Thermal combustion and oxygen chemisorption of wood exposed to low temperature long term heating 6

This section sought to analyse wood combustion from the point of a thermal problem but also explored the effect of moisture on piloted and spontaneous ignition via changes in thermophysi

Chapter Six: Results Analysis and Discussion

6 Introduction

This chapter discusses the results on the moisture effects on pure thermal model, evaluates the evaporation and moisture transport using porous model for spontaneous combustion of wood, and lastly analyses the results of oxygen chemisorption and the ignition temperature of wood chars

6.1 Heat transfer using thermal model

The comput ational model constructed as a pure thermal model in Section 4.3.2 using Fluent®6.3 provided the heat simulation in the solid slab The revised analytical

model as embodied by Equations (4.24) to (4.29) incorporating moisture-mediated terms, on the other hand, yielded data that contained the effects of moisture These data were analysed alongside experimental data obtained from Cone Calorimeter for green and preburn wood This section sought to analyse wood combustion from the point of a thermal problem but also explored the effect of moisture on piloted and spontaneous ignition via changes in thermophysical properties; surface temperatures were compared with that simulated using pure heat conduction model and those calculated by heat transfer analytical model allowing for changes in thermophysical terms

6.1.1 Piloted ignition - surface temperatures in green wood

Piloted ignition is defined as the initiation of flaming combustion in the presence of a pilot (Spearpoint and Quintiere 2001) Critical surface temperature criterion was chosen over other criteria such as mass loss rate to determine ignition for engineering analysis, since a number of researches has shown the viability of using critical surface temperature criterion (Thomas, Simms and Theobald 1959, Lawson and Simms 1952, Simms 1960, Simms 1963, Simms and Law 1967, Atreya

1983, Janssens 1991a)

In Table 6.1, the measured surface temperatures at ignition for green wood, obtained from Cone Calorimeter, have been discussed in Chapter 4 This section discusses the use of revised analytical model where the effects of moisture on ignition temperature

ig

T are taken into account through the thermophysical terms For green wood, the cooling modulusβ, as defined in Equation (4.25) ranges from 0.194 to 4.11 (to 3 significant figures) as shown in Table 6.1 The ignition temperature T igis calculated using Equation (4.24) and tabulated as “calculated surface temperature” in Table 6.1

Table 6.1 Calculated surface temperature for green wood in piloted ignition

Samples Incident

heat flux

e

q′′ (kW/m2)

Time to piloted

ignition t

(s)

Cooling Modulus β

=( / ) * (h k αt)1/ 2(- )

Calculated surface temperature

ig

T (°C)

Measured surface temperature

to 4.11 i.e β  1, the analytical model calculated surface temperatures using Equation (4.24) yielded 455.4°C for sample G6 which is in good agreement with the measured temperature of 460ºC Samples G7 and G8 showed a bigger variation between calculated (439.7°C and 400°C) and measured surface temperatures (345.4ºC and 342.4ºC) The exceedingly long heating time, which was more than two hours in G7 and G8, probably has caused changing thermal thickness, and hence deviation from thermally thick assumption For samples G1 to G3 where βranged from 0.194

to 0.543, i.e β  1, the cooling modulus was computed using the error function

erfβas given in Equation (4.28) instead of complimentary error function erfcβin Equation (4.27) In other words, the surface temperatures for samples G1 to G3 were calculated using Equation (4.29) Good values of surface temperatures were obtained: 266.5°C, 373.4°C and 374.9°C which were in reasonably good fit with the measured

surface temperatures For samples G4 and G5 where 0.944≤ β ≤1.45, no sensible values of surface temperatures were obtained using either form of cooling modulus This is because these values of βwere either too small to be considered as 1, and too close to unity to be taken as 1, which probably explained why the mathematical treatment did not work Consequently, these calculated data for samples G4 and G5 were treated as outliers

