Radiative ignition properties of food grains and surface temperatures at ignition

Abstract

Fire safety of food grains in storage, transport and processing facilities requires knowledge of their ignition properties some of which have been determined in this study. Cowpea, lentils, millet, soybean, unshelled peanut, flax (linseed), sunflower, shelled peanut and sesame, at 10% moisture content were studied. A cone calorimeter heater was used to impose radiative heat fluxes of 25, 35 and 47.5 kW/m² on the food grains to determine the time to piloted ignition concurrently with the surface temperature at ignition (T_s,ig). Ignition temperatures (T_ig) were calculated from experimentally determined critical heat fluxes. The thermal response parameter and thermal inertia ( $k ρ c$ ) essential for the characterization of the ignition of the food grains were also calculated. Thermal response parameter appears to be best suited for distinguishing the ignitability of the food grains. Ignition temperatures range between 239°C and 305°C, while thermal response parameter varies between 261 and 565 kWs^0.5/m². Small particle-sized grains showed higher ignitability.

Keywords

Food grains ignitability ignition temperature thermal inertia thermal response parameter

Introduction

There is a long history of disastrous fires around the world that involved large quantities of food grains (FGs) in storage, transport and processing (ST&P) facilities. These fires often result in huge financial losses and, sometimes, human lives. Beyond the consequent environmental pollution, there is a potential disruption of business, income, or even a complete loss of it. A number of such facilities with losses and damages upwards of US$1 billion are included in the study by Bhusari et al.¹ As already highlighted in the United Nations Food and Agriculture Organizations 2021 report,² the growing world population is facing increasing food shortages. Fires in FGs ST&P facilities are capable of worsening the situation as a consequence. In Africa, where more than 55% of livelihoods depend on agriculture, with grains comprising a huge proportion of it,³ fire safety in such facilities is critical.

Fire development starts with ignition as its first stage (although some models consider pre-heating before ignition). Ignition of agricultural food grains can be due to self-heating or exposure to external heat. Spontaneous ignition from self-heating has been extensively curtailed as regulations and guidelines based on research are now in use. Hence, interest is shifting to ignition from external sources such as hot objects, electrical faults, glowing nests and processes like drying.⁴ Recently, grain dryer fire on 26 February 2023, destroyed 50 tonnes of soybean in Pérola Independente, Brazil.⁵ On 13 June 2023, another grain dryer fire in Brazil’s Fazenda Dona Aurora, resulted in the loss of 80 tonnes of corn.⁶ It is also estimated that there are 100 grain dryer fires per annum in Germany,⁷ and yet literature is quite sparse on food grain fire research. A few studies that exist include the effect of mould on the smouldering of rice grains by Wang et al.⁸ The effect of particle size on the smouldering of corn was investigated by Rosa et al.,⁹ and the combustion of corn with surface temperature measurements by Cai et al.¹⁰ Fire parameters and comparative thermal hazards of food grains were recently investigated by Oguaka et al.¹¹ These studies have not addressed or compared the ignitability of different types of food grains, which is the focus of this work.

Ignition starts with the temperature rise of a material exposed to heat resulting in physical and chemical changes,¹² with time and temperature as key parameters. Usually, ignition is assumed to occur when a certain criterion or threshold of the material heating process is attained. Such criteria include mass loss rate, time to ignition (TTI), surface temperature, critical char depth, heat release rate, critical heat flux^12–14 and thermal response parameter (TRP) introduced by Tewarson and Ogden.¹⁵ The determination and use of time and temperature as criteria seem appropriate and can be used to predict when, and if, ignition can occur in a given circumstance. However, the ignition temperature, T_ig, is not necessarily a definitive criterion of fire initiation, nor a fixed value, due to multiple factors that play a role in the ignition.^16,17 Interestingly, among four ignition criteria ranked and contrasted in the study by Vermesi et al.,¹⁸ the T_ig was ranked the second-best predictor of ignition after the critical mass loss rate. TRP is a composite parameter that combines thermal inertia, ( $k ρ c$ ) representing the time scale,¹⁴ and ignition temperature, T_ig and is, therefore, considered to be a good criterion. These three criteria –T_ig, TRP and $k ρ c$ – are determined in this work.

Each ignition criterion has its disadvantages or difficulties in its determination. For instance, mass loss rate measurement is both difficult and to some extent, apparatus-dependent.^14,18 The measurement of heat release rate shares this challenge as it is a product of the mass loss rate and heat of combustion. A similar problem is also faced with surface temperature measurement, in addition to its usual variation with incident heat flux.^19,20 In addition, the ignition temperatures of most solids are within the same order of magnitude, 300°C–500°C.¹⁴ However, thermal inertia, which can be calculated from ignition temperature can be different by two orders of magnitude¹⁴ and so it is a more distinguishable ignition criterion. Assuming that ignition temperatures are the same, higher values of thermal inertia imply a longer time required for materials to increase in temperature and ignite.²¹

The purpose of this study is to determine, for a range of FGs at 10% moisture content (MC), the ignition criteria namely (1) surface temperature at ignition, T_s,ig, by experimental measurement, and ignition temperature, T_ig, from calculations based on critical heat fluxes (CHF), (2) thermal response parameter, TRP and (3) thermal inertia, $k ρ c$ , both derived from experimental data. In addition, these criteria are also related to the physical characteristics of the FGs to provide an indication of ignitability for other food grains. In this work, as in similar studies,^19,22 the T_ig is the temperature at the sample surface at the point of ignition determined by calculation from a heat transfer model of ignition. T_s,ig is the temperature at the surface of the sample obtained by direct measurement. Experimental measurements are taken under a radiant cone heater to determine TTI and then apply formulae based on the thermal ignition theory to derive other properties. The properties determined will be beneficial in understanding fire initiation and its propagation as well as input in fire modelling studies. Ultimately, the knowledge will help to provide engineered fire safety for ST&P facilities for FGs.

