See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/249342001

Gaseous emissions from agricultural biomass combustion: A prediction model

Conference Paper · July 2013

DOI: 10.13031/aim.20131594309

CITATIONS READS
0 149

4 authors, including:

Sébastien Fournel Bernard Marcos

Laval University Université de Sherbrooke
42 PUBLICATIONS   419 CITATIONS    71 PUBLICATIONS   845 CITATIONS   

SEE PROFILE SEE PROFILE

Stéphane Godbout
Institut de recherche et de développement en agroenvironnement
187 PUBLICATIONS   2,200 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

CIGR International Conference: Integrating Agriculture and Society through Engineering - June 14-18 2020 - Quebec City, Canada View project

Modeling and simulation of Li-ion batteries View project

All content following this page was uploaded by Sébastien Fournel on 01 June 2014.

The user has requested enhancement of the downloaded file.

 An ASABE Meeting Presentation

Paper Number: 131594309

Gaseous emissions from agricultural biomass

combustion: a prediction model

Sébastien Fournel1,2, Bernard Marcos1, Stéphane Godbout2, Michèle Heitz1

1
Department of Chemical Engineering and Biotechnological Engineering, Faculty of Engineering, Université de
Sherbrooke, 2500 boulevard de l’Université, Sherbrooke, QC, Canada.
2
Research and Development Institute for the Agri-Environment (IRDA), 2700 Einstein Street, Quebec City, QC,
Canada.

Written for presentation at the

2013 ASABE Annual International Meeting
Sponsored by ASABE
Kansas City, Missouri
July 21 – 24, 2013

Abstract. As the price of the fossil energy resources and the need to reduce the environmental impacts from
energy use increase, biomass fuels have regained interest from Quebec’s agricultural sector. Producing and
burning energy crops at the farm have become strategies to diversify incomes and decrease dependency to
fossil fuels. However, the current absence of emission factors for solid fuel combustion does not allow a
sustainable development for energy purposes. Besides, the variety of existing furnaces and biomasses
complicates the establishment of such reference values. In order to quantify emissions (CO, CO2, NOx, SO2,
CH4, NH3 and HCl) from on-farm combustion of different agricultural biomasses (short-rotation willow,
switchgrass, reed canary grass, etc.), a prediction model was established based on the calculation of chemical
equilibrium of reactive multicomponent systems. Under constant temperature and pressure, this technique has
been judged as relevant for the prediction of product compositions considering inlet conditions in several
operations and chemical processes, particularly gasification. The model was first established for wood
gasification to be able to compare and validate its results with those of existing models from the literature using
the same original data. The model was then adapted to biomass combustion and calibrated with recent results
from combustion tests held in the province of Quebec. The preliminary results of the prediction model using
data from past combustion experiments with wood, willow and switchgrass revealed good agreement between
both measured and predicted values. Other simulation tests are required to increase accuracy of the model.

Keywords. Biomass, combustion, gasification, gaseous emissions, prediction model, chemical equilibrium,
Gibbs free energy.
Introduction
As the price of the fossil energy resources and the need to reduce the environmental impacts from energy use
increase (Demirbas, 2009; Kaltschmitt and Weber, 2006), biomass fuels have regained interest, especially
from Quebec’s agricultural sector in order to develop production systems with lower energy expenses (Brodeur
et al., 2008; Lease and Théberge, 2005). Producing and burning energy crops at the farm have become
strategies to diversify incomes and decrease dependency on fossil fuels.
However, the current absence of emission factors for solid fuel combustion does not allow a sustainable
development for energy purposes (Talluto, 2009). Furthermore, the variety of existing furnaces and biomasses
complicates the establishment of such reference values since previous studies (Obernberger et al., 2006;
Villeneuve et al., 2012) have revealed that gaseous emissions from biomass combustion differ significantly
according to the combustion technology used and the characteristics of the fuel burned (chemical composition,
physical properties, etc.). From an experimental point of view, considering all the biomass possibilities would
become a laborious and expensive work.
To overcome this situation, thermodynamic equilibrium models can become useful engineering tools to assess
how fuel characteristics influence the exit gas composition (Baratieri et al., 2008; Kalina, 2011; Melgar et al.,
2007). Actually, when the chemical composition of biomass and the equilibrium temperature are specified,
thermodynamic models can simply predict the resulting emissions (Ranzi et al, 2011). To achieve this, some of
those models are based on minimization of the Gibbs free energy (Jarungthammachote and Dutta, 2008). This
technique is mainly used for determining the chemical equilibrium composition of reactive multi-component
closed systems under thermodynamic equilibrium (Néron et al., 2012). Therefore, a global minimum of the
Gibbs free energy coincides with the stable equilibrium solution under constant temperature and pressure. The
equilibrium problem is then solved as an optimization problem of a non-linear constrained system that must
satisfy the restrictions of non-negative number of moles and mole balances (Rossi et al., 2011). Lagrange
multiplier method is generally used to compute this constrained optimization problem (Baratieri et al., 2008;
Jarungthammachote and Dutta, 2008).
Even though this theoretical approach has some inherent limitations, it was judged as relevant for the
prediction of product compositions in several operations and chemical processes as gasification (Altafini et al.,
2003; Baratieri et al., 2008; Gautam, 2010; Jarungthammachote and Dutta, 2007, 2008; Kalina, 2011; Melgar
et al., 2007; Néron et al., 2012; Rossi et al., 2009, 2011; Zainal et al., 2011) and combustion (de Souza-
Santos, 2010). The model based on chemical equilibrium by minimization of the Gibbs free energy was first
established for wood gasification to be able to compare and validate its results with those of existing models
from the literature using the same original data. The model was then adapted to biomass combustion and
calibrated with recent results from wood and agricultural biomass combustion tests held in the province of
Quebec in the past few years.

