Chapter 1: Introduction ________________________________________________________________________ Chapter One: Introduction Introduction The significance the propensity of wood to combust spontaneously is that spontaneous combustion of solid materials has always been one of the major hazards in construction, transport, processing and mining industries (Wang, Dlugogorski and Kennedy 2006). Spontaneous combustion involves ignition within bulk porous solid by heterogeneous oxidation; self-heating arises as a result of low-temperature exothermic reactions, or condensation of moisture. The exothermic reactions and moisture adsorption occur on both the outside surfaces of the porous system as well as the intra-particle pore surfaces within the system. Porosity and oxidizability are fundamental qualities of materials that exhibit spontaneous combustion. Babrauskas (2001) has differentiated self-heating, thermal runaway and spontaneous combustion as a means to designate, as well as to explain, the different extent of autoignition the materials exhibit. On the other hand, spontaneous ignition within bulk solid encompasses the succession of these three events. Spontaneous combustion begins with the incipient spontaneous oxidation brought upon by chemical, biochemical or physicochemical processes; ignition within solid commences once the rate of heat generation exceeds the rate of heat dissipation to the surrounding (Walker, Taylor and Ranish 1991). The ultimate state is smoldering combustion, and when conditions are right, Chapter 1: Introduction ________________________________________________________________________ transition to flaming combustion. For this reason, spontaneous combustion is delineated to include all three stages of the process: the initial self-heating, autoignition and flaming or glowing combustion; the same approach is undertaken by Wang, Dlugogorski and Kennedy (2006). In nature, coal, charcoal and carbon black are materials that are well known for their propensity to self-heating and auto-ignition, even at temperatures close to the ambient. These types of carbon materials have in recent years come to be termed as “pyrophoric” materials (Babrauskas, 2001). Pyrophoric materials, whether in gas, liquid or solid form, are materials that self-ignite below room temperature, according to International Building Code 2006. Typical examples of pyrophoric materials include silane and phosphine gas, liquid diethylaluminium chloride and reactive substances such as cesium, plutonium, potassium and rubidium. The probable reason the carbonized materials have recently been associated as pyrophoric materials could be due to the tendency of these materials to ignite at low temperatures. The propensity of coals to combust spontaneously is underscored by their chemical composition such as the rank of coal and the percentage of fixed carbon, and enhanced by high porosity and hygroscopic adsorption of moisture (Bowes 1984, Cuzzillo 1997, Wang, Dlugogorski and Kennedy 2003a). In addition, the geometry of the solid that determines the access of oxygen plays an important role in the spontaneous combustion of this class of pyrophoric carbon. If the coal has been stockpiled in a compact manner, allowing no room for oxygen access, self-heating could be delayed. Bowes (1984) has made an excellent review on the configuration of stockpiling of charcoal for shipping to minimize self-heating. On the other hand, wood Chapter 1: Introduction ________________________________________________________________________ generally is not pyrophoric, although cellulose can undergo direct oxidation at low temperatures (Moussa, Toong and Garris 1976), but these reactions are very slow. However, when wood has been exposed to prolonged heating, even at low temperatures, they become more prone to combustion – a phenomenon that has gained acceptance based upon fires originating from the ignition of wood and other cellulosic materials that have come into contact with low pressure steam pipes and hot water pipes (Babrauskas 2001). The wood members were noted to spontaneously ignite after months to 15 years after installation (Bixel and Moore 1910, Matson, Dufour and Breen 1959). It has since been hypothesized that low-temperature, long-term heating of wood leads to the formation of pyrophoric carbon, where these partially carbonized wood would ignite more readily at low temperatures than virgin wood (Ranke 1873). Pyrophoric char and chemisorption The conversion of wood to pyrophoric carbon occurs after prolonged heating. Walker (1967) believed that prolonged exposure reduced the induction period for ignition to occur. According to Walker (1967), extended storage at relatively high temperature enabled oxidation reactions to pass through the period of induction and thus attain a rapid rate of chemical heating; but exactly how oxidation is accelerated was not explained. There are cases of practical importance in built environment that provide evidences to such claims. Examples include wooden