimed at reacting to the customers. The customers could grade the performance of a company.

(30)

4.4 Definitions

Here the definitions of the key concepts of this research are summarized in this section.

Throughput time: The longest time required to convert materials into finished products (Psossl, 1988). For this research that is the time from the moment the raw materials enter the site at MB until the finished goods are ready for shipping, but excluding the actual shipping.

Lead time: The total elapsed time from the determination of the need of customer for an item until it is available for use (Psossl, 1988).

Stock size: The value of all products and materials that are stored in the warehouses or are being processed.

(31)

5. Measured Throughput Times

In this section, two questions will be answered. These questions are:

- What are the current throughput times at Mallinckrodt Baker? - Which part of the process has the most potential for improvement?

First an overview of the evaluated process is presented, followed by the method of analyzing this process. That results in an overview of the characteristics of the process, including the throughput times. From these results it should be possible to determine the subject which shows the most potential for improvement.

5.1 The Production Process

The sequences of the required steps for production are shown in the picture below.

Figure 5.1, primary process

At MB there are three basic production steps; these are the production of the Basic Mixtures, the production of the Intermediate Products (“Half-Fabrikaten”) and the production of the Finished Goods. The production of finished goods includes all packaging activities.

(32)

At arrival of raw materials and after some production steps a sample of some batches is send to the quality department. It is dependent on the product and its prescriptions how often a product goes to the Quality department. Some need a final inspection and others do not.

Batches that are “on inspection” are held in separate temporary inventories. Raw materials are an exception, because the materials are moved into the warehouse already, while the batch still has the status “on inspection”. However, the materials cannot be used until the Quality department approved the batch and the batch is thus released and the status is changed in “AVLB” (available). This results in a theoretical inventory of raw materials on inspection.

5.2 The data

In order to answer the second and third research questions an initial scan will be performed. This will provide insight in the throughput times and its characteristics. As was described in section 3.2, several steps in the primary process will be analyzed separately. The three main processes that are chosen to be evaluated are:

- Handling and acceptation of Incoming Materials. - Production of Finished Goods.

- Quality Control process.

All products has to go through at least one of the these processes, so they will all be part of the analysis. Also, it is possible to derive comparable data for these three processes from Prism. The processes of the handling on incoming materials and Quality Control do have some overlap. It has to be evaluated what the effect is of that overlap.

The source of the data is the companies ERP system, Prism. In Prism, every change in the status of batches in material handling, production and traffic is recorded. These status changes indicate where in the process a batch is at a certain moment of time. For this research, the recorded moments of importance are;

- Arrival of the products at the warehouse

(33)

The data used for this scan are from the period starting on 01-01-2008 till 24-02-2009. The period was determined based on the requirement that it should provide a representative set of data. It had to be at least one year, in order to capture possible seasonal trends. So, the complete year of 2008 and the first months of 2009 till the moment of request of these data was provided.

The derived data is not very precise; it shows the throughput times in days, but that is sufficient for this phase of the research. That is sufficient, because this phase of the research aims at making a choice between three process steps. The data is for each process derived in the same way, so it is a fair comparison. More detailed data of a specific process will be obtained if one process is chosen to be the focus of this research.

5.3 Results: General Analysis

In this section a summary of the results of the first research part can be found. The detailed results and analyses can be found in appendix III. These results will provide an answer on the first two research questions, which aims to uncover the throughput times of MB.

Incoming materials

Handling the incoming materials is the first step of the operations process. It involves all activities that are needed to get the materials ready for use by the production process. These are activities like unloading of trucks, transport, administration, labeling and inspection. The materials that are handled here also include the packing materials, like bottles, drums, containers and boxes.

The materials that arrive get a certain classification, based on that classification a certain activity has to be performed. If that is completed, the batch gets the classification AVLB, which stands for available, which indicates that the batch is ready for use. Some materials get the classification AVLB directly, so these products do not need any additional handling or activity. The possible classifications can be found in appendix I.3

(34)

Classification Amount (batches)

Percentage Average Throughput Time (days) AVLB 2530 31% 0,1 CNTR 2181 25% 0,7 INSP 2256 26% 4 CHEC 152 2% 0,9 HFETIK 1387 16% 0,9 Total 8508 100% 1,4

Table 5.1, Incoming materials

It can be seen that the batches that have to inspected by the quality department take the longest time. As was mentioned before, the unit of measurement for the throughput time of this step is measured in number of days. That is the smallest unit that is stored in the system for this kind of data.

