United States
Environmental Protection
Agency
Office of Solid Waste
and Emergency Response
(5104)
EPA 550-F99-004
March 1999
www.epa.gov/ceppo/
EXPERT REVIEW OF
EPA/OSHA JOINT CHEMICAL
ACCIDENT INVESTIGATION
REPORT
Napp Technologies, Inc.
Lodi, New Jersey
Chemical Emergency Preparedness and Prevention Office
'* Printed on Recycled Paper
-------
CONTENTS
Introduction Page 1
Executive Summary Page 3
Expert Review Page 5
Chair's Report Page 6
Background and Statement of Purpose Page 6
Process of the Review Page 6
Chair's Summary of Comments and Recommendations Page 7
Individual Reviewers' Comments (Final Reports and Initial Comments) Page 9
Gablehouse Comments Page 9
Cox Comments Page 14
Scannell Comments Page 34
Freeman Comments Page 42
Sprinker Comments Page 66
EPA Response to Expert Review Page 75
I. Response to Comments and Recommendations from Chair's Summary Page 75
II. Response to Comments and Recommendations Noted by Several Reviewers.... Page 78
III Response to Comments and Recommendations Noted by Individual Reviewers.. Page 80
Appendix A-EPA Direction to Reviewers Page 81
-------
Page 1-NAPP
INTRODUCTION
Under a Memorandum of Understanding (MOU), EPA and OSHA have worked together
to investigate certain chemical accidents. The fundamental objective of this joint effort is to
determine and report to the public the facts, conditions, circumstances, and causes or probable
causes of any chemical accident that results in a fatality, serious injury, substantial property
damage, or serious off-site impacts, including large scale evacuation of the general public. The
ultimate goal is to determine the root causes in order to reduce the likelihood of recurrence,
minimize the consequences associated with accidental releases, and to make chemical production,
processing, handling and storage safer. ( Section 112 (r) (6) of the Clean Air Act Amendments
of 1990 established an independent Chemical Safety and Hazard Investigation Board to
investigate and determine the cause or causes of chemical accidents and recommend steps to
prevent similar incidents. At the time of the Napp accident, the Board had not been formed. The
Chemical Safety Board is now in operation and conducting accident investigations.)
On April 21, 1995, an explosion and fire took place at the Napp Technologies (Napp)
facility in Lodi, New Jersey, resulting in deaths, injuries, public evacuations, and serious damage
both on and off site. The accident occurred when Napp employees were attempting to blend
sodium hydrosulfite, aluminum powder, potassium carbonate and benzaldehyde, in order to make
a gold precipitating agent. Napp was performing the blending operation under a contractual (or
toll) arrangement with the owner of the gold precipitating agent, Technic, Inc (Technic) of
Cranston, Rhode Island.
EPA and OSHA formed a joint chemical accident investigation team (JCAIT) which
undertook an investigation of this accident because of the serious consequences and
characteristics of the substances involved. In October of 1997, the JCAIT published the
EPA/OSHA Joint Chemical Accident Investigation Report: Napp Technologies, Inc., Lodi,
New Jersey (EPA 550-R-97-002). In the report, the JCAIT noted six findings as root causes and
contributing factors of the event. The JCAIT also developed six recommendations that address
the root causes and contributing factors to prevent a reoccurrence or similar event at other
facilities. For copies of the report, contact www. epa. gov/ceppo/ or call 1-800-490-9198.
This Expert Review of the investigative report presents the comments and
recommendations of five expert reviewers with whom EPA contracted to provide an independent
examination of the EPA/OSHA Joint Chemical Accident Investigation Report. Included are
EPA's and OSHA's responses to their comments and recommendations.
We asked Timothy Gablehouse, chair of the Jefferson County Local Emergency Planning
Committee and member of the Colorado State Emergency Response Commission, to chair the
review and to suggest other reviewers who could, like himself, render independent opinion about
the EPA/OSHA report. Upon his recommendations and our own examination, we selected the
-------
Page 2- NAPP
four other reviewers: Dr. Wade Freeman of the University of Illinois, Dr. Geraldine Cox of
Ampotech Corporation, Michael Sprinker of the International Chemical Workers Union Council,
and Jerry Scannell of the National Safety Council. All five reviewers then independently
commented upon the report in writing, and the Chair circulated all comments to all reviewers.
EPA and OSHA investigators then briefed all reviewers in Washington on September 14, 1998
and answered questions about the conduct of the investigation. The reviewers then met alone to
discuss their comments. The Chair wrote a summary of the meeting and recommendations, which
were then forwarded to EPA. EPA and OSHA, in this Expert Review document, are publishing
the Chair's report, all reviewers' initial and final comments, and EPA/OSHA's response.
-------
Page 3- NAPP
EXECUTIVE SUMMARY
This report contains the complete analyses of all five external expert reviewers of the joint
EPA/OSHA investigative report of the Napp Technologies. Also included are EPA and OSHA's
response to their major recommendations and comments.
EPA and OSHA have determined that the comments of the reviewers did not change the
determination of root causes and contributing factors (which were the focus of the investigation).
Therefore, there is no need to materially change the Napp investigation report. However, the
agencies will utilize the comments and recommendations in upcoming accident reports as well as
in Alerts and other products that stem from this investigation.
The following is a summary of the general comments and recommendations of the panel
and a summary of EPA's and OSHA's responses to these comments.
Major reviewer comments and recommendations (from the Chair's summary):
• The report appeared to correctly state the root causes of the accident both in terms of
technical mechanisms and technical failures.
EPA and OSHA are encouraged to consider detailed recommendations on the special risks
associated with tolling operations and the handling of water reactive chemicals, and
consider rulemaking by either EPA or OSHA.
Several elements which could have enhanced the report's usefulness:
—discussion of the types of chemical analyses done,
—discussion of the rationale used to eliminate plausible scenarios,
—clearer depiction of difficulties presented by the extent of destruction, which
made certain analyses impossible and information difficult to obtain,
—better tracking of individuals involved through the chronology of the accident,
—time line of events.
The pre-release review which EPA and OSHA allowed Napp, but not the other
stakeholders, should have been shared by all stakeholders, and the report should have been
peer-reviewed.
EPA/OSHA response:
• In the report, we cite specific management deficiencies (i.e. lack of training, inadequate
process hazard analysis, inadequate SOPs) as the root causes and contributing factors of
the incident. We agree with those reviewers who noted that we could have made more
explicit fundamental management system failure as a root cause, with the various specific
management system failures as subparts.
-------
Page 4- NAPP
The statutes (Occupational Safety and Health Act and the Clean Air Act Amendments)
that provide the authorities for the agencies' accident prevention programs make the
owner or operator of the stationary source (or the employer) who is handling the
hazardous chemicals solely responsible for compliance with safety regulations at the
facility. Hence, owners or operators of tolling operations, provided they fall under the
regulations promulgated under these statutes, are already regulated, and for this reason,
the agencies do not agree that further regulations specific to the tolling industry are
necessary. However, the Agencies agree that more can be done to increase the safety of
performing tolling operations and handling water reactive chemicals. Each agency is
considering the addition of chemicals that were involved in this incident to the list of
chemicals that are subject to each agency's accident prevention program. Specifically,
OSHA has announced an upcoming Advanced Notice of Proposed Rulemaking (ANPRM)
that will discuss the regulation of reactive chemicals under the Process Safety
Management Standard. EPA is also in the process of reviewing its list of regulated
substances promulgated in the Risk Management Program regulations under the Clean Air
Act Amendments, section 112(r). In addition, the agencies are taking more immediate
steps to address the risks of tolling operations by a combination of actions: first, working
with the American Institute of Chemical Engineers' (AIChE) Center for Chemical Process
Safety to develop guidance for the chemical industry; this would include identification of
risks and procedures recommended for better safety in tolling operations. EPA is also
developing an Alert directed to local responders regarding information resources during
emergency responses, and OSHA has issued a Hazard Information Bulletin describing the
potential hazards of utilizing MSDSs as the primary sources of information for conducting
hazard analyses for chemical process activities. Finally, EPA has worked with the
National Oceanic and Atmospheric Administration to develop and promote the use of a
database on reactive substances.
The suggested additions to the report would have enhanced its clarity, although some of
the suggestions would have been impossible to include given the level of physical
destruction at the Napp facility.
The factual information that formed the basis of the EPA/OSHA investigation report was
obtained through the authority of Section 8 of the Occupational Safety and Health Act
[P.L. 91-596]. OSHA regulations at 29 CFR part 70, implementing Executive Order
12600, require OSHA to show the factual part of a report—not its recommendations and
conclusions— to the facility before publication, to allow for their identification of trade
secrets. Sharing this with other parties (i.e. an external expert review panel) prior to the
facility's review would not have allowed protection of trade secrets.
-------
Page 5- NAPP
EXPERT REVIEW
-------
Page 6- NAPP
EXPERT PEER REVIEW
OF EPA/OSHA JOINT CHEMICAL ACCIDENT INVESTIGATION REPORT
"NAPP TECHNOLOGIES, INC., LODI, NEW JERSEY
BACKGROUND AND STATEMENT OF PURPOSE
An explosion and fire took place at the Napp Technologies facility at Lodi, New Jersey, on April 21, 1995,
resulting in deaths, injuries, public evacuations, and serious damage both on and off site. The accident
involved a commercial chemical mixture, a gold precipitating agent, identified as ACR 9031 GPA, owned
by Technic Inc. of Cranston, Rhode Island and comprised of sodium hydrosulfite, aluminum powder,
potassium carbonate and benzaldehyde (GPA). At the time of the accident at the Napp facility, Napp was
performing a toll blending operation for Technic.
The EPA report investigating the incident was developed as part of the Agency's ongoing responsibilities
under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) and the
Clean Air Act, Section 112(r), as a component of EPA's chemical safety programs and in conjunction with
OSHA's enforcement investigation. The report was released to the public in October 1997 to become a
part of the examination of the causes of chemical accidents and efforts to prevent them. The report was
prepared by staff at EPA Headquarters and Region II office and at OSHA Headquarters and the Area
Office.
As part of its investigative program, EPA arranged for this review by a panel of experts. The purpose of
the review is for each of the reviewers to comment on the scope, approach and conclusions of the report
and its implications from their individual perspectives and disciplines.
PROCESS OF THE REVIEW
EPA contracted with Timothy R. Gablehouse to Chair the expert review panel. Mr. Gablehouse then
selected, Drs. Geraldine Cox and Wade Freeman, Mr. Jerry Scannell and Mr. Michael Sprinker to form the
rest of the review panel.
It is important to note that it was not the function of the panel to reach a consensus point of view on any of
the following issues. Instead, each member of the panel prepared independent evaluations based upon their
experience, the materials provided by EPA and OSHA, impressions gained during a group meeting, and the
text of the report. The members of the panel did not perform any independent investigation of facts specific
to the Napp Technologies accident.
As an initial effort each member of the panel was asked to prepare their initial thoughts based only upon a
review of the report. These initial thoughts were shared among the panel members and agency
representatives planning to participate in a group meeting. This sharing of initial thoughts allowed the
panel members to consider additional issues and to help prepare the agency participants for the group
discussion.
-------
Page 7- NAPP
The panel members met with the principle investigators from OSHA and EPA and other senior level agency
representatives on September 14, 1998 at EPA Headquarters in Washington, DC. At this meeting the
panel members viewed a video tape prepared by the agencies intended to depict the chain of events leading
to the accident as well as participating in a detailed discussion of the findings of the report. Agency
participants were well prepared and extremely cooperative in sharing their thoughts and observations with
the panel members. The panel members held a private discussion during the afternoon to exchange ideas,
concerns and comments.
Following the group meeting each panel member prepared their individual comments in written form.
These comments follow this introductory material and form an integral part of the report of this panel.
In its charge to reviewers, EPA asked that the following questions be considered:
Comment on the report's organization. In general, was there a logical progression in the chain of
reasoning - were conclusions adequately supported by the facts?
Was the focus on identifying potential sources of heat and water appropriate? Were the sources of
reaction initiation which were identified plausible? Were all possible sources of the reactions identified?
Was the discussion of root cause adequate? Were root causes and contributing factors
appropriately and correctly identified?
Are recommendations appropriate and drawn logically from the preceding discussion and
conclusions? Are recommendations sufficient to address the potential of a recurrence of this kind of
accident in other facilities? How will other facilities be able to apply the findings and recommendations of
this report to their particular circumstances?
Were the appendices sufficient and appropriate? Were the photos appropriate to illustrate the
narrative, clear, and properly presented?
Were all external factors considered? Were human factors and management issues considered
appropriately?
What aspects of this report could help inform future investigations? Was the approach sufficiently
broad for application to other industry sectors? Were roles of all stakeholders properly addressed in the
report, including roles of federal, state and local agencies, the community, labor and any others? Are
recommendations sufficiently broad to include all elements in addressing prevention of like accidents in the
future?
CHAIR'S SUMMARY OF COMMENTS AND RECOMMENDATIONS
With the caveat that the other members of the panel have discussed many of these issues in greater depth in
the following materials, the Chair believes that it is reasonable to state the following summary observations
and conclusions:
-------
Page 8 - NAPP
The investigation was complicated by the catastrophic level of destruction of the facility and
records, the death of many of the people with direct knowledge of events immediately prior to the accident,
and the fact that this was the initial cooperative investigation between EPA and OSHA under a formal
MOU. The investigators are to be commended for performing a thorough and useful analysis given these
hardships.
The report appears to correctly state the root causes of the accident both in terms of technical
mechanisms and management failures. Later discovered information discussed during the group meeting
support the findings of the report.
In addition to the root causes identified in the report, it appears that there were additional problems
with communications and relationships between the local emergency response agencies and Napp. While it
cannot be known with certainty, it appears that these problems may have caused Napp personnel to respond
in a less-than-adequate fashion as the emergency developed.
The usefulness of the report to other companies and later investigators could have been enhanced
by a more detailed discussion of the types and results of the chemical analyses performed during the
emergency response to the incident, and by a more detailed discussion of the rationale used to eliminate
plausible scenarios. It would also have been useful to note the difficulties presented by the level of
destruction and death by more clearly describing where information may have existed but was destroyed
and where information is third-hand or even more remote because of the death of persons with direct
knowledge. Certainly the investigators might have wanted to perform more analysis or interviewed more
people, but the conditions made this simply impossible.
It was difficult to track the various individuals through the chronology leading to the accident. A
tabular time line of events with details on which people were involved would have been useful.
The agencies should have considered and discussed more detailed recommendations on the special
risks that seem to exist in tolling operations and in the handling of water reactive materials. While some
formal and informal initiatives seem to be under consideration, the risks seem to justify a meaningful effort
to better allocate the accident prevention responsibilities in these situations. While guidance is certainly a
possibility, rulemaking by either EPA or OSHA should be considered.
The photos were not very useful as the reproduction process removes many details. The agencies
should consider the use of higher quality reproduction or the use of the internet to post higher quality
photos using color. High resolution photos posted on the internet as an appendix to the report would
appear to be a low-cost solution to this problem.
The video was helpful to understanding the incident. Even though it may contain technical
inaccuracies due to financial constraints, video recreation is a useful tool to understanding complex
incidents.
-------
Page 9 - NAPP
COMMENTS OF TIMOTHY R. GABLEHOUSE, CHAIR
EXPERT PEER REVIEW OF EPA/OSHA JOINT CHEMICAL ACCIDENT
INVESTIGATION REPORT
NAPP TECHNOLOGIES, INC., LODI, NEW JERSEY
In summary the investigators are to be commended for their efforts and analysis. The level of
destruction and the death of so many people with necessary information about the events shortly
before the accident made the investigation very difficult.
1. Comment on the report's organization. In general, was there a logical progression in the
chain of reasoning - were conclusions adequately supported by the facts?
In general the report is well organized. It would have been helpful for the report to contain
greater detail regarding why certain scenarios were discarded. It also would have been useful for
the investigators to state when the level of destruction or lack of first-hand knowledge made it
impossible to conduct certain analyses they might have desired.
It was difficult to track events in conjunction with the people involved as one goes through the
text of the report. In some cases it is not clear when information is third-hand due to the death of
individuals with primary information or when information is limited or no longer exists due to
destruction of the facility. It would have been useful for events to be tracked along with the
people that were directly involved so that the reader could determine who was present and
whether or not they were a survivor of the incident. Clearer discussion of where conclusions
were limited due to lack of surviving information would have been useful.
2. Was the focus on identifying potential sources of heat and water appropriate? Were the
sources of reaction initiation which were identified plausible? Were all possible sources of the
reactions identified?
It is not certain that all possible sources of the reactions were identified. It appears fairly certain
that all the most plausible sources of the reactions were identified. It seems most likely that water
was introduced into the blending vessel as suggested by the report.
The possibilities of product impurities, including water, being present could have been explored.
These impurities could have been present in the raw materials depending upon the grade
purchased, contamination could have occurred during storage (package damage apparently did
occur), or during other handling activities. It is not at all clear whether such studies could have
been conducted due to the level of destruction and loss of records.
-------
Page 10-NAPP
3. Was the discussion of root cause adequate ? Were root causes and contributing factors
appropriately and correctly identified?
While the discussion is adequate, it appears that a more fundamental management failure was
present and should have been described as the root cause with the various specific management
failures as subparts. All of the failures of process hazards analysis, operating procedures, training,
emergency response and decision making correctly identified in the report, are due to the absence
of a comprehensive and coordinated health, safety and environmental management system. The
failures identified are symptoms of this larger failure.
The discussion of the relationship between Napp and the local emergency response agencies is
limited. It appears that this relationship may have contributed to Napp's failure to notify the fire
department during the early stages of the incident. The EPCRA compliance of Napp and whether
the awareness of the local fire department could have been improved by LEPC activities should
have been discussed. These sources of information might be useful to agencies and companies
seeking to prevent future accidents.