Surface temperatures at ignition were also simulated by considering the pure heat conduction model constructed in Section 4.3.2 using Fluent®6.3 The simulated

surface temperatures were compared with the calculated surface temperatures using revised analytical model according to Equations (4.24) and (4.29) as well as the experimental surface temperatures in Table 6.2 The simulation produced the lowest surface temperatures at ignition at 300°C and the highest at 470°C This simulated temperature range was consistent with results obtained from other heat conduction

thermal models (Lawson and Simms 1952) These surface temperatures at ignition T ig

simulated from pure heat conduction however were larger than the measured values for green wood At high heat fluxes ≥ 20kW/m 2

thermocouple-, these simulated

surface temperatures T ig were, in average, one hundred degree Celsius larger in

magnitude than the measured surface ignition temperatures T ig

Table 6.2 Comparison of surface temperatures for green wood in piloted ignition

Samples Incident

heat flux

e

q′′ (kW/m2)

Time to piloted

ignition t

(s)

Simulated surface temperature

ig

T (°C)

Measured surface temperature

ig

T (°C)

Calculated surface temperature

Figure 6.1 illustrates the discrepancy in surface temperatures obtained from the three different methods, with large variations indicated from the longer error bars shown between the simulated and the calculated values, and relatively closer fit of surface temperatures between those calculated and measured data

Figure 6.1 Comparison of surface temperatures for green wood in piloted ignition

θF (°C) Simulated surface temperature

θF (°C)

The comparison of surface temperatures obtained from the pure conduction comput ational model constructed in Section 4.3.2 using Fluent®6.3 and that of revised

analytical model calculated using Equations (4.24) and (4.29) allowing for

moisture-mediated thermophysical terms suggests that moisture, inter alia, affects the heat

transfer in oven-dry wood Because both the computational and analytical models assumed wood to be inert, the closer fit of calculated values with the measured results strongly point towards the sole interplay of moisture effects on wood ignition, downplaying other factors such as endothermicity and pyrolysis in the scope of analysis The effects of moisture on thermophysical properties produce a significantly better agreement of results between measured and calculated values, suggesting the importance of moisture in the heating of oven-dry green wood

6.1.2 Piloted ignition - surface temperatures in preburn wood

Preburn wood has been preheated in oven resulting in 52% mass loss The partially charred wood has a smaller moisture content of 6.65% as compared to 13.83% of oven-dry green wood, and reduced specific heat capacity and thermal conductivity when compared to green wood, as shown in Chapter 4 Heat transfer in preburn wood could be increased by reduced thermal inertia because of the reduced density (Cuzzillo 1997); and that conductivity strongly correlates with reduced density and moisture content (Janssens 1991b)

When solved using the revised analytical model according to Equations (4.24) and (4.29), samples with β  1yielded feasible surface temperatures using erfcβ For samples with β  1, it was found that their surface temperatures at ignition were

successfully solved by using erfβ Both calculated and measured surface temperatures were tabulated in Table 6.3

Table 6.3 Calculated surface temperature for preburn wood in piloted ignition

Samples Incident

heat flux

e

q′′ (kW/m2)

Time to piloted ignition t

(s)

Cooling modulus β

=( / ) * (h k αt)1/ 2(- )

Calculated surface temperature

ig

T (°C)

Measured surface temperature

For preburn wood, the surface temperatures at ignition T ig obtained from simulations

using pure heat conduction model and that calculated from analytical heat balance were very similar; the computed surface temperatures from both methods were in relatively good agreement with the experimental surface temperatures The minimised discrepancy among the three types of surface temperatures could be seen from the reduced high-low bars indicated on Figure 6.2 Table 6.4 listed all the surface

temperatures obtained from the three different methods for preburn wood

When compared to green wood, the preburn wood samples showed a good agreement between the calculated and simulated temperatures, and these two types of

“theoretical” temperatures also agreed well with measured temperatures The good agreement between the calculated temperatures with the measured temperatures suggests that piloted ignition problem for preburn wood is more strongly a thermal case, as compared to that of the oven-dry green wood For partially charred wood, the reduced pyrolysable content, indicated by lower heat of release rates for fire modelling (Moghtaderi and Kennedy 1998, Chow and Han 2006), validates the underlying assumption of negligible chemical effects that is adopted in the formulation of a thermal model In preburn wood, ignitability at a given heat flux is related to the thermal properties of the material, in particular the thermal inertia, kρc

