A C
LEAN
SLATE
Reinventing Production Planning and Control at LRP
PREFACE
PREFACE
This thesis finalizes the course program of the Master Technology Management at the University of Groningen. I have started to follow this program one-and-a-half year ago. The program offered challenging courses within the field of operations, innovation management and strategy, accounting, and systems. The field of operations, in particular lean manufacturing, aroused my interest. My interest was fulfilled by the lean operations research center (LO-RC). The LO-RC offered me a research project at Lankhorst Recycling Products (LRP). In this preface I wish to thank those who enabled me to complete my research project.
First, I would like to thank my supervisors who have supported me during the research project. My first supervisor, prof. dr. ir. Jannes Slomp, came up with the idea of capacity oriented planning at LRP. His feedback regularly put me back on track when I lost focus. My supervisor at the firm, Dirk
Baarda, introduced me to the organization and the production department in which the case was conducted. His practical knowledge and feedback enabled me to see the case from a different angle. I also wish to thank the co-assessor of my project, dr. Linda Zhang, for her useful comments on my master thesis.
Further, I wish to express my gratitude to the people at LRP who allowed me to get familiar with their daily work and to ask questions whenever I wanted. First, I wish to thank Marcel Janssen, production manager of LRP, for his help in interpreting the data from the spreadsheet ‘mania’ and pleasant work environment; and Peter de Jong, assistant production manager, who helped me in unraveling the technical complexity of production equipment. Also, I wish to thank Janneke Hiemstra, for the interviews and discussions we had about planning. Finally, this research would not have been completed without the moral support of the factory workers.
ABSTRACT
ABSTRACT
This paper presents a case study at Lankhorst Recycling Products (LRP). LRP is a manufacturer of finished goods from recycled plastics. The firm produces more than 200 stock keeping items, using a centrally located mixture facility and eight process oriented product lines. The production environment of LRP can be typified as process manufacturing.
Problem situation. At LRP, there is uncertainty whether capacity suffices to meet the increase in demand. The structural increase in demand causes high utilization of capacity. Because of high capacity utilization, the firm struggles with a huge order backlog. The backlog complicates the prediction of an order’s cycle time. The cycle time varies because orders in the backlog are
rescheduled or delayed because of outages. Therefore, predicted cycle times frequently exceed allotted lead times, resulting in bad delivery performance. Delivery performance is also a problem in inventory control. Bad control causes a divergence of inventory levels. Inventory levels are extremely high or critically low. Critically low inventory levels frequently result in stockouts.
Research approach and goal. The problem situation revolves around the areas of operation strategy and production planning and control (PPC). Low delivery performance results from current PPC. Decisions in PPC are made to match available capacity with demand in the future (planning) and in the presence (control). The presence of a rigid capacity and uncertain demand complicates the decision making. In order to make the right decisions PPC requires support. At LRP, there is little support; there are no clearly stated objectives or standards. In other words, there is no transparency in PPC. Therefore management is looking for methods which bring back transparency and, above all, improve the delivery performance. The purpose of this case study research is to develop a structured method for PPC. The method must handle available capacity more efficiently and improve LRP’s delivery performance.
Analysis. Production manager is not supported by a documented operations strategy. Management seems to pursue multiple universal manufacturing objectives at same time without assigning relative priority to them. Therefore, production lacks direction which makes it difficult to make tradeoffs to resolve conflicts. Conflicts also occur in PPC. There is no consistent response to customer requests. Salespersons do not know whether customer require immediate delivery or not.
ABSTRACT
requests. Replenishment orders are generated after the weekly inventory check. Released production orders in the queue are arranged with scheduling rules. The release sequence frequently changes. It is difficult to evaluate performance with current measures. Both throughput and cycle time
performance lie between the performance of the worst-case and practical worst-case, meaning that the line is performing ‘bad’. The causes of ‘bad’ performance are explained by queuing and batch
production. Low throughput performance is caused by random outages, setups, and rework. Delivery performance is not measured at LRP because there is no consistent response to customer requests. Therefore, it is difficult to choose between the fill rate and service level.
Diagnosis. Orders have to be delivered on-time to achieve acceptable delivery performance. The cycle time of an order consist of two major components: queue time and process time. The queue time is the time an order waits for a free spot on the rotator. Fortunately, queue time only exists in the backlog. But the queue time is highly variable due to changes in the release sequence and length of the planning horizon, especially in the presence of a huge order back log. The presence of the backlog force
salespersons to plan far into the future. Therefore, salespersons anticipate long outages in determining effective throughput. This results in a lower effective throughput. The lower effective throughput inflates the cycle time of an order. Because of the inflated cycle time, lead times are too long and orders are frequently finished unpredictably early.
Conclusions. A customer driven operations strategy is required to support PPC in making trade-offs to resolve conflicts and support salespersons to consistently respond to customer requests. Poor
CONTENTS
CONTENTS
1 Introduction ... 7
1.1 Problem situation... 7
1.2 Lankhorst Recycling Products ... 7
1.3 Thesis structure ... 8 2 Research Design ... 8 2.1 Area of concern ... 9 2.2 Framework of ideas ... 9 2.3 Methodology ... 13 3 Operations strategy ... 15 3.1 Business strategy ... 15 3.2 Operations strategy ... 16 3.3 Manufacturing strategy ... 18 4 Production system ... 19 4.1 Demand characteristics ... 21 4.2 Production characteristics ... 23
4.3 Production planning and control ... 25
5 Performance evaluation ... 28
5.1 Performance evaluation at LRP ... 28
5.2 Throughput, cycle time, and WIP ... 29
5.3 Delivery performance ... 31
6 Diagnosis ... 31
7 Concepts of Redesign ... 33
7.1 Capacity oriented approach ... 33
7.2 Cyclical production ... 35 7.3 Inventory control ... 36 8 Improvement Policies ... 37 8.1 Planning framework ... 38 8.2 Inventory items ... 39 8.3 (T, kj, Sj)-strategy ... 40
8.4 The production sequence of families... 41
8.5 An example for Palen 9 ... 42
8.6 Evaluation of cyclical production ... 43
9 Conclusions ... 45
10 Recommendations ... 46
11 Bibliography ... 48
12 Appendices ... 49
Appendix A Product Market Combinations ... 49
Appendix B Product families Palen 9 ... 50
Appendix C Demand product families ... 51
CONTENTS
Appendix E Internal benchmarking ... 53
Appendix F Atkins & Iyogun’s (1988) procedure ... 55
Appendix G Order-up-to level procedure ... 58
INTRODUCTION
1
INTRODUCTION
This thesis presents a case study at Lankhorst Recycling Products B.V.. Lankhorst Recycling Products (LRP) is part of the Lankhorst Engineered Products (LEP) division of the Royal Lankhorst Euronete Group.
