Assessment of the Self-Heating Potential of Fresh Wood Using the Pulse Flow Calorimetric Method

Abstract

The self-heating propensity of the fresh wood of ten tree species (two coniferous, eight deciduous) was studied calorimetrically using oxidation heats, q₃₀, at a temperature of 30 °C. Values of q₃₀ in the range between 0.45 W kg⁻¹ (dry) and 1.1 W kg⁻¹ (dry) were found. The lowest evolution of the oxidation heat proved two coniferous wood types—spruce and pine. On the other hand, the highest value of the q₃₀ heat manifested willow wood, which exceeded (as the only one of the samples) the level of 1 W kg⁻¹ (dry). Water was confirmed to promote the generation of oxidation heat, while completely negligible oxidation heat effects were found in dry wood samples. A rise in the heat evolution with increasing moisture content can be explained not only by a change in the mechanistic pathway of the chemical oxidation of wood in the presence of water, but also by the restoration of the activity of microorganisms in wood, which occurs only at a sufficient level of moisture content. Tree bark appears to be probable carrier of a diverse microbiome. Based on the experiments with debarked wood samples, it can be estimated that the part of the heat produced by microorganisms constitutes a remarkable 35–55% of the global oxidation heat q₃₀, as determined for fresh wood samples.

1. Introduction

The phenomenon of self-heating was habitually related to coal [1,2] as a traditional source of energy in many countries over the world. However, because of climate changes and awareness of the overexploitation of the earth’s finite resources, demand for biomass as an alternative energy source to replace traditional fossil fuels is increasing. As a consequence, the number of spontaneous heating incidents caused by biomass is rising. For example, in the Czech Republic, spontaneous combustion incidents of coal clearly prevailed twenty years ago over those of biomass/agriculture products; however, since the year 2015, the ratio is quite the inverse (see Figure 1).

To deal with the spontaneous heating of lignocellulosic biomass, experience with the self-heating of coal matter is usually applied [3,4]. As sites of primary chemical attacks of oxygen on the lignocellulosic structure, unsaturated wood extractives like fatty acids and/or resins are usually reported, which form hydroxyl radicals undergoing further reactions [3,5,6]. However, the oxidation behaviour of biomass differs from coal oxidation due also to the possible microbial activity of the microorganisms, mainly in the low temperature stage of the self-heating process [6,7]. Thus, knowledge of the oxidation behaviour of coal can hardly be directly transferred to the spontaneous heating of biomass. For example, unsaturated hydrocarbons, which play a major role in indicating the initial stage of coal self-heating, have proven to be useless for indicating the self-heating of lignocellulosic biomass [8].

An inherent requirement of industrial producers and users is assessing the self-heating potential of a substance as an important factor influencing its “spon-com” risk in practice (in addition to external/technological aspects). As far as coal is concerned, a number of different methods have been proposed (e.g., [9,10]) among which calorimetry is of exclusive importance. The main argument for calorimetry is the possibility to direct evaluate the oxidation heat, which is considered the main cause of the self-heating occurrences [11,12,13]. The calorimetric approach was also recommended to assess the spontaneous heating risk of the wood pellets [14,15,16,17,18]. The method of the isothermal calorimetry, which was developed for the purpose, works with pellets of moisture content to ca 10% at a temperature of 60 °C. The pellets are placed into ampoules filled with air and sealed [16,17,18]. Heat evolution with elapsing time is then registered and the maximum specific heat release rate (HRR_max, mW g⁻¹) is evaluated for ranking the different pellet batches.

As an alternative to the isothermal calorimetry, the method of pulse flow calorimetry (PFC) can be considered. The PFC method was established at the University of Ostrava in the late 1980s [19], and, in addition to coal samples, metallic powders were also studied to assess their self-heating potential [20]. It enables measurements of samples with natural moisture content at ambient temperature (30 °C) and convincingly separates oxidation heat from moisture evaporation and/or condensation heat effects. It can also distinguish between the physical (reversible) and chemical (irreversible) interaction of oxygen with the sample. In comparison with isothermal calorimetry, the method of PFC does not require a preheating period before the sample enters the calorimeter, thus eliminating initial temperature disturbances [16].

