Keywords

1 Introduction

Manufacturing organisations are forced to always continuously monitor their production processes in order to identify the inefficiencies present within their processes and thereafter implement strategic changes that could help to alleviate these inefficiencies, thereby improving the productivity of a manufacturing organisation (Li 2013). The study of Lodding (2013) indicated that there is a high chance that some or all of the production processes/workstations in an organisation could experience limited manufacturing capacity, which thus make these processes/workstations to be tagged as bottlenecks/constraints experienced in the organisation. According to Kolinski and Tomkowiak (2010), Hsaio et al. (2010) and Betterton and Silver (2012), bottlenecked workstation is defined as a workstation that limit the efficiency of a production process, thereby exhibiting lowest level of productivity amidst other workstations present within an organisation. Bottlenecked workstations tends to dictate the throughput rate exhibited by a system, thereby contributing to the phenomena of high waiting of WIP inventories, backlogged customer order and significant amount of underutilisation of subsequent workstations (Li 2009).

Theory of Constraint (TOC) has been identified by previous researchers to address this issue. According to Goldratt et al. (2004) and Buddas (2014), the various steps that needs to be undertaken to apply TOC in an organisation include identification of the system bottleneck, decision on how to exploit the system bottleneck, subordination of bottlenecked workstation and elevation of the bottlenecks. A lot of studies such as Urban and Rogowska (2018), Sari et al. (2019), and Lizarralde-Aiastui et al. (2020) have used myriads of methods to identify bottlenecks present in an organisation. However, to the best of the authors’ knowledge, there are no case studies within the trailer manufacturing organisation that has applied simulation models in this regard. Simulation is defined as a means of imitating the behaviour of a system by means of a model (Kikolski 2016). Simulation model could be deployed as a means determining the performance of the current system used in an organisation as well as facilitate decision making that will help improve the performance of a system. Simulation promote a holistic picture of the behaviour of processes used in an organisation. It could also be used, when it is not feasible to obtain a solution to an industrial problem by means of analytical method or experiments (Ojstersek and Buchmeister 2020). Simulation could also be used to track myriads of operations carried out in an organisation for a period of few hours to years, if need be, hence, suitable to achieve short and long term decision making (Ojstersek et al., 2020). Simulation could be deployed as a means of pinpointing anomalies that could negatively affect the performance of production processes used in an organisation (Parv et al., 2021). However, in the context of this study, simulation will be deployed, on the one hand, to identify the bottlenecks contributing to high customer order backlogs experienced in a trailer manufacturing organisation. On the other hand, it will be used as means of validating the strategies that could be used to elevate the identified bottlenecks and subordinate other workstations, towards reducing the backlog experienced in the organisation as well as exponentially improve the productivity achieved in the organisation.

The structure of the paper is highlighted as follows. Section 2 present the methodology deployed in this study towards developing a simulation model for the case study trailer manufacturing organisation. Section 3 present and discuss the results of the simulation model while the last section conclude based on the results obtained from the study and thereafter unveil the future research work.

2 Methodology

A quantitative research design approach premised on the use of a discrete-event simulation modelling approach was deployed in this study. The first phase of the simulation modelling approach involves the observation and note taking of the production processes used in the case study trailer manufacturing organisation by the authors. The second phase involves the collection of historical data such as the work station cycle time, number of customer orders per day, capacity of the work station and the work order release method used in the organisation. A summary of the data collected through system observation and elicited from archival data for use in this study is depicted in Table 1.

Table 1. Summary of data collected from a Trailer Manufacturing Organisation
Full size table

The third phase involves the modelling and parametrization of the current work ordering release and production mechanism of the trailer manufacturing organisation using AnyLogic software package. The summary of the procedure used to model and parametrize the current work ordering release and production mechanism of the trailer manufacturing organisation is highlighted as follows:

  1. 1)

    Drag a source node, twelve stations, eleven queues (where customer order and WIP inventories will wait when the stations are busy), and a sink node into the AnyLogic interface.

