1576 Ind. Eng. Chem. Res.

2005, 44, 1576-1584

Removal of NH3 from Biomass Gasification Producer Gas by Water

Condensing in an Organic Solvent Scrubber
Tobias Pro1 ll,* Ingmar G. Siefert, Anton Friedl, and Hermann Hofbauer
Institute of Chemical Engineering, Vienna University of Technology, Getreidemarkt 9, 1060 Wien, Austria

Producer gas from biomass gasification contains NH3, H2S, and high molecular weight organic
compounds (tars), which must be removed prior to the gas utilization step. A highly effective
approach toward the removal of tars is absorption into organic solvents. As a secondary effect,
condensation of water takes place in the scrubber, allowing removal of gaseous trace components
such as NH3 and H2S from the gas stream. The present work focuses on the coabsorption of
NH3 and CO2 into condensing water. The general aim is to decrease the clean gas NH3
concentration by appropriate adjustment of the operating parameters without additional gas-
cleaning steps. A mathematical model of a countercurrent column, which covers mass and energy
balances, gas-liquid equilibrium, and electrolyte dissociation, is presented. The scrubber
geometry does not enter the model; therefore, it promises universal applicability. Property data
on electrolyte dissociation and Henry’s law constants are taken from the literature. A computer
code has been developed in order to study the sensitivity of NH3 removal toward certain process
parameters. The results agree well with data measured at the 2000 Nm3/h (gas load) scrubber
at the biomass gasification plant in Guessing, Austria.

1. Introduction Table 1. Typical Composition of a Producer Gas from

the FICFB Gasification Process before Gas Cleaninga
Biomass steam gasification allows the conversion of H2O 30-45 vol %
solid biomass to medium calorific gas consisting mainly CH4 10-11 vol % (dry)
of hydrogen, carbon oxide, methane, carbon dioxide, and C2H4 2-2.5 vol % (dry)
water. Either the gas may be used for electricity C3 fraction 0.5-0.7 vol % (dry)
production in turbines or engines or it may be used for CO 24-26 vol % (dry)
the synthesis of high-quality liquid fuels or synthetic CO2 20-22 vol % (dry)
H2 38-40 vol % (dry)
natural gas. N2 1.2-2.0 vol % (dry)
In the fast internal circulating fluidized bed (FICFB) H2S 130-170 vppm (dry)
gasification process,1,2 the solid fuel enters a steam- NH3 1100-1700 vppm (dry)
fluidized stationary bed where thermal conversion takes tar 2-5 g/Nm3 (dry)
place at temperatures between 850 and 900 °C. The particulates 20-30 g/Nm3 (dry)
LHV 12.9-13.6 MJ/Nm3 (dry)
energy for the endothermic gasification reactions is
introduced by the circulating-bed material heated in a a All quantities except the water concentration are related to

second fluidized-bed reactor by combustion of residual the dry gas fraction.

char. Gas cleaning is necessary in order to remove
particles and high molecular weight organic compounds higher NOx emissions in the exhaust gas. For alterna-
(tars) from the producer gas. An 8 MW (fuel power) tive gas applications such as Fischer-Tropsch synthesis
combined heat power (CHP) plant, coupling an FICFB for liquid fuel production, the required NH3 limits are
biomass steam gasifier and a 2 MWel gas engine, has as low as 1 vppm.4
been realized in Guessing, Austria, and has been in The nitrogen content of wood for the different species
operation since 2002. The plant uses wood chips as fuel. is 0.10-0.45 wt % (waf).5 Nitrogen is concentrated in
The gas-cleaning technology consists of a bag filter for the bark. Soft wood generally contains less nitrogen
the removal of particulates and an organic solvent than hard wood. Waste wood may contain much more
scrubber for tar removal. The process is described in nitrogen, up to 3.5 wt % (waf).5 According to the nitrogen
detail by Hofbauer et al.3 balance for the Guessing plant, up to 70% of the fuel
Table 1 shows the typical composition of the producer nitrogen is found as NH3 in the producer gas, even
gas of the Guessing plant if untreated wood chips are though the equilibrium of the inverse ammonia synthe-
used as fuel. The water content is strongly influenced sis reaction
by the ratio between total water (fluidization steam +
fuel humidity) and dry fuel in the gasifier (steam/fuel 2NH3 S N2 + 3H2 (1)
ratio).
Apart from solids and tars, there are also gaseous is far on the side of nitrogen and hydrogen under
trace components in the producer gas, which may cause gasification conditions [∼4 vppm (dry) of NH3 in equi-
problems in gas utilization. One of them is NH3, which librium at 880 °C; 1 bar]. Apparently, kinetics does not
occurs in concentrations up to 1700 vppm (dry) in the allow significant conversion of NH3 to N2. After leaving
raw producer gas. In the case of gas utilization in the the gasifier, the producer gas is cooled in an indirect
combustion engine, NH3 in these concentrations reduces heat exchanger to a bag filter inlet temperature of about
the operating lifetime of the engine oil and leads to 170 °C, freezing the NH3 concentration at the value of
10.1021/ie049669v CCC: $30.25 © 2005 American Chemical Society
Published on Web 01/26/2005
 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005 1577