panels and beams that are exposed to flue pipes (McGuire 1969), wood members that have been exposed to high intensity radiation lighting i and storage of fibreboards (Frangi, Fontana and Schleifer 2005). i Ming Pao News, Hong Kong (13 Nov 2007) reported that a fire broke out on wood cabinet due to highintensity lighting; source: www.npinews.com Chapter 1: Introduction ________________________________________________________________________ From the chemical viewpoint, there must be certain chemical process responsible for the exothermic reaction in pyrophoric wood that causes it to ignite. Bradbury and Shafizadeh (1980a) have identified chemisorption of oxygen onto the partially carbonized wood to be a significant factor in controlling the solid phase combustion. Their experiments showed that heat generated by initial oxygen chemisorption is sufficiently intense to induce ignition at lower temperatures for chars prepared under anaerobic conditions. Indeed, oxygen chemisorption has long been identified as the primary reaction responsible for spontaneous combustion of coal chars that are prepared at severe conditions (e.g. high temperature) or from synthetic carbons (e.g. Saran char) (Teng and Hsieh 1999). However, the kinetics and mechanisms for coal chars chemisorption are on the basal planes, while these mechanisms are probably not dominant in the more amorphous wood chars (Bansal, Vastola and Walker 1970); these two classes of carbon chars are entirely different in chemical functionality and morphology. Different types of active sites, chemical functional groups and reaction pathways are responsible for chemisorption of oxygen onto wood chars. Chemisorption studies on low temperature wood char would therefore focus on characterizing the hazardous properties that promote chemisorption and determining the conditions that facilitate it. In doing so, it becomes important to elucidate the kinetics of oxygen chemisorption in order to evaluate the autoignition potential of low temperature wood chars that have been pre-exposed to prolonged heating. Thermal models To completely understand spontaneous combustion, no mathematical treatment is ever enough to deal with the complexity of phenomena involved. Research in solid-phase Chapter 1: Introduction ________________________________________________________________________ combustion phenomena includes modeling chemical reactions beyond a single chemical reaction represented in Arrhenius form; secondary reactions that arise from interaction between the primary pyrolysis products; char oxidation mechanism and deformation such as fissuring, shrinkage and cracking (Di Blasi 1993). Access of oxygen and adsorption of moisture further complicates the solid-phase model with pore opening and closure. Simplifications of varying degree have to be made in order to keep the mathematical treatment tractable. Sometimes, simple models that neglect all chemical effects and are purely thermal offer an analytical framework for an otherwise hopelessly complex problem. Thermal explosion theory is and remains important in that regard for the investigation of spontaneous combustion of wood (Bowes 1984, Blomqvist and Persson 2003, Wang et al. 2006). This is because an investigation on chemisorption may be plausible from a chemical viewpoint, but the proposition would collapse under physicochemical framework if the reaction could not generate sufficient heat or the heat generation is less than the heat loss in the bulk solid, both of which would not result in an ignition. Thermal explosion theory was first developed for spontaneous ignition or autoignition of gases by Semenov (1940) where he observed a critical ignition temperature for flammable mixtures. Semenov model deals with situations involving low Biot number where the heat conduction inside the system is much faster than the heat conduction away from its surface; temperature gradient is therefore negligible inside the system. Heat losses are controlled by convection from the surface to the surrounding atmosphere. Chapter 1: Introduction ________________________________________________________________________ Frank Kamenetskii (1969) developed a high Biot number model (Bi > 10) of thermal explosion where heat losses within the system are controlled by thermal conductivity, and that the temperature of the surface and the ambient temperature are equal. Both models are mathematical treatments of the critical heat balances between heat generation and heat dissipation during thermal explosion of reacting materials. However thermal explosion of gases, where Semenov model was originally developed for, involves very short induction period in the order of milliseconds. In that regard, Semenov model cannot be applied to the spontaneous combustion of bulk solid, where the induction period varies between hours, days or weeks, depending on the mass and storage temperature (Drysdale 1985). Babrauskas (2001) noted the long induction period involving weeks and months for spontaneous combustion