Flare Overload Study for Simultaneous Relieving Scenario of Pressure Relief Valves Using the Monte Carlo Model

The optimization of flare design is a significant challenge in terms of safety, environment and cost. Pressure Relief Valves (PRVs) shall be designed for each relieving case and the flare header is designed for multiple relief scenarios. However, the simultaneous relieving scenarios such as total power failure can incur extraordinary flaring rates. Although, Safety Instrument System (SIS) is used to avoid simultaneous relieving such as by isolating the heat source in reaction to a HH Pressure detection, the SIS does not necessarily operate when needed. Further, when plants are expanded, the new units are often connected to the existing flare, and chance of overload exceeding flare capacity becomes higher due to additional relief sources from the new units. Therefore, it is necessary to confirm that the overload frequency is sufficiently low by performing a “flare overload study”. In this study, the overload frequency is calculated by considering the Probability of Failure on Demand (PFD) for each Safety Instrument Function (SIF).

In the conventional calculation method, the flare overload frequency is calculated against all combination of each SIS success/ failure by using the spreadsheet. However, since large-scale plants are required to use dozens of SISs per plant, calculation is complicated. (If thirty (30) SISs are used in the plant, over one billion (2³⁰) relieving scenarios need to be calculated). Since such an enormous number of relieving scenarios cannot be processed in a conventional method, the calculation effort can be reduced by aggregating relieving scenarios. Nevertheless, it takes much time to aggregate the scenarios, and it is difficult to modify the tool when there is a design change such as the number of relieving scenario and/or the design relieving rate for each PRV. Therefore, a simples / more time efficient calculation method is desirable.

This paper discusses the use of the Monte Carlo method to calculate flare overload frequency in simultaneous relieving scenarios. Since this method applies computational algorithms relying upon repeated random sampling to obtain a probability, the overload frequency can be calculated without processing such an enormous number of combination that increase exponentially with the increase of flare relieving scenario as a conventional method. Hence, the applicability of this Monte Carlo method is evaluated by comparing with the conventional method in terms of modelling, programming, calculation time and accuracy based on the onshore LNG plant data.

Keywords: Monte Carlo Method, Flare overload, Simultaneous Relieving Scenario, Pressure Relief Valve, Safety Instrument System, Probability of Failure on Demand