PARALLEL SIMULATION FOR CONCURRENT DEVELOPMENT OFMANUFACTURING FLOW AND ITS CONTROL SYSTEM

Abstract: Companies nowadays must innovate to achieve or retain a competitive position in the market. In manufacturing companies the introduction of a new product often requires design of the manufacturing system itself, which greatly increases product development time. Manufacturing system development has been relying lately on simulation models, which decrease the need for hardware testing, but the engineering applications needed for development are isolated. Concurrent engineering has found applications in the interface between the product and its manufacturing system. However, little has been researched in the concurrent development of the several steps of manufacturing system. This report presents a communications method to connect two simulation models in parallel, in two different computers, in what can be called a distributed simulation. One of the models is the flowsimulation modelled as a discrete event simulation (DES), while the other model represents the controlsystems modelled as finite state machines (FSM). Both models run in Matlab/Simulink. This concept allows two developers to work simultaneously in otherwise sequential development tasks, and get validation of their implementation while developing. The communication between the systems is achieved with the OPC protocol, an established technology in networked control systems (NCS). With a simple example model, the system is able to run in parallel, and the effectiveness of the parallel development was observed as the model was adjusted to its distributed format. The main difficulties found during implementation are related with the DCOM configuration necessary for the OPC technology and the setup of data exchange modes (synchronous/asynchronous). The distributed simulation requires real-time execution to run properly and reliably, which results in longer simulation times than single platform simulation. Finite State Machines were also successfully used to model control systems. This technique simplifies development and debugging due to its formal structure and visual interface. Overall, the results of this implementation offer good possibilities of further studies in the application of distributed simulation in concurrent development. This report also lays the path for more complexsimulation using this concept, both in the models used and the number of computers connected in parallel.