Modelling and simulation of supply chain resilience using complex system approaches

Due to the globalization and increased collaboration between firms, supply chains (SCs) are evolving into supply chain networks (SCNs). The dynamic business environment and growing complexity of global SCNs lead to increasing vulnerabilities in the SCNs to disruptions. Supply chain resilience (SCRES...

Full description

Saved in:
Bibliographic Details
Main Author: Tan, Wen Jun
Other Authors: Cai Wentong
Format: Thesis-Doctor of Philosophy
Language:English
Published: Nanyang Technological University 2020
Subjects:
Online Access:https://hdl.handle.net/10356/137144
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Due to the globalization and increased collaboration between firms, supply chains (SCs) are evolving into supply chain networks (SCNs). The dynamic business environment and growing complexity of global SCNs lead to increasing vulnerabilities in the SCNs to disruptions. Supply chain resilience (SCRES) offers an approach to build an SCN such that it can mitigate the impact of disruptions and provide effective response to recover from the disruptions in a timely manner. However, due to their complexities, it is a challenge to design resilient SCNs. An SCN can be generally characterized into microscopic behaviours and macroscopic properties. In this thesis, macroscopic properties of an SCN are modelled using graphs or system dynamics (SD) models; the microscopic behaviours of the SC operations are modelled using agent-based model (ABMs). Firstly, a top-down approach is proposed to design SCN topologies for hierarchical networks based on SC strategies. A bottom-up approach using ABM is proposed to evaluate the operational performance of the SCN topologies. An SCN is modelled as a hierarchical network and an ABM is developed to model individual SC operations for each entity in the SCN. Through simulation-based analysis, effective SCN topology can be identified to mitigate a particular risk scenario. Secondly, to study the external effects on its internal operations, a firm often needs to construct an SCN model. The firm may have sufficient data to build a detailed model of its operational processes. However, it is difficult to construct the rest of the SCN model at the same level of details since the firm has no control or visibility over the external entities. Hence, a hybrid model of the SCN using integration design is proposed to model the whole SCN using SD and the focal firm using ABM. This hybrid model is, therefore, able to capture different levels of details of the same system. Trade-offs between mitigation and contingency strategies and between backorder cost and ordering cost are analysed through simulation of the hybrid model. Thirdly, hierarchical networks with the same type of material flow between stages limit the applicability of the SCN model on complex multi-stage SCNs. To overcome this limitation, a top-down approach to model an SCN using graph theory is proposed. The proposed model is capable of representing the structural redundancy for strategic planning. Based on the structural analysis of an SCN, all the SCs in the SCN can be identified. An approach to assess SCRES is proposed to measure the structural redundancy using the number of SCs in an SCN. The vulnerability of an SCN can also be assessed by identifying critical plants. To evaluate the proactive mitigation strategy, random disruptions are simulated using real-world SCNs, and it shows that the SCRES can be improved by adding redundant plants. Contingency strategy is also developed to response reactively to the criticality of the disrupted plant. Finally, using structural analysis alone cannot capture the dynamics of recovery of SCN during the disruption. To consider the dynamics of disruption-recovery behaviours, a simulation model is developed by combining the graph model with ABM of SC operational behaviours. Based on structural analysis, mitigation strategies are designed to build redundancy. Contingency strategies are analysed to prioritise recovery on affected SCN. New SCRES indexes are proposed by evaluating the SC performance measures for disruptions of each plant and aggregating the measures based on the criticality of plants in the SCN. Simulation results show that these strategies can be used to build resilience by enabling the SCN to recover faster after disruptions. In addition, mitigation strategies are suitable for long-term disruptions while contingency strategies are more effective for short-term disruptions. In summary, the contributions of this thesis to the current state-of-the-art are: (1) modelling SCN using complex system approaches and (2) building a resilient SCN. SCN models have been developed to represent the complexities in real-world SCNs. These models are used to design a wide variety of different SCRES strategies: proactive mitigation strategies to improve resilient against disruptions and reactive contingency strategies to recover the SCN from the impact of disruptions. Through analysis of SCRES, this enables decision makers to build a more resilient SCN.