| The Investigation Process Research Resource Site |
A Pro Bono site with hundreds of resources for Investigation Investigators
|Home||Site Guide||FAQs||Tutorials||Old News||Contribute||Forums|
to advance the
Search site for::
Launched Aug 26 1996.
The widespread distribution of the escaped materials and car pieces observed from the air also left an abiding impression on this investigator. It was clear that responders were powerless to change the outcome during the post derailment course of events in that accident. This accident raised questions about what could be done in emergencies involving these materials and quantities when they escaped from their "containers." During the investigation I learned that firefighters were burned in explosions while pursuing their tactical objective - cool the tank shell above the liquid level to prevent a BLEVE.
It was a scenario that I was to see repeated often in subsequent accidents.
In the Crescent City investigation, I did some quick approximations.
As I recall, these
tank car shipments contained around 2 billion BTUs of energy. When that
energy was released, there was no way firefighters could put anywhere near
the amount of water required to absorb that much heat energy on the ensuing
fire. And because of the unpredictability of the tank ruptures at the time,
it was impossible to tell when they would be faced with this massive heat
output. This raised obvious questions about the tactical objective. It
was unclear to me what responders to emergencies involving these materials
and quantities could do when the contents escaped from their "containers"
and what their response objectives should be. It was clear why the prevailing
"attack and extinguish" approach could lead to disaster.
Subsequent to the Study's release, I participated in 3 highway hazmat accident investigations including the Waco Georgia accident. Two firemen (attacking the burning truck wreckage), a wrecker truck driver and two civilians were killed by the exploding dynamite cargo in a semi-trailer van that had been struck by a car with subsequent fire. It demonstrated again problems with the warning system for providing responders data they needed to make their risk taking decisions. In this case both fire fighting and rescue concerns confronted responders arriving at the scene. The Board recommended, in September 1972, that NFPA develop new guidelines dealing with explosives in emergencies. and bring them to the attention of emergency services personnel at all levels. It called for special emphasis on assessing the situation on arrival at the scene, determining alternative courses of action, evaluating the risks associated with each alternative, and selecting the alternative which presents the minimum risks to people, facilities, and fire fighting crews and their equipment. With the responder casualties in this case, it was becoming clearer to me that there were common challenges for hazmat accident responders across modes.
Franklin County, Missouri
In December 1970, LPG released after a pipeline rupture in Franklin County Missouri ignited and produced an open air detonation. The explosion, equivalent to 100,000 pounds of TNT and fire resulted in extensive property damage within a 2 mile radius. This was the first reported case of this phenomenon. I raised questions in my mind about the regulatory classification scheme within which a hazardous material classed as a flammable could detonate like an explosive. It was not to be the last time a material not classed as an explosive material detonated. The implications for emergency responders, who depended on knowing the class of material they were dealing with, seem profound to me at the time. But a sample of one is not evidence enough to challenge the system.
Another accident I investigated in April 1971 was the gassing of factory workers when a truckload of chemical was connected to the wrong receiving pipe at a tannery, and a fatal reaction occurred in the mixed chemicals. I had seen improper connections before, also with potentially lethal consequences, but the accidents were over before the responders could get involved. But it stuck in my mind - by then I was starting to think about describing the accident process observed in general terms so the lessons learned could be disseminated widely across industries.
In this August 1971 accident, a highway collision between a passenger car and truck resulted in hazmat containers in the truck shifting and being breached during the accident. Four people in the immediate vicinity of the accident were fatally gassed as the escaping methyl bromide/chloropicrin mixture contaminated the air around the wreck. As I consulted with the investigator about the hazmat behaviors, the dispersion of the gas in this and the Berwick cases, in the absence of fire, suggested to me that the range of harm from hazardous materials cargoes needed to be considered for all kinds of hazardous materials, not just for flammable materials and explosion threats. My inventory of dispersion issues was building rapidly.
