A Theory of Resilient Smart Manufacturing Systems Optimization Framework

Thesis Abstract

In the face of increasing demand volatility, supply chain disruptions, and rapid technological advances, smart manufacturing systems require robust optimization frameworks that integrate resilience with real-time decision-making and digital twin capabilities. This study addresses the gap in theoretically grounded yet practically implementable models that unify resilience engineering, Industry 4.0 technologies, and stochastic optimization for continuous improvement of production performance. The aim is to develop a Theory of Resilient Smart Manufacturing Systems Optimization Framework (R-SMSOF) that simultaneously enhances operational resilience and efficiency under uncertain conditions. Specific objectives are to (i) formalize a resilience-oriented optimization model that incorporates predictive maintenance, dynamic scheduling, and robust logistics; (ii) integrate a digital twin and cyber-physical systems layer to enable real-time situational awareness and adaptive control; (iii) establish a multi-objective optimization approach balancing throughput, reliability, energy efficiency, and cost; (iv) validate the framework through a real-world case study in a medium-scale electronics manufacturing plant and a simulated benchmark with varying disruption scenarios; and (v) derive governance and implementation guidelines for organizational uptake. The methodology adopts a mixed-methods design underpinned by the resilience engineering and systems theory perspectives. The population consists of manufacturing plants deploying smart factory capabilities across Europe and North America, with a purposive sample of 12 facilities for in-depth study and a supplementary synthetic dataset of 200 disruption scenarios generated through Monte Carlo simulation. Data collection instruments include structured interviews with plant managers (n=24), on-site observations, archival performance data (uptime, cycle time, defect rate, energy consumption), sensor logs from manufacturing execution systems (MES) and digital twins, and archival maintenance records. Instrument validity is established via content validity with a panel of four subject-matter experts and test-retest reliability assessed through a pilot study in two facilities. Analytical methods comprise (i) systemic codebook thematic analysis for interview transcripts, (ii) stochastic multi-objective optimization using a robust optimization (minimax) approach coupled with a two-stage decomposition framework, (iii) regression-based causal analysis to quantify relationships between resilience investments and operational KPIs, and (iv) Bayesian updating for real-time decision support within the digital twin. The model specification integrates a mixed-integer linear programming (MILP) backbone with stochastic constraints representing supply, demand, and disruption probabilities, augmented by a resilience score constructed from exposure, sensitivity, and adaptive capacity dimensions. The analytical framework further employs ANOVA to compare performance across scenarios and a sensitivity analysis to identify critical levers. Expected findings indicate that the R-SMSOF yields superior average throughput and reduced downtime under disruption scenarios when adopting a coupled maintenance-scheduling-delivery policy, enhanced by real-time re-optimization via the digital twin. The framework is anticipated to demonstrate that resilience investments yield diminishing returns beyond a threshold of digital maturity, emphasizing the need for balanced allocation between predictive analytics and human-in-the-loop governance. The study is expected to reveal that multi-objective optimization with a robust objective combination outperforms single-objective approaches in maintaining service levels while controlling energy and cost footprints, particularly under supply shocks and demand volatility. Contributions to knowledge include (i) a formalized, integrative theory linking resilience engineering with optimization in smart manufacturing, (ii) a novel multi-layer optimization framework that harmonizes predictive maintenance, adaptive scheduling, and cyber-physical coordination within a digital twin environment, and (iii) empirical insights into governance structures, investment thresholds, and implementation pathways for industry adoption. The study closes with practical recommendations for managers and policymakers, including guidelines for selecting resilience-leveraging investments, designing digital twin architectures for real-time decision support, and developing organizational routines that sustain resilient performance in the face of uncertain operating conditions. The overarching conclusion posits that resilient smart manufacturing systems can be systematically engineered through an optimization framework that embeds resilience metrics into the decision-making fabric, with digital twins acting as the central nerve system enabling proactive and adaptive enterprise-wide control.