Fuel Processing Technology 134 (2015) 243250

Contents lists available at ScienceDirect

Fuel Processing Technology

journal homepage: www.elsevier.com/locate/fuproc

Design and development of a bench scale vegetable oil based wet packed
bed scrubbing system for removing producer gas tar compounds
Prakashbhai R. Bhoi a, Raymond L. Huhnke a,, Ajay Kumar a, Krushna N. Patil a, James R. Whiteley b
a
Department of Biosystems and Agricultural Engineering, Oklahoma State University, Stillwater, OK 74078, USA
b
School of Chemical Engineering, Oklahoma State University, Stillwater, OK 74078, USA

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

Article history: The overall goal of this study was to design, fabricate and evaluate a vegetable oil based wet packed bed scrubbing
Received 30 June 2014 system to study the removal of model producer gas tar compounds, i.e., benzene, toluene, and ethylbenzene. The
Received in revised form 23 January 2015 designed column was 0.5 m diameter by 1.1 m tall and contained 6-mm raschig rings. Soybean and canola oils
Accepted 30 January 2015
were used as solvents. These oils were not statistically (p N 0.05) different in removal efciencies of benzene,
Available online 14 February 2015
toluene and ethylbenzene. Packing specic constants of Billet and Schultes (1999) correlation (CV = 0.80 and
Keywords:
CL = 2.40) were found experimentally. These constants can be used to predict height equivalent to a theoretical
Biomass producer gas tars plate (HETP) of laboratory or pilot scale vegetable oil based wet packed bed scrubbing systems. The pressure drop
Wet packed bed scrubbing system across the column was in the range of 6.7 to 18.8 mm of water column.
Vegetable oil 2015 Elsevier B.V. All rights reserved.

1. Introduction
Given the growing concerns over climate change, national energy
security and increasing global demands, there is an urgent need for de-
veloping sustainable biofuels, bioproducts, biopower, and bioenergy [1].
Biomass has been identied as a potential source of sustainable energy
due to its availability in large quantities [2,3]. Biomass conversion using
thermochemical processes is one of the pathways in producing liquid
biofuels and chemicals.
Gasication is one of the most common thermochemical conversion
technologies. The major challenge of the gasication pathway is the re-
moval of impurities including tars [46]. Tars mainly consist of aromatic
and poly aromatic hydrocarbons for the gasication temperature over
800 C [7]. Primary methods of tar removal include optimization of op-
erating and design parameters. Operating variables such as gasifying
agent, equivalence ratio (ER), temperature, and residence time have
signicant effects on the tar formation and decomposition [8]. In the
secondary methods, tar reduction measures are implemented on the
downstream side of the gasier. Secondary tar removal approaches
employ dry or wet-type equipment or high temperature (thermal or
catalytic cracking) processes [9]. Thermal and catalytic tar cracking pro-
cesses obtain high conversion efciencies; however, these processes are
energy intensive and require an external energy supply to maintain the
Corresponding author at: 223 Agriculture Hall, Department of Biosystems and
Agricultural Engineering, Oklahoma State University, Stillwater, OK 74078-6021, USA.
Tel.: +1 405 744 8417; fax: +1 405 744 6059.
E-mail addresses: prakash.bhoi@okstate.edu (P.R. Bhoi),
raymond.huhnke@okstate.edu (R.L. Huhnke), ajay.kumar@okstate.edu (A. Kumar),
patilkn@hotmail.com (K.N. Patil), rob.whiteley@okstate.edu (J.R. Whiteley).
high operating temperatures. Dry type cleaning devices include cyclone
separators, electrostatic precipitators (ESP), and various types of lters.
These equipments are mainly used for the particulate removal [1012].
Wet gas cleaning processes are often used for tar removal. These include
spray towers, impingement scrubbers, venturi scrubbers, packed bed
scrubbers, wet electrostatic precipitators, wet cyclones, and oil based
gas washers.
Spray towers and impingement, venturi and packed bed scrubbers
are the most common types of wet scrubbers applied in producer gas
cleaning application [1317]. The performance of wet scrubbing
systems is measured by wet scrubbing efciency which is dened as
the concentration difference between the inlet and outlet streams divid-
ed by the inlet stream concentration [18]. Spray towers are simple and
inexpensive; however, they are the least efcient [19]. Impingement
scrubbers are about 70% efcient. To achieve higher efciencies, im-
pingement scrubbers are connected in series [20]. Tar removal efcien-
cies of venturi scrubbers range from 50 to 90% [10]. Packed bed wet
scrubbers have the highest tar removal efciencies of 80 to 99% [21].
Wet scrubbing of tars involves absorption of gaseous tar compounds
in the scrubbing solvents. Water is the most common scrubbing solvent
in biomass producer gas based wet scrubbing studies [14,15,17,19]. The
major drawbacks in using water as a scrubbing solvent is the low solu-
bility of tar compounds and costly wastewater treatment processes [22].
Vegetable, engine, and waste cooking oils and biodiesel have been stud-
ied as solvents for the removal of tars and volatile organic compounds
such as benzene, toluene, and, methanol in a bubble column [14,
2325]. Tar compounds including benzene, toluene, and ethylbenzene,
being lipophilic in nature, can mix comparatively well with vegetable
oils because these oils have saturated (palmitic and steric acids) and un-
saturated (oleic, linolenic and linoleic acids) fatty acids which are also

http://dx.doi.org/10.1016/j.fuproc.2015.01.042
0378-3820/ 2015 Elsevier B.V. All rights reserved.
244 P.R. Bhoi et al. / Fuel Processing Technology 134 (2015) 243250

