NC State
Rousset, P., Aguiar, C., Volle, G., Anacleto, J., and De Souza, M. (2013). "Torrefaction of babassu: A potential utilization pathway," BioRes. 8(1), 358-370.


Because of its energy and mechanical properties, babassu shell is a promising energy crop for the future. Its production potential in Brazil is estimated at 6.8 million tons of fruits/year. The aim of this study was to evaluate the effects of torrefaction on the main energy and mechanical properties of Orbignya speciosa (Babassu). Three different torrefaction temperatures (220, 250, and 280°C) and two different durations (15 and 30 minutes) were employed. The influence of sample orientation was taken into account. The results showed that the energy properties of babassu are greatly improved during heat treatment. Torrefaction also led to uniform resistance to babassu shell compression. It was found that low temperature treatment was sufficient to envisage grinding and babassu use in pulverulent form.

Download PDF

Full Article

Torrefaction of Babassu: A Potential Utilization Pathway

Patrick Rousset,a* Clarissa Aguiar b, Ghislaine Volle,a João Anacleto,c and Mario De Souza c

Because of its energy and mechanical properties, babassu shell is a promising energy crop for the future. Its production potential in Brazil is estimated at 6.8 million tons of fruits/year. The aim of this study was to evaluate the effects of torrefaction on the main energy and mechanical properties of Orbignya speciosa (Babassu). Three different torrefaction temperatures (220, 250, and 280°C) and two different durations (15 and 30 minutes) were employed. The influence of sample orientation was taken into account. The results showed that the energy properties of babassu are greatly improved during heat treatment. Torrefaction also led to uniform resistance to babassu shell compression. It was found that low temperature treatment was sufficient to envisage grinding and babassu use in pulverulent form.

Keywords: Babassu; Torrefaction; Energy Yield; Mechanical Properties

Contact information: a: Biomass Energy Unit 42, French Agricultural Research Centre for International Development (CIRAD), Brasilia DF, 70818-900, Brazil; b: University of Brasilia, Brasilia DF, CEP 70910-900, Brazil; c: Laboratory of Forest Product, Brazilian Forest Service (LPF-SFB) Brasilia DF, 70818-900, Brazil; *Corresponding author:


Brazil is one of the few countries in the world that offers major potential for expansion of biomass production and use. Sugarcane bagasse and eucalyptus plantations are two main sources of biomass for energy. However, there are other species that are currently unexploited. Various palms native to Amazonia and other tropical regions of Latin America have been the subject of research and development (R&D) since the late 1970s, with extremely diverse results, extending from total failure to success in the modern market (Clement et al. 2005). More recently, Teixeira (2008) found that babassu palm fruits are a potential source of biomass for energy. He estimated the Brazilian potential at 6.8 million tons of fruits/year (main potential in Maranhão state, 92%). Availability varies from 1.6 million tons of fruits/year on the actual baseline scenario, up to 4.1 if an improved collecting system is used. The same study indicated that at least two of the fruit components have energy-use potential, with very distinctive behaviour for the epicarp (11% of the fruit, LHV of 20,238 and the endocarp (59%, LHV of 21.179 Furthermore, its production is greatly related to the small-scale extractive sector, which is roughly equivalent to wage labour and to agriculture in its contribution to households in rural areas of tropical Brazil (Zylbersztajn et al. 2000).

Efficient conversion technologies may be the key to taking advantage of babassu’s characteristics for various applications in the iron and steel industry or pulverized systems requiring small-sized samples. Among existing technologies to improve biomass for energy production, torrefaction is one of the most efficient means of recovering energy from biomass. Torrefaction is a thermal pretreatment process that enables energy densification of biomass and biomass homogenization. It is a mild pyrolysis process carried out at temperatures ranging from 200 to 280°C under atmospheric pressure in the presence of limited or no oxygen (Van der Stelt et al. 2011). The advantages of torrefaction include removing water uptake properties, eliminating biomass decomposition, reducing grinding energy requirements (Chen et al. 2011), and creating a more uniform fuel for pulverized systems.

The purpose of this study was thus to increase our knowledge of babassu (Orbignya speciosa) torrefaction for energy production. Experiments were carried out in a laboratory scale reactor adapted for small and large samples. We investigated changes in the physical, mechanical, and chemical characteristics of babassu samples when subjected to mild pyrolysis treatment in the temperature range of 220, 250, and 280°C for 15 and 30 minutes. Weight loss was measured during the process. The torrefied babassu was further characterized in terms of its composition and calorific value. Both proximate and ultimate analyses were used to compare torrefied babassu characteristics with other solid fuels.

Extractive activities are important in the Amazon region as inputs to household reproduction, and are critical as a source of cash income. Access to a highly competitive market demands a different product – competing with a similar product demands competence of a high level (Fig. 1). Consequently, the authors of this study hope, through this technical study, to contribute to babassu utilization as a significant source of both use and exchange values.

