University of Groningen
Student:
L. Gan
Student number:
s1546414
First supervisor:
prof. dr. ir. J. Slomp
Second supervisor:
dr. J.A.C. Bokhorst
Company supervisor: H.J.M. Busschers
Study:
Msc BA Operations and Supply Chain Management
Faculty:
Economics and Business
Research center:
Lean Operation Research Center (LORC)
Company:
Company E
Date:
15 March 2010 – 15 July 2010
Abstract
This research investigates direct line delivery (DLD) implementation in sheet metal unit (SMU) of EGS for warehouse inventory reduction. Management expects appropriate implementation design from this study. First, SMU should have short lead time, low process complexity, and high capacity adaptability while assembly line should promise fixed assembly schedules within certain period before implementation. We build a financial benefit and cost comparison model to select suitable articles for implementation. In general, suitable articles have low demand, low order frequency, high value and large size. Top management should apply a top-down control structure to enhance cooperation among departments for implementation.
Acknowledge
Content of Figures and Tables
Figure 1.1 Energy Distribution Network ………..…8
Figure 1.2 Layout of BSC (Balance Score Card)……….9
Figure 1.3 Logistic Model EGS………...12
Figure 1.4 Current Stock Value from SMU……….12
Figure 1.5 Average Lead Time Splited……….13
Figure 1.6 Current State VSM SMU……….14
Figure 1.7 Illustration of DLD Concept……….15
Figure 1.8 Future State VSM SMU………16
Figure 1.9 Work Load Control………16
Figure 1.10 Conceptual Model……….18
Figure 1.11 Research Questions Related to the Conceptual Model………20
Figure 1.12 Action Research Cycle……….20
Figure 2.1 Theme Illustration of Chapter 2………..22
Figure 2.2 Entire Order Flow………..23
Figure 2.3 Two-day Time Bucket Releasing……….23
Figure 2.4 OTR Improvement by Deducting Order Administrative Lead Time after Actual Production……….25
Figure 2.5 Layout of SMU………..28
Figure 3.1 Theme Illustration of Chapter 3………..32
Figure 4.1 Theme Illustration of Chapter 4………..36
Figure 4.2 Scheme of Implementation………...38
Figure 5.1 Theme Illustration of Chapter 5………..54
Figure 5.2 Classification of Inventory in Warehouse “003” in April………55
Figure 5.3 Expected Order & Delivery Pattern with Warehouse 003 Compared to Actual Patterns………56
Figure 5.4 Short Term Delivery Pattern Change………57
Table 1.1 Overview of Groups EGS……….9
Table 1.2 Overview of Processes in SMU………11
Table 1.3 Operationalizations of Lead Time, Process Complexity and Capacity Adaptability……….……….18
Table 1.4 Operationalizations of Expected Performance Influence………19
Table 1.5 Operationalizations of DLD Implementing Decision………19
Table 2.1 From/To Matrix SMU……….27
Table 2.2 Top 10 Routings SMU………27
Table 2.3 Capacity Information for Each Process in SMU………28
Table 3.1 Potential Performance Changes due to DLD Implementation……….34
Table 4.1 Proxies in Benefit & Cost Model………40
Table 4.2 Notation……….41-43 Table 4.3 Common Parameter Value………43
Table 4.4 Inputs and Outputs of Calculation under Delivery Pattern “via Warehouse”………45-46 Table 4.5 Inputs and Outputs of Calculation under DLD in Short Term………48
Content
1. Introduction………8
1.1 Company Introduction……….8
1.1.1 Company E General Supply (EGS)……….8
1.2 Introduction to the Research Problem………..11
1.2.1 Current Situation………..12
1.2.2 Direct Line Delivery (DLD) Concept………..14
1.2.3 Conceptual Model………17
1.2.4 Research Question………..19
1.2.5 Boundary Conditions……….20
1.3 Research Design……….20
1.3.1 Outline……….21
2. Preconditions: Getting Ready for DLD Implementation………22
2.1 Lead Time………22
2.1.1 Lead Time Reduction as Important Precondition………..22
2.1.2 Lead Time Analysis………..23
2.1.2.1 Data………23
2.1.2.2 Methods and Results……….23
2.1.3 Actions to Reduce Lead Time………25
2.2 Process Complexity……….26
2.2.1 Low Process Complexity as Important Precondition………26
2.2.2 Process Analysis……….26
2.3 Capacity Adaptability……….28
2.3.1 High Capacity Adaptability (Sufficient Capacity) as Important Precondition……….28
2.3.2 Capacity Analysis………..28
2.4 Fixing Assembly Schedule………30
2.4.1 Assembly Line Fixing Assembly Schedules within Certain Time as Important Precondition………30
2.4.2 Possibility of Fixing Assembly Schedules……….30
2.5 Chapter Summary……….30
3. Performance Changes due to DLD Implementation………..32
3.1 Quality………..32 3.2 Speed……….33 3.3 Flexibility……….33 3.4 Dependability………..33 3.5 Costs………..34 3.6 Chapter Summary……….34
4. DLD Decision Support Framework: Benefit & Cost Model……….36
4.1 Reason for Deciding on Financial Results……….36
4.2 Scheme of Implementation………37
4.3 Method of Model Develop……….38
4.4 Formulation of Model………41
4.4.1 Logic of Model………41
4.4.2.1 General Equation……….43
4.4.2.2 Calculation under Delivery Pattern “via Warehouse”………45
4.4.2.3 Calculation under DLD………..47
4.4.2.4 Comparison of Two Delivery Patterns………52
4.5 Chapter Summary……….53
5. An Illustrative Example: Action Plan for Assembly Line MMS………...54
5.1 MMS Assembly Line...………54
5.2 Implementation: Action Plan Making……….55
5.3 Implementation and Stock Reduction………59
5.4 Chapter Summary……….59
6. Management Recommendation……….60
6.1 Change Management Strategies………60
6.1.1 Early Stage……….60 6.1.2 Middle Stage………61 6.1.3 Later Stage……….61 6.1.4 Section Summary………..61 6.2 Future DLD Improvement………61 6.3 Chapter Summary……….62
7. Conclusion………..63
Reference……….64
Appendix………65
Appendix A: Value Stream Map of SMU EGS……….……65
A1: VSM_Current State……….65
A2: VSM_Future State………66
Appendix B: Normal Distribution of Lead Time………67
B1: Normal Distribution of Administrative Lead Time Before Production (full sample)………..67
B2: Normal Distribution of Administrative Lead Time Before Production (Kanban orders)………68
B3: Normal Distribution of Administrative Lead Time Before Production (MRP & PRP orders)…………69
B4: Normal Distribution of Production Time (actual production time + order administrative time after actual production)……….70
Appendix C: Order Flows Before Actual Production Starts………..71
1. Introduction
This thesis delivers a research of implementing Direct Line Delivery (DLD) in the general supply department of Company E (EGS). The general intention of applying DLD is to reduce inventory. The introduction chapter of this thesis starts with the introduction of the company. The introduction of the research problem and the research design come after.
