DSpace at VNU: Experimental and modeling study on room-temperature removal of hydrogen sulfide using a low-cost extruded Fe2O3-based adsorbent

DSpace at VNU: Experimental and modeling study on room-temperature removal of hydrogen sulfide using a low-cost extruded...

Experimental and modeling study on room-temperature removal

of hydrogen sulfide using a low-cost extruded Fe2O3-based

adsorbent

Nguyen Quang Long1 • Tran Xuan Loc1

Received: 5 February 2016 / Revised: 26 February 2016 / Accepted: 11 March 2016

Ó Springer Science+Business Media New York 2016

Abstract In order to prevent corrosion and catalyst

deactivation, commercial adsorption processes for

hydro-gen sulfide (H2S) removal operate at relatively high

tem-perature (350–450°C) by ZnO adsorption This paper

reports firstly data from experimental and modeling studies

of the dynamic performance of a low-cost extruded Fe2O3

-based adsorbent for H2S removal at room temperature The

H2S adsorbent containing iron (III) oxide (Fe2O3) and

bentonite was prepared by hydrothermal-precipitation

method and the material has been characterized by several

techniques including scanning electron microscopy (SEM),

X-ray diffraction (XRD), and low temperature N2physical

adsorption The Fe2O3-based extruded adsorbent was used

in a continuous fixed-bed column experiments to evaluate

the efficiency for removal of H2S under the effect of various

process parameters including the bed depth, the flow rate

and the initial H2S concentration The results showed that

the total H2S uptake slightly increased with increasing the

bed depth and the initial H2S concentration, and decreased

with increasing the flow rate In the modeling part, the

dynamics of the adsorption process was modeled by two

adsorption models namely, Thomas model and BDST (bed

depth service time) model The kinetic parameters obtained

from the models were used to predict the breakthrough time

for a larger column which contained the adsorbent about 40

times more than that in the mini column experiments The

Thomas model was the suitable model for prediction of

10 % C0breakthrough time with an error about 4 %

Graphical Abstract

0 0.1 0.2 0.3

Time (minutes)

Experimental data Thomas model BDST model

Keywords H2S removal  Fe2O3 Extruded adsorbent  Modeling  Breakthrough time

List of symbols

Ct(ppm, mg/L) Effluent H2S concentration

C0(ppm, mg/L) Feed H2S concentration

Cb(ppm, mg/L) Breakthrough H2S concentration

F (L min-1) Volumetric flow rate

kTh (L mg-1 min-1)

Thomas rate constant (Thomas model)

K (L mg-1 min-1)

Adsorption rate constant (BDST model)

N0 (mg L-1, mg g-1)

Adsorption capacity (BDST model)

& Nguyen Quang Long

nqlong@hcmut.edu.vn

1 Faculty of Chemical Engineering, Ho Chi Minh City

University of Technology, 268 – Ly Thuong Kiet St.,

Dist 10, Ho Chi Minh City, Vietnam

DOI 10.1007/s10450-016-9790-0

t0.1, t0.5 Breakthrough time corresponding to

Ct= 10 % C0, Ct= 50 % C0

q0(mg g-1) Maximum adsorption capacity

(Thomas model)

1 Introduction

The removal of toxic components from gaseous streams is

currently one of the most important environmental issues

being researched (Karmakar et al 2015; Nor et al 2013;

Nguyen et al 2008, 2010) Biogas, a potential sustainable

energy fuel and feed-stock for chemical productions, is a

product of anaerobic degradation of organic substrates The

conversion of the chemical energy contained in biogas, which

is rich in CH4, into electricity is possible through combustion

in internal combustion engines (Abatzoglou and Boivin

2009) These units can be seriously damaged by the hydrogen

sulfide (H2S), because it can cause corrosion Moreover,

hydrogen sulfide is a well-known poison for metallic

cata-lysts Thus, H2S concentration in feedstocks should be

decreased to parts per million levels before their use, for

example in fuel cell application (Liyu and King 2006)

Therefore, removal of H2S is important for protection of

pipe-line systems as well as catalysts in chemical processes

Desulfurization of gaseous stream can be done on various

adsorbents depending on the temperature of the feed gas

Most gas phase desulfurization units employ metal oxides,

such as zinc oxide (Liyu and King2006; Novochinskii et al

2004; Sasaoka et al.1994,2000), iron oxide (Xie et al.2010;

Wang et al.2011; Ren et al.2010; Najjar and Jung1995),

copper oxide (Abbasian and Slimane 1998; Slimane and

Abbasian2000), and manganese oxide (Zeng et al 2015;