(Cuzzillo and Pagni 1999) The reduced kρc as a result of pre-burning produced higher surface ignition temperatures in preburn wood as compared to that of green wood The same deduction is purported by Cuzzillo and Pagni (1999)

Table 6.4 Comparison of surface temperatures for preburn wood in piloted ignition

Samples Incident

heat flux

e

q′′ (kW/m2)

Time to piloted

ignition t

(s)

Simulated surface temperature

ig

T (°C)

Measured surface temperature

ig

T (°C)

Calculated surface temperature

θF (°C) Simulated surface temperature

θF (°C)

On the other hand, the good agreement between the simulated and the calculated surface temperatures in preburn wood re-instated the role of moisture in heat transfer, where the reduced moisture content in preburn wood simply minimised the impact of moisture on the thermophysical terms and hence the heat balance on preburn samples

6.1.3 Spontaneous ignition - surface temperatures in green wood

The spontaneous ignition of wood can also be calculated using the revised analytical model assuming a critical surface temperature criterion Spontaneous ignition is defined as the initiation of flaming combustion without a pilot (Janssens 1991a) Only sustained ignition i.e the initiation of flaming combustion that persists after the external heat source is removed, was considered in this analysis

Experimentally, spontaneous ignition to direct flaming was noted only at high incident heat fluxes of 50 kW/m2 and 40 kW/m2 Ignition at lower heat fluxes

occurred via glowing ignition is not considered in this discussion For green wood, two sustained flaming combustion were noted and the surface temperatures were

found to be 448°C and 457°C respectively These surface temperatures at ignition T ig

were much higher than that for green wood in piloted mode The results suggested that the presence of moisture increased the surface temperature of ignition for green wood For piloted ignition, ignition occurred more readily as pilot provided the energy

to ignite the combustible-gas mixture For spontaneous ignition, flaming occurred entirely upon the attainment of sufficient surface temperature (Martin 1964) It was interesting to see the interplay of moisture on ignition of green wood by comparing the two different ignition modes

For spontaneous ignition in green wood, given that the values of β  1, the surface

temperatures were therefore calculated using erfβ The calculation yielded surface temperatures of 558.2°C to 786.2°C, which were higher as compared to the surface temperatures of 448°C and 457°C measured at 50kWm-2 and 40kWm-2 respectively,

as shown in Table 6.5 The quantum of difference could be more clearly illustrated in Figure 6.3

Table 6.5 Calculated surface temperature for green wood in spontaneous ignition

Samples Incident

heat flux

e

q′′ (kW/m2)

Time to ignition

t (s)

Cooling Modulus β

=( / ) * (h k αt)1/ 2(- )

Calculated surface temperature

ig

T (°C)

Measured surface temperature

Figure 6.3 Comparison of surface temperatures for green wood in spontaneous ignition

6.1.4 Spontaneous ignition - surface temperatures in preburn wood

Three sustained spontaneous ignitions in direct flaming mode were noted for preburn wood Preburn wood, owing to its reduced moisture content, resulted in lower

surface temperatures of ignition T ig as compared to green wood The measured surface

temperatures for spontaneous ignition in preburn wood were 426.1°C at 50kWm-2 and

416.8°C at 40kWm-2 (Table 6.6), as compared to the measured surface temperatures

of 448°C at 50kWm-2 and 457°C at 40kWm-2 (Table 6.5) for spontaneous ignition of

green wood

For spontaneous ignition of preburn wood, it was found that β  1 Therefore, the

surface temperatures were calculated using erfβ The calculated surface

temperatures at ignition T ig were in good fit with the measured surface temperature, as

shown in Table 6.6 The close agreement of surface temperature values was better illustrated by the reduced high-low bars for each pair of values in Figure 6.4

Table 6.6 Calculated surface temperature for preburn wood in spontaneous ignition

Samples Incident

heat flux

e

q′′ (kW/m2)

Time to ignition

t (s)

Cooling Modulus β

=( / ) * (h k αt)1/ 2(- )

