As two major decisions in the supply chain management, both resource management and sourcing strategies decide ‘which product goes where at what amount and what time’, in order to redeploy outbound/inbound resources to synchronize product flow and eventually fulfill customers’ requirements. Nowadays, the difficulties in making such decisions are brought by many factors such as the emergence of globalization, the increasing variety of products, and uncertainties in both demand and supply. In this dissertation, quantitative methods are developed to efficiently use of all outbound/inbound resources for solving a multi-level and multi-period supply chain system design and planning problem.
Stochastic resource management model is a five-phase quantitative approach to decide facility location, capacity allocation, and production/shipping volume of the good flow in a multi-stage manufacturing supply chain. In the proposed model, manufacturing system in each facility is configured (forming manufacturing cells) at the same time when designing the supply chain network. In order to handle highly fluctuating demand, a layered mini-cellular system is developed, where product families with relatively stable demand are grouped and assigned to their dedicated cells while shared and remainder cells handle the product families with unstable demand. In addition to strategic and tactical decisions, phases IV and V also decide the production plans and shipping plans to further study the operational decisions.
Once the assignment of inbound resources has been determined by the resource management model, the long-term sourcing model provides a dynamic framework to decide the quantity allocation of outbound resources. The goal of sourcing model is to capture supply uncertainties where supplier performance changes along the time or based on purchasing decisions. The proposed framework integrates simulation module and single-period optimization module. The former module simulates various changing patterns of supplier performance. The latter one introduces fuzzy programming to identify the most ‘satisfied’ purchasing decision with multiple objectives such as minimizing cost, improving quality and service level.
The proposed models are experimented with a global jewelry manufacturing company with highly unstable demand. The results of stochastic resource management model generally outperform those of deterministic model in terms of capacity requirements, service levels and costs. The impacts of parameters (minimal demand coverage, reserved capacity, stage configuration) on both stochastic model and deterministic model are also investigated to obtain an understanding of appropriate parameters. With regard to sourcing model, various single-period optimization models are first analyzed with different parameter settings, and an appropriate optimization model is chosen. The changes of purchasing decisions and supplier profiles along the simulation time are discussed in detail. The perfect alignment of experimentation results with real-world phenomenon demonstrates the ability of capturing supplier dynamics and making the right sourcing decisions accordingly.