1.1
Problem situation
The problem was initially presented to the author as a mandate to reduce finished goods inventory and cycle times, but pretty soon it became evident that production planning and control (PPC) issues needed to be addressed. Gross of LRP’s customers order irregularly and in large quantities, creating lumpy demand and a planning nightmare. Planners, therefore, do not believe that forecasts help them to produce ahead of demand. Rather, they react to demand as it occurs by initiating production orders in response to customer requests. The usual response is to build up inventory and manipulate lead times. Inventory levels, however, are set without proper methods and lead times are allotted in the presence of variable cycle times. The system worked well for many years, but the structural increase in demand causes high utilization of capacity. Because of high utilization, orders have to wait for
available capacity. Order wait on paper, in a huge order backlog. The presence of the backlog complicates the prediction of an order’s cycle time. The cycle time varies because orders in the backlog are rescheduled or delayed because of outages. Therefore, predicted cycle times frequently exceed allotted lead times, resulting in bad delivery performance. Delivery performance is also a problem in inventory control. Bad control causes a divergence of inventory levels. Inventory levels are extremely high or critically low. Critically low inventory levels frequently result in stockouts.
1.2
Lankhorst Recycling Products
LRP is a manufacturer of finished goods from recycled plastics. The firm manufactures more than 200 inventory items, using a centrally located mixture facility and eight process oriented production lines. Production operates in five shifts, seven days a week, twenty-four hours per day. The manufactured items include poles and planks but also complete landings, picnic tables, pile sheeting, and support beams. The items are distributed to dealers and end users in both private and public sector. LRP has a sales office in Birkenhead (UK) and its own sales department in Sneek. The sales department contacts the customer and agrees upon the delivery of the products. PPC is also conducted at the sales
RESEARCH DESIGN
Figure 1-1: Material flow
The material flow starts at the receipt of the raw materials. Raw materials are supplied in big bags or bulk. Big bags are blended into different mixtures. Blending is conducted in a single facility which is a collection of several machines such as dryers and containers to blend the recycled plastic and remove undesirable material from the mixtures. The mixtures are stored in silos (RMI). Pipelines aspirate the required mixture from the silos to production. In production, items are processed. Quality approved items are moved from production to a storage depot (FGI). Gross of the items are directly expedited. Some of them are assembled or upgraded at Empatec. These parts are stored on a prespecified location on the terrain (FGI). The shipment of end items is the responsibility of expedition. Expedition workers pick orders on the storage depot and prepare them for shipment.
1.3
Thesis structure
Chapter 2 starts with a description of the area of concern. The problems are discussed in more detail and the research goal and question are defined. A framework of ideas is build to formulate sub questions. The sub questions are answered in Chapters 3 through 5. Chapter 3 deals with the operations strategy of the firm. Chapter 4 discusses the way PPC deals with demand and production characteristics. In chapter 5, the performance of the production system is evaluated. The final diagnosis is made in Chapter 6. In Chapter 7, the concepts of redesign are introduced for the
development of improvement policies. Implementable improvement policies are discussed in Chapter 8. The conclusions in Chapter 9 answer the research question. Finally, Chapter 10 sums up
recommendations for a successful implementation of improvement policies.
2
RESEARCH DESIGN
RESEARCH DESIGN
2.1
Area of concern
The problem situation revolves around the areas of operation strategy and PPC. Surely, changing current resource configuration by investing in production equipment or labor increase capacity but requires extensive amounts of capital. Capital has just been invested in labor by lengthening the workweek from 5 to 7 days. Therefore, management is looking for alternative ways to solve the problems. Both problems (i.e. variable cycle times and stockouts) affect the delivery performance. Low delivery performance results from current PPC. Decisions in PPC are made to match available capacity with demand in the future (planning) and in the presence (control). The presence of a rigid capacity and uncertain demand complicates the decision making. In order to make the right decisions PPC requires support. One form of support is a documented operation strategy. The objectives in the operation strategy help PPC to make trade-offs to resolve conflicts in production. For instance, cost and responsiveness are conflicting objectives in manufacturing. High capacity utilization is needed to keep units costs down, but low utilization for good responsiveness. Another form of support is determination of standards. The presence of standards simplifies PPC decision making. For instance, proper inventory control replenishes inventory at the right moment within the anticipated lead time. The timing of those moments can be fixed by standards. The effect PPC decision making can be monitored by performance evaluation. Performance evaluation offers management the key insight whether production is operating efficiently or not. At LRP, there is little support for PPC decision making. There are no clearly stated objectives or standards. In other words, there is no transparency in PPC.
Evidently, problems result from the lack of transparency of current PPC. Therefore, management is looking for methods which bring back transparency and, above all, improve the delivery performance. This allows us to disintegrate the problem situation into a research goal; why is the research carried
out?; and the research question; what has to be examined to reach the research goal?
Research Goal
To develop a structured method for production planning and control. The method must handle available capacity more efficiently and improve LRP’s delivery performance.
Research Question
How can LRP improve delivery performance by changing the current way of production planning and control?
2.2
Framework of ideas
The definition of the research goal and question enables us to build a framework of ideas. A
RESEARCH DESIGN
formulating sub questions1. Underneath each sub question, there is a brief explanation of: the
relevance of the question, the approach used to answer the question and the required theory and models.
Figure 2-1: Conceptual model
1. What is the impact of LRP’s operations strategy on decision making in production?
Production makes decisions in order to achieve an objective. The objective is defined in the operations strategy. The operations strategy drives PPC and supports management in investment decisions about the factory’s internal resource configuration. Thus, the operations strategy has an effect on production characteristics → planning & control characteristics as a whole.
The operation strategy is a derivative of the overall business strategy. The business strategy determines LRP’s position in the industry and the way it responds to the surrounding competitive pressures. Unfortunately, LRP does not have a formulated business strategy. The strategic choices of LRP can be typified with Porter’s (1980) generic strategies. These strategies capture the economic forces at work in the market (demand side) and in the industry (supply side). The economic forces are assessed by an industry and market analysis. A short industry analysis is needed to identify possible ways to secure and improve LRP’s current position. A market analysis is performed to show how marketing interprets the dynamics of the demand side. The business strategy provides input for a customer driven operations strategy. A customer driven operations strategy can be portrayed by the performance objectives of Slack et al. (2004). The relative priority of quality, speed, dependability, flexibility, and cost enable us to identify the strategic direction within production. The strategic direction is then linked to production characteristics → planning & control characteristics by discussing manufacturing strategies. Manufacturing strategies address the priorities of processes devoted to manufacturing (Krajewski & Ritzman, 2002). In manufacturing process, there are three basic strategies: make-to-stock, assemble-to-order, and make-to-order. The manufacturing strategies determine the response to customer requests.
