Bioresource Technology 102 (2011) 1246–1253

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: [Link]/locate/biortech

Impact of torrefaction on the grindability and fuel characteristics of forest biomass

Manunya Phanphanich, Sudhagar Mani ⇑
Biochemical Engineering Program, Faculty of Engineering, Driftmier Engineering Center, University of Georgia, Athens, GA 30602, USA

a r t i c l e i n f o a b s t r a c t

Article history: Thermal pretreatment or torrefaction of biomass under anoxic condition can produce an energy dense
Received 24 May 2010 and consistent quality solid biomass fuel for combustion and co-firing applications. This paper investi-
Received in revised form 6 August 2010 gates the fuel characteristics and grindability of pine chips and logging residues torrefied at temperatures
Accepted 9 August 2010
ranging from 225 °C to 300 °C and 30 min residence time. Grinding performance of torrefied biomass
Available online 11 August 2010
evaluated by determining energy required for grinding, particle size distribution and average particle size
were compared with raw biomass and coal. Specific energy required for grinding of torrefied biomass
Keywords:
decreased significantly with increase in torrefaction temperatures. The grinding energy of torrefied bio-
Torrefaction
Wood chips
mass was reduced to as low as 24 kW h/t at 300 °C torrefaction temperature. The gross calorific value of
Logging residues torrefied chips increased with increase in torrefaction temperature. Torrefaction of biomass clearly
Grinding energy consumption showed the improved fuel characteristics and grinding properties closer to coal.
Coal Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction duce powdered biomass supply that is often required at the throat
of the conversion chamber (Mani et al., 2004).
Efficient and economical conversion of renewable biomass into Combustion and cofiring of biomass with coal is difficult due to
high value fuels and chemicals is a top research and development imbalance in the energy density differences of coal and biomass,
priority for researchers, scientists and policy makers to enhance low burning rate and low flame temperature. It also creates addi-
the nation’s energy security, to mitigate the impact of global tional problem due to poor flowability, poor blending and reduces
warming and to boost local agriculture and rural economic devel- the thermal efficiency and capacity of existing boiler units (Hughes
opment. A recent biomass availability study conducted by The Uni- and Tillman, 1998; Tillman, 2000). Gasification of low quality bio-
ted States Department of Agriculture (USDA) and Department of mass results in poor synthetic gas quality and high tar concentra-
Energy (DOE) has reported that nearly 1.2  109 dry Mg of biomass tion. Tar is a highly viscous and higher molecular weight volatile
can be produced annually from agricultural and forestry lands for organic compound generated during biomass gasification process.
generation of biofuels and power (Perlack, 2005). The Southeast A recent study reported that feedstock composition has some effect
US contributes significantly on the sustainable production of forest on tar formation during gasification (Kiel et al., 2004). They re-
biomass and energy crops. ported that presence of high moisture and a high concentration
Structural heterogeneity, non-uniform physical properties and of hemi-cellulose and lignin in the feedstock produce more tar in
low energy density of biomass have become major problems in the syngas than that of cellulose. The hydrophillic nature of bio-
its efficient and economic transport, handling, storage and conver- mass compounds additional problem for safe storage.
sion into bioenergy products. Biomass in its original form has wide In order to address the above problems, biomass needs to be
range of moisture content (60–25%), uneven particle sizes (100– pretreated to improve its quality for efficient energy conversion.
10 mm) and low energy density (8–14 MJ kg1) and is difficult to Thermal pretreatment or torrefaction of biomass is a promising
store, transport and feed it in the existing combustion and gasifica- method to preprocess low quality biomass into high energy density
tion systems. Low bulk density (60–100 kg m3) of lignocellulosic feedstock with consistent and uniform physical and chemical char-
biomass further increases the transport and storage cost of the acteristics. Torrefaction can therefore be suitable for long distance
feedstock (Mani, 2005; Mani et al., 2006; Sokhansanj et al., transportation, long term storage and improved conversion of bio-
2009). Energy required for grinding of biomass is very high to pro- mass because of reduction in downstream cleaning and processing
costs. During torrefaction of biomass, some of the highly reactive
volatile compounds are removed from the biomass as vapors and
resulting in clean, dark brown colored solid biomass with higher
⇑ Corresponding author. Tel.: +1 706 542 2358; fax: +1 706 542 8806. energy density. The energy density of biomass could be increased
E-mail address: smani@[Link] (S. Mani). closer to the energy density of coal (22–23 MJ kg1) used for heat

0960-8524/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/[Link].2010.08.028
 M. Phanphanich, S. Mani / Bioresource Technology 102 (2011) 1246–1253 1247

Nomenclature

a Measured area covered by a particle projection P1 The pressure reading after pressurizing the reference
A Surface of all particles (mm2) volume (Pa).
C Carbon P2 The pressure after including Vc (Pa).
d16 The particle diameters obtained from the cumulative Sgw The geometric mean diameter standard deviation (mm).
distribution data at 16%. Sv The specific surface area (mm1)
d84 The particle diameters obtained from the cumulative SHPT Sphericity (dimensionless).
distribution data at 84%. T Torrefaction temperature (°C).
dgw d50 and The geometric mean diameter (mm). V Volume of all particle (mm3).
Eg Specific energy consumption for grinding (kW h/t). Vc The volume of sample cell (m3).
H Hydrogen VM Volatile matter.
HHV Higher heating value (MJ kg1). Vp The volume of solid biomass.
N Nitrogen. VR The reference volume (m3).
O Oxygen. Xc min The shortest chord or width (mm).
P Measured circumference of a particle projection.

