Generative Modeling of Holonic Manufacturing Execution Systems for Batch Plants

MILAGROS ROLÓN; MERCEDES CANAVESIO; ERNESTO MARTINEZ

Computer-Aided Chemical Engineering

Elsevier, B.V.

Lugar: Amsterdam; Año: 2009; p. 795 - 800

Problem Statement Today’s manufacturing systems are subjected to unplanned disruptive events and disturbances such as arrivals of rush orders or machine breakdowns and a multitude of interactions in the supply chain which demands automatic rescheduling and replanning strategies. Lacking adaptiviness and learning capabilities seriously limit the effectiveness of conventional control techniques in manufacturing execution systems. For better handling the dynamics at the shop-floor, holonic control architectures are seen as the alternative of choice to incorporate high flexibility and improved robustness against disturbances. Holonic architectures were proposed based on the concept of cooperative and autonomous agents associating an agent to every holon in the system. Valckenaers and Van Brussel (2005) developed an instantiation of the PROSA (Van Brussel et al, 1998) generic architecture for production control systems, augmented with coordination mechanisms inspired by natural systems. In contrast to many decentralized designs, in this system each component is autonomous and possesses local knowledge that is sufficient to accomplish its own task. The task that a single component is unable to finish alone may require the cooperation of a cluster of components. Implicit communication is a means of establishing such cooperation between the autonomous components, and this manufacturing execution system predicts future behaviour to prevent imminent problems from happening. Nevertheless, this agent-based model doesn’t have elements to validate and verify the robustness and requirement fulfilments of the control manufacturing execution system. Moreover, the model needs to be straight implemented in the manufacturing plant, without a previous test to indicate the probable impact of the Manufacturing Execution System (MES) applicability. Generative Modelling of Complex Adaptive Systems Generative modelling refers to the use of computational models to understand complex systems. This method make possible the creation, analysis and experimentation with agent-based models, as a previous test to validate the effectiveness and efficiency of the model. In this paper, a generative modelling of a manufacturing execution system inspired in the PROSA reference architecture is presented in order to evaluate the emerging behaviour and macroscopic dynamics of a holonic MES in a multipurpose batch plant. To exemplify the proposed approach the generative model was implemented in Netlogo, a software environment specifically designed for Agent Based Modelling and Simulation (ABMS). Different alternatives for the interaction mechanisms of the agents involved are compared, in particular obtaining some experimental results with regard to the kind and quantity of the available information for the other agents, the advantages and disadvantages of the heterogeneous objectives and the coordinator mechanisms role. Results and Significance Generative modelling of holonic MES for batch plants vividly demonstrate the importance of addressing the dynamic complexity of complex adaptive systems via simulation., Since current control systems in use in industry lack the ability to adapt to dynamic manufacturing environments whilst the holonic architectures designed needs a real implementation for its testing, the generative modelling proposed in this research could be instrumental for designing the next generation of batch control systems. References Valckenaers, P., Van Brussel, H., (2005), “Holonic Manufacturing Execution Systems” in CIRP Annals, Pp.427-432. Van Brussel, H., Wyns, J., Valckanears, P., Bongaerts, L., Peeters, P., (1998), “Reference Architecture for Holonic Manufacturing System PROSA” in Computers in Industry, Vol. 37, Pp. 255-274. Keywords: Generative Modelling, Agent-Based Simulation, Batch Control, Manufacturing Execution System