Production

(35)

Processing time in days

Finished on: Amount

(Batches): Percentage day 1 17634 83,5% day 2 2231 10,5% day 3 309 1,5% day 4 609 2,9% day 5 149 0,7% day 6-10 149 0,7% day 10-17 42 0,2% Total 21123 100,0% Table 5.2, production

The throughput time is measured in days, if a batch is finished on day one, that means the batch is finished on the same day that it is started. If a batch is finished on day two, that means that the batch is finished the next day after the start and so on.

It is shown that the most batches are finished within one day.

Quality control

There is a large part of batches that have to be inspected by the quality department, both raw materials and finished goods. In the analyses of the handling of the incoming materials, the quality control was already included, in order to provide a complete picture of that process. In this section all the batches that are inspected are considered. That means that the batches of raw materials are incorporated again. This is done because for the quality control department, there is no difference in handling raw materials or finished products. And as it turned out, it does make no difference for the results.

(36)

It has to be noted that there is difference between the time that is measured here and the time that was measured for quality control by the Incoming Materials section. There are two reasons for this difference. First, this section covers all products that are inspected and not only the

In sum, the throughput time for Quality Control, 2,7 days is significantly longer then the throughput times for handling the incoming materials, 1,4 days and production, 0,23 days. As was determined in section 5.2 (page 31), this data draws a incomplete, but sufficient picture about the throughput times at MB. So, based on this data, it possible to answer the question; What are

the current throughput times at Mallinckrodt Baker?,

5.4 Focus

As stated in the section Research Approach, the initial scope is too wide for the given timeframe, so the scope has to be narrowed down, something which should be done carefully. This choice will be based on the results presented in the previous section.

The area of research has to meet two requirements. First, there should be potential for a significant reduction of the throughput time. Second, a reduction in this area has a relevant (financial) impact. The choice made here is an answer on the third research question; Which part of

the process has the most potential for improvement?

Three major areas are subject of the general analysis; Handling Incoming Materials, Production and Quality Control.

Average throughput time (days)

Incoming materials 1,4

Production 0,23

Quality Control 2,7

(37)

automatically the right choice. Therefore, a verification is needed in the next research step. A first verification was already given by the supervisor of the Laboratory, who stated that a large part of the samples theoretically can be analyzed within one day.

The Quality control process is involved with all steps in the primary process, as can be seen in the figure 5.1 (p. 28). The impact of one improvement within this department effects in several places throughout the primary process. It will improve the availability of raw materials and finished goods.

The Quality department meets the two requirements mentioned before (potential in time and impact) and therefore it shows the most potential for improvement. Thus it is obvious to choose this department as the subject of further research.

This choice is based on the general analysis, however this scan was based on extensive data, it is not an in-depth analysis. It is important that the findings are confirmed by the in-depth analysis of the Quality department, so that the choice is validated.

It should be mentioned that a local optimization of a part, in this case the Quality department, is not ideal from a supply chain perspective. It can lead to a lower performance of the whole chain, which is of course not something that is wanted. That is why the improvements in this small part always should be evaluated with respect to the whole chain. Then it is possible to improve the performance of the supply chain. To reach optimal performance of the supply chain, more of these steps will be necessary.

(38)

(39)

6. The Quality Control process

In the previous section, it was established that the Quality Control process showed the most potential for possible improvement. The throughput times are significantly higher then other process steps. It has to be determined what causes this difference. This is the reason why in this section the Quality control process is examined in more detail.

The obvious step would be to directly answer the next research question, what are the current

throughput times and characteristics of this specific part of the process? However, in order to place the quality control in perspective, in this chapter it is first made clear why quality control is of such importance for the process of MB. Second, it is determined what factors might influence the throughput time in this process part. The performance, and thus answering the question, is discussed in the next chapter.