heat-exchanger entry. Up to now, no effects of the bag

filter on NH3 in the producer gas could be observed. The
organic solvent scrubber downstream of the filter cools
the producer gas again to about 40-50 °C (with the
temperature changing with ambient conditions), which
implies that the main part of the water is condensing
here, forming an emulsion with the organic phase. The
organic solvent used is rape oil methyl ester (RME),
which shows excellent tar solubility of up to 0.5 kg/kg
at 50 °C. Experimental investigations have shown that
NH3 is not significantly soluble in RME. For the present
work, it is assumed that only the condensate accounts
for NH3 removal in the tar scrubber.
The scope of the present work is to gain a deeper
insight in the mechanism of NH3 removal in the organic
scrubber. Implying that the organic solvent is not
interacting, the gas-liquid equilibrium between the
condensate and producer gas in the scrubber is of prior
importance for NH3 removal. The equilibrium in the
liquid phase is governed by the interaction of the two
weak electrolytes, NH3 and CO2. The behavior of aque-
ous solutions of weak electrolytes has been extensively
investigated by Edwards et al.,6 who present a model
for the system NH3-CO2-H2O based on the reactions
Figure 1. Dissociation constants of the NH3-CO2-H2O system:
H2O S H+ + OH- (2) constants defined according to eqs 19-22.

NH4+ S H+ + NH3 (3)

CO2 + H2O S H+ + HCO3- (4)

HCO3- S H+ + CO32- (5)

HCO3- + NH3 S NH2COO- + H2O (6)

The equilibrium at the gas-liquid interface is de-

scribed by Henry’s law for the two species NH3 and CO2:
pA ) KH,Aaa (7)

Equilibrium data covering the whole NH3-CO2-H2O

system are presented by Edwards et al.7 These authors
apply the correlations for water and ammonia dissocia-
tion reported by Tsonopoulos et al.8 (Figure 1). The
water dissociation constant by Tsonopoulos et al.8 is in
very good agreement with other literature sources9 for
the temperature range 0-225 °C. The correlation for
ammonia dissociation by Tsonopoulos et al.8 agrees well
with the correlation by Kawazuishi and Prausnitz10 up
to 150 °C. Henry’s law constant for NH3 by Edwards Figure 2. Henry’s law constants of NH3 and CO2 in water and
et al.7 (Figure 2) agrees with the correlation by Clegg equilibrium constant of carbamate formation: constants defined
and Brimblecombe11 at low temperatures and fits according to eqs 17, 18, and 23.
DECHEMA data12 at intermediate temperatures up to
about 150 °C. The correlation of Kawazuishi and Praus- and Prausnitz.10 The correlations differ by up to 50%
nitz,10 who refer to data measured by Gillespie et al.13 between 30 and 100 °C, where the correlation of Paw-
and Müller et al.,14 yields a slightly lower solubility of likowski et al.17 differs also in shape and shows a peak
NH3 in water than the Edwards correlation (difference at about 50 °C. Within the present work, the newest
of 7% at 50 °C) and is recommended up to 300 °C. Com- correlation by Kawazuishi and Prausnitz,10 recom-
prehensive data on Henry’s law and dissociation con- mended up to 160 °C, has been chosen. Anyway,
stants for the CO2-H2O system up to 250 °C are re- reaction (6) influences the total NH3 solubility only
ported by Plummer and Busenberg,15 who suggest a marginally within the temperature and pH range of the
five-parameter approach for the temperature depen- present work.6
dence. The correlation for Henry’s law constant of CO2
determined by Chen et al.16 agrees well with the 2. Experimental Section
Plummer and Busenberg formulation. The equilibrium
constant of carbamate formation (eq 6) is reported by Setup of the Organic Solvent Scrubber. Figure
Edwards et al.,6,7 Pawlikowski et al.,17 and Kawazuishi 3 shows the setup of the tar scrubber with a capacity of
1578 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005

