Experimental investigation of continuous torrefaction conditions of biomass residues for the subsequent use of torrefied pellets in domestic and district heating systems

1. Experimental investigation of continuous torrefaction conditions of biomass residues for
the subsequent use of torrefied pellets in domestic and district heating systems
Dr Jean- Bernard Michel, Professor, Professor & Head, Industrial Bioenergy Systems Unit, HEIG-VD, University
of Applied Science Western Switzerland, Avenue des Sports, 20 – 1401 Yverdon-les-Bains, Switzerland. +41 24
55 77 594 jean-bernard.michel@heig-vd.ch
Mark McCormick, Industrial Bioenergy Systems Unit, HEIG-VD, University of Applied Science Western
Switzerland, Avenue des Sports, 20 – 1401 Yverdon-les-Bains, Switzerland. +41 24 55 76 151,
mark.mccormick@heig-vd.ch
Abstract
Torrefaction of woody biomass has received considerable interest over the last few years since the work done by
Mark Jan Prins and his Ph.D thesis in 2005 [1]. This is mainly because of the high quality of torrefied pellets
compared to untreated biomass: hydrophobicity, high energy density, and high process efficiency. Torrefaction of
lower quality biomass, such as urban trimmings and forest residues, instead of wood chips provides an additional
economic benefit and this is the basis of the reported project.
This paper summarizes results obtained using a 6 kg/h experimental facility composed of a horizontal screw-
conveyor type reactor and a porous burner. The purpose of the work was several folds:
- to investigate torrefaction conditions of urban trimmings and forest residues,
- to investigate the combustion of torrefied pellets in commercial pellet stoves,
- to investigate combustion conditions of the gaseous products of torrefaction (Torgas) and the composition
of the flue gas.
The results were used to define the heat balance of the torrefaction process and forecast the overall efficiency of a
commercial scale unit. In particular it is found that with a biomass input moisture content of 30% (relative), and
heat losses lower than 5%, the process may be thermally autonomous, as the energy content of the Torgas is
sufficient for drying.
Results also show that Torgas can be burned in a porous burner when mixed with propane and air. Another
combustion system, based on a lean-gas FLOX® burner, will be used at the pilot scale.
Finally, the comparison of the combustion properties of torrefied pellets vs normal pellets was carried out in a
commercial pellet stove. Combustion properties were found to be very similar but some results obtained with
higher calorific value pellets (30% volatile mass loss) suggest that an air-fuel ratio control system will be needed
or a change of set point for the screw-feeder.
This laboratory phase led to a larger pilot-scale project (100 kg/h) that has just started with the financial support
of the State of Vaud within the framework of the program "100 millions pour les énergies renouvelables et
l'efficacité énergétique".
1 Introduction
Torrefaction is one potential treatment process and only one of many steps required to sustainably exploit woody
biomass resources to provide heat and power. In many countries, a large quantity of garden trimmings are collected,
centralized and burned or transformed into compost. In Switzerland, for example, community recycling centres
receive approximately 2000 m3
/year of woody biomass. Additionally, large quantities of forest residues are
available in Switzerland and most European countries [2]. Due to the lack of a suitable transformation process,
these resources are underexploited and in the case of garden trimmings their management is a cost to the local
communities. Moreover, the fact that heating is necessary only during the winter results in a lack of demand and
a biomass storage problem during the summer months. Torrefaction has been proposed as a solution to these
problems. As a biomass fuel upgrading technology, the energetic advantages of torrefaction are shown by
comparing the net calorific values of 1 kg of wet raw materials to the torrefied product.

2. 2
Table 1: Net calorific value increase during torrefaction
Raw material calorific value, 47% moisture (wet basis) 1
MJ/kg (NCV) 8.6
Raw material, wet mass kg 1.0
Energy in 1 kg of raw material MJ (NCV) 8.6
Raw material, dry mass kg 0.53
Mass loss during torrefaction % dm 22
Torrefied product dry mass kg 0.41
Torrefied product, wet mass, 3% moisture kg 0.43
Torrefied product calorific value, 3% moisture MJ/kg (NCV) 20.6
Energy in the torrefefied product MJ (NCV) 8.8
In the above example, torrefaction results in an increase of the net energy content of the raw material. Up to a
limit in the moisture content of the raw material, the only thermal energy cost of torrefaction is the dry mass loss
by the raw material. In practice, the upper limit of moisture content of the raw material is less than 47% due to the
need to account for heat lost to the surroundings and not available for drying.
The case for torrefaction includes process thermal efficiency, efficient exploitation of the gases produced during
torrefaction, efficient combustion of the solid fuel product and demonstration of the inoffensiveness of solid and
gaseous emissions. In this paper we summarize the results of work conducted in our lab that have led to our current
project to develop a complete lignocellulose biomass upgrading solution integrating resource logistics, drying,
torrefaction, torrefaction gas combustion, system automation and combustion of the solid fuel product. The system
requires combustion of the Torgas to produce heat for drying and also to assure compliance with air emissions
regulations. Due to its variable composition, low calorific value and presence of sulphur, chlorine and condensable
compounds, combustion of torrefaction gas presents technical challenges. In this paper we report selected results
related to the torrefaction process and to the combustion of the Torgas as well as to the combustion of torrefied
biomass pellets in a commercial stove.
Previous combustion tests had been carried-out on a commercial 50 kW pellet boiler using BO2™ pellets supplied
by ECN [3]. It was found that with no modification to the feeding and the burner parameters, the ignition and
combustion characteristics of torrefied poplar pellets were very similar to those of normal poplar pellets.
Particulate emissions per energy output were found very close and directly related to the ash content in the
feedstock. Furthermore, using the Impact 2002+ life cycle impact assessment (LCIA) method, an overall gain of
50% compared to normal pellets was demonstrated, due mainly to the improvement of the overall process
efficiency.
2 Torrefaction of urban trimmings and forest residues
2.1 Torrefaction reactor
As shown on Figure 1, the reactor is composed of a feed hopper, rotary airlock valve, and a 2000 mm long x 150
mm diameter screw-conveyor surrounded by a co-axial insulated shell. Heating is entirely by conduction through
the wall of the screw conveyor tube. Heat is supplied by electric blowers that force hot air to the chamber
surrounding the screw-conveyor. Heating power is regulated based on the weighted average temperatures
measured in the torrefaction gas and on the outer surface of the screw-conveyor tube. Gas produced during
torrefaction is removed via heated pipes connected to the top side of the screw-conveyor and transported to the
burner using a blower.
Mixed deciduous and coniferous wood chips were obtained from local suppliers. A very heterogeneous mixture
of woody garden trimmings, including green leaves, was obtained from an urban waste collection centre. The
material was chopped to obtain chips less than 20 x 10 x 2 mm and then dried to less than 12 % moisture. The
heating rate and the time and temperature of exothermic onset were selected based on the results of batch studies.
1
In the whole text, moisture content is given on a wet basis.

