Standard engineering practice or junk science?

By Richard Schulte

Schulte & Associates, Evanston, Ill.

At approximately 5:30 in the morning of March 21, 1996, a disgruntled employee ignited a fire in multiple-row storage racks at the McFrugal's Warehouse in New Orleans, Louisiana. The racks contained Class IV commodities (and, more than likely, a mixture of both Class IV commodities and Class A plastics) stored to a height of 21 feet. The multi-row racks were protected only by ceiling sprinklers, and the ceiling sprinklers protecting this portion of the warehouse were located a little more than 50 feet above the top of the storage in the multi-row racks.

The McFrugal's Warehouse, originally named the Pic 'N' Save Warehouse and also referred to as the New Orleans Distribution Center, was designed by a design/build firm in New Orleans in 1988 and 1989. The single-story warehouse had dimensions of 1,000 feet by 1,000 feet (1 million square feet) and was constructed with an attached two-story office building that was approximately 40,000 square feet in total floor area. The middle one-third of the warehouse was constructed with a ceiling height of approximately 75 feet, while the portions of the warehouse that flanked the high bay section had ceiling heights of roughly 40 feet.

The high bay portion of the warehouse was designed for the storage of Class IV commodities in double-row racks to a height of 65 feet. The sprinkler protection for the high bay portion of the building consisted of large-orifice high temperature (286 F) ceiling sprinklers and in-rack sprinkler protection, as required by NFPA 231C. The overhead system protecting the high bay storage was designed to provide a density of 0.45 gpm/SF over an area of 2,500 SF. The sprinkler systems protecting the building were supplied by a 12-inch underground yard main system that was supplied by two pumps, one electric-driven and the other diesel-driven. The electric pump took suction from the municipal distribution system, while the suction source for the diesel-driven pump was an aboveground 300,000-gallon storage tank.

Since the building was designed to address the future storage needs of the building operator, only a portion of the high bay section of the building was provided with double-row racks when the building was completed. The portion of the high bay section of the building that was not outfitted with double-row racks was intended to be used as a staging area for the racks until such time as additional high bay storage was required. Given that the high bay portion of the building would eventually be outfitted with double-row racks, the sprinkler system designer, Grinnell Fire Protection Systems Company, designed the sprinkler system protecting the high bay staging areas utilizing the same design criteria as required for ceiling sprinklers protecting the high bay racks.

The multiple-row racks were installed in the high bay warehouse some time after the building was turned over to the building operator by the general contractor. The building operator contacted Grinnell Fire Protection for an estimate on the cost to install two levels of in-rack sprinklers in the multiple-row racks (as required by NFPA 231C). However, the building operator elected not to install the required in-rack sprinkler protection.

Note: The fact that the building owner had requested a bid to provide sprinkler protection in the multi-row racks and had elected not to do so only became known after all of the experts in the case had been retained, had submitted their expert reports and had been deposed. Hence, all of the parties involved in the lawsuit were unaware of this fact until Charlie Martin, the manager of the Grinnell office in New Orleans, was cleaning out his desk and found, quite by accident, one of the key pieces of evidence in the case. He had completely forgotten that he had provided a price quote to install in-rack sprinklers in the multi-row racks in the McFrugal's Warehouse.

Given the inadequate sprinkler protection provided for the multi-row racks, what happened next shouldn't really be much of a surprise, but, actually, the events that followed are rather interesting. With a Herculean effort, the New Orleans Fire Department (NOFD) was able not only to gain control of the fire but also to save most of the building structure. By 10:30 a.m. or so, the NOFD had extinguished the fire and the cleanup of the building began. Since ceiling sprinklers throughout the warehouse had operated, all of the sprinkler systems protecting the building were shut down. The NOFD stationed a company or two of firefighters at the scene to monitor the cleanup, to extinguish any "hot spots" and as a precaution against a re-kindle.

Sometime later in the day, the building operator requested permission to turn the electric power to the building back on to facilitate the cleanup operations. Permission to re-energize the building electrical system was granted by the NOFD, and the rest of the story is probably obvious. Electrical equipment damaged in the fire ignited another fire at about 3 p.m., and, with all of the building's sprinkler systems shut down, the second fire destroyed the contents of the entire warehouse. After the second fire, only the concrete exterior walls and the concrete floors remained undamaged. The interior of the building was a mass of twisted steel. The fire had consumed all of the combustibles within the warehouse. The office portion of the building, however, remained relatively undamaged, protected by the fire separation between the office building and the warehouse.