  2. 2)

    Drag four points into the interface. Name the first two points as Start 1 and Start 2 and the last two points as End 1 and End 2 respectively. The first two points measure the time customer order spent waiting before it is been released for processing while the last two points measure the lead time and flow time to produce a trailer product respectively.

  3. 3)

    Parametrize the name, order arrival/day, agent per arrival and maximum number of order arrivals of the source node as Trailer_Orders, uniform.discrete(3, 6), 1 and 600 respectively.

  4. 4)

    Activate the enable exit on timeout, with a view to ensure that customers withdraw their orders, if a customer order is not released for production within a period of 7 days. Define the queueing rule to be FIFO and the queueing capacity to be 100.

  5. 5)

    Set the processing time of each workstation to be the processing time of each associated workstation captured in Table 1.

  6. 6)

    Use the Parameter function on the AnyLogic software to parametrize various metrics such as manufacturing cost, backlog cost and revenue generated by the organisation.

  7. 7)

    Deploy the Chart function on AnyLogic software and parametrize it with Revenue (R) parameter, Cost (C) parameter and Backlog Cost (Z) parameter respectively.

  8. 8)

    Set the value of C as sink.count() x production cost/order

  9. 9)

    Set the value of R as sink.count() x selling price/order

  10. 10)

    Set the value of Z as {sourceorder.count() – sink.count()} x backlog cost/order

  11. 11)

    Define another parameter, Profit (P) under the Chart function and set its value as R – (C + Z).

  12. 12)

    Create a time plot graph function to visually display the time spent in processing WIP units against the time spent to wait for the dispatch of the order to the production line for processing. Therefore, set the first data set name to plot as wait for manufacturing and its value as Queue_to_manufacturing.StatsSize.Mean and set the second data set name to plot as WIP and its value as queue1_CNCbendingmachine_Statsize.Mean() + queue2_treatmentstation_Statsize.Mean() + queue3_primerpaint_Statsize.Mean() + queue4_tacweld1_Statsize.Mean() + queue5_robotweld1_Statsize.Mean() + queue6_tacweld2_Statsize.Mean() + queue7_robotweld2_Statsize.Mean() + queue8_chassisassembly_Statsize.Mean() + queue9_painting_Statsize.Mean() + queue10_finalassembly_Statsize.Mean() + queue11_qualitycontrolinspection_Statsize.Mean()

  13. 13)

    Run the statistics of the average queue length for the entire system by using queue.Statsize.Mean().

  14. 14)

    Create a backlog cost function to display the number of orders backlogged, which deploys a value; {sourceorder.count() – sink.count()} x backlog cost/order, to generate the plot.

  15. 15)

    Measure the average utilisation of each workstation by means of StatsUtilisation.Mean() function. Plot the results on the Bar Chart by using the values of: lasercutting_ StatsUtilisation.Mean(), CNCbendingmachine_ StatsUtilisation.Mean(), treatmentstation_ StatsUtilisation.Mean(), primerpaint_ StatsUtilisation.Mean(), tacweld1_StatsUtilisation.Mean(), robotweld1_ StatsUtilisation.Mean(), tacweld2_StatsUtilisation.Mean(), robotweld2_ StatsUtilisation.Mean(), chassisassembly_StatsUtilisation.Mean(), painting_ StatsUtilisation.Mean(), finalassembly_ StatsUtilisation.Mean(), and qualitycontrolinspection_ StatsUtilisation.Mean(), respectively.

  16. 16)

    Lastly, create a time plot to visually display the flow time for the orders processed using the appropriate data set.

The last research phase involves the design of experiment, which comprises of simulation of suitable strategy that have the potential of elevating bottlenecks present in this manufacturing organisation as well as reducing the customer order backlog. The developed simulation model of the trailer manufacturing organisation considered in this study is presented in Fig. 1.