This paper aims to present basic results obtained in pulse flow calorimetric investigations of the self-heating propensity of different fresh wood samples. The effect of moisture content on oxidation behaviour of the woods is described, and an attempt is also made to quantify the heat effects resulting from the activity of the microorganisms under aerobic conditions. The novelty of the study can be seen mainly in the original data on the oxidation behaviour of fresh wood samples at a temperature close to ambient.

2. Materials and Methods

2.1. Samples

Wood samples from ten tree species (two coniferous, eight deciduous) were studied. The samples include trees that are abundant in the Czech Republic, which belongs to the temperate climate zone. The list of wood species is given in

Table 1. Oxidation heats of fresh wood samples, q₃₀.

Wood Sample	Oxidation Heat, q30, W kg−1 (dry) *	Moisture Content, MC, % **	Coefficient of Variation, CV, %
Willow (Salix erythroflexuosa)	1.1	47/58	8.0
Birch (Betula pendula)	0.80	47/51	10
Linden (Tilia cordata)	0.78	36–56	19
Ash (Fraxinus excelsior)	0.76	31/37	10
Maple (Acer platanoides)	0.71	41/57	12
Hornbeam (Carpinus betulus)	0.63	34/45	11
Oak (Quercus robur)	0.62	41/46	27
Beech (Fagus sylvatica)	0.48	37/46	13
Spruce (Picea abies)	0.47	34/38	3.0
Pine (Pinus sylvestris)	0.45	45/45	11

* q₃₀ values represent the average of two replicates (7 replicates for the linden sample); ** moisture content (MC) indicate actual value of MC in the sample at the replicated measurements (the moisture content range is given for the linden sample).

.

The preparation of the samples for the pulse flow calorimetric examination strictly followed the main goal of the investigations, i.e., determining the oxidation heat of the fresh wood of natural moisture content. In this respect, a wood sample was taken from a given tree by cutting a branch about 30 cm long and 1 cm in diameter (max) and sealed in an airtight plastic bag. The calorimetric examinations were carried out no longer than 24 h after in situ sample extraction. Immediately before measurement, the wood branch was cut into thin slices (thickness of ~1 mm), which were further diminished using a knife mill and then sieved to achieve a uniform grain size below 1 mm. The diminishing procedure was carried out in air at room temperature (ca 20 °C), and it took less than 30 min until the sample was put into flow of the inert gas (helium) in the calorimeter. Such a sample is considered fresh. The moisture content of the tested wood was determined after the calorimetric measurement (drying under vacuum at 105 °C for ca. 1.5 h). The possible shift in the sample moisture during the calorimetric examination was ascertained from the weight difference in the calorimetric vessel before and after the measurement.

2.2. Pulse Flow Calorimetric Method, Oxidation Heat Measurements

The principle of the pulse flow calorimetric method consists of introducing a discrete amount (pulse/dose) of oxygen into the flow of an inert gas and measuring the heat caused by the passage of the oxygen pulse through the sample [19,20].

The PFC measurements were carried out using a Setaram C80 calorimeter (Caluire-et-Cuire, France) provided by a circulation vessel. The flowing rate of helium (as an inert carrier gas) was about 2.5 mL per minute; the volume of the oxygen pulse was 80 mL and the passage of the oxygen through the sample thus took half an hour. The temperature of the experiments was 30 °C. The weight of the wood samples was about 1 g. A typical pulse flow calorimetric curve (as obtained of the measurement of the linden wood sample) is depicted in Figure 2.

From the calorimetric tests, the mean oxidation heat flow q₃₀ during the first 30 min chemical (irreversible) reaction between oxygen and the fresh wood sample was determined (J (kg·s)⁻¹, equivalent to W kg⁻¹). The index 30 indicates temperature level and the residence time of oxidation. Due to the large content of moisture in the tested woods (commonly 40–50%), values of q₃₀ heats are here exclusively related to the mass of the sample in a dry state (W kg⁻¹ (dry)).

All measurements of the fresh woods were duplicated; the sample of linden wood was measured with seven replications (values of the oxidation heat q₃₀ obtained from the replications are statistically evaluated in Appendix A). Each determination was made with a newly taken sample (not necessarily originating of the same tree plant). The coefficient of variation in the measurements, CV (representing the ratio of the standard deviation to the mean value, %), varied from 3 to 27%, and gives an average value of CV = 12.