lipophilic in nature. Soybean oil is one of the largest sources of vegetable
oils in the U.S. [26]. Additionally, the state of Oklahoma is the second
largest producer of canola oil. These oils, i.e., soybean and canola, are
less expensive than organic solvents such as acetone and isopropanol
[27]. Because very little information is available on vegetable oil based
wet packed bed scrubbing systems, the objective of this study is to
design, fabricate and evaluate such a system for the removal of tar
compounds.
2. Materials and methods
The design (operating) gas velocity of the column is often determined
using the manufacturer's pressure drop correlation for the given pack-
ings. In the absence of manufacturer's data, the diameter of the
column of 0.5 m was determined using a generalized pressure-drop
correlation (GPDC) of Eckert [3133] as modied by Strigle [34].
2.1.2. Height of the column
The height of the packed bed column was determined using equilib-
rium stages and height equivalent to a theoretical plate (HETP).

2.1. Design basis

Z N  HETP 1
Perry et al. [28] stated that the design condition (pressure, tempera-
ture and liquid-to-gas ratio) is normally selected by volatility or solubil-
where,
ity of the least soluble compound when there are no chemical reactions
involved. Solubility data are determined at equilibrium conditions. The
Z is the height of the packed bed column, m
equilibrium conditions are important for the design of an absorption
N is the number of equilibrium stages (ideal stages)
column because the absorption efciency reaches zero when the
HETP is the height equivalent to a theoretical plate, m.
equilibrium conditions are attained. An important design parameter is
the vaporliquid equilibrium ratio (K-value) which is dened as the
An equilibrium stage (theoretical plate) is dened as the hypotheti-
ratio of mole fractions of a species in two phases at equilibrium [29].
cal stage where liquid and gas phases establish equilibrium with each
The K-value of the tar compounds was determined through a ash
other. Perry et al. [28] recommended the use of a rigorous computation-
calculation using a non-random two-liquid (NRTL) model because
al method to perform design calculation of multicomponent system.
NRTL provides a better vaporliquid equilibrium (VLE) prediction than
Therefore, to estimate the equilibrium stages, an equilibrium stage
the other thermodynamic models [30].
based model was developed using the RadFrac block in Aspen PlusTM.
The compositional analyses for tars generated from uidized bed
The operating condition (1 atm and 50 C) was selected based on pre-
gasication of switchgrass is reported by Cateni [13]. The major tar
liminary experiments. The inlet stream of vegetable oil was dened
compounds in wt.% were: benzene (29%), toluene (18%), phenol
using fatty acid composition as listed in Table 1. Non-random two-
(14%), ethylbenzene (10%), methyl phenol (8%), styrene (6%), xylene
liquid (NRTL) model was used as a thermodynamic model to account
(5%), naphthalene (4%), dimethyl naphthalene (3%) and methyl naph-
for non-idealities in the system. The missing binary interaction param-
thalene (3%), which represents about 75% of the total mass of the tar.
eters were estimated using universal quasi-chemical (UNIQUAC)
To determine the least soluble tar compounds, ash calculations were
functional-group activity coefcients (UNIFAC) [30]. A detailed descrip-
performed at atmospheric conditions (1 atm and 50 C) which were de-
tion of NRTL model and its binary interaction parameters are provided
termined based on preliminary studies. The calculations showed that
in Bhoi et al. [35] and Bhoi [36]. A sensitivity analysis of equilibrium
benzene, toluene and ethylbenzene had the lowest solubilities in vege-
stage versus purity of outlet vapor stream was performed to determine
table oils; and these tar compounds represent nearly 60% of total tars.
the optimum number of equilibrium stages. Ten equilibrium stages
Therefore, benzene, toluene and ethylbenzene were used as model tar
were determined as optimal for the purity of the outlet vapor stream
compounds in the design and evaluation of the wet scrubbing system.
of 99.4%. Another method, i.e., Kremser's method [28] which is normally
Concentration levels of benzene, toluene, and ethylbenzene used were
recommended for the dilute system, was also used to determine theo-
8355, 4206 and 2439 ppmv, respectively. The K-values of benzene, tol-
retical number of stages as a check. The same numbers of theoretical
uene and ethylbenzene were 0.30, 0.10 and 0.04, respectively. Because
stages, i.e., 10, were determined through Kremser's method.
benzene is the least soluble compound due to its higher vaporliquid
The height equivalent to a theoretical plate (HETP) includes the
equilibrium ratio (K-value), it was used to design the packed bed
packed bed column mass transfer efciency. HETP can be predicted by
column.
data interpolation, mass-transfer models, or rules of thumb proposed
One of the unknown parameters in the design of wet scrubbing sys-
by Kister [37] and Perry et al. [28]. Kister [37] stated that the most
tem is the ow rate of liquid required to achieve the desired absorption
recommended method of HETP prediction is the data interpolation. In
efciency. The actual liquid-to-gas (L/G) ratio is usually greater than the
the absence of experimental data, the HETP can be predicted through
minimum L/G by as much as 25 to 100% [28]. Seader et al. [29] stated
mass-transfer models such as Billet and Schultes [38] correlation. How-
that the minimum L/G is equilibrium ratio (K value) times recovery,
ever, Billet and Schultes [38] correlation requires empirical packing
i.e., fraction of solute absorbed for a dilute system and solute free sol-
specic constants (CL and CV). In this study, these constants were deter-
vent. The minimum L/G ratio of 0.3 was determined from the equilibri-
mined using experimental data.
um ratio (K value) and using recovery of more than 99%. Researchers
suggested that the operating L/G ratio should be 20 to 50% higher than
the minimum L/G required to minimize total annual cost. Accordingly,
Table 1
operating liquid-to-gas (L/G) ratio of 0.45 (50% higher than the Properties of vegetable oils.
minimum L/G ratio) was used in this design.
Parameters Soybean oil Canola oil

