Separating Anecdotes from Science in Low Temperature Ignition of Wood

By: Delmar "Trey" Morrison, Ph.D., P.E., CFEI, Principal Engineer, Exponent, Inc.
4580 Weaver Parkway, Suite 100, Warrenville, Illinois 60555 / 630-658-7508

PDF Version


“Could this wooden structural component have been ignited at that temperature?” That is the question that underlies many residential and commercial fire investigations when trying to determine the first item ignited. For those involved in fire science or fire investigation, they may recognize the depth and limitations of potential answers to this question. What constitutes ignition, and what does “temperature” mean in this regard? The scientific and technical literature surrounding the thermal decomposition and ignition of wooden materials is expansive and covers over one hundred years of research and observation.

When relying upon the literature as a basis for fire cause hypothesis, is it appropriate to base resulting expert opinions on scientific literature that is one decade old? Two decades? Ten decades? The answer lies in the state of the art and how far science has progressed in that area since the original work as well as the strength of the underlying methods and peer review. Modern fire science is as rigorous as the other engineering sciences with university research, well-developed analytical techniques, and computational models. However, there is still a gap to be bridged when attempting to use this knowledge to investigate and determine the cause of accidental fires. By their nature in an accidental fire, the fuels involved and the circumstances surrounding ignition can be difficult to define after the incident has occurred. The fire often destroys the evidence necessary to unambiguously define these factors. Thus, a scientific fire investigation is necessary. In the field of fire science related to fire cause investigation, the rigor involved in the science of investigation has dramatically evolved since the introduction of NFPA 921 in 1992.

Scientific treatises are one basis, but what about anecdotal works such as trade magazine articles, compilations case studies, investigation reports, and personal communications? How can the scientific rigor be evaluated for such materials? The simple answer is that it can’t. For anyone involved in a fire investigation, one fact should be immediately apparent - fire destroys the evidence. Thus, scientific methods of data analysis and hypothesis evaluation must be completed to try to fill in the holes in the fire scenario puzzle left by the missing evidence. This process is highly dependent upon the skill, experience, and technical excogitation of the investigators. This inherent subjectivity can lead to challenges and introduce the basis for differing opinions of experts. A lack of experience can lead to a deficiency in an investigator’s ability to identify reasonable, feasible fire cause scenarios. A lack of technical vision can lead an expert to also develop insufficient hypothetical explanations for the fire cause. The answer to the question, “What caused this fire?” is often not obvious, especially when high-temperature ignition sources cannot be identified.


Certainly wood can ignite under a range of conditions due to various heat stimuli. A match can easily ignite a sheet of paper. A glowing coal can ignite a section of structural lumber. But, these are external ignition events caused by a heat source with sufficient heat flux to cause direct ignition, also known as piloted ignition. The physical characteristics of the wooden member, its surroundings, and the heat source all combine to affect the potential for and likelihood of ignition. When the intensity of the heat source drops, meaning a lower heat flux from the heat source into the surface of the fuel, then the likelihood of ignition drops as well. The tangible effects of a drop in heat flux are the drop in the heat source temperature and a drop in the target fuel surface temperature. When the temperature is much too low to cause direct ignition of wood, then the next possible mode of ignition is internal ignition of the wooden member due to exothermic (i.e., heat-generating) oxidation chemical reactions within the wood. If the internal temperature of the wood rises high enough, then internal ignition can occur. This phenomenon is known as self-ignition or spontaneous combustion.

Spontaneous combustion of wood, biomass, and many other materials is well known to occur in large stockpiles. Ignition of wood at low ambient temperature requires sufficient internal heat generation from exothermic reactions within the wood to raise the temperature until a sudden transition to ignition occurs. That internally-generated heat is lost from the stockpile to the surroundings through a combination of phenomena including phase change (e.g., vaporizing water), heat convection (e.g., off-gassing), and heat conduction. Depending upon the nature of the stockpile and the surroundings, the balance between heat generation and heat loss can be quite complex.

The latest edition of NFPA 921 provides examples of three modes of ignition for wood with the necessary exposure temperature range in increasing order:2

  1. Spontaneous combustion, >250°C (482°F) for ignition in less than a day
  2. Piloted ignition, 330°C to 375°C (626°F to 707°F)
  3. Flaming autoignition, 400°C to 600°C (752°F to 1112°F) range

As an example, wood flour spontaneous ignition tests indicated a temperature range based on size consistent with the literature and NFPA 921. Wood flour is a low-density form of sawdust that is well-aerated; thus, oxidation processes should readily occur. Example oven temperatures for ignition are provided in Table 1.3 Internal ignition occurred in these tests within a few hours.

In a scientific sense, the ambient temperature required for spontaneous combustion is low compared to piloted ignition. Thus, it is common to refer to this as “low-temperature” ignition of wood although to the lay person, the reported temperatures may be quite high. The question of how low this temperature can go is related to the nature of the material, its configuration, the heating method, the surrounding environment, and the heating history.

The concept of spontaneous combustion of wood material stockpiles is well-accepted, but single wooden structural members, which are commonly hypothesized first items ignited, offer a different self-heating to ignition picture. NFPA 921 offers the following commentary, “The scientific community has not reached consensus concerning the self-heating ignition of wood subjected to long-term heating,” in Section, presumably reflecting this uncertainty. Common wooden structural members may be pure wood or composite construction with a nominal shortest dimension on the order of two inches or less, such as beams, joists, or studs. Thinner structural materials such as plywood, wafer board, or oriented strand board (OSB) are also of interest as they have been offered as potential building materials involved in low-temperature ignition.

Historical anecdotes related to ignition of wooden building materials due to steam and hot water pipes are reported to be as low as 100°C to 120°C (212°F to 248°F). Babrauskas identified the lowest value for low-temperature ignition of wood as 77°C (171°F) based on fire investigation case studies and historical anecdotal references4. This concept of low-temperature ignition of wood has historically, and inaccurately, been referred to as “pyrophoric carbon.” Instead of peer-reviewed scientific literature, the weight of support for hypothetical low-temperature ignition of wooden structural members derives from these historical anecdotal references and more recent expert opinions from fire investigations.

Low-temperature, long-term ignition of wood is a result of a self-heating process related to the heat transfer, mass transfer, and chemical reaction within the wood. Rationally, if the changes to wood require weeks, months, or years of low-temperature heating (e.g., 77°C to120°C) to chemically and physically modify wood such that it will self-heat to the point of ignition at that temperature, then there are primarily three alternative explanations to be considered: (1) Long-term low-temperature heating creates a more reactive material where no oxygen limitation has been present; (2) Long-term low-temperature heating under oxygen limitations results in a reactive char, which suddenly becomes exposed to oxygen leading to measurable self-heating at low temperature; or (3) Sufficient heat has been retained within the structural wooden member to slowly raise the internal temperature to the point where ignition occurs.

The complexity of the self-heating of wood due to long-term exposure to low temperatures results in numerous questions regarding the dominating criteria. However, unless very special circumstances are identified, e.g., a hermetically-sealed enclosure, normal oxygen diffusion should be sufficient over long periods of time to preclude item (1) and item (2). Item (3) is also highly unlikely because if the timescale is very long, then internal heat generation is unlikely to outpace normal heat losses for standard insulation systems. Thus, no measurable internal temperature rise above ambient would be identified.

