Master’s Thesis
Towards Lean Dealerships:
How can workload control effectively manage different value
streams within automotive dealerships that process both new as
well as used cars?
by S.L. Dingemans
s.l.dingemans@student.rug.nl
December 27, 2016
Abstract:
Due to increased pressures within the automotive industry the competition for automotive dealerships increased. As a result, their focus tended towards aftersales, while the repair and maintenance for cars seem less often needed due to increased quality and reliability of cars. Therefore, automotive dealerships should seek for opportunities to remain competitiveness and gain maximum potential from their maintenance- and repair related activities. The results of this study are two folded. Firstly, this study determines the distinction of two newly defined value streams within the workshop of an automotive dealership. Secondly, a case-based simulation study demonstrates that workload control can effectively manage different value streams. While dispatching purely based on the different value streams decreases the performance of automotive dealerships, the results indicate dispatching based on value streams can complement existing rules to increase performance and thus effectively manage different value streams.
Keywords: value streams, workload control, automotive dealerships, dispatching rules
University of Groningen Newcastle University
Faculty of Economics and Business Business School
Table of Contents
1. Introduction ... 3 2. Theoretical Background ... 6 2.1. Automotive Industry ... 6 2.2. Automotive Dealerships ... 7 2.3. Lean Management ... 8 2.4. Workload Control ... 11 2.4.1. Dispatching rules ... 13 3. Methodology ... 16 3.1. Research Design ... 16 3.1.1. Discrete-event simulation ... 16 3.2. Data Collection ... 17 3.2.1. Interviews ... 18 3.3. Data Analyses ... 19 4. Case Description ... 21 5. Simulation Model ... 23
5.1. Assumptions and Simplifications ... 25
5.2. Model Verification and Validation ... 25
5.3. Experimental Design ... 26
5.4. Experimental Settings ... 27
5.5. Output Analysis ... 28
6. Findings ... 29
6.1. What are the differences between processing new and second-hand cars? .... 29
6.2. How can Workload Control effectively cope with these value streams? ... 31
7. Discussion ... 37
8. Conclusion ... 39
9. References ... 40
1. Introduction
For automotive dealerships, nowadays the biggest challenge is to show to the world that they are the best channel for procurement and maintenance of customers’ motorisation needs (Kiff, 2000). Besides, they need to show the manufacturer that they are the best possible route for them to market their vehicles and aftersales products. “Dealerships represent the car manufacturers at the point of sale and act as a means of continuous contact between the car producers and the customers, long after the customer has taken delivery of the vehicle” (Fraser, Watanabe, & Hvolby, 2013, p.8). The automotive industry can be a substantial part of the economy of developed western countries. “The wellbeing of the industry has long been seen as an indicator of the health of the general economy in many developed countries. It is not just the manufacturing of cars and car parts but the marketing, selling and after- sale service have an equally enormous impact on the economy” (Fraser et al., 2013, p.9). As a result, all players active in this industry need to cope with lots of different pressures and changes. Examples of those changes are sustainable matters, technological developments, and governmental regulations. These changes do not make it easier for players within the automotive industry to remain their competitiveness. Nowadays, as a result of these pressures, dealerships have to cope with pressures such as tougher sales targets set by the manufacturers, higher levels of used car stocks, tighter manufacturer’s standards for facilities within the dealerships and increased staff-training-, administration- and operations costs (Atkinson et al., 2015). In order to survive, the focus for automotive dealerships tends to be more directed to the aftersales and service- and repair related activities. However, these aftersales and service operations seem less often necessary due to increased technology, quality and reliability of today’s vehicles (Kiff, 2000; Sabbagha et al., 2016). Figure 1.1 illustrates the increasing average age of the EU car fleet from 2006 to 2015 (ACEA, 2016).
Figure 1.1 The Average Age of the EU Car Fleet (ACEA, 2016)
8 8,5 9 9,5 10 2005 2007 2009 2011 2013 2015 AVE R AG E AG E ( YE AR S)
These developments make automotive dealerships seek for alternatives to outperform their competition. The removal of wasteful activities, reduction of costs, delivering greater customer value and improving customer retention could be beneficial considerations (Kiff, 2000).
When thinking about waste within the field of operations management, the link with lean management is easily made, especially in the automotive industry. Lean production is derived from automotive manufacturers after all. Originating from Toyota in the 1950s which started with lean production to eliminate waste; those processes which do not add customer value to a product (Liker, 2007). Toyota managed to gain a competitive advantage since lean management enabled them to provide their customers with higher quality cars with the lowest number of defects while spending less time manufacturing them (Liker and Morgan, 2006).
There is a significant amount of existing literature regarding lean management within the automobile manufacturing industry, but very limited research has focused on lean management within dealerships in the automotive industry (Atkinson et al., 2015). The literature did extend the scope of field in which lean management is studied towards the supply chains in the automotive industry (Hines et al., 2002; Naim and Gosling, 2011; Thomé et al., 2014) and car manufacturers have started to explore the application of lean within their dealerships (Brunt and Kiff, 2007). Despite the success of lean management in manufacturing automobiles, lean management barely seems to be researched regarding the repair and maintenance related activities within the automotive industry.
Making use of dispatching rules can possible make dealerships manage differences between different value streams more easily.
With high environmental pressures causing minimal profit margins for dealerships, they should aim to manage their workplace-related activities to their maximum potential. With the merits of lean being proved in literature, it is interesting to see how workload control can contribute to the management of automotive distribution facilities in their journey towards lean. Therefore, this research attempts to propose two newly defined value streams within the workshops of automotive dealerships, examines the differences between them, and then examines how workload control can effectively cope with the differences between those value streams.
This report will attempt to answer the following research questions:
Research question: How can workload control effectively cope with different value streams within automotive dealerships that processes both new as well as used cars?
Sub question 1: What are the typical differences between processing new- and second-hand cars?
Sub question 2: How can workload control effectively cope with the differences between new cars and second-hand cars?
2. Theoretical Background
In this section, the relevant literature to the automotive industry, automotive dealerships, lean management and workload control is reviewed, the terminology as used is clarified, and the theoretical framework in which this research will be conducted will be developed.
2.1. Automotive Industry
Only a few industries are as large, diverse and influential as the automotive industry (Orsato and Wells, 2007). While providing the number one source of personal mobility to millions of customers, this industry had to deal with enormous environmental pressures beginning in the last quarter of the 20th century. The creation of the European Monetary Union, the block
exemption, congestion and environmental matters, the internet and the improved quality, technology and reliability of cars are examples of pressures which affected the automotive industry (Kiff, 2000). These developments, which mainly arise from the field of deregulation, globalisation, technology and fierce competition, have caused a change of rationale behind the operations of firms within this industry (Talay and Cavusgil, 2009). Section 2.2 contains a more in-depth explanation of this change in rationale.