and power generation (Lipinsky et al., 2002; Prins, 2005; Zwart bed for temperature control and was data logged to monitor and
et al., 2006). The higher energy density of torrefied biomass as control the torrefaction experiment. A nitrogen flow of 2 L min1
compared to sawdust, wood chips and conventional wood pellets was used as a carrier gas throughout the experiment to eliminate
would reduce transport costs and facilitate higher rates of co-firing the presence of oxygen, thereby avoiding oxidation and ignition.
with coal at power plants. The cement industry represents another The exhaust from the batch reactor was passed through series of
potential large application where torrefied biomass can be co-fired metal condensers immersed in an ice bath before exhausted to
with coal. Torrefied wood could also be blended with biomass the atmosphere. The purpose of condensers was to collect all the
feedstock used for production of pellets and briquettes (Arcate, potential condensable generated from the torrefaction reaction. A
2002). During the torrefaction process, biomass losses about 10– known amount of samples (1–1.5 kg) was charged into the reactor
20% of its mass through conversion into volatile gases and other at room temperature, and the reactor was heated at 10 °C min1
chemicals which can be captured and utilized to produce value under atmospheric pressure. Both clean wood chips and forest res-
added chemicals. Preliminary research work carried out on the tor- idue chips were torrefied at four different temperatures (225, 250,
refaction of wood in the batch reactor showed some evidence that 275, and 300 °C) and at 30 min residence time. The torrefaction
torrefaction improves feedstock quality (Pach, 2002; Ferro et al., residence time of 30 min was chosen to be optimal for torrefaction
2004; Prins, 2005; Bergman et al., 2005). A recent study by Felfli reaction from preliminary studies. After the treatment, samples
et al. (2005) reported that torrefaction of wood briquettes in- were cooled and weighed to determine the solid yield. The samples
creased the energy density of briquettes and decreased moisture were placed in the air tight plastic bags and stored in a desiccator
adsorption during storage. for further analysis. The moisture content of torrefied sample was
The main objectives of this study were to investigate the impact determined as per ASABE S358.2 (ASABE, 2006) standard protocol
of torrefaction treatment on the grinding energy consumption and (see Section 2.4.1). The scope of this study was primarily focused
fuel characteristics of logging residues and clean wood chips. on the analysis of torrefied clean wood chips and logging residues.

2. Methods 2.3. Grinding experiment

2.1. Materials A laboratory heavy-duty knife mill (Retsch SM 2000, Germany)

was used in the grinding experiments. The knife mill used in this
Clean pine chips were received from a local saw mill operation study consisted of a cutting blade rotor (1690 rpm, 60 Hz) which
in Oglethorpe, Georgia. The moisture content of chips as received is powered by a 1.5 kW electric motor. A computer with a data log-
was 41.7% on the wet basis (wb). The pine chips dimensions were ger assembled with a multi function transducer (CR Manetics Inc.,
in the ranges of 20.94–70.59 mm in length, 1.88–4.94 mm in thick- MO) was connected to the mill to measure the power consumption
ness and 15.08–39.70 mm in width. Southern yellow pine logging during the grinding experiment. Instantaneous power consump-
residue chips were also received from local logging operation in tion data was recorded every two seconds by the computer. The
Macon, GA and they consists of clean chips (sap and heart wood), bottom sieve opening of 1.5 mm was used in this study as this size
branches, barks, leaves and small particles. The anatomical compo- has been found to produce good quality pellets for combustion
sitions and the particle sizes of logging residues can be obtained applications (Mani, 2005; Mani et al., 2006; Kaliyan and Morey,
from Phanphanich and Mani (2010). The initial moisture content 2009). Untreated wood and coal were also tested for grinding
of the residues was 49.60% (wb). Chip samples were dried to 10% experiments under the same grinding conditions for comparisons.
moisture content (wb) for safe outside storage and were used for A known quantity of biomass samples was taken and was ground
torrefaction study. Bituminous coal sample was collected from using a knife mill. The sample was manually fed without exceeding
the Central Steam Plant at the University of Georgia, Athens, Geor- the maximum current drawn by the mill. A circuit breaker was in-
gia for comparative analysis. stalled in the grinder motor to avoid motor overloading. The time
required for complete grinding of the samples was recorded. Dur-
2.2. Torrefaction experiment ing grinding, the instantaneous power consumption data with re-
spect to time was recorded and used for energy consumption
Torrefaction of wood chips and forest residues was conducted calculation. The specific energy consumption for grinding was
in a batch torrefaction reactor placed inside an externally heated determined by integrating the area under the instantaneous power
electric furnace (Fig. 1). A thermocouple was inserted in the sample consumption curve (P–t) over the total time required to grind the
1248 M. Phanphanich, S. Mani / Bioresource Technology 102 (2011) 1246–1253

Temperature profile

Thermocouple
Non-condensable phase

Volatiles
N2

Condensable phase
Biomass

Reactor
Cold water H2O

Torrefied Biomass

Fig. 1. Laboratory torrefaction unit using a batch pyrolysis reactor.

given sample (Mani et al., 2004). Each test was repeated three 1
Sgw ¼ ½d84  d16  ð1Þ
times. 2