Model development
Gasification and combustion reactions
To develop the model, the chemical formula of biomasses is defined either as CHyOzNa or CHyOzNaSbClc. The
former corresponds to a woody biomass with negligible sulfur and chlorine contents and is implied in the global
gasification reaction for simulation comparison with other gasification models as described later. The latter term
characterizes a general biomass containing carbon, hydrogen, oxygen, nitrogen, sulfur and chlorine. It is
utilized in the combustion reaction. Both gasification and combustion reactions (eqs. 1 and 2, respectively) can
then be expressed as
CHyOzNa + wH2O+ e(O2 + 3.76N2) =
n1H2+ n2CH4 + n3CO + n4CO2 + n5H2O + n6N2 (1)
and

The authors are solely responsible for the content of this meeting presentation. The presentation does not necessarily reflect the official
position of the American Society of Agricultural and Biological Engineers (ASABE), and its printing and distribution does not constitute an
endorsement of views which may be expressed. Meeting presentations are not subject to the formal peer review process by ASABE
editorial committees; therefore, they are not to be presented as refereed publications. Citation of this work should state that it is from an
ASABE meeting paper. EXAMPLE: Author’s Last Name, Initials. 2013. Title of Presentation. ASABE Paper No. ---. St. Joseph, Mich.:
ASABE. For information about securing permission to reprint or reproduce a meeting presentation, please contact ASABE at
rutter@asabe.org or 269-932-7004 (2950 Niles Road, St. Joseph, MI 49085-9659 USA).
2013 ASABE Annual International Meeting Paper Page 1
 CHyOzNaSbClc + wH2O + e(O2 + 3.76N2) =
n2CH4 + n3CO + n4CO2 + n5H2O + n6N2 + n7O2 +
n8NO + n9NO2 + n10N2O + n11NH3 + n12SO2 + n13HCl, (2)
where y, z, a, b and c are respectively the numbers of atoms of hydrogen (H), oxygen (O), nitrogen (N), sulfur
(S) and chlorine (Cl) per atom of carbon (C) in the solid fuel, w and e are respectively the amount of moisture
and dioxygen in air per kmol of feedstock and ni are the numbers of mole of the species i. Those species
correspond to:
1. Dihydrogen (H2);
2. Methane (CH4);
3. Carbon monoxide (CO);
4. Carbon dioxide (CO2);
5. Water vapor (H2O);
6. Dinitrogen (N2);
7. Dioxygen (O2);
8. Nitrogen monoxide (NO);
9. Nitrogen dioxide (NO2),
10. Nitrous oxide (N2O);
11. Ammonia (NH3);
12. Sulfur dioxide (SO2);
13. Hydrogen chloride (HCl).
They were identified as the main products from either solid fuel gasification or combustion. All inputs on the
left-hand side of equations 1 and 2 are supposed entering the combustion system at 25°C. On the right-hand
side, the stoichiometric coefficients ni are unknown and the model consists in evaluating the concentrations of
the species to calculate them.

Assumptions
The resolution of the stoichiometric coefficients ni using a thermodynamic model was based on the following
assumptions:
 The residence time is long enough to achieve thermodynamic equilibrium. This might not be true, yet
the degree of error introduced by this assumption is acceptable and the applicability of this assumption
is confirmed in literature (de Souza-Santos, 2010; Jenkins et al., 2011; Ragland and Bryden, 2011).
The reaction system is then considered working at steady-state conditions with mass flows and
average properties of each input and output stream remaining constant.
 All carbon, nitrogen, sulphur and chlorine contents in biomass are converted into gaseous form. Mass
balances on carbon and nitrogen revealed good agreement with this statement since less than 2.5wt%
of both elements are found in ash (Godbout et al., 2012). Sulfur and chlorine are mainly present in ash
(Obernberger et al., 2006); however their low content in biomass compared to other elements involves
only a small error. An empirical factor to be included in the model could fix the problem.
 All the gaseous products are assumed to behave as ideal gases. This leads to insignificant errors
because gasification and combustion reactions are conducted at high temperature (> 700°C) and low
pressure (1 atm) (Gautam, 2010).
 The products taken into account in equations 1 and 2 are the main products formed during those
processes (Jarungthammachote and Dutta, 2008; van Loo and Koppejan, 2008). Other gases such as
hydrocarbons (CxHy), other than CH4, are assumed negligible.
 Ash in biomass is assumed inert although it holds typically only for reactions less than 700°C.
Agricultural biomasses contain silicon (Si) and potassium (K) as the major mineral content which
lowers ash fusion temperature below 700°C whereas gasification and combustion processes occurs at
temperatures higher than 700°C. Therefore, the relations derived in this study cannot be used

2013 ASABE Annual International Meeting Paper Page 2

 effectively for biomass with high mineral content (Gautam, 2010).

