Workload control of simultaneous batch processes:
Considering incompatible product families at the job release level
University of Groningen
Faculty of Economics and Business
Technology and Operations Management
1 Abstract
2 Acknowledgements
With this thesis, I complete the MSc Technology and Operations Management at the University of Groningen. It has been a challenging but rewarding project which could not have been completed without the help and assistance of several people.
First of all I would like to thank my supervisor, Hans Wortmann, for the constructive feedback he gave me during the discussions we had together. Second, I want to thank the co-assessor of my research, Jos Bokhorst, for the helpful feedback I received from him. In addition, I want to thank Sabine Waschull for her guidance at the introduction to the supervisors of the case company.
Furthermore I want to thank the people of Fokker Aerostructures that where involved and gave me the opportunity to get the insights needed for this research. In particular I want to thank Martijn Jaasma and Sibren Posthuma for their guidance during my time as an intern at the company.
3 Table of Contents
1. Introduction ... 4
2. Theoretical background ... 6
2.1 Workload control... 6
2.2 Simultaneous batch processes ... 7
2.3 Conceptual model ... 8
3. Methodology ... 9
3.1 Research design ... 9
3.2 Literature research ... 9
3.3 Single case study ... 9
3.4 Data collection ... 10 3.5 Data analysis ... 10 3.6 Design ... 10 4. Case description ... 11 4.1 Production process ... 11 4.2 Planning process ... 12 4.3 Process complexities ... 13 4.5 Customer demand ... 15 5. Analysis ... 16 5.1 MRP demand ... 16 5.2 Autoclave charges ... 17 5.3 Backwards planning ... 18 5.4 Upstream planning ... 20 6. Design ... 22 6.1 Input control ... 22 6.2 Output control ... 25 6.3 Application ... 26 7. Discussion ... 28
7.1 Related to the case ... 28
7.2 Contribution to academic knowledge ... 30
8. Conclusion ... 32
References ... 33
4 1. Introduction
In order to enhance efficiency of production, simultaneous batch operations are often involved in production environments. A simultaneous batch operation can simultaneously process several jobs in a batch (Pearn et al., 2013). Examples are ovens and heat treatment operations that are encountered in many industries, such as the manufacture of semiconductors and the aircraft industry (Li et al., 2015; Van der Zee et al., 2001). Scheduling a simultaneous batch process involves the assignment of products to batches that will be processed at appropriate times while satisfying all applying constraints (Jula and Leachman, 2010). A limited number of products can be processed in a particular batch due to volume constraints (Chiang et al., 2010). In addition, the characteristic of long processing times makes scheduling complicated and challenging (Jula & Leachman, 2010). Simultaneous batch processes are often bottleneck operations in manufacturing systems (Jia et al., 2015).
In simultaneous batching situations, the basic trade-off is between effective capacity utilization and minimal wait-to-batch time (Hopp & Spearman, 2008). This causes that in order to achieve efficient production, the workload on the shop floor has to be controlled. The concept of workload control is a production planning and control concept developed for high-variety job shops (Land & Gaalman, 1996). It integrates two control mechanisms: input control, to regulate the inflow of work to the system and output control, which uses capacity adjustments to regulate the outflow of work from the system (Oosterman et al., 2000). Much research has focused on input control. However, within input control, sequencing of orders at the release level has been largely neglected in literature (Thürer et al. 2015). Concerning output control, complexities of adjusting capacity in reality affect opportunities to increase performance and therefor needs further research (Land et al., 2015; Thürer et al., 2016).
5
the autoclave process. The described complexities need to be controlled by a combination of input- and output control. In order to address sequencing of orders at the input control and complexities to adjust capacity at the output control, this study aims to answer the following research question:
How can input- and output control be combined at the job release level to establish efficient batches for a simultaneous batch process characterized by incompatible product families?
To identify the relevant theory concerning workload control for simultaneous batch processes with incompatible product families, the literature is researched. The implications of incompatible product families and the limitations to adjust capacity are defined by a single case study at Fokker Aerostructures including interviews and observations. Analysis is performed to design a planning approach that combines input- and output control at the job release level to establish efficient batches for a simultaneous batch process with incompatible product families. This research contributes to the need for further research concerning input- and output control by studying sequencing of orders and complexities to adjust capacity at the job release level of the workload control concept.
6 2. Theoretical background
This section presents theories that serve as background to explore the opportunity to combine input- and output control in order to establish efficient batches for a simultaneous batch process with incompatible product families. Based on the identified theories a conceptual model is developed. 2.1 Workload control
The concept of workload control was developed between 1980 and 1990 (e.g. Bertrand and Wortmann 1981; Hendry and Kingsman 1989). The purpose of the concept is to ensure that orders are completed before their due dates while fully utilizing the capacity of the production processes (Fredendall et al., 2010). Many articles with a theoretical focus have followed and simulation studies demonstrated that workload control has the potential to improve performance (Thürer et al., 2011). However, Soepenberg et al. (2012) mentioned that there have been reported only few recent successful real life implementations of workload control. The most important aspects are the control levels, input control regulates the inflow of work to the production process and output control adjusts capacity to regulate the outflow of work from the process (Oosterman et al., 2000). Controlling the workload on the shop-floor can be done at different levels in the process and therefore the concept of workload control exists of three control levels: job entry, job release and dispatching as presented in figure 1.
Figure 1. Workload Control (Land & Gaalman, 1996)
7
2012). The workload control concept is developed to improve delivery performance which can be measured by lateness. Lateness can be negative or positive when an order is released early or late and zero when an order is delivered on the due date (Posthuma, 2014). Early released jobs with negative lateness result in inventory costs. Orders with positive lateness are delivered after the due date which can result in costs to compensate for the time the customer has to wait on delivery (Land et al., 2014). 2.2 Simultaneous batch processes
In theory two types of batches are described: s-batches and p-batches. At the first one, jobs in a batch are processed in serial and the processing time of a batch is the sum of the processing times of all the jobs in the batch. At the second one, the jobs in a batch are processed in parallel and processing time is determined by the job with the longest processing time in the batch (Jia et al., 2015). A simultaneous batch process is characterized by p-batches where orders are accumulated and processed together to maximize the capacity of a work center (Uzsoy, 1995). These processes are commonly found in make to order companies where workload control is an important concept (Fowler et al., 2002). The batch size is independent from the processing time of a simultaneous batch process and once the process is started, no orders can be added (Jula & Leachman, 2010). This means that the processing time is equal for all batch sizes that are smaller than the maximal batch size (Mathirajan et al., 2010).
