Developing a new layout for Neopost Technologies B.V.

(1)

Developing a new layout for Neopost Technologies B.V.

by

Jesse Kamps

University of Groningen Faculty of Economics and Business

(2)

2

ABSTRACT

(3)

3

PREFACE

This thesis forms the final culmination of the Master’s degree program of Technology Management at the University of Groningen. But before you start reading this thesis I would like to take this opportunity to thank the people who made the thesis, as presented here, possible.

First off, I would like to personally thank Ben Scholtanus and Eddy de Boer for making this research project at Neopost Technologies B.V. possible and for providing valuable feedback and new insights through the numerous conversations we have had. Furthermore, I would like to thank the rest of the staff of Operations Engineering and Assembly for their aid and making my stay at Neopost the pleasant experience that it has been. I would also like to thank my university supervisors, dr. ir. W. Klingenberg and prof. dr. ir. J. Slomp, for their feedback and for providing new insights.

(4)

4

GLOSSARY

Abbreviation Meaning A - Average to assemble ATO - Assemble-To-Order C-card - Conveyance-card CIR - Circle CM - Cellular Manufacturing

CONWIP - CONstant Work In Progress

CYL - Cylinder

DDC - Diagnose Design Change

DS - Document System

E - Easy to assemble

FCM - Focused Cellular Manufacturing

FTE - Full Time Employee

H - Difficult to assemble

IM - Incoming Mail

LeLo - Letter Opener/Letter Extractor, Also known as IM-35

LRC - Large Rectangular Cuboid

MRP - Material Requirement Planning

MTS - Make-To-Stock

NTBV - Neopost Technologies B.V.

OTH - Other

P-card - Production-card

PCA - Part Coding and Analysis

PP - Parts Production

RT - Rectangle

SRC - Small Rectangular Cuboid

(5)

5

LIST OF FIGURES

Figure 1 The DS-80 folder and inserter ... 12

Figure 2 The IM-35 letter extractor ... 12

Figure 3 Production Process ... 13

Figure 4 Diagnose, Design and Change model of de Leeuw (2005) ... 14

Figure 5 Black box of an assembly cell ... 15

Figure 6 In the black box of an assembly cell... 15

Figure 7 Subassembly section of an Assembly cell ... 16

Figure 8 Assembly line section of an Assembly cell ... 16

Figure 9 Overview of the first assembly hall with material flow ... 17

Fgure 10 DDC mode of de Leeuw (2005) ... 18

Figure 11 Conceptual model ... 19

Figure 12 Relationships between proposed cells and lines ... 34

Figure 13 Relationships in alternative configuration ... 34

Figure 14 Material flow in layouts ... 45

Figure 15 Appropriate layout for combinations of volume and variety Source: Askin and Standridge 1993 ... 46

(8)

8

LIST OF TABLES

Table 1 Advantages and disadvantages of assembly systems ... 21

Table 2 Parts, weekly demand and weekly required assembly hours of the product families ... 22

Table 3 Weekly required assembly hours per station for product family ... 23

Table 4 Six different options to be explored ... 23

Table 5 Utilisation rates and station requirements of the six options ... 25

Table 6 The proposed line configuration ... 25

Table 7 Configuration of Line 3 for normal, increased and decreased demand ... 27

Table 8 Lines 1, 2 and 3 with their number of stations, utilisation rates and product families ... 27

Table 9 Subassembly category matrix ... 29

Table 10 Required weekly hours to produce subassembly families ... 31

Table 11 Required times to produce subassembly families in FTE/week ... 31

Table 12 Cellular configuration based on difficulty of assembly ... 32

Table 13 Cellular configuration based on shape ... 32

Table 14 Most suitable cellular configuration based on difficulty ... 32

Table 15 Most suitable cellular configuration based on shape ... 32

Table 16 Final cellular configuration of subassembly ... 33

Table 17 Utilisation rates of the proposed cellular configuration with an increased and decreased demand for 2013 ... 33

Table 18 Proposed number of stations when demand increases/decreases with 30% for 2013 ... 33

Table 19 Data on cells in the alternative configuration ... 35

Table 20 Advantages and disadvantages of both alternatives ... 35

Table 21 The defined lines and cells with their number of stations, utilisation rates and product families ... 36

Table 22 Number of items per bin per product family ... 38

Table 23 General characteristics of layout types Source: Askin and Standridge 1993 ... 46

Table 24 Objectives and recommendations Source: Canel, Al-Mubarak and Khumawala 2005... 47

Table 25 Differences FCM, CM and Fractal ... 48

Table 26 Input data for Option 1, the DS 85/90 FV and DS 100/140 on eight and nine stations ... 51

Table 27 Line configuration based on the DS 85/90 FV and DS 100/140 on eight and nine stations ... 51

Table 28 Input data for Option 2, the DS 85/90 FV and DS 100/140 on four and nine stations ... 51

Table 29 Line configuration 1 based on the DS 85/90 FV and DS 100/140 on four and nine stations ... 52

Table 32 Input data for Option 3, the DS 85/90 FV and DS 100/140 on eight and six stations ... 52

Table 33 Line configuration based on the DS 85/90 FV and DS 100/140 on eight and six stations ... 53

Table 34 Input data for Option 4, the DS 85/90 FV and DS 100/140 on eight and three stations ... 53

Table 35 Line configuration based on the DS 85/90 FV and DS 100/140 on eight and three stations ... 53

Table 36 Input data for Option 5, the DS 85/90 FV and DS 100/140 on four and six stations ... 54

(9)

9

Table 38 Line configuration 2 based on the DS 85/90 FV and DS 100/140 on four and six stations... 54

Table 39 Line configuration 3 based on the DS 85/90 FV and DS 100/140 on four and six stations... 54

Table 40 Input data for Option 6, the DS 85/90 FV and DS 100/140 on four and three stations ... 55

Table 41 Line configuration 1 based on the DS 85/90 FV and DS 100/140 on four and three stations ... 55

Table 44 Input data when for an increase in demand of 30% ... 57

Table 45 Line 1 configuration based on proposed configuration with an increase in demand of 30% ... 57

Table 46 Line configuration 2 based on proposed configuration with an increase in demand of 30% ... 57

Table 47 Alternative to Line 3 in configuration 2... 58

Table 48 Input data when for an increase in demand of 30% and the DS 85/90 FV and DS 100/140 are assembled on eight and six stations ... 58

Table 49 Line configuration based assembling the DS 85/90 FV and DS 100/140 on eight and six stations with an increase in demand of 30% ... 58

Table 50 Alternative to Line 3 ... 59

Table 51 Input data when for an increase in demand of 30% and the DS 85/90 FV and DS 100/140 are assembled on eight and nine stations ... 59