Minimization of Gibbs free energy

The composition of the gas produced at thermodynamic equilibrium can be estimated using different
approaches: kinetic or dynamic models (Gøbel et al., 2007; Ranzi et al, 2011), equilibrium constants (Gautam,
2010; Jarungthammachote and Dutta, 2007; Melgar et al., 2007; Zainal et al., 2001) or Gibbs energy
minimization (Altafini et al., 2003; Baratieri et al., 2008; Jarungthammachote and Dutta, 2008; Kalina, 2011;
Néron et al., 2012; Rossi et al., 2009, 2011). The key advantage of the latter is that it has a more general
application with predictive capability and does not require the selection of appropriate chemical reactions or an
extended set of data to train the model (Baratieri et al., 2008; Néron et al., 2012).
The Gibbs energy minimization method is based on the assumption that the Gibbs energy (G) reaches a
minimum value at thermodynamic equilibrium (Jarungthammachote and Dutta, 2008; Néron et al., 2012).
Considering a closed chemical system with an arbitrary number of species present in one or several phases at
uniform temperature (T) and pressure (P) (not necessarily constant) evolving from a non-equilibrium initial state
to a closer-to-equilibrium final state and restricting the process to constant T and P, all irreversible process
(occurring at constant T and P) evolve in the direction that causes a decrease of G. The equilibrium state of a
closed system is the one for which G reaches a minimum with respect to all possible changes at the given T
and P. The Gibbs energy minimization method, then, consists in writing an expression for G as a function of the
number of moles of the species present and then finding the set of values for the mole number that minimizes
this function, subject to the constraints of mass conservation and stoichiometry. Moreover, the expression for G
can be extended for open systems, in which the number of moles of the species (ni) can vary because of mass
exchange with the surroundings. In this case, it is necessary to introduce a chemical potential (i) that is a
function of the ni moles of the different compounds (Baratieri et al., 2008):
N
G   ni i . (3)
i

If all gases are assumed as ideal gases at one atmosphere, i can be written as

i  G f○,i  RT ln yi  , (4)

where G f ,i is the standard Gibbs free energy of formation of species i (kJ kmol-1), R is the universal gas
○

constant (8.314 kJ kmol-1 K-1), T is the equilibrium temperature (K) and yi is the mole fraction of gas species i
and the ratio of ni and the total number of moles of the reaction mixture (ntot). Substituting equation 4 into
equation 3, G (kJ) becomes
N N
 n 
G   ni G f○,i   ni RT ln i  . (5)
i i  ntot 
The next step is to find the values of ni which minimizes the objective function G subject to constraints on the
allowable ni. The problem is then solved as an optimization problem of a non-linear constrained system that
must satisfy the restrictions of mole balances and non-negative number of moles corresponding to the Karush-
Kuhn-Tucker (KKT) conditions (Néron et al., 2012). Based on this principle, the Gibbs free energy minimization
is generally solved by the Lagrange multiplier method (Koukkari and Pajarre, 2006).
The first constraint of this problem is the elements conservation as follows:
N

a ni
ij i  Aj j = 1, 2, 3,…, k, (6)

where aij is the number of atoms of the jth element in a mole of the ith species and Aj is defined as the total
number of atoms of the jth element in the reaction mixture. The product of the Lagrange multipliers (j) by the
elemental balance constraints is subtracted from G in the Lagrangian function (L), defined as
k
 N 
L  G    j   aij ni  A j  , (7)
j  i 
and the partial derivatives are set equal to zero in order to find the minimum point:

2013 ASABE Annual International Meeting Paper Page 3

  L 
   0 . (8)

 i
n
Equation 8 produced a matrix that has i rows, which are solved simultaneously with the constraints defined in
equation 6. The second constraint enlightens that the solutions ni have to be real and positive numbers so that
0 ≤ ni ≤ ntot. (9)

Thermodynamic properties
o
In order to determine G f ,i , the values of the standard enthalpy of formation (  H
○
f ,i , kJ kmol-1) and standard
o
entropy of formation (  S f , i , kJ kmol-1 K-1) at the equilibrium temperature are needed because

G f○,i  H ○
f ,i  TS f ,i .
○
(10)

Both H f ,i and S f ,i can be computed by the following definitions of enthalpy and entropy changes occurring
○ ○

when a compound is formed stoichiometrically from its stable elements at T:

H ○
f (T )  H comp (T ) 
○
 H
l elem
l l
○
(T ) (11)

and

S ○
f (T )  S comp (T ) 
○
 S
l elem
l l
○
(T ) , (12)

○ ○
where l is the stoichiometric coefficient for element l. Enthalpy ( H , kJ kmol-1) and entropy ( S , kJ kmol-1 K-
1
) in equations 11 and 12 as much for the compound as for its elements are defined by
T
H ○(T )  H ○
f (T0 )  C p T 
T0
(13)

and
T
Cp
S ○(T )  S ○
f (T0 )  T
T0
T , (14)

where C p is the heat capacity at constant pressure for the standard state (kJ kmol-1 K-1) and T0 is the inlet
temperature of reactants (K). Besides, it is important to note that the arbitrary base selected for calculating the
enthalpy value is zero at 25°C in the case of reference elements and thus H ○ f (298.15) equals zero for all
reference elements.
Since heat capacity, enthalpy and entropy are functions of temperature, it would be easier for the model
computation that those properties are described in terms of polynomial equations as follows:

