IN RE M/V MSC FLAMINIA

Filing 1407

OPINION & ORDER: Based on the Court's decision as to Phase 1, the parties are to confer on the timing for Phases 2 and 3. The Court believes that those phases can be combined. The parties shall consult and provide proposed dates (with expected duration) in a letter to the Court filed not later than December 15, 2017. (Signed by Judge Katherine B. Forrest on 11/17/2017) (mro)

UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF NEW YORK
-------------------------------------------------------------------X
:
:
IN RE M/V MSC FLAMINIA
:
:
:
-------------------------------------------------------------------X

USDC SDNY
DOCUMENT
ELECTRONICALLY FILED
DOC #: _________________
DATE FILED: November 17, 2017

12-cv-8892 (KBF)
OPINION & ORDER

KATHERINE B. FORREST, District Judge:
On July 14, 2012, the M/V MSC FLAMINIA (the “Flaminia”) was crossing the
Atlantic Ocean bound for Antwerp, Belgium. The vessel had departed from New
Orleans, Louisiana fourteen days earlier and it was loaded with cargo. Early that
morning, alarms began to sound, followed shortly thereafter by an explosion. Three
members of the crew were killed, thousands of container cargos were destroyed, and
the vessel was seriously damaged. A number of lawsuits followed, seeking
compensation for death, bodily injury, loss of cargo, and damage to the vessel.
Many of the original claims have been settled, including those alleging wrongful
death and bodily injury. What remains are a host of claims relating to cargo losses
and vessel damage.
The Court has split the trial into three phases: a trial on causation in “Phase
1,” to be followed by trials establishing fault and damages. The Phase 1 bench trial
was conducted from September 11, 2017 through September 19, 2017, with closing
arguments on September 26, 2017.
At trial, three sets of parties presented related but materially different
theories of causation. All agree that the explosion occurred as a result of runaway

auto-polymerization of 80% grade divinylbenzene (“DVB80”)1 that was contained in
ISO containers2 aboard the Flaminia. The manufacturer and shipper of that cargo,
Deltech Corporation (“Deltech”) and Stolt Tank Containers B.V. (“Stolt”),
respectively, assert that runaway auto-polymerization would not have occurred
absent the storage conditions on the dock at the New Orleans Terminal (“NOT”)
(where the DVB80 was stored before being loaded onto the ship) and aboard the
vessel. In contrast, Container Schiffahrts-GMBH & Co. KG MSC “FLAMINIA” and
NSB Niederelbe Schiffahrtsgesellschaft MBH & Co. KG (together, “Conti”), which
owned and operated the Flaminia, and MSC Mediterranean Shipping Company,
S.A. (“MSC”), the time-charterer, assert that the cause of the auto-polymerization
was Deltech’s failure to deliver fully oxygenated DVB80 to the dock at NOT. The
last party that presented a causation theory was Chemtura Corporation
(“Chemtura”), a shipper of another chemical contained in cargo aboard the vessel,
diphenylamine (“DPA”). Chemtura argued that in all events, the DPA was not a
substantial factor contributing to the conditions that caused the explosion.
The parties have spent an enormous amount of time litigating this case. The
discovery was, by any measure, extensive. Each group of parties retained experts,
resulting in a classic “battle of the experts.” The Court carefully studied the
experts’ work, listened to their testimony, and poked and prodded them with

DVB is a chemical used for the synthesis of ion-exchange resins, an important component of water
purifiers. These water purifiers create clean drinking water as well as clean water for use by nuclear
power plants. DVB may also be used in the production of adhesives and polymers.
2 ISO containers—sometimes referred to as “ISO tanks”—are receptacles that can be filled with
liquid. The Court discusses the characteristics of the ISO containers further below.
1

2

questions. According to the Stolt/Deltech experts, the DVB80 was fully oxygenated
and only excessive heat conditions caused the auto-polymerization. The Conti/MSC
experts argue the opposite.
It is clear that neither the experts nor the Court will ever be absolutely
certain as to what caused the DVB80 to auto-polymerize and what ignited the
explosion. But this is a civil case—one in which the standard of proof is not
certainty, but a “preponderance of the evidence.” Based on that standard, the Court
finds that that the DVB80 was delivered to NOT in an appropriately oxygenated
state. However, the choice of NOT as the port of embarkation was a fatal one.
Together, the extended, stagnant storage under a hot sun at NOT, followed by high
ambient temperatures in the hold of the Flaminia, caused the DVB80 to autopolymerize. The Court also finds that the heated DPA, which had been placed in
containers adjacent to those filled with DVB80 at NOT and in the hold of the vessel,
was a substantial contributing factor in the auto-polymerization.
As the auto-polymerization progressed aboard the Flaminia, a white cloud of
venting DVB80 gases triggered alarms. The crew missed a final opportunity to
prevent the explosion when, lacking information as to the conditions in the hold and
instructions as to how much carbon dioxide (“CO2”) to release, it failed to inert the
venting gases. The reasonable crew response to what crew members believed was
an ongoing fire then created a spark that triggered the explosion.
The Court’s findings of fact and conclusions of law are set forth below.

3

I.

THE PARTIES
Dozens of parties have, at various points, been involved in these proceedings.

For purposes of this Phase 1 causation trial, the notable players consist of the
following groups: first, the “ship interests,” Conti and MSC; second, the parties that
manufactured and shipped the three ISO containers of DVB80, Deltech and Stolt;
and third, the companies connected to ten ISO containers of DPA. This last group is
comprised of Rubicon LLC (“Rubicon”), the manufacturer; Chemtura, the owner and
shipper; and Bulkhaul Ltd. and Bulkhaul (USA) Inc. (together, “Bulkhaul”), which
provided the ISO containers in which the DPA was stored. (Stipulated Facts at 9,
¶ 66.)3
II.

THE BATTLE OF THE EXPERTS
A total of 54 witnesses testified at trial: 35 by deposition designation; 13 by

trial declaration, live cross-examination, and live redirect; and six by trial
declaration only (because the parties waived cross-examination (see Trial Tr. at
101–03)).4 The Court also received into evidence over one hundred documents and a
videotape.

Several other companies produced other chemicals being transported on the Flaminia—by the time
of trial, these materials had been absolved of blame. Monsanto Company (“Monsanto”) was the
manufacturer of glyphosate intermediate (“GI”) carried in 30 twenty-foot dry van containers aboard
the vessel on July 14, 2012. (Stipulated Facts at 12, ¶ 1.) BASF Corporation (“BASF”) was the
manufacturer of four ISO container shipments of dimethylethanolamine (“DMEA”) carried aboard
the vessel on July 14, 2012. (Stipulated Facts at 13, ¶ 3.) Suttons International, Ltd. and Suttons
International (N.A.) Inc. (collectively “Suttons”) were the providers of the ISO containers utilized for
carriage of the Flaminia shipments of DMEA aboard the vessel. (Stipulated Facts at 14, ¶ 13.)
4 This last group includes Leon Nell (ECF No. 1292), Gerry Walsh (ECF Nos. 1294, 1334), Robert
Cohen (ECF No. 1298), Ian Wadsworth (ECF No. 1300), Tommy Sciortino (ECF No. 1302), and David
Hughes (ECF No. 1292).
3

4

A number of intelligent, articulate, and talented experts in their fields
testified at trial. Each of the individuals who testified as an expert was truly an
expert; the fact that the Court credits certain conclusions over others does not
suggest otherwise. Ultimately, the Court credits the testimony of the experts
representing Stolt and Deltech over the experts representing MSC, Conti, and/or
Chemtura.5 The Court was highly impressed with the credentials of the
Stolt/Deltech experts, as well as the engagement, rigor, and consistency with which
they approached their work and opinions; Stolt and Deltech’s experts were the most
persuasive.
While relatively early in this Opinion and technically complex, in order to set
the stage for the Court’s findings that follow and which rely heavily on the experts,
the Court now provides a brief overview of their work. The technical details will be
explained more thoroughly in the relevant sections of this Opinion.
A. Dr. Scott G. Davis
Scott G. Davis, Ph.D., testified extensively at trial. The Court was very
impressed by him. Dr. Davis has all the expertise a court could wish for:
extraordinary credentials, engagement with his assignment, and a careful,
forthright, and clear manner. The Court was particularly persuaded by the careful
scientific work that he did which reinforced many of his opinions. Dr. Davis was not

Stolt proffered Anand Prabhakaran to conduct a thermal analysis of the maximum temperature the
DVB80 could have reached in the ISO containers. While the Court found his analyses interesting,
and supportive of heat contributions from the DPA and solar radiation, it ultimately does not rely on
him. Certain of his analyses changed between his deposition and trial and while the Court credited
his explanations, it ultimately need not delve into his analyses to reach its conclusions herein.
5

5

relying solely on theory—he and the company with which he is associated, GexCon
US, Inc. (“Gexcon”), performed modeling and testing that provide a strong,
independent basis for crediting his views. The Court relies heavily on his opinions.
He did not overreach.
A summary of certain of Dr. Davis’s qualifications are as follows: he holds a
Masters of Science and a Ph.D. in mechanical and aerospace engineering from
Princeton University. Dr. Davis is a registered professional engineer in California
and New York, a licensed engineer in Texas, and an authorized professional
engineer in Maryland and Pennsylvania. He is a certified fire and explosion
investigator with the National Association of Fire Investigators National
Certification Board. He has also completed a “fire cause and origin investigation”
training with the California Office of State Fire Marshal, hazardous waste
operations and emergency training in accordance with Occupational Safety and
Health Administration (“OSHA”) standards, and confined space entry training also
in accordance with OSHA. (Davis Trial Decl., ECF No. 1304, at 1, ¶¶ 3–6.) Dr.
Davis has authored numerous scientific and academic publications. (Id. at 1, ¶ 7.)
He is President and Principal Engineer at Gexcon, where he is responsible for fire
and explosion related assignments. (Id. at 1, ¶¶ 1, 8.) This includes post-incident
investigative work, worldwide training and experimentation, risk assessments, and
safety studies for petrochemical facilities and other industries. (Id. ¶ 8.) At Gexcon,
he has performed numerous explosion risk assessments, blast and venting analyses,
assessment of combustible dust explosions, toxic/flammable gas releases and

6

dispersion, hydrogen safety, ventilation, detector placement, and carbon monoxide
dispersion. (Id. at 2, ¶ 9.)
Dr. Davis is a member of Gexcon’s “docents group,” which delivers industrial
seminars all over the world on the hazards associated with gas explosions. (Id. at 2,
¶ 10.) He has expertise in the investigation and prevention of fires, explosions, and
dispersion hazards such as flammable vapors, as well as extensive experience
evaluating the cause, origin, and dynamics associated with fires and explosions,
principally as they relate to ignition, flame propagation, chemical kinetics, and fluid
dynamic processes associated with combustion and explosion events. (Id. at 2, ¶¶
11, 12.) Dr. Davis has been the lead investigator on hundreds of fire and explosion
incidents, including chemical and industrial facilities and equipment, dust
explosions, natural gas and propane explosions, above-ground storage tanks,
unintentional ignition including thermal runaway, and residential and commercial
fires. (Id. at 2, ¶ 13.) Prior to joining Gexcon, his research focused on heat and flow
processes in fires, chemically reacting flows, flame dynamics, and combustion
phenomena in high-pressure burners and reactors. (Id. at 3, ¶ 16.) As part of his
professional experience, he has developed large-scale experiments to understand the
explosion phenomena of deflagration to detonation. (Id. at 3, ¶ 17.)
Deltech retained Dr. Davis to conduct and lead a comprehensive scientific
investigation into the cause and origin of the explosion and fire that occurred
aboard the Flaminia on the morning of July 14, 2012. (Id. at 10, ¶ 1.) His
investigation and work on this case lasted over two and a half years. (Id. at 12, ¶

7

14.) As part of his work, Dr. Davis and others from Gexcon visited the Deltech
facility in Baton Rouge, Louisiana where the DVB80 at issue in this case was
manufactured and filled into ISO containers for transport. (Id. at 10, ¶ 3.) He
reviewed the records of DVB80 production and storage and measured the dissolved
oxygen saturation levels at various stages of Deltech’s DVB80 manufacturing,
storage, and transport process. (Id. at 10, ¶¶ 4–5.) Dr. Davis performed detailed
computational fluid dynamic (“CFD”) analysis of the mixing and storage of DVB80
in Deltech’s main cooling and storage tanks. (Id. at 10, ¶ 6.)
The Court found Dr. Davis’s mathematical heat transfer models particularly
compelling. He created and used these models to ascertain the magnitude of heat
transferred to the DVB80 ISO containers during the time they were stored at NOT
aboard the Flaminia. (Id. at 10, ¶ 7.) This analysis included evaluating the
combined effects of the ambient air temperature, solar radiation on the dock,
thermal radiation from neighboring ISO containers of heated DPA, the ambient air
temperature in the hold where the containers of DVB and DPA were stored (“Hold
4”), the impact of ventilation (or lack thereof), and the heated bunker fuel wing
tanks adjacent to Hold 4. (Id. at 10–11, ¶ 7.) Dr. Davis’s heat transfer model
allowed him to estimate the approximate time it took the DVB to begin autopolymerizing. (Id. at 11, ¶ 8.)
As part of his work on the case, Dr. Davis performed thermal accelerated rate
calorimetry tests (“ARC” tests) of DVB80 to determine the time to onset of an
exothermic reaction (discussed below), which indicates the onset of polymerization;

8

tests to determine the flammability characteristics of DVB80; and tests relating to
venting and peak temperature rise under certain conditions. (Id. at 11, ¶ 9.)
Further, he used advanced computational fluid analyses as well as an analytic
model of CFD dispersion to evaluate the effectiveness of the CO2 system, the
dispersion of venting DVB80 within the cargo hold, and explosion modeling. (Id. at
11, ¶¶ 11–12.)
In addition to all of this, Dr. Davis created a full-scale model to test certain
temperature conditions. He placed an ISO container with characteristics as similar
as possible to one of the ISO containers (Tank I) aboard the Flaminia, filled with
DVB80, and placed it within a specially built structure (the “Full-Scale Test”). The
Full-Scale Test allowed Dr. Davis to establish with a high and persuasive degree of
scientific certainty: (1) the representative UA (relating to thermal resistance) of the
Stolt ISO containers filled to 80% capacity with Deltech’s DVB80; and (2) the
minimum expected time it took to auto-polymerize, or the shelf life of Deltech’s
DVB80 as prepared and shipped aboard the Flaminia. (Id. at 11–12, ¶ 13.)
Dr. Davis’s conclusion, with which the Court agrees, is that Deltech’s DVB80
would not have auto-polymerized and undergone the thermal runaway reaction it
did on July 14, 2012, if it had not sat still in the sun at NOT, if it had not been
stored next to heated DPA both on the dock at NOT and in Hold 4, and if Hold 4 had
been properly ventilated and had not had high ambient temperatures. (Id. at 13–
14, ¶¶ 1–6.) In addition, even when thermal runaway had been achieved, an
explosion was not a foregone conclusion; additional deployment of CO2 could have