Servitization, “the term coined by Vandermerwe and Rada (1988), is now widely recognised as the process of creating value by adding services to products” (Baines et al., 2009). It is used to create better mutual value through a shift from selling pure products towards selling product service systems (Baines et al., 2007; Takata and Umeda, 2007). For dealerships, it would create more value by transforming the relationship with the customer from a transaction to a long-term relationship, which eventually lead to increased customer-loyalty (Vandermerwe and Rada, 1988). “They want to maximise their customers’ ‘life time value’ and their ‘share of wallet” (Brunt & Kiff, 2007, p.3). Therefore, dealerships may include repair and maintenance service for the vehicle in sales of vehicles. For the customer, this could create more value due to more needs being fulfilled; dealerships, for example, might take the vehicle back at the end of its lifetime. BMW is already conducting research for years on the recycling of materials from existing cars and reuse of high-value parts from used cars (Thierry et al., 1995).
2.2. Automotive Dealerships
As described in paragraph 2.1, industrial developments have their effect on the operations rationale of firms within the automotive industry. Taken into account the higher degree of servitization, this especially counts for those parties who are positioned closest to the consumers; the dealerships (Tan et al., 2009). Hence, dealerships tried to create more value, for both their customers as well as for themselves, by focussing more and more on the after-sales services of the product, including maintenance- and repair related activities. However, due to technological developments, today’s vehicles became more and more reliable which results in vehicles that need less service and maintenance (Kiff, 2000). Therefore, dealerships need to make sure they manage their service operations as efficiently and effectively as possible. Despite the fact that repair- and maintenance-related tasks within the dealerships are of increased importance nowadays, not much research has been done on these specific tasks within their workshops.
The workshop of an automotive distribution facility can be seen as a job shop. “The term job shop is used to indicate a type of manufacturing situation where a large number of different products are produced according to customer specification with highly variable routings and processing times” (Land, 2004, p.1). In terms of services, such as is the case with the workshop of an automotive dealership, these situations are called service shops. “Depending on whether the job processing order (routing) implies a predetermined sequence of operations, we distinguish between the case of arbitrary routings (open shop) and the case of given job routings which may be identical for all jobs (flow shop) or non- identical (job shop) (Haupt, 1989). In the case of the workshop of an automotive dealership, the job routings will not always be the same. While one customer would like to have the windscreen wipers to be replaced, another customer would like to have his/her car painted, new tires and a complete assessment of the motor block. With a car existing of thousands of different parts, there are a lot of different repair activities possible for all the customers. These different activities will also imply another routing for these specific jobs. However, despite the differences in routings per job, all jobs seem to flow unidirectional with some jobs skipping certain steps within the workshop. Therefore, the workshop of an automotive distribution centre can be classified as a unidirectional flow shop.
2.3. Lean Management
“From a business perspective the term ‘lean’ is used to describe a philosophy of management which involves a set of tools and techniques used to optimise time, assets, and productivity, while continually improving the quality of products and services for customers” (Atkinson & Linehan, 2008, p.3) Originating from Toyota, lean management made manufacturing of their cars more efficient and thereby more effective (Jones, 2006). After the second world war, Japanese manufacturers had to deal with significant shortages in terms of materials, finances and human resources compared to manufacturers in the United States. These conditions ultimately led to the evolvement of the lean concept (Womack et al., 1990). Early Japanese industrial leaders, such as the president of the Toyota Motor Company, Kiichiro Toyoda, developed a new process-oriented system which is nowadays known as the Toyota Production System (TPS) from which lean management originates.
improvement and quality through waste reduction, and tightly integrated upstream and down- stream processes as part of a lean value chain” (Liker & Morgan, 2006, p. 5). Lean production is nowadays known as an approach which ultimately leads to increased efficiency and effectiveness through a balanced production, minimum variation, elimination of waste and minimum inventory by making use of just-in-time (JIT) principles; using only what is needed, when it is needed (Womack and Jones, 1996). Just-in-time is one of the most commonly known lean tools. A table with more examples of commonly known lean tools can be found in Appendix A.
Many manufacturing firms have already implemented lean principles, while lean has been expanding towards other industries such as services as well (Liker and Morgan, 2006). However, literature on lean management within automotive dealerships’ workplaces is very limited. “Automotive manufacturing has historically been at the forefront of initiatives that have sought to increase efficiency and reduce costs. Lean production historically arose within automotive manufacturing at companies such as Toyota and Nissan. Ever since, the focus of Lean Production in the automotive sector has been on upstream manufacturing with limited consideration of downstream activities.” (Small et al., 2015, p.1).
dealership both on their own way. With variability aimed to be reduced to a minimum already, there seem to be possibilities to make effectively use of the differences between value streams.
Due to the lack of focus on the processes within the workshop of automotive dealerships, literature has not yet acknowledged the differences between new cars and second-hand cars with respect to the costs for automotive dealerships. The first cost characteristic that differs among new cars and second-cars is depreciation. Once a car is sold to its first owner, a car will depreciate over time. In today’s automotive industry, new cars are generally sold already to a customer before the car leaves the production plant. The sales price is accepted by the customer. Therefore, new cars will not depreciate over time when being prepared for sale at the automotive dealerships. Contrary, second-hand cars are already taken into use. Therefore, second-hand cars do depreciate over time, still while being fixed or prepared for sale at the automotive dealership. “This process can be a complicated one, depending on several parameters, such as year, mileage, and car’s condition” (Alshamary and Calin, 2013, p.209). Nevertheless, with a depreciation rate of two to three percent per month on average, automotive dealerships should aim to reduce their lead times to a minimum. While depreciation seems to be a cost rather for the owner than for the dealership, it can be considered as a cost for the dealership if the supply chain as a whole is considered. Furthermore, the holding costs for cars are different, since dealerships are most often based on a franchise contract. With cars being sold to the customer, dealerships often have to pay the manufacturers of the car sixty days after the delivery of the car to the customer. This implies that dealerships can have the money, which is paid by the customer, in their possession before they must pay the manufacturer. Since dealerships could gain interest over the money in these sixty days it seems that there is another difference between new cars and second-hand cars. No research has been done towards the differences between new cars and second-hand cars and how these differences affect the performance of the workshops, while there are unexplored opportunities in terms of planning and production control (PPC) techniques.