1.1 Company Introduction
This research is done in Company E which is located in the plant of Company E in Hengelo (Netherlands). “Company E is part of E Corporation. E Corporation is a diversified power management company with 2009 sales of $11.9 billion. E Corporation is a global technology leader in electrical components and systems for power quality, distribution and control; hydraulics components, systems and services for industrial and mobile equipment; aerospace fuel, hydraulics and pneumatic systems for commercial and military use; and truck and automotive drivetrain and powertrain systems for performance, fuel economy and safety. E Corporation has approximately 70,000 employees and sells products to customers in more than 150 countries.” (E Corporation’s official website, 2010) “E Corporation is a
partner for utilities, electrical contractors and light and heavy industry.” (E Corporation’s official website, 2010) Company E has more than 1200 employees in the Netherlands and abroad (Van den Berg, 2009). The annual turnover exceeds € 11 million. Company E provides medium voltage and low voltage products and project solutions to suit installation requirements throughout utilities, industrial, commercial and residential sectors (customers are located in the Energy Line, see Figure 1.1). The company is a world leader in low voltage distribution and a dominant player in the medium
voltage market. The mission of Company E is “To be our customers’ best supplier, providing distinctive and highly valued products, services and solutions.” (Company E’s official website, 2010) 1.1.1 Company E General Supply (EGS)
Company E General Supply (EGS) is a supplier for the licensees of Company E (internal supply). Hence, EGS’ major customers are the assembly lines of Company E. EGS supplies Low Voltage Systems (LVS), Medium Voltage Systems (MVS) and Low Voltage Components (LVC). It manufactures almost all the metal parts for the production lines of Company E. Currently, there are three producing units within EGS, namely, Sheet Metal unit (SMU), Turning and Milling unit (TMU), and Copper Bar unit (CBU). The other unit for Punching (PU) which used to be under EGS has been replaced by outsourcing. The other sector has been outsourced recently is the tool shop. EGS has these units outsourced because it has to concentrate on the three capital-intensive units left (SMU, TMU, and CBU). Table 1.1 shows an overview. The vision of EGS is “being the fastest and most reliable supplier of Company E”. To put “being the best supplier” more specific, their order winner is fast and reliable delivery. Recently, EGS is asked to use a Balance Score Card (BSC) to link its own developing goals to the developing goals of the whole plant in Hengelo. Figure 1.2 presents the layout of the BSC. The BSC is also used to tract the monthly developments with the monthly goals.
The main performance indicators in the BSC are on time request (OTR), lead time (LT), productivity, budget variance, absenteeism, warranty, warranty costs, stock level and personnel number.
Units Size (Absorption Value FYR 2009) Typical Processes SMU (including PCO) Sheet metal
€3,649,594 SawingàPunchingàBendingàWeldingàPowder Coating;
CuttingàPunchingàBendingàAssemblyàPowder Coating
TMU (& PU) Turning & Milling
€2,620,756 SawingàTuring(Milling)àDrilling TappingàSurface Treatment;
Sawing(Punching)àDrillingTappingàSurface Treatment
CBU Copper
bar
€1,016,912 Sawing(Punching)àBendingàDrilling TappingàSurface Treatment
Table 1.1 Overview of Groups EGS
Figure 1.2 Layout of BSC
Workload Control System (WLC)
At present, almost all three processing units (SMU only for Group A single parts, TMU, and CBU) have implemented the Workload Control System (WLC) to control the work flow on the shop floor. The system has a lean planning and production control system. The planning and control can be separated into two elements. The release system controls the amount of released orders per day based on the number of jobs allowed in the work floor and the order pool. The production control unit is based on the concepts of ConWIP, FIFO and Takt time control. In other words, there are constant limits of the WIP in the system that follows the FIFO priority rule. Tasks are visualized by implementing the production progress screen.
stable input for the work floor. The planner assigns a specific colour to a specific start date of an order. Then the operator starts producing orders according to different colours assigned. For orders with the same due date, the operator has to follow the FIFO rule to start orders.
The first production step activates an order so that an order becomes in production. After being activated, an order will stay on the screen until it is finished. Similar to the operators for the first process, the operators of the following process steps take the orders that are longest on the shop floor, following the FIFO rule. The WIP screen indicates the late orders by marking them, which are staying in the system longer than the Takt time allowed. The operators are triggered to finish the late orders as fast as possible. After finishing the orders that are relatively long in the production process, the operators are asked to give a reason for the lateness. The information will be later used for evaluating the system. During the production, there is a production progress control. This system compares the number of orders that have been finished with the orders that should have been finished so far (based on Takt). Then the backlog is calculated. By monitoring their progress on the screen, operators are stimulated to realize the daily production (Slomp, Bokhorst and Germs, 2009).
Sheet Metal Unit (SMU)
This research focuses on the sheet metal unit of EGS. The management team intends to use SMU as a pilot for DLD implementation. The first reason is SMU has more complicated process complexities than the other two units. If the implementation is successful in SMU, the other units armed with the experience from SMU may implement later with more ease. The second reason is that one important precondition for DLD implementation is to have the lead time under control1. The lead time in SMU is expected to reduce most apparently by the new control system that will be launched later. During the research period, a new control system, dynamic machine management (DMM)2 will be launched as well. It is expected to reduce lead time in SMU more than in the other two units since DMM provides automatic nesting process which is a process step only for SMU.
Demand of SMU
Currently, SMU consists of 35 permanent and 11 temporary employees that work on two shifts. They produce approximately 24000 orders a year with an average of 600 orders per week. Among these 600 orders, there are around 150 or 200 orders for each order type (MRP orders, PRP orders, and Kanban orders). The main customers of SMU are LVS and MVS. In near future, the whole plant follows the SIOP (sales inventory & operations planning) concept, which aims to match demand and supply. Under this concept, sales (demand side) will provide demand forecast for the coming three months every month3. They can roughly define forecast in to three basic demand levels: low, median, and high demand levels. EGS (supply side) then can adjust its producing capacity according to these demand forecasts.