2.4. Analyzes of physical and fuel properties

p
V Ellipsoid ¼ X c min2  Length ð2Þ
6
2.4.1. Moisture content
4pa
The untreated and torrefied samples were analyzed for mois- SHPT ¼ ð3Þ
p2
ture content according to ASABE S358.2 (ASABE, 2006) standard
protocol. About 10 g sample was placed in a convective oven at
A
103 + 2 °C. The mass loss was recorded after 24 h and used to cal- Sv ¼ ð4Þ
m
culate sample moisture content. ASTM Standard D 3173–87 (ASTM,
1998) was used to determine moisture content of coal sample. The Where d84 and d16 are particle diameters obtained from the cumu-
moisture content data were reported in wet basis. Each test was re- lative distribution data at 84% and 16% respectively. The Particle
peated three times. size distribution and cumulative passing percentage were con-
structed against the shortest chord or width (Xc min), which is the
2.4.2. Particle size and particle size distribution dimension that fits to sieve results because with sieving, the parti-
An image analyzer (HORIBA CAMSIZER, Instruments, Inc. CA, cle passes the mesh with its smallest dimension. Each run was re-
USA) equipped with two digital cameras was used to measure peated three times.
average particle size and size distribution using a standard proce-
dure (ASABE, 2008). Since the samples were electro-statically 2.4.3. Bulk density
charged, an ultrasonic probe disperser was used for all measure- The bulk density of grinds was performed by employing ASTM E
ments to better characterize needle shaped particles. To conduct 873–82 for densified particulate biomass fuels (ASTM, 2006). The
the test, a sample of 20–40 g biomass after the grinding experi- samples were poured from a height of 0.6 m above the top edge
ment was loaded onto the hopper of the instrument. The sample of the container and then dropped five times from a height of
was conveyed and dropped via a vibratory feeder onto the mea- 0.15 m onto a hard surface to allow settling. The density was calcu-
surement field of the cameras. The size and shape of powders were lated by the ratio of mass of the sample to volume corresponding
recorded and analyzed by the software provided by the equipment to the height of the sample in the container.
manufacturer. The series of sieve size were set on the software in
the series ranges from 30 to 0.02 mm. The series of sieves and 2.4.4. Particle density
the following particle size analysis equations are followed the A gas multi-pycnometer (MPV-D160-E, Quantachrome Corpora-
Standard ANSI/ASAE S319.4 (ASABE, 2008). Four parameters were tion, FL) was used to measure the particle density of grinds by
analyzed, including: (a) geometric mean diameter (dgw) – the size measuring the pressure difference when a known quantity of he-
at 50% cumulative distribution; (b) the geometric mean diameter lium (He) under pressure was allowed to flow from a precisely
standard deviation (Sgw) – the measurement of the variation in known reference volume (VR) into a sample cell (Vc) containing
the particle sizes of the wood powders; (c) the sphericity (SHPT) the solid material. The true volume of the sample (Vp) was calcu-
(Eq. (3)) -roundness value ranges from 0 to 1 (a perfect sphere lated from Eq. (1). The particle density of the sample was its mass
has a value of 1); and d) the specific surface area (Sv) (Eq. (4)) – divided by Vp and was expressed in Mg/m3. Both bulk and particle
the value directly shows the results of size reduction and indicates density measurements were repeated three times.
combustion properties of fuel.
 M. Phanphanich, S. Mani / Bioresource Technology 102 (2011) 1246–1253 1249

p
 
2.4.8. Chemical compositions
Vp ¼ Vc  VR 1  1 ð5Þ
p2 The sample was dried and treated in a neutral detergent solu-
tion (30.0 g Sodium dodecyl sulfate, USP; 18.61 g Ethylenedi-
where Vp is the volume of solid biomass (m3), Vc is the volume of aminetetraacetic disodium salt, dihydrate; 6.81 g Sodium borate;
sample cell (m3), VR is the reference volume (m3), P1 is the pressure
4.56 g Sodium phosphate dibasic, anhydrous; and 10.0 ml Triethyl-
reading after pressurizing the reference volume (Pa), and P2 is the ene glycol, in 1 L distilled H2O). The insoluble residues were hemi-
pressure after including Vc (Pa). The tests were repeated three times
cellulose, cellulose, and lignin. Then, the separation of cellulose and
on each sample. lignin: sample was digested with acid-solution (20 g of cetyl trime-
thylammonium bromide (CTAB) to 1 L of H2SO4). After digestion,
2.4.5. Heat content the hemicelluloses dissolved and were filtered away. Secondly,
The adiabatic oxygen bomb calorimeter (IKA C 2000, IKA Works, the separation of lignin: acid detergent lignin (ADL) was measured
Inc., NC) was used for determination of gross calorific value of by further treating ADF (the insoluble fiber residues remained after
grinds based on the ASTM D 5865–03, standard test method for digesting which consisted mainly of cellulose and lignin) with 72%
coal and coke (ASTM, 2003). Initially, a known amount of grinds H2SO4, which dissolved cellulose. The cellulose content was de-
(0.5 g) was placed in a nickel crucible and burned inside the rived by subtracting ADF values from ADL values. Calculations of
bomb calorimeter surrounded by a water jacket. The sample was cellulose and hemicellulose percentages were given by.
ignited by a pure cotton thread in the presence of oxygen. Upon
the ignition, the released heat is transferred to the water jacket % Hemicellulose ¼ % NDF  % ADF ð6Þ
causing temperature to rise. The increase in temperature was used
to calculate the heating value of the sample. The calorific value was % Cellulose ¼ % ADF  % ADL ð7Þ
expressed in MJ/kg.

