Incorporating human and management factors in probabilistic risk analysis.

摘要

Complex, engineered systems, such as nuclear power plants, chemical plants, aerospace and marine transport, have the potential for catastrophic failures with disastrous consequences. In recent years, human and management factors have been recognized as a primary cause of major failures in such systems. However, current probabilistic risk analysis (PRA) techniques are unable to handle these effects adequately. This dissertation addresses this problem by extending the PRA methodology with a framework that incorporates human and management effects in a quantitative risk model. The framework provides a structure for incorporating first the actions of individuals that affect the physical system, and then the organizational and management factors that influence those actions. It develops several quantitative models of action that apply to different types of situations, and uses these to make probabilistic predictions of actor behavior in the system. These predictions are made from the perspective of management, and depend on management factors such as incentives, training, policies and procedures, and selection criteria. In this way the framework provides the capability to evaluate how changes in management factors affect the actions of individuals, and thus how they affect system risk. The probabilistic nature of the behavior predictions reflects the limits of information available to management and the inherent uncertainty associated with human behavior. The product of this research is a methodology that can characterize the ways in which management and organizational factors affect system failure risk. This is implemented in a quantitative framework that can evaluate risk management strategies that address management problems. This framework can be used as a tool to "engineer the organization" to increase the safety and reliability of complex technical systems. To guide the development of this methodology, a preliminary application looked at the risk of general anesthesia for surgery patients. The analysis included the anesthesiologist's actions and management effects, and evaluated the risk reduction benefits of several management changes. Lessons learned from that project were incorporated in a general risk analysis methodology applicable in any domain; the resulting framework is demonstrated with an illustrative example dealing with hazardous materials transportation risk.