Thermal ignition theory for data analysis

Thermal ignition theory is extensively addressed in the literature^23,24 and therefore the theoretical development and derivation of the equations will not be repeated here. Only the final correlations used to aid the analysis and interpretation of experimental results are presented in this section. When solids are suddenly exposed to heat, changes that lead to ignition start with heating and a consequent increase in temperature (thermal changes), thermal degradation and decomposition (chemical changes). The ignition of solids can be studied from thermal or chemical perspectives.¹² In the thermal model, which applies just before ignition, the material is considered inert; any degradation or decomposition is ignored. Furthermore, heat generation from chemical energy conversion due to oxidation has yet to occur.

The empirical relationship, typically in the form of equation (1), was first proposed by Buschman²⁵ as suitable for correlating experimental ignition data –TTI and ${\overset{\cdot}{q}}_{e} ″$ . On this account, n has no physical meaning attached to it even though Abu-Zaid²⁶ later theoretically justified n = 0.5. A graph of TTI and the ${\overset{\cdot}{q}}_{e} ″$ based on the power law in the form of equation (1)^14,25 assists in extracting other ignition properties such as $T_{ig}, h_{ig}, k ρ c,$ and TRP

T T I^{- n} = C ({\dot{q}}_{e}^{″} - C H F)

(1)

A common practical correlation for thermally thick materials is given by equation (2).^27,28 Thermally thick materials are resistant to heat transmission and therefore have significant temperature gradients across their thickness. This thickness in a given heating condition of the material is referred to as the thermal penetration depth, $δ_{T}$

TT I^{- 0.5} = \frac{1}{TRP} ({\overset{\cdot}{q}}_{e}^{″} - CHF)

(2)

While equation (2) can be used to predict the TTI, it is a straight line, with TRP given by equation (3) being the reciprocal of the slope

TRP = {(\frac{π k ρ c}{4})}^{0.5} (T_{ig} - T_{o})

(3)

And the thermal inertia can be related according to

k ρ c = \frac{4}{π} {[\frac{TRP}{(T_{ig} - T_{o})}]}^{2}

(4)

The energy balance from the thermal ignition theory also yields equation (5) for the determination of T_ig and holds at the moment of ignition with $T_{s} = T_{ig}$ . At the critical ignition point, the total heat loss from the surface at ignition is equal to the $CHF$

CHF = h_{c} (T_{ig} - T_{o}) + ε σ (T_{ig}^{4} - T_{o}^{4})

(5)

Also, equation (5) can be linearized by introducing the total heat transfer coefficient $h_{ig}$ as in equation (6)

CHF = h_{ig} (T_{ig} - T_{o})

(6)

The slope and intercept from the plot $TT I^{- 0.5} versus {\overset{\cdot}{q}}_{e} ″$ , together with equations (5) and (6) are used to calculate additional ignition characteristics such as $T_{ig}, h_{ig}$ and $k ρ c$ .

Materials and methods

Sample selection and preparation

In this study, FGs were selected to, as much as possible, represent varieties of commonly found FGs in Africa. The range of samples selected considers oil content ranging from 1.8% to 49.7% by mass, bulk density and bulk porosity from 327.4 to 817.0 kg/m³, and 0.38 to 0.51, respectively.¹¹ The grains as shown in Figure 1 were (1) cowpea (black-eye beans), (2) lentils, (3) millet, (4) soybean, (5) flax (linseed), (6) unshelled peanut, (7) sunflower, (8) shelled peanut, and (9) sesame. Unshelled peanuts are those of which the shells have not been removed, while shelled are those which have the beige fibrous outer coverings removed. All grains were the species grown within the geographical region of South Africa and bought from local shops in Cape Town.

Figure 1.

Photos to identify the food grains tested in this study. Further details are in the elemental analysis contained in the study by Oguaka et al.²⁹ (a) Cowpea, (b) lentils, (c) millet, (d) soybean, (e) flax, (f) unshelled peanut, (g) sunflower, (h) shelled peanut and (i) sesame.

The MC was adjusted to 10% by weight on oven dry basis. This choice of test MC considered the review of FG storage conditions by Ziegler et al.³⁰ It indicated that most grains are stored at between 10% and 14% MC depending on the desired storage duration and ambient conditions. Also, the as-received MC of some grains used in this study was as low as 4.0%. The MC adjustments were achieved by first determining the as-received MC, and calculating the mass of moisture to be removed or added. The as-received MC was determined by weighing a quantity of the grains received, drying it in a convective oven for 24 h at 104°C and then weighing the dried mass.³¹ The MC, in percentage, is the quotient of the lost (moisture) mass divided by the remaining (dried) mass of the grains.³² Distilled water, which is devoid of any minerals or impurities, was used to make up for the MC. The required amount of water was sprayed onto the grains and completely mixed. The sample was then placed in a heavy-duty Ziploc bag within an outer Ziploc bag. This was kept in the refrigerator at 5°C for 7 days for the grains to absorb the moisture without any deterioration. This is a standard method for biomass or FGs MC adjustment as used in various studies (Bitra et al.³³ and Deshpande et al.³⁴).

The as-received conditions of cowpea, lentils, and millet were 12.9%, 11.4% and 14.7%, respectively. The mass of oven-dried grains, which when mixed with a quantity of as-received grains would result in the required 10% MC was determined. The relative masses of the as-received and oven-dried samples were then mixed and placed in double Ziploc bags. They were kept at ambient temperature for 7 days. On the day of use, the grains stored in the refrigerator were transferred to warm up to the laboratory ambient condition for a minimum of 3 h, while still in the bags. Note also that there was no moisture droplet or condensation noticed when removed from the refrigerator or when warmed up to the ambient temperature.

Experimental apparatus and procedure

The experiments were carried out using a cone heater as shown in Figure 2, similar to the standard cone calorimeter arrangement. The set-up was surrounded on the sides with a mesh screen to shield against any draught of wind. This was done to ensure that the sample surface and surroundings were undisturbed. The hood over the entire ignition apparatus was connected to a variable-speed exhaust fan. This was set so that the smoke was gently extracted and practically avoided any disturbance over the sample surface. Therefore, the flow of pyrolysate and smoke arising from the sample surface was buoyant and considerably free from wind draught interference.

Figure 2.

Schematic diagram of the experimental set-up.