3. Results and Discussion

3.1. Oxidation Heat of Wood Samples

The basic results of the measurements of oxidation heats q₃₀ for fresh wood samples are summarised in Table 1.

Values of q₃₀ in the range between 0.45 and 1.1 W kg⁻¹ (dry) are seen in Table 1 for the whole series of the studied woods. The lowest evolution of the oxidation heat was found for two coniferous wood types—spruce and pine—proving values of q₃₀ equal to 0.47 kg⁻¹ (dry), and 0.45 W kg⁻¹ (dry), respectively. On the other hand, the highest value of the q₃₀ heat manifested wood of willow exceeding (as the only one sample) a level of 1 W kg⁻¹ (dry) (see Table 1). Comparing the obtained q₃₀ values with the heat of oxidation of wood reported in the literature is not easy: either it was determined at a higher temperature [16,17,18], or the history of the measured wood samples is uncertain [15,21]. In this way, the range of the measured values of q₃₀ is quite comparable with the maximum specific heat release rates (HRR_max) for highly reactive pellet batches (0.61–1.06 W kg⁻¹), as reported by Larsson et al. for isothermal calorimetry [17]; however, this resulted from investigations at 60 °C. In comparison with values of HRR_max measured at 30 °C [15,21], the oxidation heats q₃₀ are more than three times [21], or even an order of magnitude, higher [15].

Similarly, like the HRR_max parameter, the values of q₃₀ also reflect the maximum heat evolved from the oxidation of the wood sample corresponding to the heat effect determined from the first pulse of the oxygen applied to the sample. Each successive pulse then leads to a lower oxidation heat effect. On average, heat resulting from the second oxygen pulse decreased by ca 25% in comparison with that from the first pulse (see the difference in shapes of the calorimetric curves for the first and second pulse in Figure 2). Figure 3 then confirms a steady decrease in oxidation heat in the course of repeated pulses of oxygen, as was found for fresh wood of linden. Obviously, less than half the value of the initial level of the oxidation heat q₃₀ can be expected for the fresh sample after 3 h oxidation in pure oxygen.

The fresh basis of the wood samples in calorimetric experiments also allows for an immediate comparison of the measured heats q₃₀ with the self-heating ability of fresh coal and/or metals. Specifically, the q₃₀ values found in the range of 0.45–1.1 W kg⁻¹ (dry) would rank the wood in the “high risk” category according to the boundaries valid for bituminous coal. Simultaneously, all the wood species would belong (only) to the “low risk” category with regard to the classification scale of subbituminous coal, not exceeding the limit of 2 W kg⁻¹ [20]. The oxidation heats for wood are then quite comparable with the self-heating ability of the fresh iron, as evidenced by the value q₃₀ = 0.47 W kg⁻¹ [20].

As mentioned, the oxidation heats of q₃₀ are related to the dry mass of the wood. From Table 1, the differences in the moisture content, MC, of the samples at replicated calorimetric tests are seen to exceed even 10% (for the sample of linden wood with seven replications, the maximum difference in the values of MC was ascertained to be 20%). The variability of MC reflects possible changes in moisture content in a tree over the course of the year (the experiments were conducted from March 2024 to September 2025) as well as wood heterogeneity within trees of the same species. For instance, ten samples of a maple wood taken simultaneously from various twigs of the tree proved moisture contents differing by up to 11% (6% in the average). No doubt, water is the basic compound accompanying the oxidation of all wood samples; consequently, further attention is paid to a closer description of the effect of moisture on the oxidation behaviour of the wood.

3.2. Effect of Moisture on Oxidation Heat of Wood

A preliminary survey was conducted with a sample of naturally dried linden wood taken from a broken branch in situ. (The estimated time the broken branch had been lying under the tree is about six months.) The sample proved a residual moisture content of 9% and the oxidation heat effect was determined to be 0.04 W kg⁻¹ (dry), i.e., more than one order less in comparison with the fresh sample. However, the result is not only affected by the low moisture content, but also by the approximately six-month air oxidation of the broken branch in nature.

