The successful design, deployment, and management of complex engineering systems requires a hybrid framework that evaluates and analyzes evolving risk dynamics, causal interdependencies, and explicit decision-making under uncertainty. We present a formal integration of Dynamic Probabilistic Risk Assessment (DPRA), or more generally Probabilistic Risk Assessment (PRA), with multistage stochastic optimization as a tool for risk-informed decision making for long-horizon planning of complex systems under uncertainty. A demonstration of this hybrid methodology shows how causal risk and reliability modeling, combined with decision analysis can support design, maintenance, and operational decisions for liquid hydrogen fueling station deployment with the objectives of minimizing risk, and maximizing profit and availability.

Traditional PRA methods offer robust analysis of system dynamics, their failure points, and the consequences of failures, but they do not explicitly prescribe optimal actions for managing risk over time. Multi-horizon stochastic programming, a mathematical optimization framework for Decision Making under Uncertainty (DMU), models optimal actions when future states of the world are uncertain, but representable through probabilistic scenarios. By embedding PRA causal and logic models, such as Dynamic Bayesian Networks (DBN) and event sequence diagrams (ESD) as the system models and scenario generators for multi-horizon stochastic programming, this work constructs a decision-dependent PRA architecture in which risk and reliability evolve in response to design and operational decisions. This yields a fully dynamic, causal representation of system risk, as opposed to the static snapshots in non-integrated PRA and traditional techno-economic analyses.

An example case study presents a risk and reliability based long-horizon analysis of a high capacity liquid storage hydrogen fueling station design and deployment. The application of this methodology shows the trade-offs between cost, availability, and risk according to decision-makers’ tolerances and thresholds. We identify the most prevalent risk and reliability trajectories in a hydrogen fueling station’s operation and show the optimal investment and maintenance strategies for each trajectory. Further, we use risk metrics to minimize tail-losses for unwanted consequences beyond identified risk tolerances. The hybrid modeling approach links engineering risk and reliability, operational policy, and economic viability in a unified risk-informed tool for analysis of long-term liquid hydrogen fueling station deployment.