3. 3
Figure 1: photographs of the laboratory torrefaction reactor before and after insulation and connections
Typical torrefaction conditions are given in Table 2
Table 2: Overview of torrefaction conditions
Wood chips Branches
Feed material moisture content 1 - 12 % 1 - 12%
Biomass feed rate 3 – 6 kg/h 3 – 6 kg/h
Temperature of final drying and biomass heating 200 - 250 °C 200 - 245 °C
Duration of final drying and heating 0 - 12 minutes 0 - 12 minutes
Torrefaction temperature 250°C ±10°C 245°C ±10°C
Torrefaction duration 15 - 35 minutes 15 - 35 minutes
The dry mass loss was determined by weighing of the raw material and of the torrefied product and accounting for
moisture as determined according to EN 14774-3 The biomass feed rate was determined by weighing the material
before loading and measuring the time to load the material.
2.2 Determination of specific thermal energy demand and production during continuous
torrefaction
Thermal energy is supplied to the reactor by four electric blowers located outside of the reaction chamber. Blower
thermal power output is regulated individually byvoltage (0-10V). The applied thermal power is calculated by
multiplying the maximum blower power output by regulation voltage/10. The total applied thermal power is the
sum of the individual blower outputs. The thermal energy demand during torrefaction is the difference between
the total thermal energy demands of the empty reactor and of the reactor while continuously loaded with biomass.
The specific energy of torrefaction is calculated from the measured average applied thermal power supplied
divided by the biomass feed rate as follows:
Specific thermal energy demand for torrefaction (kWh/kg) = kW applied/biomass feed rate (kg/h)
Table 3 and show the measured values and results for the continuous torrefaction of branches.
Table 3: Test results used for energy demand determination
Heating blower position Entrance Middle Exit Axial
Empty reactor energy demand Watts (avg) 2315 1555 2359 1681
Loaded reactor energy demand Watts (avg) 3085 1942 1460 1663 Sum
Torrefaction energy demand Watts (avg) 770 386 -898 -19 258
Specific thermal energy demand for torrefaction kWh/kg 0.193 0.097 -0.225 -0.005 0.065

4. 4
Table 4: Feed conditions used for energy demand determination
Biomass feed moisture content % 1.5
Biomass feed temperature °C 20
Biomass feed-rate (dry) kg/h 4.0
Dry mass loss % 22
The power required from the heating blowers in the exit and the axial positions of the reactor is lower than the
power required from the same blowers to maintain the torrefaction temperature of the empty reactor. This decrease
in the power requirement is evidence of exothermic reactions in the reactor. This is the net energy release after
accounting for activation energy demand and including possible combustion of torrefaction gas in the reactor. The
specific overall energy release during thermal treatment and due to devolatilisation or combustion of organic matter
calculated from the measured decrease in thermal energy input to the final heating blower (exit) and the test
conditions described in 2 is thus:
0.225 kWh/kg *3.6 MJ/kWh * 1 kg biomass feed/0.22 kg Torgas = 3.68 MJ/kg dry mass loss.
The percentage of dry mass loss during torrefaction is an adjustable parameter. Since the total heat input increases
with the biomass residence time in the reactor, it is important to determine the residence time beyond which energy
input results in little further gain in mass loss and energy release to the torrefactor. As shown in Figure 2, between
15 and 25 minutes residence time, the dry mass loss increases with residence time.
Figure 2: dry mass loss as a function of residence time in the reactor
Definition of the thermal treatment conditions leading to controlled torrefaction of wood include consideration
of the properties of the wood such as the heat capacity and the activation energy required to initiate
decomposition and volatilization. Manipulated industrial process variables such as the wood particle size, the
oxygen concentration in the surrounding gas phase, the heat flux to the wood and the initial relative humidity of
the wood might be included in the specification of thermal treatment conditions. The response variable was the
exothermic onset temperature which is defined as the temperature at which the rate of temperature increase in
the reactor becomes greater than the reactor heating rate. Using a closed steel cylinder equipped with
thermocouples, loaded with wood chips and placed in a muffle furnace, we evaluated the effect of these
parameters on the exothermic onset temperature and subsequent temperature profile. The experimental design
was made according to the Taguchi method to determine the effects of manipulated parameters on response
variables with a minimum number of experiments [4]. Four factors with three levels were considered for the
experimental analysis. The parameters considered are Heating rate, Particle size, Moisture content and % of air
in the surrounding gas phase. Once the parameters were considered, the Taguchi L9 matrix was constructed and
the levels were set according to possible industrial conditions described in the table below.
0
5
10
15
20
25
30
35
10 15 20 25 30 35
Drymassloss(%)
Minutes