with 350 mL of 0.05 M H2SO4 each. The dry gas flow is

monitored with a gas meter. One sampling run lasts for
30 min with a sampling flow rate of 0.080 Nm3/h. The
solvent of the four impinger bottles is put together for
further analysis.
Condensate Sampling. During the gas sampling
run, three samples of condensate are taken from the
exit of the equalization tank (Figure 3). The three
samples are merged to get the condensate representa-
tive of the gas compositions.
Determination of NH3 in Liquid Samples. The
total amount of ammonia in the liquid samples is
determined by ion chromatography in Dionex DX-120.
Because the samples are introduced by an acid buffer
solution, NH3 is quantitatively detected as NH4+. The
basicity of the condensate has been determined at room
temperature with a pH meter (Mettler-Toledo pH 2-14
electrode).

3. Mathematical Model
The model aims at the description of the phenomena
Figure 3. Scrubber system at the biomass gasification CHP plant
in Guessing, Austria.
relevant for NH3 removal inside the scrubber. As
mentioned in the Introduction, NH3 has not been found
to be soluble in the organic phase and can therefore only
be absorbed by the condensing water. Because of the
complex behavior of the two liquid phases, no reliable
correlation for the mass-transfer coefficient between the
gas and aqueous phases could be found. On the other
hand, phase separation experiments show that the
organic phase tends to cover the gas-liquid interface,
which leads to the major assumption in the model: It
is assumed that water is in equilibrium with the gas
Figure 4. Gas sampling setup. phase just when condensation takes place and then
immediately emulsified in the organic phase. Once
2000 Nm3/h (dry) in the biomass gasification CHP plant emulsified, any mass transfer from condensate to gas
in Guessing, Austria. The ratio of the new organic is inhibited by the organic phase. The condensate in the
solvent to the circulating solvent is about 1:2500 in equalization tank represents a mixture of the water
practical operation. The temperatures vary according fractions occurring along the height of the column.
to the ambient conditions (winter/summer). Energetic Aspects of Condensation. If the clean
Temperatures and flow rates are automatically reg- gas exit temperature is below the saturation tempera-
istered during operation. The producer gas CO2 content ture of the raw gas mixture, the partial pressure of
is measured twice: once by an infrared online analyzer water in the clean gas will equal the vapor pressure of
(Advanced Optima by ABB Automation) and, together water at the exit temperature. The rest of the water
with other gas components, also by an online gas introduced with the raw gas leaves the scrubber as
chromatograph (Syntech Spectras GC 955; 80 °C oper- liquid emulsified in the organic phase. All aspects of tar
ating temperature; He as the carrier gas; two columns, removal are neglected in the modeling.
Chromosorb for splitting between permanent gases, The column is modeled one-dimensionally and with-
CO2, and hydrocarbons and a 5 Å molecular sieve for out axial mixing in any of the phases. This simplified
splitting the permanent gases H2, O2, N2, CO, and CH4). description of phase flow together with the assumption
The raw gas water content can be calculated from a of inhibited mass transfer introduced above allows the
water balance across the scrubber because the amount utilization of the gas-phase temperature as a coordinate
of condensate can be directly measured, and the clean instead of the geometrical height. Therefore, the model
gas water content can be calculated from the gas exit does not depend on the column geometry. In the lower
temperature assuming 100% saturation. part of the column, only heat transfer from gas to
Ammonia Sampling from Gas. The measurement emulsion occurs until the gas phase reaches the satura-
principle is based on the discontinuous sampling of the tion temperature and condensation starts. No re-
producer gas. The gas sampling points are situated be- evaporation of emulsified condensate is considered in
fore and after the scrubber (refer to Figure 3). The sam- the lower part of the column. The gas phase is modeled
pling system consists of a heated probe with a spherical as an ideal gas, which means that the product of the
valve and a series of cooled impinger bottles containing total pressure and gas-phase mole fraction yi is propor-
an acid solution to catch the ammonia. The temperature tional to fugacity and that the total gas enthalpy is a
of the cooling agent is -2 °C, so that the gas is cooled linear combination of the enthalpies of the various
to almost 0 °C and practically free of water (Figure 4). components. The saturation gas temperature TG,C can
To get representative values before and after the be calculated by eq 8 with the raw gas water content
scrubber, raw and clean producer gases are extracted yH2O,in and total pressure p.
simultaneously with separate sampling systems accord- Once condensation has started, the water content in
ing to Figure 4. The cooled impinger bottles are filled the gas decreases, according to eq 8, with decreasing
 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005 1579