Routings of SMU
Regarding to the concrete routing description, we here use the findings of the research of Van den Berg (2009). This is because the current routings do not change much compared to those during the research period of Van den Berg (September 2008 to March 2009). According to that research, SMU has a job-shop layout. Although there are only five main processes on the shop floor, namely,
1 After applying DLD, order quantity is expected to reduce while ordering times are expected to increase. This will bring in more extra costs
if the non-producing lead time (administrative lead time before and after producing) is long for every order. Hence, reducing lead time, especially administrative lead time, is one precondition for DLD implementation.
2 DMM will be introduced more in the lead time analysis chapter.
3
nesting, punching, bending, bench working and welding, there are 126 routings that consists of many customized orders. This is mainly because of the possibility of the bidirection of routings between processes. There are two producing groups within SMU. Group A is for the single part producing and Group B is for the assembled part producing. Some of the produced parts of Group A are assembled to end products by welding and bench working in Group B. Table 1.2 gives the overview of processes for both Group A and B.
Hierarchy of SMU
The logistics manager is the main person responsible for the coordination in SMU. His main task is controlling the overall logistics. To achieve this, he supports the supervisors, engineers and manages planners. The supervisor is responsible for the production of all parts. The planners construct a plan based on incoming orders from the MRP system and the engineers. Finally, the engineers customized orders, from the Project Resource Planning system, which begins with sketches and ends with CNC drawings.
1.2 Introduction to the Research Problem
As mentioned above, because of the WLC system, most of the shop floor flows are under control now. Therefore, the planner does not spend most of his time solving problems of late or lost orders and checking the material attendance. Now the management team of EGS draws their attention to a upper level planning, from shop floor control to operational planning in EGS. Figure 1.3 presents the visualisation of the logistic model that the focus moves from shop floor control (black frame) to operational planning (blue frame).
Figure 1.3 Logistic Model EGS 1.2.1 Current Situation
There are mainly three types of orders in EGS, namely, MRP orders, Kanban orders, and PRP orders. Both Kanban orders and the PRP orders are delivered to customers directly, either to the stock on assembly line or to customer specific projects. However, MRP orders are currently transported to and stored in the main warehouse “003” before they are finally delivered to customers. The current stock level of the finished products for entire EGS is around € 2.7 million, of which about € 356,000 obsolescence4. Among this € 2.7 million, about € 840,000 is from SMU. This € 840,000 from SMU can be further roughly classified into two types: stock in the main warehouse “003” and the rest (most of the rest is stock online).
The current stock level in “003” is € 400,468.03. Figure 1.4 shows the current
inventory level from SMU in each warehouse. The stock implies many costs, such as interest opportunity costs on inventory holding, the obsolescence costs per year, material handling costs to the main warehouse and so on. Briefly speaking, the present concern of the management team is how to lower stock level, especially the stock level in “003”. They consider that the product delivery (logistical) pattern change of MRP orders is one solution. Specifically, they intend to change the current situation (via “003” and then to lines) to DLD pattern (directly to line, no via “003”). Hence,
4
Most of the obsolescence is technical obsolescence which is due to product version development. When a new version applies, the older versions become obsolescence.
the overall subject of this research is implementing DLD system in SMU to reduce the finished goods inventory in the main warehouse “003”.
Management considers lead time reduction as the most important precondition for implementing DLD to reduce stock level. Under DLD concept, smaller order size but more ordering times can be expected. If the set up time and the administrative lead time for each order are long, DLD will generate great extra costs due to the decreased order quantity and more ordering times. At present, the average total lead time in SMU is approximately 100 hours, starting from order creation date till order ready date. This average 100 hour consists of 62.65 hours for administrative lead time (before process starts) and 36.79 hours for production lead time. Then further split the administrative lead time into lead time between order creation and order document print (Gepland-Doc. Print), lead time between order document print and order release (Doc. Print-Vrijgegeven), and lead time between order release and the first step of production (Vrijgegeven-Actief). We also split the production lead time into real
production time from the start of first producing step and the finishing of the last step (Actief-Logtijd) and the order administrative lead time between the finishing of last producing step and the final order readiness (Logtijd-Gereed). Figure 1.5 shows the detailed elements of the total lead time based on the order data in the first quarter of 20105.
Showing in the current state VSM SMU (Figure 1.6), Kaizen A presents the current problem of high inventory level in “003”. In order to solve this problem, the
management team thinks about implementing DLD. Then before launching DLD, administrative lead time per order before and after the real production should be reduced as much as possible. This then indicates the other two Kaizens: Kaizen B presents the reduction of administrative lead time before the process activates; Kaizen C presents the reduction of order documentation time after the process ends.
5
We first rule out some incorrectly recorded data. We calculate the networking hours between dates by excluding weekends and public holidays and by counting every workday as 24 hour producing. To split the production lead time, we exclude orders with the last producing step as powder coating (PCO) or external producing. No order administrative time after actual production can be calculated if the PCO is the last producing step since the PCO operator should have updated the logtijd after he completes the PCO process. The total lead time (all orders) is estimated to be around 120 hours since PCO is expected to take extra 20 hours on average. In addition, the order administrative lead time after process end can vary much if the last producing step is outside EGS due to the external waiting time and transporting time. We exclude these orders when calculate the average lead time, since they will bias the real internal order administrative lead time after actual production.
Splited Lead Time (Q1)
0,00 20,00 40,00 60,00 80,00 100,00 120,00 Q1 Average Month H o u r Logtijd-Gereed Actief-Logtijd Vrjgegeven-Actief Doc.geprint-Vrijgegeven Gepland_Doc. Geprint
Figure 1.6 Current State VSM SMU (for large image, see Appendix A) 1.2.2 Direct Line Delivery (DLD) Concept
DLD Concept
The direct line delivery means products will be delivered to the next process (here: assembly line) once the manufacturing is finished, so that there will not be no final product inventory from the perspective of supplier (here: EGS). Meanwhile, the direct delivery should also be a JIT delivery for customers, which means the manufacturer supplies at the exact time with the exact amount which goes directly to assembly. Therefore, DLD has apparent characteristics from a pull system. With the optimal vision of DLD, manufacturers should produce based on the real order information instead of any demand forecasts. Manufacturers should also be able to adjust its producing capacity flexibly to cope with different demand levels.