4. Are recommendations appropriate and drawn logically from the preceding discussion
and conclusions? Are recommendations sufficient to address the potential of a recurrence of
this kind of accident in other facilities? How will other facilities be able to apply the findings
and recommendations of this report to their particular circumstances?
In general the recommendations are logical but do not always reach an appropriately strong
statement. The report describes two matters that appear to present a high risk that might not be
fully appreciated by those effected. These two matters are tolling arrangements and blending of
water reactive materials with other materials that could provide reaction energy.
In the case of tolling arrangements, it appears that both OSHA and EPA should have considered
regulatory initiatives to clearly place responsibility for developing information and communicating
risk information. While all parties to tolling arrangements carry some level of responsibility, the
default application of existing regulations and the limitations of the information presented in
MSDSs did not ensure that the companies understood the risks. Possible regulatory initiatives
include increasing and eliminating contradictory information available on MSDSs for reactive
materials and requiring that tolling arrangements include a detailed process hazards analysis.
5. Were the appendices sufficient and appropriate? Were the photos appropriate to
illustrate the narrative, clear, and properly presented?
The photos were difficult to use due to the loss of detail in reproduction. The internet could be
used to post high quality photos at a reasonable cost.
-------
Page 11 - NAPP
6. Were all external factors considered? Were human factors and management issues
considered appropriately?
In general these issues were adequately considered. The discussion of the management issues
noted above could have been improved.
7. What aspects of this report could help inform future investigations? Was the approach
sufficiently broad for application to other industry sectors? Were roles of all stakeholders
properly addressed in the report, including roles of federal, state and local agencies, the
community, labor and any others? Are recommendations sufficiently broad to include all
elements in addressing prevention of like accidents in the future?
The approach was adequately broad so as to provide a basis for future investigations. It would
have been useful for the report to have noted analyses that the investigators would like to have
performed or information they would have likely to develop but could not due to the destruction
present in this case.
I am concerned about stakeholder involvement. Most specifically the pre-release review by Napp
in a non-public forum is troubling. It would be better for the report to have been prepared and
then shared with all stakeholders at the same time.
Certainly the agencies involved in the investigation need to have some sensitivity to their
respective enforcement roles and the potential impact of the report on civil or criminal litigation.
Nonetheless, the greatest benefit of this type of report is accident prevention. The users of the
report need to rely upon the fact that the report has not be so sanitized or influenced by the
facility as to be missing key information elements.
-------
Page 12 - NAPP
Initial Summary Comments on EPA/OSHANapp Accident Investigation
Tim Gablehouse
1. The report does a good job of analyzing a complex incident. Clearly the number of deaths
and the magnitude of the physical damage increased the difficulty of the job. The report does
demonstrate the value of accident investigation and evaluation of causes. The follow through is,
of course, critical.
2. It appears, but is not certain, that all potential causes of the chemical reaction and/or
sources of water in the system were evaluated. It appears that the writers of the report
discounted certain possibilities for reasons that are not always discussed in detail.
3. Analytical analysis, sampling methodology and lab reports are not discussed in depth.
4. Napp's compliance with EPCRA is not adequately reviewed. The role of the LEPC,
interactions with first responders, exercises and emergency response procedures should have been
discussed in greater depth. Knowledge of facility operations within the community and first
responders could have lead to some recommendations relevant to these programs.
5. It appears that management system failures were at the root of failures of training, hazard
analysis and maintenance. The analysis of these failures is limited and should have been expanded.
6. An analysis of the role of employees and their participation in training, emergency
response, and maintenance programs is not discussed in adequate depth. It would have been
useful to understand more about the degree to which these employees had discretion in these
areas, failed to exercise this discretion, or failed to follow established procedures.
7. More about the role of supplier of the materials to be blended would have been
appropriate. An analysis of whether they were in a better position to advise Napp about potential
hazards for this process and equipment could have lead to meaningful recommendations of
regulatory changes dealing with tolling agreements. For example, are there regulatory changes
under TSCA that could require more sharing of information and customer control in tolling
arrangements.
8. An analysis of Napp's compliance with employee health and safety regulations along with
EPCRA could have lead to an understanding of whether regulatory gaps existed that lead to this
accident. Recommendations on filing these gaps could have resulted.
9. Recommended compliance without an analysis of why noncompliance existed is not as
useful as it could be. For example, is the noncompliance due to lack of awareness, criminal intent,
failure in enforcement, complexity of the regulations or other reasons? If these could be evaluated
in greater detail some specific recommendations might have resulted.
-------
Page 13 -NAPP
10. By the time the report was written, were any recommendations being implemented?
11. It would have been better for the report to have been released prior to review by Napp.
-------
Page 14 - NAPP
Geraldine V. Cox, Ph.D.
NAPP Technologies, Inc.
EPA/OSHA JOINT CHEMICAL ACCIDENT
INVESTIGATION REPORT
A Critical Review
September 17,1998
-------
Page 15 -NAPP
Executive Summary
Overall the EPA/OSHA Joint Chemical Accident Investigation Report, EPA 550-R-97-002
dated October 1997 is a competent review of an industrial accident. Some areas of the
report would have benefited from additional information, but, given the resources available
and the other commitments of the staff who wrote the report, it is a sound document
Technical Soundness
The review of the incident included:
• A review of each material used with a summary of the potentially hazardous
properties of the individual chemicals involved in the incident;
• A discussion of the potential hazards involved in formulating the gold precipitating
agent, GPA;
• A synopsis of the previous blending experience by Napp Technologies, Inc.;
• A chronology of the events before and after the incident;
• A discussion of the hazard management procedures used in this incident;
• A postulation for the cause of the incident with supporting evidence; and
• A limited discussion of emergency response.
A good incident review should focus on the systems operating before the incident, the
chronology and probable cause of the incident itself, and the response to the incident.
The first two were covered well — although worker and supervisor training (especially for
out-of-norm events) is weak, and it appears that the assessments of cause are accurate.
The discussion of the equipment and the water leak was good, but left some areas that
would have benefited from additional discussion. For example, any information on the
analysis of the residues of reactants from the vessels would be useful. The damage
described to the equipment seems to justify the conclusions about the evolution of the
deflagration. The discussion of prevention of future incidents could be stronger.
In cases where records are destroyed in the incident and not available for review, the
situation should be specified in the accident report. In addition, when presenting the
chronology of events, specification as to source, i.e., first-hand vs. second-hand
information, is important. In this incident, some of the first-hand observers were lost in the
incident, as were operating records. Some observations stemmed from the memory of
survivors and what some of those lost in the incident told the survivors before (he
deflagration.
-------
Page 16-NAPP
Future reports should review the emergency response to the incident — from the com
pany employees and the local emergency responders — in much greater detail. Are the
responders trained properly, are the recommended methods of response appropriate, was
the level of response, e.g., local evacuations, in accord with the real hazards of the
incident? Did conflicts exist between responders as to who was in charge? Did too many
agencies respond to the incident causing problems for the responders? The investigators
should have included a discussion of the lack of notification of the neighboring community
and the local emergency responders of an incident in progress.
While I believe it to be a minor point, I am concerned about the odor observation by
employees prior to the incident Benzaldehyde is artificial almond flavoring and has a
strongly distinct odor of almonds. The employees noticed a smell of vanilla before the
deflagration, not almond. This may simply be a confusion of "food smells" and not
significant, or it may indicate that some material other than benzaldehyde was introduced
into the blender. An explanation of the inconsistency would strengthen the report. Were
chemical analyses performed on the residue to determine if the reactants were what
would be anticipated? This should be a routine part of these investigations. One
hypothesis, probably the correct one, appeared for the reaction. Were others considered
and discarded? If so, the report should present this and discuss why the reactions did not
seem to reflect the incident that occurred.
The timing of the report is ancient history relative to the incident Can the reports be
available in less than 90 days? This would be more relevant to preventing future
incidents. Staff resources were not fully available to complete the effort in a timely
manner.
Overall Approach and Completeness
The organization of the investigation and report seems sound. Two objectives seem
paramount to this type of report. The first — to identify the factors that contributed to the
incident. The second — to identify how to improve existing systems based on lessons
learned from the incident review.
The investigators logically developed an understanding of the incident. The organization
of the incident report is sound. One topic leads to the next, and the reader builds
knowledge of the process, the management, the hazards, and the incident in a logical
progression. The photographs are difficult to evaluate because the printing process does
not allow retention of detail. Unless the photographs are printed on photo quality paper —
and perhaps in color, too much detail is tost with the present printing method to give the
reader a full appreciation of the authors' intent with the illustration.
Perhaps in a formulation incident, the investigators should review the customer's hazard
assessment procedures in addition to the formulators. The report was unclear as to the
origin and completeness of the MSDS from the client, Technic, Inc. Was a product MSDS
available, or did Technic only provide Napp Technologies, Inc. with MSDS sheets for the
individual components? Did Technic perform a hazard assessment before the company
decided to use GPA for recovery of precious metals?
-------
Page 17-NAPP
Rfrcamrnm*
; erf accident
and
iCTve«t»g3tx>ns in the future
. v
f f* A' ' "* ' ^ ' -» V^"-*™
* ! ^ . ~™ j ~ &y- ™=t£^~*
tne ^'^.j ,i>
-'J ,.jfh'< , j-ic-^ss&d *r;t'~ 3 r^j^a?];^ -f,| »•>« t
,-, '•»• ii !'»»(", if ij *njr A"J_>J '•4'jUi'*? Tif • i.'."
ffiiitilp •" !h^~ mit it -iti«, *n ;• iv O
j vrjii.t uiqtr-.; . •"!••; Thi*
. i ->-.I'|MI she -<*55*iT3ti:r'n of r
-..,, HM — 5pe-tf ~a>\\ ihs "'*i;»
ti 1 ,DK H'.iUiwi s-j;-" as O£r
•r '*?>-fMV\ " • »• f ,•;»' < > •' r • •* ': - «•'„,
» -,»•, ^ : u ' "" •",'(•! • »v ?- *;c*3"~
,.'. " -V .:» f «•• K'I ' '
-------
Page 18-NAPP
Root Causes of the Accident
Underlying Prime Reasons
Accidents are almost never due to a single cause.
cascade of factors that result in an accident.
Rather most incidents involve a
This is the case at Napp Technologies, Inc. The management systems in place before
accepting the GPA formulation were structured for less reactive pharmaceutical
formulations rather ttian for toll blending of reactive components.
The sequential and solitary process safety assessment procedure practiced by Napp
Technologies management did not benefit from the exchange that a group review would
generate. The investigators identified solitary and sequential review as a problem.
However, in many firms conflicting schedules often make group hazard assessments
almost impossible. Remaps a better solution exists. Toll blenders could activate a special
group assessment procedure when reactive materials are part of a formulation. This
focuses attention on truly hazardous formulations while allowing sequental review of less
hazardous formulations. This special assessment should include identification of the norm
and discuss how to identify and respond to potential out-of-norm excursions.
The customers asking for mixtures of reactive materials should provide more than simple
material safety data sheets for each component of the mixture. They should provide a
process safety assessment analysis and a discussion of the risks; how to determine if the
process is out-of-norm; and what to do if problems appear during mixing. This information
should be reviewed as part of the process safety review by the toller and should be
reviewed with the supervisors and the workers assigned to the mixing process.
Blending equipment used primarily for Pharmaceuticals is not always appropriate for
blending reactive materials, as was the case for GPA. In addition, workers detected a
small water leak in the equipment before loading, and the workers tried to stop the leak. A
small leak should not be allowed in a pharmaceutical formulation, but it usually would not
have the fatal consequences. The decision to load the blender in spite of the continued
leak was a poor management decisbn. Maintenance protocol should not allow a leaking
blender to be used in any formulation.
Water is ttie apparent trigger for the deflagration. The employees should understand not
to allow any water to contact the blend. A stronger employee training and understanding
of the importance of water contamination might have prevented the blending. If the
workers and supervisor fully understood the reactivity of some of the ingredients, they
would not have allowed the mixing in the Patterson-Kelley 125 Blender.
The sequence of events once the blending process began to digress from the norm
indicates that no one established procedures to handle excursions from the norm. This is
-------
Pagel9-NAPP
deafly an area for improvement in this incident and in the process safety management
system in general.
The investigators provide solid evidence of the sequence of events, and the evaluation of
the actual incident seems based on an accurate evaluation of the evidence presented in
the report. Subsequent discussions with the authors of the report identified that all of the
observed chronology was not first-hand. Some of the information was reported based on
conversations with other workers who expired in the incident. What was personal
observation (and by whom) and what was reported as second-hand should be identified
and the source of the information included.
Workers reported the smell of "vanilla" before the incident. Benzaldehyde (part of the
formulation) is used in cooking as artificial almond flavoring. None of the other materials in
the formulation has a vanilla odor. The workers could have misidentified the odor by
associating a food odor — vanilla or perhaps they smelled or another compound that was
inadvertently added in place of (or in addition to) benzaldehyde. The report did not
present sufficient chemical analyses of the residues to provide further insight.
The sources of heat and water received adequate elucidation. Since both heat and water
contribute significantly to the reaction, these topics were appropriate at the level of detail
presented. The discussion of initiation seems plausible, and appropriate for the
circumstances. Because there were problems with the introduction of benzaldehyde into
the blender, can one assume that sufficient quantities of benzaldehyde were present to
cause the reactions identified in Appendix A? Decomposition of the insulation seems to
be the likely source of the large quantities of phenol and methylphenol.
In summary, the investigators presented an excellent review of the events leading to the
incident and the chemistry that supported the deflagration.
Organization of the Report
The report followed a logical plan. It allows the reader to develop an understanding of the
background, the materials involved, the history, and operation of the facility before the
incident began.
The chronology was clear and logical. The blend of technical terms with definitions as
necessary allows a lay reader to grasp the intent of the authors. I would have liked to see
the actual MSDSs for the ingredients and the GPA, but that level of detail might not be
appropriate for this report
Appendix B provided the most plausible reactions based on chemistry and observations,
but it did not provide other possible reactions. These should have been included as well
as a discussion as to why these alternative mechanisms were not considered the probable
route. There were some typographical errors in the formulas. The documentation was
adequate, but not overwhelming.
Appendix C is enlightening, however, greater detail would be helpful. How did the incident
begin, and how did emergency responders handle the incident? Was the treatment
method good or bad? Were recommendations presented about avoiding and/or mitigating
incidents of this nature?
-------
Page 20 - NAPP
6
-------
Page 21 -NAPP
Process hazards
• The firs* rsconrnenaac.ton —
:< ••• -i- ' at
CPA .v-j-> '.-«'T-i.l,iif : one* :* 1-":ie '• t !tia thr&e .e?^ t-.i'iT' f
n.'jtc'v 3*^2 'fsc-i. arc 3 vjli J -iLdi -vir his y : • -"tiim1 u\ pi ',.j--,->'-: 'A'*h '"f
^:rr« AIC
'M'vst— .-if
'.if t!
: -, -.r :ea' i* "-•: AC-".'* /«•- '••
t.r.,.,.!*-| « — • jVl '••iM-f
A- . •" ,. !''- j^;- - -Af.1 i" t"c ret,** "^ "" /«''.-' ' --! ': •.( r
»V J" '* i . ,H; * j' _"-,. v ..V-t'r i"- .=,^f- ''Li 6i~~it *' i " ,'-!'
,n-(l, i - r" -r -tt . , ! ••• * -I- Xr,*' f, -' !'f-
ffi- f''> f - i r r -. .- 'j . . -I !' ' -1 '-• '-ii i-.
' A ! -i-i •„ , -,r>- i.c'ufi fr. 3 f«' "i.c^v -^f> *%r
supervisors. ar»_, Lie Aui'ae.ii rtiiAni^ (f* iiitjiftJieiit
-------
Page 22 - NAPP
The second bullet—
Guidance is needed to address the unique circumstances surrounding tolling arrangements and the
responsibilities for hazards assessments and communication of process safety information. —
is unclear. Who needs the guidance — the toller, supervisors, workers or the
government? Businesses, especially small groups such as tollers, have more government
than they have staff to handle. The goal should be to raise awareness of the importance
of hazard analyses and imprint the need to perform process safety analyses for reactive
materials - especially to identify the norm and what to do when the reaction is no longer
within normal parameters. The government might consider an internet site to walk small
businesses through the steps for hazard assessment. This would be easily available and
tow cost for all.
• The third bullet—
Facilities should ensure that equipment manufacturers' recommendations for proper use of equipment
are followed. —
seems weak in the report. According to the manufacturer, the Patterson-Kelley 125
Blender is not appropriate for formulation of reactive materials such as those found in
GPA. As stated on page 28 of the report, the Aluminum Association recommends a
conical blender with no moving parts. The customer might have specified the type of
equipment appropriate for the mixing and might have determined if Napp Technologies
had the appropriate equipment for the job. Napp should have reviewed the equipment
specifications to determine if their equipment would be considered appropriate for the
mixing of these reactive materials.
• The fourth bullet—
OSHA and EPA should review the lists of substances subject to the Process Safety Management
standard and Risk Management Program regulations to determine whether reactive substance should
be added.—
seems like a reasonable recommendation. Perhaps some consideration of the degree of
reactivity might be appropriate in the review. This review might include perchlorates,
cholorates, and other reactive sulfur compounds. OSHA might consider a special
emphasis program on tolling operations.
