Towards a Highly Flexible Manufacturing System for Mass Personalisation: Exploring Nature-Inspired Models

Abstract

Manufacturing companies in the 21st century are faced with new challenges, such as shorter product life-cycle, market fluctuations, and demands for personalised products. The concept of mass personalisation has been proposed as a potential solution to meet the necessary requirements to address these challenges, where customers can co-create and co-design products based on personal preferences, and the resulting personalised products are produced at costs competitive to that of mass production. However, manufacturing the resulting personalised products in the real world brings about a number of feasibility issues -- such as how to manufacture the personalised products at costs comparable to that of mass production; how to automate the production process without production planning being done ahead of actual production, since order arrival time and types are not known in advance; and how to manage individual product routing as the different products may require different resources at different time during production. To address these challenges, this research aimed to develop concepts and techniques to improve material handling flexibility and layout reconfiguration, such that personalised products are independently and autonomously routed to different resources depending on their operations requirements, and layout reconfiguration is executed in real-time to minimise costs of material handling, which accounts for the larger part of production cost. The novel approach proposed here to achieve the aim of the research is as follows: First, it goes beyond the long-standing assumption of static production machines and introduces mobile production machines which can change their location in real-time depending on the mix of incoming customers orders. Second, it proposes the use of mobile, autonomous and intelligent robots to handle the movement of materials within the factory, allowing independent routing of different products. Third, a heterarchical control architecture based on holons and isoarchic interaction structures is proposed to coordinate the mobile production machines and independently routed product orders. Fourth, the resulting real-time layout optimisation problem is formulated as a quadratic assignment problem and resolved using an ant-based heuristics technique. Finally, a “digital twin” simulation layer is introduced to guide the adaptation of the system's layout by extrapolating from present conditions in “fast forward” mode. The coordination of holons in this layer is based on stigmergy, and the optimisation techniques used are inspired by the ant colony optimisation and the “beeclust” algorithm -- inspired by the behaviour of young-honeybees in a temperature gradient field. Overall, the proposed approach of mobile production machine; mobile, autonomous and intelligent robots for material handling; the two-layered architecture; and the formulated real-time layout optimisation problem including the proposed heuristics solution are the main contributions to knowledge presented here. These contributions integrate the perspectives of three different scientific areas: holonic manufacturing, manufacturing optimisation and self-organising systems. The three papers comprising this thesis cover each of these three areas. The contributions are evaluated using a realistic shoe personalisation scenario with distinct product units and constantly changing product mix. The evaluation criteria are based on two of the major requirements to achieve mass personalisation, which are the production of personalised products in the presence of frequent and unexpected changes in product mix, and competitive production cost with mass production. The results demonstrate that these two major requirements are supported by the proposed approach, and thus the aim of this research has been achieved. Indeed, the capability for mass personalisation can be supported by designing manufacturing systems with mobile production machines and t

Date of Award	3 Jan 2020
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Nikolay Mehandjiev (Supervisor) & Cesar Marin Pitalua (Supervisor)