of wood-based members. Frank-Kamenetskii model has been applied to study the thermal explosion of bulk solid. Examples include chemically activated carbon powders prepared in cubical baskets with side length ranging from 25.4mm to 610mm (Bowes and Cameron 1971), aluminum residue packed in cubic baskets with side lengths of 51, 76 and 102mm (Hill and Quintiere 2000) and 89mm cube of cooked and uncooked wood blocks, chips and sawdusts (Cuzzillo 1997). In addition to Semenov model and Frank-Kamenetskii model, Thomas (1958) developed a model that embodied both theories by Semenov (1940) and Frank-Kamenetskii (1969) to deal with a system involving conduction through solid and convective heat loss from the surface. Walker (1967) introduced the notions of dry combustion and wet combustion in the context of spontaneous combustion of solids. Dry combustion is related to fires originating from stacking of hot, dry sheets of fiber insulation boards directly from Chapter 1: Introduction ________________________________________________________________________ production line or hot linen piled in large baskets immediately after removal from industrial tumble driers. According to Walker (1967), water does not have a dominant role in thermal explosion by dry combustion; size effects and shape of solid are inter alia the more critical factors (Bowes 1984, Cuzzillo 1997). The effect of water is restricted to heat of wetting (Walker 1967), which can, under certain circumstances, raise the materials above room temperature, enabling spontaneous combustion to begin; but exactly how water could act as a trigger mechanism in dry combustion has not been explained. Thomas (1973) proposed an approximate theory of “hot spot criticality” to deal the spontaneous ignition that may arise from this type of storage of large quantities of hot materials in a cool environment. This study is interested in the notion of wet combustion which can be associated with spontaneous ignition in air-dry materials where the migration of bound water in air-dry materials can have an effect on the ignition process. Wood is hygroscopic; in air-dry state, bound water amounts to 10% by mass in equilibrium with common atmospheres (Walker 1967). Coals that are prone to spontaneous heating also contain appreciable amounts of water in air-dry state (Davis and Byne 1926a). When a hygroscopic solid naturally prone to exothermic reaction is considered, it is the availability of moisture and it consequential effects that render it liable to spontaneous ignition by self-heating mechanism (Walker 1967, Babrauskas 2001). This is because materials such as cellulose will undergo heterogeneous oxidation more rapidly when liquid water is present, thus generating heat at lower temperatures than would the dry material. Wet combustion is also concerned with the control of heat balance by thermal conductivity, and the operative value of Chapter 1: Introduction ________________________________________________________________________ thermal conductivity that is dependent on the amount of moisture present. For spontaneous ignition in wet haystacks, the mass transport of water vapour causes high thermal conductivity; heat is dissipated and ignition is inhibited. It is interesting to note that Frank-Kamenetskii model that has been commonly applied for spontaneous ignition of wood actually does not provide for water evaporation or migration of moisture in the model (Wang et al. 2006). 1.1 Research problems Spontaneous combustion in wood in its simple form is thermal explosion caused by wet combustion, rather than dry combustion (Walker, 1967). The revisited approach brings out the importance to redress the role of moisture, the evaporation and the associated heat and mass transfer mechanisms which are fundamental to a thermal model. This section examines these issues in turn and seeks to elucidate the research problems for each term. 1.1.1 Evaporation for low-temperature prolonged drying In models that considered water evaporation, evaporation has been treated as a source term in the energy balance. Because evaporation is an endothermic process, the source term thus becomes a heat sink. The incorporation of an internal evaporation term would result in heat drawn from inside the system. This approach does not differentiate evaporation that occurs at low temperature drying and that of high temperature heating. Chapter 1: Introduction ________________________________________________________________________ Consider wood that is heated at low temperature, or low heating rate; assuming that wood is initially at room temperature and is at its equilibrium moisture state, radiative heat flux and/or convective hot air causes evaporation at the surface. Heat is conducted inwards; temperature of the solid slab starts to increase until it reaches boiling point, where heat is consumed at latent heat. Evaporation takes place almost entirely at the surface; this is the evaporation model concurring with initial stage of drying when wood