Often high costs are associated with processing a batch of a simultaneous batch process due to substantial energy usage (Liu et al., 2016). Costs per job can be decreased by spreading the costs of the process over multiple jobs that are processed simultaneously (Xu & Bean., 2015). Therefore it is important to consider the maximal batch size for a simultaneous batch process (Chiang et al., 2010). In order to establish efficient batches the workload has to be controlled. The release decision of workload control can be divided in two parts: first a selection decision which determines the criteria for choosing a particular job or set of jobs for release from the job pool and second, a sequencing decision which establishes the order in which jobs are released (Thürer et al, 2015). Simultaneous batch processes are commonly categorized into two types: compatible product families and incompatible product families. The characteristic of incompatible product families causes that only jobs of similar families can be processed simultaneously (Pearn et al., 2013). This characteristic makes the sequencing decision important to establish efficient batches for a simultaneous batch process. Only considering sequencing of orders at the dispatching level requires a high workload to be able to reorder the sequence in the queue in front of the simultaneous batch process. Considering sequencing already at the job release level allows for lower workloads (Cransberg et al., 2015).
8
process can be increased by considering the bottleneck process. An important insight provided by the theory of constraints is that “an hour lost at a bottleneck process is an hour lost for the entire production process” which encourages to utilize the bottleneck process as much as possible. This makes the focus on efficient batching for the simultaneous batch process important to improve performance of the overall production process. Considering the sequence of orders already at the order release level only results in efficient batches in front of a simultaneous batch process if orders arrive on time to enable the desired sequence (Cransberg et al., 2015). Therefore the capacity of preceding processes is important to consider at the output control. Determining capacity can be complicated by resource dependent capacity which means that multiple resources are needed for a process and operations of different products can need varying resources (Stevenson & Silva., 2008). Limited availability of resources makes it complex to determine if a process is able to produce products dedicated to a batch on time for the simultaneous batch process. In addition, a preceding process can have an unsynchronized shift schedule compared to the simultaneous batch process. Differences in operating hours can cause starvation if the simultaneous batch process is not provided with sufficient orders before the end of the shift of the preceding process (Ernst et al. 2004).
2.3 Conceptual model
Based on the literature, a conceptual model is developed as presented in figure 2. Incompatibility of product families causes that the sequence of orders released into the production system is important. By input control at the job release level a sequence can be established that has to be processed on time by preceding operations to enable the desired sequence at the simultaneous batch process. Concerning output control this is affected by the complexity of resource dependent capacity of preceding operations. The described complexities influence the ability to utilize the maximal batch size which determines the efficiency of batches for a simultaneous batch process.
Incompatibility of product families
Complexity resource dependent capacity of predecessors
Utilization maximal batch size of a simultaneous batch process
9 3. Methodology
The aim of this research is to investigate how efficient batches for a simultaneous batch process with incompatible product families can be established by combining input- and output control. The methodology used to execute the research is discussed in this section.
3.1 Research design
To identify the available literature concerning workload control that is relevant for simultaneous batch processes with incompatible product families the literature is researched. The limitations to apply input- and output control due to complexities in practice are defined by a single case study. After analysis of the data, a planning approach is designed to combine input- and output control to establish efficient batches for a simultaneous batch process characterized by incompatible product families. 3.2 Literature research
In order to gain insights in the relevant workload control theory concerning simultaneous batch processes with incompatible product families, the literature is researched as presented in the previous section. Two literature search approaches are used. First, several computerized databases e.g. Emerald Insight, Taylor & Francis and Science Direct are searched for relevant academic journal papers. Second, recent publications are searched in leading journals for the researched field e.g. International Journal of Production Research, International Journal of Operations Research, and Computers & Operations Research. To search articles, keywords are used e.g. “simultaneous batch processes”, “incompatible product families” and “workload control”. An article is selected based on the title to read the abstract. If the abstract gives a promising impression, the introduction and conclusion are read. On articles that are selected, a forward search is performed to identify related articles that are published more recently. To ensure the quality of this research, especially peer-reviewed articles are selected.
3.3 Single case study
10 3.4 Data collection
To collect qualitative data concerning the production and planning processes of the case company, structured interviews are used. Interviews are executed according to the funnel model which means that an interview starts with broad and open-ended questions and becomes more specific during the progress of the interview. To develop a comprehensive understanding and increase reliability of this research, respondents with different functions are interviewed e.g. managers, planners and operators. The conclusions drawn from the interviews are returned to the interviewees for verification to ensure that interviewees agree with the conclusions. Because of the opportunity to be as an intern at the case company during the research, also unstructured interviews and observations at the shop floor are used as sources of data. Also material requirements planning (MRP) data concerning demand and releases of the previous year is required for analysis. This is collected from the information system of the case company and databases developed by a specialist of the autoclave process.
3.5 Data analysis
A limited and an unlimited backwards planning approach, which are explained in section 5, are analyzed on performance regarding efficient batching for a simultaneous batch process. MRP demand and release data of the year 2015 of a selected program of the case company is analyzed to determine if there is potential to increase utilization of the maximal batch size by efficient batching with the two approaches. Based on expertize of the specialist of the autoclave process, the number of articles of the selected program that can be simultaneously processed in an autoclave is determined. Then the potential to establish efficient batches is analyzed by comparing the results of the backwards planning approaches to the results of the MRP approach. Also the resulting inventory of the two approaches caused by planning releases backwards is analyzed. Finally, the influence of resource dependent capacity of preceding operations on efficient batching for the autoclave process is analyzed.
3.6 Design
11 4. Case description
This section provides the case description and first describes the production and planning processes of the case company. Then the process complexities are described that influence to applicability of input- and output control of the workload control concept to the case. Finally customer demand is treated because it influences the ability to balance the workload.
4.1 Production process
The case company produces a high variety of synthetic articles for the aircraft industry. This research focusses on the first processes of the entire production process of the company. A logic flow diagram of the first part of the production process is presented in figure 3. When an article is processed by the operations precutting, layup, autoclave and debagging it is transported to the operation in its routing at the next part of the production process which is called machining. When the machining operation has not sufficient capacity to immediately process the semi-finished article it is placed in a buffer before the machining process. After an article finishes all machining operations and it passes the assembly process it is ready for delivery to the customer.
Pre-cutting 1 Pre-cutting 2 Hand-cutting Lay-up Lay-up Lay-up Lay-up Lay-up Autoclave 3 Autoclave 2 Autoclave 1 Debagging Buffer Mould maintenance Lay-up moulds Jobs to release
Figure 3. Logic flow diagram
12 Articles of different article groups cannot be processed simultaneously in the autoclave because they need different heat treatments which are called curves. This means that the article groups are the incompatible product families of the autoclave process. An article group exists of multiple articles that are non-identical, therefore each article has a specific
dedicated type of mould. The amount of available moulds per article varies depending on customer demand. At debagging the article is separated from the mould and it is moved to its next operation at the machining part of the production process. Maintenance of moulds is periodically planned but issues can cause that they have to be maintained earlier and are unavailable to process articles. Each program has its own layup department with resources that are needed to operate the process. The capacity of the processes pre-cutting, autoclave and debagging is shared by the programs.