5. 5
Figure 3 Exothermic onset temperature according to factor level
Table 5: Taguchi matrix for the design of experiments
Parameter Unit Level 1 Level 2 Level 3
Heating rate °C/min 3 4 5
Particle size mm <2 2 to 20 20 to 40
Humidity % relative 10 20 30
% air % vol air 0 50 100
As expected, the exothermic onset temperature of small particles is low. However, the largest particle size
fraction did not have the highest exothermic onset temperature. The results suggest that a narrow distribution
of particle sizes might be important to the control of exothermic onset within a narrow temperature range. The
lowest humidity level resulted in a high exothermic onset temperature. Increasing humidity might increases
conductive heat transfer within the particle leading to exothermic onset once the particle is dry. The results show
that the heating rate can be increased without decreasing the exothermic onset temperature. The variation in
average exothermic onset temperature was negligible when the air concentration was increased from 0% to 50%,
but the average onset temperature decreased by 15°C when the air concentration inside the furnace was
increased to 100%. Torrefaction is an anaerobic process. The results suggest that the presence of some air is
acceptable at the start of torrefaction.
3 Analysis of Torgas
3.1 Torgas sampling
.The torgas sampling set-up consists of a tap into the conduit that links the reactor to the burner, a condensing trap
(-20°C), a particle filter and a diaphragm pump to aspire Torgas from the main conduit, a volumetric flow meter
and an infrared analyser of non-condensable gases. The ITES isokinetic sampling system (Paul Gothe GmbH (D))
was used to continuously extract a torgas sample and also to measure the total volume of the gas sample.
Two methods were used to collect samples for determination the composition of Torgas produced during the
torrefaction of branches or wood chips:
1. Collection of the entire Torgas flow during a short period (less than 30 minutes) in order to obtain a sample
for determination of substances present in low concentrations
240.0
242.0
244.0
246.0
248.0
250.0
252.0
254.0
256.0
258.0
260.0
1 2 3
Exothermiconsettemperature(°C)
Factor level (see table below)
Heat flux Particle size Humidity (RH) % air
Global average : 250.1°C

6. 6
2. Collection of a fraction of the total flow during long durations (up to 90 minutes) in order to measure the
variations during the process
Condensable components
A large amount of the Torgas components will condense when the temperature is below 100°C. Around ten grams
per hour of condensate produced from branches or from wood chips were collected by sucking torgas through two
cooling columns maintained at -20°C by means of a cryostat. The collected particulates on the filter (some
hundreds of milligrams) were also saved for subsequent analysis.
Non-condensable components
The non-condensable components were analysed continuously by infrared spectroscopy using the portable process
gas analyser from MRU instruments (model MGA5).The analyser also measures non-condensable alkanes by
NDIR and reports values as CxHy without identifying the component. An indication of the H2 concentration was
obtained using an Industrial Scientific IBRID MX6 gas detector. The instruments were placed downstream of the
filter and condenser were the gas temperature is about 20°C. Data were recorded every second.
3.2 Composition of Torgas
The sample collected by condensation of the entire gas flow during torrefaction of branches was diluted 100 or
1000 times and analysed by head space GC-MS. The semi-volatile organic compounds (aromatic compounds
having boiling points greater than 100°C) made up less than 0.5% of the total mass of the condensate. The identity
of the components was determined by comparison of the spectra to the NIST02 library and the concentration was
expressed as the relative surface area of the individual component peeks to the total surface of the peeks. Results
of the major high molecular weight components found in the condensate are reported in Table 5 as percentages of
the semi-volatile fraction of the torgas.
Table 6: High molecular weight components of torrefaction gas
Component % mass of total
semi-volatile compounds
1,3,3-trimethyl-bicyclo[2.2.1]heptane-one 3.5
4-methyl-1-1(methylethyl)-bicyclo[3.1.0]hexane-3-one 5.7
Thujone 4.8
Borneol 3.3
2-methoxy-4-methyl-phenol 9.6
2,6-dihydroxyacetophenone 4.2
4-hydroxy-2-methyl-acetophenone 3.1
2,6-dimethoxyphenol 19.0
Vanilin 2.9
4-hydroxy-3-methoxybenzoic acid 4.0
2-methoxy-4(1-propenyl)phenol 5.3
4-hydroxy-3-methoxybenzeneacetic acid 8.3
2,3,5,6 tetrafluoroanisole 1.6
2,6-dimethoxy-4(2-propenyl)phenol 4.2
1-(2,4,6-trihydoxyphenyl)-2pentanone 2.8
Tetradadecanoic acid 7.4
4,4'-(1-methylethylidene)bisphenol 5.9
The non-condensable gases found in Torgas are CO, CO2, and possibly light alkanes, NO and H2. In Table 6, we
report the typical ranges observed during torrefaction of branches between 240 and 260°C.

7. 7
Table 7: Typical measured values of non-condensable components in torrefaction gas
Gas Unit Range of typical values (STP)
CO2 % volume 1 to 5
CO mg/Nm3
1000 to 20’000
CxHy mg/Nm3
10 to 750
NOx mg/Nm3
< 4
H2 ppm 250 to 1000
The Gross calorific value of the condensate was determined by using a Parr 2.0 bomb calorimeter. The accuracy
of GCV measurement with this set-up is XX. Total organic carbon was determined using the Hach Lange
spectrophotometric method LCK 387 for determination of TOC in wastewater in conformity with UE-1484.
Total carboxylic acid functional groups, expressed as acetic acid, were determined using using the Hach Lange
method LCK 365 for determination of organic acids in wastewater by esterification and spectrophotometry. A
control of the method was made by analysing a mixture of acetic acid, acetone, isopropanol and water. The
measured result (112 g/l) was within 6 % of the spiked amount of acetic acid (119 g/l). The reported values in
Table 7 include water present in the condensate.
Table 8: Characteristics of the condensate
Measured characteristic Unit Measured values
Density g/l 993 to 1005
Gross calorific value MJ/kg 2,36 to 3,91
Total organic carbon g/l 65 to 95
Organic acids g/l as acetic acid 112
Particles trapped on fiberglass filter g/nm3
torgas 1 to 4
Semi-volatile aromatic compounds % of the total mass of the condensate organic
fraction
< 0.5
Reaction water % total mass of condensate (calculated from CO2
in torgas and not including evaporated
constitutive water)
25±20
All organics % of dry mass loss during torrefaction 12 ±10
3.3 Calorific value of torgas
It is important to know the calorific value of the Torgas in order to establish a correct energy balance of the overall
torrefaction system. The presence of solid, liquid and gaseous components at ambient temperature makes it
difficult to use a bomb calorimeter for this determination.
Two methods were used for this purpose:
1. Using the energy balance of the raw material and the solid product
2. Using the Torgas composition and the process temperatures
The first method requires determination of the Gross Calorific Value (GCV) of the raw materials and of the
torrefied product using a bomb calorimeter and measurement of the dry mass loaded and the mass of the torrefied
product.
The total energy of the raw materials loaded and of the torrefied product is obtained by multiplying the specific
energy by the mass. The difference obtained by subtracting the total energy of the solid product from the total
energy loaded = the total energy lost to the surroundings or contained in the torgas. The product of the dry mass
of raw materials and the percentage of dry mass loss gives the dry mass loss as volatile components produced

