FIRE INVESTIGATION CHANGE AND EVOLUTION
PART 3: NEW TOOLS AND OLD MYTHS
This is the third and last article of this series describing the changes that have been occurring in the fire investigation field. The first article (January 1994) discussed the changes brought about through basic fire research and the development of NFPA 921, Guide for Fire and Explosion Investigation. The second (May 1994) described the basic phenomena of a room fire and their impact on fire investigations. This article concentrates on a powerful new toot, the computer fire model, which increasingly is being applied to fire investigation.
Computer-based fire models, which became available for general use only within the past 10 years, have been widely accepted in the fire safety field for predicting conditions within a room during a fire. Previously, such conditions could be determined only by conducting actual fire tests and measuring the desired parameters. Fire researchers, engineers, and computer programmers developed these models by analyzing large quantities of data obtained through fire tests. Analyses of these tests showed that the various parameters associated with a room fire under specific physical conditions could be predicted and evaluated mathematically. The parameters of greatest interest were those that would assess the survivability chances for an individual in a particular room. These parameters include the temperatures in the room of fireorigin and adjacent rooms, the concentration of several products of combustion (carbon monoxide and carbon dioxide, for example), incident radiant flux on objects in the room, and the thickness of the hot layer that forms during a room fire. The models are designed so that these parameters are given at regular time intervals during the modeled fire, making it possible to predict the fire conditions in each specified room at any given time. This information, combined with that known about humans’ tolerance levels of fire products, presents a method for assessing survivability.
A variety of computer fire models developed for specific purposes and applications are available. The models most useful to fire investigators are those that specifically model room conditions, generally known as “compartment” models. Some are designed to model a singlecompartment and others, multiple compartments. An example of a single-cornpartment model is the FIRST model, one of the earliest models, developed from the Harvard Computer Fire Model. An example of a multiroom model is FAST (Fire And Smoke Transport), which predicts fire and smoke transport in multicompartmented enclosures or structures. Both models are available from the National Institute of Standards and Technology (NISI).1
Each model has its limitations, which users must continually keep in mind. Some models for example, do not handleareas with a high length-to-width ratio, such as corridors or rooms with high ceilings. The danger of misapplication is not that the models will refuse to run but that they will run and generate erroneous results. A basic tenet of fire modeling is garbage in equals garbage out,” a reminder that the user must be careful when selecting input data. Most models will accept an data, right or wrong, and will not warn the user about data that are beyond the capacity of the model. Fire investigators should seriously consider pursuing training in the use and application of computer fire models, since most investigators may find many of the concepts and terminology foreign, which will make independent study difficult at best.2
HAZARD I AND FPETool
The NIST-developed HAZARD I computer-based fire model, now available through the National Fire Protection Association (NFPA), probably is the “best seller” among fire models.4 It incorporates the FAST model to perform the fire calculations and generate the results. HAZARD I also contains other models that can be used to predict evacuation times from the room(s) and evaluate reliability within the rooms.
Another fire model gaining in use is FPETool, also developed by the NIST. Because this model, which is actually a collection of models and correlations— some of which are customized to specific applications and others of which can be applied to most fire situations —was developed under contract with the General Services Administration (GSA), it is distributed by the NIST without charge.
One of the models included in FPETool is ASET (Available Safe Egress Time), a room model that can be used to predict the time to the onset of hazardous conditions and the time available for occupants to escape. Also included in the FPETool are correlations (a simple model generally developed directly from experimental data to help predict such things as time to fiashover), the quantity of fuel needed to flash over a room, detector activation times, and other tools useful to the fire investigator.
INPUT DATA
For all these models, the user must input basic data that describe the compartment, the fuel, and the fire. The most important, and usually the most difficult, component of the input data is the heat release rate data for the fuel being considered. [Heat release rate (HRR) is the quantity of energy given off by a fire per unit time and typically is measured in kilowatts (kW)[. The information needed includes a profile of the growing intensity of the fire and the heat release rate as a function of time. The best source for this data is actual burn tests of the fuel involved: the heat release rate is determined from this experimental data. An example would be burning a polyurethane couch in a furniture calorimeter, measuring the weight of the couch as it burns and the quantity of oxygen consumed. The heat release rate can be determined from this data. Since most people do not have access to the equipment needed to conduct these kinds of tests, the next best source of the data is reports and other published material. Often, data related to an item similar to the one being sought will be found and can be used in the model.