is heated from room temperature. When wood is continuously heated for a long time, or exposed to high heating rate of which moisture evaporation is fast and rapid, diffusivity of water reduces to a level where outflowing water flux becomes smaller than the rate of surface evaporation. As a result, evaporation front begins to retreat from the surface and advances inwards. Evaporation now takes place inside the system where there is a large gradient of moisture content. This is the phenomenological evaporation model for high-temperature heating or drying that has occurred at the advanced stage (Ilic and Turner 1986). The two different evaporation models are illustrated schematically in Figure 1.1 and 1.2. Chapter 1: Introduction ________________________________________________________________________ Figure 1.1: Evaporation model for low-temperature drying: water flows to the surface where evaporation takes place and consumes heat from the ambient hot air. Heat conduction Zone C Zone B Zone A Moisture diffusion Heat from hot air Latent heat Vapour Figure 1.2: Evaporation model for high-temperature drying: evaporation occurs throughout the material and is the same amount as the mass change rate of total moisture. Moisture loss is realized by changing to vapour and then vapour moving to the material surface Heat conduction Zone C Zone B Zone A Latent heat Moisture diffusion Vapour 10 Chapter 1: Introduction ________________________________________________________________________ The initial drying evaporation model (hence known as “Model 1”) and the high temperature drying evaporation model (hence known as “Model 2”) represent two different physical processes of evaporation. The incorporation of evaporation as an internal heat sink in the energy balance is a mathematical description of Model 2. The underlying assumption that evaporation occurs within wood is also fundamental to the conventional high temperature model proposed by Alves and Figueiredo (1989). By incorporating an internal evaporation term (Chan et al. 1985; Fredlund, 1988; Alves and Figueiredo, 1989; Yuen, 1998), which has been common practice for pyrolysis models used to study woods burning in fire, the models have avoided low-temperature drying evaporation completely. The consequence of adopting an internal evaporation term may not be obvious for combustion phenomenon in high temperature heating regime. However, for wood combustion phenomenon occurring at low temperature drying regime, or exposed to low heating rate, the impact is discernible in a different temperature distribution (See Figure 1.3). Model produces a uniform temperature distribution. This temperature profile concurs with drying at initial stage where thermal conductivity is relatively large due to the presence of moisture content. In Model 2, since heat is drawn from inside wood, it creates a heat sink; temperature inside wood is lower and it creates a temperature gradient for which to maintain the influx of heat from the surface. The depressed temperature curve does not reflect the right temperature distribution that involves evaporation that occurs at the surface. 11 Chapter 1: Introduction ________________________________________________________________________ Figure 1.3: Temperature profile using Model and Model during initial period of wood drying Model Model o Temperature ( C) 58 54 50 46 1.00 1.01 1.03 1.04 1.05 1.06 1.08 1.09 1.10 1.12 Distance from centre (mm) (Graph taken from Zhang and Datta (2004), Copyright ©Drying Technology) For wood drying at low temperature, it requires a different description of moisture evaporation. An alternative formulation is needed to represent evaporation which occurs at the surface, before evaporation front retreats from wood surface. The low-temperature evaporation model seeks to address some problems associated with low-temperature combustion phenomenon by firstly restoring moisture for low temperature heating where high temperature evaporation model would have rapidly vaporized the moisture; secondly redressing the importance of moisture flow at low temperature regime which has been surpassed by high temperature heating model. 12 Chapter 1: Introduction ________________________________________________________________________ 1.1.2 Liquid water transport modelling High temperature drying model does not consider liquid water transport, and moisture flow takes place essentially in vapour phase (Fredlund, 1988; Alves and Figueiredo, 1989; Yuen, 1998). The model bases its argument on that there is no continuous water phase at moderate moisture content to support liquid movement. Indeed the evaporation rate in high-temperature drying is tantamount to the same amount of mass change rate of total moisture. In other words, there is no liquid water movement. Following the discussion, research shows that there are three distinct stages for wood drying based on wood moisture content (Eriksson, Johansson and Danvind 2007). The stages can be defined as Stage X ≥ X is Stage II X is > X > X fsp Stage III X > X fsp (1.1) where X denotes the moisture content in wood, Xis is the moisture content where the continuity of liquid phase breaks down and Xfsp is the fiber saturation point. First stage of wood drying is almost non-existent since wood generally has moisture content below free water continuity ([...]