Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495

Contents lists available at ScienceDirect

Journal of the Taiwan Institute of Chemical Engineers

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

Removal of ammonia from wastewater by air stripping process in

laboratory and pilot scales using a rotating packed bed at ambient
temperature
Min-Hao Yuan a, Yi-Hung Chen a,∗, Jhih-Ying Tsai a, Ching-Yuan Chang b,c
a
Department of Chemical Engineering and Biotechnology, National Taipei University of Technology, Taipei, 106, Taiwan
b
Graduate Institute of Environmental Engineering, National Taiwan University, Taipei 106, Taiwan
c
Department of Chemical Engineering, National Taiwan University, Taipei 106, Taiwan

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

Article history: In this study, a continuous-flow rotating packed bed (RPB), functioning as an efficient gas–liquid contactor,
Received 27 July 2015 was employed for the ammonia stripping from wastewater in laboratory-scale and pilot-scale systems at
Revised 23 September 2015
ambient temperature. The effects of major operating variables, such as rotational speed (ω), liquid flow rate
Accepted 25 November 2015
(QL ), and gas flow rate (QG ) on the volumetric liquid mass-transfer coefficient (KL a) and stripping efficiency
Available online 15 January 2016
(η) were elucidated. The results show that the KL a values demonstrate the greatest increase with increasing
Keywords: gas flow rate (QG ), followed by liquid flow rate (QL ) and rotating speed (ω). Although changes in KL a would
Ammonia stripping be expected to directly reflect in the η values, the increased QL results in considerable compensation effects
Rotating packed bed leading to the decreased η, predominantly due to the decreased liquid hydraulic retention time. The dimen-
Continuous flow sionless models used in this study describe the relationships of KL a and η with the major parameters for
Miniaturization ammonia stripping in the RPB, and demonstrate good agreement with the experimental data. Moreover, in
Mass transfer
the continuous-flow pilot-scale RPB, an η of 95% was achieved at 4.6 min. while KL a values of approximately
0.017–0.027 1/s and height transfer unit (HTU) values of 2.2–4.8 cm were obtained at a QL of 5 L/min, QG of
1,500 L/min, and ω of 480–1000 rpm, suggesting that the RPB is a viable alternative technology for stripping
large loadings of ammonia from wastewater.
© 2015 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction in Taiwan recently implemented new two-stage wastewater restric-

Ammonium nitrogen (NH3 –N), which primarily comes from hu-
man, animal, certain fertilizers and industrial waste discharge, is an
important member of the nitrogen-containing pollutants in water en-
vironments [1]. NH3 –N acts as a nutrient for aquatic plants and algae,
and exerts an oxygen demand in receiving waters leading to a decline
in water quality, eutrophication, and toxicity to sensitive aquatic biota
[2]. Therefore, NH3 –N has been selected as one of the indicators for
assessing river quality, such as water quality and river pollution in-
dices [3]. Moreover, environmental regulations regarding NH3 –N in
municipal and industrial wastewaters that flow to nitrogen-sensitive
receiving waters are becoming steadily more stringent in every coun-
try. For example, in China, the 12th Five-Year Plan articulates a chal-
lenging task of a 10% discharge reduction in total NH3 –N from 2010 to
2015 [4]. Furthermore, the Environmental Protection Administration
∗
Corresponding author. Tel.: +886 2 27712171x2539; fax: +886 2 87724328.
E-mail address: yhchen1@ntut.edu.tw (Y.-H. Chen).
tions for NH3 –N that involves different maximum values and grace
periods for new and existing enterprises, petrochemical, semicon-
ductor and optoelectronic industries, and science parks [5]. There-
fore, the satisfaction of more stringent standards on NH3 –N discharge
requires significant improvement in ammonia removal technologies,
particularly for existing enterprises that have space and financial lim-
itations.
Ammonia can be removed from wastewaters through biological
and physicochemical treatments. Among them, the combination of
biological nitrification–denitrification processes is regarded as the
most common method available for the removal of ammonia from
wastewater [6–8]. While biological processes are energy efficient and
do not require large amounts of chemical additives, they are very sen-
sitive to shock, toxic loads, and cold weather conditions; they also
require relatively longer retention time and larger spatial require-
ments than other methods [7]. Another alternative method, break-
point chlorination, uses chlorine to oxidize NH3 –N (in the form of
ammonia) to nitrogen gas (N2 ). To reach the chlorine breakpoint, suf-
ficient chlorine must be added to satisfy all of the demand in the
wastewater, resulting in high costs and the formation of numerous

http://dx.doi.org/10.1016/j.jtice.2015.11.016
1876-1070/© 2015 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
 M.-H. Yuan et al. / Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495 489