• The fifth bullet—
OSHA needs to review the role of MSDS s in conjunction with HazCom, HazWoper, and PSM
Standards to clarify that MSDSs should not be used beyond their intended design. Industry should
consider additional consensus standards or guidelines to address MSDS consistency and use. —
8
-------
Page 23 - NAPP
is appropriate fhs recommertciation stiil w>i re*
;«it|0 of combination o*
as these to formulate GPA £"""er"qen:\ "t^tc'-se to
reactive matenals clarification aril 'v't.i,,..
comored MSD5 and a process analysis if tr* r'orni.
great value
« The last bullet —
'i For mixtures, a
prco.*SS nt,il:?"!t bt* Crf
• iM .«-.*i—
»);• j i -..tjarce cf oj teach is fcr users to
"j\ MW.-'VSS that no-nff tfvri he VSOS is needtil 1.0
The reerret rr.ght ex? an appropriate outreacn tool to reacn sma.i businesses If OSMA
vlevefjCied a skeie'ori for a ptxress hazard evaiud^ion and snored he/*' and he* n3t to
use MSDSs, firs might as a public serves ort may no? have wide application However
tr® general ha/artl assesarnent proce-ss and the excursion 'torn ncmidf operations, fiave
VMKJB application if
what to ao when t*¥S vanes from !fe ncrm and
ho* to harvjle an emerg^rcy with the
ccutd De as aperatirxj procedure wih reactve -"-xing situations.
then m: could reduce the seventy and probably the Dumber of ire-denis ••; This assumes a
strong worker and supervisor Iraitwej co«npcr*nt to tne manaoement function ) Pf(,x;ess
fjlet1,* assessrrerst of m«mg operators w*h 'eacJrve niater.ais to ai
irxJuslry invO^ecS with hazardous materals "
e report o.d rot ex- the corrmunrty enncrg«r'icty resfxins^ or impact very much
ether thar, to give a ge-^f-ai descnpton of the suirounding The otNjt area 'hat
stould tie expan'39-d is the internal and extern* emergency respcxxJers and their
A ciscussion about tne lack of notificabor during the events
tne deflagration *ou'd strengthen the report
r "nae""s rau.*' jr. 'tis '"',• *f".5"i,i irti.
, (p, ,-»-, ,7 '-^ija'l iflf 3~C 3L 03" .*. t._
1 fie 3;.'/c" T*jr)tj -i" '*ti »(«•"»• it ti ^
.,.,,., -,. . >f. r^tf£f. ,• &•>;„_,; i- fc sss j' '
train a"j tcc.nt !"•_-•' * s;j>.".iHf, .. '
: '.-.-' "^ : pt"f-.."in -.in 33 c->",er :r
-------
Page 24 - NAPP
Ths report provide* a tor inwesbgations. The dtvetops ih*
information rationally it to tofcw. ncidenf will differ, the overall
The raaxnmandatians haw* tor bejmnd the farmutating
industry, and era f«ssons learned from tils Help prevail
or at east mni-nize proWemB,
-------
Page 25 - NAPP
Geraldine V. Cox, Ph.D.
NAPP Technologies, Inc.
EPA/OSHA JOINT CHEMICAL ACCIDENT INVESTIGATION REPORT
A Critical Review
September 2, 1998
Chapter 1
Executive Summary
Overview
Overall the EPA/OSHA Joint Chemical Accident Investigation Report, EPA
550-R-97-002 dated October 1997 is a competent review of an industrial
accident.
Technical Soundness
The review of the incident included:
A review of each material used with a summary of the potentially hazardous
properties of the individual chemicals involved in the incident;
A discussion of the potential hazards involved in formulating the gold
precipitating agent, GPA;
. A synopsis of the previous blending experience by Napp Technologies, Inc.;
. A chronology of the events before and after the incident;
. A discussion of the hazard management procedures used in this incident;
. A postulation for the cause of the incident with supporting evidence; and
. A limited discussion of emergency response.
A good incident review should focus on the systems operating before the
incident, the chronology and probable cause of the incident itself, and the
response to the incident. The first two were covered well - although worker
training (especially for out-of-norm events) is weak, and it appears that the
assessments are accurate. The discussion of the equipment and the water leak
was good. The damage described to the equipment seems to justify the
conclusions about the evolution of the deflagration. The discussion of
prevention of future incidents could be elaborated more.
-------
Page 26 - NAPP
I would like to see future reports review the emergency response to the
incident - from the company employees and the local emergency responders in
much greater detail. Are the responders trained properly, are the recommended
methods of response appropriate, was the level of response, e.g., local
evacuations, in accord with the real hazards of the incident? Did conflicts
exist between responders as to who was in charge? Did too many agencies
respond to the incident causing problems for the responders?
While I believe it to be a minor point, I am concerned about the odor
observation by employees prior to the incident. Benzaldehyde is used as
artificial almond flavoring and has a strongly distinct odor of almonds. The
employees noticed a smell of vanilla before the deflagration, not almond.
This may simply be a confusion of "food smells" and not significant, or it may
indicate that some material other than benzaldehyde was introduced into
the blender. An explanation of the inconsistency would strengthen the report.
Were chemical analyses performed on the residue to determine if the reactants
were what would be anticipated? This should be a routine part of these
investigations.
The timing of the report is ancient history relative to the incident. Can the
reports be available in less than 90 days? This would be more relevant to
preventing future incidents.
Overall Approach and Completeness
The organization of the investigation and report seems sound. Two objectives
seem paramount to this type of report. The first - to identify the factors
that contributed to the incident. The second - to identify how to improve
existing systems based on lessons learned from the incident review.
The investigators logically developed an understanding of the incident. The
organization of the incident report is sound. One topic leads to the next,
and the reader builds knowledge of the process, the management, the hazards,
and the incident in a logical progression. The photographs are difficult to
evaluate because the printing process does not allow retention of detail.
Unless the photographs are printed on photo quality paper -and perhaps in
color, too much detail is lost with the present printing method to give the
reader a full appreciation of the authors' intent with the illustration.
Perhaps in a formulation incident, the investigators should review the
customer's hazard assessment procedures in addition to the formulators. The
report was unclear as to the origin and completeness of the MSDS from the
client, Technic, Inc. Was a product MSDS available, or did Technic only
-------
Page 27 - NAPP
provide Napp Technologies, Inc. with MSDS sheets for the individual
components? Did Technic perform a hazard assessment before the company
decided to use GPA for recovery of precious metals?
Recommendations for enhancement of accident prevention approaches and
accident investigations in the future
The investigators clearly identified some weaknesses in the process safety
system - specifically:
-The need for better communication with the client about the process
hazards prior to formulation;
-Using Material Safety Data Sheet, MSDS, information on the individual ingredients while not
looking as the combined potential hazards during the blending operation;
-The need for group discussion of process hazard assessment by
management - not sequential review for reactive materials;
-Confusing information presented on MSDSs concerning appropriate response
protocol for an incident, e.g., for a compound that reacts with water - the
recommendation is to flood with water;
-Using blending equipment that may not be appropriate for the mixing operation;
-Need for better employee training - especially with proper procedures to recognize and respond
to deviations from the norm;
-A protocol on building reentry during an incident;
-Need for better emergency responder training; and
-Recommendations for improvement of federal guidance for process hazards.
Root Causes of the Accident
Underlying Prime Reasons
Accidents are almost never due to a single cause. Rather most incidents
involve a cascade of factors that result in an accident.
This is the case at Napp Technologies, Inc. The management systems in
place before accepting the GPA formulation were structured for pharmaceuticals
rather than for toll blending of reactive components.
The sequential and solitary process safety assessment procedure practiced by
NAPP Technologies management did not benefit from the exchange that a group
review would generate. The investigators identified solitary and sequential
review as a problem. However, in many firms conflicting schedules often make
group hazard assessments almost impossible. Perhaps a better solution exists.
Toll blenders could activate a special group assessment procedure when
reactive materials are part of a formulation. This focuses attention on truly
-------
Page 28 - NAPP
hazardous formulations while allowing sequental review of less hazardous
formulations. This special assessment should include identification of the
norm and discuss how to identify and respond to potential out-of-norm
excursions.
Blending equipment used primarily for pharmaceuticals is not always
appropriate for blending reactive materials, as was the case for GPA. In
addition, workers detected a small leak in the equipment before loading, and
the workers tried to stop the leak. A small leak should not be allowed in a
pharmaceutical formulation, but it usually would not have the fatal
consequences. The decision to load the blender in spite of the continued leak
was a poor management decision. Maintenance protocol should not allow a
leaking blender to be used in any formulation.
Water is the apparent trigger for the deflagration. The employees should
understand not to allow any water to contact the blend. A stronger employee
training and understanding of the importance of water contamination might have
prevented the blending. If the workers and supervisor fully understood the
reactivity of some of the ingredients, they would not have allowed the mixing
in the Patterson-Kelley 125 Blender.
The sequence of events once the blending process began to digress from the
norm indicates that no one established procedures to handle excursions from
the norm. This is clearly an area for improvement in this incident and in the
process safety management system in general.
The investigators provide solid evidence of the sequence of events, and the
evaluation of the actual incident seems based on an accurate evaluation of the
evidence presented in the report.
Workers reported the smell of "vanilla" before the incident. Benzaldehyde
(part of the formulation) is used in cooking as artificial almond flavoring.
None of the other materials in the formulation has a vanilla odor. The
workers could have misidentified the odor by associating a food odor - vanilla
or perhaps they smelled or another compound that was inadvertently added in
place of (or in addition to) benzaldehyde. The report did not present
chemical analyses of the residues to provide further insight.
The sources of heat and water received adequate elucidation. Since both heat
and water contribute significantly to the reaction, these topics were
appropriate at the level of detail presented. The discussion of initiation
seems plausible, and appropriate for the circumstances.
-------
Page 29 - NAPP
In summary, the investigators presented an excellent review of the events
leading to the incident and the chemistry that supported the deflagration.
-------
Page 30 - NAPP
Chapter 2
Organization of the Report
The report followed a logical plan. It allows the reader to develop an
understanding of the background, the materials involved, the history, and
operation of the facility before the incident began.
The chronology was clear and logical. The blend of technical terms with
definitions as necessary allows a lay reader to grasp the intent of the
authors. I would have liked to see the actual MSDSs for the ingredients and
the GPA, but that level of detail might not be appropriate for this report.
Photocopies of black and white or color photographs is a waste of natural
resources - trees. These photographs can be scanned into a computer program
and made into half-tones or printed directly with much greater clarity. I
found the photographs difficult to study because the reproduction was so poor.
Appendix C is enlightening, however, greater detail would be helpful. How did
the incident begin, and how did emergency responders handle the incident? Was
the treatment method good or bad?
The Chemistry Appendix, B, is well presented, and logical. The documentation
was adequate, but not overwhelming.
Chapter 3
Recommendations
Process hazards
. The first recommendation -
Facilities need to fully understand chemical ad process hazards, failure
modes and safeguards, deviations from normal and their consequences, and ensure that
all relevant personnel know the proper actions to take to operate the process
safely, recognize and address deviations, return to normal operations, or
safely shutdown. This is best achieved through process hazard analyses,
standard operating procedures, and training. - is the ideal. With small and
batch operations the ideal is seldom achieved. In this case, GPA was
formulated once before, more than three years earlier. Elaborate process
safety analyses are a valid goal for large, continuous processes. With
formulators who will mix a (or many) different product every day,
-------
Page 31 -NAPP
comprehensive process safety analyses are a luxury that a small staff would
cherish. Saying that, however, does not preclude appropriate process safety
analyses when the mixture or ingredients are sufficiently reactive to warrant
more intense process evaluation. In this particular incident, GPA was blended
safely in the past, so it probably received less attention from management
on the second batch. It is not clear if the workers who formulated the first
batch were the same as those who formulated the ill-fated mixture. -
As a recommendation, however, the report might wish to distinguish the level
of hazard with the level of process safety analyses. For example,
formulations using reactive materials might require a group evaluation
where less reactive materials might be reviewed sequentially on an individual basis.
The procedures should define excursions from normal and specify a procedure
to follow if an excursion occurs.
. The second bullet -
Guidance is needed to address the unique circumstances surrounding tolling
arrangements and the responsibilities for hazards assessments and
communication of process safety information. - is unclear. Who needs the
guidance - the toller or the government? Businesses, especially small groups
such as tollers, have more government than they have staff to handle. The
goal should be to raise awareness of the importance of hazard analyses and
imprint the need to perform process safety analyses for reactive materials -
especially to identify the norm and what to do when the reaction is no longer
within normal parameters. The government might consider an internet site to
walk small businesses through the steps for hazard assessment. This would
be easily available and low cost for all.
. The third bullet -
Facilities should ensure that equipment manufacturers' recommendations for
proper use of equipment are followed. - seems unsupported by the report. In
the description of the equipment, no mention appears relative to the
appropriateness of the Patterson-Kelley 125 Blender for this reaction. This
bullet implies that the Patterson-Kelley 125 Blender is inappropriate for use
to formulate GPA. If inappropriate, this should have been documented in the
report.
-------
Page 32 - NAPP
The fourth bullet -
OSHA and EPA should review the lists of substances subject to the Process
Safety Management standard and Risk Management Program regulations to
determine whether reactive substance should be added. - seems like a
reasonable recommendation. Perhaps some consideration of the degree of
reactivity might be appropriate in the review.
. The fifth bullet -
OSHA needs to review the role of MSDS s in conjunction with HazCom, HazWoper,
and PSM Standards to clarify that MSDSs should not be used beyond their
intended design. Industry should consider additional consensus standards or
guidelines to address MSDS consistency and use. - is appropriate. This
recommendation still will not address the issue of combination of materials
such as those used to formulate GPA. Emergency response advice to address
reactive materials needs better clarification and perhaps research.
. The last bullet -
OSHA and EPA should consider whether additional guidance or outreach is
needed for users to understand the limitations of MSDSs and industry awareness
that more than the MSDS is needed to conduct full process hazards analyses.
The internet might be an appropriate outreach tool to reach small businesses.
If OSHA developed a skeleton for a process hazard evaluation and showed how
and how not to use MSDSs, this might be a good public service. In fact a
training module might help those companies with limited travel budgets to train
their staff and keep current.
Additional Comments
Application of the some specifics of this report may not have wide
application. However, the general hazard assessment process and the
excursion from normal operations have wide application. If definitions of norm,
what to do when the processes varies from the norm and how to handle an emergency
with the process could be incorporated as standard operating procedure with
hazardous reactions, then we could reduce the severity and probably the
number of incidents. (This assumes a strong worker training component to the
management function.) This has wide applications to all industry involved
with hazardous materials handling.
-------
Page 33 - NAPP
The report did not dwell on the community very much, other than to give a
general description of the surrounding area. The other area that should be
expanded is the internal and external emergency responders and their
preparedness.
In incidents such as this, the federal and even state governments play a
preventative role in terms of regulations and guidance to help industry
than an immediate response role. The governments can offer advice at the time
of the incident, but only the local responders can be on site quickly enough
to assist in these incidents. The governments can help to train and equip
these responders, but in the critical first hour, there is little a remote
government person can do other than to offer advice. This report provides a good model
for future investigations. The report develops the information rationally and is easy to follow.
While each incident will differ, the overall outline should remain relatively consistent.
The recommendations identified have implications for industries beyond the formulating industry,
and the lessons learned from this incident should help other companies prevent problems, or at
least minimize problems, from reactive materials.
-------
Page 34 - NAPP
Comments of Jerry Scannell, National Safety Council
Peer Review
The EPA/OSHA Joint Chemical Accident Investigation Report
Of Napp Technologies, Inc.
Summary: The joint teams did a good job under very difficult conditions of what was left after
this destructive deflagration. Although the EPA/OSHA Joint Chemical Accident Investigation
Report explains the event that occurred at the Napp Technologies, Inc. facility on April 21,1995,
it should have examined further why the event occurred and what can be-done to avoid these
occurrences in the future. The report should have iioted any environmental consequences of the
explosion, any failures in facility's management sysfstns tc prevent, mitigate and respond to the
event, and any regulatory gaps and inconsistencies.
GENERAL
• The report was issued over two years after the incident This causes it to lose any impact
it might otherwise have! Two years is unacceptable.
• The investigation's methodology should be further defined to include its objectives and
review parameters. In addition, the investigative team should be individually identified
to include area of evpertise and organizational affiliation.
* The investigation did not indicate that its analysis was peered review or that it solicited a
review from stakeholders - other than Napp Technologies, Inc. officials. A broader
review may have added information and other perspectives on the accident analysis. The
report should indicate whether the team interviewed all stakeholders - company officials,
labor, state and community officials, and citizens.
• The report did not review Napp Technologies, Inc.'s compliance with applicable state and
Federal regulations governing the safe harAing of hazardous cher-iicals, worker safety
and public safety. Nor, did tie report review the company's accident prevention and
emergency response procedures against industry standards and best practices.
Nspleei.wpd
-------
Page35-NAPP
• The report states that the ultimate goal of the accident investigation is "...to determine the
root cause in order to reduce the likelihood of recurrence...." etc. And the report goes on
to recognize that examples of root causes include "...failure of particular management
systems, that allow faulty design, inadequate training or deficiencies in maintenance to
exist." The report, however, focuses on the faulty design, the inadequate training and the
deficiencies in maintenance but the treatment of the failure of management systems that
allow these deficiencies to exist is handled poorly at best.
Management System Issues
• Management systw^i issues that need to be developed further include:
* The qualifications, credentials and competence of the managers involved
in the decision-making, from New Product Review to the emergency
response. This review is especially important in light of the series of
apparent bad suasions by management ss the
-------
Page 36- NAPP
*More discussion on what Technic knew. Did they have more
information that could have been given to Napp to better equip Napp to
make an informed decision about the hazards of the process?
*While the report recognizes the inadequacy of the "hazard analysis", it
does a poor job of discussing the underlying and contributory causes of
this inadequacy.