Table 52 Line configuration based assembling the DS 85/90 FV and DS 100/140 on eight and nine stations with an increase in demand of 30% ... 59

Table 53 Input data for a decrease in demand of 30% and the DS 85/90 FV and DS 100/140 are assembled on four and six stations... 60

Table 54 Line configuration based assembling the DS 85/90 FV and DS 100/140 on four and six stations with a decrease in demand of 30% ... 60

Table 55 Line configuration based assembling the DS 85/90 FV and DS 100/140 on four and six stations with a decrease in demand of 30% where the number of stations on which the DS 100/140 is assembled is reduced to three ... 60

Table 56 Input data for a decrease in demand of 30% where the DS 85/90 FV and DS 100/140 are assembled on four and three stations ... 61

Table 57 Line configuration based assembling the DS 85/90 FV and DS 100/140 on four and three stations with a decrease in demand of 30% ... 61

Table 58 Line configuration based assembling the DS 85/90 FV and DS 100/140 on four and three stations with a decrease in demand of 30% ... 61

Table 59 Cellular configuration based on difficulty ... 62

(10)

(11)

11

1. INTRODUCTION

This section serves as a general introduction to Neopost technologies B.V. (NTBV). First the Neopost Group, of which NTBV is a part, will be introduced followed by a general introduction to NTBV, the products that they produce and their production process.

1.1. The Neopost Group

The Neopost Group is a supplier of mailroom equipment, offering solutions for franking, folding and inserting, opening/extracting, addressing, address cleansing and tracking and tracing, of items and of the supply chain. Next to this the Neopost Group also offers a range of services, including consulting, maintenance and financial services.

The Neopost Group was formed in 1992 out of Alcatel’s mailroom solutions division. The mailroom solutions division had been set up in 1970 following the acquisition of two companies that provided mailroom solutions to the French market. It has since then been built up through the acquisition of multiple other companies. In 1997 Neopost was taken over by another group of investors, which led to Neopost listing on the main Paris Euronext market on 23 February 1999. By 2002 Neopost was the world’s number two supplier of mail processing solutions and further acquisitions since then have allowed them to maintain and strengthen this position (Neopost 2011). The Neopost Group currently employs over 5500 people, of which around 320 are working in R&D, spread over seven main locations. The manufacturing facilities are located in Drachten (The Netherlands), Lelude (France), Loughton (United Kingdom) and Asia. Neopost has a strong market position with over 800.000 customers in 90 countries. For mailing systems Neopost currently ranks first in Europe and second globally. Neopost is the global leader in document systems.

1.2. NTBV

The history of NTBV dates back to 1924 when it was founded as HaDeWe, a small factory that made wooden shoes. The company grew significant when the stencil machine was invented and started exporting to the United States from 1934. Since then both ownership and name have changed several times until it changed to its current name, Neopost Technologies B.V., in 2005.

(12)

12 1.3. Products

The DS-machines are able to fold and fill envelopes whereas the IM-machines open envelopes and optionally remove the contents. One of each type of system will be described next in order to get some familiarity with the type of products that NTBV produces.

1.3.1. Document Systems

These systems are capable of folding documents, inserting them in envelopes and closing the envelopes. The product chosen to illustrate this type of mail solution is the DS-80, see Figure 1. The DS-80 is a system that is composed of two separate modules, with a folder on the right side and an inserter on the left. This type of machine is a vertical system, named this way because the paper is stored vertically, which in turn requires less space. NTBV also has horizontal systems in their product portfolio, these systems require more space due to the way paper is stored but are able to process documents at a higher speed. Document Systems compromise the largest part NTBV’s assembly output.

1.3.2. Incoming Mail Systems

NTBV has three machines in its portfolio for handling incoming mail, of which the IM-35 extractor, which is also known as LeLo, is shown in Figure 2. This machine opens an envelope by cutting it open on three sides and then removes the envelope before presenting the letter to the operator. The other two IM-machines cut the envelope open but leave extracting the letter to the operator. These types of machines compromise a small part of NTBV’s assembly output.

1.4. Production process

The production process of NTBV is separated into two departments, Parts Production (PP) and Assembly. PP produces in a make-to-stock (MTS) fashion, whereas Assembly produces to order, in an

(13)

13 assemble-to-order (ATO) fashion. Assembly produces low volumes that contain high variability. Because PP produces to stock, volumes are higher and variability is smaller.

Within PP raw materials are transformed into parts, which range from small brackets and axles to frames and powder coated hoods. These parts are then used by Assembly, together with buy-parts, to produce the final products. A final product will then move from Assembly to Expedition from where it will be shipped to the customer. See Figure 3 for a graphical representation.

Parts Production Part Supplier

Assembly

Expedition Raw Materials

Supplier

Warehouse Warehouse

Figure 3 Production Process

(14)

14

2. RESEARCH DESIGN

Research design is defined as the plan and structure of investigation to obtain answers to the research question (Cooper and Schindler 2008). To give structure to this research the Diagnose, Design and Change (DDC) methodology (De Leeuw 2005) will be used. See Figure 4 for a graphical representation. The first step, Diagnose, uses the management problem as input and uses conceptualisation and description to come to the problem statement. Description is defined by de Leeuw (2005) as problem-oriented mapping of the relevant system. In the second step, Design, a solution to the problem statement will be sought of, by first specifying the direction in which the solution should be sought after and then designing a solution. The final step, Change, consists of realising the change and evaluating it. While this may seem as a straightforward sequential process, iterations do occur between the phases.

The two main sources of data are NTBV’s IT system and scientific literature. Other lesser used sources are observation and participation. In the remainder of this section the problem statement will be given, followed by a description of the assembly system. Then upcoming changes and desires will be mentioned. These steps then lead to the formation of the research question and a conceptual model will be drawn up.

2.1. Problem statement

The product portfolio within the assembly department of NTBV will undergo changes in 2012. Some systems will be phased out, new systems will be launched and NTBV is also trying to get new business to Drachten. Aside from these changes NTBV is also going to start with remanufacturing activities. These upcoming changes have led to the following management question:

“How can we assemble the future product portfolio in the most optimal way?”

Wherein optimal means at minimal integrated costs, high quality and delivery within two weeks. The current system in which the product portfolio of NTBV is produced will be described next.

2.2. Description of the system

The system that is to be investigated is the assembly department of NTBV. The description of the system will start from the highest aggregation level and will gradually zoom in. In this way first major

Diagnose Design Change Management problem Research Problem Solution Improved situation

(15)

15 relationships and subsystems will be addressed. This will be followed by a general overview of the system.