Although tropical weather can facilitate the application of ammo-

Nomenclatures nia stripping in Southeast Asia and elsewhere [17], more intense
efforts to improve the design of the ammonia stripping process
aP specific area of packing per unit volume of a packed- worldwide are underway, e.g., jet loop reactor [18]. This is particu-
bed, m2 /m3 larly pertinent to many existing treatment plants, wherein the ca-
AC average cross section area of a packed bed, 2π ravg ZB , pacity of the plant cannot expand due to space limitations and
m2 economic constraints. Therefore, miniaturization of a stripping unit
CL,in liquid concentration of ammonia of inlet liquid, mg/L would be most desirable, especially for implementation by existing
CL,out liquid concentration of ammonia of outlet liquid, enterprises.
mg/L A rotating packed bed (RPB) has attracted increasing interest in
DL molecular liquid diffusion coefficient of ammonia, many processes such as absorption [19–22], distillation [23–24],
1.90 m2 /s at 30 °C stripping [25–27], ozonation [28], and esterification [29]. This is
dP stainless wire diameter of bed, 0.22 m because of the intensified mass transfer of the gas–liquid phase
g gravitational acceleration, m/s2 interface via high centrifugal force (300–10000 m/s), which is 1–3
HC dimensionless Henry’s law constant of ammonia, orders of magnitude greater than that of gravitational acceleration
8.582 × 10−4 at 30 °C [30]. Absorption of carbon dioxide (CO2 ) in RPBs has been studied
HTU height of a mass transfer unit for a packed bed, cm using a variety of absorbents such as sodium hydroxide (NaOH) [20],
HTUOL overall height of a mass transfer unit based on liquid- ethanolamines [22], and basic oxygen furnace (BOF) slag [31–32].
phase resistance, uC /KL a, cm Pan et al. [32] reported that 30% of CO2 in flue gas was removed by
KL a overall liquid volumetric mass transfer coefficient, 1/s approximately 96–99% using BOF slag within a short reaction time
ravg average radius of a packed bed, (ri + ro )/2, m of 1 min at 25 °C and 1 atm in an RPB reactor. Jassim et al. [22] also
ri inner radius of a packed bed, m showed that the conventional stripper height necessary to achieve
ro outer radius of a packed bed, m similar CO2 absorption performance to that obtained in an RPB is
RPB rotating packed bed greater by a factor of 8.4 and 11.3 in diameter, suggesting that a
S stripping factor, HC QG /QL significant reduction in equipment size and space can be achieved
tL liquid hydraulic retention time, min or h using a rotating bed. Furthermore, fouling or scaling problems caused
T stripping temperature, °C by carbonation have not been observed by those studies on the CO2
uL superficial liquid velocity at r, QL /AC , m/s adsorption. Some similar findings have been also reported by the air
VOCs volatile organic compounds stripping of VOCs from wastewater and groundwater [25,27]. Gudena
QG gas flow rate, L/min et al. [27] demonstrated that using an RPB results in enhanced
QG /QL ratio of gas and liquid flow rate, - trichloroethylene removal from groundwater at low gas/liquid ratios,
QL liquid flow rate, L/min which is preferred but difficult to obtain in conventional packed
VB volume of a packed bed, π (ro 2 – ri 2 )ZB , m3 beds. Moreover, the total capital cost of an RPB is greatly reduced,
ZB axial height of a packed bed, m being approximately 8% of the cost of a conventional stripper based
on their scenario as well as operating cost.
Greek symbols
In a RPB, gas flows through the packing channels contacting
μG gas viscosity, 1.87 kg/ms at 30 °C
with the liquid in a counter flow configuration. The characteristics
μL liquid viscosity, 0.798 kg/ms at 30 °C
of the packings, such as size, shape, material, and surface prop-
ρG density of gas, 1.166 kg/m3 at 30 °C
erty, would affect the mass transfer coefficient [33,34]. Stainless
ρL density of liquid, 995.7 kg/m3 at 30 °C
steel exhibits higher critical surface tension than glass, ceramic
νL dynamic liquid viscosity, 8.01 × 10–7 m2 /s3 at 30 °C
and acrylic, which facilitates the gas–liquid mass transfer in a RPB
ω rotational speed, rpm or rad/s (for Gravg calculation)
[33]. Further, unlike the conventional packing column using an ax-
π the circular ratio, 3.1415
ial fluid flow, main fluid flow path in a RPB is a radial flow. Hence,
Dimensionless groupings high-voidage packing of the wire mesh was often used to pro-
GrL,avg Grashof number of the liquid based on the average vide a low drop in pressure and reduce the tendency for fouling
bed radius, ravg ω2 (ravg – ri )3 /ν L 2 [34].
ReG Reynolds number of the gas, ρ G QG ln(ro /ri )/ It should be noted that these unique features of RPBs, such as
[2π ZB (ro – ri )ap μG ] high volumetric liquid mass-transfer coefficients, a reduced ten-
ReL Reynolds number of the liquid, ρ L QL ln(ro /ri )/ dency of fouling, and a reduction in equipment size, capital and
[2π ZB (ro – ri )ap μL ] operating costs, could potentially alleviate the limitations associ-
ated with ammonia stripping. However, information regarding the
performance of the ammonia stripping process using RPBs is still
disinfection byproducts [9]. As opposed to the aforementioned two unclear.
methods, ammonia stripping is a simple process of physical sep- In this study, the ammonia stripping performance of continuous-
aration that is generally utilized in wastewaters with high con- flow laboratory- and pilot-scale RPB systems at ambient temperature
tents of ammonia, e.g., pig slurry [10–11], cattle and fermented was examined. The effect of major operating variables, such as
chicken manures [12], landfill leachate [13], wastewaters from the rotational speed (ω), liquid flow rate (QL ), and the ratio of gas and
production of biogas [14] and petrochemicals [15]. The concen- liquid flow rate (QG /QL ) on the volumetric liquid mass-transfer
trated gaseous ammonia purged after stripping can be recovered coefficient (KL a) and the stripping efficiency (η), was investigated.
and absorbed by strong acidic solutions, such as sulfuric acid Regression equations are proposed to predict the KL a and η for the
(H2 SO4 ), forming mineral fertilizer for agricultural use [11]. The ma- operation of ammonia process. Furthermore, the feasibility of a
jor drawback of ammonia stripping is that it is practically inca- pilot-scale RPB in stripping ammonia from wastewater streams was
pable of operating at ambient air temperatures below 0 °C. Foul- elucidated and compared with that of a traditional air stripper. The
ing of the tower packing due to deposition of calcium carbon- findings from this study provide further insights into the feasibility
ate or other suspended solids is also often observed with un- of using an RPB for stripping ammonia from wastewater at ambient
stable, high-pH water flowing in-line through the tower [16–17]. temperature.
490 M.-H. Yuan et al. / Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495

