Development of modules for the simulation of manufacturing networks

Abstract

Simulation is widely used to design and optimize manufacturing systems. However, the creation of simulation models becomes a very complicated task when the system under study has a high number of elements and interconnections between them. This is why simulation is complex and time-consuming in the case of global production networks (GPNs). This type of manufacturing network is used nowadays by companies of any size, but the complexity of the material and information flows, which occur throughout a large number of production sites, implies that it is difficult to achieve their full potential. Therefore, the need for improving the current methods for creating simulation models arises. This way the simulation of global production networks will be more accessible regarding the required time and cost and this will allow the improvement of the design and operation of such networks. Ontologies represent domains by describing concepts (classes) from the domain and the relations between them. Among the wide variety of their applications, they have been used to represent processes from manufacturing systems, as they present the information in a structured way and with a defined terminology that supports the interoperability of different applications. This thesis proposes a method that uses these ontologies to identify simulation modules that can be later implemented in a simulation software. The method is based on the examination of the ontology classes to identify the modules and on defining the interfaces between them by looking at the ontology relations. The method is later applied for an ontology that models the order processing in GPNs, in order to identify and develop general simulation modules. The resulting modules facilitate the creation of simulation models by grouping several functions and avoiding the repetitive task that would imply modelling these functions separately. Another advantage of the modules is that they are reusable for different cases of GPNs. Finally, the general modules have been instantiated for specific examples to prove that they allow the representation of the systems in a valid way.