2.1.1. Diameter of the column Palmitic acid (16:0), % 9 4.8

The minimum column diameter is limited by ooding, and the typi- Steric acid (18:0), % 4.4 1.8
Oleic acid (18:1), % 26.4 57.4
cal design considers the operating gas velocity in the range of 50 to 70%
Linoleic acid (18:2), % 51.6 22.3
of the ood velocity, i.e., the velocity of gas which prevents the ow of Linolenic acid (18:3), % 6.8 4.5
liquid causing a substantial increase in the pressure drop across the col- Density, kg/m3 922.5 917
umn and limits the mixing of gas and liquid phase [29]. In addition, Viscosity, mm2/s 65.4 39.2
Seader et al. [29] also stated that the diameter must be determined Heating value, MJ/kg 37 37
Flash point, C N288 N230
such that the pressure drop is below 1.5 in H2O/ft of packed bed height.
 P.R. Bhoi et al. / Fuel Processing Technology 134 (2015) 243250
Fig. 1. Photographic view of raschig rings.
2.1.3. Selection of packings
Random packings were selected due to its low maintenance and
wide variety that are commercially available. Metal packings are often
recommended due to its strength and good wettability [29]. Packings
nominal size of less than one-eighth of the column diameter is recom-
mended to minimize the liquid maldistribution [29]. For these reasons,
metal raschig rings of 6-mm (stainless steel) were selected as the
packing materials. The characteristics of stainless steel raschig rings
6 6 0.3 mm in size, i.e., diameter length thickness (Fig. 1),
purchased from Raschig Jaeger Technologies, Arlington, TX are shown
in Table 2.
2.1.4. Solvents
Soybean and canola oils were used as solvents in this study. The
properties of these oils, as purchased from Jedwards International,
Inc., Quincy, MA, are given in the Table 1. The numbers in the bracket
of each fatty acid compound represent the number of carbon atoms
and double bonds. For example, palmitic acid (16:0) means that this
acid has 16 carbon atoms and no double bonds. Fatty acids are classied
based on the presence or absence of double bonds. Fatty acids with no
double bonds, one double bond and two or more double bonds are
called saturated, monosaturated and polysaturated fatty acids,
respectively.
As shown in Table 1, the saturated compounds (palmitic and steric
acids) in soybean oil are more than double than those of canola oil.
Table 2
Characteristics of raschig rings.
Parameters
Size (diameter length thickness), mm
Density, kg/m3
Surface area, m2/m3
Packing factor, 1/m
Void fraction, %
245
Consequently, the viscosity of soybean is much higher than that of cano-
la oil. The other unsaturated compounds (sum of oleic, linoleic and
linolenic acids), heating value and density are comparatively equal. An-
other important parameter is the surface tension () of vegetable oils
which is in the range of 30 to 32 mN/m for the temperature range of
30 to 50 C [39,40]. The surface tension of soybean and canola oils are
close to the recommended surface tension ( b 25 mN/m) of solvent
which improves the wettability resulting in higher mass transfer
efciencies.
2.2. Experimental set-up
A wet packed scrubbing system of 0.6 Nm3/h capacity (schematic
shown in Fig. 2) was designed, fabricated, and installed in the Bioenergy
Laboratory at Oklahoma State University, Stillwater, OK.
2.2.1. Packed bed absorption column
The wet packed bed scrubbing system as shown in Fig. 2 consists of a
stainless steel column (5.25 cm diameter), peristaltic pump (model PC2
70-7002, Harvard Apparatus, Holliston, MA), and water bath heater
(model WD28A11B, Grainger, Roanoke, TX). Stainless steel (SS) 304
was selected as the material of construction of the column due to an ex-
cellent resistance towards the model tar compounds used in this study.
A design basis of the determination of the diameter and height of the
packed bed absorption column is given in Section 2.1. The column
(5.25 cm internal diameter and 150 cm height) holds the packed bed
material (raschig rings) having a height of 50110 cm. The column
packed bed height was varied by lling 6 mm raschig rings randomly
to a desired height and accordingly a liquid distributor was extended
into the column to maintain 25 mm distance between the distributor
and the top surface of packed bed. The packing (raschig rings) is sup-
ported at the bottom of the column by a corrosion resistant SS 316
woven wire cloth (4 4 mesh, open area of 46.2%). The size (diameter)
of inlet and outlet stainless steel pipes of wet packed bed scrubbing unit
is 1.25 cm. The packed bed column was insulated using 6.5 mm thick
berglass pipe insulation (part number 45545K41, McMaster-Carr) to
minimize heat loss. The solvent pipe was wrapped with a high-
temperature heat cable with a temperature controller (part number
3631K23, McMaster-Carr) to maintain the desired solvent temperature.
The solvent pipe wrapped with heat cable was insulated using 2.5 cm
exible ceramic insulation to prevent heat loss.
2.2.2. Gas mixing
Barratt [41] indicated that dynamic methods of standard gas mix-
tures are preferred over static methods. He also stated that an injection
method is a versatile approach for preparing standard gas mixtures, and
syringe pumps are the most widely used for the injection method. In
this study, a simulated air containing a mix of tars was prepared in a
gas mixing section by injecting known amount of pure liquid model
tar compounds into a stream of hot air (Fig. 2). The gas mixing section
consists of a compressed air line, on/off valve, pressure regulator
(model 4ZK96, Grainger, Roanoke, TX) with a pressure gauge (model
4FLH6, Grainger, Roanoke, TX), air ow switch (Model FS-926 BR A
SCFH-00.50, Gems Sensors Inc., Plainville, CT), mass ow controller
