Steam Explosion Pre-Treatment of Sawdust for Biofuel Pellets

Abstract

The current study explores steam explosion pre-treatment of wood sawdust to develop high-quality biofuel pellets. In order to determine optimized conditions (temperature and residence time) for steam-treated biomass, seven test responses were chosen, including bulk, particle and pellet densities as well as tensile strength, dimensional stability, ash content and higher heating value (HHV). Parameters tested for steam treatment process included the combination of temperatures 180, 200 and 220 °C and durations of 3, 6 and 9 min. Results showed that when the severity of steam pre-treatment increased from 2.83 to 4.49, most of the qualities except HHV and ash content were favorable for steam pretreated materials. The pellet density of pretreated sawdust in comparison to raw sawdust resulted in 20% improvement (1262 kg/m³ for pretreated material compared with 1049 kg/m³ for non-treated material). Another important factor in determining the best pellet quality is tensile strength, which can be as high as 5.59 MPa for pretreated pellets compared with 0.32 MPa for non-treated pellets. As a result, transportation and handling properties can be enhanced for steam pretreated biomass pellets. After optimization, the selected treatment was analyzed for elemental and chemical composition. Lower nitrogen and sulfur contents compared with fossil fuels make steam pretreated pellets a cleaner option for home furnaces and industrial boilers. High-quality pellets were produced based on optimized pre-treatment conditions and are therefore suggested for bioenergy applications.

1. Introduction

There are concerns regarding environmental impacts of power generation plants [1,2,3,4]. This is due to the fact that conventional technologies for electricity generation involve extensive use of fossil fuels (e.g., coal) to fire large centralized power plants, thus making electricity generation a major contributor to carbon dioxide (CO₂) emissions. In fact, studies dating back to the early 1990s have shown that electric power generation was responsible for almost 34% of global CO₂ emissions, leading to a growing demand for cleaner sustainable, ecofriendly, and economically viable sources of energy in order to decarbonize the power generation sector. Bioenergy (energy derived from the processing of extant biomass) is promising among many different forms of renewable energy since it is close to carbon-neutral compared with the combustion of fossil fuel resources [5]. Forest biomass is regarded as the primary source of sustainable energy from which wood-based fuel is expected to make up approximately 9% of the electricity used by 2030, according to the US Department of Energy. This can even lead to more exploitation of wood biomass [6].

Wood waste as by-product of wood processing industry is massively available around the world. For example, the amount of wood residue in Canada is approximately 8.7 million cubic metre (M m³) per year, which can be easily used directly or converted to solid fuel [7]. The major producers of wood residues around the globe are shown on Figure 1, China is leading with 104 M m³ wood waste followed by Brazil, the United States, Sweden and Russia [7]. According to Ghafghazi et al. [8], the whitewood residue (e.g., sawdust or shavings) and hog fuel (rejected coarse wood and bark) for Saskatchewan’s industry are estimated at 59.95 and 31.66 kilotonne (kt), respectively. Apart from the annual excess waste from sawmills and pulp mills, several historic piles of wood have been collected in the last two decades. As much as 2.9 Mt is estimated to be available as stored piles that could be exploited as a significant energy source [9]. These numbers suggest that wood-based fuels have a significant potential to partially replace fossil-based fuels such as coal that are currently important but non-renewable energy sources in the province of Saskatchewan, and more broadly in Canada and worldwide.

Biofuel can be in the form of solid, liquid, or gas, and is produced from organic matter either plant or animal. Some examples of second-generation biofuels are biodiesel and bioethanol produced from lignocellulosic material through biological and chemical conversion processes [10]. Biofuels can be obtained from different plant parts and the process for converting the plant biomass defines the particular biofuel that can be delivered. Typical feedstock for producing wood pellets for gasification or combustion are sawdust, dry chips and planer shavings [11]. Every biofuel can be equivalent to specific amounts of fossil fuel in terms of energy content. During the conversion of biomass to biofuel, some by-products may also be produced with non-ideal characteristic for chemical conversions hence they are rejected. The last step in bioenergy production is using the biofuels to generate electricity, steam, or heat. Additionally, raw biomass can be burned directly in a furnace to produce heat [12]. Advancement of biomass usage regarding finance and carbon economy efficiency can result in modern application forms [13].

Some of the important challenges associated with the combustion of biomass in power plants are: (a) feedstock preparation; (b) storage issues; (c) handling; (d) milling; (e) feeding; (f) variability in combustion performance; (g) fluctuations in overall productivity; (h) deposit accumulation (fouling and slagging); (i) sintering and agglomeration; (j) erosion and corrosion, which decreases the lifetime of equipment; and (k) ash usage problems [12,14,15]. These challenges can be resolved by preprocessing the biomass to improve biofuel for energy purposes. Another practical option is to pre-treat the woody biomass before its use. A wide variety of pre-treatment methods have been described including (a) washing/leaching; (b) comminution (grinding); (c) briquetting; (d) pelletization; (e) steam explosion; and (f) torrefaction [12]. Each of these options present a unique set of advantages and disadvantages that must be carefully evaluated when considering strategies for producing biofuels. Moreover, it is important to consider the entire life cycle of the energy production process, including biomass generation and harvesting, transportation, combustion, and waste (i.e., ash) management. Lu et al. reported that amongst various pre-treatments before pelletization of wheat straw, radiofrequency alkaline and steam explosion pre-treatments were less environmentally friendly compared with torrefaction, microwave alkaline treatment and binder addition [16].