8. 8
during torrefaction. The total energy lost divided by the total dry mass loss gives the energy density of the torgas
in terms of MJ/kg dry mass loss. This energy is available to heat the torrefaction process.
The results of the torrefaction of branches with a dry mass loss of 22.2% are summarized in Table 8.
Table 9: Estimation of the torgas gross calorific value (branches)
Torrefaction of branches Masse Moisture Dry mass Water Measured energy Calculated energy
kg % kg kg GCV, dry (MJ/kg) GCV, dry (MJ)
Feed 10 1.3 9.87 0.13 19.613 193.58
Product 7.67 1 7.59 0.08 21.032 159.70
Torrefaction gas 2.33 2.3 2.28 0.05 14.9 (calculated) 33.88
The results of the torrefaction of wood chips with a dry mass loss of 16.6 % are summarized in Table 9.
Table 10: Estimation of the torgas gross calorific value (wood chips)
Torrefaction of wood chips Masse Moisture Dry mass Water Measured energy Calculated energy
kg % kg kg GCV, dry (MJ/kg) GCV, dry (MJ)
Feed 22.28 4.1 21.36 0.92 20.01 427.41
Product 17.82 1.0 17.64 0.18 21.127 372.72
Torrefaction gas 4.46 16.6 3.72 0.74 14.7 (calculated) 54.69
Using the method described above, the calorific value of torgas was determined for samples of 38 separate
torrefaction runs. The mean gross calorific value of torgas was found to be 13.3 MJ/kg (SD = 2.1) dry mass loss
during torrefaction. See Figure 5.
Figure 4 Torgas calorific value versus dry mass loss
This energy is distributed among the sub-processes and presented using the second method which considers
torgas composition and temperature. The second method of describing the energy content of the torgas.
requires listing the energy fractions during torrefaction and stating their values. The Torgas calorific value as
potential chemical energy represents the difference in calorific value between the raw biomass and the torrefied
biomass minus the heating requirements, the activation and reaction energies of the volatile components
0
2
4
6
8
10
12
14
16
18
20
0 5 10 15 20 25 30 35 40
Energy(MJGCV/kgdrymassloss)
Dry mass loss (%)
Torgas specific energy versus dry mass loss during
torrefaction

9. 9
(overall exothermic) and the internal energy of the hot torgas. Results from the continuous torrefaction of pre-
dried wood chips are summarized in Table 10.
Table 11: Energy fractions of torgas produced from dry wood chips
Biomass heating to torrefaction temperature MJ 0.18
Water heating to boiling point MJ 0.02
Latent heat of vaporisation MJ 0.11
PCS condensate MJ 0.74
Activation energy MJ 0.33
Devolatilisation MJ 0.81
Torgas enthalpy (250°C) MJ 0.08
Thermal losses to the surroundings (25%) MJ 0.98
Sum MJ 3.24
To determine the specific energy per kilogram of mass loss (torgas produced), the sum of the energies
determined per kg of dry biomass loaded is divided by the mass loss.
3.24 MJ/0.22 kg mass loss = 14.73 MJ/kg torgas.
This value should be equal to the value obtained by subtracting the energy contained in the torrefied product
from the energy contained in the raw materials. In fact the difference is 1.41 MJ/kg torgas and the values
obtained by the two methods are within 10% of each other.
Using the above results the total energy available is compared to the total thermal energy demand for
torrefaction of 1 kg of biomass having 35% moisture content and 22% dry mass loss during torrefaction.
See Table 11-Table 14
Table 12: Raw material
Torgas Gross Calorific Value MJ/kg mass loss 13.3
Raw material moisture content % 35
Dry biomass Kg 0.65
Mass of water Kg 0.35
Dry mass loss % MS 22
Dry mass loss Kg 0.14
Torgas total energy MJ 1.91
Table 13: Biomass drying
Biomass heating 15 to 100°C MJ 0.124
Water heating 15 to 100°C MJ 0.124
Latent heat of evaporation of water MJ 0.791
Total heat demand for drying MJ 1.040
Excess energy after drying MJ 0.865
Table 14: Torrefaction
Biomass heating 100 to 250°C MJ 0.219
Activation energy MJ 0.325