(Model GFC37S-VADL2-A0, Aalborg Instruments and Controls, Inc.,
Orangeburg, NY), heater (Model HT-M-050-100-120-1/8F-1/4F-TF1,
Tutco-Farnam Custom Products, Arden, NC) with temperature control-
ler (Model CC-A10, Tutco-Farnam Custom Products, Arden, NC), check
valve (part number SS-8CPA2-3, Swagelok Oklahoma, Tulsa, OK),
syringe pump (Model KDS 280.200, KD Scientic, Holliston, MA), and
Values rupture disc (part number 4412T12, McMaster-Carr).
6 6 0.3 A manual ball valve was used to start or stop the air supply before
900 the pressure regulator. The regulator controlled the pressure of the
900 inlet air. The air ow switch was installed prior to the heater and sensed
2297 air ow. If air ow was present, power was supplied to the heater to
89
heat the inlet air to 350 C to ensure that the injected liquid tar
246 P.R. Bhoi et al. / Fuel Processing Technology 134 (2015) 243250
Fig. 2. Schematic diagram of bench-scale wet packed bed scrubbing set-up.
compounds would vaporize completely. A check valve was installed be-
tween the heater and the injection port of the tar compounds to prevent
back-ow of tar vapors. As a safety precaution, a brass rupture disc was
installed prior to the absorption column to release excessive pressure.
The model tar compounds were procured from Sigma Aldrich. The
purity values of benzene, toluene and ethylbenzene were 99.5%, 99.7%,
and 99.8%, respectively [18]. A mixture of tar compounds was prepared
by weight distribution, benzene: 50%, toluene: 30%, and ethylbenzene:
20% which is comparable to the content of tar collected from a uidized
bed gasier [13]. The prepared mixture was lled in a 100-ml gas-tight
syringe (part number 009760, SGE Analytical Science, Austin, TX) with a
luer lock needle (part number 039827, SGE Analytical Science, Austin,
TX). A syringe pump (Model KDS 280.200, KD Scientic, Holliston,
MA) was used to maintain a constant injection rate of the model tar
compounds mixture into the hot air stream.
2.2.3. Liquid (solvent) distribution
The liquid distribution is the most important aspect of the wet scrub-
bing system because it inuences the mass transfer efciency of the
packing. Perry et al. [28] stated that a single spray nozzle can serve the
purpose of liquid distribution in a small column. Accordingly, prelimi-
nary tests were conducted using six different types of commercial
spray nozzles to determine the spray distribution of soybean and canola
oils. Results showed that the selected nozzles were not able to
adequately spray the oils due to the high viscosities of oils. Other
spray nozzle options claiming to handle high viscous liquids were
explored; however, these were deemed too expensive. As the result, it
was decided to design and fabricate a liquid distributor to spray vegeta-
ble oils uniformly on the top of the packings.
Perry et al. [28] stated that the orice liquid distributor is a common
type of liquid distributor for an absorption process. The liquid distribu-
tor was designed based on the head-ow correlation given below.
p
Q C D Ah n 2gh
where,
Q is the volume ow rate, m3/s
CD is the coefcient of discharge = 0.4
Ah is the cross sectional area of a hole, m2
n is the number of discharge hole of a liquid distributor
g is the gravitational acceleration = 9.8 m/s2
h is the liquid head above the orice, m.
Using the above head-ow correlation, a liquid distributor was de-
signed considering the diameter of the holes of 0.15 mm, design solvent
ow rate of 63 ml/min, pressure head of 20 psig, and coefcient of dis-
charge of 0.4. The number of holes were determined as 9. Perry et al.
[28] recommended at least 40 irrigation points per m2 with 60
100 per m2 being ideal. He also stated that a drip point density over
40 showed little difference in packing efciency. In this study, a drip
point density of 4125 drip points/m2 was determined considering that
the liquid distributor consists of 9 holes of 0.15 mm diameter and a
packed column of 50 mm diameter which is well above the required
drip point density. As a comparison, Janzen et al. [42] distributed
water in an 80-mm diameter column using a multipoint source distrib-
utor, having a drip point density of 3976 drip points/m2.
A liquid (solvent) distribution system consists of a multi-point liquid
distributor to provide a uniform distribution of solvent (Fig. 3), a sump,
and peristaltic pump (model PC2 70-7002, Harvard Apparatus,
Holliston, MA) which is compatible with the selected solvents. The
sump (0.8 l in volume) was placed in the water bath heater to maintain
the solvent at the desired temperature.
2.2.4. Instrumentation and measurements
A mass ow controller (Model GFC37S-VADL2-A0, Aalborg Instru-
ments and Controls, Inc., Orangeburg, NY) was used to measure and
control the air ow rate. Type-K thermocouples were used to measure
the temperature of inlet and outlet air streams, air at tar injection
point, and solvent. A U-tube manometer (020 of H2O) was installed
on the packed bed column to monitor the pressure drop across the col-
umn. A differential pressure transducer (Model number GC52, Ashcroft
2 Inc., Stratford, CT) was used to log the pressure drop across the column.
Temperatures, air ow rate, and pressure drop across the column were
recorded using a DaqView program (part number OMB-DAQ-55, Omega
Engineering, Inc., Stamford, CT, USA). The compositions of inlet and out-
let gas streams were analyzed using a gas chromatography/mass spec-
troscopy (GC/MS). Two replications of each experimental condition
were used. A statistical analysis was performed using a split plot
 P.R. Bhoi et al. / Fuel Processing Technology 134 (2015) 243250 247