Efficient biomass transport from points of production to the energy-producing plant is a key determinant of the viability of a bioenergy strategy. Most biomass are not energy-dense; therefore, they must be compacted to produce a product (pellet) that can be transported and combusted efficiently. However, biomass must be pretreated for effective pelleting. Steam explosion is already known as a viable pre-treatment option for biological and chemical conversion of biomass with applications such as energy-producing biorefineries, fermentation, and environmentally friendly pulping processes [17,18,19,20,21,22,23]. Steam explosion is also suggested for the pre-treatment of wood biomass before pelletization as it can produce dimensionally stable and durable pellets along with improving bonding features, hydrophobicity and heating value of the wood pellets [6,24,25,26,27]. A study by Graham et al. revealed that unlike raw pellets which needed a fully sealed settings for safe storage, steam-treated pellets could be kept in a simple covered storeroom with less mechanical deterioration [28]. Steam explosion convert raw biomass to chemically and physically consistent material which can be exploited at thermochemical and bioconversion plants [29].

Steam explosion treatment involves exposing biomass to pressurized steam for a defined time period (seconds to several minutes), followed by rapid depressurization. Steam explosion disintegrates wood into its three main components (cellulose, hemicellulose and lignin) and increases the susceptibility of cellulose to enzymatic attack [30,31]. The key operating factors of steam pre-treatment are the particle size distribution of raw material (d_p), reaction temperature (T), applied reaction pressure (P) and residence time (t). Different sets of the process conditions result in significant changes in the composition and properties of the biomass [32]. The severity level of steam explosion could affect the amount of volatile organic compounds (VOCs) such as aldehydes, terpenes and furans which were generated from the mild to severe treated softwood bark [33]. Hence, reduced amounts of VOCs at various production steps are recommended for safety concerns [34].

Recent research has concentrated on pre-treating low-quality biomass by steam explosion to improve pellet properties compared with solid fossil fuels such as coal. Shahrukh et al. modelled the net energy ratio of pellets from raw and steam-treated biomass and it was reported that pellets produced from steam-treated wood needed less energy for compression and removal from the die compared with pellets made from non-treated wood [35]. Lam indicated that steam explosion treatment of wood produces a material with desirable fuel qualities (high calorific value, low moisture uptake) but producing durable pellets needed approximately 12–81% more energy than non-treated feedstock [6]. Biswas et al. investigated the pelletizing characteristics of steam-treated Salix and reported quality improvement after the pre-treatment [36]. In their study, steam explosion caused the level of alkali metals in the biomass to decrease, and the compressed pellets had higher density, abrasion and impact resistance. Furthermore, after steam treatment, carbon content and calorific value increased while oxygen amount dropped [36]. Tang et al. investigated the various steam treatment methods for producing pellets and reported particle size reduction in poplar as a result of lignin softening and explosive decompression at high temperatures. Additionally, steam-treated poplar via autocatalysis prevailed over other pre-treatment methods to produce durable and high energy content pellets [37].

There are some favorable logistic improvements of steam pretreated pellets over non-treated pellets in the literature. Niemela et al. compared fuel properties of low grade non-treated and steam pretreated spruce bark with top-quality industrial wood pellet. They showed that pellets produced from steam pretreated spruce bark were more hydrophobic and generated less dust in the handling practices [38]. Similarly, Abelha and Cieplik investigated various handling and storage environments by comparing steam pretreated pellets with white wood pellets and coal; their results revealed that the steam pretreated pellets generated low amounts of dust after long weathering, thus restricting the explosion possibility [39]. Steam explosion may, therefore, offer the possibility of exploiting currently unused and underused biomass sources. To the best of our knowledge, studies addressing alternative use of sawdust for biofuel applications are lacking in the literature. Hence, this research assessed process parameters for producing high-quality pellets from steam explosion pretreated sawdust, using pellet physical and mechanical properties as output variables. In addition, the effect of steam explosion pre-treatment on the quality of wood pellets for fuel and energy applications was determined.

2. Materials and Methods

The methodology presented in this study includes several steps: Feedstock in the form of sawdust (spruce shavings) was delivered to a preprocessing plant (Figure 2). Then, pre-treatment via steam explosion (1.034 to 3.447 MPa) was applied to biomass. Next, the resultant wet material was dried to reach a moisture content of 8%. Densification of the steam explosion pretreated material is required to avoid moisture uptake as well as improve transportation and handling.

2.1. Feedstock Preparation

Mixed sawdust was collected from the sawmill yard of NorSask Forest Products Inc. (Meadow Lake, SK, Canada). Other wood biomass such as pine and spruce sawdust which were already available at the Bioprocessing Lab (University of Saskatchewan) were used for comparison purposes. The moisture content of wood sawdust was determined by measuring the mass loss after drying the samples at 105 °C for 24 h. Unless otherwise indicated, the percentage of all other chemical constituents in wood was measured based on moisture-free wood.