9

rendered the gas inert, and in the absence of an ignition event—a spark—no
explosion would have been triggered. (Id.)
B. Dr. Deborah Kaminski
The Court was also highly impressed by Deborah Kaminski, Ph.D., another
expert retained by Stolt and Deltech. Dr. Kaminski provided a number of opinions
relating to heat transfer—an area particularly important in this case, as the parties
are litigating whether, how, and to what extent heat transferred into or out of the
containers in which the DVB and DPA were stored (the “ISO containers”)
contributed to runaway auto-polymerization. Dr. Kaminski is another true expert,
with decades of relevant in-depth experience.6
A summary of Dr. Kaminski’s credentials are as follows: she is a Professor
Emerita of Mechanical Engineering at Rensselaer Polytechnic Institute (“RPI”).
(Kaminski Decl., ECF No. 1303, at 2, ¶ 3.) She earned her Bachelor of Science in
physics from RPI in 1973; her Masters of Science in Chemical Engineering from RPI
in 1976; and her Ph.D. in Mechanical Engineering from RPI in 1985. (Id. at 2, ¶ 4.)
Prior to obtaining her doctorate, she spent five years at General Electric Research
and Development Center, working on heat transfer. (Id. at 2, ¶ 5.) Her doctoral
research was on computational fluid dynamics in free convection. (Id. at 2, ¶ 6.) Dr.
Kaminski was the Associate Technical Editor for the Journal of Heat Transfer from
1998–2001, and she was named a Fellow in the American Society of Mechanical

The Court was also impressed by the fact that Dr. Kaminski is not a professional testifying
expert—this was the first case in which she provided testimony in court. (Tr. at 930:16-18.)
6

10

Engineers in recognition of her work in radiation heat transfer. (Id. at 3, ¶¶ 7–9.)
In 1995–96 she served as the Program Director of Thermal Transport and Thermal
Processing at the National Science Foundation. (Id. at 3, ¶ 10.) She has also coauthored a book on thermal engineering, published in 2004, and has written 82
articles in peer-reviewed publications as well as 11 additional publications in the
areas of thermal engineering and heat transfer. (Id. at 3, ¶ 11.)
Dr. Kaminski was retained to determine the temperature history and
polymerization of the DVB80 containers from the time they were filled at Deltech’s
Baton Rouge manufacturing facility until the time the first alarm went off aboard
the Flaminia on July 14, 2012. She also computed the temperature of the DPA and
measured its contribution to heat conditions within the hold where the containers of
DVB80 and DPA were stored.
Dr. Kaminski created reliable experimental and theoretical estimations of the
thermal resistance of the relevant ISO containers. She then predicted the
temperatures of the liquid DVB80 and heated DPA while the ISO containers were
at NOT and examined the influence of the heated DPA on the DVB80. She
discussed the heat transfer that occurred at NOT—where one ISO container filled
with DVB80, an unstable and heat-sensitive mixture, was stored facing three
neighboring containers filled with heated DPA; two additional containers of DVB80
were on top of the stack. Her analysis determined that all of the DVB80 containers
were affected by solar radiation, infrared radiation, and proximity to neighboring

11

DPA containers. Her opinions in this regard were rigorous, based in evidence,
clearly explained, and persuasive.
In addition, Dr. Kaminski modeled the temperatures of the DVB80
containers while they were in the hold of the Flaminia for 14 days, considering a
number of different air temperature scenarios. She then predicted the induction
time for the three ISO containers of DVB80. Her conclusions agree with those of
Dr. Davis on which the Court relies, finding thermal runaway following from the
combined conditions of heat at NOT and both heat and poor ventilation in Hold 4.
C. Dr. Hans Fauske, D.Sc.
A third expert upon whom the Court relies is Dr. Hans Fauske. He was
retained by Deltech to testify regarding the thermal stability of DVB80. His
testimony was ultimately narrow—providing experimental results that allowed for
a mathematical calculation predicting the time necessary to achieve runaway autopolymerization. Dr. Davis was persuaded as to the reliability of these equations,
and so was the Court.
Dr. Fauske is the founder, Emeritus President, and Regent Advisor of Fauske
and Associates, LLC, a world leader in nuclear, industrial, and chemical processes,
and now a wholly-owned subsidiary of Westinghouse Electric Company. He
obtained a Masters in Science in Chemical Engineering from the University of
Minnesota in 1959 and a Doctorate of Science from the Norwegian Institute of
Technology in 1963. After completing his graduate education, Dr. Fauske joined the
staff of the Argonne National Laboratory, an entity managed by the University of

12

Chicago. In 1972, he became a Senior Chemical Engineer at the lab (the equivalent
of a full professor of the University of Chicago). In 1975 he was awarded the
University of Chicago Medal for Distinguished Performance. He has won a number
of awards and recognitions from organizations and academic institutions worldwide,
has published more than 200 scientific articles, and holds numerous patents in the
areas of nuclear and chemical safety.
Dr. Fauske believes strongly in the benefits of experimental results to inform
his conclusions. He presented compelling testimony regarding testing that he
conducted for this matter. Dr. Fauske also ran a series of ARC tests, Thermal
Activity Monitor (“TAM”) tests, and Vent Sizing Package Calorimeter (“VSP2”)
testing. In his opinion, TAM tests are the most accurate means to determine the
shelf life of DVB80. (Fauske Decl. at 27, ¶ 79.) He was persuasive in this view.
Based on these experiments, Dr. Fauske was able to derive “Arrhenius equations”
that predict the “shelf life” (that is, the “induction time” or time to autopolymerization) of Deltech’s DVB80 as a function of its temperature. (Fauske Decl.,
ECF No. 1290, at 5, ¶¶ 37–38.) “Arrhenius equations” are mathematical
calculations that can account for chemical reactions occurring at increased
temperatures. (Id. at 6, ¶ 41.)
Dr. Fauske determined an Arrhenius equation applicable here based on
guidelines for the DVB published by Deltech and Dow Chemical Company (“Dow”).
(Id. at 6–7, ¶ 44.) Notably, the Arrhenius equations that he derived from the TAM
testing (assuming no headspace in the ISO container) and from the guidelines

13

provided by Deltech and Dow predicted the shelf life of the DVB80 used in the FullScale Test conducted by Dr. Davis. In addition, based on his testing, Dr. Fauske
was able to conclude that the DVB80 sample he received from Deltech, which had
been manufactured in the same manner as that which filled the ISO containers
aboard the Flaminia, was adequately saturated with oxygen. According to his
Arrhenius equations, under normal conditions, the upper bound of the shelf life for
DVB80 manufactured according to the same process as that aboard the Flaminia
was 64.9 days. (Id. at 37, ¶ 94.) This means, under normal conditions, the DVB80
aboard the Flaminia should have made it safely to port in Antwerp, Belgium.
The Court was persuaded by Dr. Fauske that the Arrhenius equation he
developed allows for a determination of how long Deltech’s DVB80 would remain
stable at various temperatures, and provides the approximate shelf life of the
DVB80.
D. Plaintiffs’ Experts
With regard to the plaintiffs’ expert witnesses, the Court again notes that all
were impressively credentialed and deserving of the title “expert.” However, for the
reasons set forth here and throughout this opinion, the Court was not persuaded by
their testimony that Deltech did not adequately oxygenate the DVB80, or that
anything other than crew activity ignited the explosion.
Plaintiffs retained Dr. Paul Beeley, a forensic investigator who specializes in
fires and explosions. Dr. Beeley presented four possible sources of ignition for the
Flaminia fire and explosion including: a spark involving the electrical system inside

14

Hold 4; crew activity on deck; discharge of static electricity; and thermal runaway of
the DVB leading to auto-ignition. (Beeley Decl. ¶ 8.) At trial, though, Dr. Beeley
could not identify any physical evidence supporting one source of ignition over
another, and he did not assign a relative probability to any of the sources. (Tr. at
774:4–6; id. at 775:3–6; id. at 788:4–16.) Further, he did not perform an
independent investigation of the ship’s electrical systems, but instead relied on
another expert’s opinion regarding the equipment.7 As such, Dr. Beeley’s testimony
cannot be relied on to prove anything beyond the fact that the explosion was
triggered in one of at least four ways. The Court’s task in this Phase 1 trial is to
determine whether these possibilities can be measured—and they can.
Plaintiffs also retained Edward Hammersley, another fire and explosion
investigator and a chemistry expert, and David Robbins, a forensic investigator and
specialist in fires and explosions. The Court found Hammersley and Robbins
similarly credible, but was still unpersuaded that auto-polymerization occurred as a
result of a flaw in Deltech and/or Stolt’s manufacturing and/or transport processes.8
Hammersley’s declaration focused on testing of samples of materials from
Hold 4 of the Flaminia, which, he concluded, demonstrated that the DVB shipments
had undergone auto-polymerization that resulted in venting of DVB80 polymer.
(Hammersley Decl. at ¶¶ 8, 44–49, 74–75, 99–100.) He also determined that no

Dr. Beeley relied on David Robbins, whose testimony is discussed below.
Both Hammersley’s and Robbins’s declarations focused, in large part, on the fact that the explosion
derived from auto-polymerization that occurred in the tanks of DVB80. The facts that autopolymerization had occurred and that the DVB80 exploded were no longer in dispute at the time of
the Phase 1 trial.
7
8

15

other cargo was involved in the explosion, and that there was not a fire in the hold
prior to the release of CO2. (Id. at ¶¶ 9–10.) With regard to the induction time to
auto-polymerization, while Hammersley testified that Dr. Fauske’s Arrhenius
equations resulted in a “prediction contrary to scientific expectation,” he was not
persuasive in this view. (Hammersley Decl. at ¶¶ 113, 116, 127.) And at trial,
Hammersley conceded that Dr. Fauske’s equations set forth a conservative view of
temperature conditions (that is, a view favorable to the plaintiffs’ interests). (Tr. at
1397:4–1400:14.) Hammersley did not perform any TAM tests of his own. (Id. at
1405:9–10.)
Robbins similarly concluded that the fire and explosion were caused by the
DVB80’s auto-polymerization and ignition due to a discharge of static electricity.
He analyzed previous incidents involving auto-polymerized DVB as well as the
preferred and calculated temperature and storage conditions for the Flaminia
DVB80. Like Hammersley, Robbins ruled out alternative sources of a fire, such as
other cargo. (Robbins Decl. ¶ 172.) Overall, the Court was not persuaded that
Hammersley’s and Robbins’s explanations supported a theory that autopolymerization occurred due to a flaw in Stolt and Deltech’s processes.
Finally, plaintiffs retained Dr. Brian Ott, a chemical engineer who opined
that it was “more likely than not that the subject DVB shipments were not fully
saturated with oxygen when they were delivered” to NOT. (Ott Decl. ¶ 17.) The
Court was not persuaded. Though Dr. Ott opined that the liquid DVB80 was not
sufficiently oxygenated, he neither modeled its mixing within the storage tank nor

16

tested the oxygen saturation levels of DVB during Deltech’s manufacturing process.
The Court is also unpersuaded by Dr. Ott’s calculation of the relevant UA values;
Dr. Davis’s values were based on direct measurement through his Full-Scale Test,
while Dr. Ott’s were reverse-engineered and based on an incorrect assumption
regarding the oxygen saturation in one of the tanks used during manufacturing.
Additionally, his model failed to incorporate the influence of solar radiation and
DPA on the containers while they sat on the dock.
Further, Dr. Ott critiqued Dr. Davis’s computation of what is referred to as
the cumulative Fraction of Inhibitor Life Consumed (“FILC”) measure. Dr. Ott
contends that because Dr. Davis’s calculations depend on precise knowledge of the
oxygen concentration and temperature of the subject DVB shipments—which is
unattainable—they are necessarily unreliable. (Ott Decl. at ¶¶ 179–80, 186.) This
position is unpersuasive. While there is uncertainty, that does not address Dr.
Davis’s careful, reasoned conclusions based on certain known facts, principles,
modeling, and experiments. In addition, Dr. Ott’s model of the DVB’s shelf life is
itself flawed. It derived a UA value based on incorrect assumptions, used an
unrealistic temperature contribution from the DPA on the ambient air, and failed to
account for diffusion from the headspace. (Davis Decl. at 100, ¶ 3.)
The Court also viewed Dr. Ott as overreaching in his answers at trial. For
instance, he posited assumptions that, if credited, (1) would support a scenario in
which most, if not all, of Deltech’s DVB shipments are so unstable that venting and
even explosions should occur frequently on trans-Atlantic voyages (and they do not);

17

and (2) predicted auto-polymerization of the Flaminia shipments almost a week
before it actually occurred. Furthermore, Court found Dr. Ott unnecessarily
argumentative, distracting from any persuasive force in his arguments.
E. Chemtura’s Expert
Chemtura presented one witness at trial, Douglas Carpenter, a mechanical
engineer who was retained to determine what DPA’s role might have been in the
fire and explosion. The Court found Carpenter credible but was not ultimately
persuaded by his views. He principally opined that the DPA did not make a
“thermal contribution” (that is, contribute to heat) in the adjacent DVB containers.
Carpenter opined that the source of heat for the DVB containers may have been reradiation from other cargo exposed to solar radiation or the pavement; notably,
however, he did not adequately explain why these sources would not have also reheated the DPA. (And no other expert pointed to these sources as heavily
influencing the DVB’s temperature.) Simply put, the weight of evidence throughout
trial is against Carpenter’s conclusions, and thus the Court does not rely on them.
III.