RESEARCH DESIGN
2. How does planning and control anticipate demand characteristics?
The decisions of PPC depend on the nature of demand. Salespersons manipulate inventory levels, order quantities, and lead times in an attempt to match supply and demand. Unfortunately, they cannot look in a crystal ball to determine exactly how many customers arrive, when they arrive, and what they will order. Instead, they make use of whatever demand information to make appropriate decisions. To that end, the demand characteristics → planning & control characteristics relation is examined to find out which information salespersons use to deal with demand.
A current state map is created to diagnose the firm’s door-to-door value stream. The value stream consists of both a material and information flow. Part of the information flow shows which information is used for PPC decision making. However, the concise information in the map is not enough to comprehensively characterize demand. A time series analysis is conducted to determine the demand characteristics. Historical sales data is used to identify patterns in demand. The demand patterns help to understand the behavior of demand. The second part of the analysis covers the factors that cause changes in demand.
In this study, demand for end items originates from outside the factory and concerns the demand for items that are sold directly to a customer. Some of these items are assembled or upgraded at Empatec. The information system generates separate orders for Empatec. These orders have to be removed from the dataset to avoid redundancy. Another demand issue is production waste. Foremen only register products with quality approval. Demand has to be corrected for waste to assure appropriate levels of supply. Note that demand is expressed in units. At LRP, sales are recorded in both kilos and units. This is because demand is not always a product of the order quantity and unit weight. Sales in weight are calculated back to units to obtain an accurate picture of demand.
D = corrected demand for products that are sold directly to a customer [end items] 3. How does planning and control deal with production characteristics?
PPC also has to consider the limitations of supply. Supply is not infinite. Orders cannot be released on to factory floor without acknowledging production constraints. In automated production lines
machines are the primary constraint. The machines are capable of producing at a limited rate, depending on different variables (not defined here). Together, these variables are referred to as production characteristics. PPC have to anticipate for these production characteristics in decision making.
RESEARCH DESIGN
1.2, we now zoom in on the part between the raw material inventory and finished goods inventory to determine the production characteristics. Production processes between these stock points are illustrated by process boxes in a current state map. Warning triangles are used to indicate
accumulation of products. Empirical data is collected to fill the process boxes and fill the blank spaces underneath the warning triangles in the map. The following data is collected.
CTp = average time a part is being worked on at a process [seconds]
THp = average throughput given by the output rate from a process [parts per hour]
WIP
= average work in process between the start and end of the product routing [parts] FGI
= average finished goods inventory at the end of the product routing [end items] RMI = average raw material inventory at the start of the product routing [parts] 4. What is the impact of the production system on performance?Production evaluates production performance in terms of measures to determine whether the
objectives are achieved efficiently or not. Because there is lot of uncertainty whether capacity suffices to meet the increase in demand, LRP limits performance evaluation to throughput. Evaluating
throughput performance offers management the key insight in how production is doing. Throughput is also used in PPC. PPC uses effective throughput to plan an order’s lead time. Clearly, throughput is an important measure, but it is not comprehensive. Too lithe throughput means losing sales, too much means building up excess inventory. So, more efficiency measures are needed to evaluate production. Therefore, the relation between production characteristics → planning and control characteristics → production performance is analyzed to identify appropriate measures for LRP.
First, the current way of performance evaluation is analyzed. Thereafter, data in the current state map is used to diagnose performance. The performance is diagnosed with the three cases of Hopp and Spearman (2000). The performance of the best- and worst-case enables us to bracket actual performance. The performance of the practical-worst-case is intermediate. Actual performance is compared to the performance of the practical worst-case to determine whether the line is performing ‘good’ or ‘bad’. If production is diagnosed ‘bad’, there are still ways to increase effective throughput and lower the utilization of capacity. Two key parameters are required for diagnosing the line: the bottleneck rate and the raw process time. Additional empirical data from the current state map is used to compute performance of the cases. The operational definitions of the bottleneck rate and raw process time are:
RESEARCH DESIGN
Only throughput, cycle time, and WIP are used to compute the performance of the three cases. To penalize throughput shortages, delivery performance measures are also considered.
2.3
Methodology
The third element in the research design is the methodology. The research methodology marshals the research instruments, used to answer the sub questions. Here, we choose to follow a case research methodology. The case research methodology consists of five stages (Stuart et al., 2002): define the research question (1), develop research instruments (2), data gathering (3), analyze data (4), and disseminate (5). Note that stage 1 and 2 were already conducted in paragraphs 2.1 and 2.2.
Before conducting stage 3 and 4, we first have to select a suitable case. Unfortunately, it is not feasible to examine all production lines within the time frame of the research project. After communicating with the LRP management, agreement was reached on the case. The examined case is an automated production line, Palen 9. Palen 9 has the highest throughput of all the lines within the factory. The throughput of the line comprises one-third of total factory throughput. Data is gathered from Palen 9 from both qualitative and quantitative sources over an extended period of time. The sources comprise interviews, business records, internal company documents, and observation. The relevant sources to answer the sub question are discussed below2:
1. Unfortunately, the operations strategy isn’t documented. Unstructured interviews with managers, salespersons, and production workers are conducted to collect the data for the first sub question. The interview starts with a participant narrative in which the purpose, the subject, and the context are briefly summarized. The questions are linked to associated theory. The participant’s answers are noted down for later analysis.
2. LRP has an extensive amount of business records with current and historical sales data. The analysis of these business records constitutes a large part of the data collection for the second sub question. The purpose of the analysis is to identify valid and understandable demand patterns in the records. Weekly work schedules reveal part of the working methods at planning. Additional interviews are needed to find out the exact mechanics behind the decision making.
3. A brief guided tour through the factory suffices to draw a current state map. The tour starts where the finished products are shipped to the customer and finishes where raw material is received. A first draft of the map is drawn during the tour on the factory floor. Additional information is noted and recorded. The information available from observation suffices to capture the required data. 4. Technical documentation of equipment, business records, along with a time/motion analysis of
manufacturing processes is needed for the fourth research question. Production workers are observed while performing their task. In this study, the production workers are directly observed.
RESEARCH DESIGN
The observer is physically present and personally monitors what takes place. The presence of the observer is known and enters the social setting on the work floor as both an observer and a
participant. To cover the risk of potential bias from participant awareness, building trust is vital for success. A brief data collection plan is developed for the time/motion analysis of the
manufacturing processes. It answers the question who, what, when, how and where (Cooper & Schindler, 2006):
a. Who? Machine operators should have sufficient skill and experience to perform tasks at the examined production line. Production management of LRP helps with introducing the observer to the foremen and operators and fulfills the ethical responsibilities, if needed.
b. What? The behavior of the Palen 9, an automated production line. Samples of the elapsed process times are drawn once during each shift randomly over a prolonged period to determine the average cycle time for a product. The throughput of the line and utilization of the
workstations are derived from the firm’s business records.
c. When? The number of shifts does not vary. There are three shifts per day. Samples are drawn during the workweek.
d. How? The data is recorded and noted down for later analysis. The observation does not stop when detractors, such as machine breakdowns, occur.
e. Where? The observation takes place in the LRP factory.