Cheah et al.2011) as active sorbents Among them, while

ZnO is a well-known and commercial adsorbent used to

capture H2S from fuel gas streams in a moderate temperature

range (300–500°C), Fe2O3is a potential candidate for

low-temperature H2S removal It is because of the fact that iron

oxide exhibits very favorable thermodynamics in reaction

with H2S at low temperature The mechanism of H2S

adsorption on Fe2O3-based adsorbent at low temperature

consists of several steps (Eqs.1 3) (Davydov et al.1998)

Fe2O3þ 2H2S! Fe 3þ .H2S

Fe3þ .H2S

2O3! H2O Fe 3þ .S

2ðOHÞ2ðfastÞ ð2Þ

H2O Fe 3þ S

2ðOHÞ2!H2O

þ HO Fe 2þ S Fe2þ OH

Moreover, the used Fe2O3-based adsorbent can be

regenerated by reaction with oxygen according to the

fol-lowing equations (Davydov et al.1998; Wie¸ckowska1998)

Fe2S3þ 3=2O2þ 3H2O! 2Fe OHð Þ3þ3S ð5Þ 2FeSþ 3=2O2þ 3H2O! 2Fe OHð Þ3þ2S ð6Þ Although many studies on H2S adsorption by Fe2O3 -based adsorbent have been published, these researches mainly focused on (1) high temperature adsorption (Xie

et al 2010; Wang et al.2011; Ren et al.2010; Najjar and Jung 1995) and (2) adsorbent in the form of powder/pel-letized particles (Davydov et al 1998; Arcibar-Orozco

et al 2015) For practical applications in fixed-bed col-umns, powder adsorbents must be formed in granules, spheres, or extrudes in order to reduce the pressure drop

Fe2O3supporting on montmorillonite in granule form was reported (Nguyen-Thanh et al 2005) However, H2S capacity of granuled Fe2O3/montmorillonite were only (0.57–9.65) mgS/g A study on Fe2O3-based extruded adsorbent for H2S removal at room temperature has not been reported in literature as our knowledge

Additionally, the fixed-bed adsorber does not run under equilibrium conditions, so the flow conditions and mass-transfer aspects throughout the column have to be consid-ered (Serna-Guerrero and Sayari 2010) The dynamic behavior of a fixed-bed column is described in terms of the effluent concentration–time profile (the breakthrough curve) which is essential in the evaluation of the efficiency

of an adsorber The time between the absorbent start-up and the appearance of the maximum concentration of impurity

in the bed outlet determines the absorbent breakthrough (or service) time Correct prediction of the breakthrough time is critical for designing the fixed-bed adsorber

This paper reports firstly the dynamic performance for room temperature H2S removal of an low-cost extruded

Fe2O3-based adsorbent which prepared by a hydrothermal precipitation method The effect of reaction conditions and modeling approaches of the dynamic behavior of the adsor-bent have been investigated The data given in this paper provide a basis for a dynamic adsorption model that could be used to design and evaluate an adsorption column using the low-cost extruded Fe2O3- based adsorbent on a larger scale

2 Materials and methods 2.1 Sorbent preparation and characterization The Fe2O3-based adsorbent was prepared from iron (III) nitrate nonahydrate (Fe(NO3)3.9H2O, 99 %), sodium hydroxide (NaOH, 99.5 %), bentonite (Binh-Thuan Ben-tonite, Vietnam), starch and de-ionized water Firstly, 75 g Fe(NO3)3.9H2O and 10 g bentonite were dissolved in

300 mL distilled water under magnetic stirring, and were

continuously stirred at room temperature for minutes to

obtain a homogeneous solution Then, NaOH solution was

added drop-wise into the above solution until the pH was

adjusted to 7 and a reddish-brown precipitate quickly

formed The solution was stirred for 30 min before being

transferred to an autoclave for hydrothermal treatment at

temperature 120°C for 12 h The produced precipitations

were filtered and washed by de-ionized water, after being

dried at 100°C for 8 h The collected solid was pulverized

and mixed with the starch, which was 5 % weigh of solid into

a mixer for 1 h with a speed of 200 rpm Afterwards they

were extruded to particles with the same size (2-mm

diam-eter and 5-mm length) Finally, the pellets were calcined at

500°C for 4 h in a static oven The final Fe2O3-based

adsorbent was characterized by several bulk and surface

analysis techniques Scanning electron microscopy (SEM)