100

Toluene Removal Efficiency, %

80

60

40 SO - 30C
SO - 40C
SO - 50C
20 CO - 30C
CO - 40C
CO - 50C
0
0 10 20 30 40 50 60
Time, min

Fig. 5. Effect of solvent type and temperature on the removal efciency of toluene at a bed
height of 0.5 m and solvent ow rate of 53 ml/min [SO: soybean oil, CO: canola oil at
solvent temperatures of 30, 40 and 50 C].
Fig. 3. Photographic view of solvent distributor.

Cout is the concentration of tar compound at the outlet of the

model and using analysis of variance (ANOVA) procedure to evaluate column, ppmv.
the performance of wet packed bed scrubbing system. A detailed de-
scription of experimental design and the statistical analysis is given in The effect of vegetable oil type was analyzed at the lowest bed height
Bhoi et al. [18]. of 0.5 m, three levels of solvent temperature and three levels of solvent
ow rate (Figs. 46). Both soybean and canola oils follow the same trend
3. Results and discussion for all the conditions of solvent temperatures and ow rates. Statistical
analysis showed that the removal efciencies of tar compounds using
3.1. Performance of wet scrubbing system using soybean and canola oils as soybean and canola oils are not statistically (p N 0.05) different. The ef-
a solvent fect of solvent type was also determined at the highest bed height of
1.1 m with similar results.
The performance of the wet scrubbing system was evaluated in As shown in Figs. 46, as the solvent temperature increased from 30
terms of removal efciencies () of tar compounds. The removal ef- to 50 C, the removal efciencies of benzene, toluene, and ethylbenzene
ciency was determined using following equation: decreased signicantly due mainly to reduction in solubility of tar com-
pounds in vegetable oils. In addition, reduction in the density and the
C in C out
; %  100 3 viscosity of vegetable oil as the solvent temperature increases leads to
C in
a reduction in interfacial area for mass transfer of tar compounds. There-
fore, the removal efciencies of tar compounds decreased signicantly
where,
as the solvent temperature increased from 30 to 50 C. Results also
showed that the trends of removal efciencies with time were linear
is tar removal efciency, %
for toluene and ethylbenzene due to higher solubilities of these com-
Cin is the concentration of tar compound at the inlet of the
pounds in vegetable oil compared to benzene. As shown in Figs. 46,
column, ppmv
the tar removal efciencies reduce with time because the concentration

100 100
Ethylbenzene Removal Efficiency, %
Benzene Removal Efficiency, %

80 80

60 60

40 SO - 30C 40 SO - 30C
SO - 40C SO - 40C
SO - 50C SO - 50C
20 CO - 30C 20 CO - 30C
CO - 40C CO - 40C
CO - 50C CO - 50C
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time, min Time, min

Fig. 4. Effect of solvent type and temperature on the removal efciency of benzene at a bed Fig. 6. Effect of solvent type and temperature on the removal efciency of ethylbenzene at
height of 0.5 m and solvent ow rate of 53 ml/min [SO: soybean oil, CO: canola oil at a bed height of 0.5 m and solvent ow rate of 53 ml/min [SO: soybean oil, CO: canola oil at
solvent temperatures of 30, 40 and 50 C]. solvent temperatures of 30, 40 and 50 C].
248 P.R. Bhoi et al. / Fuel Processing Technology 134 (2015) 243250

of tar compounds in the solvent increases due to solvent recycling. It is
also important to note that the concentration of solute in solvent has a
signicant impact on tar absorption efciency [39].
3.2. Determination of Billet and Schultes [38] packing specic constants for
the prediction of height equivalent to a theoretical plate (HETP)
aph is the specic interface area between the phases, m2/m3
dh is the hydraulic diameter of the dumped packing, m
L is the kinematic viscosity of liquid, m2/s
V is the kinematic viscosity of gas or vapor, m2/s
DV is the diffusion coefcient of solute in gas or vapor, m2/s
DL is the diffusion coefcient of solute in liquid, m2/s.

Perry et al. [28] stated that the height equivalent to a theoretical Theoretical liquid holdup correlation of Billet and Schultes [38] is
plate (HETP) which applies to design dilute absorption columns is given below:
given below.
 1 =
1 2 3
ln hL 12 L uL a : 12
HETP HTU OV 4 g L
1
myx Billet and Schultes [38] theoretical holdup is based on the channel
L. 5 model which assumes the void volume of random packing by vertical
V ow channels. However, in real applications, the ow of phases deviates
due to the shape of random packing and the surface area of the packing
where, often partially covered by liquid. Therefore, theoretical liquid holdup
prediction deviates from the real column holdup. Billet and Schultes
HETP is the height equivalent to a theoretical plate, m [38] stated that the deviation of real holdup from theoretical holdup
is the stripping factor can be conveyed by considering hydraulic surface area of the packing.
myx is the slope of the equilibrium curve, kmol/kmol Accordingly, the real holdup of the column can be predicted by the fol-
L is the molar ow of the liquid, kmol/h lowing equation provided by Billet and Schultes [38].
V is the molar ow of the gas or vapor, kmol/h
 1 =  2
1 h =3
HTUOV is the overall height of a gas-phase mass transfer unit, m. 2 3 a
hLS 12 L uL a 13
g L a
The overall height of a gas-phase mass transfer unit (HTUOV) de-
pends on the height of a gas-phase mass transfer unit (HTUV) and height The hydraulic surface area (ah) equation described by Billet and
of a liquid-phase mass transfer unit (HTUL). The overall height of a gas- Schultes [38] is given as follows.
phase mass transfer unit (HTUOV) correlation provided by Billet and !0:1
 