10. 10
Thermal losses to surroundings (15%) MJ 0.29
Supplementary heat demand for torrefaction MJ 0.83
Table 15: Summary
Total heat demand MJ 1.87
Excess energy MJ 0.04
The total thermal and chemical potential energy available in the torgas is approximately 2% greater than the energy
requirement for drying and torrefaction of biomass having a 35% initial moisture content.
4 Combustion of torrefaction gases
Some of the data presented below was produced as part of several master thesis works [4-6]
4.1 Porous burner set-up
It was not found possible to burn such a low quality gas with a Bunsen burner. On the other hand, it was known
that porous burner combustion allows to burn gases of low energy quality. Such a burner improves the homogeneity
of the gas temperature andthe mass flow compared to a diffusion flame and increases the residence time of the fuel
gases in the combustion zone.
In order to maintain the Torgas temperature at 240°C, the Torgas supply line between the reactor exhaust and the
burner were heated electrically and thermally insulated. The experiments have shown that this temperature is
sufficient to prevent condensation of light tar components that are present.
It must be noted that the solution of electrical heating is only for the laboratory. Other more efficient methods will
be used at the pilot and industrial scale.
The porous burner is composed of an alumina plate on the bottom and a porous matrix above. The alumina plate
is perforated with a number of small holes acting as a flame arrester.
The porous foam, shown on Figure 6, is made of silicon carbide (SiC) and was supplied by the company Erbicol
in Balerna (CH). It has a rectangular shape having the dimensions of 80 mm x 80 mm x 16 mm. The foam
porosity is 87% (ratio of empty volume/total volume)
Figure 5 : Photograph of the porous matrix
The combustion chamber periphery is covered by insulating material, except for the glass window allowing to
check that the combustion is established inside the porous matrix. Leakages are arrested by filling wi t h cement.
The igniter is an electrically heated rod and is placed above the porous material for starting combustion. The
temperature of the igniter is controlled by varying the voltage of the rheostat. The temperatures in the burner are
measured by 3 thermal couples placed respectively on the bottom and the top side and 40 cm above the porous
plate. During the experiment, the glass window is replaced by an aluminium plate having a hole for inserting
the gas analyser probe so that exhaust gas emissions can be measured. A propane inlet to the conduit carrying torgas
is positioned 30 cm upstream from the burner.
During initial experiments, the combustion area was varied (by means of a cover) and the flows of air, propane,
and Torgas were varied. Finding a flow meter suitable for the measurement of Torgas is difficult to its high

11. 11
temperature and the presence of water vapour, organic acids and condensable tars. For Torgas flow rate
measurement a venturi was designed according to European Standards (NF EN ISO 5167-4, 2003). It was then
installed and calibrated using air at ambient temperature. The calibration curve is shown on Figure 7, showing
good agreement with theory. The Torgas volume flow was then calculated with a correction for its temperature
and density.
Figure 6: Venturi calibration curve
4.2 Flame stabilization and pollutants
The flame stabilization and propagation in an inert porous medium such as SiC foams are governed by the modified
Peclet number [7]. Combustion is possible if the Peclet-number is high enough ( > 65), so that quenching of the
flame inside the pores is prohibited.
The modified Peclet Number is defined by:
Pe=SL Dm*Cp*ρ/k
Where SL is the laminar flame speed, Dm is the equivalent diameter of the average hollow space of the porous
material, Cp is the specific heat of the gas mixture, ρ is the density of the gas mixture and λ is the thermal
conductivity coefficient of the gas mixture.
The following table provides the typical values obtained, measured or estimated, showing that the Peclet number
is above the critical value.
Table 16: Calculation of the Peclet number
Symbol Property Value
SL Laminar flame velocity of Torgas (m.s-1
) 0.56
Dm Pore diameter (m) 0.0040
Cp Heat capacity of Torgas (J.kg-1
.°C-1
)
1500 (estimated using NIST
database)
Ρ Torgas density (kg.m-3
) 0.92
K Thermal conductivity of Torgas (W.m-1
°C-1
)
0.04 (estimated using NIST
database)
14
12
Air flow
10
8
Air flow
(theoretical)
6
4
2
0
0 50 100 150 200
Pressure difference (Pa)
Airflow(m3/h)

12. 12
Pe Peclet number of the porous foam 77
4.3 Combustion results
Stable combustion of torrefaction gas only was not successful. Also, the presence of oxygen in the Torgas made it
sometimes difficult to adjust the correct flow of air to the burner.
However, stable combustion of a mixture of propane, air and torrefaction gas produced during continuous
torrefaction of branches was achieved. The operating conditions are summarized in the following table:
Table 17: Summary of the most stable burner operating conditions
Torgas flow rate kg dry mass/h 0.72
Air flow rate kg/h 1.78
Propane flow rate kg/h 0.09
Torgas temperature °C 240
Temperature in porous burner °C 850
Typical Torgas to propane ratio kg dry mass/kg propane 7.83
Air/Fuel ratio kg air/(kg tg + kg propane) 2.20
Propane power kW 1.18
Torgas power kW 0.60
For each test, the flue-gas composition was measured using the MRU instrument for O2, CO2, CO, NOX, CH4 and
CxHy.
The Oswald diagrams were established for each successful test and a good example of such diagram is shown on
Figure 8. From this diagram, one could conclude that the value of CO2max (at stoichiometry) was in the interval
16%-17%, which is similar to that of light-fuel oil (15.6 %) or other heavier hydrocarbons.
Figure 7: Oswald diagram obtained in one test (Torgas from branches)
y = -0.84x + 18
R² = 0.98
-10
-5
0
5
10
15
20
25
-5 0 5 10 15 20 25 30 35
CO2(%vol.)
O2 (% vol.)

13. 13
The combustion could be nearly complete combustion with 17 ppm CO in the exhaust gas achieved with a mixture
of 0.1 kg/h propane, 1.8 kg/h air and 0.7 kg/h Torgas. It was demonstrated that CO and NO2 concentrations in the
flue gas could be maintained below the Swiss limit for gas combustion of 100 and 80 mg/m3
respectively
normalized to 3% O2 in the exhaust gas. This was possible when the combustion was well established inside the
porous zone and the burner temperature was above 800°C.
5 Combustion of torrefied pellets
5.1 Experimental set-up
Torrefied pellets were produced, as previously explained, in sufficient quantities for testing in a commercial pellet
stove. The tests were carried out at the project partner company Wit SA on a pellet stove from the manufacturer
EdilKamin (Italy) on their model “Funny” [11]. In this stove, the ignition is provided by air heated electrically.
The combustion air flow is automatically regulated by adjustment of the flue-gas temperature using an extraction
fan.
Continuous supply of pellets is achieved automatically with a screw feeder. Using normal pellets, the nominal
power is 11 kW and can be manually turned down to 5.5 kW. No modification was made to the stove when burning
torrefied biomass pellets.
Flue-gas was sampled continuously during the tests and analysed with the following equipment:
• Gas analysis: MGA5 from MRU Instruments for O2 , CO, CO2, NO, NO2, CxHy
• Particulates: The isokinetic sampling unit (ITES) from the company Paul Gothe GmbH (D)
Photographs of the stove set-up are shown on Figure 9, showing the acquisition system, the MRU unit (left) and
the sampling probes (right)
Figure 8 : Photographs of the pellet stove set-up
5.2 Comparison of combustion emissions
The combustion emissions of the following fuel cases were evaluated:
1. Torrefied branches produced during the project with a mass loss lower than 20%
2. Commercial wood pellets from the company AEK.
3. Torrefied branches produced during the project with a mass loss higher than 20%
4. Torrefied poplar pellets (BO2™) produced by ECN (NL) on their 70 kg/h pilot plant