the production system, pull mechanisms are typically very effective in reducing the amount of work in process, which is supported by making use of production signals (Slomp et al., 2009). These production signals can be communicated making use of lights on the production floor, or by making use of cards such as in Kanban systems. “Card-based systems can be simple yet effective means of controlling production. But existing solutions, such as Kanban, do not typically apply to the job shops often found in make-to-order companies” (Thürer, Land, & Stevenson, 2014, p.180). In such environments, the variability in the system is relatively high due to the high degree of customization of the products. This is also the case for the workshops of automotive dealerships, where every car needs its own specific set of processes in order to leave the system as desired. Therefore, it might be interesting to focus on workload control as an alternative, when considering lean management in the workshop of an automotive dealership. Workload control aims to achieve a comparable levelling of workload to capacity as achieved by lean in repetitive manufacturing (Thürer et al., 2014). The following section will elaborate more in-depth on Workload Control.
2.4. Workload Control
Workload Control (WLC) is a planning and production control concept which has been developed primarily for systems with high variabilities, such as medium-sized make-to-order companies, which often have a job-shop configuration (Thürer et al., 2014). “WLC provides MTO companies with many of the benefits of lean’s PPC techniques by levelling demand and production over time when work is not standardised and when it is not possible to synchronise flows on the shop floor” (Thürer et al., 2014). While the first publications on WLC appeared in the early eighties of the twentieth century, the ingredients for WLC as a comprehensive concept have been developed earlier (Land, 2004). Deriving from research on both job shop control, which started early in the twentieth century, and queuing theory approaches, which have been developed when computers enabled researchers to simulate job shops in the nineteen fifties, three shop floor management decisions are related to the concept of WLC (Land, 2004).
problem size (Land, 2004). Contrary, “a dynamic model allows for a continuous stream of arriving orders in time that are intermittently released to the shop and are included in the current scheduling procedure. Reasonably, the distinction between simultaneous and intermittent job arrivals involves the one between known and fixed job data on one hand and stochastic data, in particular job inter-arrival times, on the other hand. Hence, we distinguish a static/deterministic scheduling problem from a dynamic/stochastic one” (Haupt, 1989, p.4).
Land (2004) illustrates WLC with the paradox of a bathtub, which can be seen in Figure 2.1. Based on this figure, the general concept of WLC will be explained, taking into account the general input-output model, which is a commonly accepted model to look at production and operation of organisations (Kingsman, 2000).
The water in the tub represents the work that must be performed by the system. The input of the bath can be controlled by the tap. This tap depicts the job- or order release decision in the real system. “Order release is a key component of the Workload Control concept. Jobs do not enter the shop floor directly –
they are retained in a pre-shop pool and released in time to meet due dates while keeping work-in-process within limits or norms” (Thürer et al., 2014, p.6664). “The order release decision is the main instrument for the input control. Once released, a job remains on the shop floor until all its operations have been completed. WLC concepts set norms for the workload allowed on the floor. If a job does not fit in these norms, the release decision will hold it back. This results in a pool of unreleased jobs.” (Henrich et al., 2000). The pre-shop pool in which the work will be queued, which is represented by the water in the bathtub in Figure 2.1, acts as a buffer which can guarantee a sufficient utilization of the workstations. Then, the output of the bathtub represents the capacity of the system; the wider the hole through which the water can exit the bathtub, the higher the capacity of the system. The capacity of a real-life production system would then be the output rate of the system; “the amount of processing work that can be done by the work centre per time period” (Kingsman, 2000, p.75).
However, WLC is more than just the order release decision. Land (2004) states that there is a hierarchy within the complete concept of WLC, illustrated by Figure 2.2. First, at the entry-level, the amount of orders accepted can be monitored and subsequently decisions can control
the number of accepted orders. Then, at the release level, decisions can control the workload being active within the system at a certain point in time. In principle, these decisions set the amount of work-in-progress (WIP) for a certain point in time. Lastly, at the dispatching level, decisions will determine which jobs will be processed by giving prioritising jobs. “A dispatching rule is used to select the next job to be processed from a set of jobs awaiting service at a facility that becomes free” (Rajendran & Holthaus, 1999, p.157). The following section will elaborate more on dispatching rules since that will be the focus of this research within the concept of WLC.
Figure 2.2 Workload control decision hierarchy by Land (2004)
2.4.1. Dispatching rules
others across different shop configurations, operating conditions and objective functions” (Pickardt et al., 2013, p.67).
“Research in dispatching rules has been active for several decades and many different rules have been studied in the literature” (Pinedo, 2016, p.376). In order to create some order within the chaos of all dispatching rules developed, Pinedo (2016) makes a distinction between two different types of rules; static- and dynamic rules. Static rules are not dependent on time; these heuristics are based on the characteristics of the job and machine/system data (Pinedo, 2016). An example for a static rule could be a rule that prioritises a specific product family within the schedule. On the other hand, dynamic rules, are actually taking time and thereby due dates into account (Pinedo, 2016). An example of a dynamic rule is the Minimum Slack (MS) rule. This rule prioritises the jobs that have the least slack, which is available time minus the required time to process the job.
Rules involving processing time
SPT Shortest Processing Time Highest priority is given to the queued job with the shortest imminent operation time
LPT Longest Processing Time Highest priority is given to the queued job with the longest imminent operation time
MWKR Most Work Remaining Highest priority is given to the queued job with the most total processing time to be done remaining
LWKR Least Work Remaining Highest priority is given to the queued job with the least total processing time to be done remaining
MRO Most Remaining Operations Highest priority is given to the queued job with the most remaining operations still to be done
LRO Least Remaining Operations Highest priority is given to the queued job with the least remaining operations still to be done
Rules involving due dates
EDD Earliest Due Date Highest priority is given to the queued job with the earliest due date
MS Minimum Slack Highest priority is given to the queued job with the least amount of slack
LSO Least Slack per Operations Highest priority is given to the queued job with the least amount of slack per operation that is still required for the job to be done
Table 2.1: Commonly-known elementary dispatching rules
“ranking expressions that combine a number of elementary dispatching rules (Pinedo, 2016, p.377). Thus, while elementary dispatching rules can only serve a single parameter, composite dispatching rules are aiming to provide a solution for shop floors with multiple performance indicator objectives.