FINDINGS OF FACT9
A. DVB’s Chemical Properties
The parties largely agree on the basic chemical properties of DVB80 that are

are at the heart of this case. The core disagreement relates to whether the

The Court makes its findings of fact by a preponderance of the credible evidence. This Opinion
contains some citations to evidence; these are example citations only. The Court has not attempted
to exhaustively recite all supportive citations in the record.
9

18

manufacturing process failed to adequately oxygenate the DVB80, or whether the
terminal and vessel storage conditions triggered auto-polymerization.
DVB is an “enhanced performance aromatic monomer” produced by Deltech
to include para-methylstyrene (“PMS”), vinyltoluene (“VT”), and tertiarybutylstyrene (“TBS”). (Cooper Decl., ECF No. 1295, ¶ 8.) Deltech produces DVB in
two grades: 80% (“DVB80”) and 63% (“DVB63”). The DVB that shipped aboard the
Flaminia was DVB80. DVB80 is a monomer that, depending on exposure
conditions, can undergo heat-initiated free radical10 polymerization. (Davis Decl. at
28, ¶ 61.) When exposed to heat, the DVB monomer molecule becomes unstable and
forms a reactive, free radical. This molecule can then react with another DVB
molecule to start a polymer chain. (Id. at 28, ¶ 62.) This is referred to as
“polymerization.”
The “polymerization” of DVB is an “exothermic” reaction. That is, energy
(heat) is released when DVB monomer molecules combine to form DVB polymer.
(Stipulated Facts at 1, ¶ 6.) Thus, once started, the process generates its own heat,
which results in additional polymer formation; once begun, the polymerization
process is self-sustaining. This, in turn, gives rise to “auto-polymerization.” (Id. at
1, ¶ 7; Davis Decl. at 29, ¶ 65.) In scientific terms, the “exothermic reaction” is
“self-promoting” and “auto-accelerating.” (Stipulated Facts at 1, ¶ 8; Davis Decl. at
29, ¶ 66.) Polymerized or polymerizing DVB is not a desired condition.

The term “free radical” is a technical name for a molecule that has one or more unpaired electrons.
(Davis Decl. at 28, ¶ 63.)
10

19

Polymerizing DVB is unstable and potentially dangerous. Customers order DVB in
its monomer form, without high polymer content.
When the heat generated by the exothermic auto-polymerization reaction
exceeds the heat lost to the environment, the reaction is said to have reached
“thermal runaway”; at this point, polymerization increases exponentially. (Davis
Decl. at 29, ¶ 67.) If the runaway reaction generates temperatures and pressure
that exceed the capacity of the equipment in which the product is stored (for
instance, an ISO container), a pressure relief valve is required to vent accumulated
gases. (Id. at 29, ¶ 68.) A white, smoky cloud of gas may be emitted. If exposed to
an ignition source and a specific amount of oxygen (discussed below), the DVB gas
may explode. Deltech’s manufacturing process is designed with these chemical
characteristics in mind.
TBC and oxygen halt polymerization (that is, the formation of DVB polymer)
by creating a chain-terminating reaction sequence (that is, TBC and oxygen can
stop the formation of polymer chains and thus prevent the exothermic reaction that
can lead to thermal runaway). (Id. at 29–30, ¶¶ 71–72.) During the manufacturing
process, Deltech oxygenates and adds TBC to the DVB liquid to inhibit
polymerization. (Id. at 29, ¶ 70.)
A major issue in this Phase 1 trial is the time it took the DVB80 aboard the
Flaminia to begin to auto-polymerize, referred as the “induction time” of DVB or its
“shelf life.” The induction time or shelf life of DVB liquid is the time it takes to
deplete the inhibiting materials (that is, the oxygen and TBC) below a threshold

20

value, allowing the auto-polymerization reaction to commence. (Id. at 30, ¶ 73.)
The induction time depends on the temperature of the liquid, which dictates the
consumption rate of the inhibiting material (that is, the consumption rate of oxygen
and TBC). (Id. at 30, ¶¶ 73–74, 76.)11 Once the TBC or oxygen is depleted below a
certain threshold, the chain-terminating path no longer exists and polymerization
can occur. (Id. at 30 ¶ 77.)12 Thus, the oxygen saturation level, the amount of TBC,
and temperature play a significant role in the stability of DVB80.
Even polymerized or auto-polymerized DVB80 does not explode without some
external ignition source and just the right amount—no more, no less—of oxygen. To
reach the point where a spark may ignite the liquid requires that the temperature
of the DVB80 liquid reaches what is referred to as its “flashpoint.” (Id. at 25, ¶ 51.)
DVB80 has a flashpoint of between 156.2–170°F (69–76.7°C). Its auto-ignition
temperature (that is, the point at which it ignites without an external ignition
source) is far higher—470°C. (Id. at 26, ¶ 54.)
Additionally, in order for DVB80 vapor to ignite (assuming a temperature
level at or above the flashpoint), the concentration of DVB80 vapor and air (oxygen)
must be within narrow limits; the DVB vapor can only be between 1.1% and 6.2% of

Deltech has a Material Safety Data Sheet (“MSDS”) for DVB80. The MSDS contains certain
hazard, firefighting, transport, regulatory, stability, handling, and storage information for DVB80.
The MSDS states that DVB80 is a reactive monomer with a melting point of -126.4°F (-88°C). The
MSDS also indicates that for ignition of DVB vapor to occur, the DVB in the air must be between
1.1–6.2%. (Deltech Trial Ex. 46.) Deltech has also created a Technical Bulletin for DVB. (Deltech
Trial Ex. 111.) The Technical Bulletin states that DVB80 has a boiling point of 392°F (200°C).
12 During the manufacturing process, Deltech adds more TBC than is chemically necessary to ensure
complete inhibition. Thus, the limiting factor in a closed system such as an ISO container is the
oxygen concentration level. (Id. at 31, ¶ 78.)
11

21

the combined vapor/air mixture. If the concentration is below 1.1% DVB80 vapor
per unit volume of air (that is, too much air dilutes the vapor), or more than 6.2%
DVB80 vapor (that is, not enough air), ignition is impossible. (Id. at 25, ¶¶ 48–50.)
Thus, the window for ignition of DVB80 is relatively narrow. In addition, even
when these conditions are met, an external ignition source is required (such as a
spark) to trigger a fire or explosion. Here, as discussed below and based on the
evidence at trial, the Court is persuaded that the DVB80 did not auto-ignite, but
rather that the crew lifted the lid covering the access point to Hold 4 as part of its
response to the fire alarms in order to insert a hose. Such lifting caused friction,
resulting in a spark that triggered the explosion.
B. Deltech’s DVB Manufacturing Process
Deltech’s process of manufacturing DVB80 includes sufficient chilling,
oxygen saturation, and TBC to ensure safe trans-Atlantic shipment under typical
storage, temperature, and transport conditions (including typical voyage length).
The pertinent aspects of the process are as follows: Deltech manufactures
DVB80 at its Baton Rouge facility in batches or campaigns based on customer
demand. The following figure depicts the process:

22

Figure 1.

(Stipulated Facts at 3, ¶ 19; Davis Decl. at 32, ¶ 80.)
The DVB80 is manufactured in numerous distillation columns. (Stipulated
Facts at 3, ¶ 20.) While in the columns, oxygen exposure is minimal because the
columns operate under a vacuum. (Id. at 3, ¶ 21.) TBC is added in the overhead
vapor line that connects the top of AT-307 to the TS-337 condenser and, at this
point in the process, the DVB mixture is cooled to the cooling tower water
temperature which is near ambient. (Id. at 3, ¶ 22.)
The DVB80 is then pumped from the finishing column to one of the day tanks
(MD-326 and MD-327), where it is circulated to ensure product uniformity prior to
sampling. (Id. at 4, ¶ 23.) After approximately twenty-four hours of production
(depending on the production rate), the flow of DVB80 from the column is redirected
into the other day tank. (Id. at 4, ¶ 25.) The DVB80 is then circulated and a

23

sample is taken from the day tank for analysis in the lab. (Id. at 4, ¶ 26.) If the
sample meets specifications, the product in the day tank is transferred into a
storage tank, the MV-804. (Id.)
The MV-804 storage tank plays a key role in this case, as it is the location in
which the bulk of the chilling and oxygenation of the DVB80 occurs. The MV-804 is
a cylindrical tank (30’ tall, 30’ diameter), with a capacity of 155,000 gallons. (Id. at
4, ¶ 31.) Since at least 2012 to the present (and during the production campaign
relevant here), the MV-804 storage tank is filled only at or below 80% capacity.
This allows for 20% headspace in the MV-804. Headspace, in turn, allows for the
presence of air (i.e., oxygen) above the liquid. This allows for diffusion of the oxygen
into the DVB80 while in the MV-804 tank. The tank also has a vent allowing
oxygen into the tank, as well as a circulation loop that allows the DVB to be
continuously circulated through a chiller (at a rate of 34 gallons/minute). (Id. at 4,
¶ 27; see also Davis Decl. at 33, ¶¶ 87–88; Sciortino Decl., ECF No. 1302, ¶ 16.)13
The chiller unit in the MV-804 storage tank is equipped with a permanent fixed
piping system and pump that takes DVB from the tank, runs it through an external
chiller, and then delivers it back into the MV-804 tank.
Prior to June 21, 2012, several containers were filled with DVB80. This
decreased the level of liquid in the MV-804, increasing the headspace. The DVB80

Tommy Sciortino is employed by Deltech as a “Loader” of DVB product and Shipping Coordinator.
He is knowledgable about Deltech’s manufacturing process. (Sciortino Decl. at 1–2, ¶¶ 4–7.)
13

24

ultimately destined for the Flaminia continued to circulate within the MV-804
tanks for several additional days, continuing to be oxygenated and chilled.
The temperature of the MV-804 storage tanks is monitored daily by Deltech
personnel. (Sciortino Decl. ¶ 14.) In addition, a sample from the MV-804 storage
tank is taken once per week for analysis by Deltech’s Quality Control Laboratory.
The testing checks TBC levels. (Cooper Decl. ¶ 102.)14 Deltech maintained records
of the temperatures for the product storage tanks for the period from June 18–22,
2012. (Sciortino Decl. ¶¶ 29–32.) On June 21, 2012 (the day the DVB80 was loaded
into the containers destined for the Flaminia), the temperature of the DVB80 in the
MV-804 storage tank was 44°F—a temperature well within typical and safe limits.
(Id. ¶ 34, fig. 5.)
Due to changes in the level of DVB liquid within the storage tank, DVB
residue may adhere to the walls of the tank. Lowering tank levels (for instance,
when ISO containers are filled) may leave a thin layer of liquid DVB on the tank
walls. As the liquid layer vaporizes, some DVB is left behind. In addition,
stalactites (solid formations of DVB) may be formed on the roof of the tanks.
Stalactites are removed during cleaning but sometimes fall into the liquid DVB and,
depending on the presence of factors that may have allowed it to polymerize, can
cause the polymer content of the DVB to increase.15

Greg Cooper has been a Senior Process Engineer at Deltech for over eight years. He is responsible
for quality control, including evaluating and improving Deltech’s manufacturing processes for DVB.
(Cooper Decl. at 1–2, ¶¶ 2, 6.)
15 This point is, by analogy, useful to refer to when explaining a higher than expected polymer
content of DVB80 shipped aboard a different vessel, the Ludovica, just prior to the Flaminia in June
2012. Persuasive evidence was presented at trial that the polymer content of that shipment was
14

25

C. Filling into a Storage Receptacle
The next step in Deltech’s manufacturing process is filling a receptacle with
DVB. Two aspects of this process have particular importance in this case: first, the
oxygenation (via diffusion) that occurs during and as a result of the passage of the
liquid through the air and splashing into the containers that occurs during the fill
process itself; and second, the physical characteristics of the container, such as
whether it is an uninsulated drum or an insulated ISO container. The temperature
of the stored DVB product in an ISO container is impacted by the type and
placement of insulation, vents, and the surface area exposed to heat sources.
Tommy Sciortino, a “loader” employed by Deltech, testified by declaration at
trial and all parties waived cross-examination. Sciortino personally filled the three
ISO containers of Deltech’s DVB80 that were shipped aboard the Flaminia.
(Sciortino Decl. ¶ 7.) For the purposes of this litigation, the three ISO containers
destined for the Flaminia have been designated as Tanks I, J, and K. (Id. ¶ 46.)
While Sciortino does not recall filling these precise containers, he followed the same
procedures for all loading. Deltech creates a Product Transfer Sheet for product
that will be filled into containers for a customer. The Product Transfer Sheet for
Tanks I, J, and K, which Sciortino completed on June 21, 2012, identified the TBC
level within the expected range of 1000–1,100 ppm. (Id. ¶ 42.)

based on the formation of stalactites and soluble polymer in the ADPO storage tank. (ADPO is a
storage facility for, inter alia, liquids in ISO containers. Deltech shipments are received at the Port
of Antwerp and then trucked to ADPO’s facility in Kallo, Belgium. (Jodlowski Dep. Tr. at 12:16–
16:23.))

26

ISO containers, including Tanks I, J, and K here, are filled one at a time.
Deltech uses “loading racks” for this purpose. (Id. ¶ 75.) The loading racks are freestanding and consist of a metal staircase with a platform on top. When a truck
pulls up to the loading rack, the loader drops a moveable walkway above the ISO
container. The walkway rests on the container’s frame and provides the loader with
access to its top. The walkway allows the loader to walk onto a container and to
open and close the lid on its top—referred to as a “manlid.” (Id. ¶ 76.) The manlid
is the opening through which product is filled. (Id.) Prior to filling a container, the
loader runs through a standard “pre-filling” checklist and visibly inspects the
container for cracks, dents, or damage, the container’s temperature gauge, the dome
gasket that seals the manlid for proper fit, and the container’s seal to ensure it is
properly fitted and undamaged. (Id. ¶ 79.) As reflected in the checklist retained in
Deltech’s files, Sciortino performed the typical pre-loading checks for Tanks I, J,
and K. (Id. ¶¶ 81–83.) No defects or problems were detected with any of these ISO
containers. (Id.)
The filling process itself takes 35–40 minutes. The loader inserts a 12-foot
gauge stick into an empty tank. (The stick has markings to identify the fill point.)
The loader then extends the loading arm out over the tank’s manlid, which was
previously opened during the inspection process. (Id. ¶¶ 89–90.) The loader then
lowers the loading arm through the manlid; the end of the nozzle extends
approximately 30 inches below the manlid opening. (Id. ¶ 91.) The DVB80 pours
out of the loading arm, falls through the air and splashes into and around the inside

27

of the ISO container. (Id. ¶ 97.) This process further oxygenates the liquid as it
falls or splashes into the container. (Davis Decl. at 46, ¶ 135.) The loading arm
remains above the liquid level of the DVB in the ISO container for the majority of
the filling process; it is typically submerged for about three minutes during the
entire 35–40 minute process. (Sciortino Decl. at ¶ 97.)
When filling is almost complete, the loader inserts a thermometer into the
DVB liquid in the container to determine the temperature of the product. (Id. ¶
104.) A sample of the DVB is taken and tested by Deltech’s Quality Control
Laboratory. (Id. ¶¶ 86, 104.) Samples from each of Tanks I, J, and K indicated TBC
concentration levels within the expected range of 1000–1,100 ppm and polymer
concentration at an expected level of less than 5 ppm. (Id. ¶¶ 115–118.) The
Deltech “Loading Sheet” records the temperature of the DVB after filling. The
Loading Sheets for Tanks I, J, and K indicate that the DVB80 had a temperature of
44°F. (Id. ¶¶ 129–133.) This is consistent with the temperature in the MV-804
storage tank and corroborates Deltech’s position that the manufacturing process
employed for the DVB80 here was its typical process.
After an ISO container has been filled, Deltech’s loaders then perform a postinspection of the truck, its chassis, and the filled ISO container. During this postfilling inspection, the loaders ensure, inter alia, that there are no leaks and that the
chains on the caps, plugs, and safety valves are secure. No defects or problems for
Tanks I, J, or K were noted during this post-filling check process. (Id. ¶ 107.) After
the post-filling inspection, the truck then pulls away from the loading dock.