Of course, as in all large property loss fires, the property insurer for the building and contents, a Lloyds of London syndicate, filed a lawsuit alleging negligence on the part of the sprinkler system designer, Grinnell, the design/build contractor and a lengthy list of other parties. The list of defendants in the lawsuit included the National Fire Protection Association (NFPA), Factory Mutual (FM), Underwriters Laboratories Inc. (UL), the Southern Building Code Congress International (SBCCI), the manufacturers of the rack system used in the building and, of course, the New Orleans Fire Department. The New Orleans Fire Department likely could have avoided becoming entangled in the litigation had fire fighters simply let the building burn in the morning; as they say, "no good deed goes unpunished."

Given that the multiple-row racks were installed without in-rack sprinklers by the building operator some time after the building construction was completed, the plaintiff's attorneys had a difficult time finding experts that would support their case. Eventually Dr. Craig Beyler and Richard Custer were retained as plaintiff's experts.

Both Beyler and Custer asserted that Grinnell was negligent because the sprinkler design did not adequately address the clearance between the sprinklers and the contents in the staging area of the high bay portion of the building. Beyler and Custer also asserted that, due to the size of the building, the architectural design should have included roof vents and draft curtains.

In addition to the sprinkler system design and the roof vent/draft curtain issues, Beyler asserted that the NFPA and FM were negligent because the sprinkler design standards developed by these two organizations did not adequately address the issue of clearance between sprinklers and combustibles. Beyler further asserted that both UL and FM were negligent because their listing/approval of sprinklers did not contain any warning regarding the use of sprinklers where large clearances between the top of combustibles and the ceiling exist and that the rack manufacturers were negligent because no warnings regarding excess clearances were provided on the rack structural members.

Experts for various defendants in the litigation included: John O'Neill (for Grinnell; the chairman of the NFPA 13, Chester Schirmer (for the NFPA); the former chairman of the NFPA 13 committee and Richard Schulte, the building code consultant (for the design/build contractor) for the project when it was being designed.

Custer supported his opinion on the clearance issue and on the need for roof vents/ draft curtains on general fire protection knowledge and practice.  Beyler utilized a computational fluid dynamics (CFD) model and referred to the Large Eddy Simulation (LES), to demonstrate the effect of the high ceilings/excess clearances on the ability of a sprinkler system to control a fire occurring in the multi-row racks. Beyler also used the LES model to demonstrate that the ceiling sprinklers over the multi-row racks would have failed to control the fire even if in-rack sprinklers had been provided. Beyler's rationale for this opinion was that the in-rack sprinklers would not be capable of preventing the fire from spreading to combustibles located above the in-rack sprinklers and that the ceiling sprinklers were inadequate to control a fire limited to just the combustibles in the racks above the in-rack sprinklers.

Certainly, Dr. Beyler's work with the LES model in the McFrugal's Warehouse fire litigation in 1998 was pioneering, particularly considering that, ten years later, it is still not possible to reliably predict sprinkler activation times and control/suppression by sprinklers with today's state-of-the-art models such as the Fire Dynamics Simulator (FDS), Version 5.

To determine whether or not Dr. Beyler was actually a pioneer, or whether his work as an expert witness on the McFrugal's Warehouse fire was in the realm of "junk science" masquerading as the work of an expert, the following excerpts from a draft document titled "Fire Dynamics Simulator (Version 5) Verification & Validation Guide (Volume 1: Verification)" dated May 30, 2007, are of interest:

"This guide is based in part on the 'Standard Guide for Evaluating the Predictive Capability of Deterministic Fire Models,'" ASTM E 1355 [1]. ASTM E 1355 defines model evaluation as 'the process of quantifying the accuracy of chosen results from a model when applied for a specific use.' The model evaluation process consists of two main components: verification and validation. Verification is a process to check the correctness of the solution of the governing equations. Verification does not imply that the governing equations are appropriate; only that the equations are being solved correctly. Validation is a process to determine the appropriateness of the governing equations as a mathematical model of the physical phenomena of interest. Typically, validation involves comparing model results with experimental measurement. Differences that cannot be explained in terms of numerical errors in the model or uncertainty in the measurements are attributed to the assumptions and simplifications of the physical model." (Page i)

