Anarchic Manufacturing

: A methodology for modelling manufacturing systems as distributed free market systems

Abstract

Manufacturing systems are traditionally organised hierarchically. The hierarchy works well for systems with simple and static organisation of manufacturing resources, where management layers and predetermined rules provide effective production. However, new demands for customised products with rapid delivery, have led to the rise and pervasion of smart manufacturing, where intelligent objects interact in a Cyber-Physical System. In these systems, the organisation of resources and the manufacturing environment tend to be highly complex and volatile. Consequently, centralised and hierarchical systems have exhibited shortcomings as the result of their being too rigid.
This thesis proposes the ‘anarchic manufacturing system’ as a viable alternative for such scenarios. A manufacturing system is defined as ‘anarchic’ if the production planning and control system is decentralised and underpinned by emergent synthesis, utilising a free market structure without central control, coordination or monitoring. Such systems were compared to centralised and hierarchical systems in multi-agent simulation experiments covering three scenarios: simple discrete manufacture, assembly and product transition to identify the affordances of the proposed system over the existing planning and control approaches.
The main contributions of this research are the methodology to model a manufacturing system as a distributed free market system, including advanced assembly and product transition scenarios which were previously unfulfilled, additionally the design principles for anarchic manufacturing and the associated system characteristics. The manufacturing systems were modelled within an agent based modelling environment, enabling advanced individual decision making capabilities that could operate within a free market based system. The design principles outline and justify the free market system and its mechanics, thereby defining how distributed anarchic manufacturing systems create an effective emergent outcome. Anarchic system’s effective deployment to assembly and production transition scenarios is the first of any purely distributed system and demonstrates the retention of distributed characteristics in these scenarios; most notably self-organisation and flexibility.
The experimental results in the thesis demonstrate that centralised and hierarchical systems are not inherently better than distributed systems, and that complexity and volatility can effectively be managed through distributed systems. This thesis replaces the traditional ‘simplify to improve’ mantra in production systems, with ‘embrace complexity to achieve flexibility’ through the anarchic manufacturing system.

Date of Award	26 Nov 2020
Original language	English
Awarding Institution	University of Bristol
Supervisor	Aydin Nassehi (Supervisor) & Chris M Snider (Supervisor)