WIP OPTIMISATION FOR THE FOOD PROCESSING INDUSTRY A case study of scheduling and Shop floor control methods
Bauwiena Visser
A thesis presented for the degree of:
MSc Technology & Operations management and
MSc Operations & Supply chain Management
Contents
1 Introduction 3
2 Theoretical background 5
2.1 Work in process . . . 5
2.2 Food processing industry . . . 6
2.2.1 WIP in the Food Processing industry . . . 7
2.3 Production Control . . . 8
2.3.1 Strategy . . . 8
2.3.2 Tactical . . . 9
2.3.3 Execution . . . 9
2.4 Production control in FPI . . . 9
2.4.1 Sequencing and scheduling . . . 9
2.4.2 Sequencing rules . . . 10
2.5 Shop Floor Control . . . 11
2.6 Lean . . . 14
2.6.1 5S . . . 14
2.6.2 Other visual aids . . . 15
3 Methodology 16 3.1 Research questions . . . 16
3.2 Selection of methods. . . 16
3.3 Case study . . . 17
3.4 Data collection and data analysis . . . 17
3.5 Validation of effectiveness . . . 19
3.6 Validity and reliability . . . 19
4 Analysis 20 4.1 Scheduling at the case-company . . . 20
4.1.1 Tactical decision: MPS . . . 20
4.1.2 Production schedule . . . 20
4.1.3 Scheduling decisions . . . 22
4.2 Scheduling in the FPI . . . 23
4.2.1 Applying sequencing rules . . . 24
4.2.2 Overall optimisation . . . 27
4.2.3 Additional optimisation . . . 28
4.3 Effectiveness of Shop Floor Control measures in the food industry . . . . 29
4.3.1 Material handling at the shop floor . . . 29
4.3.2 Classification of SFC-methods . . . 31
4.3.3 Lean . . . 33
5 Validation 34 5.1 Drum-buffer-rope . . . 34
5.2 Hybrid CONWIP . . . 35
5.3 Hybrid two-boundary control system (TBC) . . . 35
1
Introduction
For many years it was a common policy for food processing companies to produce in large batches to keep production costs low and limit the number of set-ups (Van Donk, 2001). However, in the last decade changes to consumer behaviour and wishes have led to an increase in packaging sizes and number of products. This increase in variety of products has resulted in more complex management and the need to increase flexibility of facilities (Nakhla, 2006).
Currently, the food processing industry is using WIP buffers to decouple the packing and processing departments in order to provide this added variability. Therefore an important area that can influence a manufacturer’s operational efficiency is in its management of its work in process (WIP) inventories (Wilson, 2013). WIP buffers, defined as inventory found anywhere after the end of the first step in a manufacturing process and before the final step in a manufacturing process (Conway et al., 1988), is one type of inventory found in manufacturing operations; the other two being raw material inventory and fin-ished goods inventory. Efforts to efficiently manage WIP inventory have been studied extensively, by researchers such as Kim and Lee (2001), Askin and Krisht (1994) and Powell (1994). Their work has contributed to our understanding of areas such as the optimal size of WIP buffers and the optimal location of these buffers.
However, the majority of research done in optimising WIP has focused on assembly-type operations, with little work done in the food processing industry (FPI) (Wilson, 2013) Due to the low margin on food products in general, only low investment solutions optimising WIP are applicable in order to still get a good return on investment. Moroever, additional complexity is due to special food processing industry characteristicsl such as the perishability of products, which require efforts to reduce lead times as much as possible (Mahalik and Nambiar, 2010) to avoid wasting product that cannot be sold as remaining shelf life expires.
Most of the facilities in the food processing industry have an unbalance,in terms of ca-pacity, between the processing and packaging department requiring WIP buffers in be-tween the processes. Dealing with these WIP buffers requires additional care in the food industry, due to spoilage, allergens and other food safety issues (Jackson et al., 2008). Furthermore, EU-regulations requires food hazards to be controlled and thus WIP buffers must follow strict regulation. One way of controlling food hazards is by implementing HACCP, which is a systematic approach to identifying, assessing, and controlling hazards (Sanders, 1999) . Last, deterioration of products during the buffer time puts a restriction between successive processes (Flapper et al., 2010).
A key step to controlling WIP in the food industry is to first optimise the work in process (WIP) level. Research is done in scheduling, by mostly scheduling rules are simplified versions of practice (Hopp and Spearman, 2011). Moreover, very complex methods are necessary to solve the problem and usually are not understand by practitioners. Moreover, scheduling alone will not eliminate buffers when dealing with an unbalanced line problem, some buffers are always required and wanted (Conway et al., 1988).
In literature several methods are discussed namely: drum-buffer-rope, kanban, CONWIP and several Lean methods. However, in reviewing the literature regarding buffer manage-ment in the food industry limited sources can be found. A review of perishable buffers was performed for the dairy industry by Wang et al. (2010). In their study they also found limited sources for dealing with buffers in the food industry.
WIP buffers in the food are specific due to the earlier mentioned regulations and other requirements. Especially due to the perishable nature of the products, that only allows them to spend a limited amount of time inside the manufacturing system. If the system flow time, or lead time, exceeds certain fixed thresholds, the product has to be considered as a defect and has to be scrapped by the system (Colledani et al., 2014).
As such, the problem of scheduling and buffering of perishable products have only spo-radically been addressed for simple systems, and no appropriate method for a simple scheduling heuristic for perishability has been found in the current literature regarding unbalanced production lines. The goal of this paper is to contribute to this end. Specif-ically, to investigate scheduling of unbalanced lines with perishability and management of buffers in the food industry.
The research question is;
“What are effective methods for reducing and controlling WIP in the food processing industry?”
2
Theoretical background
In the theoretical background literature regarding WIP in the FPI is discussed. First, the strategic value of WIP is explained. Second, the food processing industry in general and the special characteristics in relation to WIP are discussed. Next, literature regarding WIP optimisation (i.e. reduction and control) is reviewed. In this part, first, the role of production control in relation to WIP optimisation is explored. Second, the role of sequencing methods applied in the food industry are critical reviewed. Finally, the shop floor control literature is discussed.
2.1
Work in process
Interest in work in process (WIP) inventory has been high as it is one of the target criteria in production logistics. L¨odding et al. (2003) even describes is as of ”strategic importance for commercial success”. In literature WIP is normally defined as; ’inventory after the first step in manufacturing and before the last.’ (Conway et al., 1988). Different opinions exist regarding WIP. Methods such as Lean and theory of constraints (TOC) consider WIP to be unnecessary, wasteful and it’s a goal to fully eliminated WIP (Wilson, 2013). However, other methods consider WIP usesful, as inventory cannot be excess when it is the right quantity of the right goods at the right place at the right time (Crandall and Crandall, 2003).
In this research the main purpose of WIP is to give each stage of a production system, some degree of independent action (Conway et al., 1988). Independent action is needed to handle uneven production rates in decoupled serial operations (Wilson, 2013). In these situations, WIP can be used as a method of protection between working processes (Schragenheim and Ronen, 1990). Specificially, to ensure machines in the process are neither blocked (e.g. machine must wait to dispose its finished piece before it can) or starved (e.g. next machines is waiting for material). In both blocking and starving cases a continuous processing is prevented, causing loss of production.