These different rules are developed by studies due to the need for different rules which aim to achieve a specific goal of performance. While it might be important to deliver the product as fast as possible for one organisation, it might be more important to keep inventory levels low for another. In assessing the performance of a schedule, Baker (1984) states that there are two factors of primary interest; the shop time and the due-date performance. While the first is focused on the performance of the actual system itself, the latter aims to ensure that customers are served within the agreed timespan. “Lead time reduction is a common measure for increasing a manufacturing firm’s competitiveness. Its implementation is supposed to benefit both customers – as it improves a firm’s abilities in safeguarding a timely response to their demand, and the firm itself – as costs of capital associated with work-in-process and finished goods inventories are reduced” (van der Zee, 2015, p.5837).
The total time that an order has spent in the shop is called the flowtime of the order. To assess the performance of the schedule with regards to shop time, “the mean job flowtime is a basic measure of a shop's performance at turning around orders and it is therefore often used as an indicator of success in responding quickly to customers” (Baker, 1984, p.1093). With regards to the due-date performance of the schedule; it is more complicated to determine. Of course, a schedule which meets all the jobs’ due-dates is a good-performing schedule, but when it is impossible to meet all due-dates, how can the best performance then be quantified? Baker (1984) describes that “the job shop literature suggests several answers, such as the proportion of late jobs, the mean tardiness among all jobs, and the conditional mean tardiness (which is the average tardiness measured over only the late jobs)” (p.1093).
3. Methodology
This section will elaborate on the methodology used for this research. It will do this by first discussing the research questions together with its sub questions. Then the proposed methodology, together with a preliminary method for data collection will be discussed.
3.1. Research Design
To examine how WLC can effectively manage different value streams within the workshop of an automotive dealership, a case-based simulation will be used. With lean being extensively studied upstream the supply chain in the automotive industry (Womack et al., 1990; Womack and Jones, 1996; Liker and Morgan, 2006) this research attempts to complement literature with a more comprehensive understanding of the different value streams within automotive distribution facilities, and aims to find out how WLC can effectively manage them. Therefore, this research will be explorative in nature, since it aims to get a thorough understanding of the applicability of WLC within this complex environment with high variability.
This research will be a deductive, mixed-methods case study. To answer the research question, multiple methods are used. First, to get a thorough understanding of the differences of existing value streams within the workshop of automotive distribution facilities, case-company data will be analysed, and semi-structured interviews with practitioners in the field will be conducted. Then, once a thorough understanding of the different value streams is gained, a discrete-event simulation model will be used to find out how workload control can contribute to the management of these different value streams. Section 3.1.1 will elaborate more in depth on the reason why simulation is chosen as the used research method. According to Voss, Tsikriktsis, & Frohlich (2002) “there are several challenges in conducting case research: it is time consuming, it needs skilled interviewers, care is needed in drawing generalizable conclusions from a limited set of cases and in ensuring rigorous research. Interviews will be used to validate the processes modelled” (p.195). On the other side, “the results of case research can have very high impact. Unconstrained by the rigid limits of questionnaires and models, it can lead to new and creative insights, development of new theory, and have high validity with practitioners - the ultimate user of research. Through triangulation with multiple means of data collection, the validity can be increased further” (Voss et al., 2002, p.195).
3.1.1. Discrete-event simulation
the possibilities of simulation increased tremendously since these computers can also visualise the simulated system nowadays. This offers advantages that will be discussed later in this section.
The reason simulation was chosen as the used methodology is firstly that it can deal with variability, interconnectedness and complexity (Robinson, 2004). An automotive workshop can be seen as a complex environment with high variability. Vehicles being brought in by customers at any time with most of the vehicles having a specific combination of problems. Secondly, “to evaluate different dispatching rules and compare their performance, the most commonly used method is the simulation method, i.e. design simulation experiment with shop models, and compare the objective value of the resulting solution to the optimal objective value” (Fan, Xiong, Jiang, & Li, 2015, p.1929). Furthermore, it often seems to be hard for traditional organisations to imagine what the actual impact of process improvements can be. This results in decisions whether or not to make use of new techniques are very often based on either faith in the new technique, experiences of previous implementers and general acknowledgements that can be made (Detty and Yingling, 2000). These sources, often combined, may not be convenient enough for management teams. A visual simulation of the current state, as well as the future state of the manufacturing system, can be a very useful tool to persuade stakeholders of the benefits.
The simulation is based on an initial model that has been developed by dr. Bokhorst, making use of Tecnomatix Plant Simulation 12.0.6 from Siemens PLM. This software uses a discrete event simulation approach based on the programming language SimTalk and “allows to simulate discrete events and create digital models of logistic systems (e.g. production), optimize the operation of production plants, production lines, as well as individual logistics processes” (Siderska, 2016, p.64). The reason for selecting this software package is that it allows investigating the performance outcomes while experimenting with different input variables applied to the system. Furthermore, the availability of a license and in-depth knowledge of the software within reach contributed this choice as well.
3.2. Data Collection
understanding of the different value streams dealerships have to cope with, face-to-face semi-structured interviews will be held. In section 3.2.1 it will be discussed more in-depth why this is necessary. The second type of required data according to Robinson (2004) is data for model realisation. “In moving from the conceptual model to a computer model many data are required, for example, detailed data on cycle times and breakdowns, customer arrival patterns and descriptions of customer types, and scheduling and processing rules” (Robinson, 2004, p.96). For this study, data of the case company, an automotive distribution facility in the North-East of the United Kingdom, will be used. Chapter 4 provides a case description which will give the reader a better understanding of the current situation at the case company. The case company data is gathered while making use of a software system called AutoFlow. This software monitors all the activities and tasks that are performed in the workshop, and measures the time needed for each specific task. An overview of the data categories captured can be found in Appendix C. The last type of required data are data for model validation. This type of data is required to “ensure that each part of the model, as well as the model as a whole, is representing the real world system with sufficient accuracy” (Robinson, 2004, p.96). The validation and verification of the model will be discussed later in section 5.2.
3.2.1. Interviews
Since literature lacks a thorough understanding of the value streams within automotive dealerships, interviews are needed to gain a deeper understanding of the differences between the value streams. The following section will elaborate a bit more on the interviews conducted.