pyH2O,in ) pS,H2O(TG,C) (8) capacity of the organic solvent is treated as a constant

within the column, a partial heat balance in the form
TG. The saturation vapor pressure of water pS,H2O is of eq 13, starting either from the top or from the bottom
given in bar by the DIPPR 801 equation,18 which fits of the column, allows the calculation of the correspond-
the IAPWS-IF9719 data in the whole temperature range ing liquid-phase temperature to any gas temperature
in which liquid water may appear (273.16-647.13 K). between TG,in and TG,out. An interesting parameter to
check the plausibility of model assumptions and mea-
7258.2 sured data during model evaluation is the minimum
ln(pS,H2O) ) 62.1361 - - 7.3037 ln(T) + temperature difference ∆TGL,C between the gas and
T
liquid phases in the section where condensation starts.
4.1653 × 10-6T 2 (9)
For the formulation of mass and energy balances, it ∆TGL,C ) TG,C - TL,C (15)
is advantageous to relate all extensive quantities to the
flow rate of dry gas. The water load on dry gas is The partial heat balance in the lower part of the
scrubber for the calculation of TL,C can be written as
yH 2 O
Y H 2O ) (10) I
1 - yH2O CSOLV (TL,out - TL,C) + YCOND[HWAT(TL,out) -
HWAT(TL,C)] + HGAS,dry(TG,C) - HGAS,dry(TG,in) +
The upper part of the column, where condensation YH2O,in[HGAS,H2O(TG,C) - HGAS,H2O(TG,in)] ) 0 (16)
occurs, can be divided into n temperature intervals ∆T
of equal size. The specific amount of condensate ∆YCOND,k
occurring in element k can be gained from eq 11, with “C” indicates the section where condensation starts.
the indices “in” and “out” referring here to element k. Because the total pressure is approximately 1 bar, the
producer gas is assumed to behave like an ideal gas and
∆YCOND,k ) YH2O,k,in - YH2O,k,out (11) ideal gas enthalpy data are used. The pressure depen-
dence of the enthalpy of liquid water can also be
The gas-phase water loads YH2O,k,in and YH2O,k,out can be neglected. Burcat and McBride20 report coefficients for
calculated from eqs 8-10 with the temperatures at the the calculation of total enthalpies H0(T) for various
boundaries of element k. The total specific amount of species using the NASA polynomials. The molar en-
condensate YCOND is determined by the global water thalpy of ideal gas mixtures is the linear combination
balance. of the enthalpies of the different components using the
mole fractions as weights.
n NH3 Absorption. From the section where condensa-
YCOND ) YH2O,in - YH2O,out ) ∑ ∆YCOND,k
k)1
(12) tion starts upward, water is condensing in chemical
equilibrium with the gas phase and transported down-
ward as an emulsion in the organic phase. Because mass
The indices “in” and “out” refer here to the gas streams transfer between the gas and aqueous liquid phases is
entering and exiting the scrubber. assumed to be inhibited by the organic phase, the lower
Neglecting heat losses, the global energy balance can part of the column, where no condensation occurs, does
be written as not contribute to NH3 absorption.
Gas-liquid equilibrium during condensation is cal-
I
CSOLV (TL,out - TL,in) + YCONDHWAT(TL,out) + culated according to the reaction system presented by
HGAS,dry(TG,out) - HGAS,dry(TG,in) + Edwards6 (eqs 2-6). Because of the low electrolyte
concentrations in the condensate, the solution is as-
YH2O,outHGAS,H2O(TG,out) - YH2O,inHGAS,H2O(TG,in) ) 0 sumed to behave ideally. In other words, all activity
(13) coefficients in the solution are assumed to be unity and
the equilibrium is expressed by molalities m (moles per
TL refers to the liquid-phase temperature, which is kilogram of water). The set of equations to describe the
uniform for organic and aqueous phases. YCOND is the gas-liquid equilibrium is presented as follows.
specific amount of condensate per mole of dry gas, YH2O
is the H2O load on dry gas, and H(T) refers to the KH,NH3 ) pyNH3/mNH3(aq) (17)
conventional enthalpy (sensible heat including the en-
thalpy of formation). The indices “in” and “out” refer to KH,CO2 ) pyCO2/mCO2(aq) (18)
whether a stream enters or leaves the system, “WAT”
indicates liquid water, “GAS,H2O” indicates gaseous Kw ) mH+mOH- (19)
water, and “GAS,dry” indicates the dry gas composition.
The mean heat capacity of the organic solvent CISOLV is KNH4+ ) mH+mNH3(aq)/mNH4+ (20)
defined with the specific heat capacity cP,SOLV, the mass
flow of solvent m̆SOLV, and the molar flow of dry gas KCO2 ) mH+mHCO3-/mCO2(aq) (21)
n̆G,dry as follows:
KHCO3- ) mH+mCO32-/mHCO3- (22)
I
CSOLV ) m̆SOLVcP,SOLV/n̆G,dry (14) KNH2COO- ) mNH2COO-/mNH3(aq)mHCO3- (23)
If the initial gas composition and the in/out temper- mH+ + mNH4+ ) mOH- + mHCO3- +
atures of the gas and liquid phases are known, CISOLV
can be calculated from eq 13. When the mean heat 2mCO32- + mNH2COO- (24)
1580 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005