14. 14
Table 18: Combustion test results
Case 1
Torrefied
branches
Mass loss
<20%
Case 2
Wood Pellets
AEK
Case 3
Torrefied
branches
Mass loss >20%
Case 4
Torrefied
poplar ECN
Amount burned kg 2.555 0.803 0.873 0.701
Combustion duration minutes 155 50 49 37
Combustion rate kg/h 0.99 0.96 1.07 1.14
Ash collected g 49.1 2.3 28.8 4.4
Ash collection rate kg/h 0.019 0.003 0.035 0.007
Ash production g/kg burned 19.2 2.9 33.0 6.3
Ash elementary
analysis
See Table 21 See Table 21 See Table 21 See Table 21
Ignition delay Slow
(2 minutes)
Lot of smoke
Fast
(< 1 minute)
Little smoke
Very slow
(3 minutes)
Lot of smoke
Slow
(2 minutes)
Smoke
Flame quality
following stabilisation
Good
Little smoke
Good
Little smoke
Good
Little smoke
Good
Little smoke
Flue-gas analysis for CO2, NO, NOx and CxHy are presented in Figure 10 andFigure 11. The values are normalized
for the flue-gas volume calculated at 13% O2. Observed CO peaks correspond to the change in power with a factor
of 1 to 5 in power increase.
Figure 9 : Flue gas concentrations for case 1
0
50
100
150
200
250
300
350
400
0
500
1000
1500
2000
2500
3000
3500
4000
11:31:12 12:00:00 12:28:48 12:57:36 13:26:24 13:55:12
NO/NOx/CxHy(mg/m3)
CO(mg/m3)
Heure
Torrefied Branches
Mass loss < 20% - O2min=16.6%
CO/mg/m3: NO/mg/m3: NOx/mg/m3: CH4/mg/m3:

15. 15
Figure 10 : Flue gas concentrations for case 2
Figure 11 : Flue gas concentrations for case 3
0
50
100
150
200
250
300
350
400
14:38:24 14:45:36 14:52:48 15:00:00 15:07:12 15:14:24 15:21:36 15:28:48 15:36:00
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
NO/NOx/CxHy(mg/m3)
CO(mg/m3) AEK pellets - O2min=11.7%
CO/mg/m3: NO/mg/m3: NOx/mg/m3: CH4/mg/m3:
0
50
100
150
200
250
300
350
400
15:43:12 15:50:24 15:57:36 16:04:48 16:12:00 16:19:12 16:26:24 16:33:36 16:40:48 16:48:00
0
500
1000
1500
2000
2500
3000
3500
4000
NO/NOx/CxHy(mg/m3)
CO(mg/m3)
Torrefied Branches
Mass loss > 20% - O2min =11.6%
CO/mg/m3: NO/mg/m3: NOx/mg/m3: CH4/mg/m3:

16. 16
Figure 12 : Flue gas concentrations for case 4
Particulate emissions are given in Erreur ! Source du renvoi introuvable.. Power 1 corresponds to 5.5 kW and
power 5 to 11 kW (based on pellet flow-rate)
Figure 13 : particulate emissions measured in the flue-gas
Discussion:
One can observe that the trends are quite similar for the four cases with increased CO emissions for cases 3 and 4
compared to 1 and 2, this may be caused by higher calorific value of the cases 3 and 4, due to a higher mass loss
compared to the two other cases. As the stove does not regulate the air/pellet ratio, the excess air in the primary
combustion zone was lower for cases 3 and 4. However, this does not correlate to the values of oxygen measured
in the chimney: those were not plotted but instead, the minimum recorded values are given in the title; it must be
noted that this is not a laboratory installation and results should be interpreted with caution since a variable air
0
50
100
150
200
250
300
350
400
16:48:00 16:55:12 17:02:24 17:09:36 17:16:48 17:24:00 17:31:12 17:38:24
0
500
1000
1500
2000
2500
3000
3500
4000
NO/NOx/CxHy(mg/m3)
CO(mg/m3) ECN torrefied poplar- O2 min= 15.2%
CO/mg/m3: NO/mg/m3: NOx/mg/m3: CH4/mg/m3:
0
50
100
150
200
250
300
350
Branches ML
<20%
AEK pellets Branches ML >
20%
ECN pellets
mgparticles/m3
Power 1
Power 5

17. 17
dilution may have occurred before the point of measurement. Also, the normalised NOx values are quite different
in the four cases and this may be the result of a variable amount of fuel-bound nitrogen in the pellets with higher
values in branches than in forest wood. This could not be due to the effect of a variable temperature that was
relatively constant between the various cases; as a matter of fact, the regulation of the EdelKamin pellet stove is
maintaining a constant temperature.
Also, particulate emissions obtained with case 1 are quite close to those of AEK wood pellets, despite the fact that
the ash content of branches are much higher. This may indicate that most of the ash was deposited in the ash bin.
However, this is not the case with pellets of higher calorific value (case 3 and case 4); one explanation may be the
same as previously given for the CO emissions: the excess air was too small in the primary combustion zone,
leading to an increased number of unburnt particulates.
A solution for this may be to reduce the rotation velocity of the screw in relation to the increase in calorific value.
For stoves equipped by an air flow control (lambda probe) this would not be a problem because the air flow would
be automatically adjusted. This was already demonstrated during the previous test on a pellet boiler [3]
A summary of the flue-gas analysis results is given in Table 18. The Swiss allowed emissions limits are defined
in the Ordinance on Air Pollution Control (OAPC) [12] and in EN14785 for Room heaters fired by wood pellets.
The other European standards, EN 303-5 or EN 12809, are applicable to boilers fired by wood pellets,
automatically stoked, which is not the case here, but was the case on our previous trials [3]
There are no emission limit set in those standards for NOx and for total organic carbon applicable to room heaters
fired with wood pellets and to boilers below a certain size. For large boilers, the NO2 limit is set at 250 mg/m3 for
NOx mass flow greater than 2.5 kg/h (corresponding roughly to power range of 3-5 MW) and the total carbon limit
is set at 50 mg/m
3
for a power greater than 10 MW
5.3 Ash composition
Raw biomass samples, torrrefied samples and ash samples were sent to two Swiss laboratories (Scitec, Lausanne
and Inertek, Schlieren) in order to analyse the metal components and other elements of interest for agriculture.
Results are given in Table 19-Table 21. One can see that some of the metal elements are present in quantities above
limits but that also applies to raw wood chips used commercially. However, it must be pointed out that a new ISO
Standard is being prepared for thermally treated biomass fuels with limit values that may be different. Also, it has
been considered to mix various biomass sources in order to reach an acceptable composition in terms of such trace
elements.