2.4.6. Proximate analysis 2.5. Statistical analysis

Proximate analysis (Ash, volatiles, fixed carbon and moisture
content) was carried out with a micro thermo-gravimetric analyzer Specific energy consumption for grinding, average particle size
(Model TGA701, LECO Corporation, St. Joseph, MI) as similar to the and higher heating values data for torrefied pine chips and forest
procedure of ASTM D 5142–04 for coal and coke (ASTM, 2002a). residues were compared at various torrefaction temperatures
Approximately, one to 2 g of grinds was placed in an aluminum using one-way ANOVA and Duncan multiple comparison test.
crucible which was placed and heated in a muffle furnace. Ash con- SAS Non-linear Regression (NLIN) (SAS, 2006) was used for regres-
tent and volatile matter were expressed in dry basis. The fixed car- sion fits and analyses. The level of significant difference was kept at
bon, in dry basis was calculated from subtracting a hundred P < 0.05.
percentages with ash and volatile matter.
3. Results and discussion
2.4.7. Ultimate analysis
Elemental analysis of the samples was performed according to 3.1. Torrefied biomass mass and energy yields
the ASTM D3176–89 standard test method for coal and coke
(ASTM, 2002b). The C, H, N, S contents were measured using an ele- Table 1 shows the mass and energy yields of pine chips and log-
mental analyzer (LECO CHNS 932, LECO Corporation, MI). Oxygen ging residues at various torrefaction temperature conditions. En-
was estimated by the difference. All thermal and chemical mea- ergy yield data represents the ratio of actual energy retained
surements were conducted in triplicates. after the torrefaction process to initial energy content of biomass.
Table 1 shows that the mass yield of torrefied biomass (both pine
Table 1 chips and logging residues) decreased with increase in torrefaction
Effect of torrefied temperature on mass and energy yields of torrefied wood chips. temperature. The mass yield started to decline dramatically from
Biomass Torrefaction Mass yield Energy yield
temperature of 275 °C and became to only about one-half of the
temperature (°C) (%) (%) original weight when torrefied temperature reaching 300 °C. The
Pine wood chips 225 89 94
mass loss was primary due to thermal decomposition of hemi-cel-
250 82 90 lulose and some short chain lignin compounds (Bergman et al.,
275 73 87 2005). The decomposed volatiles can be captured and used for var-
300 52 71 ious applications. Detailed compositions of condensable volatile
Logging residue chips 225 88 92 products and non-condensable gases can be obtained from Mani
250 81 92 (2009). Fig. 2 shows the physical changes occurred during torrefac-
275 70 82
tion of wood chips at various temperatures. The energy yields ran-
300 52 72
ged from 71% to 94% depending on the torrefaction temperatures.

Fig. 2. Physical changes occurred during torrefaction of pine wood chips at various treatment temperatures: (a) control, (b) 225 °C, (c) 250 °C, (d) 275 °C, (e) 300 °C.
1250 M. Phanphanich, S. Mani / Bioresource Technology 102 (2011) 1246–1253

Table 2
Fuel properties of torrefied wood chips and logging residuesb.

Biomass MC (%, wb) VM (%, db) Ash (%, db) FC (%, db) C (%, db) H (%, db) N (%, db) O (%, db) HHV (MJ kg1)c
Pine chips (PC) 6.69 (0.11) 85.98 (0.01) 0.27 (0.02) 13.76 (0.02) 47.21 (0.28) 6.64 (0.01) 0.17 (0.05) 45.76 (0.53) 18.46 (0.13)a
TPC-225 °C 3.30 (0.06) 84.78 (0.06) 0.27 (0.01) 14.95 (0.06) 49.47 (0.02) 6.07 (0.02) 0.15 (0.01) 44.03 (0.03) 19.48 (0.02)b
TPC-250 °C 2.88 (0.05) 82.52 (0.41) 0.25 (0.01) 17.24 (0.40) 51.46 (0.01) 5.86 (0.37) 0.14 (0.01) 42.02 (0.34) 20.08 (0.01)c
TPC-275 °C 2.46 (0.10) 76.40 (0.27) 0.35 (0.03) 23.26 (0.23) 54.91 (0.27) 6.20 (0.1) 0.20 (0.0) 38.17 (0.13) 21.82 (0.08)d
TPC-300 °C 2.57 (0.02) 58.72 (0.16) 0.43 (0.06) 40.85 (0.22) 63.67 (0.01) 5.58 (0.14) 0.20 (0.01) 29.99 (0.16) 25.38 (0.07)e
Logging residue chips (LR) 7.94 (0.18) 82.17 (0.06) 1.77 (0.11) 16.07 (0.05) 47.29 (0.37) 6.20 (0.12) 0.42 (0.01) 45.19 (1.76) 18.79 (0.01)a
TLR-225 °C 3.11 (0.16) 80.75 (0.05) 1.37 (0.02) 17.90 (0.04) 50.15 (0.02) 6.10 (0.05) 0.30 (0.03) 42.74 (1.03) 19.79 (0.01)b
TLR -250 °C 2.66 (0.09) 78.16 (0.09) 1.49 (0.04) 20.37 (0.06) 54.91 (0.04) 5.87 (0.20) 0.31 (0.03) 40.96 (1.18) 21.21 (0.03)c
TLR -275 °C 2.64 (0.08) 71.44 (0.89) 1.88 (0.05) 26.69 (0.84) 53.24 (0.95) 5.39 (0.5) 0.30 (0.04) 40.12 (2.82) 22.03 (0.19)d
TLR -300 °C 2.36 (0.10) 52.92 (0.06) 2.32 (0.08) 44.76 (0.08) 66.07 (0.13) 4.92 (0.16) 0.48 (0.04) 27.34 (1.88) 26.41 (0.02)e
Bituminous Coala 1.57 (0.05) 36.58 (0.02) 6.52 (0.05) 56.91 (0.20) 77.63 6.30 2.00 9.16 27.72 (0.02)f
a
Sulfur content in bituminous coal was 1.66 (0.20)%, db.
b
Number enclosed in the parenthesis were standard deviations with n = 3.
c
Number with the same letters were not significantly different.