OPERATIONS STRATEGY
3
OPERATIONS STRATEGY
This chapter answers the first sub question. Several analyses are performed to determine the impact of the operations strategy on the production system (i.e. production characteristics → planning and control characteristics). First, the firm’s business strategy is examined. Porter’s (1980) generic strategies capture the firm’s distinct actions in response to dynamics in the industry (supply) and demand (market). These dynamics are assessed by a short industry and market analysis. The business strategy provides input for the operations strategy. The performance objectives of Slack et al. (2004) help to understand LRP’s operations strategy. Finally, manufacturing strategies are discussed to address the relative priority of performance objectives for manufacturing processes.
3.1
Business strategy
Unfortunately, LRP does not have a documented business strategy. However, Porter’s (1980) generic strategies can help to understand the strategic choices the firm makes. The generic strategies have two dimensions: product characteristics and market scope. The product characteristics can improve the firm’s competitive position in two ways. A firm might manufacture commodity products to enjoy cost advantages or create differentiated products that its rivals find hard to imitate. The other dimension revolves around the notion of scope. Scope becomes strategically interesting when economies of scope arise. Economies of scope arise when knowledge or other indivisible resource can be applied in multiple directions without using up that resource (McGee et al., 2005). Firms choose to operate at a broad or narrow scope.
LRP follows a differentiation strategy. During the years, LRP has developed extensive knowledge to differentiate a wide range of products in multiple directions (street furniture, stables, ground
reinforcement). To pursue the differentiation strategy, marketing captures an accurate picture of the market to ensure that there are sufficient ways in which to differentiate the product, and that the marketplace can be subdivided- and is willing to pay for the differentiation. Research and
development develops products and tests new compounds to enhance quality and reduce the unit costs in an attempt to avoid imitation.
OPERATIONS STRATEGY
LRP experiences not much rivalry in its industry. There are few other manufacturers (e.g. Gampet Plastics and Fiberplast) of finished goods from recycled plastics or equivalents. Because demand is increasing, there is more revenue to go around. The threat of entry is low in the firm’s industry. New entrants have to be able to manufacture and distribute products on a large scale. Start-up costs are high and it is difficult to gain access to suppliers of recycled plastics. There are a few suppliers of recycled plastics. Purchasers have little leverage over the suppliers to negotiate over terms. Lankhorst tries to vertically integrate (i.e. produce its own supplies) to lessen supplier bargaining power. LRP is highly differentiated and, therefore, does not rely on a few customers. This lowers the customer’s bargaining power. The threat of substitutes is high. The value of KLP® products is similar to that of its concrete of wooden equivalents. These equivalents are less expensive. Customers, however, value the extra attributes (durability, low maintenance) of KLP® products and are willing to pay for it.
Market analysis. LRP uses experience to divide the firm’s customers into market segments (see Appendix A). The customers in the market segments have enough in common to warrant the provision of products that the largest group wants. The segments are linked to article groups to establish product market combinations (PMC’s). The potential in each PMC is not assessed. This can be done by asking questions such as: What needs does the product satisfy?; By what criteria does the customer choose?; What motivates them? Answers to these questions enable marketing to identify customer
requirements, including both tangible and intangible attributes of the product. Patterns of customer requirements can then be linked to product attributes to differentiate from the competitors and improve the competitive strength. Marketing does keep track of customer satisfaction. Surveys are carried out to determine the customer’s satisfaction of product quality, communication, sales, service, and logistics. Although these variables might be interpreted as product attributes, they are not specifically linked to PMC’s.
3.2
Operations strategy
The production manager at LRP is not supported by a formulated operations strategy. This makes it hard to resolve conflicts. For instance, LRP wants high inventory for fast response, but low inventory for keeping totals assets low so that return on investment will be high. More variability is needed for greater product variety, but less to keep inventory low and throughput high. The production manager has to make trade-offs to resolve such conflicts. A customer driven operations strategy supports the manager in making trade-offs by assigning appropriate priority of performance objectives to
OPERATIONS STRATEGY
Quality. To understand quality at LRP, it is useful to draw a distinction between external and internal quality. External quality refers to the customers’ perception of product quality. LRP monitors external quality via direct and indirect measures. Salespersons keep track of customer complaints and
marketing conducts customer surveys. Internal quality refers to conformance to the design
specifications inside the plant. Internal quality at LRP is monitored through direct product measures such as weight, color, and measurements and indirect process measures such as extruder pressure. LRP tries to translate customer concerns to measures and controls for internal quality to achieve high external quality. So far, the firm does a good job. A recent survey pointed out that customers are satisfied and quality is perceived to be high. However, salespersons noticed a shift in customer requirements towards higher quality. LRP needs to develop new ways to keep perceived quality high. Thus, the quality objective is becoming more important in contributing to the firm’s overall strategy.
Speed. A speedy delivery enhances the firm’s offering. “The faster customers can have the product,
the more likely they are likely to buy it, or the more they will pay for it, or the greater the benefit they receive” (Slack et al., 2004). A few years ago, salespersons persisted in allotting short lead times to
win a customer order. Now, salespersons have more leverage over the customer. Decreased bargaining power of customers enables salespersons to negotiate longer lead times. The recent survey also
showed that customers value reliability over speed in product delivery. Speed is less important in fulfilling customer requirements. However, internally, speed is still an important objective. When parts travel faster through the factory, fewer inventories are needed to achieve the same throughput level. Also, forecasts are more reliable because the determined order quantities are intended to anticipate demand less far into the future. Speed supports low inventory and accurate forecasting. Thus, the priority of speed shifts to the production environment to enhance PPC.
Dependability. At LRP, dependability overrides all order criteria. No matter how fast a product is delivered, if the product is always late or (or unpredictably early) customers bring their business elsewhere. Salespersons at LRP try to define adequate inventory levels and quote reliable lead times for high delivery performance. Sometimes, they deliberately quote longer lead times to ensure on-time delivery. Dependability has a similar effect inside production. Foremen depend on the reliability of other processes in delivering the equipment on time. For example, if a mold is repaired by technical services, machine operators expect the mold to be available at the time set in the work schedule. When it is not, expensive capacity is lost. So, the dependability objective is of crucial importance.
OPERATIONS STRATEGY
quantity is substantial. Minor changes to the product are outsourced to Empatec where standard products are cut to customer specific size. The volume flexibility at LRP varies with the utilization of production equipment. High capacity utilization makes it difficult to quickly fill a customer order. Customers seem to accept lowered flexibility as long as the order is filled on-time. Therefore, the priority of flexibility in manufacturing is less important.