Cp
 a1  a2T  a3T 2  a4T 3  a5T 4 , (15)
R
H ○
f (T0 ) aT aT aT
2
aT b
3 4
 a1  2 0  3 0  4 0  5 0  1 (16)
RT0 2 3 4 5 T0
and

S ○
f (T0 ) a3T0 aT
2
aT
3 4
 a1 ln(T )  a2T0   4 0  5 0  b2 . (17)
R 2 3 4

2013 ASABE Annual International Meeting Paper Page 4

These equations were established by the NASA technical memorandum 4513 (McBride et al., 1993). All
coefficients for equations 15 to 17 can also be found in this report.

Energy balance
The energy balance is introduced as an energy constraint when the studied system can exchange heat with the
environment. In this case, the equilibrium temperature differs from the initial temperature (Néron et al., 2012). If
the heat duty is known, the equilibrium temperature can be obtained from the first law of thermodynamics for
the combustion process:

Qloss  n H
r  react
r r (T0 )  n
p  prod
p H p (T )  H , (18)

where Qloss is the heat loss from the combustion process (kJ) and H r and H p are respectively the enthalpies of
each reactant and each product at the specified temperatures (kJ kmol-1) . Both enthalpies can be calculated
by equation 13. The enthalpy of formation of a solid fuel can be calculated by the equation developed by de
Souza-Santos (2010) as follows

H ○
f , fuel  LHV  n
p  prod
p H○
f ,p , (19)

○
where H f , p is the enthalpy of formation of product p under complete combustion of the solid fuel (kJ kmol-1)

and LHV is the lower heating value of the biomass (kJ kmol-1) .

Solver
The thermodynamic equilibrium model using Gibbs free energy minimization approach as described before has
been developed in Matlab environment. The code first requires entering the necessary input values such as the
characterization of biomass and the initial temperature. The nonlinear programming model, comprising the
objective function to be minimized (eq. 5) and the constraints (eq. 6), is solved by using the fmincon function
contained in Matlab. The fmincon function applies the interior-point algorithm to solve nonlinear programming
problems. In different steps, the algorithm obtains Lagrange multipliers by approximately solving the KKT
conditions. A detailed description of the use of the fmincon function in Matlab can be found elsewhere
(MathWorks, 2013).
For calculating the equilibrium temperature, the initial temperature is assumed and used to perform the
minimization of Gibbs free energy. Knowing the predicted gas composition, the energy balance is calculated.
Depending on the sign of H in equation 18, the reaction temperature is adjusted. The minimization is hence
solved iteratively until H becomes zero. The complete calculation procedure is illustrated in figure 1.

INPUT: MODEL CALCULATION:

Biomass, CHyOzNbScCld
Moisture content, MC , , , ,
START Air/biomass ratio and flows ni by Gibbs free energy minimization
Initial temperature, T0 Constraints satisfied?
Initial estimate of ni, n0 H by energy balance

Adjusted
T

NO

END YES abs(H)<10-1

Figure 1. The calculation procedure.

2013 ASABE Annual International Meeting Paper Page 5

Simulation results and discussion
Validation of the model applied to gasification process
The first version of the prediction model was developed for biomass gasification considering equation 1 and
following the described procedure in the previous section. It was tested by comparing the calculation results
with experimental data reported in literature (Altafini et al., 2003; Jayah et al., 2003). The modeling results with
both data are shown in tables 1 to 4. These tables also include the predicted values of other authors’ models
(Altafini et al., 2003; Jarungthammachote and Dutta, 2007; Melgar et al., 2007) that used the same original
data.
Table 1. The comparison of predicted results with the experimental data from Altafini et al. (2003) on sawdust gasification.
Jarungthammachote
Gas composition Altafini et al. (2003) and Dutta (2007) The present model
(% mol dry basis) Experimental data SynGas model Modified model Original model Modified model
H2 14.00 20.06 18.24 21.69 18.73
CO 20.14 19.70 23.34 23.46 21.39
[a] [a]
CH4 2.31 0.00 1.66 0.03 1.72
CO2 12.06 10.15 9.82 9.57 11.21
N2 50.79 50.10 46.93 45.25 46.96
[a]
The model was calibrated by fixing the amount of methane in the model to a value derived from the experimental results.
Table 2. The comparison of predicted results with the experimental data from Jayah et al. (2003) on rubber wood gasification (MC
= 14.0wt% dry basis)
Jarungthammachote
Gas composition Jayah et al. (2003) The present model
and Dutta (2007)
(% mol dry basis) Experimental data Original model Original model Modified model
H2 12.5 18.03 19.36 17.15
CO 18.9 18.51 18.30 16.74
[a]
CH4 1.2 0.11 0.01 1.21
CO2 8.5 11.43 11.81 13.00
N2 59.1 51.92 50.52 51.90
[a]
The model was calibrated by fixing the amount of methane in the model to a value derived from the experimental
results.
Table 3. The comparison of predicted results with the experimental data from Jayah et al. (2003) on rubber wood gasification (MC
= 14.7wt% dry basis)
Gas composition Jayah et al. (2003) Melgar et al. (2007) The present model
(% mol dry basis) Experimental data Model Model Original model Modified model
H2 15.5 16.4 17.1 23.32 21.28
CO 19.1 18.3 19.3 22.4 21.02
[a] [a]
CH4 1.1 1.1 0.3 0.01 1.16
CO2 11.4 11.1 11.1 9.96 11.02
N2 52.9 53.2 52.3 44.34 45.52
[a]
The model was calibrated by fixing the amount of methane in the model to a value derived from the experimental results.
Table 4. The comparison of predicted results with the experimental data from Jayah et al. (2003) on rubber wood gasification (MC
= 16.0wt% dry basis)
Jarungthammachote
Gas composition Jayah et al. (2003) The present model
and Dutta (2007)
(% mol dry basis) Experimental data Original model Original model Modified model
H2 17.0 18.04 22.29 19.97
CO 18.4 17.86 20.82 19.25
[a]
CH4 1.3 0.11 0.01 1.30
CO2 10.6 11.84 10.76 11.97
N2 52.7 52.15 46.13 47.51
[a]
The model was calibrated by fixing the amount of methane in the model to a value derived from the experimental
results.