3.2. Fuel characteristics of torrefied biomass illustrated that torrefied logging residues had lower levels of H/C
and O/C ratios than those of torrefied pine chips. The chemical
Data from proximate, ultimate and heating value analyses of compositions of torrefied biomass were very similar to that of lig-
torrefied pine chips (TPC) and torrefied logging residues (TLR) are nite coal or sub-bituminous coal for potential co-firing application.
given in Table 2. It was observed that moisture content of torrefied Higher heating value (HHV) of torrefied biomass was significantly
biomass was decreased as the torrefaction temperature increases increased with increase in torrefaction temperature. Increase in
when stored at room temperature. It indicates that torrefied bio- energy density of biomass during torrefaction was comparable
mass can’t absorb moisture compared to untreated wood. This with other similar studies for agricultural residues (Pimchuai
behavior may be due to loss of hydroxyl (OH) groups from bio- et al., 2010) and wood chips (Prins et al., 2006). In overall, energy
mass during torrefaction reaction. The volatile matter present in density or HHV of torrefied logging residues was slightly higher
both samples was changed less at temperature of 225 and than that of torrefied chips. The higher energy density was primar-
250 °C, while noticeable reduction was observed at torrefied tem- ily due to higher percentages of lignin and extractives present in
peratures of 275 and 300 °C. Similar observation was also found in logging residues (Shafizadeh et al., 1976; Lehtikangas, 2001). Lig-
the study of torrefied wood briquette and other biomass species nin are linked by multitude of inter-unit bonds including ether
(Felfli et al., 2005; Pimchuai et al., 2010). An observed increase in and carbon–carbon linkages (Mansouri and Salvadó, 2006; van
ash content of torrefied biomass was mainly due to mass loss (con- Loo and Koppejan, 2008). In an energy density prospective, energy
centration effect) during torrefaction reaction. Logging residues contained in C–C bond is higher than C–O or C–H bonds. The rela-
had slightly higher ash content than that of pine chips due to high tive increase of C better indicates combustion characteristics as it
percentage of bark and needles contents. directly results in higher calorific value, as lower of O/C and H/C ra-
As the torrefaction temperature increased, the elemental carbon tios reduces thermodynamic losses (over-oxidation) smoke and
content of torrefied biomass was increased. At the same time, water vapor (Bergman et al., 2005; Prins et al., 2006).
hydrogen and oxygen contents of torrefied biomass decreased, Table 3 shows the change in chemical composition of torrefied
resulting in decreased H/C and O/C ratios. Lower hydrogen and biomass at various temperatures. This table supports the fact that
oxygen contents of torrefied biomass also support the fact of hemicelluloses present in the biomass samples used in this study
reduction in hydroxyl (OH) groups during torrefaction. Fig. 3 were thermally decomposed. Although it is a known fact that cel-
lulose does not decompose (Milosavljevic and Suuberg, 1995). The
acid insoluble portion of cellulose was accounted with lignin fibers
2.0 present in torrefied biomass. As a result, the acid insoluble fiber
Untreated-PC Untreated-LR content of torrefied biomass increased with increase in torrefaction
1.8 TPC-225 TLR-225
TPC-250 TLR-250
temperature. Therefore, it may be concluded that torrefied biomass
1.6 TPC-275 TLR-275 will not be suitable for extracting simple sugars and further con-
TPC-300 TLR-300 verting into ethanol.
1.4 COAL*
Peat
Fuel H/C ratio

1.2 Lignite 3.3. Grinding of torrefied biomass

1.0
Bituminous 3.3.1. Specific energy consumption for grinding
0.8 Coal Fig. 4 shows the effect of torrefaction temperature on the spe-
cific energy required to grind torrefied pine chips and logging res-
0.6
idues. Specific grinding energy consumption of untreated wood
0.4 chips and logging residues was very high (237 kW h/t) and was
comparable to other similar studies (Esteban and Carrasco, 2006).
0.2 Anthracite
Specific energy consumption for grinding was significantly reduced
0.0 with increase in torrefaction temperature for both torrefied pine
0.0 0.2 0.4 0.6 0.8 chips and logging residues. Specific grinding energy consumption
Fuel O/C ratio of torrefied biomass was linearly correlated with torrefaction tem-
Fig. 3. Chemical composition for pine chips and forest residues at various
perature (R2 > 0.9). Grindability of torrefied wood was improved
torrefaction temperatures. The comparison data of various solid fuels was obtained dramatically at 275 and 300 °C as the specific energy required for
from (van Loo and Koppejan, 2008). grinding (23–78 kW h/t) were as low as that needed to grind
 M. Phanphanich, S. Mani / Bioresource Technology 102 (2011) 1246–1253 1251

Table 3 Pine Chips

Chemical compositionsa of torrefied biomass.
Logging Residues
Biomass Hemi-cellulose Acid soluble Acid insoluble fibers 0.8

Geometric mean particle diameter, dgw

(wt.%) cellulose (wt.%) (mostly lignin), (wt.%)
0.7
Untreated-PC 15.19 48.57 26.16
TPC 225 °C 12.87 41.23 38.42
0.6
TPC 250 °C 6.93 41.90 45.70 0.5
TPC 275 °C 0.99 39.54 53.30 dgw = -7E-06T2 + 0.001T + 0.717

(mm)
TPC 300 °C 0.56 12.84 79.99 0.4 R² = 0.99
Untreated-LR 13.26 37.49 26.15 0.3
TLR 225 °C 14.77 41.04 33.20
TLR 250 °C 5.87 38.57 42.49 0.2
TLR 275 °C 5.23 34.08 52.80 dgw = -1E-05T2 + 0.001T + 0.669
0.1 R² = 0.99
TLR 300 °C 1.04 6.10 85.06
a
0.0
The reported values were presented in dry ash free percentage with a standard
0 50 100 150 200 250 300
deviation of 0.02.
Torrefaction temperature, T (oC)

Fig. 5. Effect of torrefaction temperature on the geometric mean particle diameter