Current DLD in EGS
Figure 1.8 Future State VSM SMU (for large image, see Appendix A)
Desired DLD in EGS in Future
In order to achieve the future flow shown in Figure 1.7, all phases of order (see Figure 1.9 for the theoretical model) has to change.
Order Entry
On the entry level, assembly lines and EGS have to negotiate about the order quantity and frequency, so that assembly lines can provide EGS fixed assembly schedules within certain time length. Then EGS will produce and deliver the exact amount needed on assembly lines.
Capacity Adaptation
The capacity will be adapted quickly according to different levels of demand. Roughly, it is the three demand levels given by the SIOP concept mentioned earlier, namely, the low, medium, and the high demand. With high quality forecasts in advance, EGS is able to adapt capacity requirements fast enough by switching machines, by working over time or by hiring extra labour.
Order Release and Prioritizing
With the flexibly adjusted capacity, the planner will change the current release of planning system based to an automatic Lot-to-Lot release. Then EGS applies its WLC system to have constant WIP according to the adjusted capacity levels. The controlling screen based on the FIFO rule and the Takt time can still be applied to control the shop floor flow for the production processes.
Eventually, EGS wants to have a system that can cope with assembly line demand closely, speed up delivery and reduce stock level by DLD, and still maintain the production transparency and efficiency. 1.2.3 Conceptual Model
Figure 1.10 presents the conceptual model for this study. As mentioned earlier, lead time reduction is one precondition of DLD implementation. The total lead time can be divided into four parts: administrative lead time before actual production, administrative lead time after actual production, machine set up time during the producing processes, and the cycling time (including waiting time between different processes). Under DLD concept, we expect smaller order size but more frequent orders (ordering times). If the administrative lead time and the set up time for each order are long, DLD will generate great extra costs due to more ordering times and decreased order quantity. In order to generate fewer additional costs after DLD implementation, we have to first reduce lead time as much as possible. As shown earlier in the VSM, Figure 1.6, the administrative lead time before and after the actual production are two Kaizens that require improvement actions at present. Compare to the administrative lead time, the actual producing time has less improvement margin currently. This will be explained in details in the precondition discussion chapter later.
capacity adaptation. The impact from DLD implementation on capacity can be interesting topics for future research but we do not research it in depth in this paper.
There is one more important precondition that assembly lines have to be able to fix their assembly schedules within certain time interval, for example, one week. Assembly lines can not change assembly schedules without informing EGS in advance. For instance, they should not change order due dates after the production in EGS starts. This is simply a guarantee needed from assembly lines, so that we do not do data analysis for this precondition.
Figure 1.10 Conceptual Model
Factor Operationalizing
• Lead time, process complexity and capacity adaptability: Here by lead time we mean the total time passed from the time an order is created to the time order is completed (readiness for delivery). The investigation on process complexity is mainly from the research of Ven den Berg (2009). In his research, the author analyzes many variables to present the multiplicity, uncertainty and flexibility of production. Table 1.3 summarizes proxies selected here to present lead time, process complexity and capacity adaptability.
Operationalizations
Lead Time Total time from order creation to order completion Process Complexity Number of process steps6
Capacity Adaptability
Idle Capacity;
Capacity Adapting Speed;
Risks of Machine Breakdown and Labour Absenteeism; Operators and machine’ Multiple functions.
• Fixed assembly schedules: The other precondition is simply a “yes/no” question: Can the assembly schedule of a specific article on assembly line can be (negotiated) fixed for a certain period of time? When the answer is “yes”, that specific article may be a candidate for applying DLD, vice versa.
6
There are also other proxies to present process complexity. Here we mainly discuss the number of process steps for orders. Different operations needed for one order relate to capacity requirements and (interoperation) lead time, which should be controlled first to avoid possible cost increase and dependability decrease created by implementing DLD.
• DLD Readiness: This means to check whether SMU is ready to implement DLD on a general level. This decision will depend on the readiness for DLD, which is affected by the four preconditions, short lead time, low process complexity, high capacity adaptability for sufficient capacity, and guarantee from assembly lines of fixed assembly schedules during certain time period.
• Performance: According to Slack and Lewis (2007), the five performance measurements are quality, speed, flexibility, reliability, and costs. We will discuss the expected performance changes due to DLD from these five perspectives.
Operationalizations Expected Performance Influence Quality;
Speed; Flexibility; Dependability; Costs
Table 1.4 Operationalizations of Expected Performance Influence • DLD Implementing Decision: This means the concrete DLD implementation decisions. The first decision is “which articles should switch to DLD?”. This decision can be formed according to the model analysis results by comparing the expected financial benefits and costs7. After the article selection, more concrete action decisions have to make, such as “how to deal with the existing inventory in the main warehouse?”, and “What new prices should be set when the usual order quantities decrease?”.
Operationalizations
DLD Implementing Decision Decision of article selection; Existing inventory treatment; New price setting
Table 1.5 Operationalizations of DLD Implementating Decision 1.2.4 Research Question
Based on the introduction above, the theme of this research is DLD implementation. The main research question can be formulated as follows:
What is an appropriate design for DLD for improving the performance of sheet metal unit (SMU) in Company E General Supply (EGS)?
This main research question is further divided into 5 sub questions:
• How is SMU ready for implementing DLD, in terms of lead time, process complexity, capacity adaptability and fixed assembly schedules promised by assembly line within a certain period?
• What are the potential performance changes if DLD is launched? • How to judge which articles should switch their delivery pattern to DLD?
7
• What is DLD implementation plan for the pilot assembly line MMS? • What are the management recommendations for DLD implementation? Figure 1.11 links all five sub research questions to the conceptual model.
Figure 1.11 Research Questions Related to the Conceptual Model
1.2.5 Boundary Conditions
There are a few boundary conditions to this thesis: • The thesis should be finished within 5.5 months.
• The main focus is on SMU, so analysis arguments given for the whole EGS is basically the estimation based on evidence from SMU.
• The thesis should be within the guidelines of the manual: 'thesis and graduation manual for MScBA Operations and Supply Chain Management.