Table 2. Effect of Temperature on Equilibrium Constants According to Equation 25

A0 A1 A2 A3 A4 valid T [°C] ref
KH,NH3 [bar kg/mol] 41.79296 0.003 733 62 -13.588 57 -3291.9296 0-300 10
KH,CO2 [bar kg/mol] -108.3808 -0.019 850 78 40.451 54 6919.53 -669 365.0 0-250 15
Kw [mol2/kg2] 61.2062 -22.477 3 -5839.5 0-225 8
KNH4+ [mol/kg] 60.0072 0.016 093 5 -23.974 4 -4390.82 0-225 8
KCO2 [mol/kg] -356.3094 -0.060 919 64 126.833 9 21834.37 -1 684 915.0 0-250 15
KHCO3- [mol/kg] -107.8871 -0.032 528 49 38.925 61 5157.79 -563 713.9 0-250 15
KNH2COO- [kg/mol] 8.75196 0.002 184 91 -4.017 263 262.3644 0-160 10

Table 2 reports literature coefficients for the temper- Model Implementation. To apply the model in prac-
ature dependence of the equilibrium constants according tical calculation, the upper part of the column between
to the equation the gas exit temperature and gas temperature at the
start of condensation is modeled as n discrete elements
A3 A4 represented by equal temperature intervals. The simu-
log(K) ) A0 + A1T + A2 log(T) + + 2 (25) lation algorithm for the entire column is sketched in
T T
Figure 5 for the case of known raw gas NH3 content.
with T in Kelvin. The CO2 concentration in clean gas must be known.
For known temperature and gas molar fractions of Iteration starts at the gas exit with an estimated
NH3 and CO2, the solution is determined by eqs 17- value for the NH3 concentration and proceeds downward
24. The total amount of electrolytes in the liquid is then with increasing temperature until the raw gas composi-
tion is reached at the section where condensation begins.
mNH3(total) ) mNH3(aq) + mNH4- + mNH2COO- (26) The gas-liquid equilibrium requires the solution of a
nonlinear system of eight equations (eqs 17-24) in each
mCO2(total) ) mCO2(aq) + mHCO3- + mCO32- + mNH2COO- element. To improve the convergence behavior of the
(27) commercial solver used, the system can be reduced to a
one-parameter problem with one concentration, e.g.,
The total amount of NH3 or CO2 in the condensate mOH-, as the variable. Therefore, eqs 17-23 are trans-
leaving the column can be calculated with Tk as the formed to a logarithmic scale, presenting a 7 × 7 linear
medium temperature in element k as follows.
n
1
mi(total),out )
YCOND
∑ mi(total)(Tk) ∆YCOND,k
k)1
(28)

It is important to notice that also the gas-phase

composition may change along the column’s height
influencing mi(total)(T). For the gas phase, the partial
mass balance of the electrolyte components NH3 and
CO2 in the temperature interval ∆T can be written as

∆Yi,k ) mi(total)(Tk) MH2O∆YCOND,k (29)

with the molar mass of water MH2O in kilograms per

mole.
The molar load Yi of the electrolyte in the dry gas is
related to the molar fraction yi used in eqs 17 and 18
according to

Yi ) yi(1 + YH2O) (30)

Equations 29 and 30 imply that the impact of the

concentration changes of NH3 and CO2 on the total
amount of dry gas is negligible because of the small
quantities of NH3 and CO2 removed specific to the total
dry gas flow. The same applies for the aqueous liquid
phase, where the low concentrations of dissolved species
do not affect the total liquid mass flow rate.
The model presented above allows the calculation of
the gas-phase NH3 profile along the temperature coor-
dinate if the clean gas CO2 concentration and at least
one of the NH3 concentrations in condensate, raw gas,
or clean gas are known. The main dry gas components,
the raw gas water content, and the in/out temperatures
of the gas and liquid streams are also necessary to
determine the energetic section. The required raw gas
water content YH2O,in can be substituted by the total Figure 5. Calculation algorithm for a given input NH3 concentra-
specific amount of condensate YCOND according to eq 12. tion Y′NH3,in.
 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005 1581

