In this master's thesis, we focus on a problem that is frequent in the industrial world. Since effective production management is very important to ensure business competitiveness, the importance of some known and lesser known algorithms used to improve productivity of production lines has recently been stressed in practice as well. In terms of classic mass production, this aspect is paid too little attention, which results in companies producing products with no final buyer and excessive stocks piling inside systems, which increases costs and reduces companies' profit.
The theoretical part provides a historical overview in the field of production management, describes key terms gradually developed in this field and presents key differences between different types of production process management, emphasising a currently less known algorithm, ConWIP, used to restrict the quantity of tasks released into production process and, consequently, excess work-in-process. Since the dynamics of each production process is unique, it is difficult to develop an algorithm that could be applied to multiple processes, which are similar to one another, yet in terms of management completely different. Due to the need for adaptability, in addition to the basic ConWIP mechanism, we also presented some flexible implementations. Companies may find suitability assessment of individual algorithms in a real process time- and money-consuming; in response to this, a discrete event simulation has been developed in this field, allowing us to test the algorithm on models before its implementation in the real process. In addition to a theoretical background, this paper thus includes a simulation of the ConWIP algorithm on the production line model in the Simulink SimEvents software environment, which is part of the Matlab software tool.
In order to build simulation model and analyse the algorithm, we selected two types of production lines: the first one simple with four successive workstations and second one more flexible and more complex for modelling. Before the final simulation version itself, basic requirements for mechanism operation and test on a simple test case are presented. Simulation results confirmed our hypotheses about the usefulness of such algorithm, and they showed that production costs could be reduced if the algorithm was used in a real process in which amount of work-in-process would not be controlled appropriately.
|