1.3 Research Design
The approach for answering the research questions is part of the Action Research approach (Coughlan and Coughlan, 2002) The action research model is generally presented as a loop of six phases (see Figure 1.12). Before starting a research cycle, an introduction states the context and purpose of research. The loop, namely, data gathering, data feedback, data analysis, action planning, implementation and evaluation is done during the project process. However, this research excludes the evaluation of result evaluation. Therefore, in
this paper, we can only discuss the theoretical performance changes for implementation decision making, instead of the actual performance results of execution. The main purpose for this research is to develop tool for DLD implementation decision making. The research period is too short to see the actual results of actions. Data are gathered from the Baan system or given by interviews. The researcher had weekly meetings to report and to get feedback on the research process. Based on the data analysis, the researcher together with stakeholders generate action plans for implementations. All designs and implementations are taken in realistic small steps to reach as far as possible within the agreed research time.
1.3.1 Outline
2. Preconditions: Getting Ready for DLD Implementation
As shown in the conceptual model earlier, four factors need to be under control in order to be ready for DLD implementation: lead time, process complexity, capacity adaptability and fixed assembly schedule guarantee during certain time interval. In this chapter, we will discuss all four factors to prove that SMU will be ready for the implementation. Hence, this chapter will deliver answers to the first sub research question “How is SMU ready for implementing DLD, in terms of lead time, process complexity, capacity adaptability and fixed assembly schedules promised by assembly line within a certain period?”. Shown in the conceptual model, we discuss factors and relations marked in red in this chapter. (see Figure 2.1)
Figure 2.1 Theme Illustration of Chapter 2
This chapter includes five sections, lead time (2.1), process complexity (2.2), capacity adaptability (2.3), fixing assembly schedules (2.4), and a brief chapter summary (2.5). At the beginning of each section, the reason why each factor is important precondition will be readdressed. Then the (real data) analysis of each factor will show if SMU is ready for DLD from the perspective of that particular factor. In case that any improvement is needed, detailed improving actions will be given as well. Main conclusions will be written in italics. At the end, a brief conclusion section will readdress the key conclusions of this chapter.
2.1 Lead Time
2.1.1 Lead Time Reduction as Important Precondition
Under the current delivery pattern “via 003”, either assembly line can combine long time demand in one demand order or the planner can combine several demand orders into one production order8 to take the advantage of economic producing quantity. An integrated quantity is delivered first to the main warehouse “003”. Then the actual demands online at different moments are fulfilled by each time delivery from “003” to the line. However, under DLD, assembly line will order what is needed exactly each time, while EGS will produce and delivery this exact amount. With fewer opportunities to gather demands, we may expect smaller order quantities each time but more frequent orders with probably more total ordering times after implement DLD. According to Schönsleben (2007), lead time
8
is the sum of administrative time, operation times, and interoperation time. Operation time can be further divided into machine set up time (per order) and machine process time per piece times the whole order lot. With smaller order quantities, the actual cycling time of each order will decrease. However, there will be more set ups and greater capacity utilization, which may eventually increase lead time and costs. With more ordering times, total order administrative time (non-producing time needed) also goes up. In order to reduce potential lead time waste increase due to DLD implementation, great efforts should be spent on setup time reduction and order administrative time reduction9.
2.1.2 Lead Time Analysis
2.1.2.1 Data
The lead time analysis is based on the data sample of orders that occurred in the first quarter of the year 2010. There were totally 5405 orders for SMU in this quarter, with 1772 Kanban orders (555 in January, 504 in February, and 613 in March) and 3733 MRP and PRP orders (1279 in January, 1284 in February, and 1170 in March). After ruling out orders with incorrect10 or unavailable data, we finally include 5364 orders in our whole sample. As mentioned in the first chapter, the total lead time can be divided into five elements, time from “plan” to “print”, time from “print” to “release”, time from “release” to “process active”, time from “process active” to “process end” , and time from “process end” to “order complete”. Figure 2.2 presents the visualized order flow in general. We investigate how the lead time distributes among these five elements.
Figure 2.2 Entire Order Flow
2.1.2.2 Methods and Results
First, we investigate the order administrative time before the real process starts by calculating time between “plan” and “print”, time between “print” to “release”, and time between “release” and “process active” after excluding weekends and public holidays in between for all 5364 orders in the sample. We count every workday as 24 hour producing even when the production closes at 18:30 on Fridays since the non-producing hours on Fridays are still considered as part of the lead time from the perspective
of customer. We measure the time in hours
9
Firms should reduce lead time as much as possible before implementing DLD. However, it does not imply that firms do not have to control and to reduce lead time after DLD launch. If there are potential for further lead time reduction due to technique developments, firms should keep on lowering lead time during DLD appliance.
10
For example, with recording mistake, some orders show earlier completion date than plan date.
instead of days to give more accurate analysis. The results show that there are on average 17.4 hours between an order is created (plan) and the order is printed out in documents. There are on average 5.5 hours from the order document is available to the order is ready for release by the planner. However, these two average values may not present the actual fluctuations among orders well since the normal distribution (see Appendix B) shows high kurtosis value (fat tails). This can be due to different order sizes and different drawing, version and material availability checking workload before the release. There are on average 39.8 hours for releasing an order to the nester and waiting for the nesting starts. This average time length fits the current two-day time bucket rule for releasing. The two-day time bucket rule means the planner releases orders on a certain day, while the production for these orders on the shop floor will start within two days11 (see Figure 2.3, check same colour for releasing and producing).
The other reason why the average time plan-print and the average time print-release are not very representative is that Kanban orders and MRP and PRP orders have quite different time distributions among these three elements of order administrative time before process activated. For Kanban orders, the order creation (plan) and order document printing usually take place around the same time (both steps are done by Kanban card scanning). The drawing, version and material availability will be checked later before the release. However, for MRP&PRP orders, the planner usually prints out order documents when the orders are ready for release (after all checking work has been completed). Hence, for these orders, the time interval between plan and print can be long, while print and release usually take place around the same time. These different patterns are proved by the calculation when separate Kanban orders and MRP & PRP orders: For Kanban orders, the average time between order plan and order document print is 0.09 hours. The average time between order document print and order release is 15.1 hours. For MRP & PRP orders, the average time between order plan and order document print is 25.2 hours, while the average time between order document print and order release is 1.15 hours. Appendix C presents order flows before real production starts of different order types, customer specific orders: MRP orders and Kanban orders.
Then we split the production time into actual process lead time (operation time) and order administrative time after the process finishes. The real operation time is the time between “process active” and “process end”. The order administrative time after process finishes is the time between “process end” and “order complete”. This is a period of time for booking registration to make the order finally ready for delivery. Currently, operators finish the production and transport the products in front of the inventory office. The products then wait there for the booking by the staff from the inventory office.