Calculated surface temperature

ig

T (°C)

Measured surface temperature

was judged using critical surface temperature criterion (Williams 1953, Martin 1964, Janssens 1991a), and more so when spontaneous ignition is considered In the case of spontaneous ignition of green wood, the results suggested a more comprehensive model entailing not only heat conduction, but also internal heat and mass transfer must also be considered for solid phase combustion phenomena

6.1.5 Critical heat flux

Critical heat flux (q′′cr)is an estimate of minimum heat flux (q′′min)obtained through correlation of experimental data (Janssens 1991b, Spearpoint and Quintiere 2001) Minimum heat flux is the heat flux level below which ignition under practical conditions, whether a bench-scale test or real fire can not occur (Janssens 1991a); it is therefore the quantity of interest McGuire (1965) suggested that

2 min 12.5kWm

q′′ = − for most wood materials and hitherto this value becomes the design value The minimum heat flux was customarily obtained in Cone Calorimeter where the time allocated for observation of ignition is 10-20 minutes However, Babrauskas (2001b) pointed out that lower values than 12.5kWm-2 could be found for wood This

study seeks to derive a more accurate estimate of minimum heat flux through critical heat flux value via extended correlation of longer ignition time data

6.1.5.1 Critical heat flux for piloted ignition of green wood

Critical heat flux q ′′ was derived using Equation (4.53) as shown in Chapter 4 by crplotting incident heat fluxes q ′′ /e √tig against q′′ Table 6.7 tabulates the times to eignition measured against incident heat fluxes Figure 6.5 shows the linear regression

Figure 6.5 Determination of critical heat flux for piloted ignition of green wood

The q ′′ obtained through the correlation of ignition data with incident heat fluxes is cr

14 kW/m2, which is greater than the test measurement value of 11kW/m2, the latter of

which was obtained after two hours exposure in Cone Calorimeter

However, when the same set of ignition data in Table 6.7 is examined by plotting

1 / t ig vs q′′ , a different value of e q′′ is obtained Figure 6.6 shows the linear crregression of 1 / t vs ig q′′ e

cr

Figure 6.6 Determination of critical heat flux for piloted ignition of green wood

of 11kW/m2 On the other hand, q′′ of 14kW/mcr 2

obtained through the plot of

q′′ The critical heat flux of 10.4kW/m2

found using 1 / t vs ig q′′ is substantially elower than the code-stipulated value of 12.5kW/m2 This correlation uses a longer

ignition time than the stipulated 20 minutes cut-off; however the correlation does not

cr

q =10.4kWm-2

consider the ignition time taken at 10kW/m2 since only one out of three samples tested

actually ignited, and thus the ignition data obtained at this irradiance level is no way confirmatory

The attainment of this critical heat flux at 10.4 kW/m2 through long heating exposure

in Cone Calorimeter is a challenging experimental endeavour This is because extrapolation through lower heat fluxes in search of the critical heat flux is confronted

by the issues of changing thermal thickness as well as the onset of a different ignition

mechanism (Spearpoint and Quintiere 2001) The t ig values obtained at low heat

fluxes are large, but more notably are their significant deviation from the linear

regression graph The main reason why the t ig values obtained at low heat fluxes

deviate is due to the fact that when t ig becomes substantially large, the physical

thickness of the slab Lis less than the thermal conduction length α In other words, t

a thermally thick material no longer behaves as a semi-infinite solid Drysdale (1985) has pointed out that the consequence of this changing thermal thickness on the data set is often swamped by the data scatter unless the data set includes ignition times

significantly greater than c.5min The inclusion of data set with long ignition times in

this study i.e up to 2 hours provides a case in point to illustrate this phenomenon Samples, when heated over prolonged exposure, no longer act as semi-infinite solid; the assumption of thick fuel thus breaks down and so the correlation between heat flux and time to ignition changes As a result, non-linearity was observed in the correlation Spearpoint and Quintiere (2001) however offer a different view on the non-linearity of the correlation They argue that for wood, char oxidation introduces another mechanism for ignition The deviation of data at low heat fluxes suggests a different trend dominating at these low heat flux regimes This is because when wood

is heated at low heat fluxes, wood ignites by way of a glowing ignition prior to flaming at low heat fluxes (Babrauskas 2001b) Consequently, it was suggested to correlate only the high flux data i.e more than 20kW/m2 (Spearpoint and Quintiere

