PRA
Probabilistic Risk Assessment (PRA) is a systematic and comprehensive methodology used to evaluate risks. ARES has performed PRAs for complex technological systems at all phases of the life cycle; from concept definition and pre-design through safe removal from operation. As a quantitative risk assessment, consequences are expressed numerically (i.e., the number of people potentially hurt or killed), and their likelihoods of occurrence are expressed as probabilities or frequencies (i.e., the number of occurrences or the probability of occurrence per unit time).
PRA usually answers three basic questions:
- What can go wrong?
- What and how severe are the potential detriments?
- How likely to occur are these undesirable consequences?
Risk assessments performed by ARES Corporation use event trees and fault trees to identify potential event sequences, as well as customized techniques that generate deterministic data or probability distributions to evaluate the impact to a system.
In the conventional approach, event trees are used quantify the frequencies of the postulated event sequences, and fault trees are used to quantify specific failure modes and branch point probabilities in the event trees. Probability density functions (PDFs) are used to define the uncertainties in each of the input parameters that characterize the event frequencies or component failures. These statistical distributions are propagated through the appropriate physical models using a sampling technique, such as Monte Carlo approach.
The result, either as an event sequence frequency or consequence, is typically presented as a statistical distribution.
The resulting distributions are used at several levels to develop insights and understanding about the systems and processes being evaluated. At the simplest level, either “nominal” or mean values of the frequency and consequence are used to determine a nominal or best estimate risk value.
At a higher level, the results can be displayed as either a probability distribution function (PDF) or a consequence complementary cumulative distribution function (CCDF). A CCDF represents the probability of exceeding a specific level of consequence. The total risk is the area under the CCDF, which can be expressed as the frequency or probability of exceeding the given level of consequence.
At the highest level, a double-loop Monte Carlo simulation can be performed in which the random, or stochastic, uncertainties are sampled in one loop and the systematic, or knowledge-limited, uncertainties are sampled in the other loop. The simulation results in a family of CCDFs, each of which depicts the results at a particular confidence level.
Sensitivity studies and importance measures are used to determine the parameters that contribute most significantly to the total risk. These parameters can then be managed to reduce the total risk.
