Recent ruling on the use of fire modeling in fire investigations

[Stephen Olenick]

Below is a recent ruling regarding the use of computer fire models in
fire investigation. I thought some of the FDS users and developers
would be interested in reading it. There are several aspects with
which I disagree. First, I would not characterize NIST's use of FDS
in the WTC investigations as simply "illustrative". Secondly, NFPA
921, the guide for fire and explosion investigations, which is created
by a consensus group of fire investigative personnel and interested
parties and is an ANSI standard, states that computer modeling "can be
used to test hypotheses regarding fire origin and fire development".
The fact that the defendant's expert, Mr. West, has indicated that he,
at his time at the New York Fire Department and NYC Bureau of Fire
Investigation, did not use fire modeling to aid in fire investigation
should hardly be considered an endorsement from the rest of the fire
investigation community that it does not support its use.

Fire dynamics are governed by the laws of science. If FDS and other
computer fire models are appropriate for design, it is appropriate for
use in fire investigation and reconstruction. Regardless of whether
it's pre- or post-fire, the fire dynamics are the same. Of course, it
is incumbent on the user to properly apply the model and use proper
inputs etc. I guess my overall problem with the opinion is that
instead of simply excluding this particular expert and the use of
computer fire modeling in this particular case (correctly or
incorrectly), the judge took the stance that computer fire modeling is
inappropriate for fire investigation and reconstruction as a whole. I
couldn't disagree more.

Thoughts from anyone else?

Stephen Olenick, MSFPE, PE
Senior Engineer
Combustion Science & Engineering, Inc.



---
Santos v. State Farm Fire and Casualty Company, 000790/07

Supreme Court, Nassau County
Evidence

New York Law Journal

07-02-2010

Justice Thomas P. Phelan

Decided: June 28

Wilkofsky, Friedman, Karel & Cummins, Esqs., Attention: Mark L.
Friedman, Esq., Attorneys for Plaintiff, New York, NY

Rivkin Radler LLP, Attention: Michael A. Troisi, Esq., Attorney for
Defendant, Uniondale, NY


Plaintiff moved for an order, pursuant to Frye v. United States, 293
F. 1013 [1923], excluding the testimony of defense witness Dr. Jozef
Urbas concerning the results of computer fire modeling utilized by Dr.
Urbas. At the time the motion was made, plaintiff contended that Dr.
Urbas utilized the computer fire modeling to determine the origin and
cause of a fire that occurred at plaintiff's premises known as 168
North First Street, Bethpage, New York, on January 20, 2006. By short
form order dated January 26, 2010 (Phelan, J.), this court ordered a
Frye Hearing.

Pursuant to the Frye test, expert testimony based on scientific
principles or procedures is admissible only after a principle or
procedure has gained general acceptance in its specified field (People
v. Wesley, 83 NY2d 417, 422 [1994]). A particular procedure need not
be unanimously indorsed by the scientific community but must be
generally accepted as reliable (Id. at 423). The Frye test emphasizes
"'counting scientists' votes," rather than verifying the soundness of
a scientific conclusion (Parker v. Mobil Oil Corp., 7 NY3d 434, 447
[2006]).

In 1993 the United States Supreme Court decided Daubert v. Merrell Dow
Pharmaceuticals, Inc., 509 US 579 which, in federal courts, displaced
the Frye general acceptance standard based upon the Federal Rules of
Evidence. The Frye general acceptance test, however, continues to be
the standard for determining reliability and admissibility of expert
testimony in New York (see People v. Wesley, 83 NY2d at 433).

The burden of proof is on the party challenging the evidence to make a
prima facie showing that it is a novel theory which is not generally
accepted (Matter of Seventh Jud. Dist. Asbestos Litig., 9 Misc.3d 306,
311-312 [Sup. Ct. Wayne Co. 2005]). The burden then shifts to the
proponent of the evidence to show by a fair preponderance of the
credible evidence that there is sufficient general acceptance of its
reliability (Id.).

"[T]he court may, as it did here, take the testimony of expert
witnesses" (People v. Wesley, 83 NY2d at 437). Eugene J. West, a fire
and arson consultant in the private sector since 1997, testified on
behalf of plaintiff. His experience in the investigation of explosive
and incendiary devices began in the Republic of Viet Nam where he was
an expert in mines and booby trap and the destruction of explosive
ordinance and incendiary devices. When he returned to the United
States, Mr. West enrolled in military police school where he received
additional training in investigation.

In 1971 Mr. West entered the New York City Police Academy. From 1972
to 1977 he was a corrections officer for the New York City Corrections
Department. From 1977-1987 he was a fire fighter for the New York City
Fire Department and then was promoted to the New York City Bureau of
Fire Investigation (the Fire Marshall's Office) where he attended the
Fire Marshal Training Academy. Mr. West conducted field investigations
involving the crime of arson for cause and origin of fires.