Figure 6. Specific water load in gas and specific amount of Figure 7. Partial pressures of CO2 and NH3 for different NH3
condensate versus the gas-phase temperature inside the counter- concentrations in raw gas versus the gas-phase temperature inside
current column. the column.

system that can be solved straightforward and yielding

a set of concentrations for a given mOH-. The solver finds
the mOH- for which the electric balance (24) is fulfilled.
When the algorithm has reached the section where
condensation starts, the calculated raw gas NH3 con-
centration is compared to the target value and the solver
generates a new starting guess for the exit gas concen-
tration until the values match. A minor modification in
the algorithm allows the calculation of both gas-phase
NH3 compositions if the total amount of NH3 in the
condensate mNH3(total),out is known. The algorithm is fast
and robust with respect to convergence. Numerical
errors fade with an increasing number of elements and
can be considered insignificant for n g 20.

4. Results and Discussion

Concentration Profiles in the Scrubber. As in-
troduced above, the gas-phase temperature is used as
Figure 8. NH3 load on dry gas for different raw gas NH3
a coordinate along the column’s height. In typical oper- concentrations versus the gas-phase temperature inside the
ation conditions of the Guessing scrubber, concentration column.
profiles are presented for a producer gas composition
according to Table 1, gas in/out temperatures of 170/50 Influence of Process Parameters on NH3 Re-
°C, solvent in/out temperatures of 40/80 °C, and a water moval. According to the model, the NH3 separation
content in the raw gas of 40 vol %. The raw gas NH3 efficiency of the scrubber depends on the following
content is varied from 500 to 2000 vppm (dry). process parameters: (i) clean gas temperature; (ii) raw
Figure 6 shows the water load in the gas stream and gas water content; (iii) CO2 content in gas; (iv) NH3
the specific amount of condensate versus gas-phase content in gas. The clean gas temperature and raw gas
temperature. According to the model assumption that water content determine the amount of condensate
no re-evaporation of condensate occurs in the lower part according to Figure 9. A comparison between Figures 9
of the column, the water loads are constant in that and 10 shows that the effect of TG,out and yH2O,in on the
region. The partial pressures of NH3 and CO2 in the gas NH3 separation efficiency cannot be reduced to the
phase are presented in Figure 7. The CO2 pressure is parameter YCOND only. The reason is the nonlinearity
increasing toward the top of the column because the of the gas-liquid equilibrium. The effect of the clean
absorbed CO2 does not significantly affect the gas gas CO2 content and raw gas NH3 content on the
composition and the total amount of gas is decreasing separation efficiency is shown for a fixed clean gas
with condensation. The same effect results in almost temperature and raw gas water content in Figure 11.
unchanged partial pressures of NH3 along the column’s The corresponding pH value of the condensate is
height. The actual NH3 removal in the scrubber is presented in Figure 12.
shown by NH3 in dry gas in Figure 8. The separation Comparison with Measured Data. Extensive NH3
efficiency defined as measurements have been carried out at the Guessing
plant. The solution of partial energy balances on the
Yi,out column allows the calculation of the liquid-phase tem-
ηsep,i ) 1 - (31)
Yi,in perature corresponding to gas-phase temperatures. For
the several sampling runs, the calculated pinch tem-
is 50% for 500 vppm of NH3 in dry raw gas and only peratures ∆TGL,C at the section where condensation
30% for 2000 vppm of NH3 in dry raw gas. starts vary between 0.5 and 2.0 K. If the measured raw
1582 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005

Figure 9. Specific amount of condensate versus the raw gas water

content and clean gas temperature.

Figure 12. pH value of the condensate at 298.15 K versus the

clean gas CO2 content and raw gas NH3 content.

Figure 10. NH3 separation efficiency of the scrubber versus the

clean gas temperature and raw gas water content.