In Figure 3, a close-up view of the portable data logger recorded by the video camera, and the typical placements of the thermocouple are shown. Note that a difficulty in measuring surface temperatures is achieving and maintaining perfect thermocouple tip to sample surface contact. To maintain the thermocouple’s tip steady and in contact with the grain surface, a little weight of 11.7 g was suspended and slid on its arm. In addition, visual observation was maintained throughout the experiment. When the placement of the thermocouple was not achieved satisfactorily on the surface, the test was aborted and the test data discarded.

Figure 3.

(a) Real-time surface temperature display on portable data logger recorded by video camera, (b) and (c) are typical placement of the thermocouple on test grain surfaces ensuring contact with the grain, (d) 11.7 g weight suspended to slide on thermocouple arm to maintain tip contact with grain surface.

Although not done in this work, an alternative or additional temperature measuring device is infrared camera.³⁵ It has the advantage of ease of use requiring no contact with the sample and can, therefore, be fixed steady at a location. The disadvantages with this include difficulties with accurate calibration of the infrared camera for the correct emissivity of the sample especially at ignition, camera view obstruction by apparatus, interference of radiation by emitted gas or flame, and possible reflection of incident heat from the sample surface.

Heat flux calibration with a Hukseflux SBG01 water-cooled heat flux gauge was carried out daily or when changing to a different heat flux. A minimum of 30 min was first allowed for the apparatus to warm up to a steady state. A water-cooled heat flux gauge was then used to read and adjust to the required test heat flux. Surface temperature data were measured using a K-type thermocouple with a probe overall diameter of 1.3 mm, although a much smaller diameter thermocouple should be more appropriate to limit heat conduction due to its thermal mass.³⁶ However, this would be a challenge to fix and be retained on the surface of the grains for accurate measurements unlike on wood³⁷ or plastic.³⁸ Temperature data were recorded every 200 ms. The thermocouple was also connected in parallel to a portable data logger with a display recorded by a video camera alongside the sample surface to observe ignition time. The TTI and the surface temperature at ignition were extracted from the video camera records. This provided an accurate match of ignition, ignition time, and surface temperature at the instant of ignition. The data logger records provided the form of the time–temperature history of the ignition process.

The grains were measured in a box of 100 mm $\times$ 100 mm $\times$ 17 mm to maintain consistency in the quantity and dimensions in the sample holder. However, the larger grains of shelled peanut and unshelled peanut were made deeper to about 20 and 25 mm, respectively. This was to ensure adequate coverage of the backing aluminium foil and fibre blanket as in the International Organization for Standardization (ISO) 5660³⁹ standard cone calorimeter sample holder. The grains were loosely poured into the sample holder and gently levelled at the top for each test. Four tests were carried out for each grain under each radiative heat flux of 25, 35 and 47.5 kW/m². Therefore, for the nine FGs, 108 tests in total were conducted.

Results and discussions

Important visual observations

The ignition of unshelled peanuts was unique with its distinct ‘2-stage’ ignition, as illustrated in Figure 4(a) to (i). The outer shells first glowed (see Figure 4(b)) and ignited (see Figure 4(c)), flamed (see Figure 4 (d)) and burned off within a short while, less than 8 s, and then extinguished (see Figure 4(e)) and no noticeable pyrolysis was then observed (see Figure 4(f)). Noticeable generation of pyrolysate and smoke restarted (see Figure 4(g)) and continued until a second ignition occurred (see Figure 4(h)), which then remained (see Figure 4(i)). In addition, the unshelled peanut had a glowing ignition that transitioned to established flaming. This behaviour is known to create some uncertainty in determining ignition temperature.¹⁴ It appears this behaviour was due to surface oxidation, which continued until the establishment of flaming ignition.²¹ In between, the surface temperature, T_s, rose as heterogeneous combustion progressed and subsequent flaming ignition at an isolated location or local spot, finally spreading over the rest of the sample surface. This behaviour was pointed out by Babrauskas¹⁴ as characteristic of charring materials under low heat fluxes. Heterogeneous combustion refers to the oxidation reaction occurring on the surface of the solid, solid–gas interface, in this case the grains and the oxygen.

Figure 4.

Photos of unshelled peanuts with their typical ignition sequence under 25 kW/m². They show the appearance of local glowing, ignition of the shells, flaming, and extinguishment followed by spot ignition of kernels and surface-wide flaming ignition with variations of surface temperatures, T_s. (a) Exposure at 0 s, T_s at 20°C; (b) glowing at a local spot at 27 s, T_s at 250°C; (c) ignition of shells at 30 s, T_s at 268°C; (d) flame at 34 s, T_s at 455°C; (e) flame extinguished but continued glowing and surface oxidation at 38 s, T_s at 525°C; (f) no noticeable pyrolysate or smoke at 42 s, T_s at 515°C; (g) the onset of another pyrolysis at 57 s, T_s at 459°C; (h) onset of the second-stage ignition with isolated local spot flaming at 94 s, T_s at 434°C; and (i) sustained surface-wide flaming ignition at 112 s, T_s at 579°C.

Time–surface temperature profile

Exemplary typical time and surface temperature profiles of the ignition tests for all the nine grains under the three different heat fluxes are shown in Figure 5. The profiles generally have a similar shape – a nonlinear temperature rise, which might include the period of intermittent pre-ignition flashes⁴⁰ followed by a sudden jump in temperature, which is the surface temperature at ignition, $T_{s, ig}$ . The absence of any plateau in the profiles indicates that there was no melting and that once the decomposition temperature was reached, ignition occurred.

Figure 5.

Exemplary typical time–surface temperature history of ignition of food grains under various heat fluxes. X is the surface temperature at the point of flaming ignition.

However, unshelled peanuts (see Figure 5(f)) have a unique profile. As already elaborated in the section on the important visual observations, the first sudden increase in temperature was the flaming ignition of the shells that burnt off in a short while and then extinguished. The second rapid increase involved the ignition of the kernels. As the heat flux increased, there was a shift to shorter ignition times. However, the shift in $T_{s, ig}$ is not quite noticeable from the graphs, except for unshelled peanut.