18. 18
Table 19: Summary of flue-gas emissions from the pellet stove (power of 11 kW)
Parameter or component Combustion rate Fly ash Flue gas
temperature
Combustion air
temperature
Particulates Particulates* Cl F
kg dry mass/h kg MS/h °C °C mg.Nm-3 mg.Nm-3 mg.m-3 mg.m-3
Limit OAPC Annex 4, p 72 (EN14785) 40
Fuel type
1. Torrefied branches, mass loss <20% 0.89 0.019 140 20 61.5 98.4
2. AEK pellets 0.89 0.0028 129 20 37.8 100.8
3. Torrefied branches, mass loss > 20% 0.96 0.0353 136 20 142.4 379.7
4. ECN pellets 1.07 0.0071 159 20 59.1 94.6
Parameter or component O2 CO2 CO CO* NOx NO NO2 NO2* CH4 NH3 Total org.CxHy
carbon (FID)
% vol % vol mg.m-3 mg.m-3 mg.m-3 mg.m-3 mg.m-3 mg.m3 mg.m-3 mg.m-3 mg.m-3
Fuel type
1. Torrefied branches, mass loss <20% 16 +/- 0,5 4 +/- 0,5 661 (366 à 1043) 1058 202 +/- 25 130 +/- 15 46 74 15 (9 à 21)
2. AEK pellets 17,6 (16,4 à 20,1) 3,2 (0,8 à 4,4) 752 (325 à 1592) 2005 50 (12 à 70) 31 (8 à 44) 24 64 20 (13 à 50)
3. Torrefied branches, mass loss > 20% 17,9 (16,5 à 20,4) 2,8 (0,3 à 4,1) 2165 (1079 à 3856) 5773 196 (45 à 253) 126 (28 à 163) 52 139 27 (2 à 44)
4. ECN pellets 15,9 (15,1 à 18,3) 4,9 (2,6 à 5,6) 660 (353 à 1329) 1056 140 (84 à 162) 89 (54 à 103) 33 53 17 (11 à 29)
*normalised with an oxygen content in the flue gas of 13%
Table 20: Raw biomass composition of the various cases (Nitrogen and trace elements)
Element S Cl F N P Ca Mg K As Cd Cr Cu Hg Ni Pb Zn Mo Co V
Limit EN14961-1 (raw material) mg/kg 300 200 30'000 <1 < 0.5 < 10 < 10 < 0.1 < 10 <10 < 100
Limit EN14961-1 (raw material) % 0.03 0.02 3
Sample
20130129_Torrefied chips mg/kg < 2.5 < 0.1 2.32 3.21 < 2.5 0.75 <1.0 10.1
20130129_Torrefied branches,YLB mg/kg <1000 339 67 < 2.5 < 0.1 8.86 94.3 < 0.25 5.32 <1.0 54.5
20130226_ Torrefied branches,YLB mg/kg 550 350 40 <0.8 <0.4 42 52 <0.07 5 <2 52
20130226_ Torrefied branches,YLB mg/kg <1000 1660 106 < 2.5 < 0.1 48 17.9 < 0.25 18.4 7.36 46.8
20130227_ Torrefied branches,Bettens mg/kg 430 490 50 <0.8 <0.2 37 14 <0.07 13 6 51
20130227_ Torrefied branches, washed Bettens mg/kg < 2.5 < 0.1 20.6 13.9 < 0.25 6.41 3.25 27.3

19. 19
Table 21: Biomass composition of torrefied biomass (Nitrogen and trace elements)
Element S Cl F N P Ca Mg K As Cd Cr Cu Hg Ni Pb Zn Mo Co V
Limit EN14961-2 (pellets) mg/kg 300 200 3'000 <1 < 0.5 < 10 < 10 < 0.1 < 10 <10 < 100
Sample
20130129_Torrefied chips mg/kg < 2.5 < 0.1 2.26 2.72 < 0.25 1.03 <1.0 30.7
20130129_Torrefied branches,YLB mg/kg <7'000 < 2.5 0.36 8.55 9.69 < 0.25 5.64 1.23 212
20130226_ Torrefied branches,YLB mg/kg < 1000 155 78 <7'000 <1 1.02 8.76 14.5 <0.1 4.53 <1 117
20130226_ Torrefied branches,YLB mg/kg 1600 203 58 <7'000 < 0.08 0.6 13 8 <0.07 3 <2 220
20130227_ Torrefied branches,Bettens mg/kg < 1000 440 155 1 1.59 39.7 8.3 <0.1 15.2 5.49 30.8
20130227_ Torrefied branches, washed Bettens mg/kg 1400 351 <50 <0.8 <0.2 54 18 <0.07 11 11
Table 22: Ash composition when firing torrefied biomass
Element S Cl F N P Ca Mg K As Cd Cr Cu Hg Ni Pb Zn Mo Co V
Limit value for commercial compost in Switzerland 1
1 100 100 1 30 120 400
Average wood ash composition 2
mg/kg 9'400 320'000 27'000 66'000 2,2 35 210 <0,5 45 18 380 4,5 11
Echantillon
1. Torrefied branches, mass loss <20% mg/kg 4'000 378 140 21'260 331'300 23'250 102'900 <1 0.7 225 297 <0.1 122 25.5 860 577 3.9 12.2
3. Torrefied branches, mass loss > 20% mg/kg 2'000 277 102 18'560 448'100 27'250 102'890 <1 0.59 155 169 <0.1 82.5 15.7 473 39.4 2.5 9.56
1 OSubst, ch 221, al. 1
2 Energie-bois suisse, QM standard. Table 2.15.
Cells highlighted in pink indicate a value above limit.

20. 20
6 Process energy flux
A system mass and energy balance model was created using the software tool STAN [14]. Using the measured mass and
gross calorific values of the solid input and the solid product obtained from continuous torrefaction tests and the estimated
heat capacity values for torrefaction gas, exhaust and water vapour at the experimental temperatures, it can be shown that
the torrefaction of branches and wood chips with a dry mass loss of 22% is a thermally self-sufficient process when the
moisture content of the biomass raw material is less than 30%. Following start-up using external heat supplied by
combustion of natural gas (NG in the figures below) or biomass, the process produces enough heat to operate with-out
thermal energy input. Using the experimentally determined mass ratio of Torgas:propane:air it was shown that biomass
having a moisture content up to 45% can be efficiently torrefied (without taking the losses into account)
In practice this value would be lower because only a fraction of the theoretical chemical potential energy contained in the
dry mass of torrefaction gas reaches the burner. Additional heat is lost due to transfer inefficiencies and the release of humid
air from the dryer.
Figure 15: Mass flows (kg/h) for torrefaction of branches, 35% moisture content

21. 21
Figure 16: Energy flows (kJ/s) for torrefaction of branches, 35% moisture content
Conclusions
The work conducted at the laboratory scale demonstrated the technical and energetic feasibility of upgrading wood chips
and lignocellulosic rich garden trimmings and branches to produce a solid fuel that can be burned in conventional wood
pellet stoves. Further work should seek to improve process thermal efficiency in order to increase the moisture content limit
of raw materials that can be torrefied without external energy input.
For wet biomass input having a moisture level less than 35%, the energy released during torrefaction is sufficient to heat
the process when the dry mass loss is 22%. The torrefied product has acceptable combustion properties. Disadvantages of
using branches include high ash content, presence of sulphur, chlorine, heavy metals and possibly halogenated
phytosanitary products. Blending branches with higher quality raw materials like wood chips before torrefaction would
result in a higher quality torrefied product.
7 Acknowledgements
The authors gratefully acknowledge the financial support of the Swiss Federal Institute of Energy and of the fund “Cogener”
of the Services Industriels de Genève as well as Cyril Mahmed and Isabelle Monney for their assistance in the construction,
operation and measurements. Furthermore they are grateful to Kiran Kumar, Bapusaheb Dalavi, Gowri Shankar and Chinta
Sridhar, who provided some of the data as part of their Master of Technology thesis in Thermal Engineering from National
Institute of Technology Karnataka, Surathkal. For this, the cooperation with Prof. Ashok Babu, their professor at NITK
was instrumental.
8 Reference list
1. M. J. Prins. Thermodynamic analysis of biomass gasification and torrefaction. [online]. Technische Universiteit
Eindhoven, 2005. Proefschrift. ISBN 90-386-2886-2. available from: http://alexandria.tue.nl/extra2/200510705.pdf
[accessed 10.10.2014]
2. Procurement of forest residues, Aebiom report. 2007 [online]. Available from: http://www.aebiom.org/wp/wp-
content/uploads/file/Publications/Forest_residues_August2007.pdf [accessed 10.10.2014]

22. 22
3. J.-B. Michel et al. Combustion and life-cycle evaluation of torrefied wood for decentralized heat and power production.
Conference proceedings: 9th European Conference on Industrial Furnaces and Boilers, Estoril, April 2011.
4. S. Fraley et al. Design of experiments via taguchi methods: orthogonal arrays, available from
https://controls.engin.umich.edu/wiki/index.php/Design_of_experiments_via_taguchi_methods:_orthogonal_arrays
[accessed 10.10.2014]
5. D Kiran Kumar, Valorisation of torrefaction by-product gases, MS Thesis, 2013, NITK, Bangalore
6. D Kiran Kumar et al.Continuous combustion of a mixture of propane and torrefaction gas in a porous burner,
proceedings of the 21st
European Biomass Conference and Exhibition (EUBCE), 2013
7. Bapusaheb Dalavi. Experimental investigation of torrefaction-gas combustion in porous burner, MS Thesis, 2014,
NITK, Bangalore
8. Trimis D, Durst F. Combustion ion a porous medium – advances and applications. Combust Sci Technol
1996;121:153–68.
9. C. Keramiotis, M. A. Founti. An experimental investigation of stability and operation of a biogas fueled porous burner.
Fuel 103 (2013) 278–284
10. M. Abdul Mujeebu et al. Applications of porous media combustion technology – A review. Applied Energy 86, (2009)
pp.1365–1375.
11. M.H. Akbari et al. Lean flammability limits for stable performance with a porous burner. Applied Energy, 86 (2009)
2635–2643
12. EdilKamin Pellet burning stove: Funny. Available from:
http://www.edilkamin.ch/en/stufe_a_pellet/stufa_a_pellet_funny.aspx [accessed on 10.10.2014]
13. Ordinance on Air Pollution Control (OAPC) of 16 December 1985 (Status as of 15 July 2010). Available from:
http://www.admin.ch/opc/en/classified-compilation/19850321/index.html [accessed on 10.10.2014]
14. ISO Standard: ISO/CD 17225-8, Solid biofuels -- Fuel specifications and classes -- Part 8: Thermally treated and
densified biomass fuels, Under development, planned for publication in December 2015.
15. STAN version 2.5.1202, Institute for Water Quality, Resources and Waste Management, Vienna University of
Technology. Available from: http://iwr.tuwien.ac.at [accessed on 7.12.2014]