Liquid in Gas out To hood

2 10
1
9
3 T
4 8
5 Gas in

D
Liquid out 6 7
Fig. 1. Experimental apparatus sketch of ammonia stripping in a laboratory-scale rotating packed bed (RPB). Components: 1. Ammonia-storage tank; 2. Pump; 3. Thermocouple;
4. RPB; 5. Packed bed; 6. Ion-selective electrode; 7. Motor; 8. Air flow meter; 9. Air compressor; 10. Neutralization tank.

Table 1
Specification and experimental condition of laboratory-scale and pilot-scale RPBs used in this study and a compassion with a traditional air-stripping
tower.

Items Units Laboratory-scale RPB Pilot-scale RPB Air-stripping tower [10]

Dimensions
Inner radius of a packed bed (ri ) m 0.0355 0.125 –
Outer radius of a packed bed (r0 ) m 0.0850 0.280 0.08
Average radius of a packed bed (ravg ) m 0.0603 0.203 –
Axial height of a packed bed (ZB ) m 0.0215 0.123 1.8
Volume of a packed bed (VB ) m3 3.7 × 10−4 0.02426 0.9217
Total area of a packing (ap VB ) m2 0.338 20.376 -a
Average cross section area of a packed bed (AC ) m2 0.0081 0.1565 0.9217
Specific area of packing per unit volume of a packed-bed (ap ) m2 /m3 840 840 –
Voidage (ε ) m3 /m3 0.954 0.954 –
Operating variables
Initial concentration of ammonia mg/L 1000 5630 2031
Temperature °C 30 28.4–30 18–21
pH – 11 11.5–12 11.7
Liquid flow rate (QL ) L/min 0.025–0.1 5–11.6 0.845
Gas flow rate (QG ) L/min 30–60 9000–18000 90
QG /QL ratio – 750–1800 1120–3000 106.5
Rotational speed rpm 300–1200 300–1000 –

a
The tower was packed with 250 plastic rings (properties and dimension were not reported).

2. Experimental setup
2.1. The laboratory-scale RPB system
Fig. 1 depicts the experimental setup of the laboratory-scale RPB
system employed in this study, which consists of an RPB, gas and liq-
uid feed controls, analysis of effluents and neutralization of effluent
gas. The specifications and experimental conditions of the RPB are
given in Table 1. The reactor volume (VB ) of the RPB was estimated to
be the volume of the packed-bed rotator, π (r0 2 – ri 2 )ZB , wherein ri ,
r0 , and ZB represent the inner radius, outer radius, and axial height of
the packed-bed rotator, respectively. A motor was connected to the
bed through a rotor shaft on two bearings mounted to a steel struc-
ture to produce the bed rotation and its rotating speed was varied
from 300 to 1200 rpm, resulting in a centrifugal force equivalent to
6.25 to 100 gravitational acceleration, respectively. The liquid stream
is sprayed by the liquid distributor at the center of the RPB while gas
stream flows countercurrently to the liquid in the RPB. Liquid dis-
tributor has six holes in a vertical group of two, and the groups are
spaced 180° apart. Grade 304 stainless steel wires cut in the shape
of annular rings were used for the packing, which was composed of
interconnected wires with a mean diameter of 0.22 mm.
Ammonia solution was prepared using ammonium chloride (99%,
Jin Yih Chemical Co., Ltd., Taipei, Taiwan) adjusted to pH 11 using
NaOH solution (45%, Jin Yih Chemical Co., Ltd., Taipei, Taiwan). A
diaphragm liquid pump (KNF 1.300 TT 18S, KNF, Sursee, Switzerland)
was used for the constant introduction of the ammonia solution into
the housing at a flow rate (QL ) of 0.025–0.1 L/min, whereas the flow
rate of air (QG ) was 30–60 L/min, controlled by an air compressor
(DT-175-2C, Tong Cheng Iron Works Co., Ltd., Taipei, Taiwan) and
an air flow meter (Dwyer Instruments Inc., Michigan City, IN, USA).
The ratio of gas and liquid flow rates (QG /QL ) was set between 750
and 1800. A hot plate and magnetic stirrer (C-MAG HS7 S1, IKA
Lab Equipment, Staufen, Germany), teflon heating tape (Kwo-Yi Co.,
Taipei, Taiwan), and type-K thermocouples (TES-1306, TES Electrical
Electronic Corp., Taipei, Taiwan) at the inlet of the ammonia solution
and the RPB housing were used to control the temperatures of
the solution and the RPB system at 30 °C. The concentrations of
ammonia in the inlet and outlet solutions were determined using
an ion-selective electrode (HI4101, Hanna Instruments, Woonsocket,
RI, USA). For the neutralization of the effluent gas, the stripped or
purged ammonia gas was introduced into a sulfuric acid solution
(Sigma-Aldrich, St. Louis, MO, USA) before entering the exhaust.
2.2. The pilot-scale RPB system
Similar to the laboratory-scale system, the pilot-scale experi-
mental system also comprised an RPB, gas and liquid feed controls,
 M.-H. Yuan et al. / Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495 491

analysis of effluents and neutralization of effluent gas with the a

specifications and experimental conditions provided in Table 1. 0.006
Accordingly, the geometric, dynamic and kinematic similarities
between the laboratory and pilot scale RPB are comparable in this
study. The scale ratio of the inner radius and radius of a packed 0.005
bed between laboratory-scale and pilot-scale systems was 3.3 to 3.5
and that of the axial height (5.7) was higher. Then, the VB and total
0.004

KLa (s-1)
packing area of the pilot-scale system were approximately 60 times
greater than that of the laboratory-scale system. The specific area
of packing per unit volume of a packed bed, so-called the surface-
0.003
to-volume ratio, was controlled at the same level, which provides
similar surface capacity for gas–liquid mass transfer. The dynamic
and kinematic similarities were usually estimated by dimensionless 0.002
parameters, such as Reynolds number of gas (ReG ) based on the gas
mass flux, gas density and gas viscosity, Reynolds number of liquid
(ReL ) based on the liquid mass flux, liquid density and liquid viscosity 0.001
and liquid Grashof number (GrL,avg ) based on the average bed radius, 300 600 900 1200
liquid viscosity and centrifugal acceleration. The ratio of both ReG b ω (rpm)
and ReL between for the laboratory-scale and pilot-scale RPB were 80
6–10 while that of GrL,avg were 60–80.
The treatment capacity of wastewater can be up to 200-fold
greater than that of the laboratory-scale RPB. Liquid distributor is
placed in a vertical group of four, and the groups are spaced 90° apart.
Note that the feed gas was controlled by a turbofan at the outlet of ex- 60
haust gas and the concentration of ammonia in the effluent gas after

η (%)
exiting the neutralization tower was below 1 ppm. The facility was
fabricated by SinoTech Engineering Ltd. (Taipei, Taiwan).
40
2.3. Determination of mass transfer coefficient and efficiency of
ammonia removal

In this study, the removal performance of ammonia stripping in

the RPBs was assessed using the overall volumetric liquid mass- 20
transfer coefficient (KL a) and the stripping efficiency (η). At each re- 300 600 900 1200
action conditions, more than 15 min of operation was allowed for the ω (rpm)
system to reach a steady state prior to assessment. The KL a for the
centrifugal vapor–liquid contactor was proposed by Singh [35], Singh Fig. 2. Variation of (a) KL a and (b) η for ammonia stripping with ω at QL of 0.05 L/min
and T of 30 °C. QG /QL = 600 (◦), 1200 (), 1600 () and 1800 (). Solid lines: Predictions
et al. [25] and Chen et al. [36–37] using the two-film theory and mass for KL a and η were based on Eqs. 8 and 4, respectively.
balance concepts. Note that the total resistance of liquid-to-gas mass
transfer consists of the resistances associated with the stagnant gas
and liquid interfacial films. Therefore, the resistances of both gas and ravg ω2 (ravg − ri )3
liquid films have been taken into consideration for the KL a calcula- GrL,avg = (5)
νL 2
tion. The value of η was computed from the concentrations of the
feed (CL,in ) and effluent (CL,out ) liquid streams. The applicable expres-  ro 
sions for computing KL a and η are as follows:
ρG QG ln ri
ReG = (6)
   2π ZB (ro − ri )aP μG
1 CL,in 1
ln 1− +
KL a =
QL S CL,out S  
VB 1− 1
(1) ρL QL ln rroi
S ReL = (7)
  2π ZB (ro − ri )aP μL
CL,out
η = 1− × 100% (2) where the aP denotes the specific area of packing per unit volume of
CL,in
a packed-bed. Other gas and liquid constants used in Eqs. 5–7 were
where S is stripping factor, defined as described in nomenclature section.
QG
S = HC (3) 3. Results and discussion
QL
HC is the dimensionless Henry constant for ammonia, calculated to
3.1. Ammonia stripping process in the laboratory-scale RPB
be 8.582 × 10−4 at 30 °C [38]. Then, Eq. 2 can be further modified as:
   Fig. 2 shows the dependence of KL a and η on the ω under oper-
1
1− ating at a QL of 0.05 L/min and T of 30 °C and QG /QL from 600 to
η= 1 −   S
 × 100% (4)
exp KL a VQBL 1 − 1
− 1 1800. As illustrated in Fig. 2a, KL a increased with higher ω in the
S S
range between 300 and 1200 rpm, indicating that the mass-transfer
To clarify the effect of operating variables on KL a and η, ω, QL and resistances were reduced with the increased ω. Theoretically, the am-
QG were also converted into dimensionless groups of GrL,avg , ReG and monia solution in the RPB stripper flows radially, countercurrent to
ReL , as follows: the air, and is split into very small droplets, threads, and thin films
492 M.-H. Yuan et al. / Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495

a b
0.007 0.007

0.006 0.006

0.005 0.005
KLa (s-1)