In 1992 Mr. West attended the FBI National Academy where he took
courses in forensics, photography and behavioral science. After
graduation, Mr. West was assigned to the city-wide investigations unit
in the Bureau of Fire Investigation where he was charged with
investigating major fire incidents. While there, he started the major
incident response team that was specially trained to investigate major
fire incidents. At that time Mr. West was also an instructor in the
field of fatal fire investigations at the FBI Academy. In 1997 he
partnered with the former Assistant Chief Fire Marshal of the New York
City to form Guardian Investigations Group.

Mr. West testified that in his capacity as a fire investigator he
would get many requests to utilize computer fire modeling but would
decline to use it as part of the official investigation, stating that:
"We can only speculate as to what was the composition or the exact
construction of a room or the type of materials that were used,
especially if those things are no longer available to us" (p. 16).
Mr. West stated that it was never generally accepted as an
investigative tool by the New York City Fire Department. It was his
opinion that computer fire modeling is not generally accepted in the
fire investigative community and cannot be used to determine the cause
of a fire (pp. 20, 21). Although computer fire modeling was used in
the World Trade Center investigation, it was used for illustrative
purposes (p. 30). Mr. West explained that there is a caveat by the
National Fire Protection Association ("NFPA") that the program is
essentially as good as the information put into it.

Mr. West's expert opinion is sufficient, in the court's opinion, to
make a prima facie showing that computer fire modeling, when used to
determine the cause of a fire, would be novel for that purpose and is
not generally accepted in the fire investigative community (Matter of
Seventh Jud. Dist. Asbestos Litig., 9 Misc.3d at 312). The burden now
shifts to the proponent of the evidence to establish general
acceptance of its reliability (Id.).

Jozef Urbas, an associate professor at the University of North
Carolina Department of Engineering where he teaches principles of fire
behavior or fire dynamics, testified in support of computer fire
modeling's general acceptance. Dr. Urbas uses two textbooks that
specifically refer to fire modeling: Principles of Fire Behavior and
Fire Dynamics. With regard to his background, Dr. Urbas testified that
he has a Ph.D. in chemistry and that part of his studies included fire
dynamics in fire modeling.

According to Dr. Urbas, in addition to the University of North
Carolina, programs in fire sciences which also incorporate fire
dynamics in fire modeling, are also offered at the University of
Maryland, Wooster Polytech, Lund University in Sweden, University of
Edingburgh in Scotland and at universities in New Zealand, Australia,
Japan and Europe. Dr. Urbas also testified that the following
organizations endorse fire modeling in fire dynamics: American
Standards and Testing of Materials, E5 committee; Society of Fire
Protection Engineers; National Fire Protection Association (NFPA);
International Association of Fire Safety Science and International
Standards Organization, all of which conduct conferences and lectures
attended by members of the fire science community and other scientists
and engineers. Dr. Urbas has lectured at NFPA and testified that NFPA
endorses the use of fire modeling in fire dynamics in the use of fire
investigation.

Fire dynamics was described as "the course of a fire from its ignition
to its extinguishment. It describes the time line of a fire in terms
of ignition, flame spread, release of heat, flame size, temperatures,
velocities, toxic gasses and smoke" (p. 42). Dr Urbas testified that
the underlying equations and laws of physics have been generally
accepted in the fire science community (p. 44) and that they have been
generally accepted as reliable in computer fire modeling (p. 46). Dr.
Urbas used the National Institute of Standards and Technology ("NIST")
Fire Dynamics Simulator (Version 4) ("FDS"), which was marked as
Defendant's Exhibit C. Dr. Urbas testified that fire modeling of fire
dynamics is not a new science (pp. 49-50) and that the NIST FDS is
used in the science and engineering community for design purposes
[such as buildings and their contents] and for fire re-creation (p.
60).

Although counsel for defendant stated that there are cases that have
allowed fire dynamics to be used in evidence, none were found with
respect to fire modeling that applied the Frye standard. The NIST is
part of the Department of Commerce. Dr. Urbas testified that other
regulatory agencies of the federal government also use fire modeling,
such as the Department of Energy and Nuclear Regulation Commission,
the Department of Defense, the Department of Agriculture and ATF. The
court notes that these regulatory agencies are involved in risk
assessment as opposed to fire investigation based on scientific
standards.