2.2. Steam Explosion Experiment

The lab-scale steam explosion unit installed at the Canadian Feed Research Centre (CFRC), North Battleford, SK, Canada was used in this study. The unit includes a batch reactor with an electric actuated ball valve for steam discharge and a boiler for generating pressurized steam. Approximately 1 kg of mixed sawdust at 17% moisture content (wb) was loaded into the reactor. A charge of pressurized steam at equivalent temperatures of 180, 200 and 220 °C was introduced to the reactor for a duration of 3, 6 and 9 min. The pretreated material was unloaded from the high-pressure reactor into a container at ambient conditions (1.0 bar, 25 °C) by the swift opening of the ball valve. The steam explosion pretreated wood sawdust was allowed to cool in the container and then put in plastic bags. The material was air-dried for at least 3 days and then stored at room temperature prior to pelletization.

A severity factor for the level of steam treatment developed by Overend et al. [32] is shown in Equation (1). The factor is a function of temperature and duration of steam explosion process. This severity factor is commonly applied to steam treatment optimization.

$${\mathrm{R}}_0=\log \int \exp \left[\frac{\mathrm{T}-100}{14.75}\right]\mathrm{dt}$$

(1)

where R₀ indicates the severity of the reaction, T is the reaction temperature (°C), and t is the residence time (min).

2.3. Physical and Chemical Properties of Pretreated and Non-Treated Samples

The bulk densities of pretreated and non-treated samples were estimated and measured using a standard cylindrical steel container with half-litre volume (SWA951, Superior Scale Co., Ltd., Winnipeg, MB, Canada). The particle densities of the treated and control samples were assessed. Sawdust of known mass was placed in the gas multi pycnometer (QuantaChrome, Boynton Beach, FL, USA) and the volume of the sample was determined. Hence, the particle densities were calculated by mass per unit volume of the samples according to the method described by [40].

The particle size analysis of the sawdust was performed before pelletization. The geometric mean diameter (d_gw) and geometric standard deviation of particle diameter (S_gw) of samples were evaluated following EN 15149-2 standard for solid biofuels (BSI. DD CEN/TS 15149-2:2006). The test of normality including three indicators of Shapiro–Wilk, Skewness and Kurtosis were performed for various samples retained on the sieves [41]. Calorific value and chemical composition analysis of biomass samples based upon standard procedures were conducted at the University of Saskatchewan. The neutral detergent fiber (NDF) and acid detergent fiber (ADF) along with lignin were measured according to ANKOM methods 5, 6 and 8, respectively [42]. The cellulose content was calculated by subtraction of lignin from ADF. Hemicellulose content is the difference between ADF and NDF [43]. Total ash contents were determined based on National Renewable Energy Laboratory (NREL) standard [44].

2.4. Densification

Pelletization of steam explosion pretreated and non-treated sawdust are critical as it increases the bulk density and consequently the energy density value. Pelletization also facilitates handling, transportation and storage of material [26]. The pretreated and non-treated sawdust samples were pelletized under the compression load of a single pelleting unit comprising of a plunger-cylindrical die linked to a computer that interprets and records the force-displacement curve data. The plunger was attached to the Instron universal machine (Model 3366R4848 Instron Corp., Norwood, MA, USA) in which the upper moving head provided the load to compress the samples.

A parameter that affects pellet durability colours the temperature of the biomass sample before densification. The temperature applied for preheating is often within the range of lignin’s glass transition temperature (T_g). Among the basic elements in the cell wall of the biomass, lignin compared with cellulose and hemicellulose has the lowest T_g (140 °C in dry settings and between 60 and 100 °C with 8–10% moisture) [45]. Hence, the material was conditioned to reach a moisture content of 8% (d.b.) and the die temperature was set at 95 °C for preheating the biomass sample.

Approximately 0.5–0.55 g of pretreated or non-treated sawdust was fed into the die cylinder and a predefined load (4000 N) compacted the sample corresponding to a compression pressure of ~110 MPa. Compaction pressure and residence time at that pressure significantly impact the pellet density and its features [6]. According to Li and Liu [46] at room temperature, wood sawdust needs a compaction pressure of 100 MPa to create high-quality logs; However, when compaction pressures were boosted to 138 MPa, there was no additional increment in log quality. The plunger compacted the wood material using a crosshead speed of 50 mm/min. When the pre-set load was reached, the plunger stopped and remained still for 60 s to prevent the spring back effect of the sample [43,47]. Six pellets were made from each pretreated and non-treated materials. The force-deformation information at compression and force-time data at stress relaxation were recorded. The physical properties of the densified pellets such as tensile strength, pellet density and dimensional stability were measured to assess the effect of the pre-treatment combinations. As the number of produced pellets by Instron is limited to measure the bulk density of pellets, this quality along with durability were not performed in this study.

2.5. Pellet Density and Dimensional Stability

The dimensions (length and diameter) and mass of each pellet from pretreated and non-treated samples were determined right after pelleting utilizing a digital caliper to estimate the volume and pellet density. The pellets were stored in plastic bags at room temperature at both stages for subsequent analysis. After 14 d, the height, diameter and mass of the pelleted samples were determined to estimate the dimensional stability of the resultant pellets [47] according to Equation (2).

$$\mathrm{Dimensional} \mathrm{Stability}=\frac{{\mathrm{Vol}}_{14}-{\mathrm{Vol}}_0}{{\mathrm{Vol}}_0}\times 100\%$$

(2)

where Vol₁₄ is the volume of pellets after 14 days (mm³) and Vol₀ is the volume of pellets directly after pelleting (mm³).