KLa (s-1)
0.004 0.004

0.003 0.003

0.002 0.002

0.001 0.001

0 0.025 0.05 0.075 0.1 0 0.025 0.05 0.075 0.1

c QL (L/min) d QL (L/min)
70 80

70
60

60
50
η (%)

η (%)
50
40
40

30
30

20 20
0 0.025 0.05 0.075 0.1 0 0.025 0.05 0.075 0.1
QL (L/min) QL (L/min)

Fig. 3. Variations of (a) KL a and (b) η for ammonia stripping with QL at QG /QL of 600 (solid symbol and dotted line) and 1200 (empty symbol and solid line) and of (c) KL a and (d)
η with QL at QG of 60 L/min (solid symbol and dotted line) and T of 30 °C. ω = 300 (◦ or •), 600 ( or ), 900 ( or ♦) and 1200 ( or ). Solid and dotted lines: Predictions for KL a
and η were based on Eqs. 8 and 4, respectively.

through the packing due to the action of the centrifugal force under
high ω, similar to the mechanism of VOC stripping [25]. Ammonia
then evaporates into the gas phase rapidly during this process at at-
mospheric pressure. The performance of ammonia stripping in terms
of η (Fig. 2b) indicates that the values of η were increased with the
increased ω, which is in agreement with the theory. In the mean time,
a higher centrifugal acceleration also reduces a thickness of liquid
film on the packing, resulting in the decrease in the amounts of liquid
holdup (VL ) and liquid hydraulic retention time (tL ) in the packed bed
[28]. Nevertheless, drop in VL and tL may influence the liquid–gas in-
terface and mass transfer differently. The lower liquid film and VL the
stronger is the enhancement of mass transfer. In contrary, the lower
tL may result in a compensation effect of rotating speed that reduces
the values of KL a and η in some circumstances [20]. Yet, the consider-
able compensation effects on KL a and η due to the drops in tL with the
increased ω were not significantly observed under our experimental
conditions.
Fig. 2 illustrates that the QG /QL had significant effects on KL a
and η. The results indicate that the value of KL a at a QG /QL of
1800 (0.00473 1/s) is 2.5 times higher than that at a QG /QL of 600
(0.00188 1/s), whereas the value of η at a QG /QL of 1800 (76.0%) is
also 2.1 times higher than that at a QG /QL of 600 (36.8%). A similar
trend was found at low ω of 300 rpm. The values of KL a were 0.00117
and 0.00259 1/s at QG /QL of 600 and 1800, respectively; likewise, the
values of η were 29.4% and 59.0% at a QG /QL of 600 and 1800, respec-
tively. The results reveal that the values of KL a and η increase with an
increase of QG /QL at the same level as that of ω.
It is noted that the values of ReG were 4.95 and 14.85 for QG /QL of
600 and 1800 at QL of 0.05 L/min, respectively. A higher QG /QL and
ReG represented that a higher QG was used at the same level of QL
and therefore a higher capacity to transfer the ammonia molecules
from the air–water interface to the gaseous phase was provided. Al-
though a higher QG /QL or QG also results in a smaller exposure time
of the liquid and gas currents, it is apparent that the introduction of
abundant air enhances the mass transfer of ammonia due to the fast
refresh rate of gas–liquid interface and consequently the short expo-
sure time required to attain the desired high efficiency in RPB.
The effects of QL on KL a and η as a function of ω under operation
at a QG /QL of 600 to 1200 and QG of 60 L/min are shown in Fig. 3.
As revealed in Figs. 3a and 3c, the experimental KL a increased with
higher QL at the same levels of QG /QL , QG and ω. At ω of 1200 rpm and
QG /QL of 1200, the values of KL a at a QL of 0.075 L/min (0.00492 1/s)
were 2.8 times higher than that at a QL of 0.025 L/min (0.00174 1/s).
Similarly, the considerable increasing trends on KL a were identified
by the increase of QL at a low QG /QL of 600 or the same QG .
When the higher QL enhanced the relative velocity among the
liquids and packing, the Reynolds number of the liquid (ReL ) were
simultaneously increased toward a higher degree of turbulence.
The numbers of ReL were increased from 0.082 to 0.247 as the QL
increased from 0.025 to 0.075 L/min at QG /QL of 1200. However,
 M.-H. Yuan et al. / Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495 493

a 0.16 < KL adP /DL aP < 0.85, 0.082 < ReL < 0.329, 4.95 < ReG < 14.85
and 1.80 × 109 < GrL, avg < 2.88 × 1010 . Based on the power of the di-
mensionless groups in Eq. 8, KL a is more sensitive to changes in ReG ,
0.006 than ReL and GrL,avg . Hence, the strong enhancement in mass trans-
fer by the supply of fresh air is clearly observed because of the fast
KLa, predicted (s-1)

refresh rate by the increased QG .