grasses: wheat straw, barley straw, corn, and switchgrass (43.6,
of torrefied biomass.
27.1, 19.8, and 58.5 kW h/t, respectively) using a similar sieve size
of 1.6 mm (Schell and Harwood, 1994; Mani et al., 2004) and sim-
ilar to energy needed to grind coal (7–36 kW h/t). The specific en- particles than that of untreated biomass. The particle size distribu-
ergy needed to grind the bituminous coal used as a benchmark was tion curve was skewed towards smaller particle sizes as the torre-
8.86 kW h/t. This value is comparable with other published data on faction temperature increases. The similar skewed distribution of
grinding of coal for power generation (Lytle et al., 1992). Compared ground biomass was also observed from other studies (Himmel
to untreated biomass samples, torrefaction at 300 °C reduce grind- et al., 1985; Mani et al., 2004). Cumulative percent passing curve
ing energy consumption by 10 times for pine chips and by six times also showed the similar behavior for torrefied biomass. In overall,
for logging residues. Improved grindability of torrefied biomass torrefied biomass produced relatively narrower or uniform particle
may primarily be due to relative increase in biomass brittleness size ranges compared to untreated biomass due to intense brittle-
and breakdown of fibrous compounds during torrefaction (Arias ness of torrefied biomass similar to coal. This behavior was directly
et al., 2008; Mani, 2009). supported by the least grinding energy consumption of torrefied
biomass.
3.3.2. Particle size and particle size distribution (PSD)
Fig. 5 shows the effect of torrefaction temperature with geomet- 3.3.3. Sphericity and particle surface area
ric mean particle size of torrefied biomass ground through a screen Table 4 summarizes the physical properties of ground torrefied
size of 1.5 mm. The mean particle size of ground torrefied biomass pine chips and logging residues at various torrefaction tempera-
was decreased with increase in torrefaction temperature. The aver- tures. Sphericity of pine chips was improved as the torrefaction
age particle sizes of logging residues at all torrefaction tempera- temperature was increased from 0.48 to 0.62. Higher sphericity va-
tures were slightly higher than that of torrefied pine chips. In lue of torrefied pine chips indicated that the shape of ground torr-
overall, torrefaction of biomass not only decreased the specific en- efied pine chip was changed. However, no significant difference in
ergy required for grinding but also decreased the average particle sphericity of ground torrefied logging residue was observed com-
size of ground biomass. A second order polynomial model was pared to untreated logging residues thus indicating that the shape
developed to correlate mean particle diameter of torrefied biomass of torrefied logging residue did not change compared to untreated
and the torrefaction temperature (Fig. 5). logging residue. Particle surface area of torrefied biomass was
Fig. 6 shows that percent particle size distribution and cumula- higher compared to that of untreated biomass. In overall, an in-
tive percent passing of ground torrefied pine chips and logging res- crease in particle surface area or a decrease in particle size of torr-
idues. The particle size distribution curves for pine chips and efied biomass can be the desired properties of torrefied biomass for
logging residues showed that torrefied biomass produces smaller efficient combustion and co-firing applications (Xiumin et al.,
2002). Additionally, particle size, shape and surface area had signif-
icant impact on combustion characteristics of biomass and coal (Lu
300
et al., 2010; Saastamoinen et al., 2010).
Specific energy consumption for

Logging Residues
250 Eg = -0.665T + 258.3 3.3.4. Bulk density and particle density
grinding, Eg (kWh/t)

R² = 0.94 Generally, decrease in particle size of biomass would increase

200
the bulk density and particle density. This is due to the reduction
in both inter- and intra-particle voids generated after milling
150
(Mani et al., 2004; Esteban and Carrasco, 2006). However, bulk
density and particle density of torrefied biomass behaved differ-
100
ently. Table 4 shows that particle density of torrefied pine chips
Pine Chips and logging residue particles did not change compared to that of
50 untreated biomass. Bulk density of torrefied pine chip particles de-
Eg = -0.756T + 260.0
R² = 0.98 creased until the torrefaction temperature of 250 °C. Then, it was
0 increased up to the torrefaction temperature of 300 °C. On the
0 100 200 300
o other hand, bulk density of torrefied logging residue particles did
Torrefaction temperature, T ( C)
not change compared to that of untreated logging residue particles.
Fig. 4. Effect of torrefaction temperature on the specific energy consumption for This result was contrary to the behavior observed for untreated
grinding of torrefied biomass. agricultural residues and energy crops (Mani et al., 2004).
1252 M. Phanphanich, S. Mani / Bioresource Technology 102 (2011) 1246–1253

Untreated- 100
14 90
PC
TPC-225

Cummulative passing[%]
Particle size distribution [%]
12 80
TPC-250 70
10
60
TPC-275 Untreated-PC
8 50
TPC-300 TPC-225
6 40
30 TPC-250
4
20 TPC-275
2 10 TPC-300
0
0 0 1 2
0 1 2
Particle size (mm)
Particle size (mm)
(a1) (b1)

100
14 Untreated-
Particle size distribution [%]

LR 90

Cummulative passing [%]

12 TLR-225 80
10 TLR-250 70
60 Untreated-LR
8 TLR-275 50 TLR-225
6 TLR-300 40
TLR-250
4 30
20 TLR-275
2 10 TLR-300
0 0
0 1 2 0 1 2
Particle size (mm) Particle size (mm)
(a2) (b2)

Fig. 6. Particle size distributions of various torrefaction -temperature of (a1) ground pine chips and (a2) logging residues, cumulative passing graphs of various torrefaction
temperature of (b1) for ground pine chips and (b2) for logging residues.

Table 4
Physical properties and specific energy consumption for grinding of torrefied biomassa.