2.6. Pellet Tensile Strength

The diametral compression experiment as described by Tabil and Sokhansanj [48] and Kashaninejad et al. [49] was used to evaluate the tensile strength of pretreated and non-treated wood pellets. The pellets were cut diametrically into specimens of thickness approximately 2 mm via a laser cutting machine. The single-cut pellet specimens were put at the middle of a padded platen affixed to the Instron universal tester and compressed with the upper plunger until pellet failure occurred. The Instron was fitted with a 5000 N load cell and the specimen was pressed at a crosshead speed of 1 mm/min. The broken specimens split into halves and failure took place along the axis [47,50,51]. Ten replicates were made for each sample. The fracture force was recorded, and the tensile strength was estimated from Equation (3).

$${\mathsf{\delta }}_{\mathrm{x}}=\frac{2\mathrm{F}}{\mathsf{\pi }\mathrm{dl}}$$

(3)

where δ_x is the tensile strength (horizontal) stress (Pa); F is the load at fracture (N); d is the specimen diameter (m); and l is specimen thickness (m).

2.7. Statistical Analysis

To analyze statistically the effect of temperature (°C) and residence time (min) on fuel and pellet properties, a central composite design (CCD) through analysis of variance (ANOVA) was performed [52]. The polynomial quadratic equation proposed to fit the resultant data estimating the impact of independent variables on the total responses is given in Equation (4).

$${\mathrm{y}}_{\mathrm{n}}={\mathsf{\beta }}_0+{\displaystyle \sum }_{\mathrm{i}=1}^2{\mathsf{\beta }}_{\mathrm{i}}{\mathrm{X}}_{\mathrm{i}}+{\displaystyle \sum }_{\mathrm{i}=1}^2{\mathsf{\beta }}_{\mathrm{ii}}{\mathrm{X}}_{\mathrm{i}}^2+{\displaystyle \sum }_{\mathrm{i}=1}^2 {\displaystyle \sum }_{\mathrm{j}=\mathrm{i}+1}^2{\mathsf{\beta }}_{\mathrm{ij}}{\mathrm{X}}_{\mathrm{i}}{\mathrm{X}}_{\mathrm{j}}+\mathsf{\epsilon } \left(\mathrm{n}=1, 2, 3, 4, 5, 6, 7\right)$$

(4)

where n = 1, 2, 3…, 7 is the number of response variables. The independent variables are X₁ which is the temperature (180, 200 and 220 °C) and X₂ which is the residence time (3, 6 and 9 min). The response variables y_n are tensile strength (MPa), pellet density (kg/m³), ash content (%), dimensional stability (%), particle density (kg/m³), bulk density (kg/m³) or higher heating value (HHV, MJ/kg). The other terms in this equation include β₀ (the offset term), β_i (the i linear term), β_ii (the quadratic term) and β_ij (the ij interaction term). By application of the Design Expert software (Stat-Ease, Inc., Minneapolis, MN, United States) the response surfaces of the variables in the experimental design domain can be completely analyzed [53].

2.8. Thermochemical Analysis of Biomass

Proximate and ultimate analyses were carried out to determine raw and pretreated sample properties. The samples selected for the analysis were based on high pellet qualities (e.g., tensile strength, dimensional stability and pellet density) as well as fuel properties (e.g., high HHV and low ash content). The proximate analysis was carried out following the standard analysis methods: ASTM D3175-20 for volatile matter (VM) [54], ASTM D3174-12 for ash analysis [55]. The fixed carbon (FC) was estimated by the difference of both. An automatic oxygen bomb calorimeter (Parr Instrument Company, Moline, IL) was used to measure the HHV of the non-treated and pretreated wood at the University of Saskatchewan and the results were reported in MJ/kg. ASTM Standard D5865-19 test method for HHV was used as a reference for heating value testing [56]. The carbon (C), nitrogen (N), hydrogen (H) and sulfur (S) contents were assessed by an Elementar Vario EL II (Vario Macro, Haan, Germany), whereas the oxygen content was measured by the difference between total mass and mass of C, N, H, S and ash. Proximate and ultimate analyses were repeated twice [57].

2.9. Mass and Energy Yields of Biomass

To study the efficiency of steam explosion pre-treatment, the mass and energy yields were calculated on a dry basis (d.b.) and defined by Equations (5) and (6), respectively [58].

$$\mathrm{Mass} \mathrm{Yield} (\%)=\frac{\mathrm{Mass} \mathrm{of} \mathrm{pretreated} \mathrm{biomass}}{\mathrm{Mass} \mathrm{of} \mathrm{raw} \mathrm{biomass}} \times 100$$

(5)

$$\mathrm{Energy} \mathrm{Yield} (\%)=\mathrm{Mass} \mathrm{Yield} \frac{\mathrm{HHV} \mathrm{of} \mathrm{pretreated} \mathrm{biomass}}{\mathrm{HHV} \mathrm{of} \mathrm{raw} \mathrm{biomass}} \times 100$$

(6)

3. Results and Discussion

3.1. Design Parameters

In

Table 1. Mean values of physical and mechanical properties of steam-treated vs. non-treated spruce sawdust and resultant pellets.