Counsel for defendant explained that there was a misunderstanding as
to the purpose of the computer fire modeling. Dr. Urbas was not coming
into court to state the cause and origin of the fire but rather to
apply the computer dynamics to see how the fire would spread (p. 64).
Dr. Urbas testified that the results of the fire modeling established
that there was a time line that matched a particular origin of the
fire, that the damage in the building corresponded to the results of
the modeling and that the determination of fire dynamics in that
particular theory [the time line] is generally accepted for that
purpose (p. 69). The computer fire modeling essentially verified the
hypothesis as to the ignition source or cause of the fire (p. 72). Dr.
Urbas "never said it's accepted for determining the origin of the fire…
It can—it can help determining the cause. It's—it cannot be the sole
method of determining the cause" (p.80). While computer fire modeling
may be generally accepted in the scientific community for predicting
the course of fires given a particular set of circumstances and,
therefore, useful in fire prevention and safety, Dr. Urbas has not
demonstrated its general acceptance in fire investigation.

Dr. Urbas acknowledged that Kirk's Fire Investigation is a bible in
the fire investigation field. Although Dr. Urbas disagreed with a
reading from that text on the record, the reading revealed that:
"Experienced fire scientists realize the limitation of applying models
to field fire investigations. These models in general are designed to
start with the ignition of a fire under preset conditions and predict
the time factors and conditions of growth and sometimes decay. They
are not designed to recreate a particular fire by working backward
from a set of final observations to determine what the starting or
even intermediate conditions were" (p. 79).

The NFPA Users Manual for 921 states: "To conduct valid modeling and
testing it is important that the investigator gather data that is as
accurate and complete as possible." Dr. Urbas agreed that the concept
"garbage-in-garbage-out" was applicable (p. 73). Here, the input was
based upon regulatory agency tables [for furniture, floors, walls,
etc.]*, measurements taken by Dr. Urbas, his inspection of the damage
and his reliance upon information received from fire investigators
from the insurance company. Dr. Urbas never spoke with the homeowner,
the Nassau County fire officials or local firefighters. Dr. Urbas
testified that he was unaware that there were paint thinners and
solvents in the area where the fire started and that such knowledge
would have thrown off the entire calculation (p. 75).

Fire modeling carries with it a 15 to 20 percent margin for error
assuming all conditions are correct but could be as high as 80 percent
depending upon the real conditions (pp. 76-77). Dr. Urbas acknowledged
that there could be a difference between the material represented in a
table and the actual material at the fire scene (p. 83). "If the input
in the model is correct then the output is correct" (p. 96). Dr.
Urbas' testimony indicated that incorrect input could lead to
inaccurate results.

Bearing in mind that the court's role as gatekeeper is "counting
scientists," the court finds that defendant did not present sufficient
evidence to establish that there is a consensus in the fire
investigation community that computer fire modeling is generally
accepted as reliable. "The long-recognized rule of Frye v. United
States (supra) is that expert testimony based on scientific procedures
is admissible but only after a principle or procedure has 'gained
general acceptance' in its specified field" (People v. Wesley, 83 NY2d
at 422). Here the specified field is fire investigation. The issue
before the trier of fact in this particular lawsuit involves when and
how the fire started. Defendant failed to meet its burden of proving
that its expert's use of computer fire modeling was generally accepted
in the fire investigation community (Cumberbatch v. Blanchette, 35
AD3d 341, 342 [2d Dept. 2006]). Although defendant's expert may
support a case for the acceptance of computer fire modeling in the
regulatory/design community, it does not support a conclusion that it
is generally accepted in the fire investigation community.

This decision constitutes the order of the court.

*Dr. Urbas stated that: "There might be a difference between the
material that is represented in the tables and…the actual fire scene."
He determines whether to apply those numbers or not (p. 83).
---

[Kevin]

In my opinion, fire modeling and fire science are inseparable. The
most important results of fire science are incorporated in our models.
By models, I mean anything from a simple correlation to something like
FDS. A model is nothing more than a calculation method, and to say
that "fire models" are not generally accepted in fire investigation is
akin to saying that science has no role in that practice. I agree with
Stephen that the court should consider how the model is being applied,
the same way that it would consider any other form of evidence. This
blanket dismissal, based largely on one person's opinion, sends the
wrong message.

And by the way, Stephen is also right about our use of FDS in the NIST
WTC Investigation. The temperatures predicted by FDS were used in a
sequence of calculations to assess the aircraft impact, fire, thermal
penetration of the steel, and eventual collapse of WTC 1, 2 and 7.
This was not simply done for "illustrative purposes." It was part of a
quantitative analysis to better understand why the buildings
collapsed.

[dji...@gmail.com]

Note the previous "positive" opinion on the use of FDS in an Ohio
court case.
======================================================================

NATIONAL INSTITUTE OF STANDARDS & TECHNOLOGY FIRE DYNAMICS SIMULATOR
SOFTWARE SURVIVES DAUBERT CHALLENGE IN OHIO

Source: [tedfordpond.com]

In a recent decision, Turner v. Liberty Mutual Fire Insurance Co.,
2007 WL 2713062 (N.D. Ohio)(September 14, 2007); a trial court held
that the National Institute of Standards and Technology (NIST) Fire
Dynamics Simulator (FDS)(Version 4.0) computer simulation proffered by
the defendant’s expert satisfied the Daubert reliability test
governing expert testimony.