Another approach would be to select fire growth curves the investigator knows represent an upper and a lower boundary to possible values. The results obtained will also present a boundary of possible results. This method makes it possible tc do sensitivity analyses—the analyses ol the effects of the choices on the results These data use what is referred to as a “tsquared” fire, one that grows exponential ly—that is, the heat release rate is a function of the square of the time. t The different input fires range from “ultrafast.’ such as that involving a pool of gasoline oi a pile of thin cardboard boxes, to “slow,’ such as that involving a fire retardant treated mattress or a box of books (sec Figure 1).
Tile user supplies the data related to the specifics of the room or rooms beinj studied, such as the lt-ngth. width, anc height of the room(s); the finish material! of the interior walls, ceilings, and floors and the size and placement of opening: between the rooms. The remainder of the input data needed are physical propertie: for the specific fuel being considered including heat of combustion and weight.
FIRE MODEL OUTPUT
Most fire models yield numerical information. which the individual running the model must organize into a useful form. HAZARD I contains a routine that graphs data such as temperature or oxygen concentration as a function of time. Such a graph makes it possible to see how these variables change through the course of a fire (see Figure 2).
The investigator using computer firemodeling must keep in mind that the model is only a tool and should not be used by itself to make decisions. The model must agree with the evidence, observations, and determinations made at the fire scene—not the other way around. Determining origin and cause are the primary duties of most fire investigators, and computer fire modeling can make these tasks easier and increase investigators’ confidence
OTHER NEW TOOLS
Among some other new tools being used by fire investigators are dogs, electronic sniffers, and portable gas chromatographs
gs. Able to detect accelerants, these dogs are trained in a manner similar to dogs used to detect explosives and drugs. They can indicate the location of these materials at a fire scene. California recentl has completed its first certification test of accelerant-detection dogs. This new fire investigation tool may enable investigators to find accelerant evidence in circumstances where it previously would have been impossible. Confronted with a completely burned-out building, also known as a “black hole,” it is very difficult to determine where the fire started and the areas from which to take evidencesamples Reports show that accelerantdetection dogs have been able to detect accelerant residues in these “black holes.” enabling the investigator to collect samples
ilectmnic sniffers. Although they have been around for a while, they have not been used much on the fire scene until recently They are able to detect small concentrations of many different hydrocarbons and other compounds
Portable gas chromatographs. Thee permit on-scene chemical analysis of evidence samples at the fire scene, which ma improve tile results because the samples are “fresh” and the possibility for contamination is reduced
FIRE GROWTH RATE DATA
OLD FIRE INVESTIGATION MYTHS
Fortunately, most of the many myths related to fire investigation have gone the way of leather hose and back steps on fire apparatus. However, those that still exist are being used by the ignorant fire investigator who has not kept up with the training and education that are so necessary. NFPA 921, Manual for Fire and Explosion Investigation, was written in part to refute these myths and provide the investigator with a credible reference. NFPA 921 discredits the myths by presenting the scientific explanations of the phenomena involved. Most of the myths were started based on observations made after fires. After investigators had observed these phenomena a number of times, they began to associate the observations with their determinations of origin and cause. These observations in time came to become direct “evidence” of a particular fire cause, and investigators would ignore indicators that were contrary to this “evidence.”
Spat ting of concrete. One fire investigation myth still widely held in some circles is that spalling of concrete indicates an incendiary (arson) fire. Spalling is a process wherein pieces or sections of concrete break loose from the floor or walls, leaving behind shallow craters. The exact mechanism for spalling is not precisely known, but what is known is that it is a reaction of concrete to heating at its surface. When an ignitable liquid is burned on a concrete surface, radiant heat energy from the fire plume may cause the concrete to spall. However, radiant energy from other fuel sources, such as that from a polyurethane couch or automobile tire burning near a floor, also may cause the concrete to spall. The radiant energy’ produced after flashover may be intense enough to cause spalling.