				Biomass					Pellet
Run	Temperature (°C)	Time (min)	Severity Factor	Bulk Density (kg/m3)	Particle Density (kg/m3)	Ash Content (% d.b.)	HHV 1 (MJ/kg)	M.C. (% d.b.) 2	Pellet Density (kg/m3)	Tensile Strength (MPa)	Dimensional Stability (%)
0	-	-		115.4	1420	0.23	18.44	4.40	1049.26	0.32	2.63
1	180	3	2.83	144.6	1450	1.22	17.26	9.47	1137.63	0.63	0.05
2	180	6	3.13	138.8	1480	1.30	17.33	12.76	1169.03	0.65	1.56
3	180	9	3.39	137.4	1440	0.61	17.62	8.72	1149.29	1.50	1.02
4	200	3	3.42	135.2	1440	0.43	17.67	7.89	1165.04	1.52	0.76
5	200	6 (1)	3.72	133.6	1430	0.48	18.03	3.87	1267.91	4.91	1.56
6	200	6 (2)	3.72	138.1	1440	0.44	17.92	5.08	1269.61	2.82	2.95
7	200	6 (3)	3.72	132.6	1460	0.41	17.99	2.33	1266.97	4.12	2.54
8	200	6 (4)	3.72	133.5	1430	0.32	17.94	3.68	1245.28	5.59	2.18
9	200	6 (5)	3.72	137.9	1420	0.38	17.76	5.96	1241.10	3.49	2.52
10	200	9	3.90	132.8	1450	0.43	17.99	5.24	1257.31	3.01	2.43
11	220	3	4.01	132.3	1450	0.40	17.48	21.24	1249.58	3.66	3.51
12	220	6	4.31	150.8	1480	0.40	17.70	18.54	1263.26	3.54	2.79
13	220	9	4.49	149.1	1500	0.40	17.93	19.65	1262.51	2.80	2.64

¹ Higher heating value, ² Moisture content after steam explosion treatment.

, the seven different responses are listed for all the experimental conditions. Non-treated samples showed lower bulk density, pellet density, tensile strength and ash content compared with steam pretreated sawdust. On the other hand, HHV, particle density and dimensional stability were not extensively affected by the steam explosion pre-treatment. Major improvements in pellet densities were observed in pretreated sawdust compared with non-treated sawdust (1262 kg/m³ for treated material compared with 1049 kg/m³ for non-treated material) under extreme conditions (220 °C and 9 min). Based on Adapa’s results, the unit densities were relatively lower for agricultural pellets such as barley, canola, oat and wheat which were steam exploded at 180 °C for 4 min (1169, 1163, 1165 and 1171 kg/m³, respectively) [41]. When comparing the density of non-treated and steam exploded poplar pellets, the density of steam explosion pretreated pellets was 1226 kg/m³, which is remarkably higher than the density of non-treated pellets (1086 kg/m³) [59]. Steam explosion treatment improved the density of corn stover pellets from 1012 kg/m³ to 1280 kg/m³ [27]. Lam, investigated the characteristics of steam exploded pellets and reported that steam explosion pre-treatment resulted in durable pellets [6].

One test to determine the strength of the pellets is tensile strength, which showed promising results for steam explosion pretreated pellets compared with the control (approximately 17 times stronger); e.g., 5.59 MPa tensile strength for stream pretreated pellets in comparison with 0.32 MPa for that of non-treated pellets (Table 1). According to Verhoeff, a decrease in particle size will improve pellet density and consequently the strength of pellets will be enhanced [60]. In this study, the steam pretreated material contained finer particles than the non-treated feedstock. The particles were interlocked with each other after densification (Figure 3); however, more rigid and dense pellets were produced by steam pretreated feedstock (Figure 3B) compared with the raw material (Figure 3A).

Startsev and Salin reported that after the steam pre-treatment, wood fibers were defibrillated, and their surfaces were chemically altered [61]. During hot pressing via polycondensation, the lignin–cellulose chains create new chemical linkages that promote binding features and hydrophobicity of the composite. Furthermore, polycondensated fibers have a lower T_g (between 60 and 70 °C) compared with the raw materials. This helps to explain the improvement of mechanical properties of steam explosion pretreated pellets.

Comparing bulk densities of steam pretreated sawdust with those in the literature such as steam exploded agricultural biomass, it was observed that at similar geometric mean diameter (<1 mm), agriculture residues (such as barely, canola, oat and wheat) have high volume and less bulk density compared with wood [41]. The same conclusion can be drawn for particle density since disintegrated fibers are generated as a result of steam explosion. This quality impacts the bulk density of pellets as well as their combustion performance (e.g., burning period, heat conductivity, and degasification rate). According to Lam, compared with raw Douglas fir, the relative bulk densities of steam pretreated samples varied from 1.25 to 2.51, whereas those of tapped samples ranged from 1.12 to 2.23 [6]. Steam explosion improved bulk density of spruce barks by 17% and their lower heating values (LHV) by 8% [38].

Relaxed density (pellet density after 14 d) for steam pretreated pellets was not substantially distinct from immediate density. Due to shrinkage of pretreated pellets during the storage period, pellet volume decreased; thus, dimensional stability had negative values. However, the non-treated pellets showed lower densities after two weeks of storage resulting in positive dimensional stability values.