6.1 Quality in the chemical business

MB produces specialty chemicals which are used in a wide range of applications, from the production of computer chips, analyses of blood and human tissue and as ingredients for medicines. All these applications require working on a very high precision level, which means that the products need to be exactly what is said they are.

Quality can be defined as delivering products which meet the requirements and specifications as defined for that product.

The products have certain specifications and customers rely on the given specifications when they use it. If the sold product does not meet the given specifications, this will result in a wrong analysis for example, which can have serious consequences in a medical environment. The quality has to be constant and be guaranteed. Kano (1996) describes a model in which quality attributes are measured with respect to their effect on the satisfaction of the customer. The specifications and requirements of the products of MB can be seen as must-be attributes; an

(40)

has to be documented exactly. MB International has an extensive GMP production capacity in the US. In Deventer, only a small production line is GMP certified.

Furthermore, MB has the ISO 9001:2008 certification, which means that MB complies too the regulations for quality management according to the ISO standards.

6.2 Conceptual model of the quality control process

The goal of the research is to decrease the throughput time. It is chosen to focus on the quality control department. In order to determine what factors inside the quality control department are important with regard to the throughput time. Therefore a detailed conceptual model is developed of the throughput time at the quality control department.

This conceptual model is shown in figure 6.1.

Figure 6.1, conceptual model of the throughput times at the Quality department of MB

(41)

Factor: Influence: Allowed

Throughput time

Predefined analyses time. This defines the maximum time the Quality department can take to finish the analysis

Product type Different products need different analyses

Batch size The size of the batches determines the number of batches, which determines the number of inspections Capacity The capacity defines how many analyses there can be

done. The capacity depends on the number of analysts that are available and how they do their work

Workload The demand for analyses, besides the normal demand it also includes unforeseen events

Scheduling The scheduling method determines the work order, waiting times and due dates

Transport Transport of samples from the warehouse or production to the lab

Requirements Products can have requirements with respect to the quality standards and analyses which they have to meet. This can result in longer processing times, this includes GMP standards and prescribed work methods.

Table 6.1, explanation of factors influencing the throughput time

Although the research is now focusing on the quality control process, it does not mean that the other factors, as mentioned in the higher level conceptual are not relevant anymore. These factors still have a strong influence on the throughput time and the possible reduction and will be taken into account if that is required.

6.3 How is the quality of the products controlled?

(42)

The steps of the quality control process

The primary process of MB can be viewed as a supply chain. Of course, the company is only a part of multiple larger supply chains, but for the sake of this research the processes inside MB can be viewed as one relative short supply chain. When this view is extended, all operational departments are independent entities within this supply chain. The Quality Department is then a provider of a service to the planning department. The Planning Department needs the information that the arrived or produced batches are of the quality that is wanted. That information is provided by the quality department. The input that is needed to produce this information comes from the material handling and production departments.

The Quality Department is not a part of the primary process, because it does not process al the products that need inspection, only some samples from the batches. However, the Quality control process is very important for the primary process.

The following steps are performed in order to produce the information:

− Taking Sample. A sample is taken from the arrived raw material or produced batch. − Transport. The sample is transported through the plant and warehouse to the laboratory. − Arrival at the laboratory. Material handling and Production deliver samples at the

laboratory.

− Checking and distribution. The supervisor regularly on new arrivals of samples

− Booking and scheduling commons. The samples are administrated first and then placed in the laboratory schedule.

− Analyses. This is the core business of the laboratory. All the samples are analyzed according to the requirements. Based on the schedule an analyst picks a sample from the stock. The analyst then retrieves the analysis prescription and starts the analyses of the sample.

(43)

every product type, there are documents available at the laboratory, which holds the requirements. The analyst has to look them up when he starts with the analyses.

If a sample analyses is finished, there are basically three kinds of results.

- The first is that everything is alright. In that case, it is booked in the administration and the batch is made available in Prism.

- The second case is that the sample was not alright, but it is possible to rework the batch in order to get it alright. This rework results in a new sample, which has to be analyzed again.