By studying WIP inventory it is useful to apply Little’s Law: L = λ W,
where L is the expected number of units in the system, λ the average arrival rate of items to the system, W the expected time an item spends in the system (Little, 1961). Little’s law implies that for fixed throughput, reducing WIP and reducing cycle time are directly linked. Therefore, the measures to increase the efficiency of WIP are the same as those one would use to reduce cycle times.
Hopp and Spearman (2011) refined Little’s Law in terms of WIP, cycle time and through-put:
Throughput (TH) = WIP/Cycle Time (CT),
that the part spends as WIP.
Reducing WIP is therefore directly related to reducing throughput. Moreover, decisions relating to the optimal size and optimal placement of buffers are critical.
The issue of optimal WIP size mainly focuses on not carrying excessive amounts of WIP inventory. Tsourveloudis et al. (2000) specifies four reasons for maintaining low WIP levels:
• WIP inventories tie up capital which create no profit for the company;
• High WIP inventories increase cycle times and reducing customer responsiveness; • High WIP inventories occupy more warehouse space and increase the need for
ma-terial handling equipment;
• WIP inventory increase risk of loss due to spoilage and obsolescence.
The majority of research done in optimising WIP has focused on assembly-type opera-tions, with little work done in process-type industries, such as the food industry (Wilson, 2013). A discussion regarding the special characteristics of the food processing industry and its consequence for handling WIP is discussed in the next section.
2.2
Food processing industry
The food sector is based on a very diverse group of products with different degrees of perishability, varied manufacturing lead times, and demand in different amounts at different frequencies (Dora et al., 2015). Despite the diverse nature of this industry as a whole, the huge variation in quality of raw materials and their highly unpredictable supply as well as volatile customer demands make the manufacturing sector quite unique. In the table 1 the unique characteristics of the food processing industry are described based on literature (Wezel et al., 2006; Wilson, 2013; Dora et al., 2015).
Component Characteristics of the food processing industry
Plant characteristics Expensive and single-purpose capacity coupled with small product variety and high volumes Flow shop oriented design
There are long (sequence-dependent) set-up times between different product types Plants are batch processing and have two to six production lines
Processing and packaging are separated because of food quality assurance Small and single-site factories with 30 to 100 employees
Product characteristics Variability in quality of raw materials and supply due to unstable yield of farmers. In contrast with discrete manufacturing, volume or weights are used.
Highly perishable which brings along product quality and safety considerations Production process characteristics Processes have a variable yield and processing time
Short throughput time for batches (i.e. between one to eight hours) At least one of the processes deals with homogeneous products The processing stages are not labor intensive
Production rate is mainly determined by capacity
Food industries have a divergent product structure, especially in the packaging stage Factories that produce consumer goods can have an extensive, labor-intensive packaging phase
Due to uncertainty in pricing, quality, and supply of raw material, several recipes are available for a product
2.2.1 WIP in the Food Processing industry
What typifies the FPI is their multiple serial production processes consisting of two main stages: processing followed by packaging (Akkerman and van Donk, 2009). Moreover, FPI is characterised by a divergent product structure, wherein one source material is packaged into multiple SKU’s (e.g. due to consumer wishes for multiple sizes and pack-aging forms) (Dora et al., 2015). The production and packpack-aging stages are separated by WIP buffers, which allows source materials to be packaging on multiple packaging lines in multiple SKU’s. However, high capacity differences between the production and pack-aging process can occur. WIP is therefore used as a safety stock, to reduce the impact of long setup-times and variable yield of the processing department on the packaging department (Van der vorst et al., 2007).
Food safety
An additional element in food processing industry when considering WIP is food safety. Specifically, food allergens which are considered a major health risk (Van Hengel, 2007). According to Food safety legislation and regulations EU (2002), all products require identification of ingredients on the food label to protect consumers with food allergies. Therefore a food processor must ensure proper control and separation of all processes and WIP to prevent cross-contamination of products. Moreover, production facilities that produce and store multiple products and buffers have a high risk of cross-contamination (Akkerman et al., 2010). Risk of cross-contamination in buffers is mostly due to accidental contamination caused by; mix-ups, improper production sequencing and or, inadequate cleaning of processing and packaging machines.
Another key characteristic of FPI is perishablity of WIP buffers. Dependent on the type of products, this might range from a couple of hours to multiple days. However, all WIP buffers in the FPI have a maximum shelf-life period and must be processed before the expiration date (Flapper et al., 2010). When the maximum shelf-life period is exceeded the stored product must be discharged as waste, due to food safety regulation. Not only is this a waste of valuable resource, food waste also has to be discarded properly and is therefore costly.
Storage
2.3
Production Control
In production systems, control of WIP (i.e. proper placement, etc.) is important in order to create flow in the production. Methods in the different layers of the production control hierarchy can be useful to optimise WIP. The purpose of production control is to ensure that production output closely conforms with demand (Nicholas, 2011). Ideally, it ensures that products are made in the required quantities, at the right times, with the highest quality. It should do these things for the lowest cost and enable production problems to be easily identified and remedied. Figure 1 shows the production control framework of Higgins et al. (1996) these elements of production control are described below.
Figure 1: Production Control framework
2.3.1 Strategy
The upper layer of the framework contains the long-term strategic planning. The basic function of the tools, in the upper layer, are to establish a production environment capable of meeting the plant’s overall goals (Hopp and Spearman, 2011).
As shown in figure 1 the starting point of production planning systems is forecasting. In forecasting, marketing and demand information is used to generate the future demand, and to specify when and how much to produce of a particular product (Winters, 1960). Once a forecast of future demand is created, decision are made regarding the execution, wherein the physical capacity of a production plan must be matched with the forecasted demand (Hopp and Spearman, 2011). These decisions regarding capacity concern, how much and what kind of equipment is required. Additional influencing factors for these decisions are the process requirements for making the various products and flexibility of the production facililties required.
aggregate plan is generated. This aggregate plan specifies how much of each product is produced over time, based on priorities and operating characteristics of the plan.
2.3.2 Tactical
The tactical tools take the long-range plans from the strategic level, along with informa-tion about customer orders, to generate a plan of acinforma-tion that will help the plant prepare for upcoming production (by procuring materials, lining up subcontractors, etc.) (Hopp and Spearman, 2011). Generally, it is at the tactical level a WIP/quota is set that limits the overall WIP in the process. Next, an aggregate plan is created from which a more specific schedule with the appropriate factors included for each scheduling period can be created.
Master Production schedule
In the MPS, the aggregated gross production demand is used in order to determine a weekly required production schedule (i.e. gross requirement). In general the MPS contains; current inventory status, the status of outstanding orders known as scheduled receipts.
2.3.3 Execution
In the execution layer the direct control and weekly plans are discussed, such as Shop floor control which regulates the real-time flow of material through the plant in accordance with the schedule (Hopp and Spearman, 2011). Shop floor control most important objective is to minimise lead times and WIP, primarily by accurate control over which orders are released on the shop floor and how (Higgins et al., 1996).
2.4
Production control in FPI
Due to the downward price pressure by retailers and the overall low margins on food products (Van Donk, 2001) the food processing industry focuses on low cost improvements and solutions at the operational level. As discussed in the introduction, this research will therefore focus on WIP optimisation through scheduling and WIP control via shop floor control methods.