Table 1 deals with the computational simulation of a wood waste downdraft gasifier. The sawdust ultimate
analysis results in 52.00wt% C, 6.07wt% H, 41.55wt% O, and 0.28wt% N. The sawdust presents a moisture
content (MC) nearly 10.00% (wt% on wet basis). The comparison test was done by setting the temperature and
the air/sawdust ratio at 1073 K and 1.957, respectively, as reported by Altafini et al. (2003).
Table 1 shows that the predicted values obtained by the original version of our model generally agree with the
experimental data of Altafini et al. (2003) and their SynGas model, except for the cases of H2 and CH4.
Actually, the models predicted higher amounts of H2 and lower amounts of CH4 than experimental data. Altafini
et al. (2003) noted that this result is generalized to other equilibrium models from the literature. A possible
explanation of the differences may have came from the assumptions defined in simplifying the model, such as
all gases are assumed to be ideal at a pressure of one atmosphere, ashes are inert, absence of other gases,
etc. Besides, from a chemical point of view, CH4 formation at the equilibrium temperature of 800°C is not

2013 ASABE Annual International Meeting Paper Page 6

favored because G f ,CH 4 is positive and approximately equal to 25 kJ mol-1. Therefore, the system needs to
○

absorb energy so that the formation of CH4 could happen. On the other side, the CO and CO2 formation
reactions present high values of G f at 800°C (–206 kJ mol-1 and –397 kJ mol-1, respectively) comparatively to
○

CH4. Consequently, the formation of CO and CO2 releases an important quantity of energy which allows the
system to reach a minimum level of free energy.
To increase the results’ accuracy, Jarungthammachote and Dutta (2007) calibrated the amount of CH4 in their
model by fixing a value derived from the experimental data of Altafini et al. (2003), Jayah et al. (2003) and
Zainal et al. (2001). The modified model was then used to simulate and compare with the SynGas model and
the experimental results of Altafini et al. (2003). Table 1 shows that after modifying the model, the amount of H2
significantly reduced as compared to the predicted value from the SynGas model. The amount of CH4 also
dramatically increased and was found closer to the experimental value. The same approach was used for the
modified version of the present model to finally state the same conclusions.
Tables 2 to 4 present the simulation results of Jarungthammachote and Dutta (2007), Melgar et al. (2007) and
the present model (original and modified versions) in comparison with the experimental data of Jayah et al.
(2003) on rubber wood gasification at different MC: 14.0%, 14.7% and 16.0% (wt% on dry basis). The wood
ultimate analysis results in 50.6wt% C, 6.5wt% H, 42.0wt% O, and 0.2wt% N. The comparison test was done
by setting the temperature used for the developed model fixed at 1100 K. The air/sawdust ratios were 2.29,
1.86 and 1.96, respectively for each MC, as reported by Jayah et al. (2003). As in table 1, the values predicted
by the thermochemical equilibrium models show good accuracy with experimentally obtained data even though
H2 and CH4 contents are slightly different, except if CH4 is fixed in a modified version of the model.

