(46ag) A Quick Methodology for Risk Analysis Based on Back-Calculations from QRA Techniques and Mechanical Integrity Threats

Quantitative risk analysis (QRA) and consequence modeling (CM) have come a long way since they evolved from their semi-quantitative siblings and were introduced to the chemical process industries back on the early 1980’s. However, advanced CM and modern QRAs are still complex, expensive and time consuming tasks that require large amounts of high quality data, making them unsuitable for key process safety related situations where a quick risk assessment is required (e.g. risk based decision processes on daily operations or management of change). With this in mind, we decided to flip around the risk analysis evolution curve to develop a quick, consistent and accurate semi-quantitative methodology based on back-calculations from CM and QRA techniques. To estimate the consequence of failure (CoF), generic event trees were developed to address the most significant events due to loss of containment of liquids and gases.  For each event, a key parameter was identified (i.e. the one that has the strongest influence on the consequence) and a ranking system was developed to estimate a base CoF index using this parameter. The ranking criteria were established using probabilistic effect models and back-calculations from consequence and source models. Then, the base case CoF index is corrected to account for the other parameters and passive barriers in place; the correction criteria were developed using an analogous procedure of back-calculations. To estimate the probability of failure (PoF), a similar ranking-correction scheme was developed for fifteen threats to the mechanical integrity of the equipment (e.g. external corrosion, internal corrosion, stress corrosion cracking, fatigue and erosion, among others). For each threat, an initial PoF index is assigned based on the most relevant parameter and is then corrected to account for other parameters and protection systems in place. Finally, the risk index is estimated for each piece of equipment using the PoF, CoF and the generic event trees. This paper shares the development process of the methodology as well as its application to an upstream facility, highlighting the lessons learned from this case study.