Schultes [38] is given below. uL L ah uL L 0:15 uL 2 a
ReL b5 : Ch 14
uV myx uL a L a a L g
HTU OV HTU V HTU L 6
V aph L=V L aph !0:1
 
uL L a u 0:25 uL 2 a
ReL 5 : h C h 0:85 L L 15
 1=2   a L a a L g
1=6 01=2 DL aph
L aph C L 12 u L a 7
dh a where,

3=2       Ch is the hydraulic constant

1 a uV 3=4 V 1=3 aph hL is the column holdup, m3/m3
V aph C V D V 8
hL 1=2 d1=2
h
aV DV a hLS is the loading point column holdup, m3/m3 (loading point is
dened as the point from where the liquid holdup increases
0 uL as the gas velocity increases for the constant liquid loading).
uL 9
hL
!0:75 !0:45 The hydraulic constant (Ch) depends on packing material geometry
 0:2 2
aph 0:5 uL dh uL L dh u2L and must be determined empirically. Billet and Schultes [38] provided
1:5adh 10
a L L gdh hydraulic constants for various packings; however, it is not available

Table 3
dh 4 11
a Experimental liquid holdup for air-vegetable oil wet packed bed scrubbing system for the
given bed height, solvent temperature and solvent ow rate.
where,
Bed height Temperature Liquid holdup, ml
m C
Flow rate
uV is the velocity of gas or vapor with reference to free column
cross section, m/s 53 ml/min 63 ml/min 73 ml/min
uL is the velocity of liquid with reference to free column cross Mean S.E. a
Mean S.E. a
Mean S.E.a
section, m/s 0.5 30 145 21 168 18 190 14
Vaph is the gas phase volumetric mass transfer coefcient, 1/s 0.5 40 135 7 140 0 165 7
Laph is the liquid phase volumetric mass transfer coefcient, 1/s 0.5 50 110 14 120 14 135 7
CL is the packing specic constant for liquid phase 0.8 30 270 14 290 0 315 7
0.8 40 240 0 260 0 280 0
CV is the packing specic constant for gas or vapor phase 0.8 50 215 7 225 7 243 4
uL is the mean effective velocity of liquid, m/s 1.1 30 385 7 425 7 455 7
hL is the column liquid holdup, m3/m3 1.1 40 360 14 375 21 390 28
is the void fraction, m3/m3 1.1 50 330 0 345 7 360 14
a is the specied surface area of the dumped packing, m2/m3 a
Standard error.
 P.R. Bhoi et al. / Fuel Processing Technology 134 (2015) 243250
for the raschig rings of 6-mm size. Therefore, the hydraulic constant (Ch)
was determined using experimental data of the column holdup of this
study in which the bed height ranged from 0.5 to 1.1 m, solvent temper-
ature ranged from 30 to 50 C and solvent ow rate ranged from 53 to
73 ml/min (Table 3). The average value of Ch = 2.52, found experimen-
tally using liquid holdup data provided in Table 3, was used to predict
the liquid holdup in the subsequent calculation.
The prediction of the overall height of a gas-phase mass transfer unit
(HTUOV) requires empirical packing specic constants (CV and CL). Billet
and Schultes [38] provided these constants for various packings which
were validated for an air-water system. However, these constants are
not available for the raschig rings of 6 mm size used in this research.
For the estimation of CV and CL, an experimental HETP is required. The
experimental HETP of 0.11 m was determined using the optimum bed
height of 1.1 m reported in Bhoi et al. [18] and 10 theoretical stages as
determined in Section 2.1.2. Using the experimental HETP of 0.11 m of
this study, constants CV = 0.80 and CL = 2.40 were determined using
Billet and Schultes [38] mass transfer correlations.
3.3. Pressure drop across the column
Even though the packing material provides large interfacial area for
the mass transfer of model tar compounds, it increases pressure drop
across the column due to friction between uid streams and the packing
surface. A high mass transfer and a low pressure drop resulting in low
energy consumption are very important for the performance of wet
packed bed absorption column. For the given packings, solvent, and
air ow rate, the pressure drop across the column is a function of the
packed bed height, solvent temperature, and solvent ow rate. The
pressure drop across the column as a function of bed height and solvent
ow rate for two solvent temperatures (30 and 50 C) is shown in Fig. 7.
As shown, at the solvent temperature of 30 C, as the bed height in-
creased from 0.5 to 1.1 m, the pressure drop across the column in-
creased by over 100% from 7.3 to 16.1, 8.0 to 17.9 and 8.7 to 18.8 mm
of water column at the solvent ow rates of 53, 63, and 73 ml/min, re-
spectively. It was also observed that the pressure drop across the col-
umn signicantly increased as the solvent ow rate increased from 53
to 73 ml/min and reduced as the solvent temperature increased from
30 to 50 C. A statistical analysis showed that bed height, solvent tem-
perature and solvent ow rate had highly signicant (p b 0.0001) effects
on the pressure drop across the column [36]. The pressure drop across