The fire investigator, when confronted with spalling, must consider the kinds of fuels that were available in the areas and the intensity of the fire. For spalling to be determined an indicator of accelerant use, the fire investigator must establish that one was actually present and. if so. that it was intentionally used to start the fire. The investigator should take fire-scene samples from the area where spalling occurred and from other areas as well to determine through laboratory analysis that these ignitable liquids were present. In many cases, ignitable liquids can be present at fire scenes for very normal reasons. Gasoline residue in a garage, for example, should be expected. Determining that a fire was incendiary solely on the basis of spalled concrete is totally invalid and will not hold up under scrutiny.
Depth of char. A fire investigation myth that has been fading is that the depth of char on a piece of wood found at a fire scene indicates how long that wood has burned and the location at which the fire burned the longest is the site of the fire’s origin. Fire research and testing have shown, however, that the burning rate of wood is highly variable and depends on many factors, such as moisture content, wood type, wood finish, fire ventilation, and radiant exposure of the wood.
Color of flames and smoke. In the days of wool and cotton furniture, the color of smoke generated from fires that originated in these furnishings was a little more gray than black. During those times, black smoke with yellowish flames at a fire scene was one of the indications of an incendiary fire, and an investigator would ask witnesses and firefighters at the scene whether they saw black smoke with yellowish flames. Today, however, a great deal of the furnishings in residences and businesses are made of polymers (plastics) derived from the same hydrocarbon raw materials found in gasoline. When these materials burn, they give off very black smoke and their flames tend to be yellowish. Observing these factors at a fire today, therefore, provides the fire investigator with little or no information related to the fire’s cause. Yet, some investigators still ask, “What color was the smoke?”
Low burning. This myth holds that because a fire generally burns upward, the lowest point of burning is the site of the fire’s origin. The problem with this myth and misapplication of fire dynamics is that low burning may result from other causes. In many cases, for example, fire spreads downward due to natural and expected causes such as the falling of burning material from above, melting and dripping of burning plastics, radiant ignition of materials, and effects of ventilation. The fire investigator should consider all these patterns in the room under investigation—not rely on just one—before determining the fire’s site of origin.
Multiple origins. According to this myth, establishing that there were multiple and unrelated points of origin proved the fire was incendiary. The fire investigator should understand that nonincendiary fires can produce patterns that may give the appearance that there were multiple points of origin. Cursory examination of or reliance on other myths such as low burning or spalling concrete may indicate that there were multiple origins and prompt the investigator to make an incendiary determination. A careful examination of all of the patterns and correctly attributing them to their sources often reveals that there was only one point of origin. At this point, careful investigation must rule out all accidental fire causes, and evidence that may support the cause determination must be collected. It is inadequate to make an incendiary determination based only on origin determination. A careful origin determination that does not rely on myths or invalid assumptions may reveal multiple points of origin. If the investigator has ruled out nonincendiary reasons for these patterns, then the possibility that the fire had multiple points of origin because it was an incendiary fire should be considered, and the investigator should proceed with the cause determination at each of these points.
THE FUTURE
Fire investigation is changing more now than ever before. Much of the art is being replaced by science. The myths are steadily dying, some faster than others. New tools will help the fire investigator to do a more accurate job. The ultimate purpose of fire investigation is not to determine origin and cause but to reduce and eliminate death, injury, and destruction of property caused by fire. When we fully understand how and why fires start, only then can we take steps to prevent them from happening again. Better fire investigation done by competent and properly trained fire investigators will result in fewer fires.
Endnotes
- Fire NIST can provide ready-to-use tire models with supporting documentation. The Building and Fire Research Laboratory (BFRL) at the NIST is a government research facility dedicated to providing help to the tire service, fire protection, and fire investigators. To learn more about the NISI’, write to NIST, Building and Fire Research Laboratory, Gaithersburg. MD 20899; (301) 975-6850.
- The National Association of Fire Investigators (NAF1) is holding a special one-day seminar on computer fire modeling as part of its summer fire investigator training seminar to beheld in the Chicago area July 26-29, 1994.1 w ill be the primary instructor. Contact NAF1 at (312) 427-6320 for information or a brochure
- Contact the NFPA at I Batterymarch Park; Quincy, MA 02269; (617) 770-3000.