2.4.1 Sequencing and scheduling
In the literature, scheduling is discussed separately for both flow-shop and job-shops, due to their different requirements and characteristics. In a flow shop, jobs are processed on machines in a set order. Increases in the variety of products (e.g. currently more types, mixes and sizes of consumer packaging are required) has lead to the need for greater flexibility of facilities. This result that not only flow shop designs are present in the FPI but also job shop designs are used. In a job shop environment, the sequence depends on each job. Each item has their own order to be processed on machines and may take a different path than other jobs.
Scheduling unbalanced production processes with multiple machines in literature is called Job shop scheduling (JSSP), a classical operations research problem that has been consid-ered as a hard combinatorial optimisation problem since the 1950s (Chaudhry and Khan, 2016). Furthermore, Garey and Johnson (1979) showed that in terms of computational complexity, JSSP is NP-hard in the strong sense. Therefore, even for very small JSSP instances, an optimal solution cannot be guaranteed.
In a job shop, every job may have a separate processing sequence, in the general JSSP, there is a finite set of n jobs to be processed on a finite set of m machines (Garey and Johnson, 1979). Each job comprises a set of tasks that must be performed on a different machine, and in specified processing times, in a given job-dependent order. A typical objective of this process is to minimise the total completion time required for all jobs or makespan. In recent years, several scheduling heuristics and algorithms have been developed to optimise job shop scheduling (Chaudhry and Khan, 2016). However, as was found during the case study, most of these scheduling rules are hard to implement and understand by planners.
As a result, most planners cannot or will not use these advanced scheduling heuristics or algorithms (Wezel et al., 2010) resulting in lower unoptimised scheduling and more overall WIP. In some cases simple sequencing rules such as Shortest Processing Time and Shortest Setup times are used in the food processing industry (Nakhla, 2006) by planners. However, in complex situations just applying a simple sequencing rule is not always succesful. In this case reducing complexity, using combinatorial rules, and than applying these heuristics has been found to be an effective method (Fernandes and Louren¸co, 2008).
2.4.2 Sequencing rules
Next, several simple scheduling are considered which can be usefull for reducing WIP in both flow-shops and job shops. A review by Gupta and Bector (1989) andNakhla (2006) was used to find the following method.
Shortest processing time (SPT)
Shortest setup time (SST)
When setup times are significant, it is desirable to batch together jobs from the same family in such a way that deadlines are met (Woodruff and Spearman, 1992) and setup time minised. This will reduce the time spent on setups which increases the capacity and overall throughput time. In the food processing industry long-setup times are relevant, and by applying SST, jobs will finish earlier and therefore WIP will can be decreased. Shortest weighted processing time (SWPT)
SWPT or shortest weighted processing time is an extension to the SPT with an additional weight factor included. The jobs are prioritised based on the processing times multiplied by an additional weight factor.
Lot splitting
Russell and Fry (1997) found that splitting process batches into smaller transfer batches nearly always improved shop performance criteria. Lot splitting means that the batch is not produced in one processing batch, but can be separated. Small jobs clear out more quickly than large jobs, improve performance with regard to average cycle time and machine utilisation (Hopp and Spearman, 2011). However, small batches result in lost capacity due to an increased number of setups.
Release list
The job releasing function controls which jobs are allowed to reach the shop floor as well as how many jobs are on the floor. The releasing function, then, controls the level of WIP inventory (?). By proper control of WIP the throughput can be decreased which result in lower WIP in the system.
Combination of rules
The combination of certain rules with a waiting time under a certain threshold is used to avoid excessive delays in long operations (Nakhla, 2006).
2.5
Shop Floor Control
After scheduling has resulted in an optimal production plan, which balances changeovers and buffers, Shop Floor Control (SFC) systems help to organise buffers. Burbidge (1990) defines SFC as ”the function of management which plans, directs and controls the ma-terial supply and processing activities in an enterprise”. The most important activity is ordering. Ordering is concerned with regulating the supply of both manufactured parts and bought items, in order to meet the production programme. The main contribution of a SFC-method is to coordinate the materials and information flow onto the shop floor (Fernandes and Godinho Filho, 2011).
narrowly, as purely material flow control. No scheduling system can anticipate random disruptions, but the SFC module must accommodate them anyway. Next, the range of functions of shop floor control are given.
Range of functions
Hopp and Spearman (2011) discuss the range of function of a shop floor control method. Figure 2 illustrates the range of functions one can incorporate into the SFC module: ma-terial flow control, WIP tracking, status monitoring, throughput management, capacity feedback, quality control, and work forecasting. The functions are discussed below. Material flow control
At the center of these functions is material flow control (MFC), without which SFC would not be shop floor control. Hopp and Spearman (2011) defines material flow control as ”the mechanism by which to decide which jobs to release into the factory, which parts to work on at the individual workstations, and what material to move to and between workstations”.
Figure 2: Range of functions WIP tracking
Status monitoring
Besides the WIP positions other parameters need to be monitored. Status monitoring refers to surveillance parameters such as machine status or staffing situation (Hopp and Spearman, 2011). Without these the material can not flow through the system.
Throughput management
Throughput management measures the output of the line and compares it with the set quota (Hopp and Spearman, 2011). An example of a specific mechanism for monitoring the system is statistical throughput control (STC). STC tracks progress toward making the periodic production quota (Hopp and Roof, 1998). The system collects informa-tion about real-time status which makes it a useful place to collect and process some information about future events.
Capacity feedback
A different function of the SFC module is the collection of data to update capacity measures (Hopp and Spearman, 2011). This capacity feedback function is important for ensuring that the high-level planning modules are consistent with low-level execution. Quality control
Different move points in a system give the opportunity for establishing quality assurance. This links the SFC-method with quality control. If the operator of a downstream work-station has the authority to refuse parts from an upstream workwork-station on the basis of inadequate quality, then the SFC module must recognise this disruption of a requested transaction (Hopp and Spearman, 2011). As already discussed is quality control an im-portant function in the food processing industry.
Work forecasting
When work is forecasted it is possible to specify when and how much to produce of a particular product (Winters, 1960). A function of SFC is managen the material supply and processing activities in a company (Burbidge, 1990).
SFC Methods
”Effective production control systems are those that produce the right parts, at the right time, at a competitive cost. In literature some successful production planning and control systems are discussed.
Fernandes and Godinho Filho (2011) divided shop floor control methods in four classes: order-controlled systems, stock level-controlled systems, flow-scheduled systems, and hy-brid systems. The characteristics of the systems are discussed below:
1. Order-controlled systems: There is no stock of final items, once production is carried out according to customers’ specifications.
3. Flow-scheduled systems: The release of an order is based on a centralised schedul-ing drawn up by the plannschedul-ing department. This centralised schedule pushes the production.