28

In July 2016, one of Dr. Davis’s Gexcon colleagues observed the filling
process. (Id. ¶¶ 155–156.) His observations confirmed the points in the process
during which oxygen saturation and the addition of TBC occurred, as well as the
loading protocol. No other expert observed the process.
D. Manufacturing and Filling the Flaminia Shipment
At issue in this trial is the cause of the auto-polymerization of DVB80 in
Tanks I, J, and K. The DVB80 in each of those containers was manufactured by
Deltech pursuant to the above process, and filled into three ISO containers provided
by Stolt. Plaintiffs argue that the lack of adequate oxygenation of the DVB80
caused it to auto-polymerize aboard the Flaminia. The Court here describes the
particular facts relating to the manufacturing and filling process relevant to these
three containers, based on the process set out above.
At trial, the parties focused on whether Deltech’s manufacturing process for
the DVB80 shipped aboard the Flaminia allowed for sufficient oxygen saturation
and chilling of DVB80. If not, then the DVB80 may have been doomed to autopolymerize because its inhibitors (TBC and oxygen) would have been depleted
before successful completion of the voyage; but if the DVB80 was adequately
oxygenated and had enough TBC, then external conditions must have played the
causal role.
After filling on June 21, 2012, Deltech tested Tanks I, J, and K for TBC levels
and polymer content. As stated above, the test results showed that the TBC and
polymer content were within specification levels. Evidence amply supports the

29

Court’s finding that the DVB80 filled into Tanks I, J, and K had a high level of
oxygen saturation from: (1) the mixing in the MV-804 storage tank for at least ten
days since the end of the production run, and for over six days after the headspace
in the storage tank had additional fresh air in it; and (2) agitation from the fill
process itself, from truck transport on the road, and in connection with the process
at NOT during which cargo containers were stacked. Under normal transit time
and temperature conditions, this level would have allowed for a safe arrival in
Antwerp.
The DVB80 in the ISO containers aboard the Flaminia was manufactured as
part of a production campaign that started on May 19, 2012 and ended on June 11,
2012. (Stipulated Facts at 5¸ ¶ 35; Davis Decl. at 34, ¶ 93.) During this
manufacturing campaign, the levels of DVB80 in the various tanks (including the
MV-804) rose and fell. In general, the higher the liquid level, the less oxygen can
saturate the product (as there is less room for the product to move around, allowing
for mixing with the oxygen). Before filling Tanks I, J, and K on June 21, 2012,
Deltech filled ISO containers to be shipped aboard another ship, the Ludovica, on
June 15 and June 18. This reduced the level of DVB80 in the storage tank, allowing
more oxygen to mix with the remaining product, at least some of which was
eventually shipped aboard the Flaminia. (Davis Decl. at 37, ¶¶ 105–06.) The
DVB80 that eventually filled the ISO containers destined for the Flaminia spent
several additional days in the tank, allowing for more chilling and additional

30

oxygenation compared with the Ludovica shipments (which in all events made the
trans-Atlantic voyage safely). (Stipulated Facts at 37, ¶ 110.)
The Flaminia DVB80 and the Ludovica DVB80 were produced during the
same manufacturing campaign. The Ludovica departed for Europe shortly before
the Flaminia.16 The circumstances relating to the Ludovica shipment provide
further evidence that the DVB80 manufactured by Deltech from May 19–June 11
had been sufficiently oxygenated and chilled. The key fact is that the DVB80
aboard the Ludovica did not achieve runaway polymerization en route.
Three ISO containers filled by Deltech on June 15 and June 18, 2012 were
shipped out of NOT aboard the Ludovica. These ISO containers were shipped
aboard the Ludovica eight days before the shipment aboard the Flaminia. The ISOs
aboard the Ludovica had 20% headspace, a measured temperature at filling of 44–
46°F (6.67–7.8°C), and a TBC concentration of 1020–1077 parts per million (“ppm”).
They had also been provided by Stolt. They were loaded onto the Ludovica on June
23, 2012. The total duration of the voyage was about 29 days. But, none of these
containers underwent thermal runaway. The Ludovica DVB was added to the
ADPO storage container, causing the level of the ADPO storage tank to reach its
highest level of the year.

The Court also notes that the Ludovica was one of five ships—not including the Flaminia—that
carried Deltech’s DVB80 from New Orleans to Europe during the summer of 2012. All carried
DVB80 manufactured between May 3 and June 18, 2012. None of the shipments, other than that
aboard the Flaminia, underwent thermal runaway, and none was above 25°C upon arrival at the
storage facility (ADPO) in Belgium. (Johnson Decl. ¶¶ 136–37.)
16

31

Dr. Davis found the Ludovica shipment especially informative. (Davis Decl.
at 57, ¶ 187.) The Court agrees. The DVB80 was manufactured during the same
campaign as that shipped aboard the Flaminia, yet it was shipped to France
without reaching thermal runaway. In fact, given its shipment dates, the Ludovica
DVB80 had five fewer days to chill and oxygenate in the MV-804 storage tanks than
did the Flaminia DVB80. (Id.)17
E. Dr. Davis’s Tests of the Mixing in the MV-804 Tank
Dr. Davis performed various tests to determine the oxygen saturation level in
DVB80 achieved by Deltech’s DVB80 manufacturing process—the same process
used for the May 19–June 11, 2012 campaign. The Court found this testing rigorous
and persuasive. Dr. Davis performed CFD modeling to determine how the oxygen
and TBC would mix in the DVB80 liquid within the MV-804 storage tank. (Davis
Decl. at 38, ¶ 113.) In his trial declaration, and as further explained during live
examination, he presented a persuasive set of models that showed the flow pattern
and velocity at which the oxygen mixed into the DVB fluid.

Additionally, Dr. Ott, plaintiffs’ expert, agreed that the Ludovica shipments were less oxygenated
than the DVB80 shipped in containers aboard the Flaminia. (Tr. at 226:3–8.)
Some polymer was detected in the ADPO storage tank after the arrival of the Ludovica
shipment. The amount detected was well below the specification values of 50 ppm for quality
assurance standards. (Davis Decl. at 57, ¶¶ 187–189.) The low level of polymer detected in the
storage tank at ADPO is not indicative of the onset of polymerization during the Ludovica shipment.
A more likely explanation is that the formation of polymer occurred in the ADPO storage tank over
time above the liquid surface, and when the levels within the storage tank reached their highest
levels that year with the infusion of the Ludovica liquid, the polymer that had formed on the sides of
the tank washed into the tank itself. (Id. at 57, ¶¶ 190–91.) The Court is persuaded by this
explanation. Dr. Davis, who had the requisite expertise, describes the way in which monomer vapor
could polymerize on the side of the tank and form a soluble layer, and explains that there might also
be stalactites on the sidewalls and roof of the tank. (Id. at 57–58, ¶ 191.)
17

32

Figure 2.

(Id. at 40–41, ¶ 119.) Figure 2 depicts the path and residence time of massless
tracer particles.18 (Id.) The left diagram depicts the path of the tracer particle that
stayed in the tank for the shortest amount of time, 17 minutes; the right diagram
depicts that path of the tracer particle that stayed inside the tank for the longest
amount of time, 42 hours. From this experiment, Dr. Davis concluded that particles
in the MV-804 tank completely mix from top-to-bottom in the tank and that there is
no evidence of stagnant layering (that is, no DVB80 liquid remaining in one place
without mixing with TBC and oxygen). The mixing ensured that there were near
homogenous oxygen levels at all layers (that is, heights within the tank), and that
DVB80 was also periodically rising to the surface and mixing with air. (Id. at 41, ¶¶
120–123.)

These particles have no physical effect on the liquid inside the tank, but they follow the flow and
allow analysts—and the Court—to visualize flow paths within the storage tank. (Davis Decl. at 40, ¶
119.)
18

33

When he visited Deltech’s facilities, Dr. Davis also used detectors19 to
measure the concentration of oxygen in the headspace of the MV-804 storage tank,
the recirculation line, and the ISO container after filling. (Id. at 43–53, ¶¶ 131–
165.) These measurements confirmed that DVB80 liquid is exposed to oxygen
during the manufacturing process, and that oxygen saturation is at the predicted
levels at each step. In addition, Dr. Davis found that additional oxygen continued
to be mixed into the liquid after it had been filled into the ISO container and during
transport. Such additional mixing (and thus additional oxygenation) resulted from
the fact that, as stated above, as of 2012 and thereafter, Deltech only fills containers
to near or at 80% capacity, leaving approximately 20% headspace that fills with
oxygen. The liquid can thus “slosh around” (not a technical term) and mix with that
oxygen in the headspace, providing for additional oxygenation.
Dr. Davis’s testing confirmed that the DVB80 in Tanks I, J, and K was
delivered to NOT for loading onto the Flaminia in what the Court considers a fully
saturated condition—that is, at 94–95%. He testified that this was a minimum and
likely conservative oxygen saturation level. The Court agrees. In particular, the
Court is persuaded that the manufacturing process generally used by Deltech, and
pursuant to which hundreds of voyages had been successfully completed without
incident, was utilized here.
On the morning of June 21, 2012, Tanks I, J, and K were filled with DVB80.
After being filled with DVB80, Tank I had 20.2% headspace; Tank K had 17.5%

19

These detectors had not been available in 2012.

34

headspace; and Tank J had 17.8% headspace. (Stipulated Facts at 7–8, ¶¶ 47, 53,
59.)
F. Filling Tanks I, J, and K
Stolt provided Tanks I, J, and K. An ISO container is a cylinder-shaped tank
fixed into a 20-foot long, 8-foot wide, 8.5-foot high rectangular frame. (Nell Decl.,
ECF No. 1296, ¶ 5.) ISO containers are made to certain standards and the
containers that were shipped aboard the Flaminia were similar to one another. (Id.
¶ 6.) The bodies of the tanks were constructed of stainless steel, then covered by a
layer of insulation and a protective cladding. (Id. ¶ 8.) The containers have a topside manhole, a pressure relief valve, a top discharge valve, a thermometer on the
lower portion of the rear-side of the tank, and a bottom discharge outlet on the rearside of the tank. (Id.)
Figure 3.

35

Each of Tanks I, J, and K had been periodically inspected and found to be in
working order. (Id. ¶¶ 14–17.) Through the date of the incident, the containers had
been regularly serviced and maintained. (Id. ¶ 17.) Each of Tanks I, J, and K were
set in a rectangular frame made of carbon steel. Because the frame is necessarily
larger than the container itself, when two containers are stacked on top of one
another, the space between the top of the frame on the lower tank and the bottom of
the frame on the upper tank is only approximately 3.1 inches. (Id. ¶ 9.) Under
certain conditions, such as occurred at NOT and aboard the Flaminia, this
proximity allows for radiative and convective heating.20
If and how heat can enter or exit an ISO container (that is, how well the ISO
container absorb and/or emit heat) is highly relevant to the causation issues before
the Court. Given oxygen saturation, the next question is whether heat conditions in
and outside of the ISO containers of DVB (and DPA) contributed to polymerization.
To understand this, the Court next turns to a discussion of “UA” values of the ISO
containers. The UA value is the overall heat transfer coefficient, a measure of the
ability to transfer heat into and out of an ISO container. Mathematically, it is the
inverse of thermal resistance. Therefore, if an ISO container has a higher UA
value, this indicates that heat is more easily transferred into or out of it (i.e., that
the container has a lower thermal resistance). (Davis Decl. at 60, ¶ 2(a).) Thermal
resistance of the relevant ISO containers—or, their ability to insulate DVB80—is a

Convection is the transfer of heat by movement, or circulation, of liquids or gases. Radiation is
energy emitted from a surface as particles or waves; it does not require a medium (such as a liquid or
gas) to transfer heat.
20

36

key parameter in predicting the temperature history of the DVB80. There are
various ways to measure UA values for ISO containers. Ultimately, the Court
credits that performed by Dr. Davis, which was adopted by Dr. Kaminski.21
Dr. Davis arrived at a UA value of 39 W/K for Tank I. This was calculated
through a Full-Scale Test (described below) using an ISO container of the same
make and model as Tank I. (Davis Decl. at 63, ¶ 10.) Dr. Davis examined the
average DVB80 temperature measured by four thermocouples (that is, a device that
acts as a thermometer), which yielded a UA value of 36 W/K. The two
thermocouples deepest in the tank, however, yielded an average UA value of 39
W/K for Tank I—this represents a conservatively high estimate of the overall
coefficient, as well as the shortest induction time to auto-polymerization. (Id. at 67,
¶¶ 20–21.) Gexcon calculated the UA values for Tanks J and K as approximately 48
W/K. (Id. at 60, ¶ 2(a).)
Dr. Kaminski work examined the methods proposed by various experts in the
case, and carefully and thoroughly analyzed whether their predicted thermal
resistance values were supportable. Ultimately, she determined that the “full scale
value of 39 is likely to be the most accurate because the container measured was
made on the same assembly line as Tank I of the incident voyage, was the next
container manufactured after Tank I, and closely resembles it physically.”
(Kaminski Decl. at 15, ¶ 39.) When combined with other measurements and

The Court agrees with the criticisms of Drs. Davis and Kaminski of Dr. Ott’s work. Dr. Ott arrived
at a UA value of 70 W/K, but this was derived using inaccurate assumptions such as 26,000 liter
tanks (as opposed to 25,000 liter tanks) that were 70% (as opposed to 80%) filled. (Ott Decl. ¶ 102.)
21

37

modeled, these UA values support the Court’s findings as to heat contributions into
Tanks I, J, and K.
G. Prior DVB Incidents
Much of the foregoing discussion indicates to the Court that the DVB80 was
sufficiently oxygenated and chilled when it left Deltech’s facility—as such,
something else must have caused the DVB80 to auto-polymerize and achieve
thermal runaway aboard the Flaminia. But in addition, the parties spent
substantial time at trial delving into previous incidents (all in 2006) that resulted in
auto-polymerization of Deltech’s DVB63 and DVB80. Stolt and Deltech argue that
these incidents are informative because, afterward, Deltech changed its
manufacturing process for DVB80 by, inter alia, increasing the amount of
headspace in ISO containers to about 20%; since these changes, Deltech shipped
DVB63 and DVB80 hundreds of times without incident. Plaintiffs, however, point
to at least two of these incidents (Chauny and Grangemouth) as useful analogies;
they attempt to draw out similarities between these incidents and what occurred
aboard the Flaminia.
Ultimately, the Court is persuaded by Deltech and Stolt. As the Court will
explain, each 2006 incident (described below) is distinguishable from the Flaminia
incident. Moreover, between 2006 (when Deltech changed its procedures)22 and
2012 (when the incident occurred aboard the Flaminia), none of the ISO containers
filled by Deltech underwent thermal runaway.

