Real-time probabilistic risk assessment

M Haile PhD, Vehicle Technology Director, US Army Research Laboratory, Aberdeen Proving Ground, Maryland, USA 


Assessing the reliability and performance of complex systems, such as an aircraft, involves dealing with events whose occurrences cannot be readily predicted. Not only is a good descriptive model of the system's components and connectivity required but the solution also requires some means by which the likelihood of the events can be expressed in terms of quantitative methods. There are not many practical ways of dealing with such a problem. Nonetheless, one possible way is to use probabilistic risk assessment (PRA) methods. PRA is a quantitative method for estimating what can go wrong, how likely this is to happen, and what the consequences are by solving physics models or studying sensor data. In doing so, PRA provides an insight into the strength and vulnerability of the design, operation and maintenance of complex engineering systems. A typical PRA process may involve: 1) identification of the initiating events of each potential failure and/or failure modes; 2) estimation of the probability of occurrence of each failure; 3) estimation of the consequences of each failure on the total system; and in some cases 4) remedying against an unacceptable end-state or consequence. In PRA the modelling of each failure can be accomplished with inductive logic and event trees. An event tree starts with the initiating event and progresses through a series of intermediate pivotal events until an end-state is reached. An accompanying graphical tool called event sequence diagram (ESD) is used to describe all possible failure scenarios.

Advances in sensor and sensing technology, as well as the availability of high-powered compact electronics combined with the ever increasing maintenance and sustainment cost, is causing a paradigm shift towards zero maintenance aircraft (ZMA). ZMA is just a concept. Vaguely put, a ZMA is a vehicle that operates for a much longer period (much much longer than the current!) without any maintenance or inspection. Currently, there is a strong consensus that one of the technologies needed to realise ZMA is the capability to perform real-time probabilistic risk assessment (PRA). The SHM community has a good opportunity to lead in this emerging field and a special session in PRA will be an excellent addition to EWSHM and IWSHM meetings.

Submissions may be solicited in PRA of complex systems including mechanical, aerospace and civil structures, specifically addressing:
•    Machine learning inspired event-tree and fault-tree analysis;
•    Risk-informed maintenance;
•    Maintenance-free operation period;
•    Uncertainty and sensitivity analyses; and
•    Uncertainty quantification and risk analysis.

For further information contact:
Conferences and Events Department, The British Institute of Non-Destructive Testing,
Midsummer House, Riverside Way, Bedford Road, Northampton NN1 5NX, UK.
Tel: +44 (0)1604 438300; Fax: +44 (0)1604 438301; Email: