Physical treatment of raw material
The pressed EFB fibers used for this study were physically treated to obtain the sieved EFB fibers (Test 2) and the sieved and water-washed fibers (Test 3), prior to size and moisture reduction via pre-pelletizing process. The unwanted materials, less than 25 mm in size, passing through the sieving mesh were dirt including soils and stones, short and pulverized fibers which contributed to approximately 1% (wt/wt) of total raw pressed EFB fibers fed to the sieving drum. The wastewater generated from the washing process of EFB fibers was collected and analysed. Table 3 shows the characteristics of the wastewater generated. It appeared that the wastewater had much lower characteristics, in particular BOD and COD values compared to the raw palm oil mill effluent (POME) and hydrocyclone wastewater typically generated during the palm oil milling process. This finding indicates that it is possible to treat the EFB-washed wastewater using currently practiced POME treatment system if the EFB pellets production plant is part of the integrated activity in palm oil extraction.
Production trials and physical characteristics of raw material and EFB pellets
The feedstocks (untreated and physically-treated EFB fibers) were shredded, dried and pulverized to meet the required size and moisture content as a feeding material for pelletizing process. Moisture content and particle size of the feedstocks are two important parameters affecting pellet durability and bulk density . For efficient pelletisation, many studies have suggested the allowable maximum limit of moisture content and particle size to be < 15% wt.% and 5 mm [31,13].
Table 4 summarizes the characteristics of pressed EFB fibers (raw material) and pulverized EFB fibers employing different physical treatment. The raw material appears to have heterogeneous characteristics, particularly with high variation of moisture content i.e. from 46 wt.% to > 65 wt.%. The pre-pelletizing process employed in this study managed to reduce significantly the moisture and size of EFB fibers to < 14 wt.% and < 10 mm in their pulverised forms, respectively, which are comparable to those in the literature. The bulk density of EFB fibers was therefore increased significantly from 62 to 152 kg m-3, thus able to facilitate better material flow throughout the production process.
The production trials conducted showed that binderless EFB pellets were successfully produced from untreated and physically-treated feedstock. The presence of lignin in EFB fibers, about 18%, acts as a natural binder . The high pressure and temperature applied during the pelletisation had softened the lignin to act as an intrinsic resin for improving the particles binding during pelletisation [44,45]. The size and moisture content of the feeding material were also reduced to a suitable level for effective pelletization to produce EFB pellets of desirable quality without affecting the overall production process. Based on the production trials, it was estimated that about 2.35 – 2.50 t of pressed EFB fibers and approximately 160 kWhr of electricity were required to produce 1 t of EFB pellets. The relatively lower energy consumption in this study compared to those in the literature i.e. > 200 kWhr t-1 pellet [43,9] was due to the use of partially-treated pressed EFB fibers, efficient machineries and biomass gasifier for EFB drying process.
Three types of EFB pellets were produced from this study, namely a typical EFB pellet using untreated EFB fibers (Test 1), and a low-ash EFB pellet from Test 2 and Test 3 using 2 different physical treatments. The insignificantly different physical properties of the resulted EFB pellets from Test 1 to Test 3 in Table 5 indicated that the different treatments applied did not affect the physical properties of the pellets. The EFB pellets were relatively uniform in terms of diameter and bulk density although the varying length could affect the weight of individual pellet. On average, the diameter, length and bulk density of the produced EFB pellets were 8.4 mm, 40.0 mm and 661.7 kg m-3, which complied with the requirement as stipulated in the European (EN ISO 17225-2) and International (ISO 17225-6) standards.
Pellets with high bulk density are preferred to facilitate logistics arrangement, storage and cost reduction. Pelletizing process could increase the bulk density of EFB pellet to > 650 kg m-3, far better than those of the raw and pulverised feeding materials in the range of 62 – 152 kg m-3 (Table 4). Therefore, higher bulk density relates to an increased energy density i.e. more energy per unit volume of the resulted EFB pellets as a solid biofuel . Moisture reduction, downsizing and densification process at the employed pressure and temperature in this study had largely influenced and compacted the loose EFB fibers to a much denser pellet form . The average specific density (1306 kg m-3) was higher than that reported by Tenorio et al.  and within the range required by the German standard, DIN 51731.