Currently assembly takes place in two factory halls, in two separate buildings. When looking at the assembly department one can see that it is divided into several cells. Each of these cells is responsible for the entire assembly process of a family of end-items, or product family, and can thus be called a focussed cell (Al-Mubarak, Khumawala, and Canel 2003). These cells use parts as their input and transform them into finished machines. A kanban system is used to control the flow of parts that are transported to the assembly cells from the warehouse and finished machines are transported to expedition on a conveyor belt. The number of parts that go in a bin is determined by part size, i.e. as many as fit go in a bin, not requirement. Figure 5 shows the black box of an assembly cell and the assembly department, the solid line represents material flow and the dotted line information flow.

Assembly Cell/ Department

Parts Finished Machines

Figure 5 Black box of an assembly cell

Sub-assembly Parts

Assembly line Sub-assemblies

Figure 6 In the black box of an assembly cell

When we now look inside the black box of an assembly cell we see that a cell is divided into two sections, a subassembly section and an assembly line. See Figure 6. Within subassembly parts are assembled into subassemblies that range from small axles with pulleys to large paper feeders. These subassemblies are transported to the assembly line where they are placed in the machines. Again, a kanban system is used to control the flow of subassemblies. Within subassembly tasks are, generally, smaller and more repetitive then in the assembly line. This is why the workers with more experience and skill are usually working in the assembly line section of the cell.

(16)

16

Storage locations Sub-assembly tables

Parts Parts Sub-assemblies

Figure 7 Subassembly section of an Assembly cell

Storage locations Assembly tables

Machines Parts/

Sub-assemblies Parts/

Sub-assemblies

Figure 8 Assembly line section of an Assembly cell

Now we will zoom in on the assembly line section of the cell, see Figure 8. The assembly line section consists of assembly tables and storage locations. The information input of the assembly line consists of an order sheet, which lists the details of the order. The assembly line worker can see from the order sheet what has to be added to the machine. Parts and subassemblies are taken from the storage locations and when a new bin is accessed a replenishment order is sent to either the warehouse or subassembly.

The assembly cells have been set up in the past to produce a certain number of machines and have, since then, not been adjusted. The assembly cells have not been adjusted because of several reasons. First, there has been no real necessity to make changes, as demand for products kept rising and floor space was readily available. In recent years demand for products has, however, declined. Another reason that the assembly cells have not been adjusted is that the way in which the cells are currently set up does not allow for easy changes. This is due to the fact that large fixed storage locations are used. The result from this is that all of the available floor space is in use and most stations have low utilisation rates, which has workers moving regularly between stations.

2.2.1. General overview

The way in which products are assembled has already been described in section 2.2. This section will give insight into the current layout and material flows.

Figure 9 shows an overview of the first assembly hall and the flow of material and finished goods. The assembly hall measures 45m by 77m, for a total surface of 3465m². Powder coating and Stippling, which are also present in the assembly hall, are part of PP. The second assembly hall, which is not depicted here, houses production of the IM-75, SI-92 and IM-35. The blue rectangles that have red lines going in and yellow lines going out are assembly cells and their names show which product is produced in said cell.

(17)

17 with the largest material requirements and production has been placed the greatest distance of the warehouse and expedition.

Figure 9 Overview of the first assembly hall with material flow

2.3. Upcoming changes and desires

It has been stated before that NTBV is undergoing some changes in the near future. These changes are that the IN-2D, IN-1C and FV-2A are going to be replaced by the DS-90 IN, DS-85 IN and DS 85/90 FV, respectively. The SI-92 is going to be discontinued. NTBV is going to start remanufacturing old DS-62 machines into DS-63 machines. And NTBV is also trying to get new business for their assembly department from other manufacturing locations within the Neopost Group.

All these changes give NTBV a chance and a reason to rethink their assembly process and make changes to it, if necessary. Management has also expressed some desirables for a new assembly process:

- The assembly process should no longer be split over two assembly halls

- Workers should have to move less between workstations

2.4. Research question

(18)

18 - Can a layout be developed that allows the assembly department of NTBV to be more flexible

while requiring less space?

With being flexible it is meant that the assembly department must be able to cope with changes in the product portfolio as described in the problem statement and to future changes to the product portfolio that are currently unknown and that adapting to these changes will be relatively easy, i.e. have high product flexibility (Slack, Chambers, and Johnston 2010).

2.4.1. Sub questions

The research question has been divided into the following sub questions:

1. What is the current situation and what are the upcoming changes and desires that have to be taken into account?

2. What type of layout does the literature prescribe and what tools and techniques can be used to design a layout?

3. What building blocks would such a layout consist of? 4. What production control system can be used?

Fgure 10 provides an overview of the model of de Leeuw (2005) and shows in which phase the sub questions fit and in which chapters they will be covered. Sub question 1 has already been covered in section 2.2 and 2.3. A literature review, located in Appendix A – Literature review, covers the second sub question. The third and fourth sub questions will be covered in sections three and four. The change phase of the methodology of de Leeuw (2005) will

not be covered as the

implementation is beyond the scope of this thesis. Diagnose Design Change Management problem Research Problem Solution Improved situation Sub question 1 Sub question 2 Chapter 2.2 and 2.3 Appendix A Sub question 3 Sub question 4 Chapter 4 Chapter 5

(19)

19

2.4.2. Conceptual model

Literature Company data

Can be adapted Analysis Suitable No Yes Yes Tools and Techniques Useful company data Toolbox Discard New Layout

Figure 11 Conceptual model

(20)

20

3. CREATING CELLS

Based on the literature review, a cellular layout is seen as the most fitting layout type for NTBV. First, assembly lines are identified as the most suitable way to perform final assembly. The next step entails calculating the amount of time that is required to perform the assembly operations. After the assembly times have been calculated the product families are assigned to assembly lines, using an algorithm to minimize the total number of stations. Then the subassemblies are grouped together using a modified form of cluster analysis, based on geometrical and technical attributes. The created part families were then grouped together to into subassembly cells. An alternative way to group the subassemblies that is based on the concept that is currently used is also explored. Finally, the spatial requirement of the new developed configuration is examined.

3.1. Final assembly

Final assembly can be performed in several ways. Ways that have been identified are bench assembly, moving bench assembly, assembly lines and assembly cells. Bench assembly is assembly where the product does not move. Instead all the parts and equipment are brought to it and an assembler moves around it (Baudin 2002). A variation on bench assembly is bench assembly where the bench is not in a fixed position, but moved past the parts by an assembler. An assembly line has the productive units performing the operations aligned in a serial manner. The work pieces visit stations successively as they are moved along the line (Boysen, Fliedner, and Scholl 2007). The assembly cell is a special case of an assembly line, wherein the stations are arranged in a u-shape with the operators working on the inside and parts being fed from the outside (Baudin 2002).