4.2 Planning process
The planning process of the case company has different phases as presented in figure 5. Based on negotiation with customers of the different programs a master production schedule (MPS) is established. Accepted orders are as much as possible aligned with the capacity status of the production system. However, it occurs that due dates of orders are accepted while not enough capacity is available in that particular week because customers have high negotiation power. The data of the MPS indicates when orders have to be delivered to the customers which is called the due date. Based on the due date of an order, the MRP system of the company calculates when the articles have
to be released in the production process. It also determines the latest starting times of the operations in the routing of the articles. By considering the duration of all downstream operations of a process the latest starting time is determined as presented in the equation below.
𝐿𝑎𝑡𝑒𝑠𝑡 𝑠𝑡𝑎𝑟𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 = 𝐷𝑢𝑒 𝑑𝑎𝑡𝑒 − ∑ 𝐿𝑒𝑎𝑑 𝑡𝑖𝑚𝑒 − 𝑆𝑎𝑓𝑒𝑡𝑦 𝑡𝑖𝑚𝑒
The lead time includes waiting time, setup time and processing time of an article. Safety time is added to compensate for disruptions during the operations in the routing of an article within the production process. Lead times are planned averages which are not related to the status of the shop floor. The
13
precutting process has a lead time of 3 days for each article. At the layup process each department has a lead time varying between 2 and 5 days, depending on customer demand to the program. For the autoclave process a lead time of 2 days is considered. The debagging process has a lead time of 1 day. In addition to the lead time of the processes, safety time is added for the entire production process. Based on the MRP data, a list of articles which is sequenced on latest starting time is developed. On a weekly basis this list is used by the planners to determine which articles need to be released in the next week. The planners also consider what could be produced already in advance for upcoming weeks but this is not done structurally for all programs. In the current situation the layup process has the main focus at the development of the week planning because it is the bottleneck process of the production process with the current customer demand. Added to that, the resource dependent capacity of the layup process is important to consider to determine if articles can be processed in a certain week. When the week planning is developed for each layup department, it indicates which articles are processed on which days. The autoclave process has to adjust its planning to the week planning of the layup departments. The autoclave planning indicates which curves are needed and when the charges are started. A charge is one heat treatment of a particular curve. Also the planning of the pre-cutting and debagging processes is derived from the week planning. Generally the week-planning is made final the week before but often adjustments are made later by the planners of the layup departments to utilize capacity of the process. This causes that the autoclave process also has to adjust the planning which in the short term can be problematic and leads to inefficient batches. The production process is operated five days a week with eventually overwork in the weekend. The precutting, autoclave and debagging processes are operated three shifts a day. However, the layup process is operated only one shift a day from 08:00 to 16:30. This limits the ability of the layup departments to deliver articles on time at the autoclave process which causes that additional charges of particular curves have to be added. In the current situation this does not influence output of the production system because the autoclave process has enough capacity. However, future increases in demand are expected which will cause the autoclave operation to become the bottleneck of the production process because capacity of the layup process can be increased by adding shifts.
4.3 Process complexities
14
process. Therefore, batches with articles that will be processed simultaneously at the autoclave process have to be determined at the release level of the production process. The volume of an autoclave determines the number of moulds that can be processed simultaneously in a batch. Moulds have many different dimensions which makes it complicated to determine if an autoclave is filled optimally with a certain batch. In addition, each mould needs a vacuum and/or heath connector which determines if a batch of moulds can be processed. This causes that for example a batch with many small moulds which fit in de autoclave cannot be processed because not enough connectors are available. In addition to the volume and connector constraints also technical complexities influence the amount of moulds that can be processed simultaneously. This can cause that less moulds can be processed in a batch than volume and connectors are available because otherwise the quality of articles would be insufficient through an uneven distribution of heath in the autoclave. The foregoing makes it challenging to determine how many articles can be processed in one charge.
With regard to output control the availability of resources determines the capacity of the processes. Table 2 gives an overview of the resources that are important to consider for each process. The precutting process can only cut layers synthetic material of a certain type for the layup process when the material is available. For the layup process moulds are limited available for each article because they are expensive assets which influences the amount of articles of a particular type that can be released simultaneously. In addition, moulds are periodically unavailable because of maintenance. A mould is dedicated to an article from the start of the layup process until the moment it is separated from the article at the debagging process. At the lay-up departments of the programs a limited amount of lay-up tables and laser projection systems are available. All articles need a lay-up table of a particular type and a part of the articles needs a laser projection system. Finally, the amount of workers and shifts at each layup department influences the capacity. The workers are dedicated to programs but not each worker is capable to process all types of articles because for example working with a laser projection system needs training and experience.
Precutting Available synthetic materials in stock of type s
Layup Available moulds with tooling of type m Available amount of lay-up tables of type t Available laser projection systems of type l Number of workers at each layup department
Autoclave Volume of autoclave k
15 4.5 Customer demand
Customers demand for shipsets which consist of all articles that belong to an aircraft of a particular program and have to be delivered simultaneously. Because the routing and associated lead times of articles differs in the machining part of the production process, the latest starting times for articles of a shipset can vary. This influences the opportunity to establish efficient batches for the autoclave process because articles are not released simultaneously if the latest starting times are maintained. The most important performance measure of the case company is the on time delivery to ensure that shipsets are delivered before the due dates. A second important measure is the on time start which means that an article has to be released and processed at the operations within -7 and +7 days of the latest starting time. The lateness is a performance measure to ensure that shipsets are delivered before the due date to customers. In addition, the earliness measure is to control working capital and regulate inventory because limited space on the shop floor is available.
The company has long term contracts with its customers for the delivery of shipsets and spare parts but the demand is not equally distributed per week at the job entry level which causes fluctuations in need for capacity of the production processes. In addition, customers can decide to shorten or extend the due date which causes changing latest starting times of articles. To give an indication of delivery performance, demand and release data of the A380/A400M program is analyzed. The latest starting time of an article for this program can be changed up to 10 days before the planned latest starting time. Besides customer demand there is also internal demand because articles may be scrapped through quality issues. Workloads are frequently higher than de capacity of the production process is able to handle with backlogs and violated latest starting times of articles as result. Figure 6 presents the lateness of releases of articles in weeks for the A380/A400M program. A percentage of 58,5% of all released articles in the year 2015 was released more than 1 week after the latest starting time. Concerning earliness, 11.9% of the articles was released more than 1 week early. This results in an average lateness of 1.7 weeks and a percentage of 29.6% of articles with an on time start.
Figure 6. Results on time start A380/A400M 2015
16 5. Analysis
In this section an analysis is performed to determine the number of autoclave charges that have to be planned per week for different planning approaches. In addition, the influence of resource dependent capacity of preceding operations on the applicability of the planning approaches is analyzed.
5.1 MRP demand
Based on the due date of a shipset or spare part, the latest starting times of the operations of the production process are determined by the MRP system which does not take into account efficient batching of the autoclave process. This means that it considers fixed lead times and releases of articles are only planned based on the latest starting time and no releases are planned earlier to establish full batches of compatible product families. Therefore it is expected that there is potential to reduce the amount of charges. In order to analyze this it is important to know how many articles of an article group can be processed simultaneously in an autoclave. The program A380/A400M is selected to analyze if there is potential to reduce the number of charges that are needed. This program produces wings for the A380 airplane which exists of the skins and subspars article groups which each have 16 articles. Secondly, for the A400M airplane the program produces protection panels existing of one article group of 16 articles. Each article group needs a different curve in the autoclave.