Table 5 also shows the durability index of EFB pellets produced. There was no differences in durability index of the pellets produced from untreated (Test 1) and physically-treated (Tests 2 and 3) EFB fibers. The mean value of 94.62% exceeding 80% can be regarded as high and acceptable although the value is slightly lower than those by the European and International standards . Pellet durability index is an important parameter to describe capability of the produced pellets to tolerate shock and vibration during handling and transportation. Moisture content is a major factor affecting pellet durability level. An increasing amount of moisture beyond 14 wt.% causes pellet durability to drop significantly . Other factors affecting pellet durability are particles size and type of pre-treatment of the raw material.
Table 6 shows the proximate and ultimate analyses of the three types of EFB pellets produced from this study. There were significant differences on the moisture and ash contents and the CV of the EFB pellets compared to the raw material (Table 4). Huge improvement was made on moisture content and CV of the pellets, at >79.2% and 65.0%, compared to 46.33 wt.% and 10566 kJ kg-1, respectively from the pressed EFB fibers. The pre-pelletising step to reduce EFB’s size and moisture and the pelletising process to densify the loose and pulverised fibers had proven to improve the two important fuel properties of the EFB pellets as a solid fuel.
The fuel properties of EFB pellets deriving from different physical treatment (Test 1 to Test 3) were relatively comparable except for ash content. The employed Test 2 involving sieving process had managed to reduce the ash content of EFB pellets to 3.63 wt.%, compared to 4.66 wt.% that of an untreated normal EFB pellet. A combined sieving and water washing of EFB fibers (Test 3) had significantly reduced the ash content of EFB pellets to 1.58 wt.%. This means physical treatment of raw material is necessary and should be incorporated at any commercial EFB pellets production line, in order to produce EFB pellets low in ash content for niche markets. As an agro-based by-product, EFB is naturally high in ashes plus easily contaminated by soil and dirt due to poor material handling, storage and transportation either at the mill or the pellet production plant compared to wood-based counterpart . The use of an integrated gasifier system for drying EFB fibers provides smokeless cleaner hot gases compared to existing biomass burner utilised in pellets production site.
Besides creating ash-related problem, a higher ash content also decreases the CV of the biomass pellets . An increase of 1 wt.% of ash content has resulted in CV reduction of 0.11 – 0.2 MJ kg-1 [48,49]. In such situation, more biomass fuel sources are required to generate the same amount of energy. Previous studies reported a 1.18 to 1.39 % higher biomass fuel consumption for every 1 wt.% increase in ash content [48,50]. By doing so, a higher ash production and management cost is anticipated due to frequent cleaning and machinery maintenance. High-ash content solid fuel will also influence NOx emission, in particular agro-based pellets, causing negative environmental impact . Verma et al.  reported that the NOx emitted from burning of straw pellet with 9 – 10 wt.% ash content was 7 times higher than that of wood pellet containing just 0.65 wt.% ash. This phenomenon is due to catalytic effect of the ash present; higher amount of ashes tend to provide more active sites to catalyse N into NOx during the combustion [53,52]. Therefore, minimizing ash content in biomass fuel provides significant advantages from technical, operation, economic and environmental points of view for biomass-based heat and power generation. A combined sieving and washing pre-treatment in this study was able to solve the above-mentioned shortcomings. All the EFB pellets produced exhibited < 10 wt.% moisture content (Table 6). As a result, the EFB pellets had energy value significantly higher than the raw EFB as large initial water content would affect the recoverable heat. Similarly, lower moisture content had contributed to a higher CV leading to better combustion temperature and efficiency.
The elemental contents of the EFB pellets produced (Table 6) and their initial physically-treated feeding materials (Table 4) were insignificantly different but differed significantly from those of the raw material mainly due to high moisture content. The Van Krevelen plot (Figure 2) shows that the H/C and O/C atomic ratios of EFB pellets produced from different physical treatment (Tests 1-3) were closer to those torrefied EFB, coal and lignite  and significantly better than their raw material (pressed EFB fibers). These findings postulate that the pelletizing process largely involved both the mechanical and thermal (drying) processes to improve the physicochemical properties of the raw EFB fibers, in particular the significantly reduced moisture content, leading to an increased C but reduced O contents of the EFB pellets. Like other biomass , the EFB pellets contained > 99 wt.% of O, C and H with the latter two totally ~50 wt.%. CV of a biomass fuel is much dependant on the amount of C and H present where higher contents of these two elements contribute to better energy content. As biomass combustion is an exothermic reaction, combustion of higher amount of C and H with O2 would release more heat while generate carbon dioxide (CO2) and water [55,56]. Comparing the C and H contents of the following: torrefied EFB, 58.89 wt.% and 5.12 wt.% ; coal, 64.34 wt% and 4.06 wt.% ; lignite, 58.5 wt.% and 3.0 wt.%  and pelletized EFB, 42.86 – 43.98 wt.% and 6.38 - 6.63 wt.% (this study), it is thus apparent that though the EFB pellets has a much lower CV amongst the solid fuels, its CV has been greatly improved compared to the raw material used. The results indicated that simple physical removal of moisture coupled with different sieving and washing co-treatment methods were sufficient to improve the CV of EFB pellets which then could perform much better as a fuel compared to normal EFB pellets.
N and S embedded in biomass fuel create undesirable pollutants, namely NOx and SOx during thermal combustion. Their concentration limits are important, as evidenced by the EN ISO 17225-2 (for wood pellets) and the ISO 17225-6 (non-woody pellets) standards demanding for environmental-friendly biomass pellets . The N and S contents in the EFB pellets ranged 0.24-0.54 wt.% and 0.03-0.07 wt.%, respectively are notably and relatively lower compared to the standard limits as stipulated in the EN ISO 17225-2 (≤ 1.0 wt.%) and the ISO 17225-6 (≤ 2.0 wt.%), except for S content in the EFB pellets derived from Tests 1 and 2 which was 0.01-0.02 wt.% higher than the limit (0.05 wt.%) specified in the EN ISO 17225-2, hence are of advantageous for cleaner energy generation. Tenorio et al.  reported that a combustion study of EFB pellets generated a very low emission of NOx (142.33 ppm) and none for SOx, with flame temperature varies between 500 - 700°C. During combustion, the N and S contained in biomass are oxidised and volatised to generate NOx and SOx. The physical co-treatment used in this study i.e. a combined sieving and water washing had further reduced the S and N contents of EFB pellets, and lowering further the potential of air pollutant emissions.
Table 7 shows major and minor trace elements exhibited in the pressed EFB fibers (raw material) and EFB pellets produced from Test 2 (sieving) and Test 3 (combined sieving and water washing). The trace elements present are mainly alkali and heavy metals, and their concentrations are relatively low and comparable to woody biomass pellets, except for K, Cl, Fe, Zn and Na. This difference may be due to the nature of the plant and their ability to uptake specific compounds and nutrients from the ecosystem i.e. soil, water as well as fertilizers and pesticides applied . As agro-based residues pellets, high contents of Zn and Fe of EFB pellets in particular is attributed to other natural factors concerning oil palm cultivation practices such as irrigation with treated or semi-treated effluent (wastewater from palm oil mill), chemical-based fertiliser, and organic sludge that is typically used in the soil environment and agriculture plantations for improving soil fertility . In addition, their presence serve as micro-nutrients beneficial as an alternative feed material and for soil amendment to enhance plant/animal growth .
These easily vaporized high amounts of alkali metals (K and Cl) together with the present S and silicates in the EFB pellets would result in severe ash-related problems to boilers such as slagging, fouling and corrosion [62,6]. These elements would also have adverse effects on boiler operation, ash quality, particulate matters and pollutants emission . The findings indicated that the applied sieving and water washing co-treatment could potentially reduce K, Cl and some heavy metals from the raw EFB feedstock at a minimum of 40% removal efficiency. This result corresponded with some of the previous studies, in particular concerning removal of K and Cl from EFB via other methods such as leaching and hydrothermal-washing co-treatment [29,35,34].