Welcome to the PSAM 18 Abstract Status page.
Abstract LI105Full Paper + Presentation
A Dynamic HRA Framework for Persistently Responding to Environmental Changes: A Preliminary Test of Competing Outcomes Sampling Using the Proposed Concurrent Continuous Markov Chain Monte Carlo (Co-CMMC) Method
Human Reliability Analysis (HRA) is a pivotal element in Probabilistic Risk Assessment (PRA) as it estimates human error probability (HEP) in human–machine activities such as repair. Conventional HRA methods link performance shaping factors (PSFs) to static, pre-defined scenarios, making it difficult to directly represent work contexts that evolve continuously.
In contrast, the field of dynamic PRA provides a means to capture dynamic characteristics by stochastically generating event sequences that track time-varying machine behavior. Within this field, the Continuous Markov chain Monte Carlo (CMMC) method recalibrates state-transition probabilities at each time step using environment-responsive failure rates and advances the machine state through Monte Carlo sampling, yielding time-ordered machine state histories under persistent environmental variability. To account for maintenance, earlier CMMC extensions introduced repair activities by assuming a pairwise exchange between machine states and repair-task states (e.g., a failed state mapped to repair in progress, and an operating state mapped to repair completed), allowing repair histories to be inferred while simulating only the machine process. From an HRA viewpoint, however, repair termination is a competing-outcome process (success versus failure). This exchange assumption therefore obscures outcome-specific human performance and limits HEP estimation in dynamic environments.
To enable Dynamic HRA (DHRA), this study proposes a Concurrent CMMC (Co-CMMC) method that defines a joint Markov state combining machine states and repair-task states, allowing both processes to evolve concurrently and stochastically without enforcing pairwise exchange. In addition, the concept of a competing-risks model is employed to update outcome-specific repair rates under changing environmental conditions. This preserves CMMC’s stepwise environmental updating while enabling simulation of repair completion times and outcomes. Lastly, a computational framework based on the Co-CMMC method is outlined and preliminarily examined, indicating its capability to extract dynamic repair-task histories under human–machine–environment interactions and highlighting the potential to support DHRA-oriented HEP estimation.
← Check another abstractIn contrast, the field of dynamic PRA provides a means to capture dynamic characteristics by stochastically generating event sequences that track time-varying machine behavior. Within this field, the Continuous Markov chain Monte Carlo (CMMC) method recalibrates state-transition probabilities at each time step using environment-responsive failure rates and advances the machine state through Monte Carlo sampling, yielding time-ordered machine state histories under persistent environmental variability. To account for maintenance, earlier CMMC extensions introduced repair activities by assuming a pairwise exchange between machine states and repair-task states (e.g., a failed state mapped to repair in progress, and an operating state mapped to repair completed), allowing repair histories to be inferred while simulating only the machine process. From an HRA viewpoint, however, repair termination is a competing-outcome process (success versus failure). This exchange assumption therefore obscures outcome-specific human performance and limits HEP estimation in dynamic environments.
To enable Dynamic HRA (DHRA), this study proposes a Concurrent CMMC (Co-CMMC) method that defines a joint Markov state combining machine states and repair-task states, allowing both processes to evolve concurrently and stochastically without enforcing pairwise exchange. In addition, the concept of a competing-risks model is employed to update outcome-specific repair rates under changing environmental conditions. This preserves CMMC’s stepwise environmental updating while enabling simulation of repair completion times and outcomes. Lastly, a computational framework based on the Co-CMMC method is outlined and preliminarily examined, indicating its capability to extract dynamic repair-task histories under human–machine–environment interactions and highlighting the potential to support DHRA-oriented HEP estimation.