... the surface 11 Chapter 1: Introduction Figure 1. 3: Temperature profile using Model 1 and Model 2 during initial period of wood drying Model 2 Model 1 o Temperature ( C) 58 54 50 46 1. 00 1. 01 1.03 1. 04 1. 05 1. 06 1. 08 1. 09 1. 10 1. 12 Distance from centre (mm) (Graph taken from Zhang and Datta (2004), Copyright ©Drying Technology) For wood drying at low temperature, ... investigation is to provide a chemical insight besides the thermophysical explanation on self -heating of wood chars 1. 3 Scope of work Taking into consideration of the above objectives, the focus of this thesis is on thermal combustion and oxygen chemisorption of preheated wood at low temperature, long- term heating The scope of work and the structure of discussion in each chapter are described as follows: 22... rapid heating, heat flow is different from the total energy flow There is however no reason for lowtemperature drying model to divide the moisture flow An alternative formulation is to combine the vapour and moisture flow into a total flow; a combined flow has been adopted by Zhang and Datta (2004) for both low and high temperature drying The combined approach has also been shown to produce satisfactory... in wood • Conclusions and recommendations Chapter Seven concludes on findings derived from the experimental and analytical investigation on self -heating and spontaneous combustion of wood exposed to low- temperature, prolonged heating condition Recommendations are proposed based on current work carried out which include the development of low- temperature drying models and the coupling of oxygen chemisorption. .. thermophysical terms on a revised thermal model for evaluating the ignition data of wood 2 Explore the interaction between fluid transport on temperature field development so as to examine the role of moisture transport on self -heating and spontaneous combustion of wood Long- term, low- temperature drying brings upon the transport of liquid water as well as the diffusion and convection of vapour within wood volume... included 1. 1.4 Oxygen chemisorption for cellulosic chars Prolonged exposure of wood char to the atmosphere affects its combustion properties While surface oxidation, driven by chemisorption is the exothermic reaction responsible for pyrolysis and smoldering combustion, the adsorption of water vapour onto char also affects the chemisorption process (Allardice 19 66) The propensity of wood char to ignite... spontaneous heating and Cone Calorimeter ignition tests Following the formulation of mathematical models for investigating thermal combustion, Chapter Four describes the experimental design to investigate the spontaneous combustion of wood in an isothermal oven exposed to low temperature heating Cone Calorimeter testing is carried out to examine the effects of moisture on thermophysical aspects of wood ignition... are discussed and compared with the experimental temperature field obtained from heating of wood cubes To examine the propensity of self-ignition of wood char chemisorption data and changes in function groups using FTIR absorbance spectrum are analysed The ignition temperatures determined from chemisorption experiments are further evaluated to examine the correlation of chemisorption and combustion characteristics... consequence of adopting an internal evaporation term may not be obvious for combustion phenomenon in high temperature heating regime However, for wood combustion phenomenon occurring at low temperature drying regime, or exposed to low heating rate, the impact is discernible in a different temperature distribution (See Figure 1. 3) Model 1 produces a uniform temperature distribution This temperature profile... green wood and preburn wood slabs to various incident heat fluxes Times to ignition and surface temperatures are measured for both piloted and spontaneous ignition modes • Methodology and experimental design for oxygen chemisorption Chapter Five reviews and develops mathematical models on oxygen chemisorption Experimental design for oxygen chemisorption includes testing two types of chars; inert wood . 46 50 54 58 1. 00 1. 01 1.03 1. 04 1. 05 1. 06 1. 08 1. 09 1. 10 1. 12 Distance from centre (mm) Temperature ( o C) Model 2 Model 1 (Graph taken from Zhang and Datta (2004), Copyright ©Drying Technology) For wood. 25.4mm to 610 mm (Bowes and Cameron 19 71) , aluminum residue packed in cubic baskets with side lengths of 51, 76 and 10 2mm (Hill and Quintiere 2000) and 89mm cube of cooked and uncooked wood blocks,. direct oxidation at low temperatures (Moussa, Toong and Garris 19 76), but these reactions are very slow. However, when wood has been exposed to prolonged heating, even at low temperatures, they