EMERGENCY PREPAREDNESS
• There is a much greater need for an in-depth discussion of emergency preparedness. The
report focuses on training deficiencies. But there is more to it than that How could an
emergency plan be so inadequate as to not include what action to take during deviations
from normal operations? How could deviations exist for almost 16 hours with no one on
site with the authority or willingness to safely shutdown the operation? In light of the
emergency, how could a decision be made not to notify the local fire department and
instead send employees back in to dump the batch at that stage of the crisis?
• There was no discussion of whether Napp Technologies, Inc. had an emergency plan. If
Napp did have an emergency plan, was staff trained to implement it? Did they have a
history of table top and roll-scale exercises to test and refine the plan? Did Napp have an
operational relationship with local emergency responders? There was no discussion of an
on-site incident command structure.
• The report did not review whether Napp was in compliance with the Emergency Planning
and Community Right To Know Act particularly in reference to Section 304, Emergency
Release Notification, and Sections 31 i-312, Hazardous Chemical Reporting.
• . Finally, me reported not discuss Napp Technologies emergency spill response plan.
REGULATORY REVIEW
• A review of the regulatory safety net should have been undertaken to ensure that no gaps
existed between PSM, Haz Comm, HazWoper, EPCRA and RMP.
' If as the report says, the root causes and the contributing factors "should be considered
lessons for the chemical processing industries which operate similar processes, especially
the tolling industry" then the recommendations fall far short of having the impact
necessary to minimize the likely occurrence of a similar incident in the future.
Specifically, the recommendations fail in the following areas:
* In those areas such as emergency preparedness, management
competencies and regulatory oversight, where the discussion itself was
•\ Napteel.wpd
-------
Page 37- NAPP
inadequate, the report was almost devoid of any meaningful
recommendations.
"'One of the major recommendations is essentially that companies should
comply with existing regulations such as PSM and RMP. What does this
change? Do companies not know about the rules? Don't they care?
*One recommendation is for the industry and/or government to develop
guidelines to be used in tolling contracts. This is good but needs to be
developed stronger. For instance, can OSHA and EPA take a ksd role in
serving as a catalyst to get the industry together to work on such
guidelines? If voluntary action doesn't work, is regulatory action
necessary?
*The report makes a number of recommendations for OSHA and EPA. It
is not clear what actually was done in response to those recommendations
since the memos and directives referred to in the report are not part of the
appendices. But regulatory action on the part of these agencies doesn't
seem to be enough. The report should have considered extensive outreach
activities to reach the regulated industry as part of the recommended.
actions.
EXTERNAL REVIEWS
• Finally, it is understood that a draft of this report was shared with the company before it
was finalized. That being so, then the report should have been shared with all
stakeholders.
September 24,1998
Napleel.ivpd
-------
Page38-NAPP
Comments of Jerry Scannell, National Safety Council
Peer Review The EPA/OSHA Joint Chemical Accident Investigation Report Of Napp
Technologies, Inc.
Summary: Although the EPAJOSHA Joint Chemical Accident Investigation Report explains the
event that occurred at the Napp Technologies facility on April 21, 1995, ft should have explained
further why the event occurred and what can be done to avoid these occurrences in the future. The
report should have noted any environmental consequences of the explosion, any in facility's
management Systems to prevent, mitigate and respond to the event, and any regulatory gaps and
inconsistencies.
GENERAL
The report was issued over two years after the Incident. This causes it to lose any impact it
might otherwise have! Two years is unacceptable.
The investigation methodology should be further defined to include its objectives and
review parameters. In addition, the investigative team should be individually identified to include
area of expertise and organizational affiliation.
The investigation did not indicate that its analysis was peered review or that it solicited a
review from stakeholders - other than Napp Technologies, Inc. officials. A broader review may
have added information and other perspectives on the accident analysis. The report should indicate
whether the team interviewed all stakeholders - company officials
labor, state and community officials, and citizens.
The report did not review Napp Technologies, Inc.'s compliance with applicable state and
Federal regulations governing the safe handling of hazardous chemicals, worker safety and public
safety. Nor, did the report review the company's accident prevention and emergency response
procedures against industry standards and best practices.
The report states that the ultimate goal of the accident investigation is "...to determine the
root cause in order to reduce the likelihood of recurrence...." etc. And the report goes on to
recognize that examples of root causes include "..failure of particular management
systems, that allow faulty design, inadequate training or deficiencies in maintenance to exist." The
report, however, focuses on the faulty design, the inadequate training and the deficiencies in
maintenance but the treatment of the failure of management Systems that allow these deficiencies
to exist is handled poorly at best.
The report did not recommend follow-up actions to share information resulting from the
investigation among industry, government, and other concerned groups and citizens. There are no
outreach recommendations that could result in revised engineering and management standards and
best practices to improve safety In this industry.
-------
Page 39 - NAPP
The report did not review The role that state and Federal regulatory and safety agencies
played in the incident, nor whether there were gaps and deficiencies in the regulations or in their
implementation.
Management System Issues
Management system issues that need to be developed further include:
* The qualifications, credentials and competence of the managers involved in the
decision-making, from New Product Review to the emergency response. This review is especially
important in light of the series of apparent bad decisions by management as the crisis developed.
* The existence,, effectiveness, and/or the results of the Company's audit program,
safety and health program and/or the safety committee including any finding of noncompliance
with SOPs and any recommendations for improvement.
* The role employees had or failed to have in the company's safety and health program.
* A discussion of the existence of any review of prior incidents and Federal: by the
company, in interviews with employees, or as part of the compliance investigations. Was there
anything that could be learned from any prior incidents that could have alerted management to
deficiencies in their SOPs or systems?
The report recognizes that contributing factors "facilitated the occurrence of the event." However
the report falls short in adequately dealing with the contributing factors. Contributing factors that
needed to be developed include:
No discussion whatsoever on the level of regulatory oversight, neither specific to this location
or the industry as a whole or to the type of operation-tolling.
No discussion or the violations or level of compliance found in the enforcement investigations
including company history or previous regulatory history. The same deficiency applies to the
industry as a whole,
More discussion on what Technic knew. Did they have more information that could have been
given to Napp to better equip Napp to make an informed decision about the hazards of the
process?
While the report recognizes the inadequacy of the "hazard analysis", it does a poor job of
discussing the underlying and contributory causes of this inadequacy.
-------
Page 40 - NAPP
EMERGENCY PREPAREDNESS
There is a much greater need for an in depth discussion of emergency preparedness. The
report focuses on training deficiencies. But there is more to it than that. How could an emergency
plan be so inadequate as to not include what action to take during deviations from normal
operations? How could deviations exist for almost 16 hours with no one on site with the authority
or willingness to safely shutdown the Operation? In light of the emergency, how could a decision
be made not to notify the local fire department and instead send employees back in to dump the
batch at that stage of the crisis?
There was no discussion of whether Napp Technologies, Inc. had an emergency plan. If
Napp did have an emergency plan, was staff trained to implement it? Did they have a history of
table top and full-scale exercises to test and refine the plan? Did Napp have an operational
relationship with local emergency responders? There was no discussion of an on-site incident
command structure.
The report did not review whether Napp was in compliance with the Emergency Planning
and Community Right To Know Act, particularly in reference to Section 304, Emergency Release
Notification, and Sections 311-312, Hazardous Chemical Reporting.
Finally, the report did not discuss Napp Technologies emergency spill response plan.
REGULATORY REVIEW
A review of the regulatory safety net should have been undertaken to ensure that no gaps
existed between PSM, Haz Comm, HazWoper, EPCRA and RMP.
If as the report says, the root causes and the contributing factors 'should be considered
lessons for the chemical processing industries which operate similar processes, especially
the tolling industry' then the recommendations fall far short of having the impact necessary to
minimize the likely occurrence of a similar incident in the future. Specifically, the recommendations
fail in the following areas:
* In those areas such as emergency preparedness, management competencies and regulatory
oversight, where the discussion itself was inadequate. The report was almost devoid of any
meaningful recommendations.
*One of the major recommendations is essentially that companies should comply with existing
regulations such as PSM and RMP. What does this change? Do companies not know about the
rules? Don't they care? Aren't they worried about the consequences of noncompliance? Or don't
they think they will get caught? Essentially the question is why did this company ignore the law
and what should be done to reduce the likelihood of companies ignoring the law in the future?
-------
Page 41 -NAPP
*One recommendation is for the industry and/or government to develop guidelines to be used in
tolling contracts. This is good but needs to be developed further. For instance, can OSHA and
EPA take a lead role in serving as a catalyst to get the industry together to work on such
guidelines? If voluntary action doesn't work, is regulatory action necessary?
*The report makes a number of recommendations for OSHA and EPA. It is not clear what
actually was done in response to those recommendations since the memos and directives referred
to in the report are not part of the appendices. But regulatory action on the part of these agencies
doesn't seem to be enough. The report should have considered extensive outreach activities to
reach the regulated industries part of the recommended actions.
EXTERNAL REVIEWS
* Finally, it is understood tat a draft of this report was staved with the company before it was
finalized. At best, this will undermine the credibility of the findings with stakeholders and at worse,
it could lead to avoidance of recommendations that could prevent similar catastrophes in the
future.
-------
Page 42 - NAPP
Review of
EPA/OSHA Joint Chemical Accident Investigation Report
Napp Technologies Inc., Lodi NJ
(Including points developed at the Sept. 14,1998 review panel meeting)
by Wade A. Freeman
September, 1998
This review comments on the completeness, technical soundness, and overall
approach of the report on the investigation of the April 21, 1995 explosion at the
Napp Technologies facility.
Comment on the Overall Approach
The Joint Chemical Accident Investigation Team (JCAIT) assembled
background information and gathered testimony to create a chronological
description of actions and events preceding the accident. The team evaluated
this record, physical evidence at the scene, documents describing the equipment
and chemicals in use, and descriptions of similar equipment located elsewhere
to arrive at a list of Significant Facts in the accident. JCAIT then listed
possible causes of the accident and used "engineering analyses of this
information...and professional judgement" to determine root causes and
contributing factors. This led to a set of recommendations.
This overall approach is sound. However, some possible and even plausible
causes for the unwanted chemical reactions are not explicitly considered in the
report. JCAIT should have gathered data relative to such possibilities and
sought to rule them out. The analyses described in the report do not firmly
establish the chemistry of the accident. Discussion at the Sept. 14 meeting
revealed that some of chemical analyses were selected adventitiously and not as
part of a fully conceived program of analysis in support of the investigation.
Finally, some evidence and testimony are insufficiently discussed.
Were Analyses Sufficient?
Chemical analyses of the residues left by an explosion can reveal important
details of the reactions that took place. JCAIT discusses their chemical
-------
Page 43 - NAPP
analyses in Section 3.1 (page 18) and details them in Appendix A (page 31) of
the report.
1. Insufficient detail is provided on the chemical analyses. JCAIT fails to
state or reference the analytical methods that were used. No account is given of
the selection of sampling sites, the number of samples taken, or the number of
samples analyzed.
2. Analytical results are given in non-numerical terms (such as "percentage
amounts" or "large amounts"). Numerical findings should appear in Appendix A.
3. The analyses were poorly selected. Mere elemental analysis of the residues is
unhelpful (as the report notes). A program of qualitative and quantitative analysis
for a range of inorganic compounds should have been conducted. Finding specific
substances or classes of substances in the residues would allow conclusions to be
drawn about the chemical changes within the blender over the course of the
accident. For example, finding residual elemental sulfur would support the
reaction scenario laid out in Appendix B. Detection of sulfide sulfur would show
that reactions took place other than those discussed in the report. As it is, the
report makes no mention at all of sulfur in the residues.
4. Analyses for sulfur and sulfur-containing compounds should have been
carried out.
5. Three organic compounds were identified "in large amounts" in internal and
external residues of the explosion: phenol, 2-methylphenol, and 4-methylphenol.
The report concludes that these compounds probably derived from the insulation
that lined the blender but could also have derived from the benzaldehyde that was
added to the mix. Appendix A proposes a route to phenol and the two
methylphenols starting with benzaldehyde. This portion of Appendix A should
have been omitted. It adds nothing to the conclusions of the report and is
chemically improbable, as the following comments show:
a) Other passages in the report indicate that benzaldehyde was
never in the blender. Page 20 includes, as part of a Significant Fact, the
statement that "operators were unable to inject benzaldehyde, the sole liquid
component of GPA, into the blender." Page 24 has the sentence: "However,
given that operators were not able to inject the benzaldehyde into the blender it
-------
Page 44 - NAPP
is unknown if any water in the feed line actually entered the blender."
Obviously, if benzaldehyde never got into the blender, the phenol compounds
did not derive from it. At the Sept. 14 meeting, investigators stated finally that
was impossible to conclude from the available evidence whether any
benzaldehyde actually got into the blender. Even if a small amount of
benzaldehyde made it into the blender, it would not explain the detection of
these organic compounds "in large amounts."
b) The proposed route to the phenol compounds is inconsistent with
known chemistry. Toluene would not be converted to phenol under the reducing
conditions in the blender.1 The report seems to call this conversion, which is an
oxidation, a "classic electrophilic aromatic substitution." It is not. It may be
that the report refers to the methylation of phenol as the electrophilic aromatic
substitution. Such a methylation would require acidic conditions, a methylating
agent and phenol.2 All three were absent in the blender.
Aqueous sodium hydrosulfite reduces benzaldehyde to benzyl alcohol
(C6H5CH2OH) in a two-electron reduction.3 This reaction is also plausible under
the conditions in the blender. Benzyl alcohol is presumably the "methyl
hydroxy (alcohol) intermediate" mentioned in the report, although benzyl
alcohol (and benzaldehyde) contain no methyl groups. Four-electron reduction
of benzaldehyde to toluene is also conceivable:
C6H5COH + 4 H+ + 4e —> C6H5CH3 + HOH
However, both reductions require H+ in addition to the electrons supplied by
the reducing agent. The H+ would have to come from impurities containing
active hydrogen (such as water or benzoic acid) because the nominal contents of
the blender furnish no hydrogen.
c) The chemical composition of the "rigid foam material" used to insulate
the walls of the blender and any additives in the aqueous coolant solution should
have appeared in the report. During the Sept. 14, 1998 meeting, members of
JCAIT stated that the insulation of the blender was polyurethane foam. Most
polyurethane foam is made by reacting 2,4-diisocyano-l-methylbenzene with a
1 Roberts, JohnD., and Caserio, Marjorie C., Basic Principles of Organic Chemistry, 2nd Edition, W. A.
Benjamin, Inc., 1977, page 405-10.
2 See the discussion of Friedel-Crafts alkylation in any organic chemistry text. For example, Roberts,
John D., and Caserio, Marjorie C., Basic Principles of Organic Chemistry, 2nd Edition, W. A. Benjamin,
Inc., 1977, page 1047.
3 deVries and Kellogg, J. Org. Chem. 45, 4126, 1980.
-------
Page 45 - NAPP
suitable dialcohol and then adding water.4 Oxidation of a functionalized
methylbenzene in a fire is consistent with the generation of phenol and
methylphenols.
d) The speculation concerning the source of the phenol compounds is
misplaced. Appendix A should detail experimental and observational results.
Interpretation properly belongs in the body of the report.
6. The first paragraph describing the analysis of the blender (in Appendix A)
states: "The damage initially appeared to be the result of a steam explosion inside
the water jacket lining." This statement leaves it uncertain whether a steam
explosion was excluded and, if it was, why. At the Sept. 14 review meeting,
members of JCAIT explained that they rejected the steam explosion scenario
because the portion of the blender that sustained the provocative damage housed
insulation, not coolant.
7. Analysis of the physical condition of the bodies of the victims is the sole basis
used to classify this event as a deflagration ("soft explosion") rather than as an
explosion or detonation. The analysis of the remains of the blender should have
been extended with a view to confirming or denying this conclusion.
8. JCAIT reports (page 17) that a USEPA mobile laboratory of "downwind air
samples of inorganic/acid gases, organic, and ketones" as part of the emergency
response. No other mention is made of these samples. Does this sentence mean
that acidic gases, organic compounds and ketones were in fact found in the
samples? What analyses were performed on the mobile laboratory samples? What
were the results? At the Sept. 14 meeting, members of JCAIT explained that this
sampling was for population protection, gave negative results, and had no
significance to the investigation. These facts should have been in the report.
9. The report states (page 23) that JCAIT conducted "metallurgical analysis of
the blender after the accident." This is somewhat misleading. Appendix A details
a thorough visual examination of the blender and a single microscopic
measurement (to obtain the depth of the grooves in the graphite seal).
4 Roberts, John D., and Caserio, Marjorie C., Basic Principles of Organic Chemistry, 2nd Edition, W. A.
Benjamin, Inc., 1977, page 1455.
-------
Page 46 - NAPP
10. An attempt should have been made to establish the conditions of
temperature and humidity prevailing in the blender room over the course of
the operation.
11. Members of JCAIT stated during the review meeting of Sept. 14 that the
ARC analyses discussed in Appendix B were performed not by design, but by
happenstance when professional acquaintances offered to do them. Taking
advantage of such opportunities is certainly acceptable. However, it is obviously
unsound to rely on such circumstances.
Are all sources of information properly identified?
Reports of this type should adhere to standard practices of attribution. JCAIT is
erratic in this respect. The reference on page 39 to "EPA Trip Report, July 5,
1995" does not appear in Appendix D. Did the report originate with EPA members
of JCAIT or with other representatives of the EPA? The in-text details about
Tartani and Contessa's paper on page 40 mostly duplicate the citation in Appendix
D. A flash point for powdered aluminum/air mixtures is quoted without
attribution. The NIST report on the remains of the blender is not properly cited in
Appendix A. During the Sept. 14 meeting, it was stated by an investigator that
discussion in Appendix A concerning the conversion of benzaldehyde to phenol and
the two isomeric methylphenols was a personal communication from a retired
chemist. Relying on such sources is inferior to checking facts in standard
references. The "Events and Causal Factors and Hazard-Barrier-Target
techniques" that are mentioned on page 26 as part of the engineering analysis of
the event require a reference.