2001) But the correlation using only the high heat flux data produces critical heat fluxes that are approximately twice as large the critical heat flux that would be obtained by the correlation that encompasses the lower heat flux ranges, as discussed

in Spearpoint and Quinteire’s work (2001) The critical heat flux obtained by considering only the high heat flux data would not be consistent with the measured critical heat flux in this study anyway

Despite deviation from linearity of these large t ig data, the correlation of ignition data

at such longer time scale could seek an explanation from the thermal point: based on the assumptions that linearised heat losses for an inert semi-infinite solid, the surface temperature at ignition in Equation (4.53) in Chapter 4 can be expressed as

Quintiere and Harkleroad (1984) suggested that for small t ig, based on solutions for

linearised heat losses from the surface, F(t ig ) for a semi-infinite solid can be

approximated as

2( )ig h ig t ig

Quintiere and Harkleroad (1984) proposition that for large t ig values, qcr′′ =qe′′ The good agreement between derived critical heat flux and measured minimum heat flux provides a strong ground for advancing the use of a lower critical heat flux of 11kW/m2 in lieu of the presently code-stipulated 12.5kW/m2

6.1.5.2 Critical heat flux for piloted ignition of preburn wood

The correlation of piloted ignition data (Table 6.8) in preburn wood, obtained by plotting incident heat fluxes q ′′ /e √t ig against q′′ produces a linear regression as shown e

in Figure 6.7; it yields a critical heat flux q ′′ of 14 kW/mcr 2

for preburn wood When

the ignition data is plotted using 1 / t ig vs q′′ , a critical heat flux of e

Table 6.8: Ignition data of preburn wood

Figure 6.7 Determination of critical heat flux for piloted ignition for preburn

6.1.5.3 Concluding remarks for moisture effects in thermal combustion using

thermal model

The moisture effects in thermal combustion were examined by comparing the surface temperatures predicted by pure thermal model and that computed by revised analytical model with measured thermocouple surface temperatures The findings shed lights on some important aspects pertaining to the research question on the moisture effects in thermal combustion:

 Thermal model does not accurately predict surface temperatures at ignition Moisture effects must be included through the thermophysical terms for thermal model for ignition based on critical temperature criterion

 Moisture exerts an impact on thermal ignition via heat of wetting, latent heat of vaporisation, specific heat and thermal conductivity Equilibrium moisture content not greater than 14% in green wood is able to moderate the surface temperatures at ignition to lower values

 Thermal model without moisture-mediated terms results in unrealistically higher surface temperatures at ignition

 The Cone Calorimeter experiments and revised model calculations show that it

is possible to neglect the effect of moisture migration on the temperature rise

 Cooling modulus with magnitude greater than unity can be computed by

complimentary error function erfcβ

 Cooling modulus with magnitude smaller than unity are to be computed by error

function erfβ

For derivation of critical heat flux using thermal model, the study successfully achieved the following:

 A lower critical heat flux for green wood was established at 11.31kWm-2

, and the correlation was successfully performed on an hourly time scale This critical heat flux is lower than the stipulated design value of 12.5kWm-2 for the

use of wood in construction

 Preburn wood yields a lower critical heat flux of 9.39kWm-2

than green wood

of 11.31kWm-2

 Reduced pyrolysable content in preburn wood better approximates preburn wood as a thermal case

 The surface ignition temperature T ig in preburn wood was strongly correlated

with the reduced thermal inertia kρc, yielding consistently higher surface

ignition temperatures in preburn wood than green wood

A secondary trend still dominated at the lower heat fluxes for both green and preburn wood using thermal correlation The results suggested that more tests need to be conducted at the low heat flux to validate the use of the thermal model and to elucidate the secondary mechanism responsible for the deviation of linearity in correlation