(21)

21

Assembly system Advantages Disadvantages

Bench  High job satisfaction  Picking is error-prone

 High floor space requirement

 Employees should be highly cross trained

Line  High quality

 Lower need for employees to be cross trained

 Low floor space requirement

 Parts and components are not delivered to all stations

 Lower job satisfaction Moving bench  High job satisfaction

 Less picking errors than bench assembly

 High floor space requirement

 Employees should be highly cross trained Cell  No interference between part supply and production

 Staffing flexibility

 Easy communication among assemblers for training and support

 Parts and components are not delivered to all stations

 Low floor space requirement

 Highest quality

 Only suitable for small products

 Can be operated by no more than eight employees

 Employees need to be cross trained

Table 1 Advantages and disadvantages of assembly systems

Based on the advantages and disadvantages listed in Table 1, the assembly line is chosen as the most suitable option. While assembly cells are more desirable than assembly lines (Baudin 2002) they are only suitable for the assembly of small products, which leaves the assembly line as the best option for NTBV. There are three general types of assembly lines, single product, mixed product and batch product (Rekiek and Delchambre 2006). The single product line is used to produce only one product, the mixed product line is used to produce a family of products whose main functions are similar and the batch production, or multi-model, line is used in the case of multiple different products or families (Rekiek and Delchambre 2006). Each end-item of NTBV can be delivered with or without a multitude of options, the single product line is, therefore, not possible. The choice for either a mixed, multi-product line or both depends on the time that is required to assemble a family of end-items, i.e. the demand should be large enough to warrant an own line. Therefore, the time that is required for assembly will be calculated after which a choice for a line type will be made.

3.1.1. Calculating final assembly times

(22)

22 Table 2 shows the number of parts that go in a machine and the weekly demand for machines of each product family. Using equation (1) the required assembly time in hours per week has been calculated for each product family and these times have also been included in Table 2. The weekly demand is calculated by dividing the expected yearly sales volumes by the number of weeks that are available for production. There was no information available with regards to the expected sales volumes of the machines that are currently in the Exotics section. Because these are expected to be low, i.e. less than one per week, it has therefore been decided to keep the Exotics section in place. The IM-75 product family is not present in Table 2 because this machine has been added to the Exotics cell. This has been done because of two reasons. First the weekly demand for IM-75 machines was less than one per week. And second the IM-75 is not suitable to be assembled on the same type of line as the other product groups because of its size and the way in it is assembled, e.g. workers need to get under the machine.

Table 2 Parts, weekly demand and weekly required assembly hours of the product families

From the numbers in Table 2 it becomes immediately apparent that the 63/75, 85/90 FV and DS-100/140 warrant enough work to fill a mixed model line. If the latter two, however, were to be assembled on a mixed model line, the line would only consist of two stations. The fixed lines would reduce the flexibility of assembly. And a practical drawback is that it is near impossible to create enough storage space to house the parts that are required for assembly at the stations. Another disadvantage is that, because of the high workload per station, training would take longer. If a mixed model line were to be setup for the DS-63/75 machines it would not run into these drawbacks as the tasks would still be spread over many stations. Also, if the DS-63/75 were to be included into a multi-model line, a lot of effort would have to be put in line balancing, or many stations would go unused. Therefore, the DS-63/75 will be assigned to a mixed model line and the remaining product families will be assigned to multi-model lines.

Before possible options were examined, the assembly times were recalculated to hours per station. This has been done because of two reasons. First it allows the feasibility of the solution to be seen directly; and secondly assembly times are not divided over all stations in a possible line, e.g. if the versa machine were to be produced in a 4 station line its assembly time would still be based on three stations, leaving

Product families DS85IN DS90IN DS8590FV DS100140 LeLo Versa DS63/75

Total number of parts 833 749 666 2437 1101 641 1182

Weekly demand of

machines 9,1 7,5 16,6 5 6 3,7 70,8

Required assembly time

in hours per week 42,96 31,83 62,65 69,05 37,43 13,44 474,22

Where: the number of parts; the weekly demand;

centiminutes to assemble one part, 34; centiminutes per hour, 6000;

(23)

23 the fourth unused and thus not distorting the total line time. Table 3 shows the weekly required hours per station for each of the product families. When assigning product families to a multi-model line the current number of stations has been taken into account, to keep activities relating to line balancing and training staff to a minimum. Keeping these factors to a minimum also allows for faster implementation.

Table 3 Weekly required assembly hours per station for product family

3.1.1. Assigning product families to assembly lines

NTBV has a workweek of 40 hours but the assembly staff works around 7.2 hours per day in practice, because of coffee breaks and such. This makes for an effective workweek of 36 hours. It is aimed, however, to get an utilisation rate of 85%. An utilisation rate of 85% has been chosen for the stations to cover the errors that might lie in the assumptions that have been made and to make sure that some flexibility is retained. This makes the desired total station time 30.6 hours. It can be seen from Table 3 that the total weekly required time per station is 46.45 hours. This means that with a desired 30.6 hours per station two assembly lines are required. If these hours are spread evenly over two lines they each would have an utilisation rate of 65%. The only variable that can be changed to increase this is the number of stations over which the assembly of a product family is spread. To increase the total time, the number of stations over which a product family is spread should be decreased. The two product families that are the most obvious candidates for this are the DS 85/90 FV and the DS 100/140 because their assembly is spread over twice as many stations as that of the other product families. The number of stations over which the DS 85/90 FV is spread can easily be brought down from eight to four by combining current stations. And the number of stations over which the DS 100/140 is spread can be changed to three since the DS 100/140 consists of three modules that are each assembled over three stations. Changing the number of stations on which the DS 85/90 FV and DS 100/140 can be done in several manners. These are changing the assembly of the DS 85/90 FV, changing the assembly of the DS 100/140, changing both or leaving them as is. Table 4 shows the different options that will be explored.

Number of stations

Product families Option 1 Option 2 Option 3 Option 4 Option 5 Option 6

DS8590FV 8 4 8 8 4 4

DS100140 9 9 6 3 6 3

Table 4 Six different options to be explored

When assigning a product family to an assembly line, the goal is to minimise the total number of required stations while aiming for an utilisation rate of 85%. To accomplish this, an algorithm, which is

Product families DS85IN DS90IN DS8590FV DS100140 LeLo Versa Total

Number of stations 4 5 8 9 4 3 23

Required assembly time in

hours per week 42,96 31,83 62,65 69,05 37,43 13,44 257,36

Required assembly time per

(24)

24 provided below, has been developed. The algorithm selects the product family with the lowest station requirements first and adds consecutive product families until an utilisation rate of around 85% is attained after which it moves on to another line until all product families are assigned. Parameters within this algorithm are the number of stations, S, the assembly time per station, T and the total number of stations of an assembly line, Sline which is MAX (S1, S2 … Sn).

Step 1: Select product family with lowest S.