4. Hybrid systems: These have characteristics of stock level-controlled systems and flow-scheduled systems.
2.6
Lean
Next to sequencing and SPC-methods, Lean practices in combination with visual manage-ment can be used to increase efficiency and reduce costs of production in the food sector (Mahalik and Nambiar, 2010) (Goncharuk, 2009). Lean manufacturing is a system that utilises less inputs of materials, hours and creates the same outputs while contributing more value to customers (Womack et al., 1990). Lean practices can be used to increase efficiency and reduce cost, however, only a limited number of studies have focused on the adoption of lean principles for buffers in the food sector (Dora et al., 2015). Some basic tools will work without adaption such as; short interval control (SIC) which look at if production plans are met or statistical process charts (SPC) that measure the variability in the process. However, other tools have not been discussed in literature, such as visual management in the form of 5s, visual aids (i.e. process charts, graphical representation, colour coding, symbols) and poka-yoke. A key driver of these techniques is that every person involved must be able to see and fully understand the different aspects of the process and its status at any time through visual aids (Parry and Turner, 2006)
2.6.1 5S +
One of the most common used visual control tool is 5S (Parry and Turner, 2006). 5S consist of five different elements that starts with the letter ‘S’ in the Japanese language (Randhawa and Ahuja, 2017) and that are repeatedly performed and controlled to sim-plify and organise the workplace or buffer zones.
1. Seiri – sort, clearly distinguish what is needed and what is not needed;
2. Seiton – simplify, organise systems logically to make it easier for others to find, use and return tools to the original position;
3. Seiso – sweep, keep things clean;
4. Seiketsu – standardize, maintain and improve the first 3Ss;
5. Shitsuke – self-discipline, correct procedures as habit; think about how they can be improved.
zones make them easier to maintain from a hygiene point of view (Dudbridge, 2011). Commonly food clutter (which 5S sorts) in the food production/buffer areas are;
• Last week’s production plans • Production pens
• Machine parts
• Scrapers or other hand tools
• Raw materials for a new product trial • Spare tickets/labels
• Redundant packaging
Applying 5S with a continuous improvement plan for buffers in the food industry reduces the chance of cross-contamination and organises the workplace. A more generalised ap-proach on the effectiveness of this method has not yet been researched and is considered in this research.
2.6.2 Other visual aids
Besides 5S the lean manufacturing toolsets includes visual tools that form an important part of the communication process which drives a factory. Examples of visual tools are: process flow charts, value stream mapping (VSM), Ishikawa diagrams, Andon boards (i.e. illuminated overhead displays providing information about the current status of production and emerging problems) (Parry and Turner, 2006). However, most of these tools are not directly useful for buffer management.
3
Methodology
In the introduction a research question was proposed, in the following chapter the method-ology for answering this questions is described. First, the research questions and sub-questions are re-stated and discussed. Next, a description of methods to identify new WIP reduction methods is discussed. Wherein; the data collection via a case study and the data analysis are discussed. Followed by a short discussion upon validity and reliability of the current methodology.
3.1
Research questions
The research question is; “What are effective methods for reducing and controlling WIP in the food processing industry?” In order to answer the research question multiple sub-questions are formulated to support the research question.
RQ: What are effective methods for reducing and controlling WIP in the food processing industry?
In the research questions two main topics are discussed, reducing WIP and controlling WIP in the food processing industry. Therefore, several sub-questions arise;
SQ1: What are effective methods to reduce WIP in the food processing industry? SQ2: What are effective methods to control WIP in the food processing industry?
3.2
Selection of methods.
The theoretical background already discussed some of the most important methods used to reduce WIP, such as planning, scheduling and scheduling related methods. In order to determine which methods are effective multiple characteristics of the FPI are discussed which were found in literature and from a case study company.
Next, a literature review is performed for methods applicable to reduce WIP in the food processing industry (FPI). Herein; backwards and forward search is used on key articles of (Gupta et al., 1999; Dora et al., 2015; Wilson, 2013; Johnson, 1954) to find applicable methods.
All appropriate scheduling methods are filtered and rated on key characteristics of the FPI. An overview is created which helps select the capable methods. An analysis of the selected scheduling rules is performed using a case company data.
3.3
Case study
A case study is performed in order to gain knowledge of the food processing industry, find appropriate solutions which can be generalised. Moreover, the experts in this company can provide with valuable feedback and validation of explored methods. The case com-pany involved produces peanuts and nuts for the European, Australian and Philippine markets. Duyvis was the first peanut factory who came up with special peanuts and nuts like the ‘Borrelnootjes ’, ‘TijgernootjesR , and Oven Roasted Peanut and Nuts. WithR
125 employees Duyvis produces five days a week in three shifts to supply to its customers. Duyvis has six processing lines who sent their output to multiple packaging lines. Both departments are separated by intermediate WIP buffers. From a processing line a product can be packed on different packaging lines and create multiple SKU’s, The current process result in a lot of variability in the material flow and in initial interview the managers indicate that WIP buffers have become uncontrolled an a food hazard. The case study company shows typical problems for a job-shop and is in need for better WIP control methods.
3.4
Data collection and data analysis
Figure 3 gives an overview of the methodology. Firstly, an overview was given of the existing literature of work in process with definitions, the current states of WIP literature and the function of WIP. Secondly, literature of the characteristics of the food processing industry in combination with WIP was explored. Next, literature of the optimisation of WIP is explored. This exist of exploring literature of sequencing and literature of shop floor control. Both are explored in general but also applied in a food processing environment.
Using scheduling data the company provided, a simulation of the current WIP build-up is created and several effective JSSP which are easily implementable by planners are tested. Moreover, a the effective shop floor control methods, which control WIP, at the case study company are discussed. First an analysis of the current shop floor control methods is performed and usuable methods are filtered out. Next a validation is performed, discussed later.
At the company multiple in-depth interviews are conducted with multiple stakeholders in the handling of the WIP buffer. The functions of these people are listed below:
Figure 3: Overview Methodology • Operator Peanutline 1
• Bin distributor Coated 1 • Bin distributor Peanutline 1 • Bin distributor Peanutline 2 • FLM operations
• TE processing department • Food Safety specialist • Quality specialist
3.5
Validation of effectiveness
One of the key deliverables of this research is the effectiveness of the shop floor control and scheduling methods described and discussed. Therefore, a validation at the case-study company is performed along the lines of key FPI characteristics and requirements. A validation session is held with the interviewees earlier described and already performed methods are further discussed on their effectiveness. Operational scenarios for each shop floor control methods are provided which are discussed and rated on several important food processing characteristics. These interviews are recorded and the important rated elements discussed in the validation.
3.6
Validity and reliability
A case study is performed in this research which has some validity and reliability problems which have to be addressed. Because this reports is used as knowledge exploration and validation is tested in general and not specifically for the company some generalisability can be provided. However, for future research a multiple case study would be better suited to investigate this problem.
4
Analysis
In the analysis chapter the methods for controlling and optimisation of WIP in the Food processing industry are discussed. First, the scheduling methods and rules to reduce WIP are analysed at the case-company. Next, the shop floor control methods which help manage and control WIP are analysed in the case study environment. Last, the Lean tools implemented at the case company are analysed.
4.1
Scheduling at the case-company
The case-study companies factory has six processing lines which send their output to multiple packaging processes. In figure 4 an overview of the material flow in the factory is given, left the processing lines are displayed and right the packaging lines. Each processing line sends their output to a buffer. From this buffer the product can be processed on multiple packaging lines (for multiple SKU’s). Due to the non-fixed order of machines for a job, this process can be characterised as a job-shop environment. The processing department is located on the first floor, whereas the packaging depart-ment is located below on the ground floor. As already depart-mentioned, the processes are not directly connected (i.e. via a direct feeding line) but have intermediate storage. This intermediate storage area is located at the processing department, where materials are stored in storage-bins and cars. Currently, the different processing capacity of both de-partments causes continues buffer build-up during the week. Consequently, difficulties in controlling these buffers require strict policies and additional effort.