The underlying action arose from a breach of contract and bad faith
action filed by the plaintiff against Liberty Mutual Fire Insurance
Company for failing to pay insurance proceeds after a house fire.
Liberty disclosed its liability expert, who wrote a report based on
computer software simulations showing that the fire was incendiary.
The plaintiff filed a Motion in Limine attacking Liberty’s expert’s
methodologies.

The court found that Liberty’s expert’s NIST FDS simulation satisfied
the Daubert standard governing expert testimony reliability. The
Supreme Court established the standard for admissibility of scientific
expert testimony under Rule 702 in Daubert v. Merrell Dow
Pharmaceuticals, Inc., 509 U.S. 579 (1993). The requirement that “any
and all scientific testimony or evidence admitted [be] not only
relevant, but reliable,” id. at 589, “entails a preliminary assessment
of whether the reasoning or methodology underlying the testimony is
scientifically valid and of whether that reasoning or methodology
properly can be applied to the facts in issue.” Id. at 592-93.

Under Daubert, the court provided several (non-exclusive) factors to
consider in determining reliability: (1) whether the theory or
technique “can be (and has been) “tested”; (2) whether it has been
“subjected to peer review and publication”; (3) “the existence and
maintenance of standards controlling the technique's operation”; (4)
the theory or technique's “known or potential error rate”; and (5) its
“general acceptance” in “a relevant scientific community.” Daubert,
509 U.S. at 593-94.

In Turner, the court applied these factors to the motion and found the
following. First, the software was tested. FDS is described in NIST
Special Publication 1018. The Acknowledgments section of Publication
1018 lists various individuals who have “conducted a number of small
and large scale experiments to validate FDS.” The court relied on the
fact that the NIST FDS software was used to run simulations of the
fires at the World Trade Center. The court noted that the September
2005 Computer Simulation of the Fires in the World Trade Center Towers
Abstract states, “[T]he model was validated by comparing its
predictions with measurements from a series of large scale experiments
performed at NIST.”

Second, the software was adequately subjected to peer review and
publication. The court again relied on Publication 1018's
Acknowledgment section containing three pages of peer reviews and
contributions, and noted that its bibliography listed 152 sources from
which the technical data was drawn.

Third, the software has known error rates for the court to consider.
NIST FDS cautions that two components of its calculations--flow
velocities and temperatures-- have error rates of 5-20%. However, the
court noted that this 5-20% figure doesn’t represent an overall error
rate and, thus, the matter could appropriately be raised in cross-
examination.

Fourth, the Turner Court found that Liberty’s expert's computer
simulation methodology is “generally accepted” by the “relevant
scientific community.” The court stated that NFPA 921 is a “recognized
guide for assessing the reliability of expert testimony in fire
investigations.” NFPA 921 comments on the Computational Fire Dynamics
(CFD) model: “The use of CFD models in fire investigation and related
litigation, however, is increasing. CFD models are particularly well
suited to situations where the space or fuel configuration is
irregular ... or where very fine detail is sought.”

Moreover, NFPA 921 confirms the appropriateness of Liberty’s expert's
application of the model since the court found the defendant’s expert
compared the results of the simulation to “physical and eyewitness
evidence [photographs] to support or refute the hypothesis.” The court
also stated that perhaps the best evidence of the software's
acceptance is its use in three recent nationally-recognized fires: the
World Trade Center collapse, the Rhode Island nightclub fire, and the
South Carolina sofa store fire.

While this case supports the reliability of expert testimony based on
computer simulations run on NIST FDS Version 4.0 software, counsel are
reminded that computer simulations actually created by NIST during its
own investigations are inadmissible in civil actions. See 15 U.S.C. §
281a.

On Jul 8, 11:00 am, Stephen Olenick <stephen.olen...@gmail.com> wrote:

> read more »...

[Rein]

Metaphrasing Socrates, courage is between fear and confidence.

> ...
>
> read more »

[Phil Friday]

It seems to me that if the court was 'counting scientists votes' as
Frye emphasizes, that the number of votes counted was much too low.

The way I read the decision, and this is somewhat encouraging, the
court did not issue a blanket dismissal of computer fire modeling in
fire investigation, it simply simply stated that the defense "did not

present sufficient evidence to establish that there is a consensus in
the fire investigation community that computer fire modeling is
generally accepted as reliable."

One nuance that surely led to the decision in this case is that the
input used in the model did not account for important artifacts and
information about the fire in questions-

"Dr. Urbas testified that he was unaware that there were paint
thinners and solvents in the area where the fire started and that such
knowledge would have thrown off the entire calculation (p. 75).