Cost. LRP does not directly compete on cost. KLP® products are more expensive but there some exceptions depending on the size and application of the product. Customers tend to favor KLP® products because of their durability, maintainability, and low weight. Despite the apparent low priority of cost, LRP still wants to reduce costs to maximize profitability. Capacity utilization is high to lower the direct unit costs. Inventory is reduced to lower holding costs. Holding costs include rent, utilities, insurance, employee costs and opportunity costs. Removing a euro from those costs means adding a euro to the firm’s profit. For LRP, cost reduction is a universal objective of moderate importance. The polar representation is a useful way to present the relative importance of performance objectives in manufacturing. In Figure 3-1, each scale represents the importance of each performance objective. The blue lines describe the relative importance of each objective. The closer the line is to the origin, the less important the performance objective is to the organization. Dependability is most important in manufacturing, followed by quality, cost, speed, and flexibility.
Figure 3-1: Polar representation
3.3
Manufacturing strategy
PRODUCTION SYSTEM
manufactured according a make-to-stock (MTS) strategy. A make-to-stock strategy is feasible for standardized products with high volumes and reasonable accurate forecasts (Krajewski & Ritzman, 2002). Irregular large orders are, therefore, manufactured according a make-to-order (MTO) strategy. In a make-to-stock strategy, customers allow LRP more time to manufacture and deliver the order. The extra time is referred to as the lead time. Salespersons have to allot reliable lead time to achieve acceptable delivery performance.
Resume
LRP uses knowledge as a resource to differentiate a wide range of products in multiple directions. Marketing ensures that there are sufficient ways in which to differentiate by subdividing the customers into market segments. The preferences of the customers in the segments, however, are unknown. A clear picture of the customer requirements in each market segment has to be captured to formulate a customer driven operations strategy. The operations strategy was portrayed by performance objectives. We saw that dependability overrides all other objectives at LRP followed by quality, cost, speed, and flexibility. To ensure high dependability, salespersons have to choose the right manufacturing strategy. Unfortunately, they do not know whether a customer requires immediate (MTS) delivery or not (MTO).
4
PRODUCTION SYSTEM
PRODUCTION SYSTEM ROP Supplier Daily call or fax Daily call Production Planning Customer
RMI WIP FGI
Injection Molding Cooling
Weekly Work Schedule
7 sec CTp
4355 sec CTp 18292 sec CTp 377 sec CTp 519 sec CTp
29,12 hrs WIBt 0,49 hrs Qt 0,49 hrs WIBt T0= 23550 sec CT = 55 hrs Daily shipping list 14158 parts 263 parts 18,41 hrs WIBt 1-2x Day 800-900 kg per big bag
22 big bags per freight
Last years sales
Order Picking
Prepare for delivery
Finishing 1 Packaging 1 Process Area Process Area Process Area Process Area Process Area Quality Checking 1 416 parts WIP CTp= 4355 sec THp= 14.3 units/hr Up= 0.85 17/20 positions CTp= 18292 sec THp= 14.3 units/hr Up= 0.80 12/15 drawers CTp= 377 sec 5 shifts 27.600 sec available CTp= 519 sec 5 shifts 27.600 sec available CTp= 7 sec 1 shift 3.600 sec available
Parallel processing High strategic value
Blending Receipt of raw materials 2 40 big bags 3 m = 0.5 hrs M = 16 hrs Yearly forecast 3-4 x Week
PRODUCTION SYSTEM
4.1
Demand characteristics
The first step in VSM is modeling customer requirements. In VSM, it is usual to start mapping the value stream of the family with the largest demand for end items. At LRP there are no product families. We have to aggregate items into families to determine the current state of the value stream. There are several forms of aggregation. Here, we choose two forms: production lines and setups. The first form of aggregation at LRP is to lump products with the preference for the same production line together. The preference is related to the differences between the lines. The lines differ in type, mixture, and mold length. There are three line types: manual, semiautomatic and automatic lines. Manual lines have low throughput but require little time for setups. Contrarily, semi-automatic and automatic lines have high throughput but require a lot of time for setups. As such, small orders are preferably processed on manual lines and large orders on semiautomatic or automatic lines. The use of the extruder in the production line is limited to certain mixtures. The extruder of our automatic line, Palen 9, is the only one capable of processing KLP® Geel. All items made of KLP® Geel are processed at Palen 9. The KLP® Rond 15 mixture is used for large poles. These large poles have preference for Palen 9 because of its extruder speed. The extruder speed injects the mixture at high speed into the mold, to avoid solidification of the melted mixture. All lines process KLP® Zwart. KLP® Zwart is a commonly used mixture. The mixture is used for standard items. Standard items are produced in high volumes which make it feasible to produce them on an automatic line. Finally, each line is constrained by the measurements of its water tank. Long molds do not fit in small tanks. Given these constraints, each line is assigned a number of items that ‘prefer’ that line. Appendix B lists the items with preference for Palen 9.
Aggregation by production lines results in large families. The size of the families can be reduced with another form of aggregation; setups. We distinguish between major and minor setups. Major setups occur when the production line is changed over to a different mixture. Minor setups occur when a single mold is replaced. If a changeover to another item requires a major setup, they are put into different families. If a changeover requires only a minor setup, the items are put in the same family. Appendix B lists three product families; 9-1, 9-2, and 9-3. The first number refers to the name of the production line. The second number indicates the family number.
According to Appendix C, family 9-2 is the largest. This family also serves the largest number of markets and is, therefore, suitable for our first current state map. The customer requirements, shown in the data box, are average demand and order size. But simply showing averages is not enough to characterize customer demand. A brief time series analysis is performed to comprehend the
PRODUCTION SYSTEM
than in 2005 and 2006. There are no trends and seasonal patterns leaping out the figure. However, the trend lines illustrate that demand systematically decreases every year. This is caused by the steep fall at the end of the year. Firms do not place orders at the end of the fiscal year in order to avoid
expensive inventories in their books. This fall can be interpreted as a seasonal pattern, as it occurs each year. To counteract the pattern, LRP can offer discounts at the end of the year. Demand also consistently increases in September. This increase is explained by the lower sales in the preceding months. In the months July and August, sales are lower because of the holidays.
0 2 4 6 8 10 12 14 De mand 0 2 4 6 8 10 12 14 De mand
Figure 4-2: Demand trends and seasonal patterns Figure 4-3: Cyclical patterns
Cyclical patterns concern the gradual increases or decreases in demand over multiple years (see Figure 4-3). They arise from two influences. The first is the business cycle, which includes factors that cause the economy to go from recession to expansion. The economy’s expansion has a positive effect on the demand for LRP’s items. Figure 4-3 shows a gradual increase in demand. But, the recent worldwide economic crisis has a negative effect on demand. The other influence is the product life cycle, which reflects the stages of demand from market introduction through decline. The majority of LRP’s products are in the mature stage. New products are developed to counteract the demand decrease of mature products that move to the decline stage.