ANOVA for response surface reduced quadratic model showed bulk, and pellet densities, dimensional stability as well as HHV were significantly affected by the different conditions of steam explosion pre-treatment (p < 0.05); while other qualities such as particle density, and tensile strength were not impacted by time and temperature during pre-treatment (see Supplementary Tables). Interestingly, process conditions influenced ash content values according to the ANOVA results table. However, the lack of fit was significant, and this model may have some bias at predicting data based on real experimental results. This is also evident in the three-dimensional image of ash value vs. time and temperature (see Supplementary Figure S6). The ANOVA results along with the coefficient of determination (R²) revealed that the response surface model could predict the responses such as bulk and pellet densities, dimensional stability, ash content and HHV successfully. However, other qualities were not predictable by the suggested quadratic model.

3.2. Optimization

Based on HHV and other important pellet features such as pellet density and tensile strength, the importance of the resultant responses was ranked from 1 to 5 (with 5 indicating top priority). Given the high importance of HHV (5) and lower magnitude to pellet density and ash (4 and 3, respectively), other qualities such as bulk density and dimensional stability received little or no preference for the optimization process (2 and 1, respectively). According to

Table 2. Selected optimized treatment conditions and equations for optimized response variables and values.

Variable	Value/Equation	R2	Optimized Value
Selected optimal conditions
(A) Temperature (°C)			215.13
(B) Time (min)			9
Bulk density (kg/m3)	135.52 + 1.9A + 1.2B + 6AB + 8.30A2 − 2.49B2	0.78	144.96 ± 3.80
Particle density (kg/m3)	1440.89 + 7.58A + 10.05B + 15.46AB + 26.02A2 − 2.30B2	0.70	1480.98 ± 15.54
Pellet density (kg/m3)	1253.95 + 53.23A + 19.47B + 0.31AB − 27.25A2 − 32.22B2	0.88	1266.12 ± 22.83
Tensile Strength (MPa)	3.91 + 1.20A + 0.25B − 0.43AB − 1.13A2 − 0.96B2	0.70	3.13 ± 1.09
Dimensional stability (% d.b.)	2.51 + 1.16A + 0.16B − 0.00AB − 0.18A2 − 0.81B2	0.78	2.5 ± 0.66
Ash content (% d.b.)	0.43 − 0.32A − 0.10B + 0.15AB + 0.33A2 − 0.08B2	0.89	0.32 ± 0.13
Higher heating value (MJ/kg)	19.46 + 0.16A + 0.20B + 0.02AB − 0.36A2 − 0.02B2	0.91	19.57 ± 0.11

, the chosen pre-treatment holds high desirability of 0.88 for all equally weighed responses, which is promising for the respective pre-treatment conditions (temperature: 215 °C and time: 9 min).

Along with the increasing severity of the process (e.g. high temperature and long residence time), the desirability for optimized treated material was predicted to improve as shown in Figure 4. At nearly the maximum time and temperature, the highest quality sawdust pellets were predicted to be produced (Figure 3).

3.3. Particle Size Analysis

Pine sawdust has longer fibers compared with spruce and mixed sawdust (

Table 3. Geometric mean diameter (d_gw) and geometric standard deviation (S_gw) of non-treated and treated wood particles.

Sample	dgw (mm)	Sgw (mm)
Pine sawdust	1.43	1.13
Spruce sawdust	0.92	0.94
Sawdust (mix)	0.33	0.20
Steam pretreated (Spruce sawdust)	0.39	0.35

). Steam explosion pre-treatment substantially decreased the geometric mean diameter (d_gw) of spruce sawdust from 0.92 to 0.39 mm. It was concluded that fibers were broken into small particles due to the steam explosion process, which acts to weaken and decompose the structure of fibers. This is consistent with the findings of Sokhansanj et al. [62], and Adapa [41]. According to Lam, both analytical approaches showed fragmentation of big particles throughout steam explosion [6]. This is due to devolatilization during steaming at high pressure. The particle’s inflation due to the release of volatiles throughout the steam conditions with a severity factor of 3.94 was also observed [6].

The mixed sawdust and steam pretreated material consisted of very fine particles of which 80% were <1 mm (Figure 5). The hygroscopicity of finer particles was relatively higher [6]. For instance, the moisture content of steam exploded spruce sawdust (at 220 °C for 9 min) containing finer particles was much higher (19.65% d.b.) compared with non-treated sawdust with larger particles (4.4% d.b.) (see Table 1). Since bigger particle sizes (more than 1 mm) function as predefined breaking points in the resultant pellet, the ideal particle size ranges from 0.5 to 0.7 mm [63]. In practice, a blend of various particle sizes provides the ideal pellet durability since they have stronger inter-particle binding and fewer void spaces between particles [64,65,66].

Pellet properties were reported to improve with normal particle size distribution of biomass fibers [67]. Hence, the diameters of the particles should have close to zero skewness as well as negative kurtosis statistics. According to

Table 4. Normal distribution indicators of fibers.

Sample	Skewness	Kurtosis	Shapiro–Wilk	p-Value
Pine sawdust	0.875	0.153	0.866	0.139
Spruce sawdust	0.128	−0.898	0.95	0.709
Sawdust (mix)	1.494	1.622	0.795	0.025
Steam-treated (Spruce sawdust)	1.504	1.292	0.784	0.019

, spruce fibers are ideal for the compaction process compared with other feedstocks. Additionally, the Shapiro–Wilk test was significant (p > 0.05) for pine and spruce particles, which also indicated the normal distribution of fibre particle sizes.