Were the sources of reaction initiation plausible?
JCAIT identifies two "most likely" sources of initiation: accidental wetting of the
blend and frictional heating from over-use of the intensifier bar. Both are
plausible. It is essentially certain (see below) that water was reacting in the
blender during the time preceding the explosion. Two very likely sources of water
are identified: leakage past the intensifier bar seal and residual water in the liquid
feed line. These are reasonable possibilities. Both might have contributed
concurrently to wet the blend.
-------
Page 47 - NAPP
Were all possible sources of initiation identified? Were the likely causes of the
chemical reaction explored fully?
The most likely causes of the chemical reaction were identified and explored.
Other possible causes were not explored sufficiently. The approach should have
been to consider all possibilities at first and obtain testimony, analysis and other
evidence to rule out as many as possible.
1. The report does not deal effectively with the possibility that wrong
ingredients or the contamination of ingredients contributed to the chemical events
in the blender. A few sentences suggest that the chance of inadvertent substitution
or contamination entered the deliberations of JCAIT. Page 34 states that the
benzaldehyde chemistry that the report has just elaborated "... tends to eliminate
the possibility that phenol, rather than benzaldehyde, had been inadvertently
added...." Page 8 notes the fact that benzaldehyde is oxidized to benzoic acid
when exposed to the air and inserts some descriptive chemistry of benzoic acid.
Page 23 states (correctly) that moisture present in any of the raw materials could
have sufficed to initiate a reaction. Page 31 mentions "inadvertent mixing of
different chemicals that could...occur."
Elsewhere however, JCAIT accepts the quality of the raw materials without
proof. Page 3 states: "The 1995 blending ingredients were virtually the same as in
1992." This assertion requires analytical confirmation. (Incidentally, if
"virtually" means "very nearly," then in what ways did the ingredients in the 1995
disaster differ from the ingredients in the 1992 success?) Any details that support
the "virtually" belong in the report. Page 23 mentions a quality assurance check
that Napp performed on the raw materials that did not find moisture. Details
should appear in the report. JCAIT apparently accepts the lack of apparent
reaction during the loading of the blender to rule out the presence of moisture in
the raw materials. This overlooks possible delayed onset of reaction, a common
occurrence.
Residual portions of the ingredients should have been collected (from the
bottoms of supply drums, for example) and analyzed. These drums were present in
the blending room at 7 p.m. (page 13). At the Sept. 14 review meeting, it was
established the fire that followed the blast destroyed all the supply drums. The
report should have mentioned this and any other adverse circumstances. In the
-------
Page 48 - NAPP
absence of residual ingredients, JCAIT could have attempted to confirm the
chemical identity of the materials loaded into the blender inferentially. This would
involve checking the source of the materials, conducting analyses of materials from
the same production lot, and making inquiry into conditions of transport, storage,
and handling.
2. The report does not consider the possible influence of "normal"
impurities.
a) According to Kirk-Othmer5, the highest grade of industrial anhydrous
sodium hydrosulfite contains 88% Na2S2O4 by mass mixed with 3% sodium
disulfite (Na2S2O5), 3% sodium sulfite (Na2SO3), 3% sodium sulfate
(Na2SO4), and 3% sodium carbonate (Na2CO3). A lower grade of sodium
hydrosulfite contains only 80 percent Na2S2O4 by mass. According to the
same source, anhydrous sodium hydrosulfite is produced by four methods:
formate reduction, amalgam reduction, zinc reduction and electrolytic
reduction. Each naturally leaves a different set of impurities. It would have
been informative to find out whether the GPA components in the April 1995
accident were of the same grade and produced by the same reactions as
those that were successfully blended in July 1992.
b) Depending on the way in which anhydrous potassium carbonate (K2CO3) is
prepared, it contains as much as 3% water by mass.6 Potassium carbonate
is hygroscopic; its recommended mode of storage is in bunkers ventilated
with dry air.7 The hydrate K2CO31.5H2O, which contains about 16%
water by mass and deliquesces in moist air, is readily available in
commerce as dustless crystals. Conceivably, water associated with the
potassium carbonate initiated the reaction events in the blender. This
possibility should certainly have been investigated, as members of JCAIT
agreed at the Sept. 14 review meeting.
3. Particle size and shape can affect the progress of dry blending operations.
JCAIT should have checked the state of subdivision of the materials in the
5 Kirk-Othmer Encyclopedia of Chemical Technology, 4th edition, John Wiley & Sons, New York, 1998,
Vol.
6 Ullman 's Encyclopedia of Industrial Chemistry" 5th edition, VCH Weinheim, Germany, 1993, Vol.
A22, page 99.
7 Ullman's Encyclopedia of Industrial Chemistry" 5th edition, VCH Weinheim, Germany, 1993, Vol.
A22, page 99.
-------
Page 49 - NAPP
accident to that of the materials that were successfully blended in 1992. This goes
as well to the issue of undue heating from the turning of the intensifier bar.
4. With a view to confirming the proposed reaction scenario, JCAIT should
have sought samples of authentic GPA and run experiments in which varying
amounts of water are added under the conditions prevailing in the blender. These
experiments would resemble the experiments described in Appendix B, but would
aim to identify the products as well as to measure the temperature rise.
5. JCAIT failed to consider some clues to the reactions taking place in the
blender. An employee who entered the blending room at 7 p.m. reported a smell
of "rotten eggs" (page 13). An employee who entered the blending room at 10
p.m. noticed a "dead animal" smell. Employees arriving for work the next
morning also reported a rotten-egg odor (page 14) that ".. .had escaped the
building and was noticeable in the parking lot...." The witnesses are reporting
the presence of hydrogen sulfide (H2S). The presence of hydrogen sulfide was
specifically indicated by testimony (quoted by a JCAIT member at the Sept. 14,
1998 review meeting) from another witness who named the smell as hydrogen
sulfide. It is worth nothing that H2S deadens the sense of smell,8 a fact that accords
with the prevalence of rotten-egg reports among newcomers to the scene. JCAIT
focuses on the generation of sulfur dioxide from sodium hydrosulfite (page 37).
Sulfur dioxide has a characteristic choking or suffocating odor that is never
compared to rotten eggs. It is a serious error to write off the odor of H2S as a
generic "sulfur smell" (page 28).
Sulfide sulfur (sulfur in the -2 oxidation state) would form if aluminum
reduced sodium hydrosulfite fully. Thus the reaction
10 Al + 3 Na2S2O4 —> 3 Na2S + 4 A12O3 + A12S3
might accompany or replace the second reaction on page 39. Reduction to S(-2) is
quite plausible. Such a reduction would be exothermic. The conversion of sulfides
to H2S, which boils at -60.7°C, requires a source of H+. Hence, detection of H2S at
7 p.m. indicates that a substance with active hydrogen (such as water or benzoic
acid) was in the blender by that time; the rotten-egg odor rules out the "friction-
only" scenario at the bottom of page 25 of the report. Despite the overall basic
8 Lewis, Richard J., Hazardous Chemicals Desk Reference, Van Nostrand Reinhold, New York, 1993,
page 691-2.
-------
Page 50 - NAPP
conditions in the blender, local concentrations of H+ donors could easily generate
H2S, which would out-gas rapidly because of its high volatility.
The emission of H2S does not prevent simultaneous or subsequent generation
of gaseous sulfur dioxide (SO2) according to the equations on page 37. Indeed, the
puffs of white smoke coming from the blender at 5:30 a.m. might well have been
an acid mist formed as vented SO2 reacted with moisture in the air; H2S would not
form such a mist.
6. JCAIT should have checked Napp's records to ascertain the contents of the
liquid feed system in its last prior use. Residual content might account for the
"vanilla-like odor" detected in the tank when operators prepared to add the
benzaldehyde (page 12). This odor is a loose end in the report. It is (remotely)
conceivable that the material with the vanilla-like odor entered the blender and
influenced the chemistry within. At the Sept. 14 review meeting, members of
JCAIT stated that these production records were destroyed in the accident (if they
ever existed). This point should have been included in the report.
Comment on the Discussion of Root Causes and Contributing Factors
The report does not satisfactorily exclude the possibility that one (or more) of the
raw materials originally contained water or another initiating substance or became
contaminated with water or such a substance during transportation and storage.
This point gains importance because it is known that one of the bags of potassium
carbonate had been broken open and taped over (page 10).
Reactions in the blender could have been taking place at several hot spots, of which
only one was observed. Reactions could also have been taking place throughout
the batch but with particular intensity at the observed hot spot. Therefore, the
assertion (on page 23) "...the bubbling noted towards the middle of the blender
reveals that the reactions did not take place at the walls of the blender..." is
logically faulty.
Page 10 of the report establishes that the operators knew that they were processing
water-reactive chemicals. Page 12 states that the operators found water in an
internal filter on the liquid feed line and adds: "The operators did not consider
the liquid feed line to be functioning properly. The liquid spray head and spray
system had not been completely dried prior to the charging of the blender." This
was clearly on-the-spot opinion because the operators proceeded next to attempt to
-------
Page 51 -NAPP
dry the liquid feed system by rinsing it with isopropyl alcohol and blowing nitrogen
through it. Page 24 in fact implies that other attempts took place ("...several
drying/vacuum procedures were performed"). JCAIT should have attempted to
learn what made the operators think the liquid feed system was dry enough to
permit injection of the benzaldehyde. An error in judgment on this point was
probably the proximate cause of the accident. It is not clear in the report whether
the personnel who made this judgment survived the explosion.
JCAIT should have included all it knew of the events of the last 47 minutes before
the explosion. A video describing the accident was shown at the Sept. 14 meeting.
It reported (and members of JCAIT confirmed) that during this climactic period,
the operators readied receiving drums under a nitrogen blanket and managed to
off-load four drum-fulls of material. At this point the blender before it made a
loud noise, which caused the operators to retreat from the room. The explosion
occurred after they had returned to the room and as they unloaded a fifth drum-
full of material. These facts belong in the written report.
On page 23, the report states that information was received to indicate that
operators might have used water or steam to unclog the liquid feed line. The
report then immediately states that JCAIT was able to confirm that attempts to
clear the feed line did not involve water or steam. At the Sept. 14 meeting
members of JCAIT further confirmed that investigation showed the allegation
concerning the use of water was mistaken. The sources and resolution of the
conflicting testimony should appear in more detail in the written report.
On page 28, JCAIT concludes "... there is no evidence to suggest that Napp was
aware that off-loading the blender may have exacerbated the reaction mechanisms
by exposing the contents to air or that the contents could violently erupt and
deflagrate." This is at odds with the reported prolonged efforts by
Napp to protect the blend from the air by use of a nitrogen blanket. Also,
exposure to the air would not aggravate the reactions already proceeding, but
would (and did) occasion a new set of reactions, namely, air oxidations.
"The training of the fire brigade and emergency responders was inadequate"
cannot be sustained as contributory to the accident. The investigation develops a
picture of firefighters standing ready to charge hoses and direct water on
-------
Page 52 - NAPP
command. What could they have done otherwise, no matter what their training?
JCAIT does not mention training records or other background on emergency
responders who were not members of the fire brigade. Concluding that the
training of these people was inadequate is not justified without information about
their training.
Comment on Recommendations
JCAIT makes helpful and appropriate recommendations. The recommendation
against the use of liquids to cool or purge seals in processing equipment if the
liquids are chemically incompatible with the materials being processed is
particularly important. It should appear in full in the Executive Summary. The
version "... ensure that equipment manufacturers' recommendations for proper
use of equipment are followed..." is vague.
Comments on Appendices and Illustrations
ILLUSTRATIONS
Figure 3A (page 14) shows neither the route along which the benzaldehyde was
intended to flow nor the route by which it ended up in the vacuum separator bowl.
A proper schematic diagram of vacuum collection system would help the reader
far more. A member of JCAIT at the Sept. 14 review meeting sketched an
adequate figure.
APPENDIX A
As noted in the preceding, detail on the nature of chemical analyses and their
results are lacking.
APPENDIX B
Equation 3 (showing the disproportionation of sodium disulfite) has a
typographical error. It should read 2Na2S2O5 —> 2 Na2SO4 + SO2 + S.
In the second paragraph, the formula of sodium thiosulfate is given as Na2S2O4.
-------
Page 53 - NAPP
The correct formula is Na2S2O3.
The calorimetry studies described on page 39-40 require additional detail:
a) A description of the ARC method, or a reference to a suitable background
paper should appear.
b) The sources and purity of the substances used in the studies should be
stated.
c) The amounts and the method of mixing of the reactants should be given
for each experiment.
d) The report should have established the products of the reactions when Al
was present. The observations confirm a redox reaction in which Al(0) is oxidized
and hydrosulfite reduced. Was S(-2) formed?
e) The report states that heats of reaction were determined in the ARC
experiments. If so, they should be given. These may be estimates of the number of
joules generated per gram of mixture in each experimental run. A "heat of
reaction" more usually refers to the enthalpy change associated with occurrence of
a single chemical reaction (as represented by a chemical equation). Without
knowing the products of a reaction, no chemical equation can be written and true
heats of reaction are unobtainable.
Page 39 states: "Benzyl alcohol is produced by the reaction of benzaldehyde with
sodium hydrosulfite." This is incorrect. Some source of H+ must also be present.
Page 39 states: "The reaction products expected are consistent with the results of
the chemical analysis of the site." This is technically true, but misleading. The
chemical analyses detailed in Appendix A establish only the elemental composition
of the inorganic residues and are "consistent" with any set of reactions that
includes compounds of sodium and potassium among their products.
The sentence on page 39: "The source of the large phenol concentration noted in the
grab samples from the blender does not seem to be a result of the reactions of the
reported mixture materials but most likely occurred at some time during initial
attempts to blend the GPA components." is self-contradictory. The generation of
phenol "during initial attempts to blend the GPA" would have to result from
"reactions of the reported mixture materials." That is, "GPA components" equal
"mixture materials." If the sentence means that the phenol arose from the GPA
-------
Page 54 - NAPP
components, it opposes the earlier conclusion (page 34) that "[pjhenol and the
methylphenol compounds were likely due to the insulating material remnants...."
Other Comments
Page 3. Potassium carbonate is not an alkali metal but rather a compound
of an alkali metal. Its physical and chemical properties differ sharply from those
of the alkali metals.
Page 8. A pyrophoric substance is one that takes fire on contact with air
under ordinary conditions, not "under appropriate conditions." Newspaper bursts
into fire spontaneously "under appropriate conditions" but newspaper is not
pyrophoric.
Page 8. The statement: "Benzaldehyde readily oxidizes to benzoic acid,"
should be replaced by "Benzaldehyde is readily oxidized to benzoic acid upon
exposure to the air."
Page 18. "A deflagration releases energy at a lower rate.. .and is less
destructive than a detonation." The following definitions from the literature would
clarify this discussion: "a deflagration is a soft explosion [in which] pressure are
relatively low.... Explosions involve pressures of several atmospheres....
Detonation is a severe form of explosion when pressures are much higher and are
propagated at a high rate (as much as several miles per second)."9
Page 20. The report states: "At 10:00 am on April 20, operators detected a
vanilla-like odor in the liquid feed tank...." This contradicts the Timeline of
Events exhibited on page 11 of the report, which sets the detection of the vanilla-
like odor at 12:30 p.m.
Page 22. The list of the most likely predominant reactions omits the
combustion of the hot sodium hydrosulfite, upon contact with the air.
Page 25. It is stated: "If a large amount of water was injected into the
material in the blender, the JCAIT believes a large hydrogen gas bubble would
have been formed, causing a detonation with greater energy th[a]n was released in
the accident." The belief needs justification. Presumably, the gaseous hydrogen
would result from reaction of the large amount of water with the powdered
aluminum. The rate of this reaction depends strongly in the temperature (see
report page 38). Did JCAIT estimate the temperature? How? Did JCAIT
estimate the amount of energy released in an accident? Hydrogen bubbles (as in
9 Mahn, W. J., Academic Laboratory Chemical Hazards Guidebook, Van Nostrand Reinhold, New York,
1991, page 7.
-------
Page 55 - NAPP
balloons) deflagrate when ignited in the air. If hydrogen forms, why must it
detonate?
RECOMMENDATIONS
Complete and release the report more promptly.
Include details of the analytical methods used in the course of an investigation
and at least some representative analytical findings in an appendix.
Integrate decisions about the type and extent of chemical analysis fully into the
investigations.
Take care to avoid loose ends. If observations or physical findings are judged
irrelevant, then the report should state as much, and tell why.
Use the chemical literature more aggressively to check facts.
Include full literature references in some uniform format.
Use a technical editor. This report is not very well written.
-------
Page 56 - NAPP
Review of
EPA/OSHA Joint Chemical Accident Investigation Report Napp Technologies Inc., Lodi NJ
by Wade A. Freeman August, 1998
This review comments on the completeness, technical soundness, and overall approach of the report
on the investigation of the April 21, 1995 explosion at the Napp Technologies facility.
Comment-on the Overall Approach
The Joint Chemical Accident Investigation Team (JCAIT) assembled background information and
gathered testimony to create a chronological description of actions and events preceding the accident.
The team evaluated this record, physical evidence at the scene, documents describing the equipment
and chemicals in use, and descriptions of similar equipment located elsewhere to arrive at a list of
Significant Facts in the accident. JCAIT then listed possible causes of the accident and used
_engineering analyses of this information+and professional judgement - to determine root causes and
contributing factors. This led to a set of recommendations.