Time to ignition

The TTI of the various samples, after exposure to a constant ${\overset{\cdot}{q}}_{e} ″$ are shown in Figure 6. These ranged from 25 ± 2 s (both millet and sesame) to 56 ± 2 s (unshelled peanut) under 47.5 kW/m² heat flux. While under 25 kW/m², the range was 74 ± 8 to 209 ± 10 s for millet and cowpea, respectively. Millet and sesame had the shortest TTI and were also similar ranging from 25 ± 2 to 74 ± 8 s and 25 ± 2 to 77 ± 8 s under 25–47.5 kW/m², respectively. Cowpea and shelled peanuts had very comparable ignition times. Their TTI were 209 ± 10, 94 ± 9 and 52 ± 4 s and then 200 ± 12, 99 ± 16 and 53 ± 2 s, respectively, for cowpea and shelled peanuts, respectively, under 25, 35 and 47.5 kW/m². Therefore, it appears that cowpea and shelled peanuts have a longer delay or resistance to piloted ignition within the MC and heat fluxes tested compared to the rest of the grains.

Figure 6.

Time to sustained flaming ignition of sample food grains (TTI) and measured surface temperature at ignition ( $T_{s, ig}$ ). Figures within the bars are mean values while error bars show the standard deviations.

It should be noted that the unshelled peanut had two-stage ignition, as already detailed in the section on the important visual observations. In this research, the time from heat flux exposure to the second sustained ignition was taken as the TTI for the unshelled peanuts. In all cases, the TTI decreased with exposure to higher heat flux. In addition, compared to previous research on fire behaviour and thermal hazards of FGs that used oven-dried grains,¹¹ the TTI recorded in this study was higher due to higher MC. For instance, under 35 kW/m², the previous study had a range of 28–64 s while in this research, the range was 42–99 s. The longer TTI in this study was as expected for two major reasons. First, it took time to dry the sample, although it did not have to be completely dried.¹⁴ Second, the evaporated moisture diluted or reduced the concentration of combustible pyrolysate⁴¹ thereby requiring more decomposition before ignition could occur.

Surface temperature at ignition $T_{s, ig}$

The experimental surface temperature at the point of ignition, $T_{s, ig}$ measured for the grains are shown in Figure 6. The increase of $T_{s, ig}$ with heat flux is in agreement with the trend in plastics^20,42 but at variance to the trend in a specific case of wood.²⁰ In this research, the trend of the $T_{s, ig}$ with heat flux varies with the different grains. This also reaffirms the difficulty of measuring surface temperatures⁴³ with thermocouples due to the imperfect thermocouple tip to sample surface contact⁴⁴ that allows for radiative and convective fluxes affecting the measurements.

Unshelled peanut, as shown in Figure 6(f), shows a clear trend of increasing $T_{s, ig}$ with heat flux. This can be attributable to its two-stage process to sustained ignition, as discussed in the section on the important visual observations. With a higher heat flux, the first ignition of the shells was quicker. The cooling phase that followed was shorter, and therefore, the surface temperature had not lowered before the second and sustained ignition occurred, which was taken as the ignition. This was the reason for the highest and widest $T_{s, ig}$ ranges of 418°C–556°C. For the small grains of millet, flax and sesame, (see Figure 6(c), (e) and (i)) the mild popping and darting of the grains prevented steady and constant heating of the same grains or surfaces until ignition. The extent to which this behaviour affected the variation of $T_{s, ig}$ needs further research for accurate quantification.

The millet $T_{s, ig}$ , shown as mean and standard deviations in Figure 6(c), had the lowest range and noticeably decreased consistently with heat flux, 273 ± 21°C, 243 ± 47°C and 154 ± 31°C under 25, 35 and 47.5 kW/m². This could be due to the formation of an insulating crust over the virgin grains. It was suspected that the swelling or intumescence of the grains on the surface lifted the top grains past the tip of the thermocouple such that the measurement point was below the igniting grains. This intumescing behaviour was likely more pronounced and more rapid with higher heat fluxes. Like millet, in terms of grain size and ignition behaviour, flax and sesame as seen in Figure 6(e) and (i), had the next lowest $T_{s, ig}$ which ranged from 258 ± 45°C to 288 ± 25°C and 237 ± 53°C to 276 ± 37°C, respectively. The rest of the grains had comparable $T_{s, ig}$ ranging from the minimums of 259 ± 44°C to 338 ± 47°C to maximums of 329°C–390°C, which on average was 330°C. Therefore, small grains generally have lower $T_{s, ig}$ .

Analysis – resolution of the challenge of thermal thickness

The analysis or correlation of the TTI with the external heat flux, ${\overset{\cdot}{q}}_{e} ″$ , requires the establishment of the thermal thickness, δ_T, of the material. This is because the correlations relating the TTI to the ${\overset{\cdot}{q}}_{e} ″$ , justified by thermal ignition theory, vary depending on heat flux, geometric thickness, and thermal penetration depth.^27,45 Different approaches can be used to choose the appropriate thermal thickness model or correlation for ignition data analysis, which are now considered below.

For the first approach, it is assumed that the correct thermal thickness will have a distinctly higher correlation coefficient between the plots of TTI^-1 versus ${\overset{\cdot}{q}}_{e} ″$ (thermally thin), and TTI^-0.5 versus ${\overset{\cdot}{q}}_{e} ″$ (thermally thick). However, in this work, both thermal thickness plots (see Figure 7) showed good correlation. Such behaviour has previously been observed in forest fuels,⁴⁶ electric cable insulation⁴⁷ and natural and synthetic fabrics.⁴⁸ This indicates that both models provide suitable predictions of the TTI or ${\overset{\cdot}{q}}_{e} ″$ in the range studied, meaning that neither model can be selected as providing a generalized description without further analysis.

A second approach is to determine the Biot number, Bi = h_cL/k, a dimensionless parameter rooted in conduction heat transfer and is extensively dealt with in standard heat transfer or fundamental fire dynamics textbooks.^21,49 For the calculations, L is the characteristic length or the sample test depth, and k is the sample thermal conductivity. It was considered appropriate for this practically draught-free test environment, as detailed in the section on experimental apparatus and procedure, to assume h_c as 0.015 kW/m² K.^19,50 The test sample is thermally thin if Bi << 1 or thermally thick if Bi > 0.1.^49,51,52 Results from this work are presented in Table 1.