Figure 13. Comparison of the clean gas NH3 concentration

measured with the value calculated from the measured raw gas
concentration.

such as temperatures, raw gas water content, and CO2

content in clean gas. The accuracy of the analytical
determinations of gas-phase NH3 is estimated to be
about (100 vppm (dry) for NH3 concentrations in the
range of 300-2000 vppm (dry). The same error must
be considered for the calculated values, which, of course,
depend on the measured raw gas concentrations. The
separation efficiencies, calculated from the gas concen-
tration before and after the scrubber, are compared in
Figure 14.
The model tends to overestimate the NH3 removal
effect of the scrubber, which results in slightly lower
calculated clean gas NH3 concentrations than those
found by the measurements.
Figure 11. NH3 separation efficiency of the scrubber versus the Possible reasons for the deviations between the model
clean gas CO2 content and raw gas NH3 content. and measurements could be the model assumptions and
measurement inaccuracies. One possible explanation is
gas concentration is taken as the input parameter for that, despite the model assumption, NH3 is transferred
calculation, the clean gas concentration and the separa- through the organic phase from the hot water phase
tion efficiency respectively can be calculated. Figure 13 back to the gas phase. This may happen in the hot lower
shows measured clean gas NH3 values dependent on the part of the column and also in the equalization tank. A
values calculated from the corresponding raw gas NH3 global mass-transfer parameter for NH3 could be intro-
concentrations and the specific operation parameters duced to account for these deviations. To avoid the
 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005 1583

the prediction of clean gas values even though a

tendency toward an overestimation of NH3 removal can
be observed. If gas volatilization from the emulsified
water is the reason for the lower NH3 separation
efficiencies measured, the calculation results represent
an upper bound for NH3 removal in the scrubber.
Further investigation will focus on the variation of
operating parameters in the Guessing gas-cleaning
system in order to learn more about the possibilities of
increased NH3 removal by simple measures.