PRODUCTION SYSTEM 0 2 4 6 8 10 12 1 2 3 4 5 6 7 8 9 10 11 12 De m a nd Month
KLP RECHTHOEK KLP ROND KLP SPECIALITEITEN
KLP STALLENBOUW KLP VIERKANT KLP WATERBOUW
Figure 4-4: Demand item groups 2007
KLP Rechthoek consists of a wide variety of planks and beams for different applications. Some of these plank and beams are distributed by foreign dealers or the UK sales office. Both distributors place large irregular orders to enjoy economies of scale. These orders cause peaks in demand. LRP can anticipate these peaks by sharing information. Sharing information enables the firm to stabilize demand.
The deck planks in KLP Waterbouw are suitable for pontoons and landings. As such, they are frequently applied in projects. Projects are accompanied by a large order quantity. But the order quantity is not needed at once. Salesperson and customer agree to divide the order over multiple shipments. Customers prefer multiple shipments to avoid expensive holding costs. Because the shipments are administrated separately, projects are difficult to trace. Traceability can be improved by classifying the order in the information system. Order classification helps LRP to get a more accurate picture of demand and estimate the effect of projects on demand.
4.2
Production characteristics
In the current state map, the material flow is illustrated by process boxes. Not all processes are considered in the door-to-door map. Drawing a box of each process would result in an
PRODUCTION SYSTEM
PRODUCTION SYSTEM
The processes between RMI and FGI are accompanied by a data box. The data boxes contain relevant process variables. The values of the variables in the data box are averages, which account for different types of end items. The first variable is the process cycle time. The process cycle time (CTp) represents the average time an item spends being worked on at a process, which includes time waiting for
detractors and some additional move time that will be necessary to move the products from the previous station. Major (M) and minor (m) setup times are recorded in the process box of injection molding. A typical mixture changeover at Palen 9 takes two shifts, with a lower limit of one shift and an upper limit of three shifts. A minor setup requires 15 to 45 minutes. In the process boxes of injection molding and cooling, there is also a need to distinguish throughput (THp) because the capacity of the station is not the reciprocal of the process time. The number of drawers and occupied mold positions indicate the WIP level in the process. Utilization Up and available working time are the other relevant process parameters (see Figure 4-1).
4.3
Production planning and control
The information flow adds a second dimension to the value stream. The flow is drawn from upper right corner to the upper left corner. The arrows in upper right corner illustrate the direction of
customer information towards PPC. PPC forwards information to purchase, expedition and production. PPC is divided into three basic functions: order acceptance, order generation, order release.
There are no rules for order acceptance. The absence of rules contributes to the increase of the order backlog. The huge backlog suggests that salespersons refuse to say ‘no’ to the customer. On the other hand, the customer always says ‘yes’, even if the promised lead time is substantial. Order acceptance rules enable PPC to put a cap on the backlog.
Two types of production orders are generated: replenishment and customer. Replenishment orders are generated after the inventory check. Every week, the inventory position of an item is checked against its minimum level in the Bull information system. The minimum level of an item is set to cover six weeks demand. The period of six weeks is a rule of thumb. It is not the average lead time of a replenishment order. A replenishment order is triggered when the inventory position is below the minimum level. The inventory is replenished up to the minimum level. Sometimes, the replenishment quantity is changed with an estimation of future demand. The estimation is made intuitively with historical sales data in Bull. The proximity of the estimation depends on the salesperson’s experience and knowledge of the market. Inexperienced staff members are likely to make errors in forecasts. Also, individual biases may taint the forecast; moreover, some are naturally optimistic, others more cautious. Salespersons may also underestimate their forecasts so their performance looks good.
Customer orders are generated in response to irregular large orders. The allotted lead time is
PRODUCTION SYSTEM
PALEN IX
3 shifts 5 Rotations per shift
Item
Order Quantity
Produced
Quantity Art. Nr. Weight Order Nr. Due date Molds
Available Molds In production KLP® Zwart Steigerplank 6000 5583 240888 13761 80117 2 3 20x4,7x390 13.9 days replenishment
Ronde paal met punt 25 0 230092 443 83607 wk 41 2 2
ø10x300 0.8 days sundries Plank 3000 1625 231897 9.900 74124 1 2 10x3x300 91.7 days sundries Plank 5000 4289 231990 7.679 74702 1 2 15x3x300 47.4 days replenishment Balk 2500 2075 231913 9.690 81115 1 4 20x4x325 28.3 days replenishment Veer en Groefplank 500 238 232148 2.358 83507 wk 43 1 8 14x3,2x250 17.5 days sundries Veer en Groefplank 5000 0 232148 45.000 73817 1 8 14x3,2x250 333.3 days replenishment
Out production KLP® Zwart
Balk 500 0 232680 2800 83326 1 2 7x4x250 33,3 days replenishment Balk 63 0 231940 1.222 83912 1 1 12x6x300 4,2 days diversen Balk 100 0 231940 1.940 83913 1 1 12x6x300 6,7 days replenishment Balk 150 0 232162 3.030 83914 1 2 12x7x300 10,0 days replenishment
Figure 4-5: Work schedule
PRODUCTION SYSTEM
For example, order 80117 is finished over 6000-5583/(3·5)·2) = 13.9 days, where the first part of the denominator equals the number of shifts per day. The second part indicates the number of molds, mounted on the rotator, for this particular end item. The same difference is used to calculate the weight. It is multiplied by an item’s standard weight to determine the required (remaining) weight. The required weight is recorded in the schedule for purchase. It enables purchase to trace total required weight back to the mixture components to check whether there is enough material to finish the order. Once the order is finished, foremen enter the order number from the schedule in Bull to update its status. The updated status alerts the expedition workers that the order is ready. The final column in work schedule shows how much molds there are available to manipulate throughput. Adding a mold cuts the process time in half. Note that there has to be an empty position on the rotator. The number of rotations per shift is an important PPC parameter. PPC uses this number to determine the cycle time. The cycle time basically consist of two major time components: queue time and process time. The calculation of both time components was explained in the preceding paragraph. The queue time is the time orders have to wait for a free spot on the rotor. A free spot is available when an order ‘in’ production is finished. Thus, the queue time is equal to the smallest remaining process time of an ‘in’ production. In Figure 4-5, this is order 83607 which is finished within a day. The process time is the time an order is being worked on. Despite the evident importance of the parameter, it has never been changed during the project. PPC keeps the number of rotations at a constant level whether there are 8 or 20 items mounted on the rotator.
Production orders are released in a predefined sequence. The release sequence is based on two rules. First, customer orders are prioritized over replenishment orders. Second, customer orders are arranged according to their due date (EDD), with the earliest due date first and the latest due date last. The First-Come-First-Served (FCFS) rule is used for the release of replenishment orders. PPC can decide to postpone the release of an order out production and increase the size of an order in production, thereby preventing waste by too many setups. This only happens when the orders out production solely consist of replenishment orders and the relative inventory position of the item in production is lower than the products out production.