This overall approach is sound. However, some causes of the onset of the unwanted chemical
reactions are not explicitly considered in the report.
JCAIT should have gathered data relative to such possibilities and sought to rule them out. The
analyses performed do not firmly establish the chemistry of the accident. Some evidence and
testimony are insufficiently discussed.
Were Analyses Sufficient?
Chemical analyses of the residues left by an explosion can reveal important details of the reactions
that took place. JCAIT discusses their chemical analyses in Section 3.1 (page 18) and details them in
Appendix A (page 31) of the report.
1. Insufficient detail is provided on the chemical analyses. JCAIT fails to state or reference the
analytical methods that were used. No account is given of the selection of sampling sites, the number
of samples taken, or the number of samples analyzed.
2. Analytical results are given in non-numerical terms (such as -Percentage amounts6 or large
amounts6). Numerical findings should appear in Appendix A.
3. The analyses were poorly selected. Mere elemental analysis of the
residues is unhelpful (as the report notes). A program of qualitative and quantitative analysis for a
range of inorganic compounds should have been conducted. Finding specific substances or classes
of substances in the residues would allow conclusions to be drawn about the chemical changes within
-------
Page 57 - NAPP
the blender over the course of the accident. For example, finding residual elemental sulfur would
support the reaction scenario laid out in Appendix B. Detection of sulfide sulfur would show that
reactions took place other than those discussed in the report. As it is, the report makes no mention at
all of sulfur in the residues.
4. Analyses for sulfur and sulfur-containing compounds should have been
carried out.
5. Three organic compounds were identified -in large amounts- in internal and external residues of
the explosion: phenol, 2-methylphenol, and
4-methylphenol. The report concludes that these compounds probably derived from the insulation
that lined the blender but could also have derived from the benzaldehyde that was added to the mix.
Appendix A proposes a route to phenol and the two methylphenols starting with benzaldehyde. The
following comments apply to this portion of Appendix A:
a) other passages in the report seem to indicate that benzaldehyde was never in the blender. Page 20
includes, as part of a Significant Fact, the statement that -operators were unable to inject
benzaldehyde, the sole liquid component of GPA, into the blender. - Page 24 has the sentence:
-However, given that operators were not able to inject the benzaldehyde into the blender it is
unknown if any water in the feed line actually entered the blender.
-obviously, if benzaldehyde never got into the blender, the phenol compounds
did not derive from it. Even if a small amount of benzaldehyde made it into the blender, it would not
explain the detection of these organic compounds _in large amounts.-
b) The proposed route to the phenol compounds is inconsistent with known chemistry. Toluene
would not be converted to phenol under the reducing conditions in the blender. The report seems to
call this conversion, which is an oxidation, a -classic electrophilic aromatic substitution.- It is not. It
may be that the report refers to the methylation of phenol as the electrophilic aromatic substitution.
Such a methylation would require acidic conditions, a methylating agent and phenol. All three were
absent in the blender.
Aqueous sodium hydrosulfite reduces benzaldehyde to benzyl alcohol (C6H5CH20H) in a
two-electron reduction, This reaction is also plausible under the conditions in the blender. Benzyl
alcohol is presumably the -methyl hydroxy (alcohol) intermediate - mentioned in Appendix A,
although benzyl alcohol (and benzaldehyde) contain no methyl groups. Four-electron reduction of
benzaldehyde to toluene is also conceivable:
C6HSCOH + 4 H++ 4e-—> C6H5CH3 + HOH
However, both.reductions require H+ in addition to the electrons supplied by the reducing agent. The
H+ would have to come from impurities containing active hydrogen (such as water or benzoic acid)
because the nominal components of the blender furnish no hydrogen.
-------
Page 58 - NAPP
c) The chemical composition of the-rigid foam material- used to insulate the walls of the blender and
any additives in the aqueous coolant solution should have been ascertained (for example, by asking
the manufacturer). This might rule out the insulation and coolant as a source of the phenol and
methylphenols.
d) The speculation concerning the source of the phenol compounds is misplaced. Appendix A should
detail experimental and observational results. Interpretation properly belongs in the body of the
report.
6. The first paragraph describing the analysis of the blender (in Appendix A) states: The damage
initially appeared to be the result of a steam explosion inside the water jacket lining. Did JCAIT later
retreat from this assessment? If so, why? Strong heating inside the blender conceivably boiled the
coolant and so caused the rupture of the outer jacket just before the explosion. In this sequence
coolant water might even have touched off the explosion. Such a series of events is not inconsistent
with the report (page 16) of three loud hissing noises and a -whoosh- sound preceding the explosion.
7. Analysis of the physical condition of the bodies of the victims is the sole basis used to classify this
event as a deflagration (_soft explosion - rather than as an explosion or detonation. The analysis of
the remains of the blender should have been extended with a view to confirming or denying this
conclusion.
B. JCAIT reports (page 17) that a part of the emergency response was acquisition by a USEPA
mobile laboratory of _downwind air samples of inorganic/acid gases, organic, and ketones.- No
other mention is made of these samples. Does this sentence mean that acidic gases, organic
compounds and ketones were in fact found in the samples? What analyses were performed on the
mobile laboratory samples? What were the results? Were the results evaluated with respect to the
chemistry of the explosion? Perhaps the mobile laboratory intended solely to check for toxic releases
and obtained negatives for deleterious compounds in their analyses. If so, the report should say so
explicitly.
9. The report states (page 23) that JCAIT conducted -metallurgical analysis of the blender after the
accident.- This is somewhat misleading. Appendix A details a thorough visual examination of the
blender and a single microscopic measurement (to obtain the depth of the grooves in the graphite
seal).
10. An attempt should have been made to establish the conditions of temperature and humidity
prevailing in the blender room over the course of the operation. Are all sources of information
properly identified?
Reports of this type should adhere to standard practices of attribution. JCAIT is erratic in this respect.
The reference on page 39 to -EPA Trip Report, July 5, 1995- does not appear in
Appendix D. Did the report originate wi the EPA members of JCAIT or with other
representatives of the EPA? The in-text details about Tartani and Contessa_s paper on page 40
mostly duplicate the citation in Appendix D. A flash point for powdered aluminum/air mixtures
-------
Page 59 - NAPP
is quoted without attribution. The NIST report on the remains of the blender is not properly cited
in Appendix A. The -Events and Causal Factors and Hazard-Barrier-Target techniques
mentioned on page 26 as part of the engineering analysis of the event (on page 26) require a
reference.
Were the sources of reaction initiation plausible?
JCATT identifies two -most likely_ sources of initiation: accidental wetting of the blend and
frictional heating from over-use of the intensifier bar. Both are plausible, it is essentially certain
(see below) that water was reacting in the blender during the time preceding the explosion. Two
very likely sources of water are identified: leakage past the intensifier bar seal and residual water
in the liquid feed line. These are reasonable possibilities. Both might have contributed
concurrently to dampen the blend.
Were all possible sources of initiation identified? Were the likely causes of the chemical reaction
explored fully?
The most likely causes of the chemical reaction were identified and explored, other possible causes
were not explored sufficiently. The approach should have been to entertain all possibilities and
analyze the facts to rule out as I many as possible.
1. The report does not deal effectively with the possibility that wrong ingredients or the
contamination of ingredients contributed to the chemical events in the blender. A few sentences
suggest that the chance of inadvertent substitution or contamination entered the deliberations of
JCAIT. Page 34 states that the benzaldehyde chemistry that the report has just elaborated . . .
tends to eliminate the possibility that phenol, rather than benzaldehyde, had been inadvertently
added - Page 8 notes the fact that benzaldehyde is oxidized to benzoic acid when exposed
to the air and inserts some descriptive chemistry of benzoic acid. Page 23 states (correctly) that
moisture present in any of the raw materials could have sufficed to initiate a reaction. Page 31
mentions -inadvertent mixing of different chemicals that could+occur. Elsewhere however,
JCAIT accepts the quality of the raw materials without proof. Page 3 states: -The 1995 blending
ingredients were virtually the same as in 1992. This assertion requires analytical confirmation.
(Incidentally, if -virtually_ means -very nearly,- then in what ways did the ingredients in the
1995 disaster differ from the ingredients in the 1992 success?) Any details that support the
-virtually belong in the report. Page 23 mentions a quality assurance check that Napp
performed on the raw materials that did not find moisture. Details should appear in the report.
JCAIT apparently accepts the lack of apparent reaction during the loading of the blender to rule
out the presence of moisture in the raw materials. This overlooks possible delayed onset of
reaction, a common occurrence.
Residual portions of the ingredients should have been collected (from the bottoms of supply
drums, for example) and analyzed. These drums were present in the blending room at 7 p.m.
-------
Page 60 - NAPP
(page 13). If unmixed starting materials failed to survive the accident and clean-up, then the
report should say so. In the absence of residual ingredients, JCAIT could have attempted to
confirm the chemical identity of the materials loaded into the blender inferentially.
This would include checking the source of the materials, conducting analyses of materials from
the same production lot, and making inquiry into conditions of transport, storage, and handling.
2. The report does not consider the possible influence of-normal- impurities.
a) According to Kirk-othmer, the highest grade of industrial anhydrous sodium hydrosulfite c
ontains 889. Na2S204 by mass mixed with 39. sodium disulfite (Na2S205), 3%- sodium sulfite
(Na2SO3), 3t sodium sulfate (Na2SO4), and 3t sodium carbonate (Na2CO3). A lower grade of
sodium hydrosulfite contains only 80 percent Na2S204 by mass. According to the same source,
anhydrous sodium hydrosulfite is produced by four methods: formate reduction, amalgam
reduction, zinc reduction and electrolytic reduction. Each naturally leaves a different set of
impurities, it would have been informative to find out whether the GPA components in the April
1995 accident were of the same grade and produced by the same reactions as those that were
successfully blended in July 1992.
b) Depending on the way in which anhydrous potassium carbonate (K2CO3) is prepared, it cental.
ins as much as 3'@ water by mass. Potassium carbonate is hygroscopic; its recommended mode
of storage is in bunkers ventilated with dry air. The hydrate K2CO3.1.5H20, which contains
about 16t water by mass and deliquesces in moist air, is readily available in commerce as dustless
crystals. Conceivably, water associated with the potassium carbonate initiated the reaction events
in the blender.
3. Particle size and shape can affect the progress of dry blending operations. JCAIT should have
checked the state of subdivision of the materials in the accident to that of the materials that were
successfully blended in 1992. This goes to the issue of undue heating from the intensifier bar as
well.
4. With a view to confirming the proposed reaction scenario, JCAIT should have sought samples
of authentic GPA and run experiments in which varying amounts of water are added under the
conditions prevailing in the blender. These experiments would resemble the experiments
described in Appendix B, but would aim to identify the products as well as to measure the
temperature rise.
5. JCAIT failed to consider some clues to the reactions taking place in the blender. An employee
who entered the blending room at 7 p.m. reported a smell of-rotten eggs_ (page 13). An
employee who entered the blending room at 10 p.m. noticed a -dead animal- smell. Employees
arriving for work the next morning also reported a rotten-egg odor (page 14) that -. . had escaped
the building and was noticeable in the parking lot. . . .- it is a mistake to write off these odors as a
generic -sulfur smell6 (page 28). The witnesses are almost certainly reporting the presence of
hydrogen sulfide (H2S). The odor of H2S is universally compared to rotten eggs or other decayed
material (the rotting of eggs in fact generates hydrogen sulfide). Furthermore, H2S deadens the
-------
Page 61 -NAPP
sense of smell. This fact accords with the prevalence of rotten-egg reports among newcomers to
the scene. JCAIT focuses on the generation of sulfur dioxide from sodium hydrosulfite (page 37).
Sulfur dioxide has a characteristic choking or suffocating odor that is never compared to rotten
eggs. Sulfide sulfur (sulfur in the +2 oxidation state) would form if aluminum reduced sodium
hydrosulfite fully. Thus the reaction 10 Al + 3 Na2S204 — > 3 Na2S + 4 A1203 + A12S3
might accompany or replace the second reaction on page 39. Reduction to S(2) is quite pi
ausible. Such a reduction would be exothermic. The conversion of sulfides to H2S, which boils at
+60.7oC, requires a source of H+. Hence, detection of H2S at 7 p.m. indicates that a substance
with active hydrogen (such as water or benzoic acid) was in the blender by that time; the
rotten-egg odor rules out the -friction-only_ scenario at the bottom of page 25 of the report.
Despite the overall basic conditions in the blender, local concentrations of H+ donors could easily
generate H2S, which would out-gas rapidly because of its high volatility.
The emission of H2S does not prevent simultaneous or subsequent generation of gaseous sulfur
dioxide (SO2) according to the equations on page 37. Indeed, the puffs of white smoke coming
from the blender at 5:30 a.m. might well have been an acid mist formed as vented S02 reacted
with moisture in the air; H2S would not form such a mist.
6. JCAIT should have checked Napp_s records to ascertain the contents of the liquid feed system
in its last prior use. Residual content might account for the -vanilla-like odor- detected in the
tank when operators prepared to add the benzaldehyde (page 12). This odor is a loose end in the
report. It is (remotely) conceivable that the material with the vanilla-like odor entered the blender and
influenced the chemistry within.
Comment on the Discussion of Root Causes and Contributing Factors
The report does not satisfactorily exclude the possibility that one (or more) of the raw materials
originally contained water or another initiating
substance or became contaminated with water or such a substance during transportation and
storage. This point gains importance because it is known that one of the bags of potassium
carbonate had been broken open and taped over (page 10).
Reactions in the blender could have been taking place at several hot spots, of which only one was
observed. Reactions could also have been taking place throughout the batch but with particular
intensity at the observed hot spot. Therefore, the assertion (on page 23) +the bubbling noted towards
the middle of the blender reveals that the reactions did not take place at the walls of the blender+- is
logically faulty.
Page 10 of the report establishes that the operators knew that they were processing water-reactive
chemicals. Page 12 states that the operators found water in an internal filter on the liquid feed
line and adds: -The operators did not consider the liquid feed line to be functioning properly.
The liquid spray head and spray system had not been completely dried prior to the charging of
the blender.- This was clearly on-the-spot opinion because the operators proceeded next to
-------
Page 62 - NAPP
attempt to dry the liquid feed system by rinsing it with isopropyl alcohol and blowing nitrogen
through it. Page 24 in fact implies that other attempts took place (- +several drying/vacuum
procedures were performed-); these problems are not detailed elsewhere in the report. JCAIT
should clarify what made the operators think the liquid feed system was dry enough to permit
injection of the benzaldehyde. An error in judgment on this point might have been the proximate
cause of the accident.
JCAIT should have confronted the issue of the missing 47 minutes. A great deal might have
happened in the blender room between 7 a.m., when Napp employees reentered with the intention
of unloading the blender, and 7:47 a.m., when the blender exploded, was actual progress made in
removing the contents of the blender? If facts are not available, the report should say SO.
On page 23, the report states that information was received to indicate that operators might have used
water or steam to unclog the liquid feed line. The report then immediately states that JCAIT was able
to confirm that attempts to clear the feed line did not involve water or steam. The sources and
resolution of the conflicting testimony should be given in more detail. On page 28, JCAIT concludes
there is no evidence to suggest that Napp was aware that off-loading the blender may have
exacerbated the reaction mechanisms by exposing the contents to air or that the contents
could violently erupt and deflagrate. This is at odds with the reported prolonged efforts by
Napp to protect the blend from the air by use of a nitrogen blanket. Also, exposure to the air
would not aggravate the reactions already proceeding, but would (and did) occasion a new set of
reactions, namely, air oxidations.
-The training of the fire brigade and emergency responders was inadequate is difficult to sustain
as contributory to the accident. The picture is of the fire brigade standing ready to charge their
hoses and direct water on command.
What could they have done otherwise, no matter what their training? JCAIT does not report the
employee training records or the capabilities of the emergency responders who were not
members of the fire brigade. In the absence of such information the conclusion that the training
of this group was inadequate is not justified.
Comment on Recommendations
JCAIT makes helpful and appropriate recommendations. The recommendation against the use of
liquids to cool or purge seals in processing equipment if the liquids are chemically incompatible
with the materials being processed is particularly important, it should appear in full in the
Executive Summary. The version -... ensure that equipment manufacturers- recommendations
for proper use of equipment are followed. . .- is vague.
Comments on Appendices and Illustrations ILLUSTRATIONS
Figure 3 A (page 14) shows neither the route along which the benzaldehyde was intended to flow
nor the route by which it ended up in the vacuum separator bowl. A proper schematic diagram of
vacuum collection system would help the reader far more. APPENDIX A
-------
Page 63 - NAPP
As noted in the preceding, detail on the nature of chemical analyses and their results are lacking.
APPENDIX B
Equation 3 (showing the disproportionation of sodium disulfite) has a typographical error. It
should read 2Na2S205 — > 2 Na2SO4 + S02 + S
The calorimetry studies described on page 39-40 require additional detail:
a) A description of the ARC method, or a reference to a suitable background paper should appear.
b) The sources and purity of the substances used in the studies should be stated.
c) The amounts and the method of mixing of the reactants should be given for each experiment.
d) The report should have established the products of the reactions when Al was present. The
observations confirm a redox reaction in which AI(o) is oxidized and hydrosulfite reduced, was
S(2) formed?
e) The report states that heats of reaction were determined in the ARC experiments. If so, they should
be given. These may be estimates of the number of joules generated per gram of mixture
in each experimental run. A -heat of reaction- more usually refers to the enthalpy change
associated with occurrence of a single chemical reaction (as represented by a chemical equation).
Without knowing the products of a reaction, no chemical equation can be written and true heats
of reaction are unobtainable.
