PROBABILISTIC RISK ASSESSMENT OF PROCESS PLANTS UNDER SEISMIC LOADING

Abstract eng:
The vulnerability of the urbanized territory against Na-Tech events has been recently raised because of the general unpreparedness of the countries in predicting effects and consequences in the aftermath of a disaster. This is clearly proved by the increasing interest of the scientific community, whose contributions, during the last decade, have seen a rapid increasing. The latest issue of the European Directive Seveso (III) represents an additional incentive to deal with this subject, containing an explicit reference to NaTech events. Unfortunately, despite the continuous evolution of the knowledge on this matter there is a large lack of information about possible procedures to predict damage propagation within a process plant and in the surrounding areas, and the quantification of the risk under technological/natural events. In this respect, the effects of earthquakes on chemical plants represent a strategic issue, as demonstrated by recent 2011 Tohoku Earthquake, in which many of the chemical/petrochemical plants were subjected to important damages and losses. It is known that the classical Quantitative Risk Assessment (QRA) methods cannot be applied to evaluate consequence in case of earthquakes, because of the presence of multi-damage conditions, simultaneously involving more than one equipment, which in turn can generate a multiple-chain of events and consequences. In literature, several attempts to modify the classic QRA approach, for accounting for this important aspect, have been formalized but without converging toward a unified approach. Consequently, in this paper a new tool for the probabilistic risk assessment of process plants under seismic loading is proposed, which is based on Monte Carlo simulations. In particular, starting from the seismic hazard curve of the site in which the plant is placed, a multi-level approach is proposed, in which the first level is represented by the components seismically damaged, whereas the following levels are treated through a classical consequence analysis, but including propagation of multiple simultaneous and interacting chains of accidents. This latter is applied defining for all relevant equipment proper correspondences between structural damage (i.e. limit states) and loss of containment events. The procedure has been implemented in the PRIAMUS software, which assumes that the accident dynamics may be represented by a sequence of propagation “levels”, that is the evolution of domino effects. With a series of automatically generated samples of damage propagation scenarios, the risk of the plant can be easily quantified in terms of economic losses, damage scenarios or damage propagation. The application to a petrochemical plant shows the potentiality of the method and envisages possible further evolutions.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket