Download Causes & Methods for Analyzing Spontaneous Ignition in Fire Investigations and more Study notes Literature in PDF only on Docsity!
The author(s) shown below used Federal funds provided by the U.S.
Department of Justice and prepared the following final report:
Document Title: Spontaneous Ignition in Fire Investigation
Author: James G. Quintiere, Justin T. Warden, Stephen
M. Tamburello, Thomas E. Minnich
Document No.: 239046
Date Received: July 2012
Award Number: 2008-DN-BX-K
This report has not been published by the U.S. Department of Justice.
To provide better customer service, NCJRS has made this Federally-
funded grant final report available electronically in addition to
traditional paper copies.
Opinions or points of view expressed are those
of the author(s) and do not necessarily reflect
the official position or policies of the U.S.
Department of Justice.
Report Title: Spontaneous Ignition in Fire Investigation
Award Number 2008 - DN-BX-K Author(s): James G. Quintiere, University of Maryland Justin T. Warden, University of Maryland Stephen M. Tamburello, University of Maryland Thomas E. Minnich, Technical Manager, NCFS
Abstract
The results of this project will provide the investigator with an understanding of spontaneous ignition. It presents a scientific description of spontaneous ignition, and presents the theory of the subject. Three scenarios for ignition are presented: (1) a cold material in hot surroundings, (2) a material on a hot surface, and (3) a hot material in cold surroundings. It defines the properties needed for spontaneous ignition as two parameters P and M. It presents two test methods for measuring these properties. The test requires a specialized oven and its design and operation are described. The Frank-Kameneskii method is described along with the crossing point method. The latter is simpler, and its use is illustrated in an extensive study on developing the properties of linseed oil on cotton over a range of mass concentration loadings. This is the widest range of results for this common spontaneous ignition medium. Examples are presented on how to use the material, a technique for measuring the heat transfer coefficient of the oven test, and a database of property data was compiled.
Executive Summary
Spontaneous ignition is a fire cause that can easily be missed by an investigator. It is likely the cause of many fires that have defied recognition due to its slow and nearly invisible origin. In the investigations of these cases, the phenomenon is difficult to determine due to the destruction of evidence from the fire and lack of information on the symptoms of spontaneous ignition. Spontaneous ignition is a complex process and many investigators may not fully understand the factors that promote it, and how to assess its likelihood. It occurs in solids, liquids, and gases. In the latter, it is more commonly known as “auto-ignition”. Most common to flaming fires events is the involvement of solid configurations in spontaneous ignition. The solid-based fire is the focus of this work. It can occur in well-known media as cotton impregnated with linseed oil, moist haystacks, and least known as buried woody debris, and recycled plastic storage piles. Heat transfer, or the inability to cool a hot material, initiates the process. The process starts as a chemical reaction that is not yet deemed combustion, and the consequence of ignition can be smoldering or flaming. The key variables are the size of the material and nature of the heat transfer, and the particular chemical and physical properties of the material. In fire investigation, it is important for the investigator to recognize the signature of spontaneous ignition, to learn how to estimate if it were possible in the particular fire scenario, and to appreciate how to establish samples for measurements. In addition, it is importance for forensic laboratories to have the capability and understanding to utilize methods to achieve such properties. Two main goals of this study were considered: (1) Provide investigators with a scientific understanding of spontaneous ignition and its analysis, and (2) Inform forensic laboratories on practical methods for measuring properties needed in analysis of spontaneous ignition. The latter goal was not fully achieved as the focus was on developing the laboratory techniques first. This was part of the subject in two MS theses, and that work took the bulk of the time. However, those theses and this report provide information to potential forensic laboratories that wish to adopt these measurement techniques.
A first step was to learn about the experience of investigators in the field with respect to spontaneous ignition. A survey of over 200 investigators was made that provided information about their experience with spontaneous ignition. In one case, a dwelling filled with garbage to knee-high exhibited the signs of the beginnings of spontaneous ignition, as charring was found in a portion of the garbage. The majority of the survey revealed that most of the incidents believed to have been caused by spontaneous ignition involved linseed oil infused materials, wet hay storage and problems with clothing just taken from a dryer. Other incidents involved potting soil, mulch and a variety of other materials. This project focused on linseed oil and cotton. In addition, two Master of Science theses studies were undertaken. J. Worden [1] explored methods of measuring the spontaneous ignition properties of cotton impregnated with varying concentrations of linseed oil. S. Tamburello [2] examined the procedures for predicting the possibility of spontaneous ignition for several fire scenarios, and also compiling a data set of properties for a range of materials.
Scientific description of spontaneous ignition A chemical reaction that releases energy is termed exothermic. Any substance that possesses exothermiscity by itself or in the presence of air is prone to an unstable condition in which its internal temperature can increase significantly. The substance can be a solid, liquid or gas. The chemical reaction can be a decomposition of the substance, or an oxidation due to, for example, the presence of air. For example, the decomposition of the propellant ammonium perchlorate (NH 4 ClO 4 ) is exothermic. A sufficient temperature increase can cause an accelerating reaction front propagating through the material. Gaseous product can be trapped causing large pressure increase or fissures in the material. The reaction front can travel at less than the speed of sound in the material (deflagration) or greater than (detonation); explosive materials fall into this category of exothermically decomposing solids. Wood is even known to have an exothermic decomposition reaction at high temperature. Wood fiberboard is known to have exothermiscity. Porous solids in air can oxidize exothermically throughout, and combustion gasses mixed with air can do the same. For the porous solid, the reaction may lead to a high enough temperature to initiate smoldering. The
This process for the cotton with linseed oil is indicative of spontaneous ignition in general. In this case the oven temperature and the cube size were sufficient to cause ignition in about 50 minues. The chemical reaction initially was a low level reaction and not deemed combustion. But eventually, the core temperature increased and promoted a faster chemical reaction. At about 50 minutes ignition occurred that drove the reaction even faster and this culminated in fully developed sustained smoldering. The chemical reaction associated with smolder in now clearly combustion. However, this reaction is very incomplete in terms of its products. The later flaming state is more complete, and has a much higher energy output. As the center of the cubic has the largest distance through which to lose heat, it traps the chemical energy, and the core receives that highest temperatrue and degree of chemical degradation. This internal damage indictive of the process promoting spontaneous ignition. Porous oxidizing or exothermically decomposing solids are the substances considered from here. But it should be recognized that gases and liquids can undergo the same process, and the explanations given here for the porous solids also would apply. The principle theoretical basis of the analysis for spontaneous ignition is simple, but highly mathematical in its nature. Here, this complex mathematical presentation will be avoided, but equations cannot be eliminated as a scientific method for quantitative results are the goal. The original theory on spontaneous ignition was put forth by Frank-Kamenetskii (F-K) in 1938 [3] and forms the basis for much of the analysis done and discussed here, as well as the measurement method for characterizing materials. In addition to the two MS these referenced here, there are several excellent treatises that describe many aspects of the subject [4-7]. These and other literature have been investigated in the course of this study. The simple theory, while powerful in its ability to make practical predictions, does ignore many factors such as the diffusion of oxygen through the porous material, moisture that can promote biological energy, and competing chemical reactions in mixtures such as cotton and linseed oil, as both oxidize. A graphical description of the process of achieving spontaneous ignition is represented in the figure below:
It shows the core temperature, T o, rise over the environment or oven air temperature, T A. As the air temperature is increased the early oxidation process results in a slight increase of the core temperature. So for this material, in its particular geometric configuration, up to point I, there is only a slight possible temperature increase. For 0 to I, the curve represents the various steady state temperatures in which the reaction continues and the core is slightly elevated. This range of possibilities is termed subcritical, and represents what is commonly called “self-heating”. Nothing dramatic happens. Of course, in reality this process can never be steady because the chemical reaction will cause the material to be depleted, and then will stop. Here, this is considered to happen later. Now the nature of the chemical reaction and the ability of the material to lose heat make the self-heating process unstable. This instability occurs because the chemical reaction is a very sensitive function of temperature, and increases sharply as the temperature increases. If the heat loss cannot keep up with the increase in chemical energy, the temperature will continue to rise. This feedback between an increase in the chemical reaction rate and the increase in temperature can lead to an instability and “thermal runaway” to a new state of equilibrium. This
The F-K theory with one-dimensional heat conduction and a chemical source term forms the basis of mathematical and experimental analyses. In a one- dimensional model a specific dimension such as thickness or radius will be key, and each problem defines this variable specifically. In general it will be designated as r , but it has specific meaning to each problem. The next variable of importance is the temperature that is driving the initial heat transfer, and here will called T R, a reference temperature. This could be the temperature of the environment, the temperature of a hot surface adjacent to the material in question, or the temperature of the interior of the material following some processing. Three scenarios have been depicted: o A cold material in warm surroundings o A cold material in contact with a hot surface o A hot material exposed to a cold environment. While these cases are not exclusive, they do represent many scenarios of interest. Other scenarios of interest, such as heating by radiation, would need to be developed or sought in the literature. However, for these three cases, the governing equations have been solved and the literature contains results that lend themselves to analysis, and an estimate of the conditions needed for spontaneous ignition. Here, only the condition “if” spontaneous ignition is possible will be identified. The F-K theory leads to a dimensionless parameter that governs whether ignition is possible. This is known as a Damkohler number given the symbol, δ. It represents the ratio of the chemical energy produced to the heat lost by conduction. Generally, if this number exceeds about 1, the condition is supercritical. However, the specific analyses for each of the scenarios given above result in values of the critical Damkohler number, δc. While the value for δ depends on the size ( r ), heating ( T R) and chemical and physical properties of the material for F-K theory, it can be represented as. Here M and P represent material
properties. Specifically, with the units of temperature (K) and R as the universal gas constant [8.314x10-^3 kJ/mol-K], and E known as the activation energy, in units of kJ/mol. The activation energy represents the energy needed to initiate the
" = r
2 TR^2 exp^ M^ #^
P
TR
%^ &^
(^ )
P = E R
chemical reaction. The lower it is, the easier to initiate. Typically values for self- heating reactions range from about 60 to 140 kJ/mol. The parameter M represents a
collection of physical and chemical properties in the form:
exp( M ) = E R^ " AQ # where
in this presentation it has units of (K/mm)^2. The other parameters include the density of the material ( ρ), the heat of combustion ( Q ), the thermal conductivity ( λ), and another chemical rate term ( A ). While specialized apparatus may be capable for measuring each of these properties, it is common in the study of spontaneous ignition to determine just the P and M properties from direct measurements on the material. This is particularly needed since the material usually has the structure of a porous array, as a pile in layers or particles, with air filling the void spaces. Hence it is necessary to characterize the properties of the array, not the individual solid. A measurement process will be discussed later, but for now results for the three scenarios will be summarily presented. More details are found in references [1-7]. The analytical process is to identify the ideal scenario that best matches your problem of interest. Then, assuming P and M are known for your material, the Damkohler number can be determined for your size and heating condition. This means δ in Eq. (1) is calculated. Then it is necessary to consult the literature (or solve the governing equations) to determine the critical value δc for your scenario. If δ < δc the problem is subcritical and ignition is not possible, but if δ > δc the problem is supercritical and ignition is possible. The following three sections will describe how the critical value can be determined for each of the three scenarios considered here. They all will depend on both conduction through the material and on heating from the material to the environment. This environmental heat transfer can be composed of both convection and radiation. It requires knowledge of heat transfer to fully appreciate and evaluate. Suffice that it is represented as another dimensionless group called a Biot number given by the symbol α. The Biot number is the ratio of the convective and radiative heat transfer to conduction in the material. Specifically,
! = hr ". In the literature, and for a first approximation, α is often considered to be
large (or infinite). It will show up in the following presentations on how to find δc. Material in a Warm Environment, δc
Hot Material in a Cold Environment, δ c This case is depicted in the figure below in which a slab of half-thickness r is initially subjected to a uniform temperature T i (taken as the reference temperature TR in the Damkohler number), while the environmental temperature is TA. This initial temperature could have been the result of processing or heating, e.g. clothes taken from a dryer.
The critical value of δ decreases as the initial temperature increases. Results are given for the slab, cylinder and sphere. The above three scenarios only allow the determination if spontaneous ignition is possible. For a given problem, if the Damkohler number computed is
Curve 0: initial condition Curve 1: cooling Curve 2: steady result, Curve 3: ignition event,
Curve 0: Initial State; Ti = TA Curve 1: Unsteady heating Curve 2: Steady condition; Curve 3: Thermal runaways;
greater than the critical value computed in one of the three scenarios, then ignition will occur. A formula to estimate the time to ignite is given and the time directly depends on the size dimension and inversely on the Damkohler number.
Material Properties The previous disussion gave methods for determining δc, but properties for the material are needed to complete this analysis. Obtaining the properties with sufficient accuracy to represent the material in the real spontaneous ignition scenario is not trivial. In many cases, the sample tested in the laboratory may not exactly be the same as that in the field. Moreover, the need to preserve the character of the porous matrix representative of the material is a must. This is why tiny samples representing the solid may not be sufficient for testing. Such testing might be accomplished by TGA/DSC devices, yet larger samples are preferred as more representative. Hence the usual method of choice is based on the F-K theory using larger samples representative of the real scenario.
F-K Oven Method The F-K test basis is oven testing with specific shape samples in a wire basket suspended in an oven. This test method corresponds to the first scenario discussed, i. e. a cold material in a hot environment. The oven temperature is set and the sample is inserted with the center temperature continuously monitored. Worden describes this test with safety considerations in some detail [1]. The test is repeated until the critical temperature is found for a given sample size, usually a cube. It is a tedious process that can take days to carry out. However once a minimum of three critical values have been found an analysis can be conducted. The procedure is to plot the left hand side of the equation below against 1/ TR.
ln " c^ # $^ T rR^ % &
'^2
(^ )
+^ ,^
= - TP
R
+ M
The Biot number must be known for the oven condition and a method and data to accomplish this is presented by Tamburello [2] and later discussed. An example of
basket in the oven serving as a calorimeter. The rate of temperature rise of the basket after suddenly being immerse in the oven allowed the determination of the heat transfer coefficient. This will discussed in more detail.
Crossing Point Method Using the same process as the F-K oven method, a short cut can be taken to obtain data for the property evaluation by doing essentially several oven temperature levels for one basket size. This method was first explored by Jones et al [9] and Chen et al [10]. The method requires the determination of the point at which the core temperature equals (crosses) the oven temperature [9] or the sample surface temperature [10]. At this point it is assumed that there is no heat transfer between the sample and the oven. This is an approximation. While the surface temperature criterion is better, its measurement is troublesome. The oven crossing point is easier to use and Warden pursued data for linseed oil and cotton by that means. The theory upon which the method is based considers, at the crossing point, that the sample has a uniform temperature equal to that of the core. The governing
equation is. This equation can be rearranged to introduce M
and the thermal diffusivity, a , directly:
ln " # dT dto^ $ % = & M + ln " # Pa^ $ % '^ (
*^ +^
, TP
o
. By determining
the slope of the core temperature rise at the time when the core temperature equals the oven temperature, data can be assembled for a given basket size at different oven temperatures. Then these data can be plotted according to the previous equation against 1/ To. The results give M and P , provided the thermal diffusivity of the material can be estimated. Warden [1] illustrates this method and gives first time results for linseed oil with cotton cloth over a range of oil loadings. The table below gives a portion of his results. The results form the ingredients of P and M , as he was able to probe more deeply by his analysis. This is discussed further.
Results for Linseed Oil and Cotton
Tabulation of Properties from the Literature Tamburello [2] tabulates values for P and M as found in the literature. He lists nearly 40 entries. Many are the same generic material with variations in their P and M values. Thus data for generic materials are not definitive, and this indicates that testing is needed in each case. Nevertheless, the database does provide some quantitave guidance on the values meaured for materials prone to spontaneous ignition.
Closing Remarks This study sought to bring the science and test methodology for spontaneous ignition to the fire investigator. Three scenarios were illustrated for making calculations to establish whether spontaneous ignition is possible. An approximate formula was given to estimate the time to achieve spontaneous ignition. Two methods, using an oven and samples in cubical baskets were described. The crossing point method is the simplest and is recommended for a first consideration in obtaining property data. This method was used to determine the properties of linseed oil on cotton, and showed increasing tendency for spontaneous ignition with increasing concentration.
Basket Size and Concentration (^) (kJ/mol) E (W/^ QA kg) (W/m^ λ - K) (m^ a (^2) /s) 10cm - 80% 11.73 3.59E+05 0.130 2.25E- 07 7.5cm - 77% 15.76 2.63E+06 0.127 2.38E- 07 5cm - 75% 16.97 5.30E+06 0.125 2.28E- 07 5cm - 50% 27.40 2.27E+08 0.104 5.22E- 07 5cm - 33.3% 42.37 2.60E+10 0.089 6.33E- 07 Gross and Robertson 16.6% 88 4.7E+13 0.046 1.06E- 07
investigator to recognize the signature of spontaneous ignition, to learn how to estimate if it were possible in the particular fire scenario, and to appreciate how to establish samples for measurements. In addition, it is importance for forensic laboratories to have the capability and understanding to utilize methods to achieve such properties.
1. Problem Addressed
The principal goal of this study was to review the subject of spontaneous ignition in solid arrays and provide a treatise on subject useful in fire investigation. The specific goals are listed below:
- Provide investigators with a scientific understanding of spontaneous ignition and its analysis, and
- Inform forensic laboratories on practical methods for measuring properties needed in analysis of spontaneous ignition.
As the subject is complex, a description of the science is needed in its simplest form. Still it is likely that an engineering knowledge is needed to fully grasp the analytical aspects of the subject. However, the key points should be accessible to the average investigator. Much of the subject is not presented in a quantitative manner for the investigator, and the investigator can be at a loss for how to seek analysis. Also testing is empirical and some old standard methods only give suggestive answers. In the science of spontaneous ignition, methods exist and should be translated to forensic procedures for more quantitative analyses. Data for such quantitative testing is dispersed throughout the literature, and a database on materials that have contributed to spontaneous ignition can be a useful reference to the investigator. In short, the science of spontaneous ignition needs to be translated into fire investigation.
2. Literature
The specific literature reviewed and used in this subject is contained in two Master of Science theses. The first by Justin Warden [1] addressed the design and operation of an oven for testing samples in order to obtain quantitative data for analysis. This study illustrates how to establish the measurement procedure and how to operate with safety. Such testing involves the placement of cubical baskets containing the material under a controlled oven temperature. The measurement of the center temperature of the material in the basket indicates the exothermiscity of the material. A nonreactive material’s temperature will just rise to approach the oven temperature; a reactive material will exceed the oven temperature. If it dramatically exceeds the oven temperature “ignition” is said to have occurred. Finding the lowest (“critical”) temperature for a given material size and configuration to just cause ignition is a key measurement to begin to determine quantitative property data. Warden examined this method and a simpler alternative known as the crossing point measurement. His work specifically examined linseed oil on cotton cloth with varying degrees of oil by weight. He determined property data needed to predict the behavior of this oil cloth array in general. We will describe those results later. The second thesis by Stephen Tamburello [2] addressed the theory of spontaneous ignition, the general methodology for determining properties by the critical temperature data, and compiled key property data for many materials. He also used a method to measure the heat transfer characteristics of the various size cubical baskets in the oven test. Knowledge of the heat transfer is a critical factor in using the critical oven test method, as previous investigators appear to have just estimated this factor. The general theory of spontaneous ignition comes from Frank-Kamenetskii [3]. His theory addresses a balance between the energy production rate from the chemical reaction (oxidation or decomposition) and heat transfer. Data for the chemical reaction must come from experiments such as the oven test method. The procedure in analysis is to solve the steady problem with these balancing terms. The problem can be posed for many configurations, but the ones fully studied are three that we will review here: (1) a material in a warm environment, (2) a material on a hot