More sophisticated experiments were performed in the laboratory. Specifically, a sample of fresh wood was taken and, instead of measuring the oxidation heat, the sample was dried in a vacuum at 105 °C for 1.5 h. Then, the oxidation heat was determined. Afterwards, the dry wood was immediately moistened with water mist, mixed, and the oxidation heat was measured again, including the subsequent determination of moisture content, MC. This way, a series of the oxidation heats for samples of various moisture contents was obtained. Such an approach was applied for linden and birch wood samples. The results of the investigations are presented in Figure 4.

From Figure 4, a marked rise in oxidation heat with increasing moisture content is seen. In fact, completely negligible heat effects were found for the dry sample, quite comparable to the level of sensitivity of the calorimetric experiments (~0.01 W kg⁻¹ (dry)). The oxidation ability of dry wood matter to gaseous oxygen thus appears to be practically negligible. On the other hand, gradual wetting of dry wood essentially restores its self-heating potential: for linden, it was up to a level of 50% of the original level of q₃₀ heat, and wetted birch even reached the original value of the q₃₀ oxidation heat for the fresh sample (c.f., Figure 4). A similar recovery of the oxidation heat in wetted wood was previously observed by Li et al. [21] based on the thermoanalytical examinations of wood chips. No doubt, the presence of water promotes heat evolution from the wood samples under oxygen. Such an activating effect of water is also known in the oxidation of low-rank coals [22] and metal powders [20], where it is explained by changes in the mechanistic pathway of the oxidation process in the presence of water. In principle, the involvement of the water into a mechanistic pathway can also be anticipated for the chemical oxidation of the wood texture, despite the fact that its mechanisms have not been completely clarified yet [5,6]. However, exclusively for wood, another aspect needs to be considered, which has no analogy with the oxidation of coal or metals, i.e., the action of microorganisms. Namely, when preparing the samples in dry state, vacuum drying at a temperature of 105 °C was applied. However, the temperature of 105 °C is seen as insufficient for the sterilisation of the wood to eliminate microbial organisms [18]. As a consequence, a rise in heat evolution with an increase in the moisture content (Figure 4) could thus be related not only to a change in the mechanism of the chemical oxidation, but also to the restoration of the microorganisms’ activity occurring only at a sufficient level of moisture (generally, MC ≥ 20% [5]). Such an explanation is preferred by authors dealing with the self-heating of stored wood biomass [21,23]. An effort was thus finally expended to assess the possible contribution of heat originating from microbial activity to the global oxidation heat effect of fresh wood samples.

3.3. Towards Heat Generated by Microorganisms

The basic reflection on the importance of the microorganisms in relation to the oxidation process of fresh wood was obtained from examinations of the oxidation heat as a dependence of elapsing time. In the experiments, about 20 g of the fresh linden wood (grain size below 1 mm) was spread on a dish in a desiccator at temperature of about 20 °C in an air atmosphere with a relative humidity of 98% (to ensure a stable oxidation environment). Aliquots of the sample were then taken at selected time intervals, and the oxidation heat was measured. Intuitively, a steady decrease in oxidation heat values with time was expected, similar to the heat development during repeated pulses of oxygen (see Figure 3). In contrast, a steep rise in the heat evolution was found in the first time period, reaching a maximum on the 2nd day from the start of the experiment, followed by a steady decrease and certain stabilisation of the heat effects after ~12 days at a level corresponding to approximately one third of the oxidation heat q₃₀ for the fresh sample. A similar, non-monotonous course of the development of “oxidation heat—time” also resulted from measurements with willow wood. (Other wood species were not examined in this way.) The dependence obtained for the sample of linden wood is displayed in Figure 5, including the development of the actual moisture content, MC, in the samples with elapsing time.