Resume
No forecasts of customer demand are made for planning purposes. A current state map is drawn to model the demand and production characteristics. The demand characteristics were determined with a time series analysis. The time series analysis showed that demand is fairly unpredictable and that it is not reasonable to link peaks to projects. The material was analyzed to distinguish production
PERFORMANCE EVALUATION
generated; replenishment and customer. Orders are collected and recorded in the weekly work schedule and wait in a queue for available capacity. The orders are released into the line in a
predefined sequence. Customer orders are prioritized over replenishment orders. Customer orders are released according to their due date (EDD), with the earliest due date first and the latest due date last. The release sequence of replenishment orders is determined by the FCFS rule. The priority of orders out production can change.
5
PERFORMANCE EVALUATION
In this chapter, the fourth sub question is answered. First, the current way of performance evaluation is analyzed. Thereafter, an alternative approach is presented that uses data from the current state map to evaluate performance. The approach is an addition to performance evaluation in VSM, where the current state of the value stream is determined by comparing the cycle time with the raw process time. The demonstrated approach uses cases to diagnose the production line and determine whether
performance is ‘good’ or ‘bad’. If performance is ‘bad’, sources are identified that explain poor performance. Finally, measures are proposed to indicate delivery performance. Each measure is accompanied by sources.
5.1
Performance evaluation at LRP
The production manager uses a spreadsheet to evaluate throughput, or the weekly production output. Foremen enter the produced quantities in this spreadsheet after finishing their shift. The quantity per item is multiplied by its standard weight. The total weight per item is aggregated into a total weight per line. Every week, the manager checks the sum of the weights and compares it to historical values to evaluate performance. The performance, however, depends on the product mix. When the mix in production mainly consists of heavy products, performance is high. Light weighted products result in low performance. Obviously, the manager has to anticipate for the product mix to evaluate
performance.
Table 5-1: Calculation of norm hours
Item Nr. Weigh
t
Monday Tuesday Wednesday Thursday Friday
Total Total weight Nor
m N or m hour s ES LS NS ES LS NS ES LS NS ES LS NS ES LS NS 233889 27,5 21 15 15 14 11 10 10 8 10 8 10 20 21 24 24 221 6078 157,6 38,57 231029 21 6 4 5 4 5 5 4 5 5 4 5 5 4 4 4 69 1449 157,6 9,20 231894 6 15 15 18 16 19 19 13 18 16 16 19 16 17 16 16 249 1494 157,6 9,48 240714 11,6 8 7 9 8 9 10 9 8 9 9 9 7 8 8 7 125 1450 157,6 9,20 240146 24 8 10 9 8 10 8 10 8 10 10 8 12 10 9 8 138 3312 157,6 21,02
PERFORMANCE EVALUATION
In addition, each product is accompanied by a norm in the spreadsheet. The norm denotes the expected output in kilos per hour of a item. The total weight per item is divided by the norm to compute the norm hours. The norm hours are then compared with the actual hours to measure efficiency (see Table 5-1). The actual hours are entered the same time as the product quantities. The efficiency is measured separately for each shift (early (ES), late (LS), and night (NS)). Normally, the norm hours are much higher than the actual hours, indicating a high efficiency. The norms are still programmed in the spreadsheet but there is no recollection of the values used to calculate the norms. Also, the norm values do not differ much. Apparently, new products get the same norm as comparable existing products. The norms were determined years ago and no one seems to care about the efficiency results.
5.2
Throughput, cycle time, and WIP
Evidently, throughput is an important efficiency measure for LPR. LRP wants to carefully manage capacity. The loss of capacity means lower throughput to meet demand. Currently, historical throughput is used because the upper and lower limit is unknown. Over the last year, throughput TH has averaged about 1714 parts per week, or about 14.3 parts per hour (LRP worked 24-hours a day, 5 days per week in 2007). The cycle time CT of the routing is also unknown because it is difficult to determine in practice. The cycle time consists of several variable time components (e.g. queue time, wait-in-batch time, process time) which are difficult to measure. Since throughput and WIP have been routinely been checked, the ratio WIP/TH (Little’s Law) is used to approximate the average cycle time. WIP in the line has averaged about 780 parts, which gives an average cycle time of 780/14.3 = 55 hours. This the time a part requires to pass all processes from the release at injection molding until it is stored in FGI.
Two other important parameters for diagnosing the production line are the bottleneck rate rb and raw process time T0. Since the line consists of a single routing and each process is visited once, the bottleneck is defined by the process with the lowest throughput; injection molding (Hopp & Spearman, 2000). Injection molding is capable of producing at an average rate of 16.5 products per hour (details not given here), not including detractors. So, the rb of the line is 16.5. The raw process time T0 is the sum of the average long-term process cycle times CTp of the processes in the line. This is 23550 seconds, or 6.5 hours (see Figure 4-1). The minimum WIP level W0 to achieve the bottleneck rate is calculated by multiplying the bottleneck rate and raw process time, 16.5 x 6.5 = 108 parts. Actual throughput is 14.3/16.5 = 0.87 of the bottleneck rate, cycle time is 55/6.5 = 8 times the raw process time (ratio emphasized by VSM), and WIP is 780/108 = 7.2 times critical WIP.
PERFORMANCE EVALUATION
Spearman (2000) gives us a scale to benchmark performance of Palen 9. Performance of Palen 9 is compared to the practical worst-case. The performance of the practical worst-case is always between those of the best-case and the worst-case. The ‘bad’ region lies between throughput of the practical worst-case and the worst-case. The ‘good’ region lies between throughput of the practical worst-case and the best-case. Data from the preceding paragraph can be used to simply determine whether average throughput of Palen 9 lies in the ‘bad’ or ‘good’ region. This enables management to establish internal benchmarking. An example of internal benchmarking is given in Appendix E. For cycle time as well as for throughput, the performance of Palen 9 lies in the ‘bad’ region. Unfortunately, the cases only diagnose the performance of the line. They do not tell us why a line is operating poorly and, therefore, do not help us determine how to improve performance. Some causes of bad performance are explained below.
Low throughput performance is partly explained by random outages. Random outages due to machine breakdowns happen whether we want them or not. Mold valve, scraper, lid station, mold, and water jet defects frequently occur right in the middle of an order. Defects are not monitored at production. Production does not know what the availability of a machine is. Machine setups also affect
throughput. As mentioned, there are two types of setups: major and minor. Major setups do not occur very often at Palen 9. But minor setups do. It is not uncommon to see that the machine is stopped two times for replacing a mold or that the machine is stopped before collecting the required tools to perform the setup. The average duration of both setups is not recorded in planning nor is the average number of parts between setups. Another source of low performance is rework. Only quality approved parts are entered into the weekly spreadsheet. When a part does not conform to the design
specifications, it gets quality disapproval and is thrown in the garbage. Rework is required to get the part right. Thus, rework requires additional processing and therefore has a similar effect as setups; the loss of expensive capacity.