An Anecdotal Case Study

Since there are no handbook solutions for calculating the temperature exposure necessary for wooden structural member to spontaneously ignite, investigators may resort to relying upon war stories traded between investigators or antiquated anecdotes compiled from historical sources. One such anecdotal basis was debunked as supporting low-temperature ignition of wood. A fire that occurred in an apartment building in Winnipeg, Canada on January 7, 2001 was described as “the best-documented incident of recent vintage” demonstrating low temperature ignition of wood structural materials5. The Ignition Handbook excerpt is provided below:

A recent case history was documented in an apartment house fire in Winnipeg, Canada in 2001. Fire investigation revealed the point of origin at a 1-1/4” hot water supply pipe, where it passed from one story to the next within a wall cavity. The furnace was producing water at 88 - 93°C [190°F - 200°F] and was checked out to be in proper operational order. The pipe was mostly insulated by 25 mm [1 inch] thickness of fiberglass insulation, but the insulation was omitted at the area of floor penetration. The pipe penetrated an assembly which consisted of 15.9 mm [5/8 inch] thick OSB sub-floor, on top of which was a 38.1 mm [1.5 inch] thick wall plate. The investigation was carefully done and all alternative fire causes were ruled out. In this situation, the heating conditions would not be conceptualized as a single hot surface on a body otherwise at ambient temperature. The floor-ceiling space was a small, closed volume and the trapped air space heated up to a temperature higher than ambient.

The municipal investigator concluded that the fire was caused by ignition of “pyrophoric carbon” created by contact of combustible materials (OSB flooring) with hot water heating pipes in the floor/wall cavity. A photograph of the assigned origin is provided in Figure 1. Review of the fire investigation report and 24 color photographs from the fire scene revealed other likely potential causes that the investigator either discounted or did not list in the report6. The municipal investigation was incomplete; thus, the cause of the fire was truly undetermined and not due to low-temperature ignition of wood. A summary of the available information and analysis of the cause of the fire follows.

The Fire Scene

The outer walls of the apartment appeared to be 2 × 4 wooden stud wall construction faced with gypsum board. A manufactured fireplace was located within the southeast corner of the apartment in a concealed corner space. The wall of the concealed corner space separating it from the room was gypsum board mounted on steel studs. The inside surfaces of the outer walls inside the concealed space were also covered with gypsum. The floor was constructed from 15.9 mm (5/8 inch) aspenite sheathing (i.e., oriented strand board, OSB). The two vertical hot water pipes were located adjacent to the unit’s interior wall and passed only through the OSB flooring inside the concealed space. The pipes had insulation jackets that reportedly terminated at the OSB flooring inside the concealed corner space, but this detail could not be confirmed from the photographs. The investigator noted that the pipes were “too hot to touch” during his inspection and that the boiler controls were set to 88°C to 93°C (190°F - 200°F). The actual water temperature during the time leading up to the fire was unknown; however, the ambient temperature was reported to be -30°C (-22°F).

Based upon the limited extent of fire damage, the fire originated either inside the concealed corner space or in the concealed joist space below the floor. The floor sheeting was burned through in an area that encompassed the two pipes and the corner of the fireplace and straddled two joist spaces. The exact size of the burn-through pattern could not be ascertained from the available photos, but it appears that the large opening correlated to an area under the fireplace. The two vertical hot water pipes were located on the outside edge of this damage pattern as shown in Figure 1.

From the photographs and narrative provided by the investigator, three potential sources of ignition energy within the area of origin were immediately apparent: (1) the manufactured fireplace (Figure 2), (2) branch distribution wiring in the concealed space below the floor (Figure 3), and (3) the hot water piping.

Cause Analysis Critique

Although the available investigation materials were limited, there was still sufficient information that could lead an investigator to identify alternate reasonable hypothetical causes for the fire and to conclude that the cause of the fire is undetermined:

