The closed-loop food supply chain with returnable transport items (CLFSC-RTI) is an important domain of the circular economy. The literature review identifies several research gaps, as follows: (i) no study explores an RTI sharing strategy for CLFSC-RTI, despite its applications in many sectors and its potential capacity to improve supply chain (SC) performance; (ii) no study investigates the CLFSC-RTI design problem under uncertainty, although facility location or uncertainty can together impact SC efficiency; and (iii) risk management of CLFSC-RTI under disruptions remains unexplored. To address these research gaps, this thesis investigates three novel CLFSC-RTI management problems.
Firstly, a new integrated production and transportation CLFSC problem with RTI sharing is investigated, in which multiple manufacturers and retailers are considered. The problem consists of determining production, transportation, and inventory quantities over a time horizon to maximize the total profit of the SC. For the problem, a mixed-integer linear programming (MILP) model is formulated, and a two-phase heuristic algorithm is developed. Experiments on a case study and 225 randomly generated instances are conducted to evaluate the performance of the proposed model and algorithm.
Secondly, an uncertain CLFSC-RTI design problem with multiple manufacturers and retailers is studied, in which only partial information about the uncertain parameters is assumed to be available, i.e., the mean, standard deviation, and lower and upper bounds of the uncertain parameters. The problem consists of determining the locations of manufacturers and coordinating food and RTI flows simultaneously. The objective is to minimize the total expected operational cost from a worst-case perspective. For the problem, a distributionally robust chance-constrained programming (DRCCP) model is constructed. Based on the obtained problem properties, the DRCCP model is approximately transformed into an MILP model. Subsequently, an improved Lagrangian relaxation algorithm is designed. Experiments on 160 randomly generated instances demonstrate the performance of the proposed model and algorithm. Finally, managerial insights are derived via sensitivity analysis.
Thirdly, a new three-echelon CLFSC-RTI resilience improvement problem with human-machine reconciliation and RTI inventory redundancy strategies is investigated, in which uncertain demand, and uncertain production and handling capacities caused by disruption events are assumed to be represented by moment-based ambiguity sets. The problem consists of optimally determining resilience strategies and efficiently coordinating product and RTI flows. The objective is to minimize the total expected operational cost from a worst-case perspective. The problem is formulated by a new DRCCP model. Based on the obtained problem properties, the proposed model is then equivalently transformed into an MILP model. A case study is conducted to provide valuable managerial insights. To handle large-scale instances, an improved relax-and-fix (RF) algorithm is developed. Experimental results on 240 randomly generated instances demonstrate the performance of the improved RF algorithm.