DIAGNOSIS
waiting for other parts to arrive. After finishing, products wait until the batch is formed. After the batch is formed, it is packaged and moved to prespecified location in the factory for a quality check. One huge source of waste is not uncovered in the current state map. That is the time orders out production have to wait before being released into the line. Orders have to wait for other orders to complete processing. This queue time only exists in the work schedule but is a considerable component of the cycle time under high utilization.
5.3
Delivery performance
Despite the high priority of dependability, the sales department of LRP does not measure delivery performance. A delivery performance measure is required to determine the fraction of orders that are satisfied on time. The choice for a suitable performance measure depends on the manufacturing strategy. Delivery performance in a made-to-order strategy is measured by the fraction of orders that is filled within their lead times. This measure of delivery performance is referred to as service level. In a make-to-stock strategy, delivery performance is measured by the fill rate. The fill rate is the fraction of demand filled from the shelf.
Resume
Currently, it is hard for the production manager to evaluate performance. Badly chosen units make it merely impossible to judge weekly output. Also, the efficiency measure is neglected because there is no recollection of the values used to establish the required norms to compute the measure. A new method is proposed that allows us to benchmark actual performance. Actual performance was
compared to the practical worst-case to determine whether Palen 9 is performing ‘good’ or ‘bad’. Both throughput and cycle time are in the ‘bad’ region. Bad performance is explained by outages, setups, rework, low availability of quality staff members, and batches. The time orders have to wait for freed up capacity was identified as a huge source of waste. Finally, fill rate and service level were presented as possible delivery performance measures.
6
DIAGNOSIS
Bad delivery performance is caused by unpredictable lead time and stockouts. Unpredictable lead times and stockouts are both symptoms of the root cause of the problem. The root cause of the problem is determined in the diagnosis.
DIAGNOSIS
back log. Variation is caused by two sources: rearranged release sequence and length of the planning horizon.
Figure 6-1: Order cycle time
The release sequence of orders ‘out’ production is often rearranged due to new arrivals. New orders arrivals change the priority of orders in the queue. If a free spot comes available, it is assigned to the order with the highest priority. The priority of an order is determined by rules (see paragraph 4.3). The arrival of a customer order (blue) pushes a replenishment order (red) back into the queue. EDD has a similar effect. Customer orders with an earlier due date have priority over the ones with later due dates in the queue. Orders that are placed back in the release sequence have to wait longer for a free spot on the rotator. Evidently, rearranging the sequence causes variation of the queue time. The presence of variable queue time makes it difficult to estimate the planned lead time. Especially when the order backlog is large, the queue time varies more and the lead time has to be planned over a longer horizon. The replenishment orders are released in the FCFS-sequence. The FCFS-sequence is not rearranged when new replenishment orders arrive. New arrivals placed at the back of the queue, negating the fact that urgency of replenishment changes. But urgency does change. If an unexpected large order is filled from the shelf, the replenishment urgency of orders at the back can be significantly larger than the ones at front. In a FCFS-sequence, inventory is frequently replenished too late. Late replenishment of inventory increases the risk of stockouts.
CONCEPTS OF REDESIGN
Table 6-1 shows how the effective throughput varies with a different number of occupied mold positions.
Table 6-1: Effective throughput
# Rotations per shift # Rotations per shift # Occupied mold positions 4 5 6 # Occupied mold positions 4 5 6 1 0,5 0,6 0,8 11 5,5 6,9 8,3 2 1,0 1,3 1,5 12 6,0 7,5 9,0 3 1,5 1,9 2,3 13 6,5 8,1 9,8 4 2,0 2,5 3,0 14 7,0 8,8 10,5 5 2,5 3,1 3,8 15 7,5 9,4 11,3 6 3,0 3,8 4,5 16 8,0 10,0 12,0 7 3,5 4,4 5,3 17 8,5 10,6 12,8 8 4,0 5,0 6,0 18 9,0 11,3 13,5 9 4,5 5,6 6,8 19 9,5 11,9 14,3 10 5,0 6,3 7,5 20 10,0 12,5 15,0
Apparently, salespersons use a lower effective rate to anticipate outages. Outages due to machine breakdowns, setups, planned maintenance and rework are included. The treatment of these outages, however, depends on the planning frequency. On the short term, typical outages (i.e. mold defects, water jet defects) should be included; but long outages, resulting from major breakdowns, should not. On the long-term, capacity losses due to major machine breakdowns, should be included. The
presence of the huge backlog force salespersons to plan far into the future. Therefore, salespersons anticipate long outages in determining effective throughput. Because of the lower throughput, orders are frequently finished unpredictably early. A smaller backlog enables PPC to exclude long outages andplan the lead time over a shorter horizon. A shorter horizon increases the accuracy of the planned lead time.
7
CONCEPTS OF REDESIGN
In this chapter, concepts are introduced for the redesign of policies. Not all policies are eligible for redesign. Surely, policies that reduce machine breakdowns, setups, and rework will improve delivery performance. But under the given research goal, we focus on concepts within the field of PPC. A lot of problems relate to the lack of transparency in PPC. The concepts in this chapter help to bring back transparency and, above all, integrate capacity utilization on a higher decision level.
7.1
Capacity oriented approach
CONCEPTS OF REDESIGN
ask for an approach that integrates capacity on a higher level; a capacity oriented approach. In a capacity oriented approach there is more direct control over production; orders are released with real-time information.
Currently, LRP uses a material oriented approach to planning; the generation is separated from the release of orders (Bertrand et al., 1998). Generated orders are collected and placed in a queue where they wait for available capacity. As mentioned, orders in the queue are often rearranged and outages frequently occur. From that, cycle times tend to vary and deviate from planned lead times. Especially at LRP, where utilization is high and queues are long, lead times are hard to predict and orders delays frequently occur. This results in poor delivery performance.
In the capacity oriented approach, the generated but not released production order has little meaning (Bertrand et al., 1998). When the availability of raw material is not an issue, the order quantity can be changed just before it is released into the line. At LRP, there is always enough raw material for production. Shortages rarely occur. As such, it is possible to postpone the generation of an order until it can actually be released. This leads to an alternative approach to planning where orders are released based on actual demand and production characteristics. However, this approach is more complex because these characteristics are needed at the same time to determine how many of what products have to be released. There are two ways to reduce complexity.
The first possibility for simplification is predefining the production runs. The optimal run optimizes the trade-off between setup and holding costs and can be computed with the well-known formula of Camp. The run can also be equalized to the exact size of the customer order. This is only feasible for some items, where customer orders are known well in advance.