- The last case is that the sample is not alright and that is not possible to rework the batch into something useful. In that case the batch has to be destroyed.

There are some exceptions on this process flow; these are the tank loads, samples that need special equipment and the WBP-laboratory. The main exception of the first two categories is that the samples are started almost immediately. The WBP laboratory is an exception, because it is located near the location where the products are produced which are analyzed at the WBP-laboratory. The details of these processes can be found in appendix III.

Process flow

(44)

Scheduling

At the moment, the scheduling of the Quality Department is based on the due dates. These due dates are based on the maximum period allowed for control; the maximum period can be as large as 8 days. These maximum periods are often reached. There are several reasons for that; first, the schedule is based on the due dates, which has as a result that simple analyses may be proponed until just before the due date. Second, several machines require full time attention of an employee, this consumes the capacity of the Quality Department needed for other analyses. Third, unexpected problems with batches of raw materials or problems with production, then several analyses have to be repeated a few times until the problems are solved. This can also consume a lot of capacity of the Quality Department.

(45)

7. Performance

In the previous section, the process flow of the quality control process is shown and analyzed. This is only one phase of value stream mapping of the current state. The second part is to assign relevant numbers to the process steps and inventories. This indicates how this process part is performing. This is what will be carried out in this section.

7.1 Source of the data

For the initial scan in a previous section, the data that was used came from the ERP system Prism. For a more detailed analysis of the quality control process, the Prism data is not enough.

It is recommended (Rother & Shook, 2003) that for a value stream mapping exercise all the numbers actually are measured by observation. However, in this case that is not a preferable way to determine the numbers. There is a great variation between the needed analyses for all the different products. Besides that, the needed activities can take up to several hours. In order to get a sufficient set of observations it would cost quite some time, keeping in mind the great variety of products. Therefore, it is chosen to adjust the original VSM approach into a more practical approach and use the numbers that are available in the administration of the quality department.

The Quality Department self also administrates data that can be used to determine the required performance indicators. Every sample that arrives at the Quality department is administrated at arrival and when it is finished it is also administrated. In this administration, several characteristics of the samples are recorded. Among them are:

- allowed throughput time, which determines the due date of a sample, thus how long the total throughput time may be.

- the measured throughput time, which shows how long it actually took to analyze the sample

- the number of lines, how many different analyses are needed for the sample

- the calculated analysis time, how many hours are theoretically needed for the analysis of the sample

(46)

These characteristics are not recorded for every type of sample that is processed by the Quality Department. This problem is rooted in the way the administration is done for these parts. It occurs for the categories ‘special samples’ and samples from the WBP-laboratory.

One problem that occurred when collecting the data, was that for some categories of samples, the ‘special’ samples and the WBP-laboratory.

Where it is possible, these samples are taken into account, otherwise the data is adjusted to show a correct number. In appendix IV a detailed analysis of the Quality Control process can be found. For each calculation it will be made clear, which data it embraces.

However, this problem does not influence the big picture of the Quality Control process, for that the available data is sufficient. The available data represent the standard working procedure for the majority of the samples.

As quality is an important factor at MB, as was argued in section 6.2, registration of the samples is an important feature in order to find mistakes quickly. This fact should ensure that the registered data is valid enough to use in analyses like in this research.

Besides the administration of the samples, there is also a registration of the spend hours at the quality department. The ‘hour registration’ administrates the worked hours and the distribution of these hours to the performed activities. The activities are divided in:

- Analyses, which is divided further in: o _Analyses

o _{Machine maintenance}

o _{Special analyses activities (HPLC, ICP, particles and ultra-resi)} o _{WBP activities (Dilut-it, hematological, normal solutions)} - Administration, includes management tasks

(47)

This data is available for a large number of years, however, due to changes in the structure, the data of the earlier years is not relevant anymore for the current situation. It is chosen to use the data from a period that matches the period used for the initial scan. So that is from the first of January 2008 until the fifth of March 2009(which was the date of request).

7.2 Results detailed analysis Quality Control

In addition to the obtained data, as described in the previous section, several analyzing sessions and semi-structured interviews with employees were conducted in order to understand the data. Based on the data a series of analyses can be made. These analyses can be found in the appendix III, the most important results are presented in this section.