The Lean manager and manager of the planning department stress that, low investments are preferable in the food industry. Due to the low margin on the products it is hard to get a good return on investment with high investments. Continuous improvements and low cost solutions using current resourcese is therefore preferred.
4.1.1 Tactical decision: MPS
Currently, the MPS (Master Production Schedule) is determined yearly, for a four weekly cycle. In the MPS several product families are created based on product type, allergens and changeover activities (i.e. less changeover times). Moreover, for each production family the weekly production quantities are determined using forecast data, facility pro-cessing capacity and the work-force available. Resulting in an MPS, which for Duvyis is an approximation of the production required for each product in each week. This process is shown in figure 5.
4.1.2 Production schedule
Figure 4: Material flow
Figure 5: Current situation - MPS Duvyis
for all SKU’s. Only the products determined by the MPS can be planned in that week. Next, the planner applies a forward scheduling approach to create the production schedule for each week. No additional support scheduling rules are used, however, the planner will look at the allergen changeover matrix (optimised changeover) to reduce the total setup time.
starting times of products and the total processing times.
Every Thursday the production schedule is discussed with the operations department, adding maintenance or other special activities to the schedule. After finalising the weekly production schedule, the ownership is transferred to operations, who is responsible for the execution of the production plan. In case of interruptions (due to unexpected breakdowns or maintenance) the operation department will also make decisions regarding changes to the schedule. The planning department will only function as support during this process.
Figure 6: Current situation - Planning Duvyis
4.1.3 Scheduling decisions
As previously mentioned, a planner will apply forward scheduling and tries to schedule all the jobs of the processing department in the first three days of the week. In an interview with the manager of the planning department he said;
”this decision was made when the MPS was first established, due to labour cost require-ments and the gained flexibility for the end of the week”
Currently, the intermediate storage is used to create more flexibility and to take care that the processing department finishes early in the week. This gained flexibility is needed for any unforeseen technical problems and allows more freedom in the work-force planning. Furthermore, labour cost can be reduced because at the end of the week no operators are needed. However, since the processing department has a higher average output the intermediate storage slowly fills up during the week.
Currently, the case-company does not consider this buffer build-up in their production schedule, as flexibility and labour cost were valued higher when the MPS was established. Recently, the company has discovered that this buffer causes several problems. Because there is a lot of intermediate storage at the shop floor operators lose overview which result in inefficient working and making mistakes.
Figure 7: WIP - current situation
Figure 8: WIP produced - current situation
4.2
Scheduling in the FPI
Figure 9: WIP processed - current situation 4.2.1 Applying sequencing rules
As discussed in the theoretical background, most sequencing rules are not applicable in complex situations (i.e. job shop). Therefore, complexity needs to be reduced (Hopp and Spearman, 2011), for the case-company, MPS data and production quanities of week 40 and 41 are used. First, for both weeks a one to one machine situation (i.e. flow-shop) is used as a simplification. Next, several sequencing rules are applied on this one to one machine situation namely: shortest processing time (SPT), Johnson’s rule, shortest weighted processing time (SWPT), shortest setup time (SST), SPT with a release list, SST with a release list, SPT with a release list and with lot splitting, and SST with a release list and with lot splitting. A short description of the priority rules are given below:
Shortest processing time (SPT):
Jobs are prioritised based on the length of the processing time starting with the shortest job (Gupta and Bector, 1989).
Johnson’s rule:
Johnson (1954) developed a scheduling heuristic for the n jobs on two sequential machines, so that the total time processing time is minimised for the whole operations. The time required to process each job at each machine is listed in two vertical columns (Johnson, 1954). Next, all time periods are scanned for the shortest processing time. When the shortest processing time is for the first machine, place the corresponding item first. When it is for the second machine, place the corresponding item last. Once the item is placed, cross off both times for that item. Repeat the steps until all items are placed.
Shortest weighted processing time (SWPT):
factor. For WIP reduction purposes this weighted factor is calculated by dividing the processing time of the first machine by the processing time of the second machine. Next, the SPT procedure is applied (Gupta et al., 1999)
Shortest setup time (SST):
Jobs are prioritised based on the length of the change-over times starting with the shortest change-over job.
SPT and SST with a release list :
A release list is added to the SPT and SST rule, jobs will arrive at the second machine when capacity is available at this machine (Hopp and Spearman, 2011).
SPT and SST with a release list and with lot splitting:
Lot splitting means that the batch is not produced in one processing batch, but can be separated in, for example, two batches. Lot splitting is added to the SPT and SST rule with a release list.
Results of the analysis of sequencing rules for week 40 and 41 are shown respectively in figure 10 and in figure 11. For week 40 the processing line ”BK” is producing for packaging line 9. For week 41 the processing line ”BK” is producing for packaging line 1 and 2. From these graphs it can be seen that; SPT, Johnson’s rule, the SWPT and the SST; build up the amount of WIP during the first half of the week. The highest point is reached in the middle of the week and WIP is processed at the end of the week, observations at the case study also show this process. An SPT and SST with release list and with lot splitting reduce the maximum amount of WIP kg’s and bins on the floor, processing times are spread out and larger changeover are required. However, these reules have resulted in a lower overall buffer.
Figure 10: Buffer week 40
Figure 11: Buffer week 41
for the processing department the additional variable time of the machine is displayed. Finally, for each rule the maximum KG of buffer and the amount of bins were calculated. The tables show that the rules with the release list and with lot splitting have higher added variable time, however, the maximum total of bins are half or even two third lower.
Processing Packaging WIP
Hour production finished Total processing time (hours) Changeover (hours) Variable time machine (hours) Hour production finished Total processing time (hours) Changeover (hours) Max KG Total amount of bins Shortest processing time (SPT) 66,28 63,28 3 0,00 103,51 96,51 7 38415 192 Johnson’s rule 66,28 63,28 3 0,00 103,51 96,51 7 38415 192 Shortest Weighted Processing Time (SWPT) 64,76 61,76 4 0,00 99,16 91,66 7,5 41007 205 Shortest setup time (SST) 66,28 63,28 3 0,00 102,26 96,51 5,75 40230 201 SPT with release list 85,77 63,28 3 19,48 109,26 96,51 12,75 19969 100 SST with release list 82,72 63,28 2 17,44 102,26 96,51 5,75 26289 131 SPT with release list with lot splitting 94,81 65,33 6 23,48 109,26 96,51 12,75 14300 71 SST with release list with lot splitting 91,24 59,71 6 25,53 109,26 96,51 12,75 10808 54
Table 2: Results sequencing rules week 40
Processing Packaging WIP
Hour production finished Total processing time (hours) Changeover (hours) Variable time machine (hours) Hour production finished Total processing time (hours) Changeover (hours) Max KG Total amount of bins Shortest processing time (SPT) 44,76 41,76 3 0,00 115,29 106,29 9 46988 235 Johnson’s rule 44,76 41,76 3 0,00 115,29 106,29 9 46988 235 Shortest Weighted Processing Time 44,76 41,76 3 0,00 115,29 106,29 9 46988 235 Shortest setup time (SST) 43,76 41,76 2 0,00 112,29 106,29 6 42936 215 SPT with release list 72,13 41,76 3 27,37 115,29 106,29 9 30277 151 SST with release list 96,98 41,76 2 53,21 112,29 106,29 6 26289 131 SPT with release list with lot splitting 98,13 54,82 4 39,32 112,29 103,29 9 17713 89 SST with release list with lot splitting 93,23 41,76 3 48,46 112,29 106,29 6 14844 74
4.2.2 Overall optimisation
As previously mentioned, a simplification of the overall scheduling is required to get results for this problem. First, a local optimisation of one processing line to one pack-aging line was performed and analysed. Now, the overall optimisation of this job-shop scheduling process is performed and analysed. In these situations the tested rules can be applied.