the column increased as the packed bed height and solvent ow rate in-
creased and solvent temperature decreased due primarily to increased
liquid holdup. An increase in liquid holdup reduces the column area
leading to an increase in the gas and liquid velocities which enhances
20
Pressure Drop, mm of WC
15
10
5 0.5 m - 30C 0.5 m - 50C
0.8 m - 30C 0.8 m - 50C
1.1 m - 30C 1.1 m - 50C
0 From a Pilot Scale Fluidized Bed Gasier (Doctoral Dissertation)Retrieved from
50 55 60 65 70 75
Solvent Flow Rate, ml/min
Fig. 7. Effect of bed height and solvent ow rate on the pressure drop across the column at
solvent temperatures of 30 and 50 C and using soybean oil as a solvent.
249
the friction between uid streams and packing surfaces and thus the
pressure drop across the column increases.
4. Conclusions
A vegetable oil based bench scale wet packed bed scrubbing system
was successfully designed, fabricated and evaluated to study the remov-
al of model tar compounds. The system consists of a gas mixing system
designed to achieve a desired concentration of tar in an air stream and a
liquid distributor that was designed and fabricated to uniformly distrib-
ute the high viscous vegetable oils.
It was found that soybean and canola oils were not statistically
(p N 0.05) different in removal efciencies of benzene, toluene and eth-
ylbenzene. The packing specic constants CV = 0.80 and CL = 2.40 of
Billet and Schultes [38] correlation were found using experimental
data. These constants can be used to predict the height equivalent to a
theoretical plate (HETP) for laboratory or pilot scale vegetable oil
based wet packed bed scrubbing systems.
The pressure drop across the column varied in the range of 6.7 to
18.8 mm of water column which is well within the acceptable range
of pressure drop for a bench or laboratory scale wet packed bed
scrubbing system.
Acknowledgments
This study was funded, in part, through a grant from the National Sci-
ence Foundation EPSCoR Program No. EPS-0814361 and the Oklahoma
Agricultural Experiment Station. Authors also acknowledge vital contri-
butions of the Biosystems and Agricultural Engineering Department
staff members: Mr. Mark Gilstrap, Mr. Wayne R. Kiner, Mr. Jason Walker
and Mr. Mike Veldman and Dr. Mark E. Payton from the Department of
Statistics.
References
[1] DOE, Biomass Multi-year Program Plan Ofce of the Biomass Program, US DOE,
Washington, DC, 2012.
[2] M. Ruth, T. Mai, E. Newes, A. Aden, E. Warner, C. Uriarte, D. Inman, T. Simpkins, A.
Argo, Projected biomass utilization for fuels and power in a mature market,
Transportation Energy Futures Series, Prepared for the U.S. Department of Energy
by National Renewable Energy Laboratory, Golden, CO. DOE/GO-102013-3707,
2013 (153 pp.).
[3] P. Bhoi, R. Singh, A. Sharma, S. Patel, Performance evaluation of open core gasier on
multi-fuels, Biomass Bioenergy 30 (2006) 575579.
[4] DOE, Biomass Multi-year Program Plan Ofce of the Biomass Program, US DOE,
Washington, DC, 2010.
[5] S.K. Hoekman, Biofuels in the US challenges and opportunities, Renew. Energy 34
(2009) 1422.
[6] K. Patil, P. Bhoi, R. Huhnke, D. Bellmer, Biomass downdraft gasier with internal
cyclonic combustion chamber: design, construction, and experimental results,
Bioresour. Technol. 102 (2011) 62866290.
[7] T. Milne, N. Abatzoglou, R. Evans, Biomass Gasier Tars: Their Nature, Formation,
and Conversion (Nrel/Tp-570-25357), National Renewable Energy Laboratory,
Golden, Colorado, 1998.
[8] L. Devi, K.J. Ptasinski, F.J.J.G. Janssen, A review of the primary measures for tar elim-
ination in biomass gasication processes, Biomass Bioenergy 24 (2003) 125140.
[9] S. Anis, Z. Zainal, Tar reduction in biomass producer gas via mechanical, catalytic and
thermal methods: a review, Renew. Sust. Energ. Rev. 15 (2011) 23552377.
[10] P. Hasler, T. Nussbaumer, Gas cleaning for IC engine applications from xed bed
biomass gasication, Biomass Bioenergy 16 (1999) 385395.
[11] L.P. Rabou, R.W. Zwart, B.J. Vreugdenhil, L. Bos, Tar in biomass producer gas, the En-
ergy Research Centre of the Netherlands (ECN) experience: an enduring challenge,
Energy Fuel 23 (2009) 61896198.
[12] B. Pathak, D. Kapatel, P. Bhoi, A. Sharma, D. Vyas, Design and development of sand
bed lter for upgrading producer gas to IC engine quality fuel, Int. Energy J. 8
(2007) 1520.
[13] B.G. Cateni, Effects of Feed Composition and Gasication Parameters on Product Gas
ProQuest Dissertations and Theses (UMI Number: 3302329) Oklahoma State
University, Stillwater, OK, 2007.
[14] T. Phuphuakrat, T. Namioka, K. Yoshikawa, Absorptive removal of biomass tar using
water and oily materials, Bioresour. Technol. 102 (2011) 543549.
