At every facility, there is a point during the process hazard analysis when the team inevitably dives into “the black hole” of flare/relief/scrubber systems. Following strict HAZOP analysis methodology, a team documents all deviations from normal operating limits. However, with most relief systems, this leaves the team facing numerous causes, and with each source, having multiple different relief properties. Covering every contingency and combination is infeasible.
A few things to consider when risk assessing flare/relief/scrubber systems:
- Relief Valves – Ideally, each relief valve should be “opened”. This is most effectively covered by utilizing a “PSVXXX relieves as designed” cause within each user node. The intent is to look at the relief location and identify any specific issues with that valve opening. There should be judgement checks at these locations. Some example questions to ask:
- Is the outlet piping the appropriate spec for the relief pressure?
- Is the relief temperature too cold for carbon steel piping?
- Does the PSV vent to the right location? (e.g., toxics to flare instead of a scrubber)
- Consequence Development – Relief valves are safeguards. The likelihood that they function as required is factored into any scenario in which they are applicable. Therefore, any failure that could occur within the effluent handling system, that would defeat the relief valve, should inherently already be accounted for in the failure probability. While reviewing the effluent handling node, the PHA team should perform sanity checks to confirm any cause within the flare system that results in the defeat of one or more upstream pressure relief devices has a likelihood substantially less than the Independent Protection Layer (IPL) credit taken for a pressure relief valve. If this is not the case, then the credit taken for the upstream relief devices in the individual user nodes needs to be revisited.The entire flare system must be designed so that upstream relief devices meet their specified probability of failure on demand (PFD). If there is concern that failures in the effluent handling system are frequent enough to impact and increase the probability of failure of demand on the upstream pressure relief device beyond what is being credited in the PHA, an fault tree analysis (FTA) or Failure Modes and Effects Analysis (FMEA) could be performed to validate the availability of the combined pressure relief valve/flare system as a valid IPL. Failures that could occur in an effluent handling system and potentially result in defeat of upstream relief devices include:
- Failure to drain KO drum when required or KO drum level instrumentation failure;
- Manual valve on the header inadvertently closed;
- Freezing resulting in Flare Seal Pot blockage (failure to add glycol to seal pots);
- Freezing conditions resulting in water (or other high freezing point liquid) freezing within the flare header on contacting uninsulated cold piping restricting flow;
- Hydrate formation in headers;
- Polymerization of susceptible fluids in the header;
- Flame arrestor pluggage; and
- Staggered flares with rupture disks around control valves.
- Reverse Flow – Thermal contraction should be considered. Hot days with cold rain can result in contraction within the flare system. There should be adequate controls/safeguards to ensure a positive flow rate of sweep gas out of the flare to prevent burnback. Also, consider reverse flow of relieving fluids into any systems that float on the flare.
- Flare Relieving as Designed – Where does the flare vent? Consider radiation at grade, noise, toxics against where an operator is expected to go during emergencies or where there may be an impact to the public e.g., fence lines. Is operations aware of pockets in the piping resulting in liquid traps, or unswept pipes where H2S or CO2 can cause accelerated corrosion?
A PHA is not intended to analyze risk associated with design errors or omissions. Reviewing the flare design against the requirements of API 521, is akin elsewhere in a PHA to asking if there are deviations from B31.3 or ASME VIII. If deviations against RAGAGEP are identified through the PHA process, it should be noted and recommended for correction, but detailed analysis against all RAGAGEP is not a practical expectation. Instead, it is important that the flare design be available as a process safety information (PSI) reference as well as having API 521 available when specific questions do arise.