22

Deltech’s changes are described in more detail below.

38

Travis Johnson, Deltech’s Quality Assurance and Laboratory Manager,
testified at trial. Following the incident aboard the Flaminia in July 2012, Johnson
was tasked with investigating whether the DVB shipped aboard the Flaminia had
been filled in ISO containers in a manner consistent with Deltech’s ISO container
filling procedures. (Johnson Decl., ECF No. 1305, ¶ 13.) He was also responsible
for ensuring that those practices had been consistent over time. (Id.) These
assignments fell squarely within his duties and responsibilities in Quality
Assurance. As part of his investigation, Johnson reviewed documents maintained
in Deltech’s files and was able to provide information regarding Deltech’s history of
manufacturing and shipment of DVB. The Court found Johnson credible and relies
on his testimony.
Deltech commenced manufacturing and shipping DVB80 in 2006. It had
begun manufacturing and shipping a lower grade of DVB—DVB63—in 2005. (Id. ¶
15.) In 2006, at a time when shipping DVB was still relatively new for Deltech,
Deltech experienced instances of shipments auto-polymerizing or product arriving
at their destinations with elevated polymer content (a cause for concern as the
formation of polymers happens with increases in temperature; and as set forth
above, polymers beget polymers, hence the “runaway” polymerization risk).
Following these incidents in 2006, DVB made changes to its manufacturing
and shipment procedures. First, prior to loading, Deltech chilled the DVB at a
lower temperature than it had previously. Second, Deltech increased the empty
headspace between the liquid DVB in an ISO container and the top of the tank from

39

10% to 20%. (Id. ¶ 17.) Headspace fills with oxygen, and when the tanks are moved
(whether due to transport by truck or shifting during voyage), the oxygen then
mixes with the liquid, allowing for additional oxygenation. Until 2012, Deltech had
not had additional incidents involving polymerization following these changes. In
2006, changes in the manufacturing process were implemented.
Since Deltech first began manufacturing DVB in 2005, it has made almost
800 overseas shipments without incident. (Id. ¶¶ 24, 117 tbl. 2.) Among all of the
hundreds of shipments Deltech has made, the only shipment that underwent
thermal runaway reaction during an ocean crossing was that aboard the Flaminia
in 2012. (Id. ¶ 25.) As set forth below, however, during the first summer that
Deltech shipped DVB80 in 2006, there had been several incidents. But all involved
a manufacturing process different from Deltech’s procedures in 2012, and certain of
them involved DVB in uninsulated drums (versus insulated ISO containers).
A lack of insulation fails to provide protection against heat. (Kaminski Decl.
at 22, ¶ 2.) If drums containing DVB are left in the sun, the steel drums will absorb
heat, which will easily and readily be transferred to the DVB. If left unabated,
runaway polymerization can occur. (Id.) In addition, drums have a lower capacity
than ISO containers and thus, when filled, hold a lesser amount of liquid. This
would impact the time to auto-polymerization or induction time because the more
DVB in a container, the more DVB there is to absorb energy. Further, drums have
less surface area, increasing the ratio of the surface area to mass of the DVB in the
container. Because heat is transferred where the container’s surfaces come into

40

contact with the DVB, the greater the surface area ratio (as in a drum versus in an
ISO), and the faster heat will be absorbed by the DVB, causing its temperatures to
increase. (Id. at 23, ¶ 4.)
Dr. Kaminski finds, and the Court agrees, that five of the seven incidents
discussed below are not as analogous as plaintiffs claim. (Id. at 24, ¶ 7.) The
remaining two incidents—the “Chauny” and “Grangemouth” incidents—are useful,
and they support the view that the Flaminia DVB80 cargo would have completed
the voyage safely under normal conditions.
1.

Lanxess

On July 3, 2006, Deltech shipped an ISO container of DVB80 trans-Atlantic,
to its customer Lanxess Deutschland GmbH (“Lanxess”). The ISO container had
been filled to 90% capacity. Its temperature at filling was 60°F (15.5°C). The
product did not vent during the voyage.
The ISO containers were delivered to Lanxess on July 31, 2006. Lanxess
notified Deltech that the shipment had elevated polymer levels. (Johnson Decl. ¶
31.) In fact, DVB polymer clogged the discharge valve. Despite this, the customer
(Lanxess) was able to filter the polymer product and use it. (Id.) Deltech
determined that the product that had been sent to Lanxess had spent very little
time mixing and chilling in the MV-804 storage tank. (Id. ¶ 34.)
The Court finds that the Lanxess incident is not analogous to the DVB80
aboard the Flaminia. The DVB shipped to Lanxess had been subjected to different
chilling and fill level procedures than the product aboard the Flaminia.

41

2.

Rohm & Haas

The second DVB shipment with polymerization issues occurred only three
days later, on August 3, 2006. On June 15, 2006, Deltech had filled an ISO
container shipment of DVB63, in an ISO container with a 90% fill level bound for its
customer, Rohm & Haas, located in Chauny, France. (Id. at 8–9, ¶¶ 40–42.) The
ISO container of DVB63 was transported by truck to NOT. At the terminal, it was
exposed to average daily temperatures of 83°F (23°C). It took 30 days for the
shipment to reach Chauny. Upon arrival on August 3, 2006, it had elevated
polymer levels. However, it had not achieved thermal runaway.
The Court does not find that this incident is analogous to what occurred
aboard the Flaminia. First, the product had a higher fill level (and therefore less
headspace) in the ISO container, and second, it had not been chilled to the same
extent in the MV-804 tank as the DVB80 shipped aboard the Flaminia.
3.

Ashland/Amwar in Brook Park, Ohio

Also on August 3, 2006 (the same day as the Rohm & Haas shipment above),
Deltech’s customer, Ashland, Inc. (“Ashland”) in Brook Park, Ohio, reported to
Deltech that a shipment was undergoing polymerization. Ashland had received
twenty uninsulated drums of DVB80. (Id. at 10, ¶ 48.) When the drums were filled
with the DVB80, the temperature measured 54°F (18°C); this is 10°F higher than
was the Flaminia DVB80 when it was initially filled into Tanks I, J, and K. The
drums were also filled to 90% capacity. From July 13–18, the drums were trucked
from Baton Rouge to Ohio. Sixteen days later, on August 3, 2006, Deltech received

42

notification that the drums were undergoing polymerization. Deltech did not
receive information regarding storage conditions between July 18 and August 3.
The Court does not find the circumstances of the Ashland shipment relevant
to the DVB80 aboard the Flaminia. First, the DVB80 was at a higher initial
temperature when it was filled into uninsulated drums, not insulated ISO
containers. In addition, the lack of information regarding storage conditions
eliminates the Court’s ability to draw conclusions regarding the cause of the
polymerization.
4.

Whitehall Township, PA

On July 10, 2006, at its facility in Baton Rouge, Deltech filled four,
uninsulated 55 gallon drums with DVB63 for its customer, Bowden Chemical Ltd.
(“Bowden”). Bowden arranged for the transport of the drums and did not pick them
up for nine days—that is, until July 19, 2006. (Id. at ¶ 54.) The temperature in the
truck in which the drums were shipped was measured at 120°F (48.9°C)—far higher
than the recommended 85°F. On August 6, 2006, Deltech received notice that the
product shipped to Bowden auto-polymerized.
The Court does not find that the circumstances with regard to the Bowden
product are relevant to those at issue here. First, the product was filled into
uninsulated drums, and shipment temperatures were higher than they should have
been. However, it is unclear how those truck conditions compared to temperatures
at NOT (influenced by solar radiation) aboard the Flaminia.

43

5.

Chauny, France

The parties spent significant time at trial on the circumstances relating to
this shipment, referred to as the “Chauny” incident. This incident relates to
another shipment of DVB product to Deltech’s customer Rohm & Haas. (Id. at 12, ¶
57.) The salient facts relating to this incident are as follows: Deltech filled an ISO
container with DVB80 on July 3, 2006. At the time of filling, the temperature of the
DVB80 was measured as 60–61°F (15.5–16.1°C). The DVB80 had been mixed and
chilled in the product storage tank for, at most, five days. These two facts—from
the start—distinguish this shipment from Tanks I, J, and K aboard the Flaminia.
The fill temperature was 16°F higher than Tanks I, J, and K, and the Chauny tanks
had spent five to six fewer days mixing with oxygen and chilling in the MV-804
tank.
After filling, the Chauny ISO container was transported by truck to Houston,
Texas. It sat on the dock in Houston for an additional 12 days—from July 4–15,
2006. The average temperature during that period was 85°F (29.4°C); certain days
experienced high temperatures of over 95°F (35°C). This was two days longer on
the dock than Tanks I, J, and K (with somewhat similar ambient temperatures), but
there is no indication that on the dock, this tank was proximate to tanks of heated
DPA. The ISO container was then loaded onto a vessel for trans-Atlantic shipment.
The voyage carrying this ISO container lasted 17 days—from July 15 to August 1,
2006. Upon arrival, the container sat for several additional days until it began the
final leg of its trip to Chauny, France. On August 7, 2006, and while en route to

44

Chauny, the ISO container began to vent. It was determined that the DVB80 was
experiencing auto-polymerization. It never exploded.
There are certain notable similarities between the Chauny incident and that
onboard the Flaminia. Both products sat on a dock with ambient temperatures in
excess of 80 degrees for a period of days prior to loading, and the voyages on both
trips were a similar duration. These similarities are of particular interest insofar
as the Chauny product made it safely across the ocean, and only began to vent once
en route to Chauny. Moreover, it never exploded. Additionally, the Flaminia
product auto-polymerized in 22.75 days, versus the 36 days that it took for the
Chauny product to achieve runaway auto-polymerization.
There are, however, significant differences. It is significant that again, the
fill and chill conditions during the loading and manufacturing processes were
different: the Chauny product was chilled in the day tank for fewer days than that
shipped aboard the Flaminia and it was filled to 90% in the ISO container. Also,
the Chauny product was loaded aboard the vessel 12 days after initially being
filled—two days longer than the product onboard the Flaminia.
6.

Grangemouth, Scotland

Another auto-polymerization event that received extensive attention during
the trial occurred in Grangemouth, Scotland. An ISO container was filled with
DVB63 on June 23, 2006. (Id. at 15, ¶ 74.) Sixty-two days later, on August 23,
2016, it achieved thermal runaway. The temperature at the time the ISO container
was filled was 48°F and it was filled to 90% capacity. At the time of filling, the

45

product had only chilled for a few days (fewer than the 10 allowed for Tanks I, J,
and K). After filling, the tank was then transported to NOT, where it sat for 11
days. That is to say, it was not loaded onto the ship until 12 days after it had been
filled (one day longer than the ISOs destined for the Flaminia sat at NOT). During
the time at NOT, the average daily temperature was 85°F (29.4°C); again, during
the day, certain temperatures exceeded that by several degrees. The container was
then loaded onto a vessel and successfully completed a trans-Atlantic voyage.
It arrived in Antwerp, Belgium on July 4, 2006; it was then transported to
the Netherlands before being placed aboard another vessel for transport to
Scotland. (Deltech Ex. 152 at 1.) During this period, the ISO container was
exposed to high ambient temperatures between 85–100°F (29.4–37.8°C). On August
23, 2006, 41 days after it had been off-loaded from its trans-Atlantic voyage, the
DVB63 tank achieved thermal runaway in Grangemouth.
There are notable differences and some similarities between the
Grangemouth incident and what occurred aboard the Flaminia. In terms of the
similarities, the Court notes that the Grangemouth ISO container sat at NOT for a
period of time, exposed to ambient temperatures in excess of 80 degrees prior to
loading. But the differences must not be ignored. As already discussed, the filling
and chill procedures were different. Additionally, the Grangemouth shipment
completed the trans-Atlantic voyage without incident.

46

7.

Rohm & Haas Incident #3

The final auto-polymerization incident that occurred in 2006 related to
product again destined for Deltech’s customer Rohm & Haas. This incident did not
result in thermal runaway, but in elevated polymer levels rendering the product
delivered to Rohm & Haas “off-specification.” (Johnson Decl. at 18, ¶ 93.) Here,
again, there are differences between these circumstances and what occurred aboard
the Flaminia. First, the chill and fill procedures were different—the DVB63 chilled
for fewer days and the ISO container was filled to 90%. The ISO container made
the trans-Atlantic crossing without incident. It was subsequently off-loaded and
stored in a depot in the Netherlands for 41 days. On August 23, 2006, 63 days after
filling, the ISO container was delivered to Rohm & Haas. That same day, Rohm &
Haas informed Deltech that the DVB had elevated polymer levels and was offspecification.
The Court finds that this incident is not analogous to what occurred aboard
the Flaminia. The product had the same differences in chilling and fill level as
discussed elsewhere, it made the trans-Atlantic voyage successfully, sat at a depot
in unknown conditions for 41 days, and even then, never achieved thermal
runaway.
8.