Fig. 1.
figure 1

System Model of the trailer manufacturing organisation

Full size image

3 Results and Discussion

3.1 Simulation Results of the Performance of the Manufacturing System Used in the Trailer Manufacturing Organisation

3.1.1 Results for the Revenue, Cost, Profit, Throughput and Service Level of the Organisation

During the three month simulation, the trailer manufacturing organisation generated a profit of $822909 at a revenue of $2.47 million and a total operations cost of $1.65 million. It is evident that the organisation has an opportunity cost of $3.82 million due to orders that were cancelled by customers who had waited longer than the specified lead time of seven days. The total backlog cost incurred by the organisation accumulated to $4.4 million. In addition, based on the simulation result, the organisation received a total of 370 confirmed customer orders. They managed to complete and deliver 133 of the required orders with a balance of backlog orders that amount to 237. This gives them a service level percentage of 35.95%.

3.1.2 Simulation Results of the Lead Time, Flow Time and Machine Average Capacity Utilisation at the Trailer Manufacturing Organisation

The average flow time to produce a trailer based on the total flow time distribution experienced over a period of three months is 48.58 h. Based on the result of the simulation, one could therefore conclude that the minimum, average and maximum manufacturing time that could be used to complete a customer order are 40 h, 48.58 h and 58 h respectively. After analysing the distribution of the lead time over a period of three (3) months, it is evident that the average lead time to produce a trailer is 174.84 h, which is more than the double of the average flow time. One could therefore already claim that the waiting time for orders to be dispatched is high owing to the manufacturing line experiencing capacity constraints. The capacity constraints experienced by the organisation is evident since on the one hand, there were very long queue of orders, which ranges between 23 to 43 WIP orders waiting to be sent to the manufacturing station. On the other hand, the organisation experienced unbalanced workstation capacities utilisation as depicted in Fig. 2.

Fig. 2.
figure 2

Workstation Average Capacity Utilisation for a period of 3 months

Full size image

Figure 2 revealed that the average capacity utilisation is unequally distributed among the machines with a mean total utilisation of only 28.33%. The workstation with the highest utilisation is the laser machine workstation with a 70% utilisation factor and the least utilised workstations are the final assembly and QC sections with only 0.08%. The sum of the average utilisation possess the required capacity to meet the customer demand but the organisation needs to look at line balancing of the stations that are causing a bottleneck in the process. In turn, they cause the company to suffer backlog cost and unhappy customers. In light of this, the authors pinpointed the laser machine workstation as the primary bottlenecked station while the treatment, bending and primer workstations were identified as the secondary bottlenecked workstations.

3.2 Strategies to Improve the Productivity and Reduce the Backlog Cost of the Organisation

In order to exploit and elevate the bottlenecks experienced in the organisation, the TOC and line balancing principle were deployed by the authors, which thus unveils the following strategies:

1) Increment of the capacity of lasers and bending machines used in the organisation by three and two fold respectively, 2) Increment of the capacity of the treatment and primer paint sections by three and two fold respectively, 3) Change from push to pull by utilising one-piece flow, CONWIP, and Heijunka dispatching principle, and 4) Better order management at the sales site to reduce the variability of incoming orders using workload control and customer enquiry management principle.

The ideology behind this order management principle is that the planning section now has an order release mechanism to control the amount of work on the shop floor. After applying better order management, the sales and planning team in the organisation is expected to receive live feedback from the production team and can now accept more orders per day. Therefore, only the pre-shop pool waiting time is considered to vary. Since the shop floor workload is stabilised, more shop floor throughput time is allowed. The total capital required to purchase the new machines as well as hire additional labourers required at the treatment and primer paint sections would increase the total operating cost of the organisation by $12993 which was therefore, factored into the improvement model.