The Houston, Texas accident
A railroad accident in Houston on October 19, 1971 and subsequent public hearing provided another pivotal experience. A derailment and fire, followed by the explosion during firefighting operations of tank care containing vinyl chloride, killed one firefighter and injured 50 other individuals, mostly firefighters. Significantly, injury to a firefighter on a ladder pouring water on the fire, who was overwhelmed by the explosion, was captured on film and was given prominence in the media and throughout the emergency response and transportation communities. During the investigation process we learned that the emergency response personnel were doing what they were programmed by their training. No reasonable person could complain that the firefighters were stupid to do what they did. The casualties demonstrated -- again and very convincingly -- that their programming led to the casualties. The firefighter fatally injured by the hazmat explosion was the training officer. As I heard the on-scene officers describe what they did and why they did it they provided me with many new insights into specific needs for improving emergency responses programming if losses were to be reduced in the future. The response training did not provide decision making process guidance appropriate to the events during this post-derailment emergency.
Clearly, something was amiss! The casualties to responders following existing guidance convinced me that it was the guidance that was untrustworthy, with potentially lethal consequences for those who acted in good faith on that advice. The report was issued in December 1972. The Board found that "The severity of the accident was increased by the abrupt rupture of the tank car and the lack of adequate training, information and documented procedures for identifying and assessing the threats to public safety." Threat assessment was introduced into the lexicon of emergency responders. I think it this accident and the information about the challenges faced by responders that brought to the forefront of my thinking the need to develop threat assessment concepts and procedures for emergency responders the if-then cookbook approach being pursued at that time by the DOT and CHEMTREC, focused on the properties of the materials, was not satisfying the need. This is discussed in greater detail in a 1990 paper.
Delta Air Lines
A December 31, 1971 a pint of Molybdenum 99, a radioactive material with a 66.5 hour half life, escaped from its container in the cargo hold of a Delta Convair 880 enroute to Houston, TX. The aircraft carried 917 passengers before the spill was discovered and the aircraft was taken out of service. The incident was first detected two days later by a consignee who noted his shipment was contaminated while unpacking it. This incident raised a whole new set of transportation issues and eventually resulted in legislation governing radioactive materials shipments by air.
East St. Louis Illinois
Propylene involved in this accident devastated a railroad yard and adjacent properties - including a school building in January 1972. . Walking among the rubble 8 hours later impressed me with the energy that had to be released to do all the work (damages) I was seeing. The wreckage and damage pattern suggested that the propylene may have produced a low order explosion when it ignited, based on the nature of damages to equipment near the center of the explosion. After this accident, the Board recommended that cars carrying these commodities be treated as explosives cars in switching operation, and not be "humped" even though the cargoes were classified as flammable gases. The implications for response guide writers were interesting: treat these materials as explosives or gases?
In March 1972, a tank semitrailer truck overturned on a downhill curve on a mountainous road, and slid into a rock outcrop next to the highway, puncturing the tank. LPG released from the punctured tank killed three residents in a home in a valley below the wreck when the spreading vapor cloud ignited. As I viewed the topography at the scene, it became clear to me that these bystanders never had an informed chance to recognize or react to the threat from the LPG dispersion they faced. The experience raised more questions about warnings within the range of effects of some hazmat cargoes.
New Jersey Turnpike
On September 21, 1972, about 25 minutes after a highway crash and fire, a propylene tank semi-trailer tank exploded, with one piece rocketing 1300 feet in one direction and another piece 500 feet in the opposite direction. 28 persons were injured in the explosion while responders were on scene. The Board criticized the lack of understanding needed to predict the range of the hazards facing the responders, attributing the relatively few casualties in part to luck.
Put up or shut up
One of the fire service experts with whom I explored my evolving insights and criticisms of the current emergency response guides for hazmats during these investigations was Frank Brannigan. An experienced firefighter and introspective thinker, he provided me valuable data about the fire service philosophies, perceptions and practices. He also was a valued source of information about radioactive materials and their behavior over the years.
Brannigan took vigorous exception to my radical thoughts about the inadequacy of the training approaches and materials, and especially my notion of challenging the conventional wisdom dominating the fire service's attack and extinguish mindset. In one moment of exasperation during one of our discussion he challenged me to prove
myself. His challenge:
was a fair challenge.
Accident model developments
Concurrently, when I started working for the Safety Board, I tried to learn how the "old hands" at the Board performed investigations. As I worked with different investigators
in different modes, I was surprised -- and confounded -- by the different ways investigators viewed an accident, how they scoped their investigations, and how they actually did their investigations.