Step 2: Select next product family with lowest S, add time to T and include S in Sline. If there are

multiple product families with the next lowest S, split in multiple solutions. If the added product family was the last, go to Step 5.

Step 3: If T < 30 go to Step 2. If T > 34 go to Step 4. Else, go to Step 5.

Step 4: Remove last added product family from T and Sline. And add the data of the next lowest S

to Sline and T. Go to Step 3. When all options have been explored, go to Step 5.

Step 5: Write down the product families, T and Sline. If not all product families have been

assigned to a line, go to Step 1. Otherwise, go the next step.

Step 6: Calculate ∑ line for each of the developed solutions and write them down.

Using the proposed algorithm the six different options that are provided in Table 4 will be explored. Data that was used as input for the algorithm and the results it produced are provided in Appendix B for each of the explored options.

First, not making changes to the number of stations of the DS 85/90 FV and DS 100/140 is explored. Using the algorithm with the input of option 1 created one line configuration. This configuration has two assembly lines with utilisation rates of 85.97% and 43.06% and requires a total of fourteen assembly stations.

The second option that is explored is bringing the number of assembly stations for the DS 85/90 FV down to four while keeping the number of assembly stations for the DS 100/140 at four. Using the algorithm, three different line configurations were created. The configuration that was ranked as the best from these requires thirteen stations and has utilisation rates of 85.78% and 65.00%.

(25)

25 The fourth option has the DS 85/90 FV assembly spread over eight stations while the DS 100/140 assembly is brought down to three stations. The algorithm produces one line configuration that requires a total of thirteen stations and has two lines with utilisation rates of 94.08% and 77.58%.

The fifth option changes the number of stations of the DS 85/90 FV as well as the number of stations of the DS 100/140. The number of stations over which assembly of the former is spread is changed to four, whereas assembly of the latter is spread over six stations. Using the algorithm three configurations have been created. Two of the configurations require ten workstations. One is, however, preferred over the other as it has slightly better utilisation rates, which are 85.78% and 75.67%.

The sixth and final option that is explored is assembling the DS 85/90 FV on four stations and the DS 100/140 on three. Using the algorithm leads to three different configurations. The three developed configurations all require three assembly lines and thirteen assembly stations. The most favourable configuration is chosen based on the utilisation rates, which is the configuration with utilisation rates of 94.08%, 73.33% and 26.00%.

Option 1 Option 2 Option 3 Option 4 Option 5 Option 6

Utilisation rate of Line 1 85,97% 85,78% 85,97% 94,08% 85,78% 94,08%

Utilisation rate of Line 2 43,06% 65,00% 53,72% 77,58% 75,67% 73,33%

Utilisation rate of Line 3 - - - 26,00%

Number of stations 14 13 13 13 10 13

Table 5 Utilisation rates and station requirements of the six options

Utilisation rates and the number of required stations of the best configurations of each of the options are listed in Table 5. More detailed information is available in Appendix B. Based on these figures the line configuration of Option 5 is chosen as the final configuration, see Table 6. It has been assumed that each product family can make due with one test station, despite the fact that some lines have multiple test stations in the current situation. This assumption has been made because the lines were initially designed to produce a higher volume than currently is required.

Line 1 Line 2

Product families

Hr/week

per station Stations

Utilisation rate

Product families

Hr/week

per station Stations

Utilisation rate Versa 4,48 3 12,44% LeLo 9,36 4 26,00% DS8590FV 15,66 4 43,50% DS90IN 6,37 5 17,69% DS85IN 10,74 4 29,83% DS100140 11,51 6 31,97% Total 30,88 4 85,78% Total 27,24 6 75,67%

Table 6 The proposed line configuration

(26)

26 of 100% first is Line 1. This utilisation rate will be reached if demand were to increase with 16%. This means that if demand increases with 16% changes to the proposed configuration need to be made. It is advised, however, to make changes to the configuration when Line 1 reaches an utilisation rate of 95%, which occurs when demand increases with 10%. A drop in demand does not require that changes to the proposed configuration are made. It is, however, advised to rethink the configuration when demand decreases with 10% or more. As this decrease allows the DS 100/140 to be assembled on three stations while all assembly can still be performed on two lines. It is examined next what will happen when demand increases or decreases with 30% and what changes should be made to cope with these changes. Increased demand is examined first, followed by decreased demand.

If demand were to increase with 30% changes to the line configuration need to be made as utilisation rates will raise to 112% and 98% if nothing is done. Using the algorithm again with increased demand as input leads to the definition of two configurations that both require three assembly lines. The first configuration consists of three lines that each assemble two product families. This configuration requires fourteen assembly stations and has utilisation rates of 72.72%, 72.56% and 64.56%. The second configuration consists of three lines where on the third line only the DS 100/140 is assembled. The total station requirement is fifteen and utilisation rates are 88.73%, 79.56% and 41.56%. But, because the DS 100/140 is assembled on a dedicated line the number of stations on which it is assembled can be brought down to three. This raises the utilisation rate of the third line from 41.56% to 83.11% and lowers the total number of required assembly stations to twelve. Table 44 to Table 47 in Appendix C provide more details.

It is investigated next if changing the number of stations on which the DS 85/90 FV and DS 100/140 are assembled can provide better solutions if demand were to increase with 30%. Therefore, two options, which are assembling the DS 85/90 FV and DS 100/140 on eight and six stations, and assembling them on eight and nine stations, are explored. Using the algorithm one configuration was created for both options. The first configuration consists of three lines with utilisation rates of 88.75%, 64.65% and 28.28% and requires a total of eighteen stations. The number of stations can be brought down in the same manner as was the case in the second configuration of the previous paragraph. This results in a total of fourteen required stations and raises the utilisation rate of the third line to 56.56%. Table 48 to Table 50 in Appendix C provide more detail. Assembling the DS 85/90 FV and DS 100/140 on eight and nine stations leads to a configuration with two assembly lines that require thirteen stations and have utilisation rates of 88.72% and 78.97%, see Table 51 and Table 52 in Appendix C. Neither of these solutions provide a configuration that requires less than twelve stations. It is therefore proposed that if demand were to rise with 30% to use the configuration that requires twelve stations divided over three lines.

(27)

27 stations to eight and increase the utilisation rate of Line 2 to 44.75%. It is investigated next if reducing the number of stations on which the DS 85/90 FV and DS 100/140 are assembled can provide better solutions if demand were to decrease with 30%. If the number of stations on which the DS 85/90 FV and DS 100/140 are assembled is to be reduced only one option, assembling them on four and three stations, remains. Using the algorithm, two different configurations were created. Both alternatives consist of two lines and require a total of nine stations. Differences lie in the utilisation rates and the way in which product families are divided over the lines. The first alternative has three product families per line and has utilisation rates of 83.92% and 51.47%. The second alternative has four and two product families per line with utilisation rates of 92.55% and 42.84%. See Table 56 to Table 58 in Appendix D for more detail. Both these alternatives require one station more than the configuration of Table 55. It is therefore proposed that if demand were to drop with 30% two lines, where one line assembles the DS 100/140 product family and the second line the remaining product families. An additional advantage of this configuration is that if demand were to pick up again one or two product families can easily be moved from the first line to the second.