The values of η are determined by substituting the predicted val-
0.004 ues of KL a into Eq. 4. As shown in Figs. 4a and b, most of experimental
KL a and η lie within 20% of the estimated values except for two pre-
dicted KL a values (one is at ω of 1200 rpm, QL of 0.1 L/min and QG /QL
of 600, and another is at ω of 300 rpm, QL of 0.25 L/min and QG /QL of
0.002 1200) that were underestimated by more than 20%. According to the
two-film theory, the underestimated values of KL a were the result of
the overestimation of liquid-phase or gas-phase resistances. For com-
parison, the predicted curves are also plotted in Figs. 2 and 3. As can
0 be seem, the experimental data of KL a and η was generally in good
0 0.002 0.004 0.006
agreement with the results of empirical equations.
b KLa, experimental (s ) -1
80
3.3. Ammonia stripping process at the pilot-scale RPB

70
The stripping experiments in the pilot-scale RPB system were con-
ducted at ambient temperature and atmospheric pressure. The ex-
η , predicted (%)

60 perimental results of KL a and η as a function of ω under operation

50
40
30
20
20 30 40 50 60 70
η , experimental (%)
Fig. 4. Diagonal graph of experimental and predicted (a) KL a and (b) η. Solid lines: Pre-
dictions for KL a and η were based on Eqs. 8 and 4, respectively. Dotted lines: Prediction
values within ±20%.
Figs. 3b and 3d exhibits slightly decreasing trends of η with the
increase in the QL at the same levels of QG /QL , QG and ω, which is
in contrast to the tendency of KL a on QL (Fig. 3a and c). It is often
seen that changes in KL a with ω and QG /QL are directly reflected in
the value of the separation efficiency. Yet, these results indicate that
the increased QL leads to considerable compensation effects due to
the decreased tL and consequently reduced the performance of η
for ammonia stripping. Note that the tL was decreased from 14.2 to
4.8 min when QL increased from 0.025 to 0.075 L/min, respectively.
3.2. Predicting KL a and η for the laboratory-scale RPB
The empirical dimensionless equation of ReL , ReG and GrL,avg has
been developed to predict KL a for the RPB based on Liu et al. [26],
Chen and Liu [19], and Chen et al. [28] for ethanol stripping and VOC
and ozone absorption, respectively. The predicted equation for KL a is
obtained by a linear regression of the natural logarithm of the exper-
imental data in Figs. 2 and 3 (R2 = 0.8948), as follows:
KL adp
= 0.024 ReL0.392 Re0.531
G
0.195
GrL,avg
DL ap
where dP and ap are the stainless-steel wire diameters of the
packing (0.22 m) and specific area of packing per unit vol-
ume of the packed-bed (840 m2 /m3 ), respectively. The value
of DL is the molecular liquid diffusion coefficient of ammonia.
The ranges of the dimensionless groups in this correlation are
at QL of 5 L/min and QG /QL of 3000 for synthesized wastewater with
a high content of ammonia (5630 mg/L) are shown in Figs. 5a and
5b. As expected, KL a and η increased with increasing ω. Note that the
rotating packed bed was operated between 300 rpm and 1000 rpm
giving a gravitational acceleration of 22 and 240 g in the pilot-scale
system, respectively. The maximum value of KL a in the pilot-scale RPB
(0.0273 1/s) was 4.4-fold higher than that obtained in the laboratory-
scale RPB (0.0062 1/s). Furthermore, compared to the results from
the laboratory-scale RPB, clear enhancements in terms of stripping
80 efficiency were observed, which is most likely due to a higher QG /QL
and a slightly higher pH. The value of η obtained was as high as 99.5%
at ω of 1000 rpm, while the concentration of ammonia in the effluent
wastewater was less than 30 mg/L. Furthermore, gaseous ammonia
was almost completely neutralized by the H2 SO4 solution, being be-
low 1 ppm in the effluent gas. Note that it would be economically
suitable for agricultural use when the concentration of (NH4 )2 SO4
solution is above 20 wt.%. Figs. 5c and 5d show the effect of QG on
KLa and η, respectively, at QL of 5 L/min and ω of 1000 rpm. Appar-
ently, the increasing values of KLa and η were obtained with a higher
QG into the RPB. However, an optimum compromise between energy
consumption, cost (ω, QG and QL ), pH and efficiency must be reached
before industrial-scale applications.
The prediction curves of KL a and η are also illustrated in Fig. 5. The
correlation equation of KL a is obtained with R2 = 0.949, as follows:
KL adp
= 6.38 × 10−6 Re0.392
L Re0.532
G
0.387
GrL,avg (9)
DL ap
It should be noted that the power of ReL (0.392) is adopted
from Eq. 8 due to a lack of data at different QL . Similar to the Eq. 8,
KL a is more sensitive to changes in ReG , than ReL and GrL,avg . Hence,
the model employed adequately describes the data using the di-
mensionless groups of ReL , ReG and GrL,avg . However, it seems like
the estimated results of KL a display a linear relationship with ω of
300–1200 rpm (Fig. 5a) or QG of 9–12 m3 /min (Fig. 5c). Based on
the predicted KL a, the estimated η displays a non-linear and good fit
with the experimental results (Fig. 5b and d).
(8)
3.4. Comparison with traditional stripping process
According to experimental data by Liao et al. [10] for an air-
stripping tower for swine manure wastewater containing 3021 mg-
N/L ammonia, a 90% separation efficiency is achieved after 7 h of
494 M.-H. Yuan et al. / Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495

a c
0.03 0.03

0.025
0.025
KLa (s-1)

KLa (s-1)
0.02

0.02
0.015

0.01 0.015
300 600 900 1200 9 12 15
b ω (rpm) d QG (m3/min)
100 100

95
η (%)

η (%)
95

90

85 90
300 600 900 1200 9 12 15

ω (rpm) QG (m3/min)