Biomass grinds dgw (mm)b,d Sgw (mm)b SPHTc Sv (mm1)b Bulk density Particle density Specific grinding energy
(Dimensionless) (kg m3)b (Mg m3)b consumption (kW h/t)b,d
Untreated-PC 0.71 (0.01)a 0.41 (0.01) 0.48 (99.2) 13.5 (0.3) 381.0 (1.3)) 1.40 (0.02) 237.7 (6.6)a
TPC-225 °C 0.45 (0.01)c 0.40 (0.01) 0.52 (98.2) 30.4 (1.8) 342.4 (4.1) 1.41 (0.02) 102.6 (13.4)b
TPC-250 °C 0.42 (0.0)d 0.38 (0.01) 0.51 (98.3) 32.6 (0.2) 331.8 (4.5) 1.40 (0.01) 71.4 (3.6)c
TPC-275 °C 0.27 (0.03)f 0.36 (0.03) 0.60 (95.3) 48.6 (3.3) 376.1 (1.6) 1.37 (0.01) 52.0 (7.3)d
TPC-300 °C 0.13 (0.02)g 0.31 (0.02) 0.62 (92.4) 64.2 (0.2) 400.2 (4.0) 1.34 (0.01) 23.9 (3.0)e
Untreated-LR 0.74 (0.03)a 0.43 (0.03) 0.54 (97.1) 17.2 (0.2) 351.3 (1.6) 1.35 (0.01) 236.7 (16.0)a
TLR-225 °C 0.57 (0.0)b 0.42 (0.0) 0.57 (96.4) 21.4 (0.2) 351.7 (1.0) 1.39 (0.02) 113.8 (9.0)b
TLR-250 °C 0.53 (0.01)b 0.41 (0.01) 0.53 (96.9) 21.8 (0.2) 336.6 (2.6) 1.37 (0.02) 110.4 (4.3)b
TLR-275 °C 0.46 (0.01)c 0.37 (0.01) 0.53 (96.2) 24.0 (1.2) 331.5 (0.9) 1.34 (0.01) 78.0 (10.2)c
TLR-300 °C 0.36 (0.02)f 0.33 (0.02) 0.53 (95.6) 29.2 (2.8) 330.5 (2.6) 1.33 (0.01) 37.6 (7.0)f
a
All grinds were received from knife mill with a screen size of 1.5 mm.
b
Number enclosed in the parenthesis were standard deviations with n = 3.
c
Number enclosed in the parenthesis represented fraction of sample that is less than 0.9 sphericity.
d
Number with the same letters were not significantly different at p = 0.05.

4. Conclusions Acknowledgement

Torrefaction of clean wood chips and logging residues signifi- Authors acknowledge Traditional Industries Program (TIP3) for
cantly improved the heating value compared to that of untreated forest resources and products, State of Georgia (GA), USA and Office
biomass. Proximate and elemental compositions of torrefied bio- of Vice President for Research (OVPR), University of Georgia, Ath-
mass were also improved and were comparable to coal. Specific en- ens (UGA) for financially supporting this project.
ergy consumption for grinding of torrefied biomass was decreased
up to 10 times for torrefied wood chips and up to six times for torr- References
efied logging residues compared to that of untreated biomass. Tor-
refaction temperature was linearly correlated with specific Arcate, J., 2002. Global markets and technologies for torrefied wood in 2002. Wood
grinding energy consumption and average particle diameter of Energy 6, 26–28.
Arias, B., Pevida, C., Fermoso, J., Plaza, M.G., Rubiera, F., Pis, J.J., 2008. Influence of
ground torrefied biomass. Physical and fuel characteristics of torr- torrefaction on the grindability and reactivity of woody biomass. Fuel
efied biomass was superior compared to that of untreated biomass. Processing Technology 89 (2), 169–175.
 M. Phanphanich, S. Mani / Bioresource Technology 102 (2011) 1246–1253 1253

