What to Share? A Preliminary Investigation into the Impact of Information Sharing on Distributed Decentralised Agent-Based Additive Manufacturing Networks

Distributed Decentralised Additive Manufacturing (DDAM) networks are considered a complementary method to mass/batch production paradigms providing ramp-up, robust and responsive global-local manufacturing capability. One implementation is to use Artificially Intelligent (AI) agents that represent jobs and machines, and broker on their behalf. Fundamental to the brokering process is the sharing of information about the jobs and machines such that machines are able to select appropriate jobs for which they have the capabilities and resources to complete. Here a challenge may exist, in that different jobs and machines may be unable (i.e. due to partial information) or unwilling (i.e. due to IP concerns) to share, then potentially limiting decision-making capability and system performance. This paper examines the nature of information sharing on the performance of a brokered DDAM network. To do so, AnyLogic was used to create a multi-agent simulation. The simulation modelled a DDAM network where the characteristic information shared to the machines by the jobs could be varied. The results of the simulation showed that in general more information sharing boosted system performance, but that different types of information shared by jobs and machines created a varying impact on the performance benefit that is realised and should be prioritised to maximise system throughput.