Part of the Board's mission was to make recommendations to prevent repetition of the accidents they investigated. This meant that there should be some strategy for identifying how to intervene in the accidents to change future outcomes. When I tried to add intervention points to the model that was evolving, I discovered that intervention ideas for the model were related to the intervention challenges during the emergency response decision processes. In order to know what intervention actions to take, one had to know the stage of the hazmat accident process at which intervention would begin. Aha! It also dawned on me that for emergency responders to intervene successfully, they had to know
The D.E.C.I.D.E. and GEBMO models
From mid 1972 to mid 1973, I got an investigation breather, as accidents tapered off. I
started to pull all the ideas about the accident process, the intervention
decision opportunities, the intervention decision making process and the
outcome prediction needs together. In a March 1973 paper
I pointed out the emergency response problem, noting that risks to responders
were 10 times as great to responders as to any others, and called for better
understanding of the response process. As
the accidents and classes continued, the process -- and problems -- became clearer to me.
In the hazmat emergency response decision process, predictions are needed before decisions can be made about intervening in the unfolding accident process. From observations of how losses occurred during the accidents we investigated, the data needed for those predictions became visible. In hazmat emergencies we saw that these predictions depended on data about the hazmat and container present, container stressors, the potential behavior of the stressed container and its contents, the potential behavior of the contents if released, where they will go if released, exposures they will impinge if they get where they are likely to go, what harm they will do to the impinged exposures.-- both people and objects. The, decisionmakers had to identify what options they had to change the outcome during
The rough and tumble of the Montgomery College courses in 1973 - 1975 helped me refine and expand the ideas derived from my observations of what happened in hazmat accidents where responders suffered casualties. I was in the unique position of testing my hypothetical models against observed events during subsequent accidents with investigators, with John Zercher at CHEMTREC -- who was dealing one-on-one with responders during accidents -- and with the very cooperative students in the course. The students particularly provided additional insights, such as the idea suggested student Don Morrison of forming a "mental movie" to help the on-scene commander visualize what was expected to happen at the site, from any point in time forward. This became an essential concept for applying the predictive threat assessment process model. I t helped show on-scene commanders how they needed to develop their "scripts" for their mental movies. In other words, it was needed to shift responders away from the "chem-card" if-then cookbook advice mode toward a "think your way through this new situation" mode.
As I focused on the hazmat emergency response decision process, my observations indicated that the management process at a hazmat emergency was not that much different from management of other risky processes in which I had participated or had managed. I tried to capture a generalized description of that process, in a flow chart format, for several reasons.
The 1980 GEBMO version was expanded to incorporate additional behaviors:
Accidents like Wenatchee and prompted the introduction of the "hazmat reaction products disperse" branches in the more recent GEBMO model.) I suspect more changes will be forthcoming if someone is willing to step forward to assume responsibility for keeping the model updated after I retire.
a flow chart could provide a working tool so everyone was talking the same
language during individual emergencies. I don't know how often this has
Finally, it provided guidance for safety professionals to broaden the range of safety approaches they might use to produce better future outcomes. As more accidents were investigated with this approach, new opportunities for improving predictive tools arose.
I had been trying to formalize ny work by documenting the ideas about making decisions in Hazardous Materials Emergencies as the work progressed. The ideas were passed out as handouts to students as they evolved. In 1975 Martin Grimes and Tony O'Neil decided to publish the finished paper in NFPA's Fire Journal to give the information widespread exposure in the fire service community. It was republished in NFPA's Publication SPP-49, Hazardous Materials Transportation Accidents, in 1978. That paper is posted on the internet at http://www.www.bjr05.net/papershm/DECIDE.htm It describes each element of the decision making process, and the rationale supporting each step. In it, the emergency process model with possible intervention points was published in flow chart format for the first time.
During this period, NFPA staff was actively engaged in the problems, and produced several publications describing what was happening. One of the most important was the NFPA Technical Information Bulletin 1-74, which was a progress report on NFPA's study of hazardous materials accidents. Acknowledging that the report might be considered alarmist by some, Mr. Grimes explained that the report was issued so that "at least, fire fighters should know the realistin risks they might have to face." It is my personal opinion that his action was an unusual display of courage, in that some of the major supporters of his organization were producers and shippers of the commodities that were creating these problems.