An important rule to optimise WIP is that just one machine of the processing department can produce for one machine of the packaging department. It should not be possible that two machines of the processing department simultaneously produce for the same machine of the packaging department (i.e. to keep WIP low). When a processing line finishes producing for a packaging line it can start producing for another packaging line. However,only if the packaging line can start processing the output of the processing line immediately.
A simplified one to one machine situations (i.e. flow-shop) can be created from the case-situation (job-shop) using the following steps:
1. Create a table with the total processing times of the departments (int this case for the packaging and processing department).
2. Ensure the total sum of the processing times of each production line is below the maximum total processing hours. (Example: when there is 5 days of production 5x24=120 hours).
3. First, create blocks for the production line who’s output gives input to one machine 4. Second, identify the production line with the most total processing line, plan the
largest processing time first.
5. Repeat step 4 and 5 until all blocks are assigned. 6. Apply the sequencing rule separately for each block.
Figure 12 shows a block diagram where the steps are applied on the case company data. Dark colours represent the time the processing line is producing for that batch. Lighter colours represent the time the packaging line are still producing.
Figure 12: Block diagram week 41
4.2.3 Additional optimisation
In literature, simple visual tools have not been discussed, however it was found during this study that WIP can be further decreased with some simple adjustments.
Line balancing of the overall facility output on both production and processing can be performed using the graphs. For example, simply delaying some production jobs in the week and levelling the production and packging lines results in the production of 6000 KG of products per hour and the processing of 6000 KG of products per hour as well. Applying capacity levelling balances WIP during the week and decreases the maximum buffer output for week 41 to 150 bins.
Figure 13: WIP - Line balancing
Figure 15: WIP processed - Line balancing
4.3
Effectiveness of Shop Floor Control measures in the food
industry
After limiting and optimising the WIP through scheduling, the left-over WIP still needs to be controlled on the shop-floor. In this section, first, the current way of controlling the buffer is discussed. Second, an analysis of effective shop floor control methods suitable for the food processing industry and WIP control is discussed.
4.3.1 Material handling at the shop floor
In the case-company decisions regarding the location and storage of the buffer are made at the shop floor, the process is shown in figure 16. The operator gives the bin distributor instructions about, for example, the location of the buffer. Next, it is the job of the bin distributor to fill and transport the storage bins.
There are no rules concerning the intermediate storage. The case company tried to organise the shop floor by applying 5S. On the floors lines indicate the location of a buffer. The processing operators indicated in interviews that they try to locate the buffer in an assigned buffer location and close to the pouring hole of the packaging line where it is intended for. However, this is not always possible due to the crowdiness of the storage lots and many bins located close to the packaging line. Currently the processing can be described as unorganised, and ad-hoc rules wherein operators try to manage/find placement of the storage bins.
the green squares represent the pouring holes of the packaging line. The map is created based on observations.
Figure 16: Current situation - Buffer process Duyvis
4.3.2 Classification of SFC-methods
In order to find shop floor control methods that can be used in the food industry, a two-stage filtering process is performed. First, the shop floor control methods are rated on three basic characteristics namely: if it is suitable for a flow shop, if it helps to reduce WIP on the shop floor, and if it does not require high investments. The three characteristics are discussed below.
Flow shop
A flow shop is a repetitive and continuous operation producing similar or identical items in high volumes (usually found in the FPI). Materials move through them somewhat smoothly and with few interruptions (Nicholas, 2011). The shop floor control system must take into account two separate processes in serial, with different capacities.
Reduction of WIP
The method needs to intent to control or reduce the total work in process. It needs to give overview of the work in process in the system as well to control that, no jobs are released when the capacity can not handle the flow. When the jobs can not be handled the consequence can be that the job is not processed on time.
Low investment
Last, an initial requirement of this research was to find the methods of continuous im-provement and control that require minimal investment (since SME’s do not have much investment capabilities). Moreover, the FPI deals with low margin on their products and the return on investment of most high cost SFC-methods is simply to high.
Each method requires to score a + on all three characteristics to be selected for the second filter. Methods scoring three pluses are categorised green, others are categorised in red. Results of the first filtering are shown in the first half of figure 18.
Figure 18: Shop floor control methods
For the second filtering the characteristics changeovers, perishability, variability and stor-age constraints are taken into account. The characteristics are discussed below:
Regardless of batch size, whenever a machine must produce a different product, machines and fixtures at each station must be changed over (Nicholas, 2011). Changeovers in the food industry are often sequence dependent and are critical due to the food-safety issues (Yash Dave and Nagendra Sohani, 2012). EU-regulations require control of buffers to prevent food hazards (Sanders, 1999). In production facilities where multiple products are produced in the same location, there is a high risk of cross-contamination. Cross-contamination means that allergenic proteins from one product are carried into the next product (Jackson et al., 2008). This makes it important that allergens and other un-wanted materials are cleaned and separated. In the food industry, cleaning times (due to allergens), and start-up losses result in long set-up times, therefore, long runs are preferable from a production perspective (Wezel et al., 2006). An effective SFC-method therefore should help control and optimise for changeovers and longer production runs. Perishability
In the food processing industry raw material, semi-manufactured products, and end prod-ucts are perishable (Van Donk, 2001). There is a time constraint for perishable buffers, which means that the products can only be stored for a limited period (Flapper et al., 2010). When this period is exceeded the stored product will be discharged as waste, due to food safety regulation. It is important that a SPC-method can ensure that the buffers are processed before the maximum period is reached. Due to quality, environmental requirements and product responsibility there is a necessity for traceability of work in process (Van der vorst et al., 2007). As already discussed, buffers need to consider food allergens which are considered a major health risk (Van Hengel, 2007). A SPC-method needs the ability to register and provide information of WIP to be sure the time period is not exceeded.
Variability
Two types of variability can be distinguished, namely; variable yield and variable pro-cessing times.
Variable yield:
Variable yield is due to waste during the process which has to be re-worked or scrapped. Moreover, this results in additional production quantities being planned to end up with enough final product. A shop floor control method must be capable of dealing with variable yields in their WIP systems.