[15] A. Bhave, D. Vyas, J. Patel, A wet packed bed scrubber-based producer gas cooling
cleaning system, Renew. Energy 33 (2008) 17161720.
250 P.R. Bhoi et al. / Fuel Processing Technology 134 (2015) 243250
[16] T. Phuphuakrat, N. Nipattummakul, T. Namioka, S. Kerdsuwan, K. Yoshikawa,
Characterization of tar content in the syngas produced in a downdraft type xed
bed gasication system from dried sewage sludge, Fuel 89 (2010) 22782284.
[17] M. Dogru, A. Midilli, C.R. Howarth, Gasication of sewage sludge using a throated
downdraft gasier and uncertainty analysis, Fuel Process. Technol. 75 (2002) 5582.
[18] P.R. Bhoi, R.L. Huhnke, A. Kumar, M.E. Payton, K.N. Patil, J.R. Whiteley, Vegetable oil
as a solvent for removing producer gas tar compounds, Fuel Process. Technol. 133
(2015) 97104.
[19] E. Baker, M. Brown, R. Moore, L. Mudge, D. Elliott, Engineering Analysis of Biomass
Gasier Product Gas Cleaning Technology, Pacic Northwest Laboratory, Richland,
WA, USA, 1986.
[20] D. Khummongkol, C. Tangsathitkulchai, A model for tar-removal efciency from
biomass-produced gas impinging on a water surface, Energy 14 (1989) 113121.
[21] G. Joseph, D. Beachler, Scrubber systems operation review: self-instructional
manual, Prepared for the United States Environmental Protection Agency Under
Cooperative Assistance Agreement CT-902765 by North Carolina State University,
Raleigh, NC1998.
[22] P.R. Bhoi, R.L. Huhnke, S. Kumar, A. Kumar, M.E. Payton, Solubility enhancement of
producer gas tar compounds in water using sodium dodecyl sulfate as a surfactant,
Fuel Process. Technol. 133 (2015) 7579.
[23] H. Boerrigter, S. Van Paasen, P. Bergman, J. Knemann, R. Emmen, A. Wijnands,
OLGA Tar Removal Technology. ECN-C-05-009, Energy Research Centre of the
Netherlands, 2005.
[24] S. Pierucci, R. Del Rosso, D. Bombardi, A. Concu, G. Lugli, An innovative sustainable
process for VOCs recovery from spray paint booths, Energy 30 (2005) 13771386.
[25] B. Ozturk, D. Yilmaz, Absorptive removal of volatile organic compounds from ue
gas streams, Process. Saf. Environ. Prot. 84 (2006) 391398.
[26] USDA, http://www.ers.usda.gov/topics/crops/soybeans-oil-crops.aspx (Accessed
during September 2012).
[27] EIA, http://www.eia.gov/oiaf/aeo/otheranalysis/aeo_2006analysispapers/nlf.html
Accessed during September 2012.
[28] R.H. Perry, D.W. Green, J.O. Maloney, Perry's Chemical Engineers' Handbook, 8th ed
McGraw-Hill, New York, 2008.
[29] J. Seader, E.J. Henley, D. Keith, Separation Process Principles, 3rd ed John Wiley &
Sons, Inc., Hoboken, NJ, 2011.
[30] D. Ravindranath, B.J. Neely, R.L. Robinson, K.A.M. Gasem, QSPR generalization of ac-
tivity coefcient models for predicting vaporliquid equilibrium behavior, Fluid
Phase Equilib. 257 (2007) 5362.
[31] J. Eckert, Tower packings comparative performance, Chem. Eng. Prog. 59 (1963)
7682.
[32] J. Eckert, Selecting the proper distillation column packing, Chem. Eng. Prog. 66
(1970) 39.
[33] J. Eckert, How tower packings behave, Chem. Eng. 82 (1975) 7076.
[34] R.F. Strigle, Packed Tower Design and Applications: Random and Structured Pack-
ings, 2nd ed Gulf Publishing Company, Houston, TX, 1994.
[35] P.R. Bhoi, R.L. Huhnke, J.R. Whiteley, S. Gebreyohannes, A. Kumar, Equilibrium stage
based model of a vegetable oil based wet packed bed scrubbing system for remov-
ing producer gas tar compounds, Sep. Purif. Technol. 142 (2015) 196202.
[36] P.R. Bhoi, Wet Scrubbing of Biomass Producer Gas Tars Using Vegetable Oil (Order
No. 3629669), 2014. Available from Dissertations & Theses @ Oklahoma State
University - Stillwater; ProQuest Dissertations & Theses Full Text. (1562504711).
Retrieved from http://search.proquest.com/docview/1562504711?accountid=4117.
[37] H. Kister, Distillation Design, McGraw-Hill, New York, 1992.
[38] R. Billet, M. Schultes, Prediction of mass transfer columns with dumped and ar-
ranged packings, Trans IChemE 77 (1999) 498504.
[39] M. O'Meara, Determination of the Interfacial Tension Between OilSteam and OilAir
at Elevated Temperatures(Master's thesis). Retrieved from http://www.lib.ncsu.
edu/resolver/1840.16/8150 North Carolina State University, Raleigh, NC, USA, 2012.
[40] B. Esteban, J.R. Riba, G. Baquero, R. Puig, A. Rius, Characterization of the surface ten-
sion of vegetable oils to be used as fuel in diesel engines, Fuel 102 (2012) 231238.
[41] R. Barratt, The preparation of standard gas mixtures. A review, Analyst 106 (1981)
817849.
[42] A. Janzen, M. Schubert, F. Barthel, U. Hampel, E. Kenig, Investigation of dynamic
liquid distribution and hold-up in structured packings using ultrafast electron
beam X-ray tomography, Chem. Eng. Process.: Process Intesif. 66 (2013) 2026.