The third approach is to determine the value of the thermal penetration depth (δ_T). This is based on the same principle as the Biot number. In practice, the δ_T is given by A√(αTTI) where A is an arbitrary value usually ranging from 1.12 to 4. However, Babrauskas¹⁴ indicates that the appropriate value is 1.13 to 1.2. This notwithstanding, in this work, a more conservative value of 2.0 considered adequate for fire engineering applications by Drysdale⁴⁹ was used. It is noteworthy that $α$ is almost temperature independent for solids and for non-metals is nearly constant.^53,54 This is because α is a compound property such that a change in $k$ is compensated for by a change in $ρ .$ ⁵⁵

The fourth method is to place a thermocouple at the bottom of the sample to read the temperature. The temperature reading at the time of ignition of a thermally thick sample would be close to the ambient or sample temperature prior to the test. This method was, however, not used in this study.

Figure 7.

Correlations of the mean times to ignition, TTI, with external heat flux, ${\overset{\cdot}{q}}_{e} ″$ .

Table 1.

Biot numbers, Bi and thermal penetration depths, $δ_{T}$ , of the FGs tested.

Grain	k	$α$	Reference	Test depth L	$Bi$	TTI ^a	$δ_{T}$
(–)	(W/mK)	×10^–7 (m²/s)	(–)	(mm)	(–)	(s)	(mm)
Cowpea (Black eye beans)	Unavailable	0.731	Akinola and Adeniji⁵⁶	17	ND	209	7.8
Lentils	0.191	1.900	Gharibzahedi et al.⁵⁷	17	1.34	93	8.4
Millet	0.145	1.765	Subramanian and Viswanathan⁵⁸	17	1.76	74	7.2
Soybean	0.123	0.835	Deshpande et al.³⁴	17	2.07	160	7.3
Flax (Linseed)	Unavailable	Unavailable	ND	17	ND	107	ND
Unshelled peanut	0.128	2.550	Bitra et al.³³	25	2.93	113	10.7
Sunflower	0.120	3.010	İnce et al.⁵⁹<	17	2.13	130	12.5
Shelled peanut	0.158	1.050	Bitra et al.³³	20	1.90	200	9.6
Sesame	0.100	0.663	Ashtiani et al.⁶⁰	17	2.55	77	4.5

FG: food grain; TTI: time to ignition; ND: not determined due to unavailable data.

This is the maximum TTI of each FG shown in Figure 6.

Specifically, the values of k (except for cowpea and flax) and $α$ (except for flax) for the various grains, extracted from literature are shown in Table 1. It can be reasonably assumed that k and $α$ of cowpea and flax will be in the same range as others, as they were not available in published data sets. In summary of this section, (1) the materials studied show a suitable correlation with either thermally thin or thermally thick ignition models. (2) Based on the Biot numbers which are greater than 0.1, the test samples are thermally thick as shown in Table 1. Finally, for (3) the thermal penetration depth also indicates thermally thick behaviour as shown in Table 1. Therefore, the results of the correlation based on thermally thick material assumptions will be used to determine other ignition characteristics.

Analysis – determination of $TRP, T_{ig}, h_{ig}, and k ρ c$

The TRP was calculated as the reciprocal of the slope for the thermally thick correlations in Figure 7 described by equation (2). The ignition temperatures, T_ig, were calculated from equation (5) using experimentally determined CHF from previous research¹¹ and values of $ε$ and $σ$ as 0.95 and 5.67×10⁻¹¹ kW/m²K⁴, respectively. The assumption of 0.95 as the emissivity of the grain surfaces was because the sample surfaces were coarse, they also charred and turned black before flaming ignition. Also, as indicated in the section on resolution of the challenge of thermal thickness, h_c was taken as 0.015 kW/m²K. The use of CHF from 0% MC was based on the experimental study by Khan et al.⁴⁵ concluding that CHF is not affected by MC. This is also in agreement with earlier studies by Simms and Law⁶¹ as well as Moghtaderi.²⁰ Also, at low heat flux or critical flux, ignition time will increase to allow the moisture to dry, however the CHF remains the same. The thermal inertia, $k ρ c$ , was calculated from equation (4). The total heat transfer coefficient at ignition, $h_{ig}$ , was also determined using equation (6) with T_o as 20°C and the results are presented in Table 2.

Table 2.

Ignition properties of study food grains.

Grain	TRP	Grain sieve size	^a CHF	h_ig	$k ρ c$	Calculated T_ig	^bExperimental T_s,ig
Grain	(kWs^0.5/m²)	(mm)	(kW/m²)	(kW/m²K)	((kW/m²K)²s)	(°C)	(°C)
Millet	269	1.4	6.6	0.0301	1.911	239	223 ± 51
Sesame	261	1.6	8.9	0.0333	1.218	287	263 ± 18
Flax (Linseed)	265	2.4	9.4	0.0340	1.167	296	271 ± 12
Lentils	412	3.2	9.9	0.0347	2.649	305	288 ± 30
Soybeans	314	3.8	8.9	0.0333	1.766	287	315 ± 29
Cowpea	325	3.8	8.6	0.0329	1.968	281	340 ± 28
Sunflower	313	5.0	9.4	0.0340	1.629	296	347 ± 31
Shelled peanut	338	11.5	9.9	0.0347	1.785	305	349 ± 8
Unshelled peanut	565	20	9.9	0.0347	4.993	305	489 ± 56

TRP: thermal response parameter; CHF: critical heat flux.

Experimentally determined critical heat flux in Oguaka et al.¹¹

The $T_{s, ig}$ are mean and standard deviations from experimental measurements.