Page 39 states: -Benzyl alcohol is produced by the reaction of benzaldehyde with sodium
hydrosulfite.- This is incorrect. Some source of H+ must also be present.
Page 39 states: -The reaction products expected are consistent with the results of the chemical
analysis of the site. - This is technically true, but misleading. The chemical analyses detailed in
Appendix A establish only the elemental composition of the inorganic residues and are -
consistent- with any set of reactions that includes compounds of sodium and potassium among
their products.
The sentence on page 39: -The source of the large phenol concentration noted in the grab samples
from the blender does not seem to be a result of the reactions of the reported mixture materials but
most likely occurred at some time during initial attempts to blend the GPA components._ is
self-contradictory. The generation of phenol _during initial attempts to blend the GPA- would have
to result from -reactions of the reported mixture materials.- That is, -GPA components_equal
-mixture materials.- If the sentence means that the phenol arose from the GPA components, it
opposes the earlier conclusion (page 34) that plhenol and the methylphenol compounds were likely
due to the insulating material remnants
Other Comments
Page 3. Potassium carbonate is not an alkali metal but rather a compound of an alkali metal. Its
physical and chemical properties differ sharply from those of the alkali metals.
-------
Page 64 - NAPP
Page B. A pyrophoric substance is one that takes fire on contact with air under ordinary conditions,
not -under appropriate conditions._ Newspaper bursts into fire spontaneously -under appropriate
conditions- but newspaper is not pyrophoric.
Page B. The statement -Benzaldehyde readily oxidizes to benzoic acid, - should be replaced by
_Benzaldehyde is readily oxidized to benzoic acid upon exposure to the air.
Page 18. -A deflagration releases energy at a lower rate. . and is less destructive than a detonation..
The following definitions from the literature would clarify this discussion: -a deflagration is a soft
explosion [in which] pressure are relatively low. . . . Explosions involve pressures of several
atmospheres .... Detonation is a severe form of explosion when pressures are much higher and are
propagated at a high rate (as much as several miles per second).
Page 20. The report states: At 10:00 am on April 20, operators detected a vanilla-like odor in the
liquid feed tank....- This contradicts the Timeline of Events exhibited on page 11 of the report,
which sets the detection of the vanilla-like odor at 12:30 p.m.
Page 22. The list of the most likely predominant reactions omits the combustion of the hot sodium
hydrosulfite, upon contact with the air.
Page 25. It is stated: - If a large amount of water was injected into the material in the blender, the
JCAIT believes a large hydrogen gas bubble would have been formed, causing a detonation with
greater energy th[a]n was released in the accident. The belief needs justification. Presumably,
the gaseous hydrogen would result from reaction of the large amount of water with the powdered
aluminum. The rate of this reaction depends strongly in the temperature (see report page 38). Did
JCAIT estimate the temperature? How? Did JCAIT estimate the amount of energy released in an
accident? Hydrogen bubbles (as in balloons) deflagrate when ignited in the air. If hydrogen forms,
why must it detonate?
RECOMMENDATIONS
Include details of the analytical methods used in the course of an investigation and at least some
representative analytical findings in an appendix.
Take care to avoid loose ends. If observations or physical findings are judged irrelevant, then the
report should state as much, and tell why.
2. Use the chemical literature more aggressively to check facts. Include full literature references in
some uniform format.
Use a technical editor. This report is not very well written.
I devries and Kellogg, J. org. Chem. 45, 4126, 1980.
-------
Page 65 - NAPP
2 Kirk-Othmer Encyclopedia of Chemical Technology, 4th edition, John Wiley Sons, New York,
1998, Vol.
3 Ullman-s Encyclopedia of Industrial Chemistry,_ Sth edition, VCH Weinheim, Germany, 1993,
Vol. A22, page 99.
4 Ullman-s Encyclopedia of Industrial Chemistry,_ Sth edition, VCH Weinheim, Germany, 1993,
Vol. A22, page 99.
5 Lewis, Richard J., Hazardous Chemicals Desk Reference, Van Nostrand Reinhold, New York,
1993, page 691-2.
6 Mahn, W. J., Academic Laboratory Chemical Hazards Guidebook, Van Nostrand Reinhold, New
York, 1991, page 7.
-------
Page 66 - NAPP
Michael Sprinker, CIH
Director, Health and Safety Department
International Chemical Workers Union Council / UFCW
NAPP Technologies, Inc.
EPA/OSHA JOINT CHEMICAL ACCIDENT INVESTIGATION REPORT
EPA 550-R-97-002
A Critical Review
September 23, 1998
General Observations
Overall, the report answered many questions about the incident (especially given the unfortunate and
preventable deaths of five workers with knowledge of the events, the destruction of production records,
and what appeared to be some reluctance of Napp upper management and owners to provide all
necessary information). The investigation team did a good job on the inspection and the report. It is
unfortunate that the staffing levels of both agencies was (and remains) such that more time and
personnel could not be dedicated to getting the report out in a faster manner. I agree completely with
the other peer review team members that these reports need to be written and released as soon as
possible after the incident, in order to ensure that the report can have a greater impact.
The Facility Information, Process Information, and Chemical information sections clearly provided the
necessary background information. However, it would have been useful to have (if it survived the fire
or was otherwise available) a copy of Napp's hazard analysis on the operation, a copy of the MSDSs
supplied to Napp, and a copy of any correspondence between Napp and Technic (or other companies)
which was relevant to the operation. In addition, a copy of the incident report from the Lodi Fire
Department would have been helpful. These could all be placed in the report as appendices.
A description of Technic Inc. and its expertise in chemical blending / processing would have been
useful, if that information was available and if the writers were allowed to include that information. This
would, perhaps, help to place some of the potential problems with tolling into perspective. IfOSHAand
EPA were not able to investigate Technic's expertise due to some legal reason, those reasons should
have been noted in the report (in most circumstances) and recommendations on how to eliminate such
reasons should have been included in the report.
The Description of the Accident was helped greatly by the timeline in Exhibit 1. However, it was difficult
to keep the personnel straight, given the number of workers, supervisors, foremen, etc. over the three
days and four shifts covering the incident. It was difficult to know how many shift supervisors were
involved and if the same night shift supervisor was involved on successive days (while names cannot
be used, perhaps a number or letter designation would be helpful in future incident reports).
-------
Page 67 - NAPP
The discussion of the possible sources of water was very good and the conclusions as to the causes
of the deflagration appear to be very reasonable.
The photos were difficult to understand due to poor reproduction which eliminated much of the contrast
needed to make out important details.
I would have preferred to see a greater amount of emphasis on the failures of the Napp "management
system" and how those lead to the incident. (I cannot refer to the oversight system at Napp as a
management system without putting that term in quotations, since it appears that it was a safety
management system in name only.)
Executive Summary / Overview
+ In general, the report seems to clearly identify root causes. However, I believe that while the
report does address some issues of potential problems in tolling operations (as shown in this incident),
the Executive Summary/Overview seems to de-emphasize the potential problems surrounding tolling.
These specifically are the issues of:
• accountability of the contracting company (owner of the process/technology)
versus that of the contractor (Napp) for training, hazard evaluation and oversight
of the process; and, potentially,
• the need to conduct an on-site evaluation of the contractor facility and equipment.
While I understand that there may be no rules or guidance to require/encourage this, these are
still, potentially, root causes or contributing factors.
+ My second concern with the Executive Summary/Overview is probably more one of wording.
Many employers look at the words "training" and "employees" and apply them only to non-supervisory
employees. Clearly, a major problem identified throughout the report is the lack of training of
supervisors in the areas of hazard recognition, procedures to follow when operations are "out of spec"
(even when to call responsible parties within Napp), and emergency response. I strongly believe that
the report should make clear that training was inadequate for both hourly employees and supervisory
personnel, if that is what the investigation team found. If the team had concerns as to the ability of top
management to adequately determine hazards, then that should also be clearly stated as a factor.
Chapter 1 Background
1.1 Facility Information
Facility Chemical Review Procedures
It would have been helpful to have a copy of Napp's "New Product Review" procedure included in the
report, along with a description of the general deficiencies within that review process. This could help
others in determining where their PHAs may be deficient. Also, the description of those involved shows
that no operators or other line workers had any role in the review. Trevor Kletz and other experts in
this field identify the need for line worker involvement.
-------
Page 68 - NAPP
1.2 Process Information
This section was quite thorough. It would have been useful to have Figures 2 and 3 situated vertically;
this would make it easier to refer from the text to the diagrams. In addition, a scale should be added
to provide some sense of actual dimensions. The descriptions proved here are very useful.
1.3 Chemical Information
In rereading this section, it strikes me that the instructions in the MSDS for the Gold Precipitating
Agent to "... flood the material with water to ensure complete wetting ...", could be correct for a drum
of material. If so, that should be noted as well as whether or not those instructions should apply to
much larger amounts. Many users and others may not be aware of the difference. This raises the
question of whether MSDSs should be required to note whether precautions and emergency
procedures apply to any amount of material.
I think it would have been useful to many users to have a section which showed the ideal system for
blending such reactive chemicals. A summary of why each specific piece of equipment Napp used was
improper for the job would also have been useful.
Other Background Issues / Concerns
I would have liked to have seen a broader discussion of the tolling industry, possibly as an appendix.
A listing of significant incidents at such operations would also be valuable; however, I realize that
finding such information is often very difficult since I don't believe that OSHA or EPA code such
operations in their data bases. A discussion of whether or not tolling operations should be investigated
in more depth by OSHA, EPA or the CSHIB would be useful.
If the reviewers had any specific regulatory recommendations regarding tolling operations or
recommendations for guidance (versus regulations) which were not included in the final report, they
should have been. However, I understand that it is often difficult to include such recommendations in
official reports, given the potential legal implications to the agencies. These might include such issues
such as making both employers responsible for the PHA and other requirements, as well as legal
responsibility (for civil and criminal citations and penalties). OSHA and EPA should look into this area
and, possibly, begin the rulemaking process to address such problems (e.g., a Notice of Proposed
Rulemaking).
Chapter 2 Description of the Accident
+ While the length of time over which the event occurred was extremely long, the description was
generally quite complete. However, as noted above, it would be useful in future reports to provide
some designation for each worker and supervisor so that actions and observations can be connected
to specific people. This would help the reader to better understand where there were communication
breakdowns.
-------
Page 69 - NAPP
+ The report should be specific that there was no reason why a non-cooled (or at least a non-water
cooled) blender could not have been used. Also, the report should state:
• Whether or not such blenders were available at Napp; and,
• Whether or not the blender could have been made "safe" by disconnecting the water
feed and the blanking off the flange without affecting the proper operation of the blender.
+ If there was any testimony as to why the first shift supervisor did not check the liquid feed line and
intensifier bar for leakage, that should have been included in that, given that a very slow leak might not
introduce enough water to be noticed immediately after repair. This would help others to make proper
decisions in similar circumstances.
+ It appears that there had been no testimony as to why the night shift foreman assumed the water
was condensation. It seems obvious in hindsight that this was a major error and could have been
avoided had the SOP called for some other action. One question which still remains in my mind is
whether the decision to wipe out the condensation rather than investigate all the possible sources was
based on time and cost concerns or whether it was based on a lack of training and/or real power to
make decisions. With the information that the site was expecting an FDA inspection the next Monday,
the decision to fill rather than investigate could also be due to a need to finish and clean up before that
FDA inspection. To me, this suggests that a tolling operation, knowing when inspections by a regulator
will occur, could end up taking unnecessary risks. Is this problem?
+ I strongly believe that the term "accident" should be avoided. That term implies "unforeseeable"
and/or "unavoidable" to many readers.
Chapter 3 Analyses and Significant Facts
This section is very informative and well laid out. I have only few comments.
+ A literature review and laboratory study of the hazards of the GPA mixture of the is noted on page
19. From a review of Appendix B, it appears that this refers to the OSHA Salt Lake City Technical
Center. If so, this should be clearly stated here; if not, the reviewer/ experimenter should be stated.
Also, it would be helpful to have the term "small quantities" quantified (e.g., milliliters). This would help
reviewers and others using this document.
+ On page 21, it is noted that the use of an internal alarm would have notified the local emergency
responders. It would be useful to know whether or not an automatic sprinkler system or other fire
suppression system was in place and functioning at the plant, or whether one was required by code
or could be required (if the lack of one was "grandfathered"). At one plant where ICWUC represents
workers, a fire of what turned out to be water-reactive chemicals in the warehouse caused the fire
sprinklers to activate, compounding the extent of the fire and damage. The sprinkler system was
required under NFPA and/or NFC. While this did not happen at Napp, contradictory standards for
reactive chemicals could pose a serious hazard at other tolling operations and at chemical processing
/ storage operations in general. This may be an area which OSHA, EPA and the fire code
organizations should investigate.
-------
Page 70 - NAPP
Chapter 4 Causes of the Accident
+ The discussion of possible causes of the chemical reaction did not address a possibility raised on
page 7 of this document: the potential reaction of aluminum powder with sulfur dioxide (a
decomposition product of sodium hydrosulfite). Since both compounds could have been present once
decomposition started, it seems that this should have addressed, even if only to suggest that it would
be a minor contribution, if any, to the deflagration or to be eliminated as a possible reaction.
Otherwise, this section looks very good.
+ One item seems to be missing from the discussion of root causes and contributing factors: the
seeming breakdown, or perhaps, nonexistence of a clear chain of command in decision making when
serious problems arose with the blending operation. The lack of communication within the Napp
management "structure" throughout this event (as related in the summary in this report) was
astounding. I would have liked to see some discussion of this. Napp also appeared to leave operators
and other hourly workers out of the decision making loop in evaluating hazards. OSHA strongly
advises this (and to some degree requires this) in the PSM standard, while EPA does not address this
in the RMP rules. The report should have addressed this as at least a contributory cause of the
incident.
+ I believe that is important to stress that training for management was inadequate for foremen,
supervisors and even upper management to make proper decisions or properly direct the workers in
the plant, if that was the belief of the investigators. If training had been done but was inadequate, it
would be helpful to spell out those inadequacies point by point.
Chapter 5 Recommendations
The recommendations noted in the report are all clearly supported by the report itself and should help
to reduce risk in other operations, if they are read. I agree with other reviewers that there needs to be
good, simple methods of getting this information out to those most affected: the plants, workers,
unions, supervisors, engineers and process designers, chemists, fire department personnel (including
fire marshals and investigators), etc.
In addition, the need for employers to involve their workers (at all affected levels) in the recognition and
evaluation of abnormal situations, the proper use of equipment, and in the development of PHAs,
SOPs and training, needs to be stressed. Again, in my experience, that is OSHA's policy and belief
and should be clearly stated here, in order to make these more proactive recommendations. Too many
employers are still content to tell workers to "look it up on the MSDS" rather than train workers as
required under the HazCom standard. Those same employers are usually not much better at training
supervisors.
-------
Page 71 -NAPP
Chapter 6 Outcomes of OSHA/Napp Technologies Settlement
A number of OSHA settlements have given both OSHA and the Union specific rights of entry and
involvement in evaluating the progress under the settlement agreement. If OSHA and/or the union
were given any specific rights under this agreement, this should be noted here. As written, this looks
like OSHA will rely on Napp's "good will" to ensure compliance.
As I understand our discussions, Napp's attorneys were allowed to review the preliminary report for
errors and confidential business information. A list of which outside groups were able to review the
report, prior to release, should be noted in the preface.
Appendices
Appendix B - Chemical Reactions
+ There were two significant typographical errors in this section. In the second paragraph, the
formula for sodium thiosulfate is wrong. On page 39, the reaction arrows for both aluminum reactions
are missing.
+ The third paragraph states that "only catalytic amounts of water are needed ...". I would find it
helpful to know what are considered to be "catalytic amounts" in the case of this mixture, with some
discussion of how those catalytic amounts must be dispersed in a mixture (e.g., will 0.5% in a small
portion of the mixture cause a self sustaining reaction or does it take that amount in a much larger
volume of the mixture).
Appendix E - Photos
+ The photos are helpful but would be much more so if they were reproduced better and details were
labeled. They are difficult to understand due to poor reproduction which has eliminated much of the
contrast needed to make out important details. In a number of cases, the backgrounds fade into the
surrounding page, making it difficult to orient one's view. The use of a scale would be helpful, as would
labeling of parts noted in the descriptions. The vacuum head (or spray nozzle?) in the two photos
which make up Figure 5 would have been best adjusted to the same size and then joined together.
Again a scale would be useful here.
I realize that this was the first joint investigation report done by OSHA and EPA. These reports are
quite valuable as shown by the information provided by this report. I hope that my comments are taken
in the spirit of building on the strong foundation which this report provides for future work. I know that
the authors of this report worked under a great deal of time pressure as well as the pressures of trying
to complete their other work on ongoing serious incidents. They should be commended for work well
done.
-------
Page 72 - NAPP
Michael Sprinker initial comments - September 2, 1998
Here are my initial observations regarding the EPA/OSHA Joint Chemical Investigation Report for the
Napp Technologies incident. As these are preliminary, additional readings of the report and
discussions during the upcoming meeting may result in changes to these observations.
The Facility Information, Process Information, and Chemical information sections clearly provide the
background information. As a minor point, it would have been useful to have Figure 2 situated
vertically; this would make it easier to refer from the text to the diagram.
The Description of the Accident was helped greatly by the timeline in Exhibit 1. However, it was difficult
to keep the personnel straight, given the number of workers, supervisors, foremen, etc. over the three
days and four shifts covering the incident. It was difficult to know how many shift supervisors were
involved and if the same night shift supervisor was involved on successive days (while names cannot
be used, perhaps a number or letter designation would be helpful in future incident reports).