6.2 Heat and mass transfer in porous slab

Self-heating and spontaneous combustion of wood was studied using wood cube of critical size exposed to low-temperature, long-term heating in an isothermal oven as described in Chapter 4 The temperature field and progression of self-heating was tracked by thermocouples The modified Darcy’s law porous model developed in Chapter 3, which was formulated with surface evaporation and that of an internal evaporation, was used to simulate the moisture movement and temperature field development at both the initial and late drying phase

The main objective of this section was to examine the effect of moisture movement on the temperature distribution in green wood, since the foregoing section showed that moisture effects on thermophysical terms alone have been insufficient to predict ignition temperatures, especially in the case of spontaneous ignition relying solely on critical surface temperature criterion The effects of moisture movement of free water and later combined moisture (water and water vapour) on the temperature distribution profile in green wood were analysed With the modified Darcy’s law Model, the study aims to seek an insight into the initial drying phenomenon involving moisture migration of free water which is important in the case of slow heating of wood, unlike the high-temperature drying model which normally ignored the free water as it vaporises too quickly to allow any tangible impact Spontaneous combustion of wood was then considered within the framework of moisture migration and thermophysical properties of wood, of which experiments on wood cube heating were analysed for its thermal runaway behaviour An analysis of the progression of temperature profiles

provided a validation to the proposed framework to examine self-heating of wood cube

6.2.1 The effect of elimination of evaporation term

The heat and mass transfer model for porous slab developed in this study treated evaporation as the boundary condition (See Chapter 3), since evaporation at initial drying stage, particularly for low temperature heating occurred mainly at the surface (Farid 2002, Zhang 2003) Temperature profile was generated from this modified Darcy’s law Model, and were compared to the other two model’s predictions from Zhang and Datta (2004), where one was based on conventional drying model that considered evaporation as an internal term (Model 1), and the other simply as a diffusion model treating evaporation as a boundary condition (Model 2), the latter being just as the same as the modified model proposed in this study The comparison illustrated the effect of the different treatments of evaporation in a drying model of wood The temperature profiles were shown in Figure 6.9

The different numerical approach and treatment of evaporation generated very different temperature profiles in Figure 6.9 Model 1 has a depressed curve while the temperature profiles of Model 2 and modified Darcy’s law Model were rather linear The linear temperature profiles of Model 2 and modified Darcy’s law Model showed that the internal temperature distribution within wood volume was largely uniform

Figure 6.9: Comparison of temperature profiles during the initial heating period

(Data for Model 1 and 2 were taken from Zhang and Datta (2004), © Drying Technology, 2004)

Table 6.9: Parameters used in drying of wet wood for the initial period

Model

Liquid water diffusivity

Vapour density in heating

The uniform temperature profiles within these wood volumes were the consequence

of which firstly, evaporation took place at the surface and therefore only consumed

heat from the ambient hot air; secondly the thermal conductivity was relatively large due to the considerable presence of liquid water at the low temperature drying stage However, for Model 1 which considered evaporation as an internal term, heat was collected from inside the material and acted as a heat sink The system generated a temperature gradient to draw heat flux from the surface to the inside of the material The inclusion of evaporation as an internal term created a heat sink in the system, resulting in the depressed temperature curve

The departure of the predicted temperature profile in Model 1 from that in Model 2 and modified Darcy’s law Model was caused by different ways of representing the evaporation term At the initial drying of wood, evaporation occurred mainly at the surface, and temperature distribution within wood would remain rather uniform, because of the presence of relatively considerable amount of moisture at the initial stage (Eriksson, Johansson and Danvind 2007) The comparison showed that the representation of evaporation at the boundary condition in the system would generate

a more realistic temperature distribution, consistent with the phenomenological model

of wood drying at low temperature (Fang and Ward 1999, Zhang 2003)

