Dynamic and highly-distributed configuration of supply webs.

摘要

Driven by global competitive pressures, companies are converting from vertically-integrated monoliths to core-competency-based lean organizations. This change has led to increasing dependency on outsourcing for fulfilling engineering and manufacturing needs. In such an environment, characterized by constantly changing product variety, unpredictable demands, and short life cycles, the use of dynamic supply web formation and management strategies can potentially be more efficient and effective at minimizing risk and managing uncertainties as compared to conventional approaches of long-term contracts. This process has also been influenced by use of the Internet, which has allowed unprecedented levels of flexibility in customer demand fulfillment.;This dissertation addresses the need for fundamental research on rapid and effective highly-distributed decision-making methodologies and systems for supply web interactions that take advantage of rapid exchange of information. We investigate agent-based paradigms for supply web interactions and study dynamic decision-making strategies, incorporating real-time information, that enhance the lead time and cost performance of supply webs.;We view the supply web environment as a collection of autonomous entities that exchange messages to form alliances (supply-web configuration). Once configured, these entities engage in highly-distributed collaborative decision-making and coordinating activities (supply-web orchestration) within the resulting supply web. This vision is a significant departure from conventional thinking and current practice due to its emphasis on dynamic configuration of supply-webs and the use of a highly-distributed decision-making paradigm for their orchestration.;As a part of this research, we have developed a generic agent architecture for representing interacting entities in the supply web environment, formulated modeling constructs task and entity representation, and developed decision-making strategies (a) for outsourcing, (b) for dealing with pending bids, and (c) for bid-assembly. An agent-based simulation test bed (developed as a part of this research) has been used to study the interactions among entities while they engage in supply web configuration and orchestration.;This research represents a step towards the advancement of knowledge and understanding of strategies for dynamic configuration and highly-distributed orchestration of supply webs, and will help support the development of supply web decision systems to support this new paradigm.