All that remains is to further specify the DS 63/75 line. The current situation, which has been set up to produce 40 machines per day, divides testing over two stations, one for visual inspection and the second one for the test program, and has a total of ten test stations. The new situation will be set up to produce around fifteen products per day and will combine visual inspection and running the test program into one station. This requires one test station per daily output of eight machines. Therefore only two test stations will be required in the new situation. Three test stations will be placed, however, to account for the fact that difficult repairs can be performed at the test station which make testing take longer. The number of assembly stations currently is 24. This can easily be brought down to sixteen, which leads to an utilisation rate of 82%. When demand increases or decreases with 30%, utilisation rates change to 107% and 58% respectively. This requires additional stations to be placed. Additional test stations will not be necessary. Table 7 contains the proposed configuration and proposed changes when demand increases or decreases with 30%.

Line 3 Expected demand Increase in demand of 30% Decrease in demand of 30%

Weekly required hours 474 616 332

Number of Stations 16 20 11

Test stations 3 3 3

Utilisation rate 82% 86% 84%

Table 7 Configuration of Line 3 for normal, increased and decreased demand

Number of stations Utilisation rate Product families

Line 1 5 86% Versa DS8590FV DS85IN

Line 2 7 76% LeLo DS90IN DS100140

Line 3 19 82% DS63/75

(28)

28 An overview containing the lines that have been defined in this section with their size, utilisation rates and product families is shown in Table 8. This proposed situation requires a total of 31 stations, whereas the current situation performs assembly of the same product families on 79 stations. While this is a great reduction in the number of stations the downside is that delivery flexibility is decreased. This is because the three lines can only assemble one type of product family at a time, whereas in the current situation all product families are assembled simultaneously. Now that the assembly lines have been developed the cells that will supply them have to be defined, which will be discussed in the next section.

3.2. Creating subassembly cells

Two different options to create subassembly cells are explored. The first option that is explored groups subassemblies together, based on their geometrical and technical attributes, into subassembly families. The subassembly families are then assigned to subassembly cells in multiple ways and the best alternative is chosen. This is followed by the exploration of the second option, which is based on the same concept that was used to create the cells that are currently used. When both alternatives have been explored one is chosen for the final configuration.

3.2.1. Creating subassembly families

A hierarchical agglomerative clustering method was used to group the subassemblies together. This method consists of two steps; first a form of similarity is employed to create singleton clusters, or subassembly families, and second, clusters are merged into cells (Papaioannou and Wilson 2010). This method is normally used for operations that are reliant on machines and will therefore be used in an altered way. It will be stipulated when modifications have been made to the method. The attributes on which the cluster analysis is based, have been assigned through the use of visual identification.

A subassembly family is a collection of subassemblies sharing similar design and/or manufacturing characteristics (Lee-Post 2000). As a first step to create subassembly families, all subassemblies have been identified. The distinction between subassembly and assembly operations has not been made for the smaller, less complex machines. These are, coincidentally, the machines that in the current situation do not have their own cells, such as the machines that are made in the Exotics cell. In identifying the subassemblies, the assembly operations of these machines have also been included. This has led to the identification of 1570 unique subassemblies. Next, a classification based on geometrical attributes and technical attributes, which are also known as design characteristics and manufacturing characteristics (Kusiak 1985; Choobineh and Nare 1999), was created.

(29)

29 and difficult to assemble (H). The first category, easy to assemble, should contain subassemblies that can be made by workers of any skill. The second category, average to assemble, should contain subassemblies that require more skill to assemble, e.g. subassemblies that require certain settings or more complex assembly operations. The final category should contain the most complex subassemblies that only the most skilled workers can complete. What is different here is that categories consisting of both shape and dimensions have been defined a priori, whereas in regular coding and classifications systems subassemblies will be graded on each of the attributes. This has been done for two reasons; first it makes the sorting process easier; and second not all categories require an attribute of dimension, e.g. all circular shaped objects are roughly the same size.

Now that the categories have been defined subassemblies can be added to them. Subassemblies have been assigned to categories by first assigning them to one of the geometrical categories and then to one of the technical categories. Categories within the geometrical and technical categories are mutually exclusive, meaning that a subassembly is assigned to one geometrical category and to one technical category. Adding the subassemblies to categories has been done in a two tiered process. First subassemblies have been added to categories by viewing pictures and assembly instructions. The result of this first step was used as the basis for the second step. The second step consisted of judging the result of the first step with experienced workers who have more insight in the complexity of the subassemblies and making changes accordingly. A short outtake of the result from this step is depicted in Table 9. This table shows how a technical and geometrical attribute is assigned to the subassemblies.

Technical attributes Geometrical attributes

Number E A H CYL SRC LRC CIR RT OTH

V1.002.01 x x

2600219L x x

2601034K x x x

ST1-901 x x

V3.144.01 x x

Table 9 Subassembly category matrix

Sorting the matrix, of which Table 9 shows a part, resulted in fourteen subassembly families. Each subassembly family consists of all the subassemblies that have the same technical and geometrical attributes, e.g. CYL-E and CYL-A. The time that is required to produce these subassembly families is calculated next. Based on these times, subassembly families can be grouped together into cells. The time to assemble a subassembly family is calculated in the following manner: the time that is required to assemble the individual subassemblies is calculated first; the weekly demand for subassemblies is calculated second; and finally the total weekly required time to produce a subassembly family is calculated. These times are then used to group families together and assign them to cells.

(30)

30 of by 25 centiminutes, or 15 seconds. Data on the number of parts that subassemblies are made of was not readily available. The number of unique parts that each subassembly is made of is, however, known. Because the number of unique parts of a subassembly is either equal to or smaller than the total number of parts in a subassembly, the number of unique parts is used as a base to calculate the total number of parts. This is accomplished by multiplying the number of unique parts with a factor to come to the total number of parts. The factor that should be applied has been calculated by viewing a sample of the subassemblies, which is representative of all subassemblies, and calculating the average difference between the number of unique parts and total number of parts. The sample size has been calculated using equation (2) and (3):

And

Where: Sample size for a finite population

= Sample size for an infinite population The population; 1570

The confidence level; 95%, or 1.645 Response distribution; 50%

Margin of error; 10%

A margin of error of 10% and a confidence level of 95% have been chosen because these provide enough certainty for the situation at hand and because these are widely used in practice. This means that it can be stated with 95% certainty that the values found have a 95% chance of being right. Inserting these values in (3) and (2) returns a sample size of 65. Next, using a random number generator, 65 random numbers, excluding doubles, between 0 and 1571 have been generated. The total number of parts that go into the subassemblies that corresponded with the generated numbers were counted and added up, making a total of 1100 parts. The sum of the unique parts of all of the selected subassemblies is 562. Dividing 1100 by 562 results in a mark-up of 1.96. Making the way to calculate the time to assemble a subassembly: 25 * 1.96 * the number of unique parts.