  • The investigator reported that he did not interview the “resident” of the involved apartment unit as part of his investigation because he/she reportedly “vanished” after investigators arrived. Thus, no details about the use of the fireplace prior to the fire, electrical service issues, or witness observations regarding the discovery of the fire were considered as part of the investigation or the conclusions provided in the report. Witness interviews can be very valuable and in this case could potentially either support or refute the fireplace, non-metallic (NM) wiring failures, or hot pipes as potential causes.
  • The fire investigation report states, “…floor sheathing directly in front of wall plate and the hot water pipes is charred from the bottom up.” The photos of the scene indicate that the flooring for the entire joist cavity (from joist to joist) extending west from the wall towards the fireplace was burned away. The deep charring and “VEE” pattern identified in the area of the wall plate, east of the steam pipes, is consistent with fire venting through the pipe opening in the flooring and not necessarily a consequence of the point of origin being at this location. Furthermore, there is heavy charring within the joist space directly west of the steam pipes and below the front left corner of the fireplace. These fire damage patterns do not support any one of the potential causes (i.e., hot pipes, NM cable, or fireplace) as the sole cause of the fire.
  • The assertion that the boiler was reportedly “working properly” and the controls were set to operate at a temperature of 88 to 93°C is not evidence that the temperature of the hot pipes in the area of origin was that hot around the time of the fire. The temperature of the pipes could have been different from this assumption for various reasons.
  • Although the photos indicate that the hot pipes were passing through the OSB floor, they do not indicate that the pipes were in direct contact with either the wall or the floor. There was no physical evidence to support close contact of the pipes and wood in the area of origin.
  • The report stated, “…this particular fireplace is often referred to as a ‘zero clearance’ or ‘insert’ fireplace.” The assumption that such a fireplace could be installed with “zero clearance” in direct contact with combustible materials is a common misconception. However, the manufacturer and/or model could not be identified by the investigator. Many fireplaces have very specific installation requirements for clearance to combustibles that when not followed can lead to the ignition of combustible materials in contact with the fireplace.
  • It is unclear if combustible materials were in contact with the side of the fireplace. For example, the report stated that the outer wall of the fireplace was approximately 2½ to 3 inches from the gypsum walls of the enclosure, which would have put it either in contact or in close proximity to the insulated pipes. Additionally, the fire investigation report indicates that the coiled wire plastic air intake for the fireplace examined in the downstairs unit was in direct contact with the outer shell of the fireplace and that the two fireplaces “were installed in the same fashion.” There are defined burn patterns on the left side of the fireplace that indicate that the fire may have originated in that area.
  • The photos did not indicate the presence of a steel spark/ember strip along the front edge of the fireplace to prevent embers from falling into the crevice between the fireplace and hearth tiles, igniting the OSB, and burning through to the joist space below (see Figure 4). A significant portion of the flooring beneath the front left corner of the fireplace, which was adjacent to the area where the steam pipes were located, was burned away, which is consistent with this type of fire cause. In fact, the fireplace was documented as containing partially burned wood logs, but the investigator did not further discuss this observation.

  • Although not directly addressed by the investigator, an inference can be drawn that use of the fireplace may have caused the fire within the concealed spaces. The investigator’s report noted that approximately 4 hours prior to ultimate fire discovery, the tenants of the upstairs apartment summoned the fire department because of the smell of smoke. The fire department dismissed the odor as being related to use of one of the fireplaces in the building.
  • The presence of electrically arced NM wiring in the area of fire origin was not sufficiently addressed as part of the initial fire investigation. The investigator noted that the conductors showed signs of parting arcs but stated, “…the area of the parting arcs [did] not represent heating sources or points of origin.” The unprotected wiring was stapled to the joist beneath the fireplace and was arced in the area of greatest fire damage. A significant section of the joist in the area of the arced conductors was consumed by the fire in the concealed joist space, which is consistent with a point of fire origin in that location. NM wiring can be susceptible to failure due to various scenarios when stapled down, which may be sufficient to cause lead to ignition of wood structural materials in the concealed joist space.

After reviewing the original investigation report and photographs, it is apparent that the determination of a cause being due to ignition of pyrophoric carbon (i.e., low-temperature long-term heating to ignition of wood) is not supported. Figure 5 is a sketch illustrating the spatial relationship between the potential ignition sources within the area of origin. The published case study did not mention the presence of a wood burning manufactured fireplace or severed NM wiring coinciding with the area of greatest damage within the area of fire origin. The excerpt describing this fire incident that has been presented to support the hypothesis of low-temperature ignition of wood does not fully describe the investigation and as a result is misleading at best.