Questions 1, 2 and 3 are included in order to make sure that the interviewee and the organisation in which he/she is active are indeed comparative to the case company. These questions are important to verify that the interviewee has knowledge of a similar environment. Then, questions 4 and 5 are included to gain an understanding of the various processes that are involved in the repair and maintenance of automobiles. Question 6 aims to get insights into the differences between processing a new car compared to processing a second-hand car. Questions 7, 8 and 9 are mainly focussing on the management of the different processes that are involved within the workshop of an automotive dealership. While question 7 is included to get knowledge on any operational rules applied in the organisation of the interviewee to the surface, the purpose of question 8 is to get information about the beliefs of the interviewee about process improvements within a workshop. This could be useful information in evaluating the interviewees’ attitude toward process improvement in automotive workshops. Question 9 tries to unveil the used performance measures of the interviewees’ organisation. While question 10 slightly introduces the subject of lean management by focussing on the customer satisfaction and the rework that must be performed within the workplace, question 11 tries to reveal to what degree the organisation of the interviewee applies lean principles within their workshop. Question 12 is included because requirements set by the manufacturers of automobiles may affect the processes performed on cars. The last question in the main body of the interview is question 13, which is included to make sure that the interviewee has the opportunity to elaborate a bit more on a subject of which he/she may believe that is important for the interviewer to know in the interest of this research. Question 14, 15 and 16 aim to round-up the interview and gain some information about the opinion of the interviewee on the interview.
3.3. Data Analyses
4. Case Description
The case company is a business to business P.D.I.-centre located in the North-East of the United Kingdom. P.D.I. is the abbreviation of pre-delivery inspection, which is the final check performed by a dealership before the car is delivered to its new owner. This site of the case company prepares cars for sale only. This can either be a new car or a second-hand car. In the case of new cars, the vehicles are brought in by the manufacturer and then the case company is expected to prepare the car for sale. In the case of second-hand cars, the cars are brought in by dealerships which outsource the pre-delivery inspection. A total amount of eight thousand, seven hundred and five cars arrived at the case company within the period of the first of January 2015 until the sixth of May 2016. On an average working day, approximately 25 cars will be processed.
The case company uses eight mechanical ramps and four SMART-ramps. While the ramps are used to process cars that require mechanical service, the SMART-ramps are used for processing cars with cosmetic damages. All the mechanical ramps are identical, and so are all the SMART-ramps. Furthermore, the case company uses four valet places. For all ramps and valet places, there is a worker. So, eight mechanical workers, four SMART-workers and four valets work in the workshop. Furthermore, there is a quality controller, and a photographer to make photos of the cars that are going to be sold.
The case company uses a dock at the port at which it can park a maximum of 900 cars. This dock is used as an input-buffer, or in terms of WLC a so-called pre-shop pool, for the PDI-centre itself, where they can park another 35 cars. Figure 4.1 is a simplified flow diagram of the case company.
Figure 4.1 Simplified flow diagram of the Benfield Ltd. site
in the port. The commonly known First Come First Serve (FCFS) dispatching rule is thus currently applied. The car is picked up by an employee of the PDI-centre, who performs an initial inspection of the car while driving it from the port to the PDI-centre. After a vehicle health check is performed, the case company knows which processes are required for that specific car. Some cars require multiple processes to be performed, and a decision needs to be made with regards to the planning of those activities.
The parking spaces at the P.D.I. centre are also used as a buffer. If a car needs multiple processes, that cannot be performed in succession, the car is placed on the parking lot and must wait until the required resources are available. Furthermore, if there are replacements of parts necessary, the customer must give authorisation to replace those parts and the parts often have to be ordered. If a car needs to wait for either authorisation, part ordering or both, the car will be placed on the parking lot at the P.D.I. centre and waits there until it can be processed. The mechanics will pick a car from the parking lot at the P.D.I. once a car is available to be processed. If there is no car available to be processed at the P.D.I. itself, then a car from the port will be picked to be processed. This results in a continuous WIP at the P.D.I. centre.
5. Simulation Model
The simulation model is based on the case company. As mentioned in section 3.1.1 the final model is based on an initial model developed by dr. Bokhorst. This initial model included several aspects of the current state of the case company. It needed to be adjusted in order to be a good reflection of the situation at the case company. The used distributions for processing times were based on the company data and the model was verified and validated as the current state, which will be discussed in detail in section 5.2. Then, adjustments had to be made in the SimTalk codes of the model to ensure that the different dispatching rules were followed by the model accordingly. These adjustments will be described in section 5.3 in which the experimental design of the simulation model will be described. It is recommended to keep Figure 5.1 at hand while reading the following section.
5.1. Assumptions and Simplifications
In this section, several assumptions and simplifications of reality are presented. These assumptions and simplifications facilitate in modelling complex systems while focussing on the most important facets of the problem.
• Resources can only be used for a single job (i.e. a mechanic can only work on a single car at any given time).
• Once a car is taken for processing, this car may not leave the P.D.I. before another car can be taken up for processing. Job pre-emption is not allowed.
• There are no unexpected interruptions in the workshop, e.g. no machine breakdowns. • All jobs are independent of each other; there is not assembly involved across different
jobs.
• The average value of new cars is higher than the average value of second-hand cars. • Cars are moved out of the system as soon as all the required processes are
accomplished.
• At the beginning of the simulation, 50 cars will be parked at the dock. • All mechanical repair jobs can be assigned to any mechanical ramp. • All smart repair jobs can be assigned to any smart ramp.
• Interest and depreciation rate will be the same for all cars in specific segments. • All customers will be served.
5.2. Model Verification and Validation
(Abdulmalek and Rajgopal, 2007). While the incorporated operation- and processing times are validated by a general manager of the case company. The output of the model is also compared with the output of the collected data making use of AutoFlow. Retrieved from the case company data, Benfield Ltd. processes 24,87 cars on average on a working day. This performance measure is used to verify that the output of the simulation model was a good reflection of the reality. The simulation model has an average throughput of 24,88 cars per day.
5.3. Experimental Design
To find out how WLC can effectively manage the different value streams within the workshop of an automotive distribution facility, the simulation model as described in the beginning of this chapter will be used. The different dispatching rules that will be applied to the simulation model are presented in Table 5.1.
Rule Description
FCFS First Come First Serve
VS Value Stream based dispatching
EDD Earliest Due Date
VSEDD Value Stream based Dispatching combined with Earliest Due Date
VSEDD-V VSEDD rule with an additional prioritisation based on the value of cars
Table 5.1 Experiments applied to the simulation model
composite dispatching rules will be evaluated. While the VSEDD emerges from the VS and the EDD elementary rules, the VSEDD-V rule is an extension of the VSEDD rule. With the VSEDD rule applied, jobs are prioritised first on the value stream in which they are, and then within those value streams, the cars will be prioritised on their due dates. The VSEDD-V rule extends the VSEDD rule in a sense that it will additionally check on the value of the cars in the queue, and then will prioritise the more expensive cars.