Fig. 5. Variation of (a) KL a and (b) η for ammonia stripping with ω at QG /QL of 3000 and of (c) KL a and (d) η with QG at ω = 1000 rpm. QL of 5 L/min and T of 30 °C in the pilot-scale
RPB system. Solid lines: Predictions for KL a and η were based on Eqs. 9 and 4, respectively.

recirculation at an ambient temperature of 18–21 °C. A comparison 4. Conclusions

of specifications and experimental conditions for these two meth-
ods is shown in Table 1. As can be seen, the volume and height of
these RPBs were 38 and 14.6 times smaller, respectively, than those
of the stripping tower. Moreover, the QL of 5 L/min in the continuous-
flow RPB is also greater than that of 0.845 L/min in the recirculation
stripping tower, whereas the tL was only 4.6 min in the pilot-scale
RPB.
In designing a stripper, height of transfer units (HTU) is often used
to ascertain the degree of separation difficulty for comparing the per-
formance of a packed bed. Therefore, the overall HTU based on liquid-
phase resistance (HTUOL ) in the pilot-scale RPB is also calculated, as
determined by KL a [21], as follows:
uL
HTUOL =
KL a
where uL is the liquid superficial velocity. The obtained results in-
dicate that the corresponding HTUOL values for ammonia stripping
were 2.2–4.8 cm with η > 95% depending on the QL , QG and ω. The
HTUOL values in the RPB were 2–3 orders lower than those in tradi-
tional packed beds. It is clear that the RPB system provides a minia-
turization of the stripping packed bed and greatly reduces the tL to
reach the desirable treatment efficiency under a larger wastewater
load.
The removal of ammonia from wastewater has been successfully
performed using a continuous-flow rotating packed bed (RPB) as an
air stripper in laboratory and pilot scales at ambient temperatures.
Experimental results indicated that an RPB boosted the overall vol-
umetric liquid mass transfer coefficient (KL a) and the stripping ef-
ficiency (η) for ammonia at small height of transfer unit based on
liquid-phase resistance (HTUOL ) and short liquid hydraulic retention
times (tL ). The KL a values largely increase with an increasing gas flow
rate (QG ), followed by liquid flow rate (QL ) and rotating speed (ω). Al-
though changes in KL a would be directly reflected in the η values,
the increased QL results in considerable compensation effects that
mainly decrease the η, predominantly due to the associated decrease
(10) in the tL . The empirical model comprising dimensionless parameters
of Reynolds number of gas and liquid streams, and Grashof number
of liquid stream, was employed and adequately described the data
from the laboratory and pilot scale RPB. Furthermore, KL a values of
0.017–0.027 1/s and HTUOL values of 2.2–4.8 cm can be obtained in
the pilot-scale RPB at QL of 5 L/min, QG of 1500 L/min and ω of 480–
1000 rpm, which account for the η greater than 95% at a tL of 4.6 min.
Compared to traditional air strippers, the continuous-flow RPB ap-
pears to be a very efficient mechanism for ammonia stripping in
terms of achieving the miniaturization of a stripping packed bed and
 M.-H. Yuan et al. / Journal of the Taiwan Institute of Chemical Engineers 60 (2016) 488–495
significant reduction in the time required to achieve the satisfactory
treatment efficiency for wastewaters with high ammonia loadings.
Acknowledgments
This work was funded by the Ministry of Science and Technology
of Taiwan (102-2622-E-027-024-CC3) and by SinoTech Engineering
Ltd., Taipei, Taiwan, who fabricated the pilot-scale RPB system. The
authors are grateful for these contributions.
References
[1] Rabalais NN. Nitrogen in aquatic ecosystems. Ambio 2002;31:102–12.
[2] Camargo JA, Alonso A, Salamanca A. Nitrate toxicity to aquatic animals: a review
with new data for freshwater invertebrates. Chemosphere 2005;58:1255–67.
[3] Liou SM, Lo SL, Wang SH. A generalized water quality index for Taiwan. Environ
Monit Assess 2004;96:35–52.
[4] China Ministry of Environmental Protection (MEP). MEP Explains the National
12th Five-Year Plan for Environmental Protection, PR China; http://english.
mep.gov.cn/News_service/infocus/201202/t20120207_223194.htm. (accessed Jan
2015).
[5] (EPA) TEPA. Effluent standards for chemical industry announced. Environ Policy
Mon 2014;17:6–7.
[6] Munch EV. Simultaneous nitrification and denitrification in bench-scale sequenc-
ing batch reactors. Water Res 1996;30:277–84.
[7] Zhao HW, Mavinic DS, Oldham WK, Koch FA. Controlling factors for simultane-
ous nitrification and denitrification in a two-stage intermittent aeration process
treating domestic sewage. Water Res 1999;33:961–70.
[8] Ruiz G, Jeison D, Chamy R. Nitrification with high nitrite accumulation for