Planning data of the year 2015 is analyzed to determine the amount of articles that were demanded to be processed by the autoclaves each week according to the MRP system. In figure 7 is presented how insight is gained into the demand to each article per week. The table gives an overview of demand for A380-subspar articles in the first 10 weeks of the year 2015. This overview is generated for each article group for all weeks of 2015 as presented in appendix A. The next step of the analysis is to determine the number of charges of each curve that should be planned if the MRP data is adopted to establish the planning. The results then can be compared to the results of other planning approaches.
17 5.2 Autoclave charges
The autoclave curves that are used for the A380/A400M program are the 137 for the A380-skins and 1304 for the A380-subspars, the A400M-panels are processed by the 1031 curve. This are relatively short curves with a duration around 3 hours. Based on analysis of the specialist of the autoclave process and experience of the autoclave operators is determined that 6 moulds of the A380-skins or A380-subspars can be processed simultaneously in the largest autoclave which is called AC5. There is also a smaller autoclave with the name AC3 which is able to process 4 moulds of the A380-skins or A380-subpars simultaneously. Autoclave AC3 is only used to process A380 articles when autoclave AC5 is broken down. Both autoclave AC3 and AC5 have capacity to process 8 moulds of the A400M program. For the A380-skins and A400M-panels one mould is available per article, this means an availability for each article group of 16 moulds. For the A380-subspars article group only 8 moulds are available, each mould has place for two specific articles. With the information above can be determined how many charges have to be planned if the demand of the MRP system is adopted at the development of the week-planning. In figure 8 an overview is presented of demand to articles, indicating the number of charges of curve 1304 that have to be planned each week to satisfy demand.
Figure 8. Charge demand MRP
18
For example, in a situation where 6 articles are demanded in one week but two times the same article is asked, the autoclave capacity is sufficient to process the articles in one charge but the limited amount of moulds causes that an additional charge has to be planned. These complexities cause that the number of charges needed each week is determined as presented in the equation below.
Charges = Maximum (( Σ Maximum (Article pair) / Moulds per charge , Maximum article demand))
The equation determines for a week the highest value between the charges based on the moulds that need to be processed and the maximum demand for a specific article with one available mould. The result of the equation is rounded up because a half full batch takes one charge of the same length as a full batch. The amount of charges that is needed in each week of the year 2015 is determined for the three article groups of the A380/A400M program as presented in appendix A. The total amount of charges needed with the MRP planning approach is presented in table 3. The charges are not optimally filled as shown by the measure for utilization of the maximal batch size which confirms that there is potential to reduce the amount of charges which is analyzed in the remaining part of this section.
A380-skins A380-subspars A400M-panels
Charges 99 99 81
Utilization 0.763 0.667 0.659
Table 3. Charge demand MRP 2015
5.3 Backwards planning
In order to reduce the amount of charges, each charge should be filled as much as possible with articles. This can be done by planning the release of articles in earlier weeks which we call backwards planning. With this approach for example a demand of 4 articles in week 6 and 2 articles in week 8 which normally needs 2 charges can be combined with a reduction of one charge as result. The planner of the A380/A400M program in the current situation already considers backwards planning, especially for the A380-subspars. The amount of charges actually planned and the utilization of the batch size in 2015 is presented for each article group in table 4. Also the reduction in charges of the actual result compared to the MRP result is shown. It is again assumed that only charges are planned for autoclave AC5 with a capacity of 6 articles of the A380 and 8 articles of the A400M article group. The actual releases of articles and resulting autoclave charges in the weeks of 2015 are presented in appendix B.
A380-skins A380-subspars A400M-panels
Charges 91 62 63
Reduction 8 37 18
Saving (%) 8,1% 35,4% 22,2%
Utilization 0.830 0.718 0.847
Table 4. Charge demand actual 2015
19
approach for example an article planned by MRP in week 12 is allowed to be planned backwards between week 8 and week 12. The results of the backwards-L planning approach are presented in table 5 with a percentage reduction compared to the MRP result. In appendix C an overview is shown of the backwards planned articles and resulting charges in the weeks of 2015.
A380-skins A380-subspars A400M-panels
Charges 84 53 58
Reduction 15 46 23
Saving (%) 15.1% 46.5% 28.4%
Utilization 0.899 0.934 0.920
Table 5. Charge demand backward-L 2015
When the limited amount of weeks that orders are allowed to be planned backwards is neglected the amount of charges needed can be reduced up to optimality. We call this the backwards-U approach which gives the opportunity to take optimal advantage of efficient batching opportunities. In appendix D is presented which articles are planned backwards with the resulting charges in 2015. The reduction of charges with the backwards-U approach compared to the MRP result is shown in table 6.
A380-skins A380-subspars A400M-panels
Charges 76 43 54
Reduction 23 56 27
Saving (%) 23.2% 56.6% 33.3%
Utilization 0.993 0.965 0.988
Table 6. Charge demand backwards-U 2015
Backwards planning causes that semi-finished articles have to wait after passing the first part of the production process before they can be processed further at the machining part of the production process. Analysis is performed to give an indication of the resulting amount of inventory when the backwards-L and backwards-U approaches are applied. Both backwards planning approaches determine for each article if backwards planning leads to less charges and otherwise the MRP release date is maintained. The limited and unlimited amount of weeks allowed to plan articles backwards causes differences in the average number of weeks a semi-finished article has to be stored as shown in table 7. This is calculated as presented for the A380-skins article group in appendix E.
A380-skins A380-subspars A400M-panels
Backwards-L 29 149 123 Percentage of total 6.4% 30.9% 28.8% Average weeks 1.31 1.68 1.35 Backwards-U 54 217 136 Percentage of total 11.9% 44.9% 31.85% Average weeks 4.80 7.58 2.54
Table 7. Expected inventory 2015
20
percentage of backwards planned articles of the total demanded articles is presented. Finally, the average amount of weeks that articles are planned backwards is presented. This data gives an indication of the inventory resulting from the two backwards planning approaches.
5.4 Upstream planning
At the application of backwards planning and the resulting reduction of charges for the autoclave process it is important to consider the resource dependent capacity of the layup process to ensure that all articles of a batch can be delivered on time at the autoclave. The A380-skins article group is selected to analyze the routing from the precutting to the debagging process. For each article one specific mould is available that needs a layup operation varying between 7 and 11 hours, depending on de A380-skin article. A layup table with a laser projection system and a specialized worker are needed to process the article at the layup department. Figure 9 gives a presentation of the routing of this article group and for each step the related lead time, production time and resources needed are indicated.