Literature Cited
(1) Hofbauer, H.; Rauch, R.; Loeffler, G.; Kaiser, S.; Fercher,
E.; Tremmel, H. Six years experience with the FICFB-Gasification
process. In Twelfth European Biomass Conference; Palz, W.,
Spitzer, J., Maniatis, K., Kwant, K., Helm, P., Grassi, A., Eds.;
ETA: Florence, Italy, 2002; pp 982-985.
(2) Kehlenbeck, R.; Yates, J.; Di Felice, R.; Hofbauer, H.; Rauch,
R. Novel Scaling Parameter for Circulating Fluidized Beds. AIChE
Figure 14. Comparison of the measured and calculated NH3 J. 2002, 47, 582-589.
separation efficiencies of the scrubber. (3) Hofbauer, H.; Rauch, R.; Bosch, K.; Koch, R.; Aichernig, C.
Biomass CHP plant Guessingsa success story. In Pyrolysis and
Gasification of Biomass and Waste; Bridgwater, A. V., Ed.; CPL
measurement errors, all triples of NH3 concentrations Press: Newbury, U.K., 2003; pp 371-383.
(raw gas, clean gas, and condensate) that did not fulfill (4) Boerrigter, H.; den Uil, H.; Calis, H. P. Green Diesel from
the mass balance were excluded from consideration. Biomass via Fischer-Tropsch Synthesis: New Insights in Gas
Cleaning and Process Design. In Pyrolysis and Gasification of
Biomass and Waste; Bridgwater, A. V., Ed.; CPL Press: Newbury,
5. Conclusions U.K., 2003; pp 527-536.
(5) Reisinger, K.; Haslinger, C.; Herger, M.; Hofbauer, H.
Within the present work, a relatively simple model BIOBIBsa Database for Biofuels. THERMIEsConference: Re-
of coabsorption of NH3 and CO2 into condensing water newable Energy Databases, Harwell, U.K., 1996 (http://www.
in a countercurrent organic solvent scrubber has been vt.tuwien.ac.at/biobib/biobib.html).
developed based on the following assumptions: (i) (6) Edwards, T. J.; Newman, J.; Prausnitz, J. M. Thermody-
exiting gas relative humidity of 100%; (ii) no radial namics of Aqueous Solutions Containing Volatile Weak Electro-
lytes. AIChE J. 1975, 21, 248-259.
gradients and no axial mixing in the column; (iii) pure
(7) Edwards, T. J.; Maurer, G.; Newman, J.; Prausnitz, J. M.
organic solvent at the top of the column; (iv) no mixing Vapor-Liquid Equilibria in Multicomponent Aqueous Solutions
between water and the organic phase; (v) no solubility of Volatile Weak Electrolytes. AIChE J. 1978, 24, 966-976.
of NH3 in the organic solvent; (vi) gas-liquid equilib- (8) Tsonopoulos, C.; Coulson, D. M.; Inman, L. B. Ionization
rium during condensation; (vii) condensate behaving as Constants of Water Pollutants. J. Chem. Eng. Data 1976, 21, 190-
an ideal electrolyte solution; (viii) no re-evaporation of 193.
the condensate in the hot lower part of the column; (ix) (9) Marshall, W. L.; Franck, E. U. Ion Product of Water
Substance, 0-1000 °C, 1-10,000 Bars New International Formu-
no gas volatilization from emulsified water. lation and Its Background. J. Phys. Chem. Ref. Data 1981, 10,
The gas-phase temperature is applied as a height 295-304.
coordinate of the column. Therefore, no geometry data (10) Kawazuishi, K.; Prausnitz, J. M. Correlation of Vapor-
are needed in the modeling, and the model does not Liquid Equilibria for the System Ammonia-Carbon Dioxide-
contain any empirical parameter that needs to be Water. Ind. Eng. Chem. Res. 1987, 26, 1482-1485.
determined from measurements. An algorithm is pre- (11) Clegg, S. L.; Brimblecombe, P. Solubility of Ammonia in
Pure Aqueous and Multicomponent Solutions. J. Phys. Chem.
sented that allows the calculation of clean gas and 1989, 93, 7237-7248.
condensate NH3 contents if NH3 in raw gas and certain (12) Knapp, H.; Döring, R.; Oellrich, L.; Plöcker, U.; Prausnitz,
process parameters are known. Calculations show for J. M.; Langhorst, R.; Zeck S. Vapor-Liquid Equilibria for Mixtures
the typical operation range of the producer gas scrubber of Low Boiling Substances; DECHEMA Chemistry Data Series VI;
that the partial pressure of NH3 does not change DECHEMA: Frankfurt, Germany, 1982; p 690.
significantly along the column’s height because of a (13) Gillespie, P. C.; Wilding, W. V.; Wilson, G. M. Vapor-
Liquid Equilibrium Measurements on the Ammonia-Water System
decrease in the total gas quantity (condensation of from 313 to 589 K; Research Report RR-90; Gas Processors
water). The separation efficiency is strongly dependent Association: Tulsa, OK, 1985.
on the gas exit temperature, raw gas water content, CO2 (14) Müller, G.; Bender, E.; Maurer, G. Das Dampf-Flüs-
content in gas, and raw gas NH3 content. While tar sigkeitsgleichgewicht des ternären Systems Ammoniak-Kohlen-
separation is very efficient, the NH3 separation efficien- dioxid-Wasser bei hohen Wassergehalten im Bereich zwischen
cies in the possible operation range of the scrubber are 373 und 473 Kelvin. Ber. Bunsen-Ges. Phys. Chem. 1988, 92, 148-
160 (written in German).
limited to about 50%. Quantitative removal of NH3 from (15) Plummer, N.; Busenberg, E. The solubilities of calcite,
the producer gas can be reached with further gas- aragonite, and vaterite in CO2-H2O solutions between 0 and 90
cleaning units, which may be economically critical at °C, and an evaluation of the aqueous model for the system CaCO3-
small plant capacities. On the other hand, primary CO2-H2O. Geochim. Cosmochim. Acta 1982, 46, 1011-1040.
catalytic measures in the hot raw gas can be set in order (16) Chen, C. C.; Britt, H. I.; Boston, J. F.; Evans, L. B.
to decrease the load of the raw gas and to reach certain Extension and Application of the Pitzer Equation for Vapor-
Liquid Equilibrium of Aqueous Electrolyte Systems with Molecular
NH3 limits in clean gas after the tar scrubber. Solutes. AIChE J. 1979, 25, 820-831.
Extensive NH3 measurements have been performed (17) Pawlikowski, E. M.; Newman, J.; Prausnitz, J. M. Phase
at the organic solvent scrubber in Guessing, Austria. Equilibria for Aqueous Solutions of Ammonia and Carbon Dioxide.
The model evaluation shows satisfactory agreement for Ind. Eng. Chem. Process Des. Dev. 1982, 21, 764-770.
1584 Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005

(18) DIPPR project 801 (http://dippr.byu.edu/). Report TAE 804; Technion Israel Institute of Technology: Tech-
(19) The International Association for the Properties of nion City, Haifa, Israel, 1997 (http://garfield.chem.elte.hu/Burcat/
Water and Steam. Release on the IAPWS Industrial Formulation burcat.html).
1997 for the Thermodynamic Properties of Water and Steam, Received for review April 23, 2004
Erlangen, Germany, Sept 1997 (http://www.iapws.org/relguide/ Revised manuscript received December 7, 2004
IF97.pdf).
Accepted December 10, 2004
(20) Burcat, A.; McBride, B. Ideal Gas Thermodynamic Data
for Combustion and Air-Pollution Use; Aerospace Engineering IE049669V