Testing the model with combustion results

The prediction model developed for biomass gasification was adapted to biomass combustion now considering
equation 2 instead of equation 1. Data from Godbout et al. (2012) technical report on recent agricultural
biomass combustion was used for the simulation tests even though some important input values are missing.
This situation is widespread in the literature. Already that there is not many references on the gaseous
emissions from agricultural biomass combustion, most of scientific articles on the subject do not report all these
essential results for modeling: biomass ultimate and proximate analyses, air flow rate, combustion rate and
temperature. Unknown values were therefore estimated when necessary.
Combustion tests were held in April 2011 in an on-farm combustion laboratory in Deschambault, Quebec,
Canada. The experiments were carried out in a 60,000 BTU h-1 (17.58 kW) output biomass pellet stove (Enviro
Omega model, Sherwood Industries Ltd., Saanichton, BC, Canada). A Fourier transform infrared spectrometer
(FTIR; model FTLA2000, ABB, Quebec city, QC, Canada) was used to analyze concentrations (ppmv) of eight
gases (CH4, CO, CO2, HCl, NH3, NO2, N2O and SO2) during the combustion of wood, short-rotation willow and
switchgrass.
The three biomasses were all pelletized and their characteristics are described in table 5. This table also
includes the combustion conditions for each combustion test done with these biomasses. Table 6 presents the
measured emissions at the laboratory and the predicted values obtained with the model. For most gases, the
model seems generally accurate even if the experimental error on the measured values is high. According to
the authors, it reaches almost 100% mainly because of the low flue gas flow rates which were in the detection
limits of the iris damper.
Table 5. Characteristics of experimental biomass pellets and combustion conditions during combustion test with three
biomasses.
Element or property Units Wood Short-rotation willow Switchgrass
BIOMASS
C wt% dry basis 49.8 48.8 46.9
N wt% dry basis 0.114 0.632 0.672
Cl wt% dry basis 0.001 0.004 0.014
S* wt% dry basis 0.05 0.06 0.10
H* wt% dry basis 6.2 6.2 5.5
O** wt% dry basis 43.875 44.204 46.814
Moisture content wt% wet basis 6.10 11.36 13.05
-1
Gross calorific value MJ kg 17.9 18.0 18.7
COMBUSTION
3 -1
Air flow rate m min 0.23 0.21 0.20
-1
Combustion rate kg h 1.20 1.92 1.56
*Estimated from mean concentrations proposed van Loo and Koppejan (2008).
**By difference.

2013 ASABE Annual International Meeting Paper Page 7

Nevertheless, the model obtained really good results in predicting carbon compounds, especially for CO2
where the difference between the measured and the predicted value is about 7%. CO, as an intermediate in
the conversion of fuel carbon to CO2, is oxidized to CO2 if oxygen is sufficiently available. Given that the air
flow is low, only a small difference in the quantity of air entering the combustion chamber can drastically
change the modeling results. For instance, a simple increase of 0.03 m3 min-1 to reach an air flow of 0.26 m3
min-1 for the wood combustion test implies a drop of predicted CO to less than 1 g kg-1. In these circumstances,
the CO results are acceptable even though they do not seem as good as for CO2. Since the emission of CH4 is
a result of too low combustion temperatures, too short residence times or lack of available oxygen (van Loo
and Koppejan, 2008), the higher emission value measured during wood combustion is likely due to these
factors. Theoretically, the real behaviour of the combustion process and possible unfavourable conditions such
as poor mixing are hard to predict. With sufficient air and temperature, the predicted CH4 emission approaches
zero like it was the case for fast-growing willow and switchgrass experiments. Therefore, it is possible to think
that the combustion conditions during wood combustion were sufficient to reach a lower CH4 value like the one
obtained by the model.
-1
Table 6. Measured and predicted emissions (g kg ) from the combustion of three biomasses.
Gas Wood Short-rotation willow Switchgrass
species Measured Predicted Measured Predicted Measured Predicted
CH4 0.3012 0.0077 0.0043 0.0079 0.0000 0.0003
CO 21.8 17.4 17.4 18.6 2.41 4.67
CO2 1949 1800 1870 1760 1610 1710
HCl 0.018 0.010 0.000 0.041 0.016 0.144
NH3 0.0016 0.0082 0.0063 0.0082 0.0007 0.0003
NO2 0.0066 0.0100 0.1284 0.0097 0.3616 0.0004
N2O 0.0000 0.0086 0.0343 0.0084 0.0000 0.0003
SO2 0.71 1.00 0.53 1.20 1.70 2.00

Nitrogen oxides (NOx) emissions increase with increasing nitrogen content in the fuel (van Loo and Koppejan,
2008). This tendency can be seen in table 6 as measured NO2 for agricultural biomass combustion are much
higher than for wood combustion. However, the simulation results for agricultural biomass combustion do not
match the values obtained in laboratory. From a chemical point of view, NO2 formation at the simulated
equilibrium temperature (≈ 875°C) is not favored because G f○is positive. It can then be supposed that under
agricultural biomass combustion conditions most nitrogen was rather converted to NO or N2 by the model. This
assumption suits to the fact that NO emissions represent about 90% of NOx emissions during biomass
combustion (van Loo and Koppejan, 2008). Also, in fuel-rich conditions, NO react with intermediates like NH3
and hydrogen cyanide (HCN) to form N2. When the primary excess air ratio, temperature and residence time
are optimized, a maximum conversion of NH3 and HCN to N2 is achieved. Moreover, a nitrogen mass balance
reveals that only 15% to 30% of nitrogen is emitted as NOx, the rest being emitted as N2 (AILE, 2012). The low
accuracy of the model to predict NO2 could result from these phenomena, but the absence of data on NO and
N2 do not allow confirming anything. Future experimental tests should consider acquiring the necessary
equipment to measure these gas species. However, the prediction model obtained good results for NH3 and
NO2, both gases being only present in small amounts.
Agricultural biomasses contain significant amounts of chlorine comparatively to wood. This difference echoes
on HCl emissions as the model predicted higher HCl levels for switchgrass and short-rotation willow
combustions. Nevertheless, the relation between fuel chlorine and HCl is not clear by looking to experimental
results. In fact, fuel chlorine is not completely converted to HCl. The main fraction is rather retained in salts by
reaction with potassium (K) and sodium (Na) (van Loo and Koppejan, 2008). This can explain the difference
between measured and predicted values. More work is expected to adjust the model so that it takes into
account this fact. A better result for wood combustion is certainly due to its very low chlorine content.
SO2 emissions are directly linked to the content of sulfur in the biomass. Since sulfur content was estimated, it
is hard to conclude something on the basis of the present results. However, it is possible to note that predicted
values are always higher than measured values. As for HCl, fuel sulfur is only partly released in its main
gaseous form (SO2). In fact, a significant portion of sulfur remains in ashes while a minor fraction is emitted as
potassium sulphate (K2SO4) or hydrogen sulphide (H2S). Measurements at two district heating plants in
Denmark using straw as fuel showed that 57% to 65% of the sulfur was released in the flue gas (Obernberger
et al., 2006). Future work should also consider these numbers.