3.4. Chemical Composition

After applying pressurized steam to spruce sawdust, all chemical components except ash content decreased (

Table 5. Chemical composition of pretreated and non-treated sawdust.

Components	Sawdust (Mix)	Spruce Sawdust	Pine Sawdust	Steam Pretreated (Spruce Sawdust)
Ash (% d.b.)	0.230	0.101	0.047	0.400
Lignin (% d.b.)	25.167	25.332	25.565	20.447
Hemicellulose (% d.b.)	14.470	15.413	13.759	1.045
Cellulose (% d.b.)	50.389	51.016	45.544	36.271
Solubles (% d.b.)	9.74	8.14	15.08	41.84

). The decrease was evident for hemicellulose and cellulose which were 14 and 15%, respectively. This contrasts with findings for agricultural straws in which cellulose content of barley, canola, oat and wheat were increased after steam treatment [41]. Hemicellulose showed a similar pattern for canola, barley, and oat straws, while wheat hemicellulose decreased after pre-treatment [41]. These results are consistent with observations made using steam exploded poplar, in which the apparent lignin content increased after steam explosion at 200–205 °C [68]. Only a small fraction of lignin (5%) was dissolved and mostly holocelluloses are recoverable with other chemical substances in the downstream processing such as integrated biorefinery and bioenergy plants. The huge amount of non-structural solubles in dissolved liquor (41%) indicates a potential resource for high value end-products such as furfural, hydroxy methyl furfural (HMF) and ethanol from fermentable carbon sources. Tooyserkani et al. suggested that steam pretreated pellets can be used to produce bioethanol [69].

The amount of lignin and hemicellulose as well as their structural configuration impacts pellet strength [70]. According to Lehtikangas, the durability of pellets made from sawdust, logging bark and wastes are highly correlated to the lignin concentration of the biomass [11]. The higher the lignin concentration of the pellet, the greater is their durability [11]. During steam pre-treatment, the activated pseudolignin could act as a binding agent between wood fibers and consequently form stronger pellets [29,71]. Pine sawdust contained less ash and cellulose compared with mixed sawdust and spruce sawdust (Table 5).

3.5. Thermochemical Analysis of Treated and Non-Treated Biomass

Carbon and oxygen contents were not strongly affected by steam pre-treatment (

Table 6. Thermochemical properties of raw and treated sawdust.

Sample	HHV 1 (MJ/kg)	M.C. (% d.b.) 2	Proximate Analysis (% d.b.)			Ultimate Analysis (% d.b.)
			Fixed Carbon	Volatile	Ash	N 3	C 4	H 5	S 6	O 7	O/C	H/C
Sawdust (mix)	17.49	4.57	17.4	77.80	0.23	0.06	49.25	6.27	0.02	44.39	0.90	0.13
Spruce sawdust	18.44	4.40	18.6	76.90	0.23	0.03	48.81	6.29	0.02	44.84	0.92	0.13
Pine sawdust	18.70	4.76	19.8	75.40	0.04	0.09	51.52	6.43	0.04	41.91	0.81	0.12
Steam treated (Spruce sawdust)	17.93	5.00	22.9	71.70	0.4	0.07	50.76	6.22	0.04	42.89	0.85	0.12

¹ Higher heating value, ² Moisture content, ³ N: nitrogen, ⁴ C: carbon, ⁵ H: hydrogen, ⁶ S: sulphur, ⁷ O: oxygen.

). This suggests that steam pre-treatment did not have a large impact on the quality of biofuel pellets compared with raw sawdust. Additionally, a slight increase in sulfur content was detected after steam explosion. Sulfur and chlorine can lower the ash melting point, leading to technical issues such as slagging or deposit forming in burners [72]. Due to their corrosive properties, they can also erode the oxide layer from the surface of the burner. For almond tree-derived biomass, C and H contents were 48.11% and 5.77%, correspondingly [73]. The percentages of C and H were within the range of biomass categories that have been reported in the literature (C = 42–71%; H = 3–11%) [74]. He et al. reported that after steam treatment of rice straw at 200 °C for 10 min, the carbon content improved from 42.14% to 46.47%. Although hydrogen and oxygen decreased from 5.60 and 51.45% to 5.35 and 47.40%, respectively [75]. Low nitrogen content (0.07%) in steam exploded samples is favorable in terms of environmental impacts as less nitrogen dioxide is emitted compared with other bio-based fuels (N = 0.1–12%) [76].

There was a small increase in fixed carbon amount of approximately 4% for the steam pretreated spruce sawdust compared with non-treated (

Table 7. Mass and energy yield of steam-treated compared with non-treated spruce.

Temperature (°C)	Time (min)	Severity Factor	Mass Loss (%)	HHV 1 (MJ/kg)	Mass Yield (%)	Energy Yield (%)
Non-treated
-	-	-	-	18.44	100	100
Steam-treated (Spruce sawdust)
180	3	2.83	18.90	17.26	81.10	78.70
180	6	3.13	25.30	17.33	74.70	74.00
180	9	3.31	28.50	17.62	71.50	72.10
200	3	3.42	26.00	17.67	74.00	70.88
200	6	3.72	27.00	18.03	73.00	75.20
200	9	3.90	22.70	17.99	77.30	75.43
220	3	4.01	30.30	17.48	69.70	69.60
220	6	4.31	40.40	17.70	59.60	60.30
220	9	4.49	31.90	17.93	68.10	69.80

¹ Higher heating Value.