Variable processing times:
Especially in the packaging phase, the food industries have a divergent product structure (Van Donk, 2001). From one source product multiple SKU’s are produced with different characteristics and processing times (i.e. packaging size, etc.). The SFC-method must take into account differences in production capacity (between multiple SKU’s), packaging on different production lines, and between the processing and packaging department. Storage
need to be stored near the work floor/process and are constrained by limited storage space and storage bins. In the food industry, when the next machine cannot process the incoming capacity, additional products are stored in specialised silos or food storage crates. Storing in silos is common, but unfavourable, since a silo can only be used for a single product (e.g. due to cross-contamination), whatever the quantity in the silo (Flapper et al., 2010). Subsequently, the silo has to be emptied, cleaned and checked before re-use. Storage crates are used for other temporary smaller size buffers, however properly organising these buffers is a challenge. The SFC-method need to take into account the limited storage space and must be suitable for multiple storage types. Validation
Similar to the first filtering round each of the SFC methods is scored using + or - and a minimal of three pluses are set as the limit for validation. In the next chapter these methods are validated with case-company representatives and rated for their effectiveness.
4.3.3 Lean
In the case company an additional Lean tool is used namely ”QCDM”. QCDM stands for Quality, Cost, Delivery, and Moral. Every one hour, four hours, twenty-four hours, week, and period a ”QCDM” meeting is held. In every meeting different people of different layers of the organisation are present. The aim of ”QCDM” is continuous improving. It starts at the shop floor. Every operator is responsible for a zone. For every zone a KPI is set on Quality, Cost, Delivery, and Moral. An indicator turns red when a KPI is not met. Every hour the department manager checks the boards and asks the operator to explain the red indicators. What is the problem? What did you already do to solve the problem? Can you solve the problem? When do you expect the problem is solved? When the operator cannot solve the problem the department manager discuss the problems with the other department manager and the team manager in the four hour meetings. All the departments have their own KPI’s. In the twenty-four hour meeting the managers of all the departments discuss the red KPI’s. Also the managers mention what the cause is of the problem and when the problem is solved. When the root cause is unknown 5why’s are used to find the root cause. In the weekly meeting the management team discusses problems that cannot be solved in the twenty-four hour meeting. In the weekly meeting projects are set-up to improve the processes on the items that came from the twenty-four hour meeting. Also Gemba-walks are done to spot if everything is still on standard. Lastly, the period meeting discusses the problems that cannot be solved in the weekly meeting.
5
Validation
In this chapter the shop floor control methods drum-buffer-rope, hybrid CONWIP, hybrid two-boundary control, and minimal-blocking are validated. During the validation the four methods are discussed with the Lean manager, manager of the planning department, and a co-operator of the Lean department who also has experience with operations.
5.1
Drum-buffer-rope
The first SFC method is drum-buffer-rope. The method is based on the Theory of Con-straints (TOC) and focuses on the systems constraint (Goldratt and Cox, 2004). Three major components of DBR are the drum, the buffer, and the rope. Lower capacity sta-tion, which governs the throughput rate of the entire manufacturing line, are known as the ”drum”. The drum has to include a detailed schedule of the constraint in order to ensure the exploitation of the constraint. A ”buffer” is protection time (Schragenheim and Ronen, 1990) and is placed before the constraint. The purpose of this buffer is to protect the drum against unexpected fluctuations in the flow of materials. Last, the ”rope” is a material release mechanism to force all parts of the system to work up to the pace dictated by the drum and no more (Spearman et al., 1990).
DBR enables better scheduling and decision making on the shop floor (Schragenheim and Ronen, 1990). It only releases jobs when the bottleneck can handle the capacity. This prevents excessive WIP build-up in the process by limiting the input for the the bottleneck process. DBR is tested in job shop environments (Spearman et al., 1990) which makes it suitable for processes with variability.
A basic version of drum-buffer-rope is practised at the case company. Currently, the drum and the buffer are present. Only the rope is not present. According to the Lean manager there is no signal that the intermediate storage is too little or too high. Applying the rope will give more insights.
Changeovers can be taken into account when scheduling the constraint. But in general no changes to either cleaning, set-up requirements or start-up /stop losses are taken into account.
5.2
Hybrid CONWIP
With hybrid CONWIP the total work in process is limited by the number of cards. An available card has to be present to authorise a job entering the production line (Spearman et al., 1990). The card is attached to the job that is being routed through workstations. When the processing of the job is completed, the card is removed and made available to authorise another job to enter the line. The orders that need to be processed in the production line come from a backlog list (Fernandes and Godinho Filho, 2011). The backlog list dictates what goes to the line, and the card decides when.
According to the manager of the planning department, the system could be applied in the food production industry when the plant has a flow shop design. In case of Duyvis the system is difficult due to the job shop design of the factory. This causes variability in the material flow.
Furthermore, there is a chance of a lot of start and stop moments which result in waste. High utilisation is preferable and CONWIP risks that the machines stand still because no authorisation card is available. This will influence the utilisation rate. There is also a risk that the total amount of jobs for that week cannot be produced when the processing line is not able to start before the packaging line is finished. Because there is a maximum amount of WIP in the system the throughput time will decrease, which is positive when taking into account perishability. The co-worker of the Lean department added that it is important that the bins are processed based on the First in First out (FIFO) priority rule. This will secure that the product is not to long on the shop floor and this will enhance the quality of the product.
Last, changeovers can be considered when applying hybrid CONWIP. The backlog list determines the sequence of the jobs. Changeovers can be considered in the backlog because the length of the changeover is sequence dependent. This means that this method can consider the optimal changeover times.
In conclusion, the hybrid CONWIP system could be applicable in the food processing industry when the plant has a flow shop design. When the factory has more job shop characteristics the hybrid CONWIP method will influence the utilisation rate.
5.3
Hybrid two-boundary control system (TBC)
system. The authorisation cards are based on scheduling (Fernandes and Godinho Filho, 2011).
The reasoning for the hybrid TBC is partly the same as for the CONWIP system. The system could be applied in the FPI when there is a flow shop design and not a job shop design. Moreover, the it can be implemented when the utilisation rate is not important. The difference is that kanban is not suitable for environments with a lot of variability and long changeover times. Partly the changeover time can be reduced because in the FPI changeover times are mostly sequence dependent. The sequence can be determined in the backlog list.
Likewise the changeovers, perishability works the same for the TBC system as for the CONWIP system. To be applicable for the FPI the products already in the system needs to be processed based on the FIFO priority rule to prevent that the product is to long at the shop floor.
To conclude, the hybrid two-boundary control system is not applicable to the food pro-cessing industry. Especially the kanban part needs ideal conditions to work effective. Kanban does not work well with a lot of variability and long changeover times which is the case in the FPI.
5.4
Minimal-blocking system
The minimal-blocking system has a lot in common with the kanban system. The difference with the systems discussed before is: ”if the machine upstream finishes its operation before the machine downstream, and the demand occurs at the downstream machine in the meantime, the upstream machine can start a new operation” (So et al., 1988). Therefore, as a machine can start its operation as a result of either a requisition from the downstream machine or a schedule.
The minimal-blocking system solves the problem of the hybrid CONWIP system and the two-boundary control system. Because the upstream machine can already start a new operation when the downstream machine is still producing the utilisation rate is higher. This makes the system not only applicable to the food processing industry, but also interesting for the case company.
Changeovers can be taken into account because there is a schedule involved in this system. Also in this system the changeover times can partly be reduced because the sequence of the jobs can be determined in the schedule.
When taking into account perishability. It is important that FIFO is applied to control the time the product is placed in the intermediate storage.