(31)

31 chosen. The required hours per week in Table 10 are calculated to FTEs by dividing them by 30.6, which is 85% of 36 hours. The result of this can be seen in Table 11.

Table 10 Required weekly hours to produce subassembly families

Table 11 Required times to produce subassembly families in FTE/week

3.2.2. Assigning subassemblies to cells

(32)

32 Cells based on technical attributes Required FTEs in 2013 Required FTEs in 2014 Number of unique parts E 13,62 14,06 1332 A 8,75 9,06 1092 H 2,48 2,59 506 Total 24,86 25,71 2930

Table 12 Cellular configuration based on difficulty of assembly Cells based on geometrical attributes Required FTEs in 2013 Required FTEs in 2014 Number of unique parts CYL 4,52 4,69 587 SRC 4,80 4,96 590 LRC 11,17 11,58 1264 CIR 0,59 0,61 104 RT 2,93 3,01 196 OTH 0,83 0,85 244 Total 24,86 25,71 2985

Table 13 Cellular configuration based on shape

When using difficulty as the basis for creating cells it becomes immediately apparent that a cell with only H-subassembly families does not meet the size requirements for a cell. Therefore, H-subassembly families must be combined with A-subassembly families, which in turn have to be combined with E-subassembly families. Several combinations of E-subassembly families, using the data in Table 11, have been explored. The most suitable combination is depicted in Table 14. This configuration was chosen over other options, of which three can be seen in Table 59 to Table 61 which are located in Appendix E, because the total number of required workstations is below that of the others. The cellular configuration of Table 14 contains two cells of eight workstations and one of nine, for a total of twenty-five. The other options that have been identified all required twenty-six workstations. For 2014, each cell requires one additional workstation. The cell sizes can, however, be kept at their original sizes, since the increase raises the utilisation rate of the cells to 87%, for cell a, and 88%, for cells b and c, which should still be acceptable.

FTEs in 2013 Required Required FTEs in 2014 Number of unique parts Cell a 8,89 9,22 1095 Cellb 7,98 8,25 1016 Cellc 7,99 8,24 790 Total 24,86 25,71 2901

Table 14 Most suitable cellular configuration based on difficulty

a contains SRC-H, LRC-A, LRC-H and OTH-H

b contains SRC-E, LRC-E, CYL-A, SRC-A, CIR-A, RT-A and OTH-A c contains CYL-E, RT-E, CIR-E and OTH-E

FTEs in 2013 Required Required FTEs in 2014 Number of unique parts Cell a 6,80 7,00 730 Cell b 8,73 9,06 1080 Cell c 9,33 9,65 1084 Total 24,86 25,71 2894

Table 15 Most suitable cellular configuration based on shape a contains LRC-E, CIR, RT and OTH

b contains LRC-A and LRC-H c contains SRC and CYL

(33)

33 option requires the same amount of workstations but fewer unique parts than the other options, which are depicted in Table 62 to Table 64 in Appendix E.

The next step is to choose the final cellular configuration. After comparing the alternatives of Table 14 and Table 15 the final choice was made for the former option. The option of Table 14 was chosen because the configuration requires one workstation less and less storage locations for parts. Another advantage is that difficult jobs are separated from easier jobs, which makes sure that difficult jobs are handled by more adept workers. The final cellular configuration of subassembly is depicted in Table 16 with the required number of workstations.

Required FTEs in 2013 Required FTEs in 2014 Number of unique parts Number of workstations Cell a 8,89 9,22 1095 9 Cell b 7,98 8,25 1016 8 Cell c 7,99 8,24 790 8 Total 24,86 25,71 2901 25

Table 16 Final cellular configuration of subassembly a contains SRC-H, LRC-A, LRC-H and OTH-H

b contains SRC-E, LRC-E, CYL-A, SRC-A, CIR-A, RT-A and OTH-A c contains CYL-E, RT-E, CIR-E and OTH-E

The proposed cellular configuration and the number of workstations are based on a forecast. And because forecasts are always wrong (Hopp and Spearman 2008), it will now be examined how this configuration will perform if the projected demand drops or rises with 30%. Utilisation rates for 2013 with both an increase and decrease in demand are given in Table 17. This shows that the proposed configuration needs to be modified if demand increases with 30%, i.e. stations should be added, while change is not needed when demand decreases with 30%. Table 18 shows the proposed number of stations when demand increases or decreases with 30% and the corresponding utilisation rates.

In this section subassembly cells have been identified that require 25 workstations divided over three cells. Before this configuration is proposed, however, a second alternative which is based on the same

Cells of proposed configuration Utilisation rate with +30% demand Utilisation rate with -30% demand Cell A 109% 59% Cell B 110% 59% Cell C 110% 59%

Table 17 Utilisation rates of the proposed cellular configuration with an increased and decreased demand for 2013

Demand +30% Demand -30% Cells of proposed configuration Required number of stations Utilisation rate Required number of stations Utilisation rate Cell A 11 89% 6 88% Cell B 10 88% 6 79% Cell C 10 88% 6 79%

(34)

34 principle that has been used to define the cells that are used in the current situation will be explored. The next section will go into more detail. After this a final cellular configuration will be chosen.

3.2.3. An alternative cellular configuration

The relationships between the cells and the lines that have been defined in the previous sections can be seen in Figure 12. It can be noted from Figure 12 that each cell transfers subassemblies to every other cell or line, with the exception of Cell a, which does not transfer subassemblies to Cell c. As this will put a lot of strain on the logistics department an alternative configuration is explored before making a final decision. The alternative is based on the same concept that has been used to create the cells that are currently used. It assigns all the subassemblies that are required in one line to one cell. Figure 13 shows the relationships between the cells and lines in this alternative configuration. In the alternative configuration subassemblies are transferred over three different routes, whereas in the earlier developed configuration subassemblies can be transferred over fourteen different routes.