). However, ash content increased, which is not a desirable feature of the final pellet product. Similarly, other researchers [75,77,78] reported increased ash content which may be due to removal of other compounds after steam explosion pre-treatment of lignocellulosic biomass. The HHV for steam pretreated sample decreased slightly (4% compared with non-treated spruce), which could be attributed to the near total loss of hemicellulose along with a proportion of cellulose (15%) and lignin (5%) losses compared with the non-treated sample. Demirbas suggested a linear relationship between HHV and lignin amount of biomass [79]. As lignin decreased in steam pretreated sample consequently its HHV decreased as well. In comparison, the HHV of steam-pre-treated agricultural crop residues including barley, canola, oat and wheat straw was 6%, 10%, 9%, and 5%, respectively, higher than non-treated material [41].

The volatile content of spruce sawdust after steam pre-treatment at 220 °C for 9 min decreased from 76.90 to 71.70%. As a result of Lam’s study on steam exploded Douglas fir, volatiles for raw samples compared with treated samples at 200 and 220 °C were 85.4, 84.2 and 78.9%, respectively [6]. The mass loss (20–30%) after steam explosion could be due to unrecovered material stuck to the inside surface of the discharge unit as part of the batch steam explosion unit. On the contrary, this issue is hardly noticeable in a continuous steam explosion unit in a commercial system which is capable of minimizing or removing material sticking to parts of the system. Additionally, the lengthy air-drying step could lead to the removal of volatile constituents of pretreated sawdust. Stahl et al. discovered that long drying time caused volatile loss, emissions of terpenes and reduction in the calorific value of biomass [80]. In another study, Emmel et al. reported that following explosive expansion, the volatile elements could not be properly recovered as a result of partial loss to the environment [30]. The majority of the original wood extractives were lost as volatile substances, while secondary extractives evolved as by-products of structural polymers derived from lignin and carbohydrates [81].

3.6. Mass and Energy Yield

Average severe conditions for steam explosion can result in higher calorific value and less mass reduction after treatment (Figure 6A). There was no general trend between the severity factor and the HHV of steam pretreated sawdust. However, as severity increased, the mass and energy yields dropped from 80% to 60% for R₀ values of 2.83 and 4.31, respectively (Figure 6B).

Formerly, it has been noted that estimating the solid yield of steam exploded biomass is challenging [82]. When the particle size material was smaller, the yield loss of retrieved samples was significantly higher. Approximately 17–26% solid recovery loss (d.b.) for red oak chips after steam treatment with a severity of 4.3–4.54 has been reported [82]. Decomposition of the biomass is initiated at low severity of R₀ less than 2. The degree of cellulose polymerization is substantially reduced for R₀ higher than 3. In high severity reactions (R₀ > 4), the hemicellulose undergoes a dehydration and condensation process; therefore, more soluble sugars will be generated as by-product [6].

Maximum mass and energy yields were achieved in less severe conditions and, as expected, with a harsh steam explosion pre-treatment, the recovered mass decreased (Table 7). The energy loss for higher temperatures was more evident based on the results. Kobayashi et al. found that compressed hot water pre-treatment at 240 °C resulted in yield loss of below 50% for wood residue [83]. Losses are often attributed to inaccuracies in material recovery and volatile losses throughout steam explosion [30,82]. Lam reported the solid yield at a range of 51–84% for steam explosion pretreated Douglas fir. Between steam pretreated samples at 220 °C and 200 °C, the mean value difference for solid yield was statistically significant (p = 0.05) with a lower amounts for higher temperatures [6]. Additionally, Pérez et al. reported a linear relationship between mass loss and process temperature during steam torrefaction of Eucalyptus globulus [84].

4. Conclusions

Steam explosion, like any pre-treatment process, is an energy-consuming procedure. Nevertheless, the advantages offered by steam explosion from the perspective of improving biomass quality may justify the high energy input and related expenses caused by moisture uptake, dust generation and gas emissions during storage and transportation. This study has determined that steam explosion pre-treatment of spruce sawdust dramatically improved the density and tensile strength of pellets compared with non-treated sawdust pellets. The other responses such as dimensional stability, ash content, bulk and particle densities were in the range of raw biomass. After applying severe steam explosion conditions to spruce sawdust, most of the quality attributes were desirable for producing high-quality pellets, although the energy and mass yields were reduced to 68–70%. Chemical composition analysis showed extensive removal of main wood constituents such as holocelluloses and lignin breakage throughout the steam explosion process. A low amount of ash as well as lower nitrogen and sulfur contents relative to other solid fuels on the market, make steam explosion pretreated wood pellets eco-friendly and more beneficial for power generation. Since gross energy values of samples decreased after applying steam, it is advised that the entire process be subjected to technoeconomic analysis to determine its feasibility for industrialization. Considering the carbon footprint of the whole process can help to understand the environmental impacts of steam-treated bio pellets compared with other fossil fuels like coal. It is worth noting that the chemical recovery of dissolved liquor after steam explosion can be of huge importance to wood and paper companies, which can benefit from different by-products originating from this side stream besides the main product.