Continuing accident feedback
As the ideas for hazardous material emergency response guidance to support the decision making needs of responders were evolving I had the unique opportunity to test and refine my ideas against what was observed in subsequent accident investigations. I was fortunate to be an "outsider" doing this because I was not bound in any way by the experiences and implicit assumptions built into the responders' perceptions, viewpoints and expectations. Each accident and many class sessions added something new or forced me to refine the ideas about the D.E.C.I.D.E. and GEBMO models and others that evolved, over the period I taught at Montgomery College. until 1977.
In May 1973, sparks from railroad brake shoes ignited the wood floor of a box car, which resulted in the explosion of military bombs in that and a number of adjacent box cars over a five mile stretch of track. Fortunately the explosions occurred in a deserted area. minimizing damage. The incident demonstrated the difficulties responders would face in trying to control the fire - inside a closed box car - before the bombs exploded. Interestingly, not very many bombs detonated, raising questions about the lingering risks posed by the ones that had been heated but had not exploded. A whole new element had to be considered for the decision process. Hmmm.
On July 5, 1973, a tank car containing LPG parked on an industrial rail siding in near Kingman, started leaking during unloading. The leaking gas ignited, knocking two unloaders off the car. The fire began to intensify, engulfing the loading rack and the top of the car. The fire department began trying to get water on the car to cool the tank, when about a half hour later the tank car exploded, The fireball extended 15- to 200 feet from the car, leaving 13 dead and 95 injured. While the Board did not investigate this accident, since the car was no longer in transportation, we followed the investigation closely. .
Pan Am Boston 707 Freighter
This accident occurred over a 20 minute span in November 1973, as nitric acid started a fire in the cargo area after the aircraft was airborne. As our investigation proceeded, I observed that the emergency response challenges for the crew were not unlike challenges in other hazmat accidents with respect to the decision making process involved. The crew was confronted with something they had never experienced, and had no basis for recognizing that hazmat were involved -- they first thought they were dealing with electrical overheating of fire, and acted on that assumption until other symptoms appeared. The commonality included the hazmat detection, threat prediction and assessment, selection of action options and the influence they had on the evolving outcome. The closed environment was different, but the accident fit the models that were evolving in my mind. The lessons learned were learned again a generation later in the Air Florida crash into the everglades. 1974 was a bad year.
Oneonta, New York
Another "Crescent City" type accident occurred in a rural setting in February 1974. Still the LPG explosions and fire injured 54 firemen and members of the press during the emergency response. It was their first experience with this kind of emergency. The firemen had not been trained to cope with an emergency of this magnitude. This is the accident to which Zercher referred when he observed that firefighters should climb the nearest hill to watch it, because they would never see another one like this again, and there wasn't anything they could do about it. The accident produced some spectacular photos, widely published and circulated in the fire service. The fire service community leadership was gradually but surely beginning to lose patience with these kinds of accidents and casualties, and the risk creators.
In July 1974, a switching accident in a railroad yard released isobutane from a tank car, It vaporized for about 10 minutes, exploded at 5:03 a.m. and damaged 700 homes, making some uninhabitable. A nearby school was one of the most severely damaged buildings, and 31 commercial establishments reported damages. Losses were estimated to be $18 million. 7 railroad employees near the explosion were killed, and numerous injuries were reported outside the railroad yard. This was one of those accidents about which responders could have done nothing to change the course of events. Had they had arrived on scene and pursued their customary tactics, more casualties would have occurred. .
The following month, on August 6, 1974, a contents of a 10,000 gallon tank car detonated unexpectedly in a switch yard. The monomethylamine nitrate in solution was classified as a flammable solid. Here was another example of a classification scheme failure, based on the detonation. It acted like a duck, but it wasn't classified as a duck. The implication for everyone handling the shipments were readily apparent. Standing in the deep crater of a flammable solid detonation left a strong and lingering impression on this investigator. What if firefighters had responded to a wisp of smoke visible during the few minutes before explosive reaction and approached it in the customary manner? The complication of the responders' threat prediction challenge due to the hazmat classification reinforced my perception of the need to change the basis for response guidance then being circulated. .