In table 7.1 the performance numbers of the quality control process can be found. The columns in the table reflect the steps of the quality control process as are shown in figure 6.2 (page 40). For the inventory points the average waiting time is calculated, for the steps that require processing, the processing time is calculated. The time is represented in hours. It is important to mention that this figures are based on the data that is recorded by the Quality Department. Thus there is a difference with the tables in chapter 5, which were based on the Prism data.

Process steps of Quality Control Transport Incoming Stock Admini- stration To Do list Analyses Final stock End control Total Processing Time (hours) 0,25 0,1 2,7 0,2 3,25 25,25 Waiting Time (Hours) 3,5 2 14,5 2 22

Table 7.1, VSM table: average throughput times

Also the Work-in-Progress (WIP) is calculated, via Little’s Law (1960). In table 7.2 it is shown how much WIP there is on average for each process step.

(48)

Tables 7.1 and 7.2 embrace the Analytical Laboratory, including the ‘special’ samples. As can be seen in the table, the majority of the average time that a sample spends in the quality control process, is spend waiting. The largest part is the waiting time before the analysis starts.

The averages are not the only numbers that are important. It is also useful to calculate the variability. Variability influences the throughput and the WIP (Hopp and Spearman, 2002). The variability is measured here by the dimensionless coefficient of variation (CV), which is obtained by dividing the standard deviation by the mean.

Coefficient of Variation

Processing time 1,04

Arrival rate 0,56

Table 7.3 Variability

The figures in table 7.3 indicates that the arrival of samples is a process which has low variability, while the processing time has moderate variability. The VUT equation (Hopp and Spearman, 2002) can be used to predict the cycle time of the queue (CTq). This equation holds a variability term (V), a utilization term (U) and a time term (T).

The Variability term uses the CV’s given in table 7.3. The utilization level was difficult to determine. There are eight analysts working at the laboratory, who all can perform these analysis. Based on the assumption that all these analysts are available for these analyses, the determined utilization level is 48%. However, on a day-to-day basis, there are about four analysts working these standard analyses. The others are occupied with other tasks. In that case the utilization level is 96%. But it is not a rigid system, on a busy day more analysts can work on the standard analyses and vice versa.

The time term is the mean effective processing time, which is the processing time mentioned in table 7.2: 2,7 hours.

(49)

Cycle time queue (hours)

Utilization = 48% _0,04

Utilization = 96% _9,8

Table 7.4 Expected queue time

From table 7.4, it can be derived that the waiting time at the laboratory is performing worse than it might be expected based on the variability and the VUT equation. That indicates that it might be possible to improve the waiting time.

7.3 Areas of attention

In this section conclusions about the quality control process are made. These conclusions are based on the results presented in the previous section and the earlier presented description of the process flow. These conclusions are presented as areas of attention; in the terminology of VSM these are called Kaizen points.

A potential area of attention is a specific area which is not performing very well and which can be improved. Hereby it is important that one keeps in mind that the one of the basic thoughts of VSM method is to determine what steps does not add value to the product. The product in this case is the information provided to the other departments.

(50)

- Transport. The samples of the Production and Material Handling department are transported directly to the laboratory during the period there are analysts working at the laboratory. If the laboratory is closed, the samples are collected and transported the next morning. This results in a peak in the arrival in the morning and a unpredictable arrival pattern during the day. That causes waiting time before the administration.

- Administration. The handling of the incoming samples is another possible area of attention. Although the administrative action does not take long, there is on average two hours of waiting time. The administration does not add direct value to the information that QC provides, however it is a (legal) requirement.

While the incoming samples are waiting to be administrated, the analyses cannot start yet.

The same can be observed with respect to the end-control. There is also waiting time and again this step does not add value to the information.