Figure 9. Production steps A380-skins
The A380-subspar and A400M-panel article groups need a similar routing but the duration of the operations and the need for resources differs. Only one article of the A380 sub-spars needs LPS and no articles of the A400M-panels need this resource. In table 8 the duration of the layup operation for each article is presented which is varying for the A380 article groups because there is difference in layup complexity between articles. A layup operation of one mould is executed by one specialized worker and the duration to complete an operation cannot be speeded up by adding additional workers.
A380-skins A380-subspars A400M-panels
Article Hours LPS Article Hours LPS Article Hours LPS
21 L57450710000 11.0 Y L57450956000 5.0 N M53070583200 4.0 N L57450714000 11.0 Y L57450957000 5.0 N M53070583201 4.0 N L57450810000 7.0 Y L57450960000 4.0 N M53070584202 4.0 N L57450814000 7.0 Y L57450961000 4.0 N M53070584203 4.0 N L57450910000 7.0 Y L57450964000 4.0 N M53070585200 4.0 N L57450914000 7.0 Y L57450965000 4.0 N M53070585201 4.0 N L57451010000 7.0 Y L57450977000 4.0 N M53070586200 4.0 N L57451014000 7.0 Y L57450978000 4.0 N M53070586201 4.0 N L57451110000 11.0 Y L57451112000 3.5 N M53070587200 4.0 N L57451114000 11.0 Y L57451113000 3.5 N M53070587201 4.0 N L57451210000 11.0 Y L57451212000 3.5 N M53070588202 4.0 N L57451214000 11.0 Y L57451213000 3.5 N M53070588203 4.0 N
Table 8. Duration layup operations
In order to determine the capacity of the layup department the available resources and hours per week have to be known. In the current situation the A380/A400M layup department is operated one shift 5 days a week from 08:00 to 16:30 which results in 40 hours a week. A limited number of each resource is available and the number of layup tables and LPS systems is fixed while the number of available workers can vary per week. With the available resources and workhours an indication is given of the capacity of each resource in hours per week as presented in table 9. The capacity needs to be shared between the A380-skins, A380-subspars and A400M-panels article groups. In addition to the resources of the layup department also the limited available number of moulds is important to consider. With the available capacity of the resources can be determined if the articles that are demanded in a week can be processed by the layup department. When all articles are released once, the demand for layup table and worker capacity is 284 hours and the demand for LPS capacity is 157 hours. This indicates that the capacity of one shift per day is approximately sufficient to process all articles once in a week.
Available Capacity (hours)
Layup tables 7 280
LPS systems 4 160
Workers 3 - 7 120 - 280
Table 9. Available capacity of layup resources A380/A400M
22 6. Design
Demand to synthetic aircraft parts is expected to increase. This causes that the autoclave process becomes the bottleneck of the production process because the capacity of the layup process can be extended by adding one or two additional shifts. The focus on the layup process at the development of the planning in the current situation has to be moved to the autoclave process. In the new situation the autoclave process pulls orders from the layup and precutting operations such that they are arriving on time to be processed with a charge in the autoclave. In this section the design is described for the new planning approach by considering input and output control of the workload control concept. 6.1 Input control
At the input control a distinction has to be made between programs and related article groups that need charges of different curves in the autoclave. Releases of articles have to be determined focusing on the efficient batching for the autoclave. Analysis showed that the MRP demand does not consider the potential to establish efficient batches resulting in additional charges as presented in figure 10. Also the current planning performance of the A380/A400M program has potential for improvement. In this research two planning approaches are analyzed, limited and unlimited backwards planning. Analysis showed that the backwards-U planning approach has the highest potential to reduce charges that have to be planned. However, customers not always demand for shipsets and spare parts equally divided over a year. This causes that application of the backwards-U approach results in higher inventory than backwards-L planning which only releases work demanded by the MRP system for a limited amount of weeks ahead. The foregoing indicates that a tradeoff between charge reduction and inventory costs has to determine the choice for a planning approach. For the backwards planning approach a design is developed and in addition a description is given of a cyclic planning approach. The applicability of the approaches for the case company is discussed in section 7.
23 Backwards planning approach
The backwards planning approach is able to substantially reduce charges compared to the MRP planning while considering minimization of resulting inventory. However, it is a complex and time consuming task to plan the releases of articles backwards. Therefore a mathematical linear programming model is developed for this problem
as presented in model 1.1. This model is developed for the A380-skins article group. Characteristics of other programs and article groups can need modifications of the model in order to deal with additional complexities.
Two sets are used, set N for the articles i and set M for the weeks j. The variable Xij determines the amount of released articles i in week number j. The variable Cjn is binary and counts the amount of charges that have to be planned based on the amount of articles that are released in a week. Dij is the only parameter of this model which represents the demand to article i in week j. The objective of the model is to minimize the total amount of charges that are planned in weeks j while considering the amount of articles that are planned backwards because this results in inventory (𝐼). Constraint (1) ensures that demand in week j is satisfied by articles that are processed in that week or in earlier weeks. With the factor k all weeks until week j are considered. The second constraint (2) causes that articles that are demanded in week j are maximally released four weeks in advance for the backwards-L approach. For the backwards-U approach this constraint can be neglected. Constraint (3) activates the first binary variable Cj1 for an autoclave charge if the amount of articles released in week j is larger than zero. The fourth constraint (4) adds with the binary variable Cj2 an
𝑺𝒆𝒕𝒔 𝑆𝑒𝑡 𝑁 = 𝐴𝑟𝑡𝑖𝑐𝑙𝑒𝑠 𝑖 𝑆𝑒𝑡 𝑀 = 𝑊𝑒𝑒𝑘𝑠 𝑗 𝑽𝒂𝒓𝒊𝒂𝒃𝒍𝒆𝒔 𝑋𝑖𝑗 = 𝑃𝑙𝑎𝑛𝑛𝑒𝑑 𝑎𝑚𝑜𝑢𝑛𝑡 𝑜𝑓 𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑖 𝑖𝑛 𝑤𝑒𝑒𝑘 𝑗 𝐶𝑗𝑛= 𝐴𝑚𝑜𝑢𝑛𝑡 𝑜𝑓 𝑎𝑢𝑡𝑜𝑐𝑙𝑎𝑣𝑒 𝑐ℎ𝑎𝑟𝑔𝑒𝑠 𝑖𝑛 𝑤𝑒𝑒𝑘 𝑗 𝑷𝒂𝒓𝒂𝒎𝒆𝒕𝒆𝒓s 𝐷𝑖𝑗 = 𝐷𝑒𝑚𝑎𝑛𝑑 𝑡𝑜 𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑖 𝑖𝑛 𝑤𝑒𝑒𝑘 𝑗 𝑶𝒃𝒋𝒆𝒄𝒕𝒊𝒗𝒆 𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 ∑ 𝐶𝑗1+ 𝑀 𝑗=1 ∑ 𝐶𝑗2+ 𝑀 𝑗=1 𝐼 𝐼 = 0.001 (( ∑(𝐷𝑖𝑗∗ 𝑗) − 𝑀 𝑗=1 ∑(𝑋𝑖𝑗∗ 𝑗 )) 𝑀 𝑗=1 𝑪𝒐𝒏𝒔𝒕𝒓𝒂𝒊𝒏𝒕𝒔 (1) ∑ 𝑋𝑖𝑗 ≥ 𝑘 𝑗=1 ∑ 𝐷𝑖𝑗 𝑘 𝑗=1 ∀ 𝑖 (2) ∑ 𝑋𝑖𝑗 ≤ 𝑘 𝑗=1 ∑ 𝐷𝑖𝑗 𝑘+4 𝑗=1 ∀ 𝑖 (3) ∑ 𝑋𝑖𝑗 𝑁 𝑖=1 ≤ 𝐶𝑗1 𝑀 ∀ 𝑗 (4) ∑ 𝑋𝑖𝑗 𝑁 𝑖=1 ≤ 6 + 𝐶𝑗2 𝑀 ∀ 𝑗 (5) 𝑋𝑖𝑗 ≤ 1 ∀ 𝑖, 𝑗 (6) ∑ 𝑋𝑖𝑗 𝑁 𝑖=1 ≤ 12 ∀ 𝑗 (7) 𝐶𝑗= 𝐵𝑖𝑛𝑎𝑟𝑦 𝑋𝑖𝑗= 𝐼𝑛𝑡𝑒𝑔𝑒r 𝑋𝑖𝑗 ≥ 0
24
additional charge to a week if more than six articles are demanded. To control the workload for the layup departments and consider the limited availability of moulds, constraint (5) ensures that no more than one of each article number is released each week. To balance the autoclave charges over the weeks, constraint (6) allows a maximum of twelve articles to release each week which is two charges for the autoclave. When customer demand increases, constraints (5) and (6) can be adjusted. Finally, constraint (8) ensures that Cij is binary and Xij has integer values and is bigger or equal to 0.