On September 1, 1974, colliding trains resulted in the release of isobutane from a punctured tank car, with ensuing fire. Firemen evacuated eight families in a 1 mile area, but allowed the wreckage to burn. A first. They will probably never see another fire like that one.
On September 21, 1974, another explosion in a railroad switch yard when a tank car carrying butadiene was punctured during railroad yard switching operations. The butadiene vaporized, ignited and in 2 to 3 minutes, exploded. One person died, and 235 were injured in the blast. This one blew before responders could get on scene. The Board reiterated its recommendation that these tank cars be treated like explosives in railroad switch yards. The observed range of effects in these accidents was really bothering me more and more as I pondered the implications for emergency responders.
Eagle Pass Texas
In April 1975, a tank semi-trailer with LPG separated from the truck tractor, struck a concrete abutment and ruptured, releasing the LPG which vaporized, ignited and exploded, fatally injuring 16 and burning 51 others in the area. The Board recommended
research to try to reduce the severity of such accident. Emergency response
was not addressed in this case.
Glen Ellen Illinois
In May 1976 ammonia escaped from a tank car derailed in a populated suburban Chicago community, and I did get on scene there. 14 persons were treated for exposure to the released ammonia. I noted with great interest that in this accident, persons who remained indoors were well protected against injury from exposure to the released ammonia, providing evidence that evacuation was not always the only option to be considered in hazmat accidents. In the report, the Board stated that when hazardous materials are present, firemen might have to abandon their traditional "attack ad extinguish" approach and adopt alternative emergency response procedures. I felt comfortable proposing that to the Board because I know by then from my work and consultations that options were available.
In September 1976, a tank-semitrailer load of anhydrous ammonia was released in a single truck accident on an interstate highway cloverleaf bridge, resulting in 6 fatal exposures, hospitalization of 48 and treatment of 100 others. The ammonia dissipated fairly rapidly. It was not a good year for ammonia transportation. The emergency resonse was not investigated because there was not much that could be done to combat the widespread dispersion of the vapor cloud around the clover leaf intersection before it naturally dissipated that day.
Derailing cars in this November 1976 railroad accident crashed into fuel storage tanks adjacent to the track, starting fires that burned for 12 hours. Several explosions occurred as the fire burned the derailed cars and nearby buildings. Firefighters were unable to determine the contents of the cars involved for more than 8 hours after the crash. Because it happened in a remote area, the accident did not receive much publicity. However, it did prompt serious dialogue among responder, carrier and shipper experts about the communication of needed data about the shipments involved in these accidents.
Rockingham, North Carolina
This March 1977 derailment involved challenges for the emergency response involving a radioactive material shipment. It resulted in a Board special study of released in 1979 calling on a critical review of all existing emergency response plans to determine their adequacy. In this accident, the fear factor aroused by the presence of a radioactive material container seem to create unnecessary stresses, in my view. It emphasized the need for better predictive threat assessment capabilities.
A November 1977 derailment accident resulted in another release of a tank car load of anhydrous ammonia, 2 fatalities, 46 injuries and the evacuation of about 1000 persons. The Board noted that the preplan helped responders act effectively when they intervened, and again reported the lack of information on how to determine the potential danger zone of exposure to released hazardous materials. Another occasion when the wide geographic range of the gaseous threat posed special challenges to the responders.
The February 1978 derailment near Youngstown Florida was the first involving the release of liquefied chlorine in large quantities. The released gas killed 8 and injured 138 others. A three mile long gas cloud formed after the release. The cloud gave no discernible indication of its lethality to exposed victims. We kept seeing these accidents where there was little responders could do, due the large areas affected and the speed with which they were affected. It added pressure to take action to prevent the releases in the first place, or alt the very least, reduce the release rate from breached containers of these liquefied gaseous commodities.
This February 1978 derailment and explosion turned out to be the turning point among emergency response disasters. It was the last one in which firefighter fatalities
occurred during the response operations. 16 died and 43 were injured on the second day after the accident, when a tank car was being handled during reponsders' efforts to reduce the residual risks posed by the damaged car full of LPG.
