PurposeThe aims of this paper is to investigate simulation‐based optimisation and stochastic dominance testing while employing kanban‐like production control strategies (PCS) operating dedicated and, where applicable, shared kanban card allocation policies in a multi‐product system with negligible set‐up times and with consideration for robustness to uncertainty.Design/methodology/approachDiscrete event simulation and a genetic algorithm were utilised to optimise the control parameters for dedicated kanban control strategy (KCS), CONWIP and base stock control strategy (BSCS), extended kanban control strategy (EKCS) and generalised kanban control strategy (GKCS) as well as the shared versions of EKCS and GKCS. All‐pairwise comparisons and a ranking and selection technique were employed to compare the performances of the strategies and select the best strategy without consideration of robustness to uncertainty. A latin hypercube sampling experimental design and stochastic dominance testing were utilised to determine the preferred strategy when robustness to uncertainty is considered.FindingsThe findings of this work show that shared GKCS outperforms other strategies when robustness is not considered. However, when robustness of the strategies to uncertainty in the production environment is considered, the results of our research show that the dedicated EKCS is preferred. The effect of system bottleneck location on the inventory accumulation behaviour of different strategies is reported and this was also observed to have a relationship to the nature of a PCS's kanban information transmission.Practical implicationsThe findings of this study are directly relevant to industry where increasing market pressures for product diversity require operating multi‐product production lines with negligible set‐up times. The optimization and robustness test approaches employed in this work can be extended to the analysis of more complicated system configurations and higher number of product types.Originality/valueThis work involves further investigation into the performance of multi‐product kanban‐like PCS by examining their robustness to common sources of uncertainties after they have been initially optimized for base scenarios. The results of the robustness tests also provide new insights into how dedicated kanban card allocation policies might offer higher flexibility and robustness over shared policies under conditions of uncertainty.
The cost of acquiring commercial simulation packages is considerably high and this might explain why organisations are often reluctant to make further investments on training and retraining of employees on simulation modelling. Ironically, the level of benefit derived from simulation is highly dependent on experimental, analytical and statistical skills of the user. These cost and skill requirements make simulation an unattractive decision support tool to SMEs and small multinational organisations. Proposed in this study is ManPy, a semantic-free opensource approach to discrete event simulation (DES), such that users with different levels of skills can derive considerable benefits from simulation. ManPy eradicates the high investments required for simulation modelling by making it possible for low skilled users to benefit from readily available generic modelling objects which are contributed to an open source platform by highly skilled simulation practitioners, statisticians and academics. Another benefit of ManPy is the ability to integrate with other enterprise planning tools for system knowledge extraction and real time simulation input data. Some of these benefits are demonstrated through the implementation of ManPy in a SME that specialises in rapid prototyping and rapid tooling.
The work presented in this paper is part of a project aimed at streamlining and improving the process flow at a leather furniture manufacturing company. The manufacturing throughput time is highly variable, and this makes planning difficult for the assembly of components at the downstream stages. Throughput time predictability at the upstream stages where the components are manufactured would facilitate the planning of their assembly according to their expected arrival times for specific product models. Research conducted in a previous phase of the project showed that the application of the CONstant Work In Progress (CONWIP) control mechanism to regulate inventory yielded significant improvements in the throughput time's mean and variation. However, as it is the case with tighter control of inventory in manufacturing, previously unrealised problems were exposed in relation to the selection of the product model to release into the CONWIP loop. This has significant impact on the balance of the distribution of workload across the system's workstations and among the multi-skilled teams at one of the workstations. This research implements a nested configuration of the Paired-cell Overlapping Loop Of Cards with Authorisation (POLCA) and the Generic Kanban control mechanisms to achieve a balance of the workloads. This ensures a synchronised flow of the different product mix through the entire manufacturing system.
This research is focused on improving the flow of items through the cross-trained teams of a leather furniture manufacturing company, which manufactures a high mix of products. Presently, a Push control strategy is applied to control production, but this causes uneven build-up of items for processing (i.e. workload) at the cross-trained teams. Hence, this research investigates the application of the CONstant Work In Process (CONWIP) strategy to control Work in Progress (WIP) and at the same time ensure balanced workload distribution amongst the teams. It uses the release signal from downstream to monitor the work rate of individual teams and regulate the release of new items for them to process. Results of simulation experiments conducted on the system show that the application of CONWIP, particularly with consideration of the cross-trained teams in its item release decisions, ensures a balanced distribution of workload amongst the teams. This eradicates the constant need for human intervention to redistribute items between the cross-trained teams, which is a current challenge for the case study company.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.