"Evaluation is critical to establishing both the acceptable uses and the  limitations of a model. Throughout its development, FDS has undergone various forms of evaluation, both at NIST and beyond. This guide provides a survey of work conducted to date to evaluate FDS." (Page i)

"The software package is a computer model that may or may not have predictive capability when applied to a specific set of factual circumstances. Lack of accurate predictions by the model could lead to erroneous conclusions with regard to fire safety. All results should be evaluated by an informed user." (Page iii)

"FDS does not limit the range of most of the input parameters because applications often push beyond the range for which the model has been validated. FDS is still used for research at NIST and elsewhere, and the developers do not presume to know in all cases what the acceptable range of any parameter is. Plus, FDS solves the fundamental conservation equations and is much less susceptible to errors resulting from input parameters that stray beyond the limits of simpler empirical models. However, the user is warned that he/she is responsible for the prescription of all parameters. The FDS manuals can only provide guidance." (Page 31)

"The grid size is the most important numerical parameter in the model, as it dictates the spatial and temporal accuracy of the discretized partial differential equations. The heat release rate is the most important physical parameter, as it is the source term in the energy equation. Property data, like the thermal conductivity, density, heat of vaporization, heat capacity, etc., ought to be assessed in terms of their influence on the heat release rate. Validation studies have shown that FDS predicts well the transport of heat and smoke when the HRR is prescribed. In such cases, minor changes in the properties of bounding surfaces do not have a significant impact on the results. However, when the HRR is not prescribed, but rather predicted by the model using the thermophysical properties of the fuels, the model output is sensitive to even minor changes in these properties." (Page 31)

"The sensitivity analyses described in this chapter are all performed in basically the same way. For a given scenario, best estimates of all the relevant physical and numerical parameters are made, and a "baseline" simulation is performed. Then, one by one, parameters are varied by a given percentage, and the changes in predicted results are recorded. This is the simplest form of sensitivity analysis. More sophisticated techniques that involve the simultaneous variation of several parameters are impractical with a CFD model because the computation time is too long and the number of parameters too large to perform the necessary number of calculations to generate decent statistics." (Page 31)

"The most important decision made by a model user is the size of the numerical grid. In general, the finer the numerical grid, the better the numerical solution of the equations. FDS is second-order accurate in space and time, meaning that halving the grid cell size will decrease the discretization error in the governing equations by a factor of 4. Because of the non-linearity of the equations, the decrease in discretization error does not necessarily translate into a comparable decrease in the error of a given FDS output quantity. To find out what effect a finer grid has on the solution, model users usually perform some form of grid sensitivity study in that the numerical grid is systematically refined until the output quantities do not change appreciably with each refinement. Of course, with each halving of the grid cell size, the time required for the simulation increases by a factor of 24 = 16 (a factor of two for each spatial coordinate, plus time). In the end, a compromise is struck between model accuracy and computer capacity." (Page 32)

"Some grid sensitivity studies have been documented and published. Since FDS was first publicly released in 2000, significant changes in the combustion and radiation routines have been incorporated into the model. However, the basic transport algorithm is the same, as is the critical importance of grid sensitivity. In compiling sensitivity studies, only those that examined the sensitivity of routines no longer used have been excluded." (Page 32)

"In an effort to validate FDS with some simple room temperature data, Zhang et al. [33] tried different combinations of the Smagorinsky parameters, and suggested the current default values. Of the three parameters, the Smagorinsky constant Cs is the most sensitive. Smagorinsky [34] originally proposed a value of 0.23, but researchers over the past three decades have used values ranging from 0.1 to 0.23. There are also refinements of the original Smagorinsky model [35, 36, 37] that do not require the user to prescribe the constants, but rather generate them automatically as part of the numerical scheme." (Page 34)

"Radiative heat transfer is included in FDS via the solution of the radiation transport equation for a non-scattering gray gas, and in some limited cases using a wide band model. The equation is solved using a technique similar to finite volume methods for convective transport, thus the name given to it is the Finite Volume Method (FVM). There are several limitations of the model. First, the absorption coefficient for the smoke-laden gas is a complex function of its composition and temperature. Because of the simplified combustion model, the chemical composition of the smoky gases, especially the soot content, can affect both the absorption and emission of thermal radiation. Second, the radiation transport is discretized via approximately 100 solid angles. For targets far away from a localized source of radiation, like a growing fire, the discretization can lead to a non-uniform distribution of the radiant energy. This can be seen in the visualization of surface temperatures, where "hot spots" show the effect of the finite number of solid angles." (Page 34)