Changes to the Manufacturing Process After the 2006 Incidents

The above incidents confirm the Court’s view that what occurred aboard the
Flaminia was unusual and not the result of the Deltech manufacturing process.
These incidents reveal what occurs when the manufacturing process has issues: all

47

of the incidents occurred during a 24-day period in 2006, and none occurred in the
years after Deltech changed its manufacturing and fill process.
Following the 2006 incidents described above, Deltech changed certain
aspects of its manufacturing and fill processes. In 2007, Deltech roughly doubled
the amount of headspace in its shipments (from 10% headspace to approximately
20% headspace), allowing more oxygen into the tank to mix with the product.
Deltech also implemented additional chilling procedures. It began chilling the
DVB80 to cooler temperatures: from about 61°F (16.1°C) to 45°F (7.2°C). This
increases the time that it would take to heat up the product and thus increase the
time to polymer formation or thermal runaway. When product arrives at the
storage facility in Belgium (“ADPO”), its temperature is measured and is
consistently below 25°C. A report of the temperature of the Deltech DVB product
measured at ADPO for 2012 does not show any product above 25°C. ADPO also
measures polymer levels of the Deltech DVB product. Other than the Flaminia
shipment, none of the 300 shipments that ADPO has measured experienced a
thermal runaway or were found to have polymer levels elevated above the
customer’s specification.
The Court also finds persuasive evidence that shows that other containers of
DVB80, filled in approximately the same timeframe as the Flaminia’s containers,
did not experience thermal runaway. Between May 3, 2012 and June 18, 2012,
Deltech filled thirteen other ISO containers with DVB80. These tanks were
destined for shipments aboard five other vessels, also departing from NOT. When

48

the contents of these ISO containers were transferred to the ADPO storage tanks,
the DVB within the storage tank was within specification.
H. The Conditions at New Orleans Terminal
After being filled with DVB 80 on June 21, 2012, Tanks I, J, and K were
trucked to NOT.23 In light of other factors, Deltech’s fateful decision to ship Tanks
I, J, and K out of NOT—as opposed to a more northern port—was a substantial
contributing factor in the auto-polymerization event aboard the Flaminia.
(Stipulated Facts at 16, ¶ 1.)
Deltech utilizes various ports and routes for shipping its product overseas.
Shipping via NOT at New Orleans takes longer to reach Antwerp than do other
routes. (Deltech Trial Ex. 139 at 18.) Based on its investigation into other autopolymerization events that had occurred in 2006, Deltech understood that the
combination of time-to-destination and temperature exposure risked the stability of
the product. As it was summer (June 2012), temperatures at NOT could have
almost certainly been expected to be higher than the recommended maximum of
85°F. The choice to ship from NOT therefore risked (1) a longer time-to-destination,
and (2) higher NOT exposure temperatures than a more northern port (such as
Newark’s).
A Deltech document set forth transit times as follows:

This transport process would have additionally oxygenated the DVB80 as it “sloshed” around the
tank. (See Kaminski Decl. at 34, ¶ 10.)
23

49

Route
New York to Antwerp
Charleston to Antwerp
New Orleans to
Antwerp

Table 1.
Range Number
Median
Maximum
of Days in
Number of
Number of
Transit
Days in Transit Days in Transit
7–12
9.5
13
11–13
12
16
15–17

16

31

(Id.) In fact, after the 2006 auto-polymerization incidents, Deltech implemented a
policy of shipping DVB80 to Europe only through Newark during the hottest time of
year: April 16 to November 14. (Pl.’s Ex. 59 at 2–3.)
In March 2012, when a Deltech employee was consulted about shipping
DVB80 from NOT to Antwerp, he stated: “Since we have no control of the container
once it leaves the plant and the possibility of sitting on a dock at 100 F, I vote no.”
(Pl.’s Ex. 141 at 2.) Then, forty minutes later: “If I understand the transit route the
shipment from NO [New Orleans] goes south and exposure time to heat is much
longer. We changed to the NJ route to avoid that exposure in the summer.” (Id. at
1.) When asked at his pre-trial deposition whether it was safer to ship out of
Newark, the same employee unequivocally answered “Yes.” (Fluharty Dep. Tr. at
224:19–23.) He also stated, “If we were going to ship out of New Orleans? That
[sic] we probably should ship in refrigerated containers in hot weather conditions.”
(Id. at 222:19–23.) Based upon a desire to fulfill a customer order as soon as
possible, however, Deltech’s owner and President, Bob Elefante, nonetheless
decided to ship from NOT. (Levine Dep. Tr. at 106:4–9; id. at 237:2–18; id. at
240:19–25; id. at 241:2–16.)

50

Tanks I, J, and K arrived at NOT on June 21, 2012 and sat still on the dock
until July 1, 2012, when they were loaded aboard the Flaminia. A principal
question in this Phase 1 trial is whether this 10-day period of storage at NOT
substantially contributed to the auto-polymerization of the DVB80 on July 14, 2012.
The Court finds that there is convincing evidence that it did.
As discussed below, several aspects of the storage at NOT lead to this
conclusion. First, once on the dock, the liquid became still—previously, it had been
circulated in a tank and sloshed around in a truck. This diminished ongoing oxygen
diffusion. Deltech had previously recognized such circumstances as having been a
contributing factor to the 2006 auto-polymerization events. (Pl.’s Ex. 113 at 3.) In a
Deltech email analyzing the cause(s) of such events, an employee noted:
One thing that all of our DVB polymerization incidents had in common
besides temperature was that they were held quite still for some days or
weeks. As the temperature of the bulk liquid rises, the rates of TBC and
oxygen consumption rise. If the solution is still, then the only method
for oxygen to be distributed through out the bulk liquid is diffusion.
Diffusion is not a very fast method of transportation and so it is possible
that at some temperature, the oxygen is being consumed faster than it
can diffuse into the entirety of the liquid. The implications of this are
simple: for a container sitting still at an average temperature, if the
oxygen is consumed at the bottom of the container, where it is farthest
away from the headspace air, then the inhibition is lost in the bottom of
the container, and thus the polymerization could begin unchecked by
TBC.
(Id.)
In addition, at NOT, the DVB80 containers were stored outdoors and exposed
to consistently warm ambient temperatures. On June 21, the high daytime
temperature was 92°F (33.3°C), and the average temperature was 85°F (29.4°C).

51

On June 26, 27, 28 and 29, the high temperature reached 96–98°F (35.5–36.6°C).
(Earle Decl., ECF No. 1289, ¶ 3.)
Chemtura’s expert, Douglas J. Carpenter, also testified that while at NOT,
the DVB containers were subjected to temperatures substantially higher than the
recommended 65°F (18°C). (Carpenter Decl., ECF No. 1293, ¶ 43.) He further
stated:
the DVB containers exposed to the sun at NOT also would have received
thermal energy from the uninsulated spill box manlid on the top of the
shilling containers. This exposed section of the tank can transfer more
thermal energy since it is not thermally insulated. This is especially
true in the scenario where an ISO container is located on the top of the
pile with exposure to direct thermal radiation from the sun for an
extended period of time (in this case, approximately 10 days).
(Id. ¶ 44.) He added, “This solar contribution would be analogous to having a
heated plate inside the container with a surface temperature of 90°F to heat the
vapor space and ultimately the DVB liquid in the container.” (Id.)
Based on its prior investigations into the 2006 auto-polymerization incidents,
Deltech recognized that exposure to high temperatures was a contributing factor.
For instance, a Deltech “Root Cause Analysis and Corrective Action Report” stated:
Root Cause 1
Shipping:
The product was left still and unmonitored in Antwerp for 30 days,
which is not recommended. The weather in Antwerp was not above the
temperatures that Deltech normally stores its DVB, but the fact that the
materials was [sic] not handled, circulated, aerated or sampled AND had
seen about 15 days of temperatures in the 80’s leads us to believe that
the material could have had significant polymer formation during
shipping and handling. Other incidents with DVB . . . lend support to
the main cause as improper handling leading to polymerization.

52

(Pl.’s Ex. 85 at 2.) Another “Root Cause Analysis” for a separate incident found
similarly:
Root Cause 1
Polymer formation during shipping and handling:
The product was held in ambient conditions unsuitable for storage of the
material for 66 days. The most likely cause of this autopolymerization
is the consistent elevated temperature.
(Pl.’s Ex. 87 at 2.) An additional Deltech study concluded that DVB transportation
times needed to be kept to 15–20 days, and less if the temperature was greater than
85°F:
The conclusion is simple and inconsistent with current handling and
storage procedures included in either Deltech or Dow’s MSDS or product
bulletins. We believe that DVB has to be handled in a controlled
environment where periods without direct care are kept to a minimum
typically less than 15–20 days. The 15–20 days is based upon the
product being exposed to average daily temperatures in the 75-85
degrees F. Should temperatures exceed 85°F then this time without
direct care needs to be shortened. When temperature exposure is less
than 75°F then the time without direct care can increase. However, we
do not recommend leaving the product without direct care for more than
35 days.
(Deltech Ex. 139 at 3.)
Third, at NOT, the DVB80 containers were adjacent to containers of heated
DPA and other cargo. The following diagram shows the arrangement of the DVB80
containers (Tanks I, J, and K) and the relevant DPA containers (Tanks L, M, and
N):

53

Figure 4.

(Davis Decl. at 17, ¶ 9.) As evidenced from the diagram, while on the dock at NOT,
Tank I was exposed to thermal radiation from three neighboring ISO containers
filled with heated DPA: Tanks L, M, and N.
It must be kept in mind that the ISO containers are 20 feet long and thus
“border” adjacent containers in close proximity for a full 20 feet. The DVB80 ISO
container on the top left, designated Tank J, was exposed to solar radiation and had
a narrow line of exposure to one neighboring ISO container filled with heated DPA
(Tank N); Tank K, on the top right, was exposed both to direct solar and thermal
radiation from the heated DPA in the ISO container beneath it, Tank N; and Tank

54

I, in the middle on the left, was exposed to direct solar and thermal radiation from
two heated DPA containers, Tanks N and L, as well as Tank M through a narrower
line of exposure. (See Kaminski Decl. at 17, ¶ 42.)
I. Measuring the Effect of the Conditions at NOT on Tanks I, J, and K
To analyze the effect of these conditions on the temperature of the DVB80
and DPA while they sat at NOT for 10 days, Dr. Kaminski performed various
simulations. She broke the ISO container’s thermal resistance down into “internal
thermal resistance” and “external thermal resistance.” (Kaminski Decl. at 18, ¶ 44.)
Internal thermal resistance consisted of natural convection of DVB80 with the
inside wall of the container, conduction through the container wall and various
support members, and conduction across the insulation. External thermal
resistance consisted of natural convection and radiation from the outer surface of
the 20-foot container. Both resistance values are needed to determine the surface
temperature of the container. (Id.)
At the time the ISO containers were filled, as set forth above, the DVB80 in
the ISO containers was measured at 44°F (6.67°C), with the appropriate amount of
TBC. Dr. Kaminski considered the particular position of each container, the
average air temperature at NOT, and the temperature of the DPA. She included
the absorption of solar energy by the ISO container while at NOT. (Id. at 19, ¶ 48.)
She calculated the absorbed solar flux on an hour by hour basis and considered any
portions of the ISO containers exposed to the sun, including ends and flat panels.

55

(Id.) She also took into account the angle at which the sun struck the surface of the
container. (Id.)
While the ISO containers were stored at NOT, there was also metal grating
above them. Dr. Kaminski took this structure into consideration in her analysis by
considering that the metal grating on top of Tanks J and K acted to partially block
the incoming sunlight. (Id. at 19, ¶ 49.) Her model further accounted for the
varying position of the sun, the night sky temperature, and the absorptivity of the
paint on the exterior of the container itself. (Id. at 19, ¶¶ 49–51.)24
J. Properties and Loading Condition of DPA Shipments
The DPA has a melting point of 125.6°F (53°C). (Stipulated Facts at 9, ¶ 65.)
It is solid until heated. To fill an ISO container with DPA, the DPA is first heated
to a level above its melting point. On June 21–25, 2012, as they were filled with
DPA, Tanks A–F, L, M, N, and P, had temperatures between 71–74°C. (Stipulated
Facts at 9–11, ¶¶ 67–76.)
On cross-examination, Chemtura’s expert, Douglas J. Carpenter, agreed that
while stored on the dock at NOT, the ISO containers of heated DPA would have
made a thermal contribution to the temperature of the adjacent ISO containers of
DVB. (Tr. 1198:9–18.) He agreed that the pavement underneath the ISO
containers, as well as the ambient temperature in the hold, may have caused the

The Court agrees with Dr. Kaminski’s criticisms of the work of Dr. Ott regarding the impact of
solar radiation on the liquids in the ISO containers. She correctly points out that his work does not
include a radiation model, the solar radiation that follows from the exposure of portions of an ISO
container to the sun, and the liquid capacity of the container and its impact on absorptivity.
(Kaminski Decl. at 20, ¶ 54.)
24

56

DPA to increase or maintain its temperature. (Tr. 1189:20–22, 1190:1–3, 1196:10–
1197:1.)25 Carpenter also acknowledged that ISO containers could receive thermal
energy from the uninsulated spill box and manlid. (Carpenter Decl. at 18–19, ¶ 44.)
While he was positing that this would have led to thermal energy heating the DVB,
his conclusion in this regard is applicable to the ISO containers that were filled with
DPA. In addition, he opined that when ISO containers are stacked on one
another—as was the case here with the DPA and DVB containers at NOT—solar
radiation of one container could, by radiative heating and/or convection, contribute
to the temperature of the contents of an adjacent tank. (Id.) He gave this testimony
in connection with opinions regarding the effect of solar radiation hitting a tank,
and then being emitted from an uninsulated spill box or manlid. However, the
evidence at trial supported solar radiation hitting certain ISO containers along the
sides and ends, as well as the tops. In these circumstances, which Carpenter did
not specifically address in his report, his conclusions are applicable to ISO
containers of DPA impacted by solar radiation, and emitting heat through the
uninsulated spill box and manlid to the adjacent containers of DVB.
In sum, the ISO containers of DPA arrived at NOT in a heated, liquid state,
with a temperate above 125.6°F (53°C). The DPA was itself then further exposed to
warm ambient temperatures, solar radiation, radiation from the pavement, and

In his trial declaration and at various points during his examination, Carpenter opined that the
casualty would have occurred independent of the presence of the DPA. (See, e.g., Carpenter Trial
Decl. at 15, ¶ 38 et. seq.) The weight of the evidence is to the contrary. Indeed, statements made by
Carpenter at trial, as recited herein, are supportive of the opposite conclusion.
25

57

radiative heat and convection from each other, which prevented the DPA liquid
from cooling as rapidly as it otherwise would have. As a result, the ISO containers
of DPA emitted heat through their tank walls, uninsulated manlids, and spill boxes,
contributing to the temperature of the liquid DVB80 in the adjacent containers.
The Court finds that given their proximity in storage, the ISO containers filled with
heated DPA were a substantial factor in the heat conditions that led to DVB
runaway polymerization.
Chemtura commissioned a study from Willbros Engineers LLC (the “Willbros
Study”). The purpose of the study was to determine how quickly DPA filled into an
ISO container at 175°F (79°C) would cool to its melting point, and to show that
sufficient cooling would have occurred to eliminate any temperature contribution to
the DVB80. (Carpenter Decl. at 21, ¶ 59; Carpenter Decl., Ex. C, ECF 1293-3
(“Willbros Study”).) The Court is not persuaded that this study was sufficiently
close to actual conditions to have utility. In the Willbros Study, an ISO container
was filled with heated DPA (with a measured temperature of 175°F) and placed in a
warehouse. The warehouse did not have mechanical ventilation, but from
photographs it is clear that it was far roomier open than the Flaminia and had a
large garage door that opened to the outside.
The DPA was 10°F (6°C) hotter than that in the ISO containers loaded
aboard the Flaminia. After seven days, the DPA had cooled to 127°F (53°C). The
exterior shell of the ISO container in the Willbros Study also never reached more
than 18°F (10°C) above ambient temperature.

