Each pressure safety valve (PSV) was sized correctly. The decanter PSVs vented to the decanter crude drum and the decanter crude drum PSV vented to the pipe still. But there was no serious back-pressure — what was wrong?
A 210-psig relief vented to a 75-psig drum that vented to a 50-psig tower — that’s what was wrong! This could be a cascade event, a fire circle, if it weren’t for a single pressure control valve (PCV) protecting these vessels. However, a PCV isn’t intrinsically safe; no law of nature assures it will open as is required by ASME code.
How could this risky setup been overlooked? Refineries consist of interconnected interdependent units. API codes and standards don’t help in complex situations, e.g., in circumstances where a PSV operates correctly but its actions initiate unsafe events.
For example, API assumes a fire can be contained within a 50-ft diameter — the 2,500-ft2 rule referenced in API-521, Section 184.108.40.206. Anyone who’s been very near a flare can testify that 50 ft is way too close. When your relief vents to the atmosphere or ground instead of a flare, the vent discharge becomes a potential flare. API rules can adequately protect individual equipment; sometimes, though, it seems they protect the tree while the forest may be at risk.
This is just one situation where API and other guidelines fail. Let’s consider others. In January 1972 at Montreal East, Quebec, natural gas back-flowed into a nitrogen purge system after a PSV popped during the defrosting of a compressor. The gas entered the control room. An operator’s cigarette initiated an explosion. In June 2005 at a Praxair facility in St. Louis, Mo., propylene cylinders relieved in response to heat, resulting in an ignition and explosion (www.csb.gov/newsroom/detail.aspx?nid=167). The PSVs were sized correctly but nobody anticipated what would happen when 800 cylinders vented to the atmosphere simultaneously at the same location.
Recently, I was asked to review PSV calculations for a refinery. I noticed that some PSVs vented in the same semi-enclosed area. So, I did a lower-explosive-limit (LEL) calculation of the area around the vent pipe and determined that concentrations of the vented gas were within its explosive envelope. A spark could set off a detonation. PSVs vented to a congested area at Kleen Energy, Middletown, Conn., resulted in a natural gas explosion on February 7, 2010. Such dangerous design undoubtedly afflicts other sites, too.
Other unintended consequences can occur. In 1971 in La Spezia, Italy, liquefied natural gas of different temperatures mixed in a storage tank. The PSVs vented. The flow of 2,000 tons of vapor collapsed the roof of the tank.
During a debottlenecking study of a propylene system an off-hand comment by an operator — that the plant had had fires from the venting of PSVs because of the summer’s heat — alarmed me. Loading and unloading of tanks and tankcars caused venting. The friction of the gas molecules, a speck of dust, a flow disturbance or a patch of rust can suffice to ignite propylene and some other flammable gases.
How can we avoid unintended consequences? There really isn’t a simple answer to complicated situations, but I’ll give it try. Start by not crowding equipment together or at least keeping relief vents separate unless they go to a flare header. Don’t vent flammable gas or vapor to the atmosphere, ground, an atmospheric tank or low-pressure vessel — sometimes manual vents use the same header. Include pressure relief devices, vents and flares in electrical-area-classification surveys (“Survey the Sources of Electrical Risks"). Keep flammable vapors and ignition sources apart.
In addition, stabilize dangerous or unstable contaminants and prevent them from becoming a hazard. When I worked at Millennium Inorganic we killed phosgene with methane. The trouble is methane forms methyl chlorides, which could easily escape our scrubbers, which were designed for chlorine. Methyl chlorides are fairly lethal on their own.
Another idea to prevent the consequences of a relief scenario is to isolate relief systems or other dangerous equipment with fireproof barriers like earthen bunkers. Along the same line, installing flammable-gas sensors and establishing sniff protocols and inspections to spot leaks also could identify potential relief issues. Lastly, consider locating equipment and, especially, critical instruments far from where flammable liquid pools could accumulate or where a fire could be passed from point-to-point — 50 ft probably is grossly inadequate.
Looking at the big picture is key to avoiding creating secondary risks.
DIRK WILLARD is a Chemical Processing Contributing Editor. You can e-mail him at firstname.lastname@example.org.