Observations - Executive Summary/Overview
1. In general, the report seems to clearly identify root causes. However, I believe that while the
report does address some issues of potential problems in tolling operations (as shown in this incident),
the Executive Summary/Overview seems to de-emphasize the potential problems surrounding tolling.
These specifically are the issues of accountability of the contracting company (owner of the
process/technology) versus that of the contractor (Napp) for training, hazard evaluation, oversight of
the process, and, potentially, the need to conduct an on-site evaluation of the contractor facility and
equipment. While I understand that there may be no rules or guidance to require/encourage this, these
are still, potentially, root causes or contributing factors.
2. My second concern with the Executive Summary/Overview is probably more one of wording. Many
employers look at the words otrainingo and oemployeeso and apply them only to non-supervisory
employees. Clearly, a major problem identified throughout the report is the lack of training of
supervisors in the areas of hazard recognition, procedures to follow when operations are oout of speco
(even when to call responsible parties within Napp), and emergency response. I strongly believe that
the report should make clear that training was inadequate for both hourly employees and supervisory
personnel, if that is what the investigation team found. If the team had concerns as to the ability of top
management to adequately determine hazards, then that should also be clearly stated as a factor.
General Observations - Description of the Accident
1. I would have liked to have seen a broader discussion of the tolling industry, possibly as an
appendix. This would be useful if it included a listing of significant incidents at such operations;
however, I realize that finding such information is often very difficult since I don/Et believe that OSHA
or EPA code such operations in their data bases. Perhaps some discussion of whether or not tolling
operations should be looked at in more depth might be useful.
-------
Page 73 - NAPP
2. Did the reviewers have any specific recommendations regarding tolling operations and the
recommendations for guidance (versus regulations?) which did not make it into the report? These
might include such issues such as making both employers responsible for the PHA and other
requirements, as well as legal responsibility (for civil and criminal citations and penalties).
3. Was there any reason why a non-cooled (or at least non-water cooled) blender could not have
been used? Were such blenders available at Napp? Could the water feed have been disconnected
and the flange blanked off with affecting the proper operation of the blender? If so, this might have
provided a positive isolation from water. I don't recall reading this in the report.
4. Was there any testimony as to why the first shift supervisor did not check the liquid feed line and
intensifier bar for leakage, given that a very slow leak might not introduce enough water to be noticed
immediately after repair? (Pages 9-10)
5. Was there any testimony as to why the night shift foreman assumed the water was condensation?
Was there anything in the SOP which called for some other action on the foreman's part? Is there any
conclusion which pointed to whether this decision to wipe out the condensation rather than investigate
all the possible sources was based on time and cost concerns or whether it was based on a lack of
training and/or real power to make decisions? (Pages 10-12)
6. In my opinion, the term accident should be avoided. That term implies unforeseeable and/or
unavoidable to many readers.
General Observations - Analyses and Significant Facts
This section is very informative and well laid out. I have only few comments.
1. A literature review and laboratory study of the hazards of the GPA mixture of the is noted on page
19. From a review of Appendix B, it appears that this refers to the OSHA Salt Lake City Technical
Center. If so, this should be clearly stated here; if not, the reviewer/ experimenter should be stated.
Also, it would be helpful to have the term small quantities quantified (e.g., milliliters). This would help
reviewers and others using this document.
2. On page 21, it is noted that the use of an internal alarm would have notified the local emergency
responders. I could not find any discussion earlier of whether such an alarm was present or even of
what alarms were present. In addition, it would be useful to know whether or not an automatic sprinkler
system or other fire suppression system was in place and functioning at the plant. At one plant where
ICWUC represents workers, a fire of what turned out to be water-reactive chemicals in the warehouse
caused the fire sprinklers to activate, compounding the extent of the fire and damage. While this did
not happen at Napp, this could be problem at other tolling operations and chemical operations in
general.
-------
Page 74 - NAPP
General Observations - Causes of the Accident
1. The discussion of possible causes of the chemical reaction do not address a possibility raised on
page 7 of this document: the potential reaction of aluminum powder with sulfur dioxide (a
decomposition product of sodium hydrosulfite). Since both compounds could have been present once
decomposition started, it seems that this should have addressed, even if only to suggest that it would
be a minor contribution, if any, to the deflagration. Otherwise, this section looks very good.
2. One item (in my mind) seems to be missing from the discussion of root causes and contributing
factors: the seeming breakdown, or perhaps, nonexistence of a clear chain of command in decision
making when serious problems arose with the blending operation. The lack of communication within
the Napp management structure throughout this event (as related in the summary in this report) was
astounding. I would have liked to see some discussion of this.
General Observations - Recommendations
My comments above may, to some degree, affect the recommendations section. The
recommendations noted in the report are all clearly supported by the report itself and should help.
General Observations - Appendices
Appendix B - Chemical Reactions
1. There were two significant typographical errors in this section. In the second paragraph, the
formula for sodium thiosulfate is wrong. On page 39, the reaction arrows for both aluminum reactions
are missing.
2. The third paragraph states that oonly catalytic amounts of water are needed ...6. I would find it
helpful to know what are considered to be catalytic amounts in the case of this mixture, with some
discussion of how those catalytic amounts must be dispersed in a mixture (e.g., will 0.5% in a small
portion of the mixture cause a self sustaining reaction or does it take that amount in a much larger
volume of the mixture).
Appendix E - Photos
1. The photos are helpful but are somewhat difficult to understand due to poor reproduction which
has eliminated much of the contrast needed to make out important details. In a number of cases, the
backgrounds fade into the surrounding page, making it difficult to orient ones view. The use of a scale
would be helpful, as would labeling of parts noted in the descriptions. The vacuum head (or spray
nozzle?) in the two photos which make up Figure 5 would have been best adjusted to the same size
and then joined together. Again a scale would be useful here.
I do appreciate the excellent work of the investigation team and look forward to meeting with the team
and the reviewers to discuss this document further.
-------
Page 75 - NAPP
EPA/OSHA RESPONSE TO REVIEWERS
EPA/OSHA have summarized and consolidated reviewers' comments and recommendations for the
purpose of providing EPA/OSHA responses. Several comments were shared by several reviewers, and
the Chair's report summarizes comments common to the group.
I. Comments and recommendations from the Chair's summary statement
II. Additional comments and recommendations noted by several reviewers
III. Additional comments and recommendations noted by individual reviewers
I. Comments and Recommendations from Chair's Summary
The Chair in his summary listed the following comments and recommendations as emphasized by all
reviewers:
Reviewers noted that the report appeared to state correctly the root causes of the accident in terms of
both technical mechanisms and technical failures.
They enumerated several elements which could have enhanced the report's usefulness:
—discussion of the types of chemical analyses done,
—discussion of the rationale used to eliminate plausible scenarios,
—clearer depiction of difficulties presented by the extent of destruction, which made certain
analyses impossible and information difficult to obtain,
—better tracking of individuals involved through the chronology of the accident,
—time line of events.
EPA/OSHA Comment: We agree that these elements would have made the report clearer and will
consider inclusion of such elements in any future reports on accidents.
The Agencies agree that information regarding the chemical analyses performed adds value and
understanding to the accident investigation report. However, the agencies chose to summarize the
findings of these analyses in the report and provided the names of the chemical analysis reports in the
reference section of this report. The reports on the chemical analyses are quite extensive and were simply
too large to feasibly incorporate into the report.
-------
Page 76 - NAPP
The Agencies agree that a clearer depiction of the difficulties presented by the physical destruction
at the facility would have provided a greater understanding of the investigatory process. Due to the
extent of the fire, most of the documentation of the blending process was destroyed (or could not be
located). Because most of the information for this report came from interviews, and since the witnesses
of the explosion perished, the investigators used what information they could find. There was a large
amount of information that the investigators were unable to obtain. The Agencies agree that noting the
information that was not available in the report would have provided value.
In the report, the Agencies had to strike a balance between tracking individuals who were involved
in the accident and maintaining the confidentiality of these persons. The Agencies recognize that other
ways of identifying the individuals (e.g. operator#l, manager #1, etc.) could have added clarity to the
report, without compromising confidentiality.
The report contained supporting information for the investigators' rejection of plausible scenarios and
a time line of events. The agencies agree that other tools (e.g. events and causal factors chart, MORT
chart, etc.) could have enhanced the understanding of the investigation process.
Reviewers strongly encouraged EPA/OSHA to consider more detailed recommendations on the special
risks associated with tolling operations and in the handling of water reactive materials. While guidance
is certainly a possibility, rulemaking by either EPA or OSHA should be considered.
EPA/OSHA Comment: EPA and OSHA have taken several steps to address the risks associated with
tolling operations, as a result of the information gathered during this accident investigation.
1. EPA is working with the Center for Chemical Process Safety (CCPS) (at the American Institute
of Chemical Engineers) to develop guidance for the industry, to define the risks more precisely and to
lay out practices and procedures in this important area. This is an important and logical first step to
examine aspects of the problem and to then determine the best approach to disseminating information and
ensuring better safety in the industry. We anticipate that CCPS will complete this project within one year.
2. EPA is developing an Alert for local officials that provides guidance on information resources
during emergency responses, as well as managing reactive chemicals.
3. EPA and the National Oceanic and Atmospheric Administration (NOAA) have worked together
to develop a database of reactivity information for more than 4,000 common hazardous chemicals. The
database includes information about the special hazards of each chemical and whether a chemical reacts
with air, water, or other materials, (http://response.restoration.noaa.gov/chemaids/react.htm)
4. OSHA is currently developing an Advanced Notice of Proposed Rulemaking (ANPRM) which
seeks further comment on the applicability of the PSM Standard to reactive chemicals.
5. EPA is also reviewing the list of regulated substances subject to the Risk Management Program
(RMP) regulations and will consider reactives.
-------
Page 77 - NAPP
However, EPA and OSHA do not believe that regulations specific to the tolling industry are
necessary. The Occupational Safety and Health Act and the Clean Air Act Amendments, which are the
authorities for the agencies' accident prevention programs, make the owner or operator of the stationary
source (or the employer) who is handling the hazardous chemicals solely responsible for compliance with
safety regulations at the facility.
The reviewers did not find the photos very useful and recommended higher quality photos using color,
with high resolution photos posted on the Internet as a low-cost appendix to reports.
EPA/OSHA Comment: EPA has taken steps to acquire such capabilities and expects to have higher
quality photographs in future reports.
The investigators are currently finishing a computer animated video based on the accident
investigation report. This video is meant to support presentations where the audience can discuss the
report with the investigators.
-------
Page 78 - NAPP
II. Additional Comments and Recommendations raised by Several Reviewers
Reviewers: Thepre-release review which EPA andOSHA allowed Napp, but not the other stakeholders,
should have been shared by all stakeholders, and the report should have been peer-reviewed.
EPA/OSHA Comment: Prior to releasing investigation reports, OSHA and EPA must ensure that the
report contains no confidential business information. The Freedom of Information Act (FOIA), the Trade
Secrets Act, and Executive Order 12600 require federal agencies to protect confidential business
information from public disclosure. OSHA has issued regulations in 29 CR part 70, specifying the review
process. To meet these provisions, OSHA and EPA have established a clearance process in which the
companies mentioned in the report are provided a factual portion of the draft report. This portion
contains only the factual details related to the investigation (not the findings, the conclusions or the
recommendations). Companies are asked to view this factual portion to confirm that the draft report
contains no confidential business information (CBI) or trade secrets.
Napp, Technic and Patterson-Kelley had ten business days to review the document for trade secrets
and inform the Agencies. Napp claimed that the report contained CBI, but both OSHA and EPA found
that these claims were without merit. Neither of the other employers claimed that publication would
disclose trade secrets. Therefore, the Agencies published the report.
Prior to publication, the investigative report was extensively reviewed within the Agencies by
technical and management staff. Release of the report to peer reviewers constitutes public release.
Therefore, once the CBI review process described above was complete, the report was immediately made
public and an expert review process began. The expert review, which is documented in this publication,
is the external scrutiny which we agree is necessary for establishing the credibility of the investigative
report and its conclusions.
Reviewers: The accident investigation report should have been published more promptly after the
incident itself.
EPA/OSHA Comment: We agree. In the future, staff resources will be marshaled for any publications
whose effectiveness can be blunted by a lack of timeliness. As a matter of Agency practice, EPA and
OSHA, upon becoming aware of a hazard or safety issue during the course of an investigation, have
promptly published Alerts to the stakeholder community. These precede publication of accident
investigation reports. During the course of this investigation, OSHA issued a Hazard Information
Bulletin describing the potential hazards of utilizing MSDSs as the primary sources of information for
conducting hazard analyses for chemical process activities.
Reviewers: Information should have been included in the report about the actions taken during the
response operation and about the notification of local authorities and their relationship with the
company.
EPA/OSHA Comment: Early in the EPA/OSHA investigation, the Agencies decided to limit its scope
to the events leading up to and including the explosion. This included the actions of persons at Napp
-------
Page 79 - NAPP
Technologies, as well as Technic Inc., that contracted for the tolling operation. More information about
the emergency response would have allowed the reader to understand the context better, and will be
considered for inclusion in any future reports.
Reviewers: While the discussion [of root cause] was adequate, it appears that a more fundamental
management failure was present and should have been described as the root cause with the various
specific management failures as subparts. Elements that should have been developed further in the
report include:
—qualifications of managers,
—SOPs, audits, safety and health programs, hazard analysis, and employees' roles in them,
prior incidents,
—training (of both hourly employees and supervisors).,
—accountability of the contracting company (for this tolling operation).
EPA/OSHA Comment: In the identification of the root causes of the accident (for example, inadequate
hazard analysis, inadequate SOPs and training not addressing emergency shut-down procedures),
management failures are implied since management is responsible for these actions. The subsequent
recommendations focus on steps management should take in the future to address the problems
encountered by Napp in this scenario. Although the Agencies' identification of root cause could have
been expanded to make explicit the management failure, the steps to be taken in future would be the
same. Since the root causes and contributing factors of this accident were management system failures,
the Agencies considered it inappropriate that they be ascribed to individuals or their qualifications.
-------
Page 80 - NAPP
III. Additional Comments and Recommendations from Individual Reviewers:
Scannell
Scannell: The report did not include the company's compliance with applicable state and federal
regulations concerning hazardous materials, nor did the report review the company's prevention and
response procedures against industry standards.
EPA/OSHA Comment: The report's purpose and focus were on root cause of the accident, and the
report was intended to document an event and its causes, rather than to discuss overall regulatory
compliance issues. Expanding the report to address such issues may well result in unnecessary delays in
publication, as an employer's state of compliance often becomes a subject of litigation following an
incident of this nature.
The Agencies agree that a review of applicable regulations and standards could be included in the
accident investigation report. The agencies will incorporate a more detailed review in future reports.
Freeman
Freeman: Use the chemical literature more aggressively to check facts; include full literature
references in some uniform format; use a technical editor.
EPA/OSHA Comment: EPA/OSHA will adhere to these conventions in future reports. The agencies
included a reference section in Appendix D of the Napp report.
Sprinker
Sprinker: Certain further information (appendices for Napp's hazard analysis, MSDS, description of
tolling operations, and other information) would have been helpful.
EPA/OSHA Comment: Inclusion of these elements would have lengthened- the report. Since this
information is available elsewhere, the decision was made not to include it
Sprinker: Need for employers to involve their employees (at all levels) in aspects of safety and proper
procedures should have been emphasized.
EPA/OSHA Comment: We agree that this point could have been made more explicit.
-------
Page 81 -NAPP
APPENDIX A
CHARGE TO REVIEWERS
for the EPA/OSHA Joint Chemical Accident Investigation Report
Napp Technologies, Inc.
Lodi, New Jersey
October 1997
EPA and OSHA jointly released the above report in October 1997, concerning an accident on April 21,
1995 occurring at Napp Technologies, Inc. at Lodi, New Jersey. The report is 67 pages long and
includes an executive summary, background and description of the accident, analysis of the event,
discussion of causes of the accident, and recommendations. Also included are a description of the
OSHA/Napp settlement, appendices containing references, and figures. The principal investigators were
John Ferris and Paul Kahn of EPA and Michael Marshall, Michael Yarnell, and Efraim Zolden of OSHA.
As a reviewer of this document, you should use your technical knowledge and professional judgment to
comment on the technical soundness, overall approach, and completeness of the report and to derive
recommendations for enhancement of accident prevention approaches and accident investigations in the
future.
The report seeks to ascertain the root causes of this accident to further the goal of preventing future
accidents in similar facilities. Your review should address the following aspects of this concern.
Comment on the overall approach taken in the report and its organization.
Comment on the analyses undertaken. Were these sufficient? Were methodologies appropriate? Were
methodologies identified? Were all sources of information properly identified?
Were the sources of reaction initiation plausible? Were all possible sources identified? Were the likely
causes of the chemical reaction explored fully?
Was the discussion of root cause adequate? Were root causes and contributing factors appropriately and
correctly identified? Was evidence for the conclusions drawn sufficient and plausible? Were
methodologies identified? Were all alternative root causes explored?
Are recommendations appropriate and drawn logically from the preceding discussion and conclusions?
Are recommendations sufficient to address the potential of a recurrence of this kind of accident?
Were the appendices sufficient and appropriate? Were the photos appropriate to illustrate the narrative,
clear, and properly documented and presented?
Were all external factors considered? Were human factors and management issues considered
appropriately?
-------
Page 82 - NAPP
What of the overall approach could be used for future investigations? Was the approach sufficiently
broad for application to industry sectors? Were roles of all stakeholders properly addressed in the
report, including roles of federal, state and local agencies, the community, labor and any others? Are
recommendations sufficiently broad to include all elements in addressing prevention of like accidents in
the future?
------- |