58

The conditions in which the ISO container in the Willbros Study was placed
did not approximate those at NOT or aboard the Flaminia. First, and most
significantly, the ISO container was never placed out in the open, where solar
radiation would directly hit a portion of the container, and the container would be
exposed to ambient temperatures that fluctuated into the 90s (Fahrenheit). Second,
when in the warehouse, the container with DPA was not proximate to any other
containers with heated DPA that might have emitted some heat. When both on the
dock at NOT and in storage aboard the Flaminia, the ISO containers filled with
DPA were proximate to one another. For instance, DPA (N) was directly below DVB
(K), and directly adjacent to DVB (I). DVB (I) was also directly above another
container with DPA, DPA (L). DPA (M) and DPA (N) were each diagonal from DVB
(I) and (J), respectively. (See fig. 4.)
Finally, there was no evidence that the ambient temperatures in the
warehouse would have been similar to the ambient temperatures in Hold 4 of the
Flaminia. Indeed, the photographs of the Willbros Study show a large, garage-like
door that opened into the outside, suggesting a level of ventilation that was not
present within Hold 4. (Willbros Study at 18–20.) Taken together, the Court is not
persuaded that the conditions in the Willbros Study were sufficiently similar to
those at NOT or aboard the Flaminia to provide meaningful information regarding
the speed of cooling. Accordingly, the Court draws no conclusions based upon the
results of the Willbros Study.

59

Carpenter testified that the DVB was sufficiently saturated with oxygen that
but for the addition of heat, it would have arrived safely at its destination. (Tr.
1151:4–11; Carpenter Decl. ¶ 52.)
K. Stowage and Conditions in Hold 4 of the Flaminia
The cargo was loaded aboard the Flaminia on July 1, 2012. (Stipulated Facts
at 18, ¶ 16.) Hold 4 was one of the holds designated for dangerous cargo. The
containers with the DVB and DPA were loaded into this hold. Including the
containers of DVB80 and DPA and the other cargo, 262 containers were stored
below deck in Hold 4.26
Below is a diagram showing the arrangement of the relevant cargoes in Bays
29 and 31 of Hold No. 4 of the vessel:

The other cargoes stowed in Cargo Hold 4 on the Flaminia played no role in causing the DVB80 to
undergo a thermal runaway reaction and there is no evidence that they contributed to the explosion
and fire that occurred on July 14, 2012.
26

60

Figure 5.

(Id. at 22, ¶ 31.) This diagram shows that in Hold 4, Tank J was surrounded on two
sides by Tanks B and L, and diagonal to Tank D, all of which contained DPA.
Similarly, Tank A—of heated DPA—was stored directly on top of the least insulated
portion of Tank I, and Tank C was diagonal to it; Tank K had top, side, and
diagonal DPA containers.
The Flaminia was propelled by heavy fuel oil or diesel oil stored in wing
tanks along the port and starboard sides spanning from cargo hold 3 to cargo hold 7.
(Id. at 23, ¶ 2; see also Davis Decl. at 18, ¶ 11.) Wing tanks were therefore located
on either side of one of the holds designated for dangerous cargo, and the hold which

61

the heat-sensitive DVB80 was stored, Hold 4. The wing tanks were equipped with
heating coils, which were heated by steam generated in the engine room.
(Stipulated Facts at 31, ¶ 33.) The fuel oil is heated during the voyage (and was
during the voyage at issue here) in order to lower its viscosity so that it can be more
easily pumped. (Davis Decl. at 18, ¶ 12.) Dr. Davis, Dr. Kaminski, and Carpenter
all agreed that the fuel tanks adjacent to Hold 4 contributed to ambient
temperatures in the hold. (Id. at 100–01, ¶ 4; Kaminski Decl. at 27, ¶ 3; Carpenter
Decl. ¶ 49.)
Hold 4 was comprised of two principal compartments, or hatches, each of
which was approximately forty feet long and had multiple cells. (Stipulated Facts
at 28, ¶ 15.) Each cell was capable of holding two twenty-foot containers or,
alternatively, one forty-foot container lengthwise. (Id.)
The hold also had three mechanical exhaust ventilation fans to accomplish at
least two air changes per hour based on the volume of air in the hold when empty of
cargo, with all three fans being operated simultaneously. (Id. at 29, ¶ 19.) Under
normal circumstances, the nozzles and piping continuously sampled the air in the
holds, drawn by fans, as a smoke detection system. (Id. at 29, ¶ 20.)
During the Flaminia’s voyage, ambient outside temperatures were generally
in the 80s (from July 2–9), with one day in the 60s (July 12), and three days in the
70s (July 10, 11, and 13). (Earle Decl. ¶ 11.) The average daily temperature during
the Flaminia’s voyage was higher than that for the five previous trans-Atlantic DVB
shipments. (Id. ¶ 19.)

62

L. The Temperature Inside the Hold
The containers of DVB80 loaded into Hold 4 had already sat—motionless—on
the dock at NOT for ten days, exposed to summer ambient heat, solar radiation
from the sun, and thermal radiation from adjacent DPA containers. Neither the
temperature of the DVB80 liquid nor its polymer content were measured when
Tanks I, J, and K were loaded into Hold 4. No measurements were taken of the
ambient temperature or the cargo in Hold 4. Accordingly, the parties in this matter
presented expert opinions as to temperature based on known, likely, and theoretical
conditions. The most significant expert opinions combined the manufacturing
process, conditions at NOT, and conditions in Hold 4 to explain the likely cause of
runaway polymerization.
All parties prepared models based on the fact that on the day of the
explosion, July 14, 2012, Tanks I, J, and K had been in transit for 23 days. Notably,
this was already three days beyond the 15–20 day preferred transit time and known
ambient temperature conditions had, at times, exceeded the recommended
maximum of 85°F. (Deltech Ex. 139.) Apart from these known conditions, the
parties agree on little. Their respective experts offered substantially different
opinions as to whether at the time Tanks I, J, and K were loaded into Hold 4, they
were already doomed to auto-polymerize, (see Ott Decl., ECF No. 1306, at 4–9, ¶¶
10–37), or whether the ambient conditions in Hold 4 were a necessary contributing
factor.

63

The Court has already found that but for the extended storage at NOT, the
auto-polymerization event would not have occurred. But it also would not have
occurred but for higher than normal temperatures in Hold 4. The autopolymerization resulted from a “perfect storm” of combined circumstances. Below,
the Court first sets forth its findings with regard to the temperature in the hold.
Following that, it turns to the expert modeling it finds most persuasive with respect
to the onset of auto-polymerization and thermal runaway.
The temperature in the hold was influenced by a number of factors. First,
ambient temperatures. During the Flaminia voyage at issue, the outside air
temperatures were in the 80s Fahrenheit from July 2–9, in the 70s on July 10 and
11, and in the 60s on July 12 and 13. (Earle Decl. ¶ 11.) Both Dr. Davis and Dr.
Kaminski found that the difference between the outside air temperature and that
inside the hold was strongly dependent on whether the ventilation system was
working, its capacity, and whether the ventilation flaps to the outside were open.
Dr. Davis’s modeling, which the Court found persuasive, demonstrated that even
assuming ventilation, by July 7, 2012, the air temperature within the hold would
have been 8.3°F higher than the outside air temperature. (Davis Decl. at 78, ¶
5(a).) But if, as Dr. Kaminski assumed, the ventilation flaps to the outside were
closed or the ventilation system was not operating, the temperature would have
been higher. According to Dr. Davis, without ventilation, the hold temperatures
would have been 11.2°F higher than outside ambient air temperatures.

64

Ventilation (or a lack thereof) would have exerted additional influence on the
hold’s temperature. MSC’s Dangerous Cargo Manager, Dirk Vande Velde, testified
that the ambient air temperature within a ship’s hold is maintained by the ship’s
ventilation system. (See also Tr. at 1060:15–20.) There is mixed evidence as to
whether the ventilation flaps were open but, on balance, the Court is persuaded
that they were closed. All experts agree that if they were closed, the hold
temperature was likely higher than it would have been if they were open. (See, e.g.,
Tr. at 624:24–25 (Davis); id. at 1055: 9–16, 1060:21–1061:6 (Kaminski); id. at
1256:17–23 (Robbins); Ott Decl. ¶ 146.) A crew member reported to the Captain
that on the morning of July 14, 2012, the flaps on the hatch covers for the passive
ventilation were found to have already been in the closed position when the crew
arrived to close them in preparation for deployment of CO2; the captain testified
that he was “astonished” when he heard this. (Langer Vol. 1 Dep. Tr. at 72:22–
73:12; 203:8–16.)27 The third engineer confirmed that all four flaps for the passive
ventilation were already closed at the time he arrived to close them. (Deltech Ex.
306 at 2.) Moreover, had the ventilation flaps been open, crew should have observed
“smoke” (or venting DVB gas) emitting from them prior to the explosion; they did
not. (Casandra Dep. Tr. at 100:4–102:17; id. at 105:10–106:5; id. at 112:2–13.)

The Court here relies on certain designated deposition testimony, portions of which was objected to
by one or more parties. The Court overrules those objections. Here, for instance, the shipping
parties made a hearsay objection. However, the testimony fits comfortably within Federal Rule of
Evidence 803(3).
27

65

The closed ventilation flaps meant that the mechanical ventilation fans had
not been in use—as the flaps must be open for that to occur. (Dalomias Dep. Tr. at
108:6–8.) Further, crew testimony indicated that it was standard operating
procedure for the Flaminia to use mechanical ventilation only while in port, not
while at sea. (Tr. at 1255:1–7.) The Court finds that during the Flaminia’s voyage
from NOT en route to Antwerp, the mechanical ventilation for Hold 4 was not being
operated.
Additional conditions added to the temperature within the hold: the heated
wing tanks, (Davis Decl. at 100–01, ¶ 12), and the DPA. The fuel oil was heated to
a temperature between 113–140°F (45–70°C). (Kirstein Dep. Tr. at 187:1–188:2; id.
at 216:12–23); Tarnowski Dep. Tr. at 33:24–34:2; Langer Vol. 1 Dep. Tr. at 188:9–
16.) Based upon usage, the fuel in the wing tank adjacent to Hold 4 would have
been heated a day before it was needed, or by July 5, 2012. (Pokusa Dep. Tr. at
197:21–199:9; id. at 231:5–13.) As of July 13, 2012, the wing tanks adjacent to Hold
4 were almost empty. Thus, the heated wing tanks increased the temperature in
Hold 4. (See Tr. at 990:13–18.)
And finally, as depicted in Figure 5 above, ten ISO containers of heated DPA
were stowed in Hold 4: Tanks A, B, C, D, E, L, M, N, P, and FF. As described in
more detail above, the three containers I, J, and K were therefore surrounded by
containers of DPA. The DPA containers contributed to an increase in both the
ambient air temperature in the hold and, via radiative heat, in the adjacent tanks
themselves. (See Davis Decl. at 94, ¶ 48; id. at § 6.4; Kaminski Decl. at 17, ¶ 42.)

66

Dr. Davis persuasively modeled the influence of the DPA and found that between
the DPA and reduced ventilation, the DVB80 liquid temperature would have risen
to between 87.4–90.7°F (30.8–32.6°C). (Davis Decl. at 80, ¶ 9.)28
M. Measuring the DVB80’s Shelf Life Aboard the Flaminia
As discussed above, an Arrhenius equation predicts the rate at which the
polymer inhibitors are consumed as a function of temperature. (Davis Decl. at 98, ¶
1(b).) Based on the temperature of the DVB80, the Arrhenius equation therefore
predicts the fraction of inhibitor life depleted in the DVB80 liquid, and hence the
amount of shelf life that remains for the liquid. (Id.)
1.

Dr. Fauske’s and Dr. Kaminski’s Arrhenius Equations

Dr. Fauske, using his TAM tests, developed one Arrhenius equation for DVB
in a container with no headspace (a conservative and restrictive estimate) and
another for DVB in a container with 20% headspace. (Fauske Decl. at 8, ¶ 45.) Dr.
Kaminski derived her Arrhenius equation based on data from the Chauny DVB80
incident, in which a shipment of DVB80 auto-polymerized in 2006. The DVB80
containers in the Chauny incident had 10% headspace. She used the UA value of 39
W/K, as determined by Gexcon’s Full-Scale Test, and assumed the air temperature
in the hold was 31°F (another conservative estimate), as put forth by plaintiff’s
expert, Robbins. (Kaminski Decl. at 35–36, ¶ 23.)

Dr. Davis’s modeling accounted for both convective and radiative heating from the containers of
DPA.

28

67

Dr. Kaminski calculated the thermal history and “fraction of life consumed”
by the DVB80 during the 14 days of the voyage. She used a model to simulate the
heat transfer to the DVB80 from the containers filled with DPA, and she compared
the Arrhenius equations put forth by Drs. Fauske and Ott to her own. According to
her analysis, which the Court credits and found both thorough and persuasive, the
temperature of the air in the hold was affected by many factors. These included the
ambient air temperature, the sea temperature, the temperature of the heated fuel
in the wing tanks, heat transfer from and to all of the containers in the hold, and
the status of the ventilation fans and passive ventilation flaps. (Id. at 27, ¶ 3.)
Dr. Kaminski utilized these three equations to calculate the shelf life of the
DVB80 aboard the Flaminia. She concluded that the Arrhenius equations that
account for the DPA from the nearby ISO containers predict auto-polymerization on
July 14, 2012 only if the ship’s ventilation had also been sufficiently impaired.
2.