5.4. Experimental Settings
To make the outcomes of the simulation model more realistically, it is important to apply the right experimental settings to the simulation model. The simulation model was built as a non-terminating process. In other words, if a car is not finished by the time the employees of the dealership go home, the car will still be there in the morning the following day, and the process will continue as soon as the service provider gets back to work.
Having all the minimum requirements for the simulation determined above, the following experimental settings will be applied. Due to the fact that the case company, such as most of the automotive dealerships, is not open seven days a week, the run length should be counted in weeks. In order to include rare events, it is preferable to make the simulation longer than just one hundred and five days, which is the minimum amount of days above the minimum run length, and still is dividable in weeks. Therefore, the run length of the simulation is set to fifty-two weeks, which is three hundred and sixty-four days, plus a warmup period of ten days. The number of runs is determined to be ten at a minimum. However, Wilson VanVoorhis & Morgan (2007) recommend applying a minimum of 30 replications to have an effect power of 80% when analysing samples making use of the independent sample test, the paired sample t-test or the one-way or factorial ANOVA t-tests. Since a paired t-t-test will be performed to examine the significant difference of the results, the recommendation of Wilson VanVoorhis & Morgan (2007) becomes valuable to this research. Therefore, a total of 30 replications will be applied to the simulation model.
5.5. Output Analysis
The output of the simulation model will be split in three. First of all, the general performance of the dispatching rule will be assessed. The outputs used for the general performance analysis will be the mean flowtime (F), the maximum flowtime (Fmax), the percentage of tardy jobs
(%T), the mean tardiness (Tmean) and the maximum tardiness (Tmax). While the first two are
mainly used to assess effect of the dispatching rule on the flowtimes, the latter three are used to determine whether the rule serves sufficient with regards to due date fulfilment. Once the dispatching rule is assessed on these outputs, the effect of the dispatching rules on the different value streams will be assessed. This will be done by making use of the mean flowtime per segment of car (F(x)), the maximum flowtime per segment of car (F(x)max), the percentage of
tardy jobs of each segment (%T(x)) and the mean tardiness of each segment (T(x)mean). Assessing these output variables will give insights in the effect on the flowtimes and tardiness of different car segments.
Lastly, in order to make the differences between the dispatching rules more tangible in terms of this research, an overview of the mean total costs per car will be given. The output variables are the mean depreciation costs per car (mDep) and the mean penalty for tardiness per car
(Tpenmean) and will be given in order to assess the overall effectiveness of the dispatching rule
6. Findings
This section will be structured by following the sub questions as stated in the introduction of this study. First, the differences between processing new cars and second-hand cars will be declared. Then, the results with regards to the second sub question will be presented.
6.1. What are the differences between processing new and second-hand
cars?
While analysing the dataset retrieved from the case company, there was a clear distinction between the processes that were required by new cars compared to the processes required by second-hand cars. While new cars mainly are processed within the workshop only once, it was clear that second-hand cars more often must wait at the parking lot at the P.D.I.-centre. This implies that second-hand cars more often require multiple processes, authorization by the customer or parts that need to be ordered. Table 6.1 illustrates the processes that are related to new- and second-hand cars.
* These activities do not necessarily have to be performed in the exact same sequences as presented
Processes to perform for a new car*
Arrival of the car Activity
Parking at port Buffer
Vehicle Health Check Activity
Pre-delivery inspection Activity
Deliver car to the customer Activity
Processes to perform for a second-hand car*
Arrival of the car Activity
Parking at port Buffer
Vehicle Health Check Activity
Mechanical service activities Activity
SMART-service activities Activity
Authorization Buffer
Parts ordering Buffer
Processes within P.D.I.-centre Activity
Take a photo Activity
Deliver car to the customer Activity
The reason that second-hand cars often require more processes is that these cars are already used by a consumer. Therefore, it is more likely that the state in which a second-hand car arrives compared to a new car varies much more. While manufacturers aim to ensure that all new cars arrive at an automotive dealership in the exact same state, namely perfect, this cannot be the case for second-hand cars. Since a car is a relatively complex system with lots of different components, it is hard for dealerships to determine what the content of the work to be done exactly is, and thus what the corresponding processes in the workshop are.
These issues are validated by the interview. From the interview, it is gained that “preparing a car for delivery is the main activity that has to be performed on new cars. This mainly includes checking whether the manufacturer did their job appropriately; making sure all the settings of the car are correct, and double-check whether all the nuts are steadily installed. Further work on new cars is mainly done by working down a checklist provided by the manufacturer.” This implies that the conversion of the transporting-state of the car into the usable-state of the car can be standardized to a certain degree. “In the case of a second-hand car you are never completely sure what the work content is to restore the car to its original state. And besides, chances that a mechanic will find issues that were not clear to the customer yet are relatively high.”
The uncertainty with regards to the work content of the second-hand cars also results in another management problem for automotive dealerships; the authorization by the customer. With cars arriving in an unknown state, customers must be contacted for authorization if any issues arise while the car is serviced. Furthermore, it might be necessary to order parts to restore the car to its original state. Since there are a lot of different models of cars from multiple different makes, automotive dealerships will not be able to keep all parts of potential customers in inventory. Both these issues will require time. The authorization process can take up to six hours and the ordering of parts can take up to twenty-four hours. With an average flowtime of forty hours in the current state, the authorization- and parts ordering processes take a relatively large share of the total flowtime, while the exact amount of time is depending on external factors.
the workshop.” While the work content for new cars is often known, and the degree of variation being much higher for second-hand cars, there is a distinction to be made not only between predictable and unpredictable work.
With the work content and the processing times having more variety in the case of second-hand cars, the actual processes within the workplace of an automotive dealership are different. However, the most important differences between new cars and second-hand cars is their effect on the performance of the workshop. It was expected that the costs for holding second-hand cars would be higher compared to the costs for the new cars. This is confirmed by the interview conducted. “It is often said that a car loses one-fifth of its value once you drive it off the dealership’s ground. This is not exactly the truth, but indeed; a car starts to depreciate from the moment it is taken into usage.” Therefore, new cars will not depreciate when they are waiting to be prepared for the sale. Second hand cars do. Therefore, dealerships should aim to make the second-hand-cars flow through their system as fast as possible, in order to generate a minimum of depreciation costs.