5.5
Conclusion of validation
Concluding from the validation, all four methods are applicable to the food processing industry. ALl methods are applicable in a flow-shop and can at least work in a job-shop environment, controlling the WIP level. For the case-company drum-buffer-rope and the minimal-blocking system has the most potential to control the WIP level in their production facility. Mainly because these methods keep the utilisation rate high, whereas the other methods cannot guarantee a high utilisation rate.
6
Discussion
In this chapter an overview is given of the contribution to literature, contribution to practice, and the limitations and suggestions for further research.
6.1
Contribution to literature
The results presented in this research show how WIP can be optimised in the food pro-cessing industry. This research shows that the FPI has different characteristics compared to other processing industry’s. Due to these characteristics different methods need to be applied while handling WIP.
The research extend the WIP optimisation methods used in flow-shop scheduling of per-ishable products by researching the scheduling heuristics in the food processing industry. Finding the important parameters and influencing factors for WIP reduction. Also a method for simplification of job-shop scheduling rules have been shown, which help im-prove WIP without applying complex algorithms. Moreover, a visual technique based on capacity balancing, which can be easily applied by planner is shown to reduce WIP in the FPI.
Furtermore, research regarding the applicability of shop floor control methods in the food processing industry are given and validated in a food company. In conclusion of the validation, the drum-buffer-rope and the minimal-blocking system are applicable in the FPI with a job shop environment. In both methods perishability needs to be taken into account and FIFO needs to be applied. Another important characteristic of both methods which make the methods suitable for the FPI is that the methods keep the utilisation of the machines high. However, more research in the general applicability. In addition, the case company showed that ”QCDM” is a useful Lean tool to constantly review processes. Duyvis showed its relevance to the FPI and its contribution to control-ling the WIP buffer.
6.2
Contribution to practice
6.3
Further research
Due to the single case study this research has several limitations, generalisability is needs to be considered. Furtermore, rules and methods are tested in a job shop environment and need to be tested in a flow shop environment to expand the research of WIP methods applicable in the food processing industry.
7
Conclusion
Currently, the majority of research in optimising WIP has focused on assembly-type op-erations, with little work done int the food industry. Moreover, most of these scheduling rules are not applicable by companies due to their complexity or the need for advanced software. Methods for optimising and controlling work in process in the food processing industry are presented in this research. The FPI needs to apply low investment solu-tions, due to the low margin products but also due to the complex characteristics. This researched showed that by breaking down the complexity, simpler sequencing rules can be applied to optimise WIP. Furthermore, simple visualisation can help planners control WIP in the food processing industry.
References
Akkerman, R., Farahani, P., and Grunow, M. (2010). Quality, safety and sustainability in food distribution: A review of quantitative operations management approaches and challenges. OR Spectrum, 32(4):863–904.
Akkerman, R. and van Donk, D. P. (2009). Product mix variability with correlated demand in two-stage food manufacturing with intermediate storage. International Journal of Production Economics, 121(2):313–322.
Bonvik, A. M., Couch, C. E., and Gershwin, S. B. (1997). A comparison of production-line control mechanisms. International Journal of Production Research, 35(3):789–804. Burbidge, J. (1990). Production Control: a universal conceptual framework. Production
Planning & Control, 1(10):3–16.
Chaudhry, I. A. and Khan, A. A. (2016). A research survey: Review of flexible job shop scheduling techniques. International Transactions in Operational Research, 23(3):551– 591.
Colledani, M., Angius, A., and Horvath, A. (2014). Lead time distribution in unreliable production lines processing perishable products. 19th IEEE International Conference on Emerging Technologies and Factory Automation, ETFA 2014.
Conway, R., Maxwell, W., McClain, J. O., and Thomas, L. J. (1988). The role of work-in-process inventory in serial production lines. Operations Research, 36(2):229–241. Crandall, R. and Crandall, W. (2003). Managing Excess Inventories: a life-cycle approach.
Academy of Management Perspectives, 17(3):99–113.
Dora, M., van Goubergen, D., Maneesh, K., Molnar, A., and Gellynck, X. (2015). Ap-plication of lean manufacturing tools in the food and beverage industries. Journal of Technology Management and Innovation, 10(3):120–130.
Dudbridge, M. (2011). Handbook of lean manufacturing in the food industry.
EU (2002). Regulation (EC) No 178/2002 of 28 January 2002. Official Journal of the
European Communities, 31:1–24.
Fernandes, F. C. F. and Godinho Filho, M. (2011). Production control systems: Litera-ture review, classification, and insights regarding practical application. African Journal of Business Management, 5(14):5573–5582.
Fernandes, S. and Louren¸co, H. R. (2008). A simple optimised search heuristic for the job shop scheduling problem. Studies in Computational Intelligence, 153:203–218. Flapper, S. D. P., Fransoo, J. C., and Broekmeulen, R. a. C. M. (2010). Planning and
Garey, M. and Johnson, D. (1979). Computers and intractability: a guide to NP-completeness.
Goldratt, E. M. and Cox, J. (2004). The Goal: A Process of Ongoing Improvement, volume 3rd Editio.
Gupta, M. C. and Bector, C. (1989). International Journal of Computer Integrated A review of scheduling rules in flexible manufacturing systems. (May 2016).
Gupta, S. M., Al-Turki, Y. A., and Perry, R. F. (1999). Flexible kanban system. Inter-national Journal of Operations and Production Management, 19(10):1065–1093. Higgins, P., Le Roy, P., and Tierney, L. (1996). Manufacturing planning and control:
Beyond MRP II. Springer Science & Business Media.
Hopp, W. J. and Roof, M. L. (1998). Setting WIP levels with statistical throughput control (STC) in CONWIP production lines. International Journal of Production Re-search, 36(4):867–882.
Hopp, W. J. and Spearman, M. L. (2011). Factory physics.
Jackson, L. S., Al-Taher, F. M., Moorman, M., DeVries, J. W., Tippett, R., Swanson, K. M. J., Fu, T.-J., Salter, R., Dunaif, G., Estes, S., Albillos, S., and Gendel, S. M. (2008). Cleaning and other control and validation strategies to prevent allergen cross-contact in food-processing operations. Journal of food protection, 71(41):445–458.
Johnson, S. M. (1954). Optimal two- and three-stage production schedules with setup times included. Naval Research Logistics Quarterly, 1(1):61–68.
Little, J. D. C. (1961). A Proof for the Queuing Formula: ¡i¿L¡/i¿ = λ ¡i¿W¡/i¿. Operations Research, 9(3):383–387.
L¨odding, H., Yu, K. W., and Wiendahl, H. P. (2003). Decentralized WIP-oriented man-ufacturing control (DEWIP). Production Planning and Control, 14(1):42–54.
Mahalik, N. P. and Nambiar, A. N. (2010). Trends in food packaging and manufacturing systems and technology. Trends in Food Science and Technology, 21(3):117–128. Moe, T. (1998). Perspective on traceability in food culture. Trends in Food Science &
Technology, 9:211–214.
Nakhla, M. (2006). Production control in the food The need for flexibility in operations scheduling. 15(8):73–88.
Nicholas, J. (2011). Lean production for competitive advantage: A comprehensive guide to lean methodologies and management practices. Productivity Press.