After the data investigation, the split up is not possible for orders of which the last process is powder coating. The operators of powder coating (PCO) should have recorded the log time (process end time) just like other operators. However, the operators of PCO are usually not good at operating computer, so they do not do the data updating after their producing operations finish. Hence, we have to analyze these two lead time components based on another data sample. We rule out all orders with PCO as the last production step and orders with external process12, then we have totally 2446 orders left. Concerning the actual processing time and the order administrative time after process end, Kanban orders and MRP and PRP orders do not have apparent pattern differences. Thus, it is not necessary to further split sample for Kanban orders and MRP and PRP orders. Based on
11
The two-day time bucket does not mean every order will activate for production in exact 48 hours. It just implies that if the planner releases an order on Monday, this order should start for production on Wednesday. However, at which time exactly on Wednesday operators will start production on the shop floor really depends on the order priority (FIFO, always choose the oldest order in the ConWIP system) shown on the control screen.
12
the 2446 observations, we calculate the average actual process time is 28.42 hours13 and the average order administrative time after process end is 8.37 hours.
Based on the investigation above, we can also spilt the total lead time into two parts, the actual producing time (28.42 hours) and the non-producing time (71.07 hours = 17.4 + 5.5 + 39.8 + 8.37). At present, the non-producing time is 2.5 times of the actual producing time. There should be large margin for lead time reduction. In other words, before implementing DLD improvement actions must be taken to lower the lead time.
2.1.3 Actions to Reduce Lead Time
Action 1: Reduce Time Bucket for Releasing Order
Starting from April 2010, the planner changed the release time bucket from two days to one day. In other words, from then on if the planner releases an order today, the order will start its production tomorrow. Then we expect time between “release” and “process active” can be reduced by on average 12 hours14.
Action 2: Skip Order Administrative Lead Time after Actual Production
Also in April 2010, the booking pattern to complete an order after the production finishes is also changed. Now operators have to do the booking registration by themselves after they complete the production processes (automatic documentation). Therefore, products do not have to wait in front of the inventory office for booking. Then the original administrative lead time after process end can be deducted, which is expected to lower the lead time further by 8 hours. The deduction of documentation hours will also increase OTR in SMU. Based on the 2446 orders selected from Quarter 1 2010, the actual OTR is 63.45% by comparing the “gereed” to the due date. However, OTR can increase to 74% if we skip the time between “process end” (log tijd) and “order complete” (gereed) but only compare the “log tijd” with the due date. (see Figure 2.4)
Figure 2.4 OTR Improvement by Deducting Order Administrative Lead Time after Actual Production
13
This is the average operation time for orders which do not have PCO as the last step. It is estimated that the average process time for PCO is around 20 hours. Hence, the operation time for all orders (including PCO as the last step) should be roughly around 48 hours.
14
Action 3: Automatic Nesting by DMM
Starting from June 2010, EGS launches the software “Dynamic Machine Management” (DMM) in SMU. With the help of DMM, SMU will be able to do the nesting automatically by the software instead of manually by the nester. The software will do the nesting right before punching, so with DMM, the nesting will be done by the operators on the shop floor15. DMM will range nests (among orders of certain time interval, time interval can be customized) to max material utilization. It will first nest large parts and then fill the margin of sheet with small nests. It assigns different priorities to orders, urgent orders, split orders, normal orders and ect. for operations. With DMM, the nesting will take only a couple of minutes instead of the average 8 hours if it is done manually. DMM is currently only linked to the punching machine, operators can start punching according to the priority and the nest drawings shown in the screen right after the nesting is done. Hence, this will again save hours since operators do not need to wait for the nester to bring the nests to start punching. Because of DMM, there will be another 8 hour reduction expected between “release” and “process active”. All three actions mentioned above are related to administrative lead time reduction. It is always good to improve lead time performance. Hence, further lead time reduction can continue during DLD implementation. Later on, great efforts will be put on actual producing time reduction. The dynamic nesting from DMM will nest several orders simultaneously on one sheet. Then those orders actually share the same set up. Then the set up time for each order can be shortened. However, how the further set up reduction really works is not the focus within the period of this research of EGS. Hence, here we do not discuss those actions as part of the preconditions of DLD implementation. Concerning interoperation lead time reduction, it is linked with process complexity and capacity sufficiency. Thus, we will discuss it in the coming sections.
2.2 Process complexity
2.2.1 Low Process Complexity as Important Precondition
Low process complexity is another important precondition for DLD implementation readiness. First, complex process (variant steps and more machine involved) increases the exposure to machine breakdown or other capacity shocks. Then for products with complex processes, stock holding is more important to protect production against capacity shocks and to keep process independency than products with simple production steps. However, by implementing DLD, no inventory should be held. In other words, it may be less suitable to launch DLD if the producing is too complex. Secondly, complex processes are likely to introduce longer non value added interoperation waiting time during producing. For instance, the average waiting time between two operations in SMU is around 4 hours. Apparently, the more operations are needed, the longer non value added interoperation waiting time is expected. Longer total lead time also implies that assembly line has to provide fixed schedules for longer period (the fourth precondition). The longer period, the more difficult for assembly line to guarantee. With more potential problems of capacity adapting and lead time control, high process complexity may decrease dependability performance by applying DLD.
2.2.2 Process Analysis
In 2009, Van den Berg did an in-depth study on producing processes in SMU. The current processes are similar to those one year ago. Hence, we use his findings to prove that the complexity of processes in SMU is suitable (low enough) for DLD implementation. He analyzes the routings in SMU: A From/To matrix for SMU (see Table 2.1) provides an overview the distribution of routings. The
15
main route is between punching and bending. Over 90% of orders start with punching and about 9% of orders finish with this process step. Over 60% of orders finish with bending. Together, there are about 70% of orders finish with either punching or bending. There are some product flows between Group A and B, however, the main product flows do not involve in processes from both groups. Processes in Group B take only a small portion in the whole EGS, since the demand of single parts (Group A products) is much higher than that of assemblies (Group B products). Among the 126 routings possible, there are 10 main fixed routings (2 in Group B and 8 in Group A) which take 80% of the capacity (booked production time), 90% of the total product quantity and 88% of order steps between processes (see Table 2.2). He also calculates that the mean number of process steps of one order is 2.3, where 55% of orders in Group A have only 2 process steps. In terms of process complexity (number of stations needed for production), it is low enough for SMU to be ready for DLD implementation.
* All proxies of machines used in Table 2.1has presented in Table 1.2. In order to ease reading. Here we present Table 1.2 as well.
