Japanese cedar softwood (sugi, Cryptomeria japonica) and Japanese oak hardwood (konara, Quercus serrata) were used as feedstock.

Dried cedar and oak wood chips were ground in a commercial mill, whereupon water was then added to adjust the moisture content (about 15% - 20%-wet) in order to pelletize the feedstock smoothly. Figure 1 shows the pelletizer used in this study (Earth Engineering Corp., Japan). It is of the flat die type with a hole of ϕ 5.9 mm, 40 mm thickness, and production capacity of 30 kg/h. No binders other than water were used during the pelletization.

During laboratory scale production, wood pellet samples were torrefied using an inert gas oven (Advantech Toyo Kaisha, Ltd., Japan) under a nitrogen flow of 30 L/min. The weight of the sample was 1kg for cedar pellet and 0.5 kg for oak pellet. The torrefaction temperatures were 170˚C - 320˚C and the heating rate was about 8˚C/min. Figure 2 shows typical profile of torrefaction for cedar pellet at 260˚C. After maintaining the torrefaction temperature for about 5 min, the oven was slowly cooled to room temperature overnight, whereupon the characteristics of the torrefied pellets, including their mass yield, calorific value, volume density and bulk density were all evaluated according to the literatures [13] [14] . However in case of bulk density measurement, we used a 1L of plastic cylinder and read the scale of the top of the pellet layer, due to the small amount of the torrefied samples. During larger scale production, a conventional bamboo charcoal oven was used for torrefaction. 12 units of stainless vessels with cedar pellets (about 9 kg per unit) were placed in an oven, which was heated using firewood as fuel.

Initially small samples (approx. 1 mg) were employed for differential thermal analysis (DTA, DTG-60, Shimadzu

Corp., Japan) to check the exothermal behavior of torrefied material under the air flow. Subsequently, conventional cedar and torrefied pellets were mixed (80:20 by weight) and the mixture was used as fuel for a commercial pellet stove (FFP-701DF, Sunpot Co., Ltd., Japan) with output of 4.7 kW. To prevent the stove from overheating, we applied the mixing ratio as our first-time trial. The temperature of the combustion zone was monitored. The torrefied pellet used here was prepared by heat treatment of 6 kg cedar pellet at 240˚C for 12 h at 13 kPa (100 torr) in the oven (Yasujima Corp, Japan).

Table 1 and Table 2 show the characteristics of the torrefied pellets under various temperatures for Japanese cedar and oak, respectively. Generally, the calorific value per weight increased with increasing torrefaction temperature. For the Japanese cedar, calorific values were improved by heat treatment up to 260˚C. By heat treatment at 240˚C, the upgrade ratio of higher heating value (HHV) by weight was nearly 1.3 and the energy yield was 97% (Table 1). For the Japanese oak, calorific values were improved by heat treatment up to 320˚C. By heat treatment at 280˚C, the upgrade ratio of HHV exceeded 1.3 and the energy yield was 84% (Table 2). The moisture content was almost zero for torrefied pellets, due to water evaporation of the torrefaction process, which means water content need not be taken into consideration when calculating the lower heating value (LHV).

Conversely, volume and bulk density gradually decreased with rising torrefaction temperature. Figure 3 shows changes in bulk density and calorific value (by weight and volume) for torrefied oak pellets. Calorific value by volume (energy density) decreased gradually although that by weight increased with torrefaction temperature. It means that the benefit of transportation cost per unit of energy for the pellet torrefaction is smaller than that for chip torrefaction followed by pelletization.

Figure 4 shows scheme of the production. Although there were some problem such as heating control and continuous production, wood pellet with 25 MJ/kg of calorific value could be produced under 350˚C of torrefaction.

Figure 5 shows DTA profiles for untreated and torrefied samples for Japanese cedar and oak. The temperature ranges of exothermic peaks were almost the same between untreated and torrefied samples.

^aCalculated by ((HHV of torrefied)/(HHV of untreated (25˚C)) × 100; ^bCalculated by ((HHV of torrefied) × ((Mass yield of torrefied)/100))/(HHV of untreated (25˚C)) × 100.

^aCalculated by ((HHV of torrefied)/(HHV of untreated (25˚C)) × 100; ^bCalculated by ((HHV of torrefied) × ((Mass yield of torrefied)/100))/(HHV of untreated (25˚C)) × 100.

Figure 6 shows a commercial pellet stove and mixed fuel (conventional pellets:torrefied pellets = 80:20). Figure 7 shows the temperature profile of the combustion zone for conventional pellets and a mixture of conventional and torrefied pellets. A small increase in combustion temperature for the mixed pellet was observed. On the contrary there was no significant difference in ignition time, extinction time, and exhaust gas composition, meaning the torrefied wood pellets produced in this study might be usable as fuel for conventional pellet stoves. Further studies are necessary in order to apply 100% of torrefied pellet to the stoves.