TRP is an indication of the resistance of a material to decomposition and generation of ignitable pyrolysate. Therefore, it is a property that can be considered to be strongly associated with material ignition⁶² and fire propagation.⁶³ The higher the TRP, the more resistant the material is to ignition. Sesame, flax (linseed) and millet have the lowest values at 261, 265 and 269 kWs^0.5/m². These low values are not surprising, as these also had the lowest TTI, T_ig and $T_{s, ig}$ as shown in Figure 6 and Table 2. This suggests that small grains readily generate sufficient combustible pyrolysate and quickly reach the lower flammability limit for ignition to occur. This is comparable to the instance of a bed of pine needles igniting as a thermally thick fuel while a single or individual needle behaves as a thermally thin fuel.²² Sunflower, soybean, cowpea and shelled peanut are comparable at 313, 314, 325 and 338 kWs^0.5/m², respectively. The lentil TRP determined was higher at 412 kWs^0.5/m², while the unshelled peanut value was very high at 565 kWs^0.5/m². The reason for the unshelled peanut can be attributed to the adoption of the second-stage sustained ignition temperature and time values for analysis. This behaviour has been elaborated on in the sections on TTI, and surface temperature at ignition. These distinct values of TRP suggest that it is a better definite indicator of ignitability than T_ig or $k ρ c$ . This is because it considers the combined properties of T_ig, and $k ρ c$ .

The calculated values of millet, lentils and soybean are within the bounds of uncertainties of the measured values. Sesame, flax and sunflower are 6°C, 13°C and 20°C, respectively outside the bounds of the uncertainties of the measured values and can be considered satisfactory.⁶⁴ Cowpea and shelled peanut are 31°C and 36°C, respectively outside the bounds of the uncertainties of the measured values. Note should be taken that these calculated or theoretical values are based on assumptions of thermal ignition theory as well as assumed values of emissivity and convective heat transfer coefficient. Therefore, the various T_ig or T_s,ig obtained for the grains are considered to be largely acceptable. The wide difference in the T_ig and the mean T_s,ig of unshelled peanuts is attributable to the adoption of the second-stage ignition data. The deviation for lentils, millet, soybeans, flax and sesame is less than 30°C. In addition, the values of the total heat transfer coefficient, h_ig of 0.0301–0.0347 kW/m²K are all similar and in the range for different species for woods of 0.0306–0.0359 kW/m²K determined by Janssens.²³

Thermal inertia is a critical property that indicates how readily a material can transmit heat or resist the development of a temperature differential or gradient across it. A material with low thermal inertia will retain heat received or absorbed on the surface and resist transmitting or transferring it into the body of the material. This implies that the surface temperature of such materials exposed to heat will rise fast and attain their ignition temperature and speed up fire spread.⁴⁹ It is provided as input in some fire models. The grains in this study have values that ranged from 1.17 to 2.65 (kW/m²K)²s, as shown in Table 2, except for unshelled peanuts at 4.99 (kW/m²K)²s. This high value is expected as it took a long time for the sustained ignition of the unshelled peanuts because of the two-stage ignition process as discussed in the section on the important visual observations. Apart from unshelled peanuts, based on thermal inertia, the grains are not distinctly dissimilar and their behaviour is relatively consistent. In comparison, the values of $k ρ c$ of some wood species are about 0.5 (kW/m²K)²s,^20,65 while some plastics – PMMA (poly(methyl methacrylate)) and glass-reinforced plastic – 1.00 and 0.32 (kW/m²K)²s,⁶⁶ respectively. These suggest that food grains at 10% MC are more likely to be slow to ignite and spread fire than wood and plastics if they have the same T_ig,²¹ as would be expected.

Conclusion

For decades now, fires in FG ST&P facilities in different parts of the world have been known to have caused huge damage or loss of lives and properties. There may not be many of these fires in a year, but whenever they occur their consequences are usually extensive. This is also in addition to the negative impact of smoke and residues on the environment. If these accidents are to be minimized, an understanding of the fire behaviour of these FGs is necessary. A quantitative study on ignition is, therefore, a vital step in limiting the occurrence and extent of damage that these fires can cause.

The ignition properties of nine different food grains at 10% MC have been derived from experimental data under a radiant cone heater. The ignition temperatures T_ig varied from 239°C (millet) to 305°C (lentils, unshelled peanut and shelled peanut). Other ignition properties of TRP (261–349, and 565 kWs^0.5/m²), and $k ρ c$ (1.17–2.65, and 4.99 kW/m²K²s) were determined from a correlation of TTI with ${\overset{\cdot}{q}}_{e} ″$ . The ignition characteristics of this study show that the T_ig is typically in the lower range of most solids, which is between 300°C and 500°C. In addition, $k ρ c$ values were found to be about the same for all the grains, except for unshelled peanuts, but generally higher than some wood species. This indicates that these food grains can be slower to ignite should their T_ig be the same as the woods. However, TRP, which combines T_ig and $k ρ c$ seems to be a more distinguishing property for the ignitability of the FGs. The small particle size grains have lower TRP, which suggests ignitability can possibly be inferred from FG particle size. However, studies focussed on the effect of particle size on TRP are required to confirm this.

This study has been limited to FG at 10% MC, which is generally the storage condition of most FGs. The effect of MC on these characteristics is yet to be determined. As with natural materials of this type, variations with different species are expected and could be considered in future studies. Finally, this study was conducted on a bench-scale apparatus with a controlled environment. Large-scale studies are needed to determine how these results scale to real fire situations.

Footnotes

Appendix

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Royal Academy of Engineering / Lloyd’s Register Foundation under the ‘Engineering Skills Where They are Most Needed’ grant [Grant ESMN 192-1-141]; and the Society of Fire Protection Engineering (SFPE) Frederick W. Mowrer Global Scholar Award endowed by Kathleen Almand.

ORCID iD

Anene Oguaka

Author biographies

Anene Oguaka has been a consulting mechanical engineer in the built environment for upwards of 15 years. Before this, he was an academic at Ekiti State University, Ekiti State, Nigeria. He has recently completed his PhD studies at Stellenbosch University where he focussed on the fire behaviour of agricultural food grains.

Natalia Flores-Quiroz is a postdoctoral researcher in the fire engineering team of Fire safety at Stellenbosch University. Her main research areas are reconstruction of incidents in low-income settlements (i.e. informal settlements, refugee camps) and wildland urban interface (WUI) fires.

Richard Walls is the arsonist-in-chief of the fire engineering team at Stellenbosch University. After being a practising structural engineer, he took up incendiary research interests focussing on building safety, material development for high temperatures, investigations and other topics. The team runs masters and PhD programmes in fire safety which are expanding the field into developing countries.

References

Bhusari

Goh

, et al. Process safety incidents across 14 industries. Process Saf Prog 2021; 40: e12158.

