CSB Releases Final Report on Arkema Fire

Last week the Chemical Safety Board (CSB) released their final report [.PDF download] on the Arkema chemical facility fire that occurred as a result of the flooding during Hurricane Harvey last year. The detailed report is lengthy and detailed and deserves the attention of safety professionals, but the Executive Summary is surprisingly detailed and should be read by everyone in the chemical industry. And, as usual, the CSB has produced another excellent video providing a good review of the incident. I cannot say enough good things about CSB incident videos.

High Points

I am not going to go into much detail here because I do not want to give anyone an excuse for not reading the report, but there are a couple of things that need to be highlighted.

First, the Arkema facility team did do a good job in analyzing the safety issues and preparing for the storm. While the CSB report does raise some specific questions and identifies some things that could have been done better, the pre-planning and on-site reactive measures that were taken demonstrate that Arkema was proactive and properly reactive regarding this incident.

Second, the police officers that were exposed to smoke from the decomposition fires during this incident were well within the mile and a half evacuation zone established around the plant when the plant lost the ability to cool the organic peroxides. This was due to the fact that the road outside the plant fence was kept open during much of the evacuation because it was one of the few remaining accessible roads in the area. The affected officers were patrolling that road to monitor the potential effects of the Arkema incident on users of that route.

Finally, the flooding levels seen in the area of the plant exceeded the ‘500-year flood’ level. While it appears that Arkema was not aware of what the 100- and 500-year flood levels were for the facility, this does demonstrate the magnitude of the disaster that was the proximate cause of the Arkema incident.

Commentary

The CSB makes the point that all of the protective measure put in place to prevent the organic peroxides from reaching their self-accelerating decomposition temperature (SADT) failed from a common cause; the flooding at the site. They then go on to recommend (Recommendation 2017-08-I-TX-R1) that the Arkema Facility:

“Reduce flood risk to as low as reasonably practicable (ALARP). Ensure that any safeguards for flooding meet independent layer of protection requirements.”

The fact that Arkema was not aware that a significant part of the facility was within the “500-year” flood plain is a point well taken, but it is not clear that before this incident anyone would have considered preparing for a 500-year flood to be a reasonable standard for preparedness.

Having said that, this incident and the whole Harvey flood calls into question the efficacy of the use of historical flood data in predicting future flooding. Sections 12 and 13 of the report deal with the issue of increasing risks related to ‘extreme weather’ events while being very careful to avoid any discussion of climate change. While the CSB is an independent agency not directly responsive to the Trump Administration (Could that have anything to do with the attempts to defund the organization?) it appears that the agency was very careful to avoid getting caught up in that controversy.

Unfortunately, anyone with a modicum of intellectual honesty has to admit that there has been an ongoing increase in the severity of rain events in the southeast (and along the Texas and Louisiana coast in particular) in recent history. And, that increase is significantly outside of the historic norms. While there may be some room for debate as to the cause of this recent change, the fact that the change exists cannot be ignored. The hard part, however, is going to try to determine what the 500-year flood plain is in the current reality.

Now, I suspect that for political reasons, the FEMA flood maps for the area flooded during Hurricane Harvey will not reflect the fact that the areas flooded now represent the defacto 100-year flood plain, but planners, specifically including emergency response planners, will have to accept that as the current reality.

One final note here worthy of consideration by chemical facility planners (and I will be taking this up in more detail in a future blog post) is the fact that the CSB recommendations do not limit themselves to extreme flooding events. In the Executive Summary (though carefully missing from the official recommendations in the report) the CSB also recommends that (pg 8):

“Seismic hazard maps should be evaluated to determine the potential risk of earthquake. Risk of other extreme weather events such as lightning strikes and high wind events should also be considered.”

CSB’s Arkema Investigation Update

Earlier this week the Chemical Safety Board (CSB) held a news conference (note this link is to a copy of the email that I received about the press conference, the Sutherland statement is not currently available on the CSB web site) to provide an update on their investigation of the fires at the Arkema site in Crosby, TX after Hurricane Harvey (discussed in this blog here). As part of that news conference, CSB released a video showing the time-line of activities that took place during the incident.

The Time Line

CSB shared the following time-line graphic at that news conference

Commentary

Sutherland concluded her statement by saying: “There is a valuable lesson that facilities in the Gulf and elsewhere should note:  Reassess continuity of operations plans and worst case (sic) scenario assumptions.  Plan and plan again. Don’t be lulled into a false sense of safety by thinking that ‘it can’t/ won’t happen here.’”

A key part of that planning process is the identification of the key assumptions made during that process. Here, for example, the assumption was that flood waters would not exceed 2-ft. I would be surprised if this assumption was made and documented in any formal fashion, but it was made when it was decided to elevate the backup generators by that much.

If the facility had documented the reasoning process that lead to that decision, a periodic review of the plan may have noted that the increased rainfall that storms have been producing in the north-western Gulf Coast in recent years might have called for a revision of that assumption.

