Self-resilient production systems : framework for design synthesis of multi-station assembly systems

Product design changes are inevitable in the current trend of time-based competition where product models such as automotive bodies and aircraft fuselages are frequently upgraded and cause assembly process design changes. In recent years, several studies in engineering change management and reconfigurable systems have been conducted to address the challenges of frequent product and process design changes. However, the results of these studies are limited in their applications due to shortcomings in three aspects which are: (i) They rely heavily on past records which might only be a few relevant cases and insufficient to perform a reliable analysis; (ii) They focus mainly on managing design changes in product architecture instead of both product and process architecture; and (iii) They consider design changes at a station-level instead of a multistation level. To address the aforementioned challenges, this thesis proposes three interrelated research areas to simulate the design adjustments of the existing process architecture. These research areas involve: (i) the methodologies to model the existing process architecture design in order to use the developed models as assembly response functions for assessing Key Performance Indices (KPIs); (ii) the KPIs to assess quality, cost, and design complexity of the existing process architecture design which are used when making decisions to change the existing process architecture design; and (iii) the methodology to change the process architecture design to new optimal design solutions at a multi-station level. In the first research area, the methodology in modeling the functional dependence of process variables within the process architecture design are presented as well as the relations from process variables and product architecture design. To understand the engineering change propagation chain among process variables within the process architecture design, a functional dependence model is introduced to represent the design dependency among process variables by cascading relationships from customer requirements, product architecture, process architecture, and design tasks to optimise process variable design. This model is used to estimate the level of process variable design change propagation in the existing process architecture design Next, process yield, cost, and complexity indices are introduced and used as KPIs in this thesis to measure product quality, cost in changing the current process design, and dependency of process variables (i.e, change propagation), respectively. The process yield and complexity indices are obtained by using the Stream-of-Variation (SOVA) model and functional dependence model, respectively. The costing KPI is obtained by determining the cost in optimizing tolerances of process variables. The implication of the costing KPI on the overall cost in changing process architecture design is also discussed. These three comprehensive indices are used to support decision-making when redesigning the existing process architecture. Finally, the framework driven by functional optimisation is proposed to adjust the existing process architecture to meet the engineering change requirements. The framework provides a platform to integrate and analyze several individual design synthesis tasks which are necessary to optimise the multi-stage assembly processes such as tolerance of process variables, fixture layouts, or part-to-part joints. The developed framework based on transversal of hypergraph and task connectivity matrix which lead to the optimal sequence of these design tasks. In order to enhance visibility on the dependencies and hierarchy of design tasks, Design Structure Matrix and Task Flow Chain are also adopted. Three scenarios of engineering changes in industrial automotive design are used to illustrate the application of the proposed redesign methodology. The thesis concludes that it is not necessary to optimise all functional designs of process variables to accommodate the engineering changes. The selection of only relevant functional designs is sufficient, but the design optimisation of the process variables has to be conducted at the system level with consideration of dependency between selected functional designs.