AbstractQuantitative modelling of reverse logistics networks and product recovery have been the focus of many research activities in the past few decades. Interest to these models are mostly due to the complexity of reverse logistics networks that necessitates further analysis with the help of mathematical models. In comparison to the traditional forward logistics networks, reverse logistics networks have to deal with the quality of returns issues as well as a high degree of uncertainty in return flow. Additionally, a variety of recovery routes, such as reuse, repair, remanufacturing and recycling, exist. The decision making for utilising these routes requires the quality of returns and uncertainty of return flow to be considered.
In this research, integrated forward and reverse logistics networks with repair, remanufacturing and disposal routes are considered. Returns are assumed to be classified based on their quality in ordinal quality levels and quality thresholds are used to split the returned products into repairable, remanufacturable and disposable returns. Fuzzy numbers are used to model the uncertainty in demand and return quantities of different quality levels. Setup costs, non-stationary demand and return quantities, and different lead times have been considered.
To facilitate decision making in such networks, a two phase optimisation model is proposed. Given quality thresholds as parameters, the decision variables including the quantities of products being sent to repair, disassembly and disposal, components to be procured and products to be repaired, disassembled or produced for each time period within the time horizon are determined using a fuzzy optimisation model. A sensitivity analysis of the fuzzy optimisation model is carried out on the network parameters including quantity of returned products, unit repair an disassembly costs and procurement, production, disassembly and repair setup costs. A fuzzy controller is proposed to determine quality thresholds based on some ratios of the reverse logistics network parameters including repair to new unit cost, disassembly to new unit cost, repair to disassembly setup, disassembly to procurement setup and return to demand ratios. Fuzzy controller’s sensitivity is also examined in relation to parameters such as average repair and disassembly costs, repair, disassembly, production and procurement setup costs and return to demand ratio. Finally, a genetic fuzzy method is developed to tune the fuzzy controller and improve its rule base. The rule base obtained and the results of sensitivity analyses are utilised to gain better managerial insights into these reverse logistics networks.
|Date of Award
|Dobrila Petrovic (Supervisor) & Keith Burnham (Supervisor)
- quantitative modelling
- product recovery
- reverse logistics networks