Early-stage design decisions in complex engineering systems play a critical role in defining the system’s lifecycle performance, cost, and viability. As engineering systems are becoming increasingly interconnected, cyber-physical and multi-disciplinary, traditional decision-making approaches based on historical data and expert judgments often fall short. To navigate the early-stage design decision challenges and support better decision-making, model-driven decision support systems have emerged as promising tools that allow for integration of simulations, optimization, and data-driven models. Yet, in the context of complex engineering systems, the effective implementation of decision support systems remains limited due to socio-technical challenges. 

This thesis systematically investigates and identifies challenges through a literature review and empirical case studies in an industrial setting. The research identified interrelated barriers that hinder effective decision support systems integration and utilization. First, difficulty in integrating heterogeneous simulation models across varying levels of granularity remains technically and methodologically challenging. Second, the uneven maturity level of decision support systems across engineering domains limits their consistent use in collaborative environments. Third, a lack of accessible and intuitive interfaces negatively impacts the usability for non-expert stakeholders. Fourth, the absence of a real-time feedback mechanism limits their function as a boundary object. Fifth, the lack of metrics to gauge and communicate model maturity and reliability creates risks for misinformed decisions. Lastly, the lack of lifecycle management, including evolution, traceability, and reliability of the underlying models, is rarely supported, which undermines long-term sustainability and trust in the decision support systems.