Towards Self-adaptive Production Systems: Modular Generic Simulation Models for Continuous Replanning and Reconfiguration

Changing factors within markets, such as continually shortening product life cycles, increasing competition due to shorter development and order processing times, manufacturing in intercompany networks and the individualization of customer needs, create new requirements which manufacturing organizations are expected to fulfil. To be successful, manufacturing organizations and business processes have to be transformable to cope with these environmental changes [1]. That's why the creation and implementation of self-adaptive production systems represents one vision of the near future, offering the ability to detect important changes, simulate the impacts of these changes and generate suggestions for an optimised configuration. A first step on the way to self-adaptive production systems is the transition from a nonrecurring, static planning und operation cycle to a continuous replanning and reconfiguration of work systems, supported by new planning methods and tools. In order to realize continuous replanning and reconfiguration, planning methods and tools have to meet new requirements: production modelling should be fast, and the model structure modular in order to enable quick and economical changes by the production planner. Within internal or external production networks, defined sectors have to be planned, simulated and optimised both separately and in combination with others. In temporary inter-company production, as realized in virtual companies, different production models must be able to communicate online with each other using new information technology such as the Internet. The paper presents a solution for an integrated planning tool, based on object-oriented, generic simulation models. The tool supports the production planner in adjusting the production system to changing product demand or corporate goals. It enables the planner in a decentralized organisation to model, simulate and evaluate his area of accountability, and to generate proposals for reconfiguration. Because no specialized knowledge of simulation programming is needed, model composition or input and structure parameter modification can be performed by the planner himself. Flexible production planning modules (FPPMs) are the core elements of the solution. By using these modules, machines and groups of machines can be compounded to subsystems which can then be replanned by a decentralized planning group. Because of the hierarchical composition of such a model, it is also possible to group the FPPMs repeatedly in order to improve the production system as a whole. The interconnection between the individual FPPMs, as well as between the FPPMs and the evaluation and optimisation unit, takes place through plug-and-play interfaces. The generic simulation model is generated using structure data and the FPPMs. After combining different FPPMs to a practical simulation model, the evaluation unit and the optimisation unit are added automatically. During the optimisation cycles, the simulation results are compared to firm-specific goals. The optimisation unit, based on an evolutionary algorithm, creates a population of possible new configurations which are the starting point for the next optimisation cycle. Thereby future production systems will be flexible, mobile, agile and temporary, thus meeting the requirements of the turbulent market environment. The breadth and depth of the production planner's tasks will increase substantially within this environment. With this proposed planning tool, the planner obtains an instrument to realize self-optimisation, to react to changes in his company's objectives and environment