6.2.2 Velocity and temperature distribution of free water in Initial Drying

Phase (IDP) Model

To examine the velocity and temperature contours in the initial drying phase, the Initial Drying Phase (IDP) Model developed in Chapter 3, Section 3.1.1 produced contours at three locations measured from the mid-plane (z =0mm) of the wood cube (see Figure 4.3) The contours showed the results of the convective flows and temperature field associated with the longitudinal flow in the domain The contours at

the x plane representing the transverse flows were not discussed since the flows were

negligible in the transverse directions

For free water movement, because the IDP model considers evaporation at the boundary, instead of an internal term, the velocity contours of free water did not exhibit large variation (see Figures 6.10, 6.11 and 6.12) The velocity contour near the surface (Figure 6.12) however achieved more uniformity than any other inner location, which implied that evaporation occurred at the surface and vapour convected without resistance at the boundary in the computational domain These results would only be possible if the liquid water content in those domains did not vaporise too quickly The free water movement simulated using IDP model agreed well with the phenomenological understanding of the initial drying stage of wood (Fang and Ward

1999, Zhang 2003), thus endorsing the validity of evaporation term used in the IDP model

A direct implication was noted between velocity contours and the temperature profiles

in the IDP model The more consistent the velocity contours, the more uniform the temperature field became The results showed that the convective flows of free water within the domain was rather stable, enabling the wood cube to possess rather uniform temperature fields The temperature contours were shown besides the corresponding velocity contours in Figures 6.10, 6.11 and 6.12

The simulations could be justified on physical basis Velocity contours of free water movement were the direct consequence of the treatment of the evaporation term; the corresponding temperature profile however must be justifiable on phenomenological

basis The internal temperatures simulated showed least variation; the simulations thus complied with the presence of still large amount of water content at the initial drying stage The results thus confirmed the validity of the IDP model

Figure 6.10: Velocity and temperature contours on the mid-plane (z = 0mm) for longitudinal flow using IDP Model

Figure 6.11: Velocity and temperature contours on z = 20mm from the mid-plane for longitudinal flow using IDP Model

Figure 6.12: Velocity and temperature contours near wall from the mid-plane for longitudinal flow using IDP Model

Temperature contour

Temperature contour Velocity contour

Velocity contour

6.2.3 Velocity and temperature contours of moisture movement in Extended

Drying Phase (EDP) Model

To examine the velocity and temperature contours of moisture movement in the extended drying phase, the Extended Drying Phase (EDP) Model developed in Chapter 3, Section 3.1.2 was used Combined moisture (water and water vapour) has more pronounced convective flow pattern as compared to free water convection (Figures 6.13

to 6.15) The flow patterns suggested that evaporation now occurred within the volume of wood instead of evaporating through the boundary surfaces as in the case of early stage heating At this late drying stage, water flux inside the wood decreased The evaporation front recessed from the surface and retreated inwards (Hukka 1996) Evaporation was now taking place throughout the material and occurring as in the same amount as the mass change rate of the total moisture (Eriksson, Johansson and Danvind, 2007) The flows within the domain should become much more erratic as compared to the movement

of free water

The temperature field also became highly non-uniform as a consequence of higher speed

of water vapour convection within the domain Unlike the initial drying stage where the temperature within the domain was rather uniform, the temperature field arising from combined moisture convection in the EDP model was more varied in tandem with the stronger convective flow The EDP model showed that the surface zone has a higher temperature than that of IDP model Diffusivity of liquid became small due to diminishing moisture content (Plumb, Spolek and Olmstead, 1985) Less water flux diffused through the surface and the surface became drier and therefore hotter The EDP

model therefore suggested an advancing temperature front towards the center This result provided two crucial explanations on spontaneous ignition of wood experiment: firstly, when a porous mass of oxidizable material was considered for exothermic reaction, local variations in water content throughout the volume increased its propensity to spontaneous ignition by self-heating mechanisms that would not otherwise operate if it were at uniform water content The retreating evaporation front also accounted for a characteristic S-curve temperature profile observable in the progression of self-heating in wood slab that would be discussed in the next section

Figure 6.13: Velocity and temperature contours on the mid-plane (z = 0mm) for longitudinal flow using EDP Model

Figure 6.14: Velocity and temperature contours on z = 20mm from mid-plane for longitudinal flow using EDP Model

Figure 6.15: Velocity and temperature contours near wall from mid-plane for longitudinal flow using EDP Model