The model is implemented in Microsoft Excel by use of the Solver application. This application has a limited amount of variables and constraints that can be considered so the tool has the ability to determine an optimal charge planning for 5 weeks. A second limitation is that the objective to consider inventory makes the model too complex for the software to find an optimal solution. A more advanced software will be able to consider the whole model with more variables. With the model the MRP demand of week 22 to 26 of the A380-skins in the year 2015 is considered to determine if there is potential to establish a planning that needs
less charges. As shown in appendix A the A380-skins article group needs 10 charges divided over the weeks in that period. The MRP data is implemented in the model and after running the solver it appears that there is potential for a reduction of 30% to 7 charges as shown in figure 11. Based on the MRP data, the backwards planning model is able to periodically reestablish the backwards planning such that changes in demand are considered.
Cyclic planning approach
A cyclic planning approach is able to reduce the amount of charges with the same number as the backwards-U planning approach because both consider unlimited weeks to plan the releases of articles backwards. However, the cyclic planning approach does not consider resulting inventory when articles are released before the latest starting time. In addition, a cyclic approach is most appropriate for a situation with stable demand without backlogs because otherwise backwards planning is needed in order to meet the due dates. An advantage of a cyclic planning is its repetitive behaviour by dividing the releases of articles in a balanced way over a year. This causes that variability is reduced and routine is established. Weeks can be planned free between cycles to execute maintenance or to process rework through quality issues. In figure 12 the cyclic planning for the first 14 weeks of 2015 is presented
25
for the A380-skins article group. However, the variability in demand causes that this planning will lead to due date violations. The fluctuations in demand as shown in appendix A will cause that the cyclic planning needs to be adjusted in order to meet the high demand in for example the first weeks of 2015. This indicates that the fluctuations in customer demand cause that the cyclic planning approach is not appropriate in the current situation.
Figure 12. Cyclic planning A380-skins
6.2 Output control
With output control it has to be ensured that the layup department of a program has enough capacity to deliver the articles that have to be processed in a charge on time at the autoclave. A program can exist of multiple article groups which need charges of different curves in the autoclave. This causes that the demand of the different article groups of a program has to be considered simultaneously to determine if the capacity of the layup department is enough to process the demanded articles of a week. This means for the A380-A400M that the demand for A380-skins, A380-subspars and A400M-panels has to be considered simultaneously at the application of backwards planning.
26
software. Once this is done at a future project, the values of the capacity parameters have to be determined, based on one shift as in the current situation. However, if the capacity appears to be insufficient to satisfy demand, the parameters have to be adjusted by adding capacity. For example additional resources or an additional shift can be added.
Because not for each program an increase in demand is expected, the need to add capacity to the layup department will vary per program.
When the amount of needed charges is determined and capacity of the layup department is sufficient, the starting times of the autoclave charges have to be determined. At the development of the autoclave planning again the resource dependent capacity is important to consider because the limited number of available moulds per article causes that charges of the same curve cannot be planned close to each other. Otherwise the capacity of a week is sufficient but an uneven distribution of workload causes delays. Based on the autoclave planning the latest starting times of the preceding operations layup and pre-cutting need to be determined by upstream planning according to lead times of the processes.
6.3 Application
With the backwards planning approach a week planning for each program can be developed as in the current situation of the case company. This planning indicates which articles have to be processed in a week by the autoclave process in order to establish efficient batches and utilize the maximal batch size. Based on this planning, the starting times of the autoclave charges and preceding operations have to be determined. Application of the backwards planning model saves time for the planners to perform the complex task of planning the releases of articles backwards. This research focusses on the A380/A400M program, characteristics of other programs can cause additional constraints and a need to extend the backwards planning model. To establish a week planning with the backwards planning approach a number of steps has to be followed as presented in figure 13 and explained below. Customer demand is known for the upcoming years for each program and every week the MRP data of the case company is refreshed to include changes in demand. Based on the MRP data concerning the latest starting times of operations (1), the backwards planning model can develop an efficient
𝑷𝒂𝒓𝒂𝒎𝒆𝒕𝒆𝒓s 𝐿𝑖= 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑙𝑎𝑦𝑢𝑝 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑖 𝑅𝑚𝑖𝑗 = 𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 𝑚𝑜𝑢𝑙𝑑𝑠 𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑖 𝑖𝑛 𝑤𝑒𝑒𝑘 𝑗 𝑅𝑡𝑗= 𝐻𝑜𝑢𝑟𝑠 𝑙𝑎𝑦𝑢𝑝 𝑡𝑎𝑏𝑙𝑒 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑤𝑒𝑒𝑘 𝑗 𝑅𝑙𝑗= 𝐻𝑜𝑢𝑟𝑠 𝐿𝑃𝑆 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑤𝑒𝑒𝑘 𝑗 𝑅𝑤𝑗= 𝐻𝑜𝑢𝑟𝑠 𝑤𝑜𝑟𝑘𝑒𝑟 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑤𝑒𝑒𝑘 𝑗 𝑪𝒐𝒏𝒔𝒕𝒓𝒂𝒊𝒏𝒕𝒔 (8) 𝑋𝑖𝑗 ≤ 𝑅𝑚𝑖𝑗 ∀ 𝑖, 𝑗 (9) ∑(𝑋𝑖∗ 𝐿𝑖) ≤ 𝑅𝑡𝑙 𝑁 𝑖=1 ∀ 𝑗 (10) ∑(𝑋𝑖∗ 𝐿𝑖) ≤ 𝑅𝑙𝑗 𝑁 𝑖=1 ∀ 𝑗 (11) ∑(𝑋𝑖∗ 𝐿𝑖) ≤ 𝑅𝑤𝑗 𝑁 𝑖=1 ∀ 𝑗
27
planning with regard to utilization of the maximal bath size of the autoclave process (2). With the output data of the model, the planners can determine which articles of each program have to be processed by the autoclave process in a week. The model is able to weekly determine the backwards planning for each program based on the refreshed MRP data to consider the latest changes in demand. In order to ensure that enough time is available to plan the latest starting times of the preceding operations of the autoclave process the planning of a particular week has to be developed two weeks in advance. In addition, the week planning has to be developed before a strict deadline to prevent for late adjustments which can cause inefficient batches and additional charges for the autoclave process. The backwards planning model simultaneously considers input- and output control at the job release level of the workload control concept. This means that the efficiency of batches and the availability of capacity is considered simultaneously. When customer demand for a program increases, the constraints of the backwards planning model concerning availability of resources and workload control have to be adjusted. If this is neglected, the model cannot come up with a solution when the parameter of customer demand exceeds the constraints. This causes that when the backwards planning model cannot give a solution, capacity has to be added to the layup process. By changing the capacity and workload control parameters of the model it can be indicated if enough capacity is available to process the additional articles resulting from increasing demand.