Dr. Davis’s Full-Scale Test

Dr. Davis performed the Full-Scale Test with an ISO container filled with
DVB80 exposed to certain heat conditions. This test provided him—and the
Court—with further support for his opinion as to what is likely to have occurred
aboard the Flaminia and why. Gexcon’s Full-Scale Test allowed Dr. Davis to test
his views with regard to the UA value for the ISO container, the induction time
necessary to deplete the inhibitor and oxygen during thermal exposure, and various
aspects of auto-polymerization reaction in an ISO container.

68

Gexcon applied the ambient temperatures at NOT to the ISO container used
for the Full-Scale Test. (Davis Decl. at 68, ¶ 23.) He started with DVB80 at 6.7°C.
To take into account the time at NOT, the exposure to the DPA, and the time spent
in the hold, Gexcon would increase the exposure temperature to accelerate the
induction time. Dr. Davis assumed that the DPA had a temperature of 71–73°C.
Storage of the DVB ISO containers in proximity to these temperatures raised their
temperature to between 24.1–24.9°C. (Id. at 74, ¶ 38.) The heated DPA increased
ambient temperatures by approximately 2.9°C. (Id. at 74, ¶ 40.)
Dr. Davis used the Full-Scale Test to make predictions about the induction
time (i.e., shelf life), accounting for the number of air changes per hour. Like Dr.
Kaminski, he determined that the DVB80 shipments aboard the Flaminia would
not have undergone thermal runaway unless the hold was excessively hot. He also
determined that, accounting for the DPA and reduced ventilation, the DVB80 would
likely auto-polymerize around July 14, 2012. If the ventilation system was
completely off, auto-polymerization was predicted to have occurred on July 14,
2012—the precise day of the incident.
3.

Dr. Ott’s Calculations

The Court agrees with Dr. Davis’s criticisms of Conti/MSC’s expert, Dr. Brian
Ott. Dr. Ott’s principle opinion was that if the DVB80 had been properly
oxygenated, then the other heat sources would not have caused run away autopolymerization. He reached this conclusion without, however, ever visiting
Deltech’s manufacturing facility and without doing any of the sorts of

69

measurements undertaken by Dr. Davis at that facility. Dr. Ott opined on the UA
value of the ISO containers without running any actual tests on an ISO container.
Notably, Dr. Ott agreed that: the DPA would have added to ambient
temperatures; solar radiation would have increased the heat of the DVB liquid by at
least 1°C; the combination of ambient air and solar radiation would have increased
the temperature from 6.7°C to 24 or 27°C; and the wing fuel tanks would have
contributed to the temperature in the hold. He also agreed that the DVB80 liquid
filled into the ISO containers destined for the Flaminia would have had 5–6% more
oxygen saturation than the DVB liquid transported by the Ludovica.
Dr. Ott’s assumptions, put into Dr. Davis’s model, result in a prediction that
auto-polymerization would have occurred after 15.7 days—almost a week earlier
than it did occur. Moreover, Dr. Ott’s own prediction uses an assumption that the
oxygenation level is at 94%; if it were lower, as Dr. Ott claims it was, autopolymerization should have occurred even sooner than his 15.7 days. (Davis Decl.
at 110–11, ¶ 48.) In the end, a fundamental problem with Dr. Ott’s oxygenation
theory is that, unlike those of the Stolt/Deltech experts, it is entirely unsupported
by any actual testing.
N. The Alarms, CO2 Discharges, and Subsequent Explosion
The Flaminia is fitted with a smoke detection system. This system monitors
the air in the cargo holds. (Stipulated Facts at 33, ¶ 43.) The system is located in
the CO2 room and to the right of the three-way valve manifold for the CO2 system
for the cargo holds. (Id. at 33, ¶ 44.) The cargo hold smoke detection alarm is

70

linked to the ship’s main monitoring and alarm system. (Id. at 34, ¶ 49.) The
output from the smoke detection system to the main monitoring and alarm system,
whether indicating a fire or a fault, is in the form of one alarm channel stating
“SMOKE DETEC. CARGO SYS FAIL.” (Id.)
At approximately 5:42 in the morning on July 14, 2012, the smoke detector
alarm activated on the bridge of the Flaminia. (Id. at 35, ¶ 56.) At approximately
6:00 a.m., the Chief Officer sounded the general alarm and announced on the public
address system that a fire had been detected in Cargo Hold 4. (Id. at 35, ¶ 60.) At
6:42:27 a.m., the door for the cabinet containing the master release ball valve for
the cargo hold CO2 system was opened. (Id. at 35, ¶ 63.) The second release of CO2
to the cargo hold commenced at 7:07:52 a.m. as confirmed by the machinery
monitoring system alarm log: CO2 BOTTLE CO2 LEAKAGE ALM ALARM M2
07:07:52 B. (Stipulated Facts at 36, ¶ 67.)
At 7:40 a.m., the Captain sent an email message to NSB headquarters that
read:
vessel in position
048 13 N / 027 59 W
fire in hatch NO 4
fire not under control
stopped engine, used CO2 to destinguish (sic) the fire without result.
Crew is ok
(Id. at 36, ¶ 69.) Sometime shortly after 8:00 a.m., the Chief Engineer was ordered
to release additional CO2 into Cargo Hold 4. (Id. at 36, ¶ 70.) The third release of
CO2 to the cargo hold commenced at 8:06:32 a.m. (Id. at 36, ¶ 71.)

71

The Flaminia was equipped with a fixed gas fire extinguishment system, a
fire detection system, an alarm system, and a water supply fire main system with
two fire pumps. (Id. at 32, ¶ 34.) The Flaminia’s fixed High Pressure Carbon
Dioxide Gas (“CO2”) System was intended to protect the engine room and all cargo
holds and included 330 cylinders of CO2, each of which contained 45 kilograms of
CO2; these were stored in the CO2 Room. (Id. at 32, ¶ 35.) According to the vendor
drawing, the quantity of CO2 carried onboard was intended to ensure that the
largest volume compartment requiring protection receives a sufficient concentration
of CO2 to suppress a fire. (Id. at 32, ¶ 36.)
Release of CO2 into a cargo hold requires manual opening of the individual
cylinders. The number of CO2 cylinders released into the cargo holds varies by the
size of the specific hold and the volume of cargo containers carried in the hold. (Id.
at 33, ¶ 40.) The parties spent substantial time at trial on the questions of (1)
whether the CO2 release was inadequate, and (2) whether the DVB80 vapors could
have been rendered inert and the explosion avoided if more tanks had been
released.
The explosion occurred one to two minutes after the Chief Engineer finished
opening the CO2 cylinders for a third release of CO2. (Id. at 36, ¶ 72.) The Chief
Engineer was alone in the CO2 room when the explosion occurred. (Stipulated
Facts at 36.) The explosion occurred in the vicinity of Hold 4. (Id. at 36, ¶ 74.)
After the explosion, on deck, the Second Officer looked back in the direction of
Cargo Hold 4 to see a large amount of dense black smoke on the upper deck,

72

containers falling into the sea, and, eventually, flames. (Id. at 37, ¶ 75.) After the
explosion, witnesses on the bridge saw dense black smoke, containers in the sea,
and flames coming from the area of Cargo Hold 4. (Id. at 37, ¶ 76.) At 8:08 a.m.,
the Captain sent another email message to NSB headquarters that included the
original language from the prior message sent at 7:40 a.m. and the additional text:
CARGO HOLD four ecxploded (sic)
container over board will lesave (sic) the vessrel (sic)
(Id. at 37, ¶ 77.)
After the Flaminia incident, its CO2 system was found to have been defective.
(Paffenhoff Dep. Tr. at 21:18–21.)29 Immediately after the incident, an engineer
from NSB determined that piping leading to a booster unit (designed to assist in the
release of CO2) was incorrectly installed. This resulted in an activation of levers
triggering an alarm and an unexpected shutdown of engine room equipment.
(Dehde Dep. Tr. at 48:17–49:13; id. at 84:25–85:24; id. at 109:17–111:6; id. at 133:9–
135:7; Deltech Ex. 315; Deltech Ex. 316.)
The Flaminia had four sets of instructions stating four difference sets of
numbers regarding how many CO2 cylinders needed to be released immediately and
how many were to be released at 30-minute intervals. Two sets (a Fire Control and
Safety Plan and the Nautical Audiovisual Emergency Control Support) both advised

The Court in this Section cites the deposition transcripts of a number of plaintiffs’ employees, some
of whom were Flaminia crew members: Pompeyo Dalomias was an electrical engineer aboard the
Flaminia; Joerg Dehde and Joerg Erdtmann are both project engineers for NSB; Lars Paffenhoff was
a fleet engineer for NSB at the time of the Flaminia incident; Maciej Pokusa was a second engineer
on the Flaminia; Steve Sabandal was a fitter on the Flaminia; Janusz Tarnowski was an engineer
aboard the Flaminia.
29

73

an initial release of 189 cylinders, with 32 cylinders every 30 minutes thereafter.
(Erdtmann Dep. Tr. at 75:17–76:14; Hall Dep. Tr. at 207:10–211:1030; Deltech Ex.
266; Deltech Ex. 318; Deltech Ex. 319.) Additional instructions directed that only
31 cylinders should be released immediately into the hold, and 31 additional
cylinders every 30 minutes thereafter. (Tarnowski Dep. Tr. at 69:25–70:8; Deltech
Ex. 265; Deltech Ex. 320; Deltech Ex. 330.)
The evidence supports confusion surrounding the releases of CO2. Several
individuals testified that they were involved in the manual opening of CO2
cylinders, but were not sure how many. (See, e.g., Dalomias Dep. Tr. at 65:3–23; id.
at 78:13–15; Sabandal Dep. Tr. at 22:6–20; id. at 51:6–11); Pokusa Dep. Tr. at
36:12–14; id. at 38:2–4.) The weight of the evidence supports fewer than 30
cylinders having been released during the first release. Following the first release
of CO2, the Flaminia’s main engine room shut down. (Tarnowski at 194:6–15.) The
vessel’s electrician was able to restart the auxiliary boiler only shortly before the
explosion. (Dalomias Dep. Tr. at 58:11–59:13; id. at 60:1–61:5.) And as noted
above, a second release of CO2 cylinders did occur at 7:07 a.m.—but again, none of
the personnel employed in the task could recall how many cylinders were operated.
(Tarnowski at 126:4–14; id. at 198:23–199:2; id. at 229:14–19.) The evidence
supports a release of only 7 cylinders during the second release. (Davis Decl. at
134–36, § 9.2.3.)

Brian Hall was employed by the Merchant Marine Academy and was retained as an expert in
shipboard firefighting.
30

74

As noted earlier, the Court is persuaded that the DVB80 did not auto-ignite—
rather, crew activity led to a spark that caused the explosion and fire in Hold 4.
The parties agree that the thermal runaway of the DVB80 led to the discharge of
flammable vapors into the hold. (See, e.g., Davis Decl. at 121, ¶ 1; Ott Decl. at 54, ¶
199.) Then, the Court finds, the explosion was ignited when the crew opened the
manlid to insert firehoses.
Opening the access point would have allowed more oxygen into the hold,
which likely brought the DVB80 vapor within the narrow concentration range (1.1–
6.2%) that allows it to ignite. A spark was then created through the opening of an
access point or by dropping a metal object into Hold 4. The physical act of opening
an access point—which can be heavy and “track back when you open” it—can create
a spark. (Tr. at 826:3–6.) And while there is uncertainty as to whether hoses were
inserted into the hold—the hoses were certainly being prepared and positioned—the
Court is persuaded by Dr. Beeley’s testimony that, when he examined the Flaminia,
he found evidence of burnt hoses that had been attached to the hydrants, such as
the remains of the spigot and burnt bits of hose. (Id. at 834:1–16; id. at 826:22–
827:4; Robbins Decl. ¶ 50; Langer Dep. Tr. at 135:2–23.) The insertion of the hoses
itself could also have initiated the spark. Thus, the Court finds that this crew
activity—again, in response to what crew members believed was an ongoing fire—
created a spark that triggered the explosion.

75

IV.

CONCLUSIONS OF LAW
The parties agreed that the Court’s primary task in Phase 1 of this multi-

phase trial is factual: the Court is to decide what occurred during the production
and transport of the DVB80, and which external conditions ultimately contributed
to the explosion and fire aboard the Flaminia. The conclusions of law discussed
herein are therefore limited to very basic concepts of causation.
In order for a party to be found liable on any of the claims against it, the
claimants must establish causation. See In re M/V DG Harmony, 533 F.3d 83, 96
(2d Cir. 2008) (“In addition to proving duty and breach, the plaintiff must prove
causation. . . . [and] show that the dangerousness of the cargo . . . caused the
harm.”). Causation requires claimants to prove both cause in fact as well as legal
cause. See Jurgens v. Poling Transp. Corp., fa113 F. Supp. 2d 388, 397 (E.D.N.Y.
Sept. 19, 2000) (“To prevail on a claim for negligence under the general maritime
law, the burden is on the plaintiff to establish duty, breach of duty, causation (both
cause in fact and proximate cause) and damages.” (internal quotation omitted).
The fire and explosion that occurred aboard the Flaminia on the morning of
July 14, 2012 were caused by a number of factors. All parties agree that the DVB80
onboard underwent auto-polymerization, which led to thermal runaway. The Court
now finds that the following were substantial contributing factors that led to the
DVB80’s auto-polymerization:

76



The decision to ship the DVB80 out of NOT, which necessitated a longer
voyage than would have a more northeastern port and exposed it to
undesirable conditions;



The fact that the DVB80 was left still on the dock at NOT for 10 days in the
sun, in hot weather, and next to a number of tanks of heated DPA31;



The placement of the DVB80 in Hold 4, where it was stored next to
containers of heated DPA and near the ship’s heated fuel tanks; and



The lack of proper ventilation, leading to hotter-than-typical ambient
temperatures in Hold 4.
The Court additionally finds that the DVB80 was adequately oxygenated and

chilled when it left Deltech’s facility, and that it did not auto-ignite aboard the ship.
Rather, crew activity—through, inter alia, opening the access point to Hold 4—
created a spark that ignited the fire.

Dr. Davis specifically estimates that the storage conditions at NOT subjected the DVB to “a
needless and avoidable 2.5°C increase in the bulk liquid temp of the DVB80 at loading.” (Davis Decl.
at 77, ¶ 1.)
31

77

V.

CONCLUSION
Based on the Court’s decision as to Phase 1, the parties are to confer on the

timing for Phases 2 and 3. The Court believes that those phases can be combined.
The parties shall consult and provide proposed dates (with expected duration) in a
letter to the Court filed not later than December 15, 2017.
SO ORDERED.
Dated:

New York, New York
November 17, 2017
______________________________________
KATHERINE B. FORREST
United States District Judge

78