Table 2.1 From/To Matrix SMU (Source: Van den Berg 2009, pp.22)
Table 1.2 Overview of Processes in SMU (Source: Van den Berg 2009, pp.21)
2.3 Capacity Adaptability
2.3.1 High Capacity Adaptability (Sufficient Capacity) as Important Precondition
In order to apply DLD, SMU should be able to maintain sufficient capacity by quickly adapting capacity requirements when facing frequently changing demands. Under the situation of DLD, there will be no opportunities to schedule orders forward to produce in advance and to hold pre-finished products in inventory. Hence, SMU should have the ease of capacity increasing and it should also be able to adjust its capacity rapidly to cope with order demands. Capacity sufficiency maintenance is also relevant to possibilities of machine breakdown and unexpected labour absenteeism. These possibilities in SMU also should be low.
2.3.2 Capacity Analysis
Figure 2.5 shows the producing area layout of SMU. There are six work centres shown in the figure. Table 2.3 gives capacity information for each work centre.
Figure 2.5 Layout of SMU Process Nr. of machines Nr. of operators No. of Shifts
Punching 3 6 2 Bending 7 9 2 Bench working 4~5 1 Welding 8 2 Welding robot 1 4~5 2 PCO 1 9 2
Idle Capacity (Capacity Increase Margin)
Currently, the usual capacity demand in SMU is around 800 to 900 hours per week, while the standard capacity available is around 1100 to 1200 hours per week. This indicates that it is easy to increase capacity by about 30% without huge capacity rearranging. In addition, as stated earlier, due to delivery pattern "via 003" EGS has actually produced more than what is actually needed on assembly line. Capacity has been wasted on this over producing, especially when the extra inventory becomes obsolescence. Processing costs are usually counted as one third of product value, so the roughly calculated processing costs for the total €356,000 obsolescence is €106,800. If machine costs per hour count as €73, the obsolescence has wasted about 1500 hour capacity. After implementing DLD, much lower obsolescence will be expected. In this way, DLD itself can save producing capacity. Meanwhile, according to the current machine and labour combination shown in Table 2.3, it is possible that one operator has to run several machines simultaneously (especially for bending machines). Machine may be idle since it has to wait for being operated on. This implies that labour increase is the key of capacity increase in SMU. With more operators, capacity can be increased greatly. In conclusion, there is extra capacity in SMU to prepare for potential capacity pressures that may be created by applying DLD.
Capacity Adapting Speed
The question of whether SMU can maintain sufficient capacity to apply DLD is also related to how fast it can adapt its capacity demand. On one hand, DLD may balance off order quantities among different time periods since with more volume flexibility assembly line does not need to order for far future. Then it is possible that SMU will receive orders with smaller order quantities each time with smaller differences. On the other hand, facing demand peaks, SMU can not prepare in advance. Hence, SMU has to be able to increase its capacity quickly within short time intervals. Talking about extra arrangement for capacity increase, first, it is easy to range extra shifts on Saturdays and Sundays, usually a 6 hour day shift on each day. Secondly, it is also possible to let punching machines run stand-alone (without operators) during nights. These two options are relatively easy to execute. Thirdly, sometimes it is also possible to borrow labour from other units. However, it may need coordination with other units in advance. Outsourcing can also be an option when EGS can have some control on the external producers to maintain its dependability. If there is continuous tendency of demand increase, EGS can consider hiring new people. However, the training usually takes half a year. Considering all options above, it should be easy and quick for SMU to adapt its capacity requirements. In fact, with SIOP concept (see Chapter 1), EGS has demand forecasts of the coming three months every month from sales. Though with DLD it can not produce in advance, it indeed has long time interval (three month) to prepare for the capacity requirements.
Risks of Unexpected Machine Breakdown and Labour Absenteeism
Operator Multiple Skills and Machine Substitutability
When unexpected capacity shocks occur, SMU will have more protection if operators have multiple skills and machine substitutability is high. There is a cross functional training program to develop multiple functional employees (see Appendix D for skill matrix in SMU). Hence, it is possible for punching operators to operate bending machines or for bending operators to do bench work as well. The welding operators can bend or do benchworking while the bench workers and bending operators can not weld. Each welder has their own specialties for different kinds of products and they manage the concrete task division by themselves. As shown in Table 2.3, there are several similar machines for each process except the welding robot. The substitutability among machines belong to same process is high. For instance, 95% punching tasks can be done on any of the three punching machines. The seven bending machines perfectly substitute each other. SMU is capable to switch tasks to different operators and machines if unexpected shocks happen.
2.4 Fixing Assembly Schedule
2.4.1 Assembly Line Fixing Assembly Schedules within Certain Time Length as Important Precondition
The success of launching DLD also depends on the cooperation from assembly line. EGS needs a fixed assembly schedule (production orders) within a certain time length. This means assembly line has to provide correct order requirements and does not change them within that period, usually at least for five working days (current average lead time in EGS to produce an order). If assembly line changes technical requirements during that certain time length, it is possible that EGS has already started production with different technical design. Without warehouse available for inventory holding, the incorrect products have to pipe up on assembly line. An other example would be if assembly line puts forward or postpone the due date suddenly (within that period), dependability it receives may be lower that delivery will be either too late or too earlier compared to the new due date.
2.4.2 Possibility of Fixing Assembly Schedules
Asking for fixed assembly schedules indeed puts some extra constrains on assembly line. However, this is a basic requirement for customer to order. No matter from which suppliers assembly line orders, it should not change orders after production begins or ask for due dates which are not sufficient to complete production. This does not mean that EGS will scarify customer flexibility to make its own life easier. As departments within a same company, two departments have to cooperate to serve final customers better. Assembly line has to promise correct demand information for at least a short time. It is still able to adjust its demand but just has to inform EGS early enough. Furthermore, different articles are allowed to have different time length for schedule fixing, which can be negotiated between assembly line and EGS in advance based on the final customer demand patterns. In conclusion, it is possible for assembly line to fix its assembly schedule for a certain period for EGS to be ready to apply DLD. This precondition for each specific article can be addressed in more details during the discussion for DLD implementing decision making (see Chapter 4 and 5). 2.5 Chapter Summary
3. Performance Changes due to DLD Implementation
The essential reason to implement DLD is to improve the performance in SMU. Referring to the performance of SMU, we use the five indicators of Slack and Lewis (2007): quality, speed, flexibility, dependability and costs. In this chapter, we will discuss the potential performance changes due to DLD implementation through the five aspects. The potential changes of these five performance objectives will then influence the DLD implementing decision (discuss more in Chapter 4). This chapter will provide answers to the second sub research question: “What are the potential performance changes if DLD is launched?” Figure 3.1 shows the role of this chapter (marked in red) in the conceptual model.