All emergency response plans need to be formally reviewed a recurring basis. For example, along the Gulf and Atlantic Coast, chemical facilities should formally review their hurricane response plans every spring, well before the start of the season. That review should include:

• Lessons learned from previous seasons;

• Assumptions about storm action levels;

• Assumptions about worst-case scenarios;

• Shutdown decision points;

• Evacuation decision points;

• Coordination activities with local community responders;

• Facility protection plans; and

• Recovery plans.

Worst-case scenario planning, it must be remembered, should not start with an assumption about what is the worst thing that could happen to the facility. It should start with an analysis of what is the worst thing that can happen at a facility. At the Arkema facility this would have been the decomposition/fire/explosion of the material in one of the cold storage buildings. From that worst-case incident the planning process needs to identify possible routes to that incident and the mitigation measures necessary to prevent those causes.

Was the worst-case planning at the Arkema facility adequate? In hind sight, it is easy to come to the conclusion that it was not; easy, but not necessarily correct. Three feet of flood water had never been documented in that area, so it is hard to fault Arkema (or any of its neighbors) for planning for such flooding. Plans going forward will have to be revised for that eventuality, but it was not reasonable pre-Harvey.

The other thing about emergency planning that we see clearly from this event is that plans need to be modified on the fly as situation change. It is clear from the timeline presented by CSB, that Arkema continued to recognize that the situation was changing for the worst and that they adapted in a timely and proactive manner to those changes. This is one of the reasons that facility evacuation plans in the face of storms like these must consider leaving a team on site to respond to changing conditions.

Any stay behind team needs to include knowledgeable management and operations/maintenance personnel that have the authority and skills to react to changing conditions. They must be protected against the potential storm effects and provided with communications tools to be able to coordinate with local response agencies if required.

As I have noted before in discussing this event, the CSB investigation of the incident should focus on the planning process that was in place for this facility. The result of the investigation should include recommendations for emergency planning actions that chemical facilities should take to prevent damage (with off-site consequences) from predictable natural disasters like hurricanes, floods, tornados, and earthquakes (all in appropriate areas of the country).

Harvey Chemplant Explosion – Part I

It looks like this organic peroxide plant situation will be continuing news. It seems that late last night there were two ‘explosions’ at the facility (see here and here for news reports) and a number of police officers are being treated for chemical exposure issues.

NBC News Tweeted® a copy of the Arkema statement about last night’s incident. It makes a very important point: “We want local residents to be aware that product is stored at multiple locations on site, and a threat of additional explosions remains.”

Health Effects

First, we need to remember that the smoke from any fire contains some number of toxic elements and should be avoided. This is especially true when you see thick black smoke; that indicates incomplete combustion and you are going to have a wide variety of chemicals and physical particles that will, at the very least, irritate the lungs.

I am not an industrial health expert, by any stretch of the imagination, but organic peroxides will almost certainly have some level of toxicity due to their chemical nature. The free radicals produced in the initial decomposition are very reactive and will almost certainly react with body tissues. Fortunately, they also react very quickly with oxygen in the air, so this toxicity is typically greatly decreased the further you get from the un-decomposed organic peroxide.

Police officers are always going to be at risk from smoke inhalation injuries due to the nature of their duties since they do not have ready access to necessary personal protective equipment. The standard issue protective mask (used mainly for riot control situations where tear gas may be employed) may not be effective protection against all components of the smoke of an industrial fire. This is why fire fighters carry the heavy and awkward breathing air tanks on their backs.

For more information on the toxicity of organic peroxides you can visit the Arkema web site and find the Safety Data Sheets for the Luperox® line of organic peroxides. I am not sure which of those are manufactured at this particular facility (the local fire department has that list), but you can get an idea of the types of solvents used and the general toxicity information.

SADT

Those SDS also contain another interesting bit of information, the temperature at which a self-accelerating decomposition reaction (SADR) begins {referred to as the self-accelerating decomposition temperature (SADT)}. This is the decomposition reaction that I described in last night’s blog post. This is the critical temperature that I talked about. Looking at a random selection of the Luperox products this morning, it would seem that most have a SADT in excess of 100° F.

Unfortunately, even below the SADT some level of decomposition remains, and the exothermic nature of that decomposition reaction will raise the temperature of the mixture. The SADT is the point of no return. When it reaches the SADT there is essentially nothing that can be done to prevent catastrophic decomposition rates.

The higher the SADT, the longer it is going to take for those containers to reach their failure point, prolonging the current problem. Of course, a fire on the site will change all of that as it would quickly raise the temperature well above the SADT point while weakening the structure integrity of the storage containers.

Storage Issues

One last item that needs to be taken into consideration. Organic peroxides are normally shipped in five-gallon plastic containers. I would expect that this facility would store those on pallets with the containers stacked two or three high. The containers at the center of the stack are going to generally be the first to fail as they are insulated from the cooling effects of the air surrounding the stack.

It would not be unusual to expect that, depending on how the pallets are stacked in relation to each other, that we could see several small ‘explosions’ of individual containers before the bulk of a certain product releases. This could also cause a relatively small fire in the storage area that could expedite other products reaching their SADT.

It will be interesting to see how much detail is included in the monitoring of these storage areas. We could get some very important data on failure rates and effects that could be beneficial in preventing future incidents at these types of facilities.