Fig. 1. Products from the industrialization of babassu nuts (Oliveira et al. 2009)


Material Preparation

Babassu (Orbignya speciosa) biomass is a very abundant palm tree in the northern-central region of the country, especially in the State of Maranhão. The tree is large in size, up to 20 m, having a trunk with a crown containing a number of fruits in ellipsoidal form. Each fruit consists of a thin peel (epicarp) that has a starchy secondary layer (mesocarp), endocarp, and kernel amounting for 11, 23, 59, and 7% of the mass, respectively (Cirad and Brasileiro 2008).The fruits weigh 90 to 280 g (Fig. 2). The babassu nuts used in this study were harvested by a local cooperative in the Maranhão region. Ten fruits were tested for the study. They were cut lengthwise and widthwise. The nomenclature for each sample is given in Table 1. It can be seen that for the low temperature (220°C), only the 30 min residence time was tested. Before undergoing torrefaction, all the samples were previously dried to a constant weight at 103°C.

After torrefaction, the material was tested to evaluate its mechanical properties compared to the control. The samples were then ground and sieved to a maximum particle size of 60 mesh and stored in a desiccator at room temperature pending chemical and physical analysis.

Fig. 2. Transversal section of the babassu (

Torrefaction Protocol

The thermal treatment of babassu nuts was carried out using a laboratory‑scale reactor that measured simultaneously the dynamics of biomass weight loss and the temperature inside the biomass throughout the course of the experiment. A programmable PID (proportional-integral-derivative) controller was used to control the temperatures and heating rate of the reactor by way of three thermocouples: one near the electric heater, one inserted through the lid of the reactor, and the last one inside the sample to monitor its temperature, as described in Rousset et al. (2012). Finally, two samples were tested in each trial run, one coupled to a precision balance to monitor mass loss and the other to a thermocouple to monitor the internal temperature.

The temperature programme, consisting of an isothermal heating period, was applied to perform torrefaction. The temperature was raised from ambient temperature to the appropriate torrefaction temperature (i.e. 220, 250, and 280°C) with a constant heating rate (3°C min-1) and held for 15 and 30 min, followed by a natural cooling period. Thermogravimetric analyses versus biomass temperature were obtained from the recorded distribution of weight and temperature.

Chemical Analysis

Gross Calorific Value (GCV) was determined on the basis of thermal energy generated by complete combustion of the sample in a constant pressure chamber. A PARR 1261 bomb calorimeter was used in compliance with standard NBR 8633/ 84. The energy yield (ƞe) was also determined. It relates the mass yield to the higher heating value (HHV) of wood and is calculated by the formula (Rodrigues and Rousset 2009), where HHVtorr and HHVcont represent respectively the higher heating value of torrefied wood and the untreated sample.


Gravimetric yield (ƞm) is the ratio of torrefied wood mass (MTorr) to initial feedstock mass (MCont), dry at 0%.


A proximate analysis was conducted to determine the fraction of ash (Ash) and volatile matter (VM). The fixed carbon (fC) content was obtained by difference. The values of the proximate analysis were obtained in compliance with the procedure of the Brazilian Association of Technical Standards ABNT NBR 8112/86. The ultimate analysis provided the composition of the biomass in wt.% of carbon, hydrogen, and oxygen (the major components) as well as sulphur and nitrogen. The ultimate analysis values were obtained in compliance with the European procedure XP CEN/TS 15104.

Specific Surface and Pore Size Distribution

The quality of activated carbons is evaluated in terms of their physical properties of adsorption and of superficial area, using different analytical methods for liquid and for gas phase adsorption. This property is very important for use in metal oxide reduction in an industry such as steelmaking. Of all the raw materials existing in Brazil, babassu is probably the most typical and traditional raw material for the preparation of activated biomass. Adsorption-desorption isotherms for nitrogen adsorption onto carbon were acquired using a Micromeritics ASAP 2010 instrument. The samples were previously outgassed at 200°C for 2 h. Specific surface areas were calculated by the multi-point Brunauer-Emmett-Teller (BET) method at relative pressures of P/P0 = 0.06 to 0.3. The pore size distributions were obtained from the desorption branch of the nitrogen isotherms using the Barrett-Joyner-Halenda (BJH) method. The sample weight was around 1 g.

Mechanical Analysis

Compressive static tests were conducted using a universal testing machine, Instron model 1127, and the maximum load (ML) until crushing was determined. In this test the specimen was compressed over its surface area at a speed of 0.3 mm/min. Figure 3 depicts an example of the test carried out with babassu heated to 280°C for 30 min. A virtually linear increase in load was found up to shell breakage at around 800 kgf, corresponding to the maximum load that babassu shell can withstand.