Figure 3.1 Theme Illustration of Chapter 3 This chapter will consist of five sections to discuss the five performance indicators respectively. Main conclusions will be written in italics. At the end, a brief conclusion section will readdress the key conclusions of this chapter and motivate how findings in this chapter associate with the later chapters.
3.1 Quality
feedbacks and quality corrections. With DLD, (smaller) order quantities are expected to be used on assembly line in short time. Hence, if there are quality problems, it will be easier and quicker to track down where the problems have occurred. Then the corrections will also be easier and in time.
3.2 Speed
In DLD implementation preparation phase, we have already listed several actions to lower the administrative lead time. The interoperation lead time should not increase if the number of routing is small and the capacity is sufficient. It may also reduce due to smaller order sizes. The cycle lead time is expected to decrease significantly due to smaller order (producing) quantities expected under DLD. However, there are more set up times expected. For some articles, ordering and producing the exact demand quantities online each time may increase total set up time more than the total cycle time reduction. These articles then may not suit to implement DLD. Though DLD may also expose SMU to more risks of demand peaks, lead time congestion will not occur if the capacity can adjust to be sufficient quickly. In conclusion, we should consider the trade offs between set up time increase and cycle time reduction to select suitable articles to switch to DLD. For those articles selected, the lead time is expected to decrease by implementing DLD.
3.3 Flexibility
The performance indicator flexibility can be further divided into product flexibility, mix flexibility and demand flexibility (Slack et al., 2007). Product flexibility means the ease by which new products can be introduced. Mix flexibility means the ease by which a firm can vary the scope of products. Volume flexibility means the ease by which the production volume can increase or decrease. Regarding to the product and mix flexibility, there may be not many changes. Apparently, assembly line will have more volume flexibility to order the exact amount needed each time instead of ordering based on the economic order size of each article. However, DLD indeed reduces freedom for assembly line to change orders. As one important precondition, assembly line has to fix the assembly schedule for a certain time length for EGS' production. Comparing to the current situation, it will be impossible to change due dates, demand quantity or other ordering information during that certain period (usually after the production in EGS begins). Otherwise, it will reduce the dependability of EGS and cause problems for assembly line itself, either delivery delays or early delays (piping up additional online stock). However, from the perspective of the whole company, this less freedom is actually a benefit since it forces assembly line to organize better to demand more accurately. After all, all departments should cooperate to serve customers better, instead of loading all pressure to EGS.
3.4 Dependability
3.5 Costs
In short term, DLD may not show the effect of lowering the product prices charged on customers. This is because the practice of price setting in Company E: prices of products from SMU are set once a year in September. Price is set based on a standard order quantity or on the economic producing size for each article. The quantity used for price setting must not be lower than 10 pieces. If the actual order quantity is lower than the quantity for price setting, there will be a negative production result for EGS showing in its financial reports, vice versa. Hence, after implementing DLD, EGS is expected to increase its negative production results due to more smaller quantities expected. The increasing negative production results are the major costs of implementing DLD. Nevertheless, this will be reduced in long term if set up time is further reduced and price is set based on more realistic quantity estimation.
On the other hand, DLD is supposed to reduce the actual manufacturing costs of EGS. Manufacturing costs usually include machine costs, labour costs, material costs, and inventory costs. According to full cost account method (FCA), the price setting in Company E’s Baan system follows the practice: integrate direct costs (labour costs and machine costs) and indirect costs (overheads and inventory costs) into process costs. Then price includes two parts: material costs and process costs. DLD implementation will affect two types of manufacturing costs: machine costs and inventory costs. The possible machine costs increase due to more set up times (smaller order quantities) has already been presented in the increasing negative production results. Inventory costs will be reduced greatly, since DLD will no longer generate new stocks. Inventory cost reduction is the major source of DLD benefits.
3.6 Chapter Summary
In this chapter, we discuss potential changes of five performance objectives (see Table 3.1). Objectives Effects
Quality Less scrap expected;
Less exposure to damage and obsolescence; Easier and quicker quality control.
Speed Shorter cycling time; More set up times.
Flexibility Increased volume flexibility;
Less freedom to change orders after production starts.
Dependability No lower dependability if two preconditions are reached (high capacity adaptability and fixed assembly schedule during certain time interval)
Costs Increased negative production results; Much lower inventory costs.
which articles are more suitable for DLD based on these performance changes16. However, there can be apparent trade off between positive and negative lead time changes. The same story also applies to cost performance changes. Different articles will show pronounced different trade off value on these two performance objectives. Therefore, it will be more powerful and meaningful to compare changes of these two performance objectives to distinguish different articles for implementing DLD. These two performance objectives actually closely link to each other. More set up times are the reason of machine cost increase and negative production result increase. We will make article selection based on cost performance changes based on a financial benefits and cost comparison model. This selection will then be discussed by a DLD project team (including preventatives from departments relevant to implementation) to make sure that there will not be negative impacts on the other four performance objectives (mainly dependability) due to DLD. The comparison model and the detailed implementing plan making process will be presented in the coming chapters.
16
4. DLD Decision Support Framework: Benefit & Cost Model
As discussed above, there may be some trade offs among the performance changes of implementing DLD. This indicates that not all articles are suitable to switch their delivery pattern to DLD to improve performance of SMU eventually. Hence, a DLD decision support framework is required to assistant the candidate article selection to implement DLD. Though some selecting logic has already been mentioned briefly in Chapter 3, we will build up a more concrete model for decision making in this chapter. In addition, we will present a clear scheme that how the model should be used. The scheme answers questions such as “when will the model involve in?”, “who should use the model?” and “What actions should be taken for implementation?” This chapter will answer the sub research question “How to judge which articles should switch their delivery pattern to DLD?” Figure 4.1 presents the theme of this chapter in the conceptual model.
Figure 4.1 Theme Illustration of Chapter 4 First, we will explain why we build up the decision support framework mainly based on the financial result changes. Then we will present the scheme of how the model should involve in the implementing decision making process. After that, we will also briefly introduce how the model is developed. Thereafter, the formulation of the framework will be presented. At the end of this chapter, the model logic and all matters need attention for using the model will be mentioned once more.