"Errors of 100 % in heat flux were caused by errors of 20 % in absolute temperature." (Page 34)

"Moisture content of wooden fuels is very important and difficult to measure." (Page 35)

"Flame spread over complicated objects, like cables laid out in trays, can be modeled if the surface area of the simplified object is comparable to that of the real object. This suggests sensitivity not only to physical properties, but also to geometry. It is difficult to quantify the extent of the geometrical sensitivity." (Page 35)

"As a rule of thumb, in simulations of limited resolution FDS predictions are more reliable in the farfield because the substantial numerical diffusion mimics the unresolved sub-grid scale mixing. This is hard to quantify other than through comparisons with experiment. In some of the sensitivity studies discussed above, the authors conclude that the model works best with a cell size of a given value, and often this cell is not the smallest one tested. In these cases, the authors have found a flow scenario where the unresolved convective mixing is almost exactly offset by numerical diffusion. This is fortuitous, but the conclusion does not necessarily extend to other scenarios. The disadvantage of any turbulence model, large eddy simulation included, is that good results are not guaranteed on grids of limited resolution. The advantage of LES over other turbulence models is that the solution of the actual governing equations, not a temporal or spatial average, is obtained as the mesh is refined." (Pages 35 and 36)

"The same can be said for phenomena closer in to the fire. However, grid resolution is more critical for near-field phenomena because numerical diffusion near the fire on coarse grids does not have the same fortuitous effect as it does on far-field results. In general, coarse resolution will decrease temperatures and velocities by smearing the values over the large grid cells. This can affect the radiative flux, convection to surrounding solids, and ultimately flame spread and fire growth."(Page 36)

Excerpts from the "Fire Dynamics Simulator (Version 5) Technical Reference Guide" (NIST Special Publication 1018-5) dated October 1, 2007, also of interest in this discussion are as follows:

"Sufficient evaluation of any model is necessary to ensure that users can judge the adequacy of its technical basis, appropriateness of its use, and confidence level of its predictions. This document provides the theoretical basis for the Fire Dynamics Simulator (FDS), following the general framework set forth in the "Standard Guide for Evaluating the Predictive Capability of Deterministic Fire Models," ASTM E 1355 [3]." (Page I)

"The US Department of Commerce makes no warranty, expressed or implied, to users of the Fire Dynamics Simulator (FDS), and accepts no responsibility for its use. Users of FDS assume sole responsibility under Federal law for determining the appropriateness of its use in any particular application; for any conclusions drawn from the results of its use; and for any actions taken or not taken as a result of analysis performed using these tools." (Page iii)

"The idea that the dynamics of a fire might be studied numerically dates back to the beginning of the computer age. Indeed, the fundamental conservation equations governing fluid dynamics, heat transfer, and combustion were first written down over a century ago. Despite this, practical mathematical models of fire (as distinct from controlled combustion) are relatively recent due to the inherent complexity of the problem. Indeed, in his brief history of the early days of fire research, Hoyt Hottel noted 'A case can be made for fire being, next to the life processes, the most complex of phenomena to understand' [5]." (Page 1)

"The basic idea behind the LES technique is that the eddies that account for most of the mixing are large enough to be calculated with reasonable accuracy from the equations of fluid dynamics. The hope (that must ultimately be justified by comparison to experiments) is that small-scale eddy motion can either be crudely accounted for or ignored."(Page 2)

"Version 1 of FDS was publicly released in February 2000, version 2 in December 2001, version 3 in November 2002,\ and version 4 in July 2004. The present version of FDS is 5, first released in October, 2007."(Page 5)

"The major assumptions of the model, for example the large eddy simulation technique and the mixture fraction combustion model, have undergone a roughly 40-year development and are now documented in popular introductory text books. More specific sub-models, like the sprinkler spray routine or the various pyrolysis models, have yet to be developed to this extent. As a consequence, all documents produced by NIST staff are required to go through an internal editorial review and approval process. This process is designed to ensure compliance with the technical requirements, policy, and editorial quality required by NIST. The technical review includes a critical evaluation of the technical content and methodology, statistical treatment of data, uncertainty analysis, use of appropriate reference data and units and bibliographic references."(Page 13)