Conclusion or Summary
An equilibrium model based on minimization of the Gibbs free energy was developed to predict the composition
of gaseous emissions from biomass gasification and combustion. The general calculation procedure includes

2013 ASABE Annual International Meeting Paper Page 8

an energy balance to calculate the reaction temperature and the minimization of the Gibbs free energy based
on this temperature. Given that the equilibrium approach has been mostly used in biomass gasification, the
model was validated by comparing the simulation results of the six major components (CO, CO2, CH4, H2O, H2
and N2) with those of other researchers who utilized the same original data for their model. The predicted
values and the experimental data showed good accuracy, except for H2 and CH4. The slight differences
obtained can be reduced by modifying the model by fixing the CH4 concentration in the gas mixture, based on
experimental results. The model was then adapted to combustion and tested with data from past experiments
on wood and agricultural biomass combustion. The results of simulation from the model were closer to the
experimental data for CO, CO2, NH3 and N2O. The model also showed a certain deviation from the
experimental data for CH4, NO2, HCl and SO2. At the moment, the model allows to know the theoretical
combustion conditions that allow reaching a complete combustion by keeping the products of incomplete
combustion (CO, CH4 and NH3) as low as possible. Future experimental tests considering the measure of
hydrogen and sulfur contents, combustion temperature, NO and N2 emissions are needed to calibrate the
model in order to improve results’ accuracy.

Acknowledgements
The authors thank the “Ministère de l’Agriculture, des Pêcheries et de l’Alimentation du Québec” (MAPAQ) and
the “Fonds de recherche du Québec – Nature et technologies” (FQRNT) for their financial contributions. The
authors gratefully acknowledge the Research and Development Institute for the Agri-Environment (IRDA) and
the Université de Sherbrooke which provided in-kind contributions for this study. The authors also recognize
the professional support provided by IRDA research staff (Joahnn Palacios and Patrick Brassard). Authors also
thank Dr. Matthieu Girard for the linguistic revision.

References
AILE. 2012. Résultats du projet Green Pellets. Available at: http://www.aile.asso.fr/cultures-energetiques/GreenPellets/colloque-
final-green-pellets. Accessed 9 Janaury 2013.
Altafini, C. R. 2003. Prediction of the working parameters of a wood waste gasifier through an equilibrium model. Energy
Conversion Management 44: 2763-2777.
Baratieri, M., P. Baggio, L. Fiori, and M. Grigiante. 2008. Biomass as an energy source: Thermodynamic constraints on the
performance of the conversion process. Bioresource Technology 99: 7063-7073.
Brodeur, C., J. Cloutier, D. Crowley, X. Desmeules, S. Pigeon, and R. M. St-Arnaud. 2008. La production de biocombustibles
solides à partir de biomasse résiduelle ou de cultures énergétiques. Publication n° EVC 032. Québec City, QC, Canada:
Centre de référence en agriculture et agroalimentaire du Québec (CRAAQ).
Demirbas, A. 2009. Biofuels securing the planet’s future energy needs. Energy Conversion and Management 50: 2239-2249.
de Souza-Santos, M. L. 2010. Solid fuels combustion and gasification: modeling, simulation, and equipment operations. 2nd ed.
Boca Raton, FL, United States: CRC Press.
Gautam, G. 2010. Parametric study of a commercial-scale biomass downdraft gasifier: Experiments and equilibirum modeling. M.
Sc. thesis. Auburn, AL, États-Unis: Auburn University.
Gøbel, B., U. Henriksen, T. K. Jensen, B. Qvale, and N. Houbak. 2007. The development of a computer model for a fixed bed
gasifier and its use for optimization and control. Bioresource Technology 98: 2043-2052.
Godbout, S., J. H. Palacios, J. P. Larouche, P. Brassard, and F. Pelletier. 2012. Bilan énergétique, émissions gazeuses et
particulaires de la combustion de la biomasse agricole à la ferme. Project # 100 035. Québec City, QC, Canada: Research
and Development Institute for the Agri-Environment (IRDA).
Jarungthammachote, S., and A. Dutta. 2007. Thermodynamic equilibrium model and second law analysis of a downdraft waste
gasifier. Energy 32: 1660-1669.
Jarungthammachote, S., and A. Dutta. 2008. Equilibrium modeling of gasification: Gibbs free energy minimization approach and
its application to spouted bed and spout-fluid bed gasifiers. Energy Conversion and Management 49: 1345-1356.
Jayah, T. H., L. Aye, R. J. Fuller, and D. F. Stewart. 2003. Computer simulation of a downdraft wood gasifier for tea drying.
Biomass and Bioenergy 25: 459-469.
Jenkins, B. M., L. L. Baxter, and J. Koppejan. 2011. Biomass combustion. In Thermochemical processing of biomass: conversion
into fuels, chemicals and power, 13-46. R. C. Brown, ed., Chichester, United Kingdom: John Wiley & Sons.
Kalina, J. 2011. Modelling of fluidized bed biomass gasification in the quasi-equilibrium regime for preliminary performance studies
of energy conversion plants. Chemical and Process Engineering 32(2): 73-89.
Kaltschmitt, M., and M. Weber. 2006. Markets for solid biofuels within the EU-15. Biomass and Bioenergy 30: 897-907.
Koukkari, P., and R. Pajarre. 2006. Introducing mechanistic kinetics to the Lagrangian Gibbs energy calculation. Computers and
Chemical Engineering 30: 1189-1196.
Lease, N., and L. Théberge. 2005. Le secteur agricole au Québec : une source d'énergie pour l'avenir? FrancVert 2(1): 1-6.
MathWorks. 2013. Find minimum of constrained nonlinear multivariable function - MATLAB fmincon. Natick, MA, United States:
The MathWorks, Inc. Available at: http://www.mathworks.com/help/optim/ug/fmincon.html. Accessed 28 May 2013.