Fig. 3. Compressive static test with babassu (1B2) in the longitudinal section treated at 280°C for 30 min. Compression speed: 0.3 mm/min. 1 kgf = 9.8 N

Statistical Analysis

Statistical analyses were conducted using a specific software product, XLSTAT (addinsoft), to enhance the analytical capabilities of Excel. Eight variables in response to the experiments (temperature, time, and section) were analysed and discussed: the mass loss (Wt%), the fixed carbon content (FC%), the gross calorific value (GCV), the compression value (Cv), the carbon content (C%), the hydrogen content (H%), the nitrogen content (N%), and the oxygen content (O%).

Untreated and treated nuts were subjected to a variance analysis (ANOVA) and the Tukey test at a probability of 5%. Tukey’s test is one of the main applications of an ANOVA to check whether or not the parameters for the various categories of a factor differ significantly. It is a single step multiple comparison procedure and a statistical test to find which means are significantly different from each another. The test compares the means of each treatment to the means of every other treatment, i.e. it applies simultaneously to the set of all pairwise comparisons and identifies where the difference between two means is greater than the standard error would be expected to allow.


Overall Results

Table 1 gives the averages obtained for the eight variables studied considering two replicates per treatment; gravimetric yield (ƞm), fixed carbon content (fC), gross calorific value (GCV), compression value (Cv), and ultimate analysis (C, H, and O).

Table 1. Average of Duplicate Samples for Five Response Variables Considering Two Replicates per Treatment

Dir.:L/T: Longitudinal/Transversal section; Control = Untreated

Means followed by the same letter are not significantly different (α = 5%)

Type III sum of squares analyses were carried out to determine which parameters interfered significantly (α = 0.05) with four response variables, along with the effect on the torrefied babassu properties (Table 2). Of the parameters studied, temperature and duration were the factors with the greatest impact on the four response variables. Sample orientation (L or T) did not influence the variables.

Mass Loss

Figure 4 shows the mass loss of heat-treated babassu nuts. As expected, the mass loss increased with temperature and time of torrefaction. Exposure of wood to tempera-tures of >150°C during drying may cause thermal degradation of its structure, often accompanied by mass loss. Devolatilization and carbonization of hemicelluloses, deploy-merisation, and devolatilization/softening of lignin and depolymerization and devolatiliz-ation of cellulose are observed. The degree of structural damage depends on the tree species as well as the process parameters, such as treatment duration, temperature, and relative humidity. All structural changes observed in the biomass due to moisture loss influence its mechanical properties, such as bulk density and grindability. The lowest gravimetric yields (ƞm) were obtained at the highest temperatures or treatment times. For instance, it was found that for a treatment time of 30 minutes at 250°C, mass loss was equivalent to that obtained at 280°C for 15 minutes with 63.5 and 63.9%, respectively. These results confirm those obtained in the literature (Rousset et al. 2011).

Table 2. Analysis of Variance of Temperature (T), Duration (D), and Direction (L/T) along with their First Order Interactions for the Four Response Variables Analyzed *

*Significant at 5%; ns = not significant at 5%; P-values are in italics

Fig. 4. Thermogravimetric analysis for each treatment

Despite some more pronounced differences for the high temperatures, up to a difference of 10%, the orientation of the sample did not have any significant impact on this variable, as shown by the variance study (Table 2). Indeed, the Direction factor (L/T) did not seem to be significant (Table 1), as the only major difference between L and T seemed to occur between 2A and 2B. This was not enough to deduce statistical significance. However, the significant difference was induced by the duration factor D (15/30 min). The figures show that when heating for 15 min, the gravimetric yield ƞm values were always greater for direction L and the opposite when heating for 30 min. Figure 5 gives the changes in mass loss depending on the two parameters “duration and direction”. The explanation may come from heat transfers within the sample that were facilitated in the direction of the fibres at low temperature.

Fig. 5. Average weight loss depending on time for each of the directions, L and T

Energy Characteristics

Table 3 summarizes the descriptive statistics for four variables: ash, volatile matter, fixed carbon, and gross calorific value for the set of treatments. These variables were highly correlated to the model for the third interaction T×D×Dir with a P-value <0.0001 and a coefficient of determination over 0.99.

Table 3. Descriptive Statistics for the Four Variables Analysed

= Temperature; = Duration; Dir = Orientation of samples (L/T)

Figure 6 shows the energy yield as a function of gravimetric yield. The linear regression model obtained shows that the two parameters were highly correlated with a coefficient of determination over 0.98. These figures tally with the literature (Pierre et al. 2011). More severe treatments will lead to a displacement of the points to the origin of the graph, up to obtaining properties similar to those of wood charcoal.