The planners of the programs have to assess the feasibility of the backwards planning (3). If the planning of a certain week appears to be not feasible by unexpected circumstances as breakdowns of moulds or loss of worker capacity by illness, the planning of the model has to be adjusted and the MRP system will replan the unreleased articles. When it is determined which articles of the programs are processed in a week, the autoclave planning has to be developed containing the starting times of the charges of different curves (4). At the development of the autoclave planning it is important to consider the limited availability of resources for the layup process. Based on the starting times of the charges, the latest starting times of the layup and precutting processes have to be planned upstream from the autoclave process based on the lead times of the operations (5). If an autoclave charge is planned in the beginning of a week, the latest starting times of the layup and precutting operations have to be planned in the preceding week. When the autoclave planning is developed and the latest starting times of the preceding operations are determined, the week planning is ready to be released (6).
MRP planning Backwards planning
Check feasibility planning Determine starting times charges autoclave Determine latest starting times of predecessors Release week planning (1) (2) (3) (4) (5) (6)
28 7. Discussion
The results of this research and the associated implications are discussed in this section. In the first part the results related to the case and opportunities for future research are discussed. The second part discusses the contribution of this research to academic knowledge and its limitations.
7.1 Related to the case
Research is performed to efficient batching for the autoclave process while considering the resource dependent capacity of the layup process. Based on the MRP data, the number of charges that have to be planned each week for the autoclave process is determined. Analysis showed that the MRP planning results in an average utilization of the maximal batch size of 0.696 for the autoclave process. In order to identify if there is potential to increase the utilization and reduce the number of charges needed, two backwards planning approaches are analyzed. Analysis showed that there is substantial potential to reduce the needed charges compared to the MRP planning, up to 46.5% for the limited backwards approach and even 56.6% for the unlimited backwards approach. This results in an average utilization of the maximal batch size of 0.918 for the limited and 0.982 for the unlimited backwards planning approach. The implications of efficient batching for preceding processes of a simultaneous batch process and the limitations of the planning approaches are discussed below.
MRP data of the analyzed program A380/A400 shows fluctuations which is caused by an uneven distribution of demand over the weeks. In addition, spare parts, backlogs and rework cause additional work in particular weeks. Analysis showed that the backwards planning approaches establish a planning with more stable workloads distributed over the weeks compared to the MRP planning. However, there is potential for more stable workloads if backlogs and rework are reduced and customer demand is better regulated at the job entry level of the workload control concept. In such a situation also a cyclic planning approach can be applied which has the advantage of reduced variability and established routine by its repetitive behaviour. The fluctuations in demand cause that the cyclic planning approach is not appropriate in the current situation.
29
department has actually sufficient capacity to process the demanded articles on time for the autoclave process. Based on the autoclave planning, the starting times of the layup and precutting processes have to be planned upstream from the autoclave process based on the lead times of the operations. Planning releases of articles backwards causes inventory at succeeding processes in the machining part of the production process. For the backwards-L planning approach on average 22.0% of the articles of the analyzed program A380/A400M have to be stored approximately 1.5 weeks after the debagging process. The backwards-U planning approach results in substantially more inventory because no limited amount of weeks to plan articles backwards is considered. This causes that on average 48.5% of the articles is buffered after the debagging process approximately 5.5 weeks. The limited amount of available square meters causes problems to store semi-finished products and controlling working capital is important for the case company. The cyclic planning approach will cause even more inventory because it does not consider the inventory objective of the backwards planning approach. Posthuma (2014) proposed in his research a buffer after the debagging process with a release decision. This causes that semi-finished articles have not to be stored in front of operations at the machining process but instead can be stored in a separate warehouse. Further analysis has to be performed to determine which backwards planning approach has to be selected based on the tradeoff between the cost saving by efficient batching for the autoclave and increased costs of resulting inventory.
A limitation of this research is that in reality unexpected circumstances such as rework and mould unavailability can affect the applicability of the backwards planning approach. Secondly, only data of the year 2015 is analyzed which causes high workloads in the first months of the year. However, in reality, the high workloads in the beginning of the year will be lower because articles then are planned backwards to the preceding year. This results in more balanced workloads but also causes additional inventory because more articles are planned backwards. At the application of the backwards planning approach it is important to consider that materials for the precutting process have to be purchased earlier than the MRP planning indicates. In addition, late adjustments of the week planning can cause inefficient batches for the autoclave process which makes a strict deadline for the final week planning important. Finally, when backwards planning is applied, the measure for the on time start has to be adjusted to allow earliness resulting from releases of articles before the latest starting time.
30
the backwards planning approach can be extended to be applicable for all programs. This will be the next step towards a planning model that establishes efficient batches for the autoclave process while considering all applying constraints of the production process of Fokker Aerostructures.
7.2 Contribution to academic knowledge
Workload control integrates input and output control to regulate the in- and outflow of work of a production process (Oosterman et al., 2000). Research is needed to sequencing of orders at the input control job release level (Thürer et al. 2015). For a simultaneous batch process characterized by incompatible product families, sequencing of orders is important to establish efficient batches (Posthuma, 2014). In addition, preceding operations of the simultaneous batch process need sufficient capacity to process the orders on time such that the desired sequence is enabled at the simultaneous batch process (Cransberg et al., 2015). Research is needed to the complexities that affect the opportunity to adjust capacity in reality (Land et al., 2015; Thürer et al., 2016). The involved case company of this research has a simultaneous batch process with incompatible product families. In addition, the preceding processes are characterized by resource dependent capacity. The mentioned gaps in literature are addressed by researching how input- and output control can be combined at the job release level to establish efficient batches for a simultaneous batch process characterized by incompatible product families.