For over one hundred years, investigators have been searching for the answer to low-temperature, long-term heating of thin wood structural members. Thick or massive timbers have been reported to ignite due to contact with embers and steam pipes, but the necessary dimensions, temperatures and time scales have not been established in the scientific literature. Oxygen diffusion, heat transfer, and chemical changes to wood may all play a significant role in self-heating. However, relationships between low-temperature (i.e., <100°C (212°F)) heating and spontaneous combustion have not yet been established in the scientific literature for wooden structural members. No scientific studies have definitively proven that (1) wood can be ignited in ambient temperatures below 100°C or (2) relatively thin structural wood (such as joists and beams) can be ignited when in close proximity or even direct contact over long time periods with a heat source below 100°C. Previous claims of this fact have been primarily anecdotal and based on unscientific observations.

One of the published case studies reported to support low-temperature ignition of structural members as a fire cause was reviewed in detail in this article. The review noted several key observations that were omitted from the published case study and demonstrated that low-temperature ignition of wood was inconsistent with several fire patterns and other evidence related to the case.

As a result of close review, one can conclude that it is inappropriate to use anecdotal cases as proof that thin wood structural members can be ignited at temperatures less than 100°C. As a point of caution, reliance on such anecdotes and unsupported conclusions can become a circular argument where the unproven hypothesis of the cause of a fire is taken to be the proof of the condition (i.e. low-temperature ignition of wood) that is then used as proof of the initial hypothesis of the cause.

Babrauskas stated that, “…it does not seem credible that all fires ascribed to low-temperature, long-term ignition of wood involved incompetent fire investigation”7 Open-ended statements such as this leave the door open for fire investigators to opine that since the phenomenon hasn’t been disproven that it is a real fire cause, almost as though it were a basis in negative corpus. It is not that these fire investigators are incompetent when they take this track; they are simply wrong about the underlying science. They are advancing an ignition hypothesis for which there is no scientific support.

What is the most significant lesson to be taken from a review of the pyrophoric carbon literature? Always examine the underlying technical work product including the facts and fire science before relying upon a published case study (even this one!). The hallmark of scientific investigation is to query the facts, assumptions, and hypotheses advanced in a fire investigation. Only through this approach can scientifically defensible fire cause opinions be developed.

1 The current article was updated and adapted from the manuscript, “A Review of the Hypothesis of Low-Temperature Self-Ignition of Wood,” which was published in the proceedings and presented at the 2011 Fire &

2 Chapter 5, NFPA 921 Guide for Fire and Explosion Investigations, 2014 edition.

3 Dee SJ, Cox BL, Hart RJ, Farina R, Morrison DR. Effects of cooking on the thermal ignition behavior of vegetable oil. Proceedings, 2015 Fire and Materials Conference, San Francisco, CA, Interscience Communications Limited, London, February 2015. pp. 889-904.

4 Babrauskas, V, 2003, Self-Heating, 'Pyrophoric Carbon,' And Ignitions from Hot Pipes in Chapter 14, Ignition handbook, Fire Science Publishers / Society of Fire Protection Engineers, Issaquah, WA.

5 Babrauskas, V, Gray, BF & Janssens, ML 2007, ‘Prudent practices for the design and installation of heat-producing devices near wood materials’, Fire and Materials, vol. 31, pp. 125-135.

6 Swan, K 2001, Fire investigation report, KGS200101071, Office of the Fire Commissioner, Winnipeg, Manitoba. [inc. 24 photographs, requested separately]

7 Babrauskas, V 2003, Ignition handbook, Fire Science Publishers / Society of Fire Protection Engineers, Issaquah, WA.

Categories: Volume 5 #4

The information on this website is for general information purposes only. Nothing on this site should be taken as legal advice for any individual case or situation.

This information on this website is not intended to create, and receipt or viewing of this information does not constitute, an attorney-client relationship.

© 2018 MDTC. All Rights Reserved.