With the processes, process times and the costs for holding new cars being different than is the case for second hand cars, there are meaningful differences that affect the performance of the automotive dealership. With the differences, both in terms of processing and cost characteristics, it can be stated that new cars and second-hand cars can both be seen as two different value streams. The following section will present the findings with regards to the capabilities of workload control to manage these differences effectively.
6.2. How can Workload Control effectively cope with these value streams?
In the current state situation, the commonly known First Come First Serve (FCFS) dispatching rule is applied. Therefore, the results of the FCFS dispatching rule will be used as a benchmark when comparing the other dispatching rules.
Rule
F
F
max%T
T
meanT
maxFCFS
39,193 47,027 0,247% 7,318 16,100VS
56,956 93,202 1,534% 17,632 53,093EDD
51,358 67,280 0,040% 3,337 10,191VSEDD
51,327 63,699 0,029% 4,031 11,249VSEDD-V
51,177 64,261 0,075% 4,349 15,316Table 6.2 General performance of the dispatching rules
With regards to the performance of the evaluated dispatching rules, it is noticeable in Table 6.2 General performance of the dispatching rules that not all processed cars fulfil their due date limitations. This applies for all the evaluated rules. The value-stream (VS) dispatching rule does not contribute in improved performance with regards to the due date fulfilment. Actually, the VS dispatching rule is performing tremendously worse than the currently applied FCFS rule. This mainly is a result of the fact that nothing within the VS rule takes into account the due date of the cars. Since due dates are set based on their arrival time, the FCFS dispatching rule does indirectly considers the due date to some degree. The EDD rule is outperforming the FCFS rule with regards to due date fulfilment. With a decrease of almost 90% in the number of jobs that fail to fulfil their due date, the EDD rule performs, as expected, best on the fulfilment of due dates.
Furthermore, Table 6.2 demonstrates that the mean flowtimes will increase with at least 30% compared to the current FCFS rule if the other evaluated rules are applied. This can be declared by the fact that the alternative rules will likely give priority to second hand cars for the following reasons. While the VS rule attempts to make second-hand cars flow through the system as fast as possible, the EDD rule will prioritise second-hand cars due to their tighter due dates set. With both rules being included in the composite dispatching rules, this will also count for the VSEDD and the VSEDD-V rules. However, since the aim is to find the dispatching rule which most effectively makes the different value streams flow through the system, it is important to analyse the flowtimes per car segment. This can be demonstrated by comparing the mean flowtimes (F(x)) and maximum flowtimes (F(x)max) per car segment as
Variable
FCFS
VS
EDD
VSEDD
VSEDD-V
F(1)
70,345 67,110 63,222 63,072 62,908F(2)
69,894 67,121 62,956 62,810 62,724F(3)
69,860 67,630 62,892 62,973 62,918F(4)
31,845 54,481 48,588 48,560 48,397F(1)
max 79,768 74,649 66,615 66,476 65,618F(2)
max 78,170 74,488 65,803 65,484 65,283F(3)
max 79,572 73,977 65,006 65,332 65,083F(4)
max 39,438 97,689 68,386 64,147 64,765Table 6.3 Mean flowtimes (F(x)) and maximum flowtimes (F(x)max) for evaluated rules
With regards to the flowtimes of the different value streams, it is noticeable in Table 6.3 that the flowtimes of second-hand cars, which are categories 1, 2 and 3, for the alternative dispatching rules are more than twice as long as the flowtime of new cars with the FCFS rule applied. This can be explained by the fact that second-hand cars need more processes and generally have more waiting times than new cars have, as explained in section 6.1. Furthermore, another important thing to notice is that all the alternative dispatching rules decrease the flowtime of second-hand cars, while the flowtime of new cars is increased. In that respect, all rules seem to increase the performance of the dealership regarding the depreciation costs for all cars. Additionally, Table 6.3 demonstrates that the dispatching rules with the EDD rule incorporated balances the maximum flowtimes of all different car segments. Table 6.4 provides an overview of the mean depreciation cost for all cars and per car segment.
Variable
FCFS
VS
EDD
VSEDD
VSEDD-V
m
Dep
£37,12 £35,59 £33,39 £33,34 £33,27m
Dep(1)
£72,14 £68,82 £64,83 £64,68 £64,51 mDep(2)
£47,72 £45,82 £42,97 £42,88 £42,81 mDep(3)
£28,62 £27,71 £25,76 £25,79 £25,77Table 6.4 Mean depreciation, both total (mDep) as well as per car segment (mDep(x)) for the evaluated rules
belongs to, the FCFS rule is outperformed by the VS rule with regards to the mean depreciation costs for all car segments. So, while Table 6.2 demonstrated that the VS dispatching rule was outperformed by the FCFS rule with regards to the mean flowtime, the mean depreciation costs for the VS rule are lower than is the case for the FCFS rule. However, since due date fulfilment of the VS rule is very disappointing, the VS rule on its own will not be suitable for automotive dealerships. However, due to its relatively good performance on the depreciation costs, the VS rule can be off value within a composite dispatching rule. Table 6.4 demonstrates that the VSEDD rule outperforms the VS rule on all the segments and thus also the overall mean depreciation, while compared to the EDD rule the VSEDD rule performs better on segments 1 and 2, with a slightly better overall performance for the mean depreciation. In imitation of the VSEDD rule, the VSEDD-V rule arose. It is an amplification in the sense that it basically schedules jobs on the same way the VSEDD rule does, with the value of the car added as a priority criterion.
With the VSEDD-V rule performing best with respect to the mean depreciation of the cars, it is now time to examine how the evaluated dispatching rules perform with regards to the tardiness of jobs of the different value streams, which is the second important performance criterion.
Variable
FCFS
VS
EDD
VSEDD
VSEDD-V
%T(1)
1,910% 1,268% 0,114% 0,176% 0,128%%T(2)
1,491% 1,449% 0,280% 0,928% 0,860%%T(3)
1,404% 1,369% 0,204% 0,252% 0,173%%T(4)
1,795% 1,688% 0,000% 0,000% 0,000%T(1)C
mean 6,531 5,273 0,928 1,427 0,971T(2)C
mean 7,162 6,314 2,469 2,778 3,919T(3)C
mean 7,285 7,354 2,510 3,069 2,436T(4)C
mean 1,986 18,795 0,000 0,000 0,000Table 6.5 Percentage of tardy jobs (%T(x)) and mean conditional tardiness (T(x)Cmean) per segment
While the VS rule outperforms the FCFS rule with regards to the conditional mean tardiness of segments 1 and 2, the VS rule will perform slightly worse on the mean tardiness of segment 3. While the VS rule performs marginally better with regards to the percentage of tardy jobs in segment 4, the mean tardiness of the VS rule is almost ten times the mean tardiness of the FCFS rule. It is expected that this will influence the overall performance of the VS rule. Furthermore, with regards to due date fulfilment it was expected that the dispatching rules in which the EDD rule is involved will outperform the rules in which it is not. This, off course, can be declared by the fact that de EDD rule is developed to achieve a high level of due date fulfilment. Therefore, it was expected that the EDD rule would perform best on this criterion when the elementary dispatching rules are considered.