"Any user of the numerical model [FDS] must be aware of the assumptions and approximations being employed. There are two issues for any potential user to consider before embarking on calculations. First, for both real and simulated fires, the growth of the fire is very sensitive to the thermal properties (conductivity, specific heat, density, burning rate, etc.) of the surrounding materials. Second, even if all the material properties are known, the physical phenomena of interest may not be simulated due to limitations in the model algorithms or numerical grid. Except for those few materials that have been studied to date at NIST, the user must supply the thermal properties of the materials, and then validate the performance of the model with experiments to ensure that the model has the necessary physics included. Only then can the model be expected to predict the outcome of fire scenarios that are similar to those that have actually been tested." (Page 71)

Discussion

Good judgment and an ability to have a "feel" for the answers to engineering problems are traits that are essential to develop in the practice of engineering. One of the problems with those involved with modeling is that modelers have a tendency to believe the results of their modeling efforts, regardless of whether or not the results are logical and make sense.

In the litigation involving the fire at the McFrugal's Warehouse, Dr. Beyler stated that the installation of in-rack sprinklers in the multi-row racks would have had little impact on the first fire that occurred in the warehouse, and he utilized the LES model to support his opinion. Certainly, you don't need to be an expert (or a Ph.D.) to have serious doubts about Beyler's opinion regarding the capabilities of in-rack sprinklers.

It seems obvious that it's possible, though not necessarily probable, that the initial fire in the warehouse could have been controlled by the operation of a single in-rack sprinkler. It likewise seems obvious that, even if the fire had spread above the in-racks sprinklers as predicted by Beyler, in-rack sprinklers would have operated sooner than the ceiling sprinklers. This would have caused the water flow alarm in the system to operate sooner than it actually did and would have resulted in the NOFD arriving at the building sooner. Given that the size of a fire typically grows exponentially with time, if the NOFD could have arrived at the building just 30 seconds earlier, the entire fire scenario that firefighters confronted could have been markedly different.

If the NOFD was able to control and extinguish the fire without in-rack sprinklers, it is highly probable that the NOFD would have been able to control and extinguish the first fire much more easily and quickly with the assistance of in-rack sprinklers. It is even possible that, with the assistance of in-rack sprinklers, the staff at the building would have been able to control the fire prior to the arrival of the NOFD. Given the above, Beyler's opinion, based upon the results of the LES model, that in-rack sprinklers would not have affected the results of the first fire is, quite frankly, silly.

Fortunately for Beyler, Charlie Martin's discovery of Grinnell's price quotation for the installation of in-rack sprinklers in the multi-row racks effectively ended the case. (As is typical, the litigation was settled out of court.) While the legal rankling over a fire that occurred more than 12 years ago is of little consequence at this point in time, the reason for interest in the McFrugal's fire litigation now is that Dr. Beyler is once again utilizing fire modeling to predict the capabilities of sprinklers to control a fire. The purpose of this modeling is to conclusively demonstrate that the "ganged" operation of roof vents will not have any adverse effects on the operation of a sprinkler system. (More on that in Part 2 of this article.)

One final point that should be made regarding Beyler's work in the McFrugal's litigation. While both the NIST "Verification & Validation Guide" and "Technical Reference Guide" for Version 5 of the FDS state that experiments should be conducted to verify that the model will actually produce valid results, in the McFrugal's litigation Beyler conducted no experiments to verify the model's predictions. It should be noted that the National Institute of Standards and Technology (NIST) conducted experiments to calibrate and assess their use of the FDS in the World Trade Center collapse investigation, The Station Night Club investigation and the NIST portion of the investigation of the fire at the Cook County Administration Building.

With regard to the predictions of sprinkler operating times by the FDS in Beyler's research on the "ganged" operation of roof vents, Beyler simply ignores the experimental results on sprinkler operating times determined in the National Fire Protection Research Foundation large-scale tests (Tests P-1 through P-5) on the interaction of sprinklers, roof vents and draft curtains conducted in 1997/1998.

Certainly, the Fire Dynamics Simulator is an important tool used in building fire safety design. It would be a shame if the FDS were given a "black mark" by engineers who use the model outside the range of its capabilities.

Richard Schulte is a 1976 graduate of the fire protection engineering program at the Illinois Institute of Technology. After working in various positions within the fire protection field, he formed Schulte & Associates in 1988. His consulting experience includes work on the Sears Tower and numerous other notable structures. He has also acted as an expert witness in the litigation involving the fire at the New Orleans Distribution Center. He can be contacted by sending e-mail to rschulte@plumbingengineer.com.