AS, S., 2006. SAS User’s Guide: Statistics. Ver. 9. SAS Institute Inc, Cary, NC. Mani, S., 2009a. Integrating biomass torrefaction with thermo-chemical conversion
ASABE, 2006. ASAE S358.2 moisture measurement – forages. In: ASABE Standards, processes. In: Proceedings of The 2009 AIChE Annual Meeting. Nashville, TN
608. American Society of Agricultural and Biological Engineers, St. Joseph, MI. [Paper # 160229].
ASABE, 2008. ANSI/ASAE S319.4 Method of determining and expressing fineness of Mani, S., KC. Das, J. Kastner & R. Adolphson. 2009. Development of biomass
feed materials by sieving. In: ASABE Standards, 608. American Society of torrefaction technology to produce biocoal for electricity generation. Final
Agricultural and Biological Engineers, St. Joseph, MI.. Project Report. Georgia Traditional Industries Program – Forest Resources &
ASTM, 1998. ASTM Standards D 3173–87 – Standard test method for moisture in Products (TIP-FRP), Atlanta, GA.
the analysis sample of coal and coke. Annual Book of ASTM Standards, Section 5, Mani, S., Tabil, L.G., Sokhansanj, S., 2004. Grinding performance and physical
vol. 05.05. American Society for Testing and Materials, West Conshohocken, PA, properties of wheat and barley straws, corn stover and switchgrass. Biomass
pp. 301–302. and Bioenergy 27 (4), 339–352.
ASTM, 2002a. ASTM- D5142–04 Standard test methods for proximate analysis of Mani, S., Tabil, L.G., Sokhansanj, S., 2006. Effects of compressive force, particle size
the analysis sample of coal and coke by instrumental procedures. Annual Book and moisture content on mechanical properties of biomass pellets. Biomass and
of ASTM Standards, Section 5, vol. 05.06. American Society for Testing and Bioenergy 30 (7), 648–654.
Materials, West Conshohocken, PA. Mansouri, N.E.E., Salvadó, J., 2006. Structural characterization of technical lignins
ASTM, 2002b. ASTM- D3176–89 – Standard practice for ultimate analysis of coal for the production of adhesives: application to lignosulfonate, kraft, soda-
and coke. Annual Book of ASTM Standards, Section 5, vol. 05.06. American anthraquinone, organosolv and ethanol process lignins. Industrial Crops &
Society for Testing and Materials, West Conshohocken, PA. Products 24 (1), 8–16.
ASTM, 2003. ASTM D5865–03 – Standards test methods for gross calorific value of Milosavljevic, I., Suuberg, E.M., 1995. Cellulose thermal decomposition kinetics:
coal and coke. Annual Book of ASTM Standards, vol. 05. American Society for Global mass loss kinetics. Industrial & Engineering Chemistry Research 34 (4),
Testing and Materials, West Conshohocken, PA, pp. 517–527. 1081–1091.
ASTM, 2006. ASTM E 873–82 – Standard test method for bulk density of densified Pach, M., R. Zanzi, E. Björnbom. 2002. Torrefied biomass a substitute for wood and
particulate biomass fuels. In: Annual Book of ASTM Standards. American Society charcoal. In: 6th Asia-Pacific International Symposium on Combustion and
for Testing and Materials, West Conshohocken, PA, pp. E871–E872. Energy Utilization May 20, 2002, Kuala Lumpur.
Bergman, P.C.A., Boersma, A.R., Zwart, R.W.R., Kiel, J.H.A., 2005. Torrefaction for Perlack, R.D., L.L. Wright, A.F. Turhollow, R.L. Graham, B.J. Stokes, D.C. Erbach. 2005.
Biomass Co-firing in Existing Coal-Fired Power Stations, ‘‘Biocoal”, ECN-C-05– Biomass as feedstock for a bioenergy and bioproducts industry: the technical
013. The Netherlands, Energy Research Center of the Netherlands (ECN). feasibility of a billion ton annual supply. Report prepared for DOE under
Esteban, L.S., Carrasco, J.E., 2006. Evaluation of different strategies for pulverization contract DE-AC05-000R22725. 59. p. 2005.
of forest biomasses. Powder Technology 166 (3), 139–151. Phanphanich, M., Mani, S., 2010. Drying characteristics of pine forest residues.
Felfli, F.F., Luengo, C.A., Suarez, J.A., Beaton, P.A., 2005. Wood briquette torrefaction. Bioresources 5 (1), 108–120.
Energy for Sustainable Development 9 (3), 19–22. Pimchuai, A., A. Dutta and P. Basu. 2010. Torrefaction of agricultural residues to
Ferro, D.T., V. Vigourous, A. Grimm and R. Zanzi. 2004. Torrefaction of agricultural enhance combustible properties. Energy Fuels, doi:10.1021/ef901168f.
and forest residues. Cubasolar 2004 Conference Publication 4. Guantanamo, Prins, M. J. 2005. Thermodynamic analyses of biomass gasification and torrefaction.
Cuba. Ph.D. Thesis, Eindhoven. Eindhoven University of Technology, The Netherlands.
Himmel, M., Tucker, M., Baker, J., Rivard, C., OH, K., Grohmann, K., 1985. Prins, M.J., Ptasinski, K.J., Janssen, F.J.J.G., 2006. More efficient biomass gasification
Comminution of biomass: hammer and knife mills. Biotechnology and via torrefaction. Energy 31, 3458–3470.
Bioengineering Symposium 15, 39–58. Saastamoinen, J., Aho, M., Moilanen, A., Sorensen, L.H., Clausen, S., Berg, M., 2010.
Hughes, E.A., Tillman, D.A., 1998. Biomass cofiring: status, prospects 1996. Fuel Burnout of pulverized biomass particles in large scale boiler – single particle
Processing Technology 54 (1–3), 127–142. model approach. Biomass and Bioenergy 34 (5), 728–736.
Kaliyan, N., Morey, R.V., 2009. Factors affecting strength and durability of densified Schell, D.J., Harwood, C., 1994. Milling of lignocellulosic biomass: results of pilot-
biomass products. Biomass and Bioenergy 33 (3), 337–359. scale testing. Applied Biochemistry and Biotechnology 45/46, 159–168.
Kiel, J.H.A., van Paasen, S.V.B., Neeft, J.P.A., Devi, L., Ptasinski, K.J., Janssen, F.J.J.G., Shafizadeh, F., Sarkanen, K.V., Tillman, D.A., 1976. Thermal Uses and Properties of
Meijer, R., Berends, R.H., Temmink, H.M.G., Brem, G., Padban, N., Bramer, E.A., Carbohydrates and Lignins. Academic Press, American Chemical Society, New
2004. Primary Measures to Reduce Tar Formation in Fluidized-Bed Biomass York.
Gasifiers. ECN-C-04–014. The Netherlands, Energy Research Center of the Sokhansanj, S., Mani, S., Turhollow, A.F., Kumar, A., 2009. Large scale production,
Netherlands (ECN). collection and supply of switchgrass (Panicum virgatum L.) – Visioning a mature
Lehtikangas, P., 2001. Quality properties of pelletised sawdust, logging residues and technology. Biofuels, Bioproducts and Biorefining 3, 124–141.
bark. Biomass and Bioenergy 20 (5), 351–360. Tillman, D.A., 2000. Biomass cofiring: the technology, the experience, the
Lipinsky, E.S., Arcate, J.R., Reed, T.B., 2002. Enhanced wood fuels via torrefaction. combustion consequences. Biomass and Bioenergy 19 (6), 365–384.
Fuel Chemistry Division Preprints 47 (1), 1–3. Van Loo, S., Koppejan, J., 2008. The Handbook of Biomass Combustion and Co-firing.
Lu, H., Ip, E., Scott, J., Foster, P., Vickers, M., Baxter, L.L., 2010. Effects of particle Earthscan Publications, London, UK.
shape and size on devolatillization of biomass particle. Fuel 89 (5), 1156–1168. Xiumin, J., Chuguang, Z., Che, Y., Dechang, L., Jianrong, Q., Jubin, L., 2002. Physical
Lytle, J., Choi, N., Prisbrey, K., 1992. Influence of preheating on grindability of coal. structure and combustion properties of super fine pulverized coal particle. Fuel
International Journal of Mineral Processing 36 (1–2), 107–112. 81 (6), 793–797.
Mani, S. 2005. A systems analysis of biomass densification process. Ph.D. Zwart, R.W.R., Boerrigter, H., van der Drift, A., 2006. The impact of biomass
Dissertation. Department of Chemical and Biological Engineering, University pretreatment on the feasibility of overseas biomass conversion to Fischer–
of British Columbia, Vancouver, Canada. Tropsch products. Energy Fuels 20 (5), 2192–2197.