Food and Agriculture Organization. The impact of disasters and crises on agriculture and food security: 2021. Italy: Food and Agriculture Organization of the United Nations Rome, 2021, http://www.fao.org/documents/card/en/c/cb3673en

Nukenine

Stored product protection in Africa: past, present and future. Quedlinburg: Julius-kühn-Archiv, 2010.

Xie

Gao

Huang

Fire risk and behavior of rice during the convective drying process. Fire Saf J 2020; 115: 103013.

Jornal Do Oeste. Fire in soybean dryer mobilizes firefighters in the region, https://www.jornaldooeste.com.br/seguranca/incendio-em-secador-de-soja-mobiliza-bombeiros-da-regiao/ (2023, accessed 17 July 2023).

G1 and Globo. More than 80 tons of corn are destroyed during a fire in a grain dryer in MT. G1, https://g1.globo.com/mt/mato-grosso/noticia/2023/06/15/mais-de-80-toneladas-de-milho-sao-destruidas-durante-incendio-em-secador-de-graos-em-mt-video.ghtml (2023, accessed 17 July 2023).

Brüning

High inflamability: safety in grain dryer handling. Fire and Explosion, https://fire-and-explosion.com/blog/2019/04/10/high-inflamability-safety-in-grain-dryer-handling/ (2019, accessed 27 July 2023).

Wang

Xing

Huang

, et al. Smoldering of storage rice: effect of moldy degree and moisture content. Combust Sci Technol 2022; 194: 1395–1407.

Rosa

Hammad

AWA

Qualharini

, et al. Smoldering fire propagation in corn grain: an experimental study. Results Eng 2020; 7: 100151.

10.

Cai

Zhao

Jin

, et al. Evolvement investigation of combustion behavior of corn based on cone calorimeter. Combust Sci Technol 2020; 192: 229–239.

11.

Oguaka

Flores Quiroz

Walls

Fire parameters, behaviour, and comparative thermal hazard of food grains based on the cone calorimeter tests. Process Saf Environ Prot 2023; 179: 928–940.

12.

Lyon

Janssens

. Polymer flammability. In: Mark

Seidel

(ed.) Encyclopedia of polymer science and technology. Hobokn, NJ: John Wiley & Sons, pp. 1–70.

13.

Dao

Luche

Richard

, et al. Determination of characteristic parameters for the thermal decomposition of epoxy resin/carbon fibre composites in cone calorimeter. Int J Hydrog Energy 2013; 38: 8167–8178.

14.

Babrauskas V. Ignition handbook: principles and applications to fire safety engineering, fire investigation, risk management and forensic science. Issaquah, WA: Fire Science Publishers2003.

15.

Tewarson

Ogden

SD.

Fire behavior of polymethylmethacrylate. Combust Flame 1992; 89: 237–259.

16.

Babrauskas

Ignition of wood: a review of the state of the art. J Fire Prot Eng 2002; 12: 163–189.

17.

Torero

. Flaming Ignition of Solid Fuels. In: Hurley

Morgan J.

Gottuk

Hall

, et al. (eds) SFPE handbook of fire protection engineering. New York, NY: Springer New York, pp. 633–661.

18.

Vermesi

Roenner

Pironi

, et al. Pyrolysis and ignition of a polymer by transient irradiation. Combust Flame 2016; 163: 31–41.

19.

Fangrat

Hasemi

Yoshida

, et al. Surface temperature at ignition of wooden based slabs. Fire Saf J 1996; 27: 249–259.

20.

Moghtaderi

Novozhilov

Fletcher

, et al. A new correlation for bench-scale piloted ignition data of wood. Fire Saf J 1997; 29: 41–59.

21.

Quintiere

. Fundamentals of fire phenomena. 1st ed. Hoboken, NJ: John Wiley & Sons.

22.

Fateh

Richard

Batiot

, et al. Characterization of the burning behavior and gaseous emissions of pine needles in a cone calorimeter – FTIR apparatus. Fire Saf J 2016; 82: 91–100.

23.

Janssens

A thermal model for piloted ignition of wood including variable thermophysical properties. Fire Saf Sci 1991; 3: 167–176.

24.

Quintiere

Harkleroad

New concepts for measuring flame spread properties. Fire Saf Sci Eng ASTM STP 1985; 882: 239–267.

25.

Buschman

. Ignition of some woods exposed to low level thermal radiation: second progress report. NBS Report 7848. Gaithersburg, MD: National Bureau of Standards.

26.

Abu-Zaid

Effect of water on ignition of cellulosic materials. PhD Thesis, Michigan State University, East Lansing, MI, 1988.

27.

Mikkola

Wichman

IS.

On the thermal ignition of combustible materials. Fire Mater 1989; 14: 87–96.

28.

Tewarson

. Generation of heat and chemical compounds in fires. In: DiNenno

(ed.) SFPE handbook of fire protection engineering. Boston, MA: National Fire Protection Association, 2002, pp. 82–158.

29.

Oguaka

Flores Quiroz

Walls

Comparative pyrolysis characteristics and kinetics of agricultural food grains by thermogravimetric analysis. Process Saf Environ Prot 2023; 179: 559–574.

30.

Ziegler

Paraginski

Ferreira

CD.

Grain storage systems and effects of moisture, temperature and time on grain quality – a review. J Stored Prod Res 2021; 91: 101770.

31.

Luo

Guo

Zhang

, et al. Insight into pyrolysis kinetics of lignocellulosic biomass: isoconversional kinetic analysis by the modified Friedman method. Energy Fuels 2020; 34: 4874–4881.

32.

Dietsch

Franke

, et al. Methods to determine wood moisture content and their applicability in monitoring concepts. J Civ Struct Health Monit 2015; 5: 115–127.

33.

Bitra

VSP

Banu

Ramakrishna

, et al. Moisture dependent thermal properties of peanut pods, kernels, and shells. Biosyst Eng 2010; 106: 503–512.

34.

Deshpande

Bal

Ojha

TP.

Bulk thermal conductivity and diffusivity of soybean. J Food Process Preserv 1996; 20: 177–189.

35.

Janssens

. 2 – Fundamental measurement techniques. In: Apte

(ed.) Flammability testing of materials used in construction, transport and mining. 2nd ed. Sawston: Woodhead Publishing, pp. 23–61.

