Quantitative Risk Analysis - A Realistic Approach to Relief Header and Flare System Design
Abstract
As process industry facilities increase production capacity and add processing units, existing relief headers and flare systems are frequently found to no longer meet the same conservative design criteria used for the original design of the facility. This presentation will show how Quantitative Risk Analysis (QRA) may be used to develop a more detailed understanding of the safety issues associated with the design of such systems. Examples will be presented in which QRA has resulted in large cost savings by revealing that proposed multi-million dollar relief header and flare system modifications would have resulted in insignificant reductions in a facility's risk to personnel safety. Most existing pressure relief headers and flare systems were originally designed with little or no consideration of the probabilities or consequences of specific design scenarios, and without taking credit for the operation of the multitude of safeguards present in a typical operating facility that would have a mitigating effect on these probabilities and consequences. The primary reason for this conservative approach was the lack of sophisticated tools and reliability data required to analyze the relationships among initiating events, event probabilities, safeguard reliabilities, and consequences. However, recent advances in modeling techniques, in the collection of reliability data, and in the availability of computing power have now rendered these complex relationships tractable and have made QRA a practical tool for relief header and flare system analysis. The traditional methods for performing relief header and flare system design made no effort to distinguish among initiating events based on their anticipated frequency. Relatively common events, such as a compressor failure (expected to occur perhaps once per year) were treated in the same manner as such rare events as total electric power failure (expected to occur perhaps once in the facility's lifetime). On the other hand, with the QRA approach presented here, a higher level of a consequence (for example, pressure relief valve back pressure) may be deemed acceptable for the power failure event, due to its low frequency of occurrence. By differentiating events based on anticipated probability, the relief header system may not be required to be designed to the same criteria for each initiating event. In addition, this approach allows a more accurate assessment of the aggregate risk for the relief header system as the frequency/consequence relationship for the sum of all the initiating events can be defined. QRA also allows the inclusion of the impact of safeguards - such as shutdowns, conventional instrumentation, and operator intervention - on the likely consequences of an initiating event. Although the assumption that no safeguards are present is reasonable when evaluating the pressure relief for an individual piece of equipment, it becomes more unnecessarily conservative as the number of safeguards involved in the event increases. For example, the probability of failure on demand (PFOD) of a single shutdown valve on a reboiler's steam supply line may be 10%. Certainly, this is frequent enough to require the installation of a relief device to provide overpressure protection to the individual reboiler and its associated distillation equipment. However, when a flare system is designed to handle the relief device effluent from ten such distillation systems, the simultaneous PFOD for all ten steam shutdown valves is 0.110 or once every ten trillion demands. The author- as well as the management of many operating companies - would assert that this is not frequent enough to justify significant expenditure on relief header or flare system upgrades. In this relatively simple example of a single safeguard on each often distillation systems, there are 210, or 1,024 possible outcomes. This presentation will show how the QRA process can be used to define more clearly the probability/severity relationship of the many possible outcomes of significantly more complex process facilities.
Description
PresentationSubject
Flare System DesignCollections
Citation
Perez, John T. (2001). Quantitative Risk Analysis - A Realistic Approach to Relief Header and Flare System Design. Mary Kay O'Connor Process Safety Center; Texas &M University. Libraries. Available electronically from https : / /hdl .handle .net /1969 .1 /193799.