Cell a Cell b Cell c Line 1 Line 2 Line 3

Figure 12 Relationships between proposed cells and lines

Cell a Cell b Cell c Line 1 Line 2 Line 3

Figure 13 Relationships in alternative configuration

(35)

35

Required hours

per week FTEs

Number of stations

Utilisation rate

Number of

unique parts Product families of the subassemblies

Cell a 175,67 5,74 6 81,33% 894 Versa DS8590FV DS85IN

Cell b 185,58 6,06 6 85,92% 1786 LeLo DS90IN DS100140

Cell c 399,25 13,05 13 85.31% 516 DS63/75

Total 760,5 24,85 25 3196

Table 19 Data on cells in the alternative configuration

The two alternatives will now be compared. In order to avoid confusion the cellular configuration that was developed in section 3.2.2 will be called Alternative 1 and the configuration that was developed in this section Alternative 2. Alternative 1 performs better when it comes to the number of unique parts, as Alternative 1 requires 9% less unique parts. Alternative 2 does, however, have some advantages over Alternative 1. Overseeing transport movements will be clearer, communication between subassembly and final assembly will be easier and direct, and the workforce is already familiar with the way of working. Volume flexibility will be lower in Alternative 2; a change in volume that requires the assembly lines to be reconfigured has no impact on subassembly in Alternative 1, whereas a change in line configurations in Alternative 2 requires that changes are made to the subassembly cells as well. Both alternatives have their strong points with regards to product flexibility. Adding a new machine in Alternative 2 is easier, all subassemblies go to one cell, but in Alternative 1 less training effort will be required as workers will be familiar with the type of assembly operations that need to be performed. Alternative 1 has the subassemblies separated based on difficulty, which ensures that adept workers perform the assembly operations, which will result in less defects, i.e. higher quality. Alternative 1 has two teams of eight and one of nine, all of which lie in the optimal team size range of five to ten (Peters and Waterman 1984), whereas Cell a in Alternative 2 lies outside of this range. Table 20 shows a summary of the advantages and disadvantages of both alternatives.

Advantage Disadvantage

Alternative 1  Fewer parts stored in multiple

locations

 Higher volume flexibility

 Higher quality

 Team size within optimal range

 Movement of subassemblies less

orderly

 Higher strain on logistics

department

Alternative 2  Clear connection between

subassembly and assembly

 Easier to oversee transport

movements

 Easier on logistics department

 Staff familiar with way of working

 Has more parts in multiple

locations

 Lower volume flexibility

 Team size Cell a not in optimal

range

Table 20 Advantages and disadvantages of both alternatives

(36)

36 the movement of subassemblies will be easier to oversee and will place less strain on the logistics department. This, however, does not outweigh the advantages that Alternative 1, such as higher quality and higher volume flexibility, provides. Alternative 1 one is, therefore, chosen over Alternative 2 as the most favourable alternative.

The decision for Alternative 1 brings the total number of required stations to 56. These are divided over three assembly lines and three subassembly cells. Table 21 provides more detail for the defined lines and cells. When compared to the current situation the proposed situation requires 58% less stations, 134 in the current situation versus 56 in the proposed situation. A reduction in the number of stations is not the only advantage that proposed situation offers. Product flexibility and volume flexibility will go up, as will quality. The spatial requirements of the proposed situation will be examined next and compared to the current situation.

Stations Utilisation rate Produces

Line 1 5 86% Versa DS8590FV DS85IN

Line 2 7 76% LeLo DS90IN DS100140

Line 3 19 82% DS63/75

Cell a 9 84% SRC-H LRC-A LRC-H OTH-H

Cell b 8 85% SRC-E LRC-E CYL-A SRC-A CIR-A RT-A OTH-A

Cell c 8 85% CYL-E RT-E CIR-E OTH-E

Table 21 The defined lines and cells with their number of stations, utilisation rates and product families 3.3. Spatial requirement

The definition of the three assembly cells brings the total cell count, including Exotics, to seven. And the number of required workstations, excluding Exotics, is 56. In the current situation these same product families are produced on a total of 134 workstations. This means a reduction in workstations of 58% can be realised. What this means in terms of square meters will be examined next.

Currently, 1965m² is used in the first assembly hall to assemble all of the product families that have been added to the cells, except for the IM-35. The rest of the floor is filled with Powder coating, Stippling and offices (750m²), Exotics (150m²) and a module line for the SI-92 (175m²). The remaining space is used as aisles. The configuration that has been defined in section 3.1 and 3.2 requires only 965m² of shop floor space. The 965m² is based on spatial requirements of 15m² for an assembly station and 20m² for a subassembly station. Both include space for storage locations. This means that even if the size of Exotics were to double, because the IM-75 and SI-92 modules are added, over 1000m² will still be available for other activities in the first assembly hall, while the second assembly hall will be available for other activities in its entirety.

(37)

37

4. PRODUCTION CONTROL SYSTEM

In this section a production control system that will direct interactions between the blocks that have been defined in the previous section will be proposed. First a control system that will direct interactions between the blocks will be detailed. This will be followed by a system to control the assembly lines. And finally conditions that need to be met for the system to work and implications that the system will have will be discussed.

4.1. Kanban

Kanban has been found to be the most suitable production control system for feeding an assembly line (Akturk and Erhun 1999) and will therefore be used to control the movement of parts and subassemblies on the shop floor. Two general types of kanban systems exist. These are the single-card and the dual-card kanban system. The advantage of the dual-card kanban system is that it provides strong control over the production system; it is, however, difficult to implement (Akturk and Erhun 1999; Nicholas 1998). The single-card system is, when compared, easier to implement and provides shorter information lead times (Akturk and Erhun 1999). Because the single-card system is easier to implement and a card system is currently used between the cells and the warehouse, a single-card system will be developed.

A single-card kanban system works in the following manner: First, when a worker in cell b accesses a full bin, the kanban card is taken from it and placed in the kanban mailbox. This card specifies the type of material needed and from which cell it should be taken. Second, a material handler reads the card and takes the card and an empty bin to the specified upstream cell, cell a. And third, the card is attached to a full bin and is then sent back to cell b.

(38)

38 And the number of parts that are placed in a bin have to be counted, which will increase the workload of the logistics department even further. To spare the logistics department, initially one kanban card per item will be used. Once the system is functioning, variables can be changed to increase performance of the system.

(4) Where: Demand; Lead Time

Safety Factor; Items per bin

Product families DS85IN DS90IN LeLo Versa DS8590FV DS100140 DS63/75

Items per bin 11 9 7 5 19 6 10

Table 22 Number of items per bin per product family

Table 22 shows how many items should be placed in each bin, which depends on the product family the bin is for. Table 22 is based on the assumption that a part or subassembly is used once in an end-item. If in an assembly line a subassembly or part is used more than once per station, the number of items per bin should be multiplied by the number of times it is required. This lies a bit more complicated for the subassembly cells, as it depends on the combination in which the parts will be stored. Next to the countable stock there are also small parts that cannot be counted, such as screws and fasters. These parts, which are found in every type of machine, should be delivered in standard amounts.

4.1.1. Signal

Developing a new layout for Neopost Technologies B.V.