With regards to the composite rules evaluated, at first glance it is an exciting contest. First of all, all the rules managed to fulfil all jobs of segment 4 within the due date. This can be clarified by the wide span set for due dates of new cars. With regards to the other variables, it is hard to tell which rule performs best overall. With the EDD and the VSEDD-V rules both having managed to outperform each other on specific variables, the VSEDD rule did not manage to outperform the other two on any single variable. However, it is hard to tell which dispatching rule did perform best on this performance criterion. Therefore, to be able to tell which one does perform best, the performance criteria presented so far will be translated to monetary values according to the explanation in section 5.5. A monetary performance indicator will make the differences more tangible and give better insights in the overall performance of the dispatching rules within automotive dealerships. With the depreciation being directly linked to the flowtime of jobs and the penalties for tardiness being directly related to the due date fulfilment of jobs, a combination of these two criteria provide a good criterion for the overall performance assessment of the evaluated dispatching rules. Table 6.6 illustrates the total cost of depreciation and penalties for tardiness for the evaluated dispatching rules. The results will again be benchmarked against the FCFS rule since that is the rule currently applied.
Variable FCFS VS EDD VSEDD VSEDD-V
Depmean £37,11775 £35,58606 £33,39 £33,34 £33,27
Tpenmean £0,11741 £1,99490 £0,00790 £0,00788 £0,00628
TCmean £37,23516 £37,58096 £33,39964 £33,34704 £33,27959
First thing to notice is that the VS rule performs worse than all the other rules. With the total cost of depreciation being lower than is the case for the FCFS rule, the tardiness of the jobs causes a major source of costs which makes the VS rule perform worst with regards to the mean total costs per car. The penalties given for the tardiness of late jobs regarding new cars with the VS rule applied, account 95,49% of the total penalties awarded for lateness. Therefore, we could state that the VS rule is overemphasizing the importance of second-hand cars flowing through the system faster, while neglecting the importance of due date fulfilment.
Furthermore, Table 6.6 demonstrates that the performance of the EDD rule outperforms the FCFS rule both on the mean total cost of depreciation as well as on the mean penalties given for tardy jobs. With less tardy jobs, and second-hand cars flowing through the system faster the total costs decreased with 10,44%. Furthermore, while the VS rule performs worst being an elementary dispatching rule, Table 6.6 demonstrates, again, that the VS rule can complement other dispatching rules. The VSEDD rule performs better than the EDD rule on its own. Its extension, the VSEDD-V rule, performs slightly better than the VSEDD rule. The following section aims to explain this difference.
The better performance for the VSEDD-V rule can be explained by the fact that a relatively large share of the total costs is derived from the total depreciation costs. Since those are closely related to the value of the car, the results show that adding the value of the car as a prioritization criterion can have a positive effect on the total costs made. However, with cars having a relatively high value, the differences between the total depreciation costs of the VSEDD rule compared to the VSEDD-V rule are very small. By performing a paired samples t-test, it is determined that the differences between the total depreciation costs for both rules is not significant with a p-value of 0.886. Therefore, it could be stated that the VSEDD-V rule does not perform significantly better than the VSEDD rule.
7. Discussion
In this chapter, the findings as presented in the previous chapter will be discussed and related to the literature. Furthermore, the limitations of this research and recommendations for future research will be given.
In the first section of the previous chapter, it is determined that there is a new distinction to be made within the value streams of workshops of automotive dealerships. While literature made a distinction between predicable and unpredictable work (Brunt and Kiff, 2007), this study proves that there is a distinction between new cars and second-hand cars to be made.
From the second part of the previous chapter, it can be derived that the evaluated dispatching rules all can have their effect on the performance of the workshop of automotive dealerships. However, the performance of the workshop differs among each rule. While aiming to examine how workload control can effectively cope with the different value streams within the workshop of an automotive dealership, the elementary VS dispatching rule was expected to most clearly define the effect of the dispatching rules with regards to the different value streams. Therefore, the VS rule will be the first rule to be discussed in this section.
The VS dispatching rule was developed to make the second-hand cars flow through the system faster, and it is demonstrated in the previous chapter that the VS rule succeeded on that objective. With a better performance on the total depreciation costs for all the different car segments, the rule proves to reduce the lead times for all cars. However, since the rule does not take into account the due date for the different cars at all, it is assumable that a relatively large share of cars will not fulfil their due date. The costs derived from penalties for late cars will be very high, which makes the standalone VS dispatching rule perform very poor.
rule prioritize the second-hand cars very often.
However, both rules can still be combined into a composite rule, which lays focus even more on the categorisation based on the value streams. The findings demonstrate that with the standalone elementary VS dispatching rule being outperformed by the currently applied FCFS rule, the VS dispatching rule improves the performance of other rules, when they are combined. This is in line with the literature, since it has been proven that composite rules are able to serve more objectives of the schedule (Land, 2004). The VSEDD rule serves customers on time, while there is enough emphasis on the minimization of the flowtimes of second-hand cars to decrease depreciation cost.
The VSEDD-V rule did not perform significantly better than the VSEDD rule. It is expected that the prioritisation of cars based on their value did not change the schedule very much. While it is very unlikely in the used simulation model that cars will have the exact same due date, the prioritisation on value will probably have been redundant. For dealerships that have to cope with batch arrivals and deliveries with varying values of the cars, the value of the car could possibly increase the performance of the schedule. Furthermore, this unveils that it is important to base dispatching rules on characteristics of inputs in which not too much variance is possible. While due dates of cars van vary only a millisecond, the category of the car can only be four different options. With the prioritisation based on the categories being very clear, the differences between the VSEDD and the VSEDD-V rule were not significant.
With the findings of this study having practical, as well as theoretical relevance, there are some limitations and recommendations for further research to be mentioned. First, there are many more dispatching rules that can be applied to the environment of automotive dealerships. With this study aiming to examine whether workload control can effectively cope with the different value streams, future studies can aim to determine which dispatching rules perform best for automotive dealerships in general.