the treatment of wastewater with high ammonia concentration. Water Res
2003;37:1371–7.
[9] Jafvert CT, Valentine RL. Reaction scheme for the chlorination of ammoniacal wa-
ter. Environ Sci Technol 1992;26:577–86.
[10] Liao PH, Chen A, Lo KV. Removal of nitrogen from swine manure wastewaters by
ammonia stripping. Bioresour Technol 1995;54:17–20.
[11] Laureni M, Palatsi J, Llovera M, Bonmati A. Influence of pig slurry characteris-
tics on ammonia stripping efficiencies and quality of the recovered ammonium-
sulfate solution. J Chem Technol Biot 2013;88:1654–62.
[12] Zeng L, Mangan C, Li X. Ammonia recovery from anaerobically digested cattle ma-
nure by steam stripping. Water Sci Technol 2006;54:137–45.
[13] Cheung KC, Chu LM, Wong MH. Ammonia stripping as a pretreatment for landfill
leachate. Water Air Soil Pollut 1997;94:209–21.
[14] Lei X, Sugiura N, Feng C, Maekawa T. Pretreatment of anaerobic digestion effluent
with ammonia stripping and biogas purification. J Hazard Mater 2007;145:391–7.
[15] Chang SY, Tu MY, Hsiao AF, Tsao LK, Huang TL. A real case of ammonia and nitrate
removal improvement in refinery wastewater treatment. J Petrol 2013;49:91–110
In Chinese.
[16] Kutzer S, Wintrich H, Mersmann A. Air stripping – A method for treatment of
wastewater. Chem Eng Technol 1995;18:149–55.
495
[17] Liu B, Giannis A, Zhang J, Chang VWC, Wang JY. Air stripping process for ammo-
nia recovery from source-separated urine: Modeling and optimization. J Chem
Technol Biot 2015;90:2208–17.
[18] Deĝermenci N, Ata ON, Yildiz E. Ammonia removal by air stripping in a semi-
batch jet loop reactor. J Ind Eng Chem 2012;18:399–404.
[19] Chen YS, Liu HS. Absorption of VOCs in a rotating packed bed. Ind Eng Chem Res
2002;41:1583–8.
[20] Lin CC, Liu WT, Tan CS. Removal of carbon dioxide by absorption in a rotating
packed bed. Ind Eng Chem Res 2003;42:2381–6.
[21] Lin CC, Wei TY, Hsu SK, Liu WT. Performance of a pilot-scale cross-flow rotat-
ing packed bed in removing VOCs from waste gas streams. Sep Purif Technol
2006;52:274–9.
[22] Jassim MS, Rochelle G, Eimer D, Ramshaw C. Carbon dioxide absorption and des-
orption in aqueous monoethanolamine solutions in a rotating packed bed. Ind
Eng Chem Res 2007;46:2823–33.
[23] Kelleher Y, Fair JR. Distillation studies in a high-gravity contactor. Ind Eng Chem
Res 1996;12:4646–55.
[24] Lin CC, Ho TJ, Liu WT, Liu HS. Distillation in a rotating packed bed. J Chem Eng Jpn
2002;35:1298–304.
[25] Singh SP, Wilson JH, Counce RM, Villiers-Fisher JF, Jennings HL, Lucero AJ, et al.
Removal of volatile organic compounds from groundwater using a rotary air strip-
per. Ind Eng Chem Res 1992;31:574–80.
[26] Liu HS, Lin CC, Wu SC, Hsu HW. Characteristics of a characteristics of a rotating
packed bed. Ind Eng Chem Res 1996;35:3590–6.
[27] Gudena K, Rangaiah GP, Lakshminarayanan S. Optimal design of a rotating
packed bed for VOC stripping from contaminated groundwater. Ind Eng Chem
Res 2012;51:835–47.
[28] Chen YH, Chang CY, Su WL, Chen CC, Chiu CY, Yu YH, et al. Modeling ozone con-
tacting process in a rotating packed bed. Ind Eng Chem Res 2004;43:228–36.
[29] Chen YH, Wang LC, Tsai CH, Shang NC. Continuous-flow esterification of free fatty
acids in a rotating packed bed. Ind Eng Chem Res 2010;49:4117–22.
[30] Rao DP, Bhowal A, Goswami PS. Process intensification in rotating packed beds
(HIGEE): an appraisal. Ind Eng Chem Res 2004;43:1150–62.
[31] Chang EE, Pan SY, Chen YH, Tan CS, Chiang PC. Accelerated carbonation of steel-
making slags in a high-gravity rotating packed bed. J Hazard Mater 2012;227–
228:97–106.
[32] Pan SY, Chiang PC, Chen YH, Tan CS, Chang EE. Ex situ CO2 capture by carbonation
of steelmaking slag coupled with metalworking wastewater in a rotating packed
bed. Environ Sci Technol 2013;47:3308–15.
[33] Lin CC, Wei TY, Liu WT, Shen KP. Removal of VOCs from gaseous streams in a
high-voidage rotating packed bed. J Chem Eng Jpn 2004;37:1471–7.
[34] Chen YS, Lin FY, Lin CC, Tai CYD, Liu HS. Packing characteristics for mass transfer
in a rotating packed bed. Ind Eng Chem Res 2006;45:6846–53.
[35] Singh SP. Air stripping of volatile organic compounds from groundwater: an eval-
uation of a centrifugal vapor-liquid contactor, Knoxville, TN: University of Ten-
nessee; 1989. Ph.D. Dissertation.
[36] Chen YS, Lin CC, Liu HS. Mass transfer in a rotating packed bed with viscous New-
tonian and non-Newtonian fluids. Ind Eng Chem Res 2005;44:1043–51.
[37] Chen YS, Lin CC, Liu HS. Mass transfer in a rotating packed bed with various radii
of the bed. Ind Eng Chem Res 2005;44:7868–75.
[38] National Institute of Standards and Technology (NIST), Ammonia, http://
webbook.nist.gov/cgi/cbook.cgi?Name=ammonia&Units=SI. (accessed Jan 2015).