Densification process of Merbau (Intsia bijuga) and Matoa (Pometia pinnata J.R. Forster & J.G Forster) Sawdust Waste for Biomass Based Solid Fuel Source in West Papua Indonesia: Optimization using Response Surface Methodology (RSM)

Merbau (Intsia bijuga) and matoa (Pometia pinnata J.R. Forster & J.G Forster) are two amongst many prominent biomass sources from West Papua, Indonesia. With their versatile characteristics, merbau and matoa wood are used in many industries such as furniture, music instrument, and many other specialty products. However, wood processing industries can emit up to 60% of the residue. In this study, the usage of both merbau and matoa sawdust wastes as solid fuel was studied using response surface methodology (RSM). merbau and matoa sawdust are sieved to get the desired particle size (-20+50 mesh). Two kinds of the sawdust are then mixed in various compositions (70, 50, and 30% merbau). The mixed sawdust is then mixed with varied sticky starch solutions (1, 5, and 10%) to be formed in pellets with various moulding compression force (50, 100, and 150 kg/cm2). The pellets are then analyzed for its characteristics such as ash, moisture contents, and calorific value to be compared with its initial conditions. A full three-level factorial design under RSM was applied to explain the correlation between independent and dependent variables. The results show that statistically, merbau content, binder content, and compressive force showed relatively significant effects on the studied responses (ash content, moisture content, and calorific value). In addition, ANOVA analysis proved that each variable has significant effects on the responses that are confirmed by practically zero P-value. The coefficient of determinations (R2) are all above 0.96 and the normal probability plots confirm that the proposed models adequate the experimental results.


Introduction
In recent years, there has been an increasing interest in renewable and sustainable energy utilization around the world to slowly supersede fossil fuels usage.
Several main issues about fossil fuels such as carbon emission, environmental and health impacts, and its shortage in the future are starting to be highly considered (Dunlap & Jorgenson, 2012;Heede, 2014;Kampa & Castanas, 2008). One of the most promising energy sources to be utilized as a substitute is the biomass as solid fuel due to its abundance and easy implementation (Henderson et al., 2017;Stelte et al., 2012). However, one of the most important things to be considered is its sustainability. Sustainable biomass solid fuel usage should not compete with food crops, cause land-clearing, or create a higher possibility of CO2 emission (Tilman et al., 2009).
The raw material itself should not be used directly as a fuel. In terms of physical characteristics, the quality of a solid fuel is strongly influenced by its density (Kaliyan & Morey, 2009;Poddar et al., 2014;Wang et al., 2018). A densified solid fuel will have smaller volume and uniformed shape that will optimize the cost of handling, transportation, and storage (Kaliyan & Morey, 2009).
However, most importantly, it will have a higher energy density. A work by Majid, et al. (2017) suggested that densification can be done with binder and fructose was found to be the best binder for spruce wood shavings (Soleimani et al., 2017). Another work that was conducted by Weerapong, et al. (2011) successfully increased solid fuel heating value by 21.8% by torrefaction and almost 400% by densification process (Wattananoi et al., 2011), suggesting that densification process played a vital role on its energy density. Several works also had been done to compare different biomass sources as solid fuel. A work by Miranda et al. (2015) suggested that wastes from wood industry and forest produced higher heating value than herbaceous and fruit biomass wastes due to higher C percentages gives higher High Heating Value (HHV). Another work found that in general, bamboo and pine have almost the same net calorific value. However, pine wood has a much lower heat release rate from its bark content, giving it a benefit to maintaining combustion time (Liu et al., 2016).
On the other hand, West Papua, Indonesia holds a significant amount of biomass residue. Forest residues and sustainably harvested wood are among its main sources.
With a total forest production area of 2.19 million hectares (Forestry, 2015), its own wood processing production reached 150,537.50 m³ in 2015 (BPS, 2017).
Unfortunately, the normal percentage of biomass residue that will be produced from wood processing can be 60% (FAO, 1990), suggesting that there is more residue generated than that of its own production.  (Alakangas, 2011). For general uses, EN 14961-1 "Solid biofuels -Fuel specifications and classes -Part 1: General requirements" is used. As for international standards, the specifications are determined in ISO 17225-2 "Solid biofuels -Fuel specifications and classes -Part 2: Graded wood pellets" (Alakangas, 2015). Source: (Alakangas, 2011(Alakangas, , 2015 From the table, it is clear that the moisture of the raw materials is above the suggested standards. However, after densification and drying process, the result can be different and satisfying. Ash content of merbau Wood is slightly above the third-class standard but matoa wood is still acceptable. However, both raw materials have considerably satisfying calorific value. In addition, the calorific value of merbau wood is higher than the matoa, suggesting that the mixture between them can improve the pellet's quality.

Solid Fuel Making Process
In this research, merbau and matoa wood powder wastes were formed into wood pellet to be used as the solid fuel. The powders were ground and screened into -20+50 mesh. On the other side, the starch that was obtained from a local market in Yogyakarta, Indonesia was used as a binder. As much as 25 grams wood powder with varied compositions as described in Table 2 was mixed with 25 grams of water and various amounts of starch.
Several studies proved that the addition of the binder, especially starch, will enhance the briquettes' strength in the densification process. During the process, heating is an important part because it will help the mixture become gelatinous and sticky (Bazargan et al., 2014). Several studies also tested the binder with considerably high mass percentage compared to the solid fuel, such as 10%, 25%, or even 50% (Arzola et al., 2012;Bazargan et al., 2014). In this research, the starch was varied from 1% to 10% (starch weight/water weight) in order to focus the study on energy density upgrading via densification of the raw material itself.
As much as approximately 16 grams of the mixed wood powder and starch slurry was then formed with various mould forming compressions using Universal Testing Machine (Torsee's UTM AMU-5DE) as it can be seen in Figure 2. With a total of 27 samples were created and tested, the overall process and the variations can be seen in Figure 1 and Table 2. After the moulding process was done, the pellets were then dried in an oven at 102°C for 3 hours. The pellets were then analyzed by proximate analysis.  (Rahayuningsih et al., 2018). There are three independent variables used (wood content, binder content, and compression force) and three levels denoted as +1, 0, and -1 that represent as high, mid, and low level respectively as it can be seen in Table 2. (1) (Bezerra et al., 2008).
where Y is the response; xi and xj are the factors (i and j is the range from 1 to k); β0 is the constant coefficient; βi, βii, βij are the coefficients for the linear; quadratic and interaction effect; and ε is the error. The model's accuracy was analyzed using the coefficient of determination (R²). The R² has a value, ranged from 0 to 1. If R² is close to 1, it indicates that the model is highly accurate (Vedaraman et al., 2017).      (Maran & Manikandan, 2012). In addition, F-test and p-value are also assessed to determine the statistically significant variables which can be seen in Table 4.

Results and Discussion
As it can be seen from For a value lower than the level of significant, that is 0.05, the variable is considered as statistically significant for the proposed model (Sinha et al., 2012).
The results are also shown in response and contour plot in