31
successive processes. Limited availability of such resources causes that articles demanding for the same resource cannot be batched but have to be allocated to different batches. Otherwise orders will compete for the same resources. The resource dependent capacity has to be sufficient to deliver orders on time such that the desired sequence is enabled at the simultaneous batch process.
32 8. Conclusion
This research explores the opportunity to establish efficient batches for a simultaneous batch process characterized by incompatible product families by combining input and output control at the job release level. With this research the need for insights in sequencing of orders at the job release level is addressed. In addition, it investigates the complexities that affect the opportunity to adjust capacity in reality. For a simultaneous batch process it is important to consider the utilization of the maximal batch size. Two planning approaches to establish efficient batches are designed and analyzed: limited- and unlimited backwards planning. Also the possibility of application of a cyclic planning approach is considered. The results indicate that the efficiency of batches can be substantially improved by planning articles backwards. Fluctuating customer demand results in inventory and increased working capital when articles are planned backwards. This causes that a cyclic planning is most appropriate in a situation with stable demand because otherwise substantial costs of inventory are incurred. Considering sequencing to establish efficient batches for a simultaneous batch process by backwards planning results in more balanced workloads. In addition, application of a backwards planning approach results in a more balanced workload for preceding and succeeding processes. In order to deliver orders on time and enable the desired sequence at the simultaneous batch process, capacity of preceding processes has to be considered. Resource dependent capacity causes complexities to determine the capacity of a process. The availability of each resource has to be considered individually at the development of a planning because different orders can demand for varying resources. In addition, it is important to consider resources such as moulds that are allocated during operations at successive processes. Limited availability of such resources causes that orders that demand for the same resource cannot be batched. The capacity of the preceding operations has to be sufficient to deliver orders on time and enable the desired sequence at the simultaneous batch process.
33 References
Bertrand, J.W.M. and Wortmann, J.C., 1981. Production control and information systems for component-manufacturing shops. Dissertation, Elsevier, Amsterdam. Tile Netherlands.
Chiang, T.C., Cheng, H.C., Fu, L.C. 2010. A memetic algorithm for minimizing total weighted tardiness on parallel batch machines with incompatible job families and dynamic job arrival. Computers & Operations Research 37(12): 2257–2269.
Cransberg, V., Land, M., Hicks, C., Stevenson, M. 2015. Handling the complexities of real-life job shops when implementing workload control: a decision framework and case study, International Journal of Production Research, 54(4):, 1094-1109.
Ernst, A. T., Jiang, H., Krishnamoorthy, M., Sier. D. 2004. Staff Scheduling and Rostering: A Review of Applications, Methods and Models. European Journal of Operational Research 153(1): 3–27.
Fowler, J.W., Hogg, G.L., Mason, S.J. 2002. Workload Control in the Semiconductor Industry. Production Planning and Control 13(7): 568–578.
Fredendall, L. D., Ojha, D., Patterson, J.W. 2010. Concerning the Theory of Workload Control. European Journal of Operational Research. 201(1): 99–111.
Goldratt, E.M., Cox, J. & Whitford, D. 1992. The goal: the theory of constraints, a process of ongoing improvement. North River Press.
Hendry, L.C. and Kingsman, B.G. 1989. Production planning systems and their applicability to make to order companies. European Journal of Operations Research. 40: 1–15.
Hopp, W., M. L. Spearman. 1996. Factory Physics: Foundations of Manufacturing Management. Irwin, Chicago, IL.
Jia, Z., Li, K., Leung, J.Y.T. 2015. Effective heuristic for makespan minimization in parallel batch machines with non-identical capacities. International Journal of Production Economics. 169: 1-10. Jula, P., Leachman, R.C., 2010. Coordinated Multistage Scheduling of Parallel Batch-Processing Machines Under Multiresource Constraints. Operations Research 58(4): 933-947.
Karlsson., C. 2009. Researching Operations Management. Taylor & Francis, Inc.
Land, M.J. and Gaalman, G., 1996. Workload control concepts in job shops - a critical assessment. International Journal of International Production Economics 46–47: 535–548.
Land, M. J., and G. J. C. Gaalman. 1998. The Performance of Workload Control Concepts in Job Shops: Improving the Release Method. International Journal of Production Economics 56–57: 347–364. Land, M. J., M. Stevenson, and M. Thürer. 2014. Integrating Load-based Order Release and Priority Dispatching. International Journal of Production Research 52(4): 1059–1073.
34
Li, Z., Chen, H., Xu, R., Li, X. 2015. Earliness–tardiness minimization on scheduling a batch processing machine with non-identical job sizes. Computers & Industrial Engineering. 87(9): 590-599.
Liu, J.J., Li, Z.T., Chen, Q.X., Mao, N. 2016. Controlling delivery and energy performance of parallel batch processors in dynamic mould manufacturing. Computers & Operations Research. 66(2): 116-129. Mathirajan, M., Bhargav, V., Ramachandran, V. 2010. Minimizing total weighted tardiness on a batch-processing machine with non-agreeable release times and due dates. International Journal of Advanced Manufacturing Technology 48(9–12): 1133–1148.
Oosterman, B., Land, M.L. and Gaalman, G. 2000. The influence of shop characteristics on workload control. International Journal of International Production Economics 68(1), 107–119.
Pearn, W., Hong, J., Tai, Y. 2013. Burn-in test scheduling with batch dependent processing time and sequence dependent setup time. International Journal of Production Research. 51(6): 1694–1706. Posthuma. S. 2014. Workload control with simultaneous batching of incompatible product families. Master Thesis, University of Groningen.
Soepenberg, G.D., Land, M.J., Gaalman, G.J.C. 2012. Workload Control Dynamics in Practice. International Journal of Production Research 50(2): 443–460.
Stevenson, M. 2006. Refining a workload control (WLC) concept: a case study. International Journal of Production Research. 44(4): 767–790.
Stevenson, M., and C. Silva. 2008. Theoretical Development of a Workload Control Methodology: Evidence from Two Case Studies. International Journal of Production Research 46 (11): 3107–3131. Thürer, M., C. Silva, M. Stevenson, and M. J. Land. 2012. Improving the Applicability of Workload Control (WLC): The Influence of Sequence-dependent Set-up times on Workload Controlled Job Shops. International Journal of Production Research. 50(22): 6419–6430.
Thürer, M., Land, M.J., Stevenson, M., Godinho Filha, M. 2015. Concerning workload control and order release: the pre-shop pool sequencing decision. Production and Operations Management. 24(7): 1179-1192.
Thürer, M., Stevenson, M., Land, M.J. 2016. On the integration of input and output control: workload control order release. International Journal of International Production Economics. 174(4): 43–53. Uzsoy, R. 1995. Scheduling batch processing machines with incompatible job families. International Journal of Production Research. 33(10), 2685–2708.