14 persons were injured by hazmats released during this April 8, 1979 derailment. The derailment occurred adjacent to a waterway, creating grave environmental problems during the cleanup operations. The report dealt extensively with emergency response issues, ranging from on-scene commander identification, dissemination of guidelines, evaluation of response preparedness and others.
The M.A.P.S. Project
Observations of threat assessment problems during these accidents suggested that the mapping of past spills might be helpful to responders trying to figure out where released hazmats might migrate. The Board developed and then began publishing its Spill Maps series. The intent was to use them for other purposes too, such as validating spill model predictions that were evolving at the time, and for data about the range of effects for various risk assessment projects like routing guidance, etc. It may not be entirely clear to the casual observer, but much of the effort by the Board's Hazardous Materials Division was to encourage implementation of the D.E.C.I.D.E and GEBMO models.
Implementation of the models
The course developed
at Montgomery College formed the basis for the instruction of over 150,000
firefighters by the railroad industry in years that followed. For several
years after 1979, when the instruction became widespread through efforts
of the Association of American Railroads, Federal Emergency Management
Association and others, to my knowledge no emergency responders trained
in the elements of the DECIDE process were fatally injured or disabled
by hazmats in transport accidents. Rather than bearing risks 10,000 tunes
greater per hour of exposure than others, the risk levels for responders
declined to background levels for this activity. The
ultimate result was what is now known and apparently widely taught as the
D.E.C.I.D.E. process for hazmat and other emergencies where intervention
can make a difference in the outcome. The
D.E.C.I.D.E. process was applied experimentally to pilot emergency training in later years,
with some apparent success. The D.E.C.I.D.E. and GEBMO models were used
to identify needed changes in a state's emergency response agencies' assigned
roles in accidents.
View study at http://www.iprr.org/HazMatdocs/sts71.html
Study can be viewed at http://www.iprr.org/HazMatdocs/sts71.html
 See Chauncy Starr's seminal work on societal risk in An Overview of the Problems of Public Safety, presented to the Symposium on Public Safety - a Growing Factor in Modern Design at the National Academy of Engineering, May 2, 1969
 See flow chart at http://www.iprr.org/HazMatdocs/LOXchart.jpg
 See paper introducing multilinear events sequencing concepts at http://www.iprr.org/papers/mes75.html
 See LP GAS EXPLOSION, Kingman AZ July 7 1973, published by NFPA. 12 firefighters were fatally injured in this one.
 See SAFETY TRAINING'S ACHILLES HEEL found at http://www.bjr05.net/papers/AchillesHeel.htm
 see Fire Officer 's Guide to Dangerous Chemicals, ,Bahme, Charles W., N.F.P.A., Boston,MA,1972
See Flammable Hazardous Materials, Meidl,J.H.,Glencoe Press, Beverly Hills,CA 1970 and his accompanying volume Explosive and Toxic Hazardous Materials
 See 1975 paper Accident Investigation: Multilinear Events Sequencing at http://www.bjr05.net/papersa/mes75.htm Note: this is now obsolete: see Guides referenced in Footnote .
 See HAZARDOUS MATERIALS TRANSPORTATION: CURRENT AND FUTURE ISSUES posted at http://www.bjr05.net/papershm/hmtissues.htm
 NFPA Fire Journal, volume 6:9, July 1975. See article posted at .http://www.bjr05.net/papershm/DECIDE.htm
 HAZARDOUS MATERIALS EMERGENCIES, first published by Lufred Industries in 1976. It is posted for viewing as a PDF file - Preview publication
 HAZMAT INVESTIGATIONS, is one of four investigation booklets originally published by Fire Protection Publications. It is now available from the Emergency Film Group as part of a video set (see the bottom of the page for separate pricing), or bulk orders are available from the author.
 The Time /Loss Analysis method for assessing emergency response performance is described in a publication found at http://www.starlinesw.com/product/Guides/MESGuide07.html
 The NTSB Spill Map program is described in a paper posted at http://www.bjr05.net/papershm/MAPS.html