Multi-agent Framework for Manufacturing Sustainability Analysis and Optimization

Manufacturing sustainability is related to complex adaptive systems crossing different layers; sustainability issues, and their potential resolutions, are transversal to supply and shop-floor levels, often including interaction between them; in addition, multiple actors across the supply chain interact displaying non-linear and non-rational behaviours characterized by feedbacks and time lags. This complexity is nowadays addressed with classical simulation approaches such as System Dynamics (SD), Dynamic Systems (DS) at very high and aggregate level, or Discrete Event Simulation (DES) at very low and disaggregated level. Agent Based Modelling (ABM) approach may potentially address such issues in a more effective way because it exploits the strengths of both aggregated and disaggregated models, while minimizing the drawbacks. The purpose of this paper is to present a self-developed multi-agent based framework as a basis for sustainable manufacturing modelling, simulation and analysis. A literature review on agent based modelling applied to sustainable manufacturing issues is reported to justify choices adopted in the developed tool. Then an illustrative case of a semi-automated food production line is described with the aim of evaluating the self-developed framework through some sustainability and productivity KPIs, and showing easiness and adaptability. The complete agent based simulator and the interaction sequence are reported and validated by an experimental campaign.