2013 ASABE Annual International Meeting Paper Page 9

 McBride, B. J., S. Gordon, and M. A. Reno. 1993. Coefficients for calculating thermodynamic and transport properties of
individual species. NASA Technical Memorandum 4513. Cleveland, OH, United States: National Aeronautics and Space
Administration.
Melgar, A., J. F. Pérez, H. Laget and, A. Horillo. 2007. Thermochemical equilibrium modelling of a gasifying process. Energy
Conversion and Management 48: 59-67.
Néron, A., G. Lantagne, and B. Marcos. 2012. Computation of complex and constrained equilibria by minimization of the Gibbs
free energy. Chemical Engineering Science 82: 260-271.
Obernberger, I., T. Brunner, and G. Bärnthaler. 2006. Chemical properties of solid biofuels – significance and impact. Biomass
and Bioenergy 30: 973-982.
Ragland, K. W., and K. M. Bryden. 2011. Combustion engineering. 2nd ed. Boca Raton, FL, United States: CRC Press.
Ranzi, E., S. Pierucci, P. C. Aliprandi, and S. Stringa. 2011. Comprehensive and detailed kinetic model of a traveling grate
combustor of biomass. Energy & Fuels 25: 4195-4205.
Rossi, C. C. R. S., L. Cardozo-Filho, and R. Guirardello. 2009. Gibbs free energy minimization for the calculation of chemical and
phase equilibrium using linear programming. Fluid Phase Equilibria 27: 8117-128.
Rossi, C. C. R. S., M. E. Berezuk, L. Cardozo-Filho, and R. Guirardello. 2011. Simultaneous calculation of chemical and phase
equilibria using convexity analysis. Computers and Chemical Engineering 35: 1226-1237.
Talluto, L. 2009. Des champs d'énergie? Le coopérateur agricole 38(7): 50-53.
van Loo, S., and J. Koppejan. 2008. The handbook of biomass combustion and co-firing. 1st ed. Washington, DC, United States:
Earthscan.
Villeneuve, J., J. H. Palacios, P. Savoie, and S. Godbout. 2012. A critical review of emissions standards and regulations regarding
biomass combustion in small scale units (<3 MW). Bioresource Technology 111: 1-11.
Zainal, Z. A., R. Ali, C. H. Lean, and K. N. Seetharamu. 2001. Prediction of performace of a downdraft gasifier using equilibrium
modeling for different biomass materials. Energy Conversion and Management 42: 1499-1515.

Nomenclature
a number of atoms of nitrogen per number of atom of carbon in the feedstock
Aj total number of atoms of jth element in reaction mixture
aij number of atoms of jth element in a mole of ith species
b number of atoms of sulfur per number of atom of carbon in the feedstock
c number of atoms of chlorine per number of atom of carbon in the feedstock
-1
Cp specific heat at constant pressure, kJ K
e amount of oxygen per kmol of feedstock
G total Gibbs free energy of a system, kJ
H enthalpy, kJ
L Lagrangian function
LHV lower heating value, kJ
n numbers of moles
P pressure, atm
Qloss heat loss, kJ
-1 -1
R universal gas constant, 8.314 kJ kmol K
-1
S entropy, kJ K
T temperature, K
w amount of moisture per kmol of feedstock
number of atoms of carbon in the
x
hydrocarbon
y number of atoms of hydrogen per number of atom of carbon in the feedstock
yi mole fraction of gas species
z number of atoms of oxygen per number of atom of carbon in the feedstock

Greek letters
 show a difference
 stoichiometric number
 chemical potential
 Lagrange multiplier

Superscripts
-1
– quantity per unit mole, kmol
○ standard reference state

Subscripts
comp compound
elem element
f of formation
i ith gas species
j jth chemical element
k total number of elements
l lth gas species
N total number of gas species

2013 ASABE Annual International Meeting Paper Page 10

 r reactant
p product
tot total

2013 ASABE Annual International Meeting Paper Page 11

View publication stats