Thursday, December 10, 2009

CSB Issues Urgent Recommendations

Yesterday I addressed the CSB vs SSI issue reported in yesterday’s press conference and press release about the Chemical Safety Board’s investigation of the CITGO explosion and fire on July 19th at their Corpus Christi refinery. There was another, equally important issue raised in that press conference (more immediately important to people living and working around the CITGO refineries mentioned), the apparent inadequacy of the hazard mitigation procedures to protect people from catastrophic releases of hydrogen fluoride gas from storage. According to the press release:
“The refinery used a water spray system to absorb the released HF, but the CSB cited scientific literature to conclude that at least 4,000 pounds of HF likely escaped from the unit into the atmosphere and left the facility. Investigators determined that during the first day of response efforts CITGO nearly exhausted the stored water supply for the water mitigation system. Approximately eleven-and-a-half hours after the initial release, before the water supply was completely exhausted, the refinery began pumping salt water from the ship channel into the refinery fire water supply. Multiple failures occurred during the salt water transfer including ruptures of the barge-to-shore transfer hoses and water pump engine failures.”
While the investigation has yet to be completed, the CSB took the unusual step of issuing and “urgent safety recommendations calling on CITGO to immediately improve its emergency water mitigation system in the event of another release of potentially deadly hydrogen fluoride (HF) vapor”. They also recommended an independent safety audit be conducted of the hydrogen fluoride units at all of their refineries. The Board only issues such early recommendations when they “determine an imminent hazard may be present and has the potential to cause serious harm unless rectified in a short timeframe”. Water Spray Mitigation Systems A number of high risk chemical facilities across the country with toxic release COI use similar type water spray systems for release mitigation. Such water sprays convert a toxic gas cloud to a toxic gas dissolved in water that presents a much smaller risk to the neighboring or downwind community. The effectiveness of such systems depend on a number of complex issues including (but certainly not limited to), rate of gas release, amount of gas released, wind speed, wind direction, water flow rates, water spray dispersion patterns and the amount of available water. Since many of these factors cannot be determined in advance the design process for these systems requires the engineering staff to make a number of assumptions about the release incident that might occur. Safety system designers will make these assumptions based on a wide variety of factors including probable failure modes based on past histories. If a new failure mode is ‘discovered’ during a process incident it is likely (as in this case) that the design basis will be inadequate for that new scenario. What CSB is essentially saying in this instance is that the CITGO process engineers did not take this type release scenario into account when this water mitigation system was designed. With the actual data now available for this release case, those engineers need to go back and redesign the system. Other facilities with similar systems will need to model their systems against the CITGO release data to see if their design would be adequate to mitigate a similar type release at their facility. Counter Terror Water Spray Mitigation I have long advocated that this type of water spray system should be included in the Site Security Plan for facilities with toxic release COI. The idea being that in the event of a terrorist attack, this type system would be in place to backstop a failure in security measures that could allow for a successful terrorist attack. Again, it would be just another layer in the site defenses. The apparent failure of the safety mitigation system at Corpus Christi emphasizes how difficult it is to adequately design such a system; even when there is a process history to provide input into the design model. Since not terrorist attacks have been conducted against such facilities, there is no current history to draw from for modeling a system for mitigating a terrorist attack. This means that even more assumptions must be made about the variables that affect the efficacy of such a system. Perhaps it is time for the Federal Government and industry to fund some studies of potential storage tank failure modes from terrorist attacks so that spray mitigation designers have more data upon which to base their designs.


Anonymous said...

Mr. Coyle,

Water sprays and other consequence-reduction mitigation measures are sensible for many cases, but they wouldn't be a silver bullet for toxic release attacks: Attackers might sabotage a facility's spray system before releasing toxic chemicals stored there.


Fred Millar said...

The question is the adequacy of such systems and the reliability of these. Even an accident in an HF refinery reportedly often damages and renders inert the water systems. For HF, there has actually been some (sobering) field tests of what is required:

"A second major way to assess toxic cloud disaster risks from HF facilities is to use the data from the oil industry's own $2 million HF release field test [the "Goldfish" test series] in 1986, which revealed surprisingly long distances for HF clouds (extrapolated to reach downwind at a lethal level of 50 ppm as far distant as 5 miles, or 8 km).

In early 1987 Friends of the Earth learned from industry experts that after the Bhopal disaster worried Amoco officials spearheaded these tests which had stunning results.
As one industry group summarized later: "In 1986, tests were conducted in the Nevada desert to determine the dangers of a possible HF liquid release. Under conditions similar to those that exist in an alkylation unit, lethal concentrations of an HF aerosol were present up to 8 km (5 miles) from the release points. It was during these tests that HF releases were observed to be much more dangerous than anticipated." [6]

The results of the huge potential downwind hazard were not disseminated, by anyone involved, to workers, fire service, local officials, much less media in the roughly 50 HF refinery at-risk communities. "We have no budget for that", said the federal experts involved. The test results were in the public domain, but obscurely: only a few studies were done for industry expert meetings and journals. US EPA in 1993 also soft-pedaled the Goldfish test results, so environmentalists had to disseminate the information to regulators and the media. [7]

In the "Hawk" HF release field tests that followed the "Goldfish" HF tests, a US petrochemical industry consortium tried to show that engineered water sprays could be effective in mitigating HF releases. But other industry experts have pointed out uncertain reliability of this mitigation method:

"Due to the risk, many refiners are implementing water mitigation and
detection devices in an effort to remove any HF that would vaporize in the
event of a release. With water/HF ratios of 40:1, nearly 90% of the HF can
be removed. However, these systems are expensive and there is the concern
that the water sprays could become inoperative as a result of an accident. In
addition, details have not yet been obtained, or at least reported, on the fate
of the HF that is not removed by the water sprays. For a major leak (200 lb/s
100 kg/s) that might result from a 4 inch (10 cm) hole at process conditions,
water systems are thought to be less effective. Major HF leaks have been
rare in the industry, and when they have occurred, there has usually been a
major fire event that has dissipated the HF cloud as it formed. However, the
impact of a major HF release should always be considered." [8]
[such as the 1987 Marathon release at Texas City]. See the US EPA's 1993 Report on HF Risks.

PJCoyle said...

My reply to the comment by Anonymous can be found at:

PJCoyle said...

My reply to the comment by Fred Millar can be found at:

/* Use this with templates/template-twocol.html */