36.

Weickhardt

Ritchie

Batterham

, et al. Surface temperature measurement. Chem Eng J 1971; 2: 212–214.

37.

Dietenberger

Grexa

White

RH.

Reaction-to-fire of wood products and other building materials: part II, cone calorimeter tests and fire growth models (FPL-RP-670). Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Products Laboratory, 2012.

38.

Hopkins

Quintiere

JG.

Material fire properties and predictions for thermoplastics. Fire Saf J 1996; 26: 241–268.

39.

ISO 5660-1: 2015. Reaction-to-fire tests – heat release, smoke production and mass loss rate – Part 1: heat release rate (cone calorimeter method) and smoke production rate (dynamic measurement).

40.

Olson

Beeson

Haas

, et al. An earth-based equivalent low stretch apparatus for material flammability assessment in microgravity and extraterrestrial environments. Proc Combust Inst 2005; 30: 2335–2343.

41.

Mouritz

AP.

Durability of composites exposed to elevated temperature and fire. Dur Compos Civil Struct Appl 2007; 2007: 98–125.

42.

Thomson

Drysdale

Beyler

CL.

An experimental evaluation of critical surface temperature as a criterion for piloted ignition of solid fuels. Fire Saf J 1988; 13: 185–196.

43.

Tran

White

RH.

Burning rate of solid wood measured in a heat release rate calorimeter. Fire Mater 1992; 16: 197–206.

44.

Smith

King

JB.

Surface temperature of materials during radiant heating to ignition. J Fire & Flammability 1970; 1: 17.

45.

Khan

De Ris

Ogden

Effect of moisture on ignition time of cellulosic materials. Fire Saf Sci 2008; 9: 167–178.

46.

McAllister

Grenfell

Hadlow

, et al. Piloted ignition of live forest fuels. Fire Saf J 2012; 51: 133–142.

47.

Rantuch

Štefko

Martinka

Critical heat flux determination of electric cable insulation. Res Pap Fac Mater Sci Technol Slovak Univ Technol 2018; 26: 11–20.

48.

El Gazi

Sonnier

Giraud

, et al. Fire behavior of thermally thin materials in cone calorimeter. Polymers 2021; 13: 1297.

49.

Drysdale

An introduction to fire dynamics. 3rd ed. Hoboken, NJ: John Wiley & Sons, 2011.

50.

Staggs

JEJ

. Convection heat transfer in the cone calorimeter. Fire Saf J 2009; 44: 469–474.

51.

Bergman

Levine

AS.

Fundamentals of heat and mass transfer. 8th ed. Hoboken, NJ: John Wiley & Sons, Inc, 2019.

52.

Fateh

Rogaume

Luche

, et al. Characterization of the thermal decomposition of two kinds of plywood with a cone calorimeter – FTIR apparatus. J Anal Appl Pyrolysis 2014; 107: 87–100.

53.

Rhodes

Quintiere

JG.

Burning rate and flame heat flux for PMMA in a cone calorimeter. Fire Saf J 1996; 26: 221–240.

54.

Spearpoint

Quintiere

JG.

Predicting the piloted ignition of wood in the cone calorimeter using an integral model – effect of species, grain orientation and heat flux. Fire Saf J 2001; 25: 391–415.

55.

Delgado

Sun

D-W

Rubiolo

. Thermal physical properties of foods. In: Sun

D-W

(ed.) Thermal Food Processing: New Technologies and Quality Issues. Taylor & Francis Group: LLC, 2006, pp. 3–32.

56.

Akinola

Adeniji

OB.

Estimation of the thermal diffusivity of beans using a simple apparatus. FUOYE J Eng Technol 2019; 4: 43–47.

57.

Gharibzahedi

SMT

Ghahderijani

Lajevardi

ZS.

Specific heat, thermal conductivity and thermal diffusivity of red lentil seed as a function of moisture content: thermal properties of red lentil seeds. J Food Process Preserv 2014; 38: 1807–1811.

58.

Subramanian

Viswanathan

Thermal properties of minor millet grains and flours. Biosyst Eng 2003; 84: 289–296.

59.

İnce

Güzel

İnce

Thermal properties of some oily seeds. J Agric Mach Sci 2008; 4: 399–405.

60.

Ashtiani

SHM

Emadi

Sanaei-Mohammad

, et al. Effect of moisture content and temperature on thermal behaviour of sesame seed. Ann Univ Dunarea Jos Galati Fascicle Technol 2014; 38: 87–103.

61.

Simms

Law

. The ignition of wet and dry wood by radiation. Combust Flame 1967; 11: 377–388.

62.

Mowrer

FW.

An analysis of effective thermal properties of thermally thick materials. Fire Saf J 2005; 40: 395–410.

63.

Khan

Tewarson

Chaos

. Combustion characteristics of materials and generation of fire products. In: Hurley

Gottuk

Hall

, et al. (eds) SFPE handbook of fire protection engineering. New York: Springer, 2016, pp. 1143–1232.

64.

Taylor

JR.

An introduction to error analysis: the study of uncertainties in physical measurements. 2nd ed. Sausalito, CA: University Science Books, 1997.

65.

Janssens

Improved method for analyzing ignition data from the cone calorimeter in the vertical orientation. Fire Saf Sci 2003; 7: 803–814.

66.

Quintiere

JG.

Principles of fire behavior. 2nd ed. Boca Raton, FL: CRC Press, 2016.

Radiative ignition properties of food grains and surface temperatures at ignition

Abstract

Keywords

Introduction

Thermal ignition theory for data analysis

Materials and methods

Sample selection and preparation

Experimental apparatus and procedure

Results and discussions

Important visual observations

Time–surface temperature profile

Time to ignition

Surface temperature at ignition T s , ig

Analysis – resolution of the challenge of thermal thickness

Analysis – determination of TRP , T ig , h ig , and k ρ c

Conclusion

Footnotes

Appendix

Declaration of conflicting interests

Funding

ORCID iD

Author biographies

References

Surface temperature at ignition $T_{s, ig}$

Analysis – determination of $TRP, T_{ig}, h_{ig}, and k ρ c$