Obviously, such development of the “oxidation heat versus time” dependence does not reflect the reaction kinetics, as expected for chemical reactions. On the other hand, such a shape is quite typical for systems with an ongoing fermentation process that occurs only in the presence of favourable microorganisms [24,25]. Evidently, such microorganisms must have been present already in the fresh linden sample. Although surprising, such a scenario is quite possible. Namely, recent studies reveal that the bark of the trees across various climate zones hosts a stable and diverse microbiome comprising bacteria, fungi, and yeasts as well as algae or cyanobacteria [26]. In the light, the values of the oxidation heats q₃₀ from measurements of the fresh wood samples can, in principle, be affected by heat coming from the action of the microbiome present in the bark. From the maximum value of the determined oxidation heat, 1.3 W kg⁻¹ (dry) (Figure 5), one can further deduce that the action of the microorganisms under optimal conditions caused a rise in heat evolution by 0.5 W kg⁻¹ (dry) (i.e., by 75%) in comparison with the mean level of the q₃₀ heat for the fresh sample of linden. The finding appears serious and in practice means that, depending on the actual concentration of microorganisms in the bark, differences in q₃₀ values in replicated measurements of the order of tens of percent are completely legitimate. In Table 1, the largest values of the coefficient of variation, CV = 27%, and CV = 19%, for oak and linden wood, respectively, are shown. Rather speculatively, such a large experimental scatter could be precisely explained by the different bark contamination in the linden/oak samples in replicated tests. To examine the aspect more closely, additional calorimetric experiments were performed with fresh samples of linden, oak, and birch wood, which had been carefully debarked beforehand. Such pre-treatment was aimed at the elimination of the (possible) bark-present microorganisms from the wood samples during the measurements of the oxidation heat. The results of the investigations are shown in

Table 2. Oxidation heats of fresh wood samples with/without bark.

Sample	Fresh Wood Sample (with Bark)		Fresh Wood Sample (Debarked)
	Oxidation Heat q30, W kg−1 (dry) *	Moisture Content, MC, % **	Oxidation Heat q30, W kg−1 (dry)	Moisture Content, MC, %
Birch (Betula pendula)	0.72–0.89	47/51	0.57	36
Linden (Tilia cordata)	0.60–1.00	36/43	0.45	53
Oak (Quercus robur)	0.46–0.79	41/46	0.48	34

* The q₃₀ heats of fresh birch and oak wood samples represent values of two replicates, and two values of q₃₀ heats for fresh samples of linden represent minimum and maximum heat effects of 7 replicates; ** values of MC indicate actual moisture content of the sample at the measurement.

, comparing minimum/maximum values of the q₃₀ heats from the previous measurements of the fresh woods (including bark) with the heat effects of q₃₀ obtained from experiments with debarked fresh samples.

Indeed, debarked birch and linden wood showed evidently lower oxidation heats (at least by 20%) compared to the q₃₀ of fresh samples, while the oxidation heat of debarked oak was at the minimum q₃₀ level found for the fresh sample. However, when comparing the q₃₀ heats of debarked wood with the upper values of the q₃₀ heat for fresh samples (see Table 2), one can estimate that the value of the heat produced by microorganisms represents 0.32 W kg⁻¹ (dry) (=36%) for birch, 0.31 W kg⁻¹ (dry) (=40%) for oak, and/or 0.55 W kg⁻¹ (dry) (=56%) for linden of the global oxidation heat q₃₀ of fresh wood samples. Although these values are rather estimated, they are comparable to (theoretical) heat production rates resulting solely from microbial activity in the biological oxidation of wood, as reported by Fan et al., 0.63 W kg⁻¹ (dry) [27] and/or by Caizán Juanarena et al., 0.6 W kg⁻¹ (dry) [28].

The experiments thus confirm the significant role of heat originating from the activity of microorganisms even in the (global) oxidation heat q₃₀ of fresh wood.

4. Conclusions

Based on the performed investigations of the self-heating propensity of fresh wood samples, the following findings can be considered principal:

• Oxidation heat q₃₀ of fresh wood taken from ten tree types was found to vary between 0.45 W kg⁻¹ (dry) and 1.1 W kg⁻¹ (dry).
• Moisture in wood acts as an accelerator of oxidation heat production. For dried wood, virtually no measurable evolution of the oxidation heat was observed.
• In the global value of the oxidation heat of fresh wood, q₃₀, a significant part can originate from the action of microorganisms. On the basis of experiments with debarked wood samples, it is estimated that the share is approximately 35–55%.

Based on the result, the drying of wood before its storage can also be positively recommended as an efficient pre-treatment procedure to prevent self-heating incidents in practice.