3.2.1 Simulation Results of the Revenue, Cost, Profit, Throughput and Service Level for the Improved Manufacturing System of the Trailer Manufacturing Organisation Using the Proposed Strategies

For the improvement model, during the three month simulation, the trailer manufacturing organisation is expected to generate a profit of $2.48 million at a revenue of $8.26 million and a total operations cost of $5.78 million. It is evident that the organisation is expected to have an opportunity loss cost of $0. The total backlog cost is expected to reduce to $1.04 million. In addition, based on the simulation result, the organisation is expected to receive a total of 501 confirmed customer orders. The organisation is expected to complete and deliver 445 of the required orders with a balance of backlog orders that amount to 56. This will give the organisation a service level percentage of 88.82%.

3.2.2 Simulation Results of the Lead Time, Flow Time, and Machine Average Capacity Utilization for the Improved Trailer Manufacturing System Model

The average flow time to produce a trailer based on the total flow time distribution experienced over a period of three months is 51.09 h. One could therefore conclude that the minimum, average and maximum manufacturing time that could be used to complete a customer order are 36 h, 51.09 h and 66 h respectively. After analysing the distribution of the lead time over a period of three (3) months, it is evident that the average lead time to produce a trailer is 83.88 h, which is closer to the average flow time, as compared to the simulation result of the current manufacturing process of the organisation. A bottleneck occurred at the waiting order station where orders wait to be released into the manufacturing process with an average WIP of 9.84 ≈ 10. Therefore, the organisation is expected to experience a semi-equally distributed workload among the workstations (see Fig. 3) with a mean total utilisation of 57.58%, if the proposed strategies are implemented in the organisation.

Fig. 3.
figure 3

Workstation Average Capacity Utilisation for the improved system model for a period of 3 months

Full size image

The station with the highest utilisation is the laser machine workstation with an 80% utilisation factor and the least utilised workstations are the final assembly and QC sections with 26% utilisation. It can also be observed that the utilisation is more evenly distributed than that of the simulation result of the current trailer manufacturing process. This indicates that the trailer production line is balanced compared to that of the current trailer manufacturing process. The total mean utilisation is much higher than that of the current trailer manufacturing process, which indicates that the company would achieve better throughput. In light of this, it could be inferred that the organisation: (i) trailer throughput in terms of customer order received and processed, and total revenue would increase by 35.41% and 234.59%, (ii) backlog cost, opportunity cost and lead time would reduce by 76.37%, 42.23% and 52.02%, and the (iii) service level, and mean capacity utilisation would increase by 52.88%, and 29.25% respectively, if the trailer manufacturing organisation considered in this study implement the strategies proposed in this study. The process simulation exercise conducted in this study unveil cost-effective solutions that are tailored towards improving the productivity and service level of the trailer manufacturing organisation at a minimized operations cost, while maximizing the capacity utilisation of the organisation.

4 Conclusion

In this study, on the one hand, we explored how system modelling and simulation could be used to validate the strategies that could be used to reduce backlog cost owing to high customer order backlog experienced in a trailer manufacturing organisation. On the other hand, we also investigated how system modelling and simulation could be used to elevate bottlenecks as well as balance the production line of a trailer manufacturing organisation towards improving its productivity and revenue. The result of the simulation exercise pinpointed that the organisation experienced high fluctuations in machine capacity utilisation. It was also observed that the laser machine is the primary bottleneck workstation while the treatment, bending and primer paint processes are the secondary bottleneck workstations. In order to address this aforementioned issue, it was asserted by means of simulation that increment of the capacity of the laser and treatment workstations by three folds, increment of the capacity of the bending machine and primer paint workstations by two folds, and the use of a pull system would reduce the backlog cost experienced in the organisation by 76.37%, as well as increase the machine capacity utilisation and service level of the organisation by 29.25% and 52.88%, if these proposed strategies are deployed in the trailer manufacturing organisation considered in this study. Implementation of the proposed strategies at the trailer manufacturing organisation and measurement of their impact need to be explored in future studies.