− Waiting list. The waiting list before the analysis is the greatest part of the throughput time of the quality control process. The average input and output are equal to each other, so this waiting list seems unnecessary. The forming of this queue is caused by variation. If there is a difference between the input rate and the processing rate and the input rate will thereby vary significantly, then a queue will be formed (Bertrand et al. 1998). This is the case at the laboratory. The input and the analysis rate both vary. At the days that the input rate is higher than the analysis rate the queue will grow. The input is determined by the Planning and Purchasing department, while the output is determined by the scheduling at the laboratory.

o_{Scheduling. In order to match the capacity of the laboratory with the input} scheduling can be used. However, in the current situation, it is not used in such a way. The schedule is just a list of samples that have to be done before a certain date. There are three things that influence the scheduling.

(51)

structurally use the information about what is in production and what is being delivered. Only with the tankloads the knowledge of when it is arriving is used in a structural way.

Priority. Not all products are equally important; some products are going to stock, while others are directly shipped after approval of QC. At the current situation, QC does not know which product deserves priority. Two things are important for this:

• Allowed throughput times. The allowed throughput times (ATT) define how long a product may stay at the laboratory. Based upon these times, the scheduling method defines when the sample has to be finished. However, the current ATT’s does not represent the priority of the products. So it is possible that a certain product that has to be shipped as fast as possible has an ATT of 8 days and thus stays for 8 days at the laboratory. • ABC-MTO/S classifications. The products have two classifications

which determine the ‘importance’. These are the ABC-classification and the MTO/S ABC-classification. These ABC-classifications are not used by the laboratory.

o_{Planning/purchasing. This department determines the input for the laboratory.} Either by planning the production or by ordering new batches of raw materials. This is done based on the demand of the customers, either directly (MTO) or via the stock (MTS). Obviously the demand from the customers is not influenced by the planning or purchasing department.

Batch size. The batch or order size is determined by the Economic Order Quantity (EOQ) model. This model aims to optimize the order quantity, and thus determines the batch size. That is in fact a trade-off between the stock costs and the order costs, with a known demand (Bertrand et al. 1998). The order costs are the costs associated with ordering a batch, these costs are independent of the quantity of a batch.

(52)

but is certainly not correct for the Quality Control costs. It can be seen in appendix III that the analyzing times are differ strongly among the different products. So the costs for Quality Control will also differ. This results in batch sizes that are not optimal for the laboratory. Products with long analysis times can be ordered in small quantities and so put significant workload on the analysts.

- Product usage (production/shipping). When QC approves a sample, the Prism status is changed in AVLB. It can then be used for further production or shipping. However, in the current situation there is a gap between the moment that the sample is approved and the moment that the related batch is used. Planning does not know when a sample is going to be approved. In Prism, it can be seen what the date is that a sample has to be inspected, according to the Allowed Throughput Time, but samples are often finished before that date and thus they can be used or shipped earlier then the Planning department expects.

In the last two chapters, it was tried to find an answer on the research question: what are the

current throughput times of this specific part of the process? The specific process part was the quality control department.

It was found that the throughput time is more than 25 hours. Of these 25 hours, more than 20 hours is spend waiting. Several causes can be appointed for this waiting time. One of them is that there is no balance between the input and output of the department.

(53)

8. Improvements

In this section the fifth research question; How can the throughput times be reduced?, is answered. The reason to initiate this research was to increase the responsiveness and to decrease the stock size. In this section several solutions will be proposed, which can help to reach that goals. These solutions are not only evaluated with respect to the throughput time decrease, but also with respect to the higher goals of responsiveness increase and stock size decrease.

First it is determined on what measurements the solutions will be evaluated with respect to the stock size and the responsiveness. Subsequently, each solution is elaborated and evaluated.

8.1 Effects

The effects of the solutions on the stock size and the responsiveness need to be measured or captured in some way. Before the solutions are presented, it is first determined how that can be done in this part.

Stock size

In section 4.2, it was discussed that the gain of the throughput time reduction with respect to its effects on the stock size is with a reduction of the stock costs.

As was established in section 2.1, the costs associated with holding stock are 25 percent of the stock value per year. Via this percentage, the costs can be determined. So, if the effects on the stock value are clear, then it is straightforward to determine the effect on the stock costs.

Master Thesis Technology Management Throughput times at Mallinckrodt Baker