analysis was carried out by using Keyence VE8800

appa-ratus was for characterization of the structure properties of

the synthesized material with different magnifications at an

electron acceleration voltage of 20 kV in a high vacuum

condition The crystallization phase analysis was executed

by X-ray diffraction (XRD) technique using Rigaku

Multi-flex diffractometer operating at 40 kV, 20 mA, and CuKa

monochromatic radiation (k = 1.54 A˚ ) N2 adsorption at

low temperature (77 K) was used for specific surface area of

the material (BET method) using Autosorb-1

(Quanta-Chrome) equipped with an analysis software

2.2 Hydrogen sulfide removal test

The adsorption test system for H2S is described in Fig.1

The system consists of gas cylinders for production of H2S/

N2 model mixture The adsorbent was packed in a Pyrex

U-tube reactor (ID = 0.8 cm) which located in a

temper-ature controllable furnace H2S concentrations were

measured continuously by a H2S sensor system purchased from Alphasense, England and be calibrated by Gastec

H2S-test kit (Japan) Prior to the adsorption test, the sample was dried at 100°C overnight The effects of various process parameters: (1) bed depth (h = 3–6 cm), (2) inlet

H2S concentration (C0= 600–1200 ppm), and (3) flow rate (F = 3.3–8.3 9 10-2L/min) were investigated to evaluate the performance of breakthrough on H2S adsorp-tion by Fe2O3and the breakthrough of H2S was 10 % of the inlet H2S concentration All the experiments were carried out at room temperature (30 ± 2°C), atmospheric pressure The H2S adsorption capacity of the material was calculated by the following equation:

CS¼ 103 MS

madsorbent

F:P R:Tr

t b 0

CS in CS out

where Cs—H2S adsorption capacity (mgS/g); Ms —molac-ular weight of S (=32); msorbent—mass of the adsorbent;

CSin—H2S concentration of the input stream; CSout—H2S concentration of the output stream; F—total gas flow rate;

tb—time of the adsorption until the concentration of the output stream higher than the breakthrough point (break-through time)

2.3 Modeling 2.3.1 Mathematical analysis Detailed information on the composition, structure, physi-cal and chemiphysi-cal properties of the stationary phase is important for appropriate understanding and description of surface diffusion The surface diffusion therefore is com-plicated by a complex nature of interactions between the adsorbate and the surface adsorbent as well as by the complexity of the surface The mass transfer coefficient

Fig 1 Hydrogen sulfide adsorption test system

(km) was calculated by the Ranz-Marshall correlation

(Gutie´rrez et al.2014)

where:

Sh¼kmdp

Dm

; Re¼ udpqg

1 e

ð Þlg; Sc¼

lg

qgDm

If Re [ 100, the flow rate of gas through the porous

media is turbulent flow

Dm¼

103T1:75 1

Mgasþ 1

MH2S

P V gas1=3þ VH1=32S2 ð9Þ

where Dm, dp, qg, lg, e, M and V are the molecular

dif-fusivity, particle diameter, gas density, gas viscosity and

bed void fraction, molecular weight, the diffusion volume,

respectively

The loading behavior of H2S to be removed from biogas

containing the Fe2O3media in a fixed-bed column is shown

by breakthrough curves The shape of the concentration–

time profile of breakthrough curves is the important

char-acteristics for determining the operation and the dynamic

response of an adsorption column A typical breakthrough

curve is illustrated in Fig.2a When the volume of the

mixture gas begins to flow through the column, the mass

transfer zone varies from 0 % of the inlet concentration

(corresponding to the free adsorbent) to 100 % of the inlet

concentration (corresponding to the total saturation)

(Calero et al.2009; Taty-Costodes et al.2005)

The breakthrough plot is usually expressed in terms of

adsorbed H2S concentration (Cads(ppm) = inlet H2S

con-centration (C0)—effluent H2S concentration (Ct)) Thus,

the total adsorbed H2S quantity (maximum column

capacity) (qtotal, mg) in the column for a given feed

con-centration and flow rate can be estimated as follows

(Kundu and Gupta 2007; Thomas 1944; Bharathi and Ramesh 2013):

qtotal¼ F

Zt total 0

Cadsdt¼ F

Zt total 0

C0 Ct

where F is the volumetric flow rate of the mixture gas (L min-1), ttotal is the total time of flow (min) and the volume of the effluent (Veff) can be calculated from the following equation:

The total amount of H2S fed to the column (X, mg) is as follows:

The total percent removal of H2S by the column can be calculated from the following equation:

total H2S removal %ð Þ ¼ qtotal

mtotal

2.3.2 Modeling of breakthrough curves Prediction of the breakthrough curve for the effluent is the predominant factor for the prosperous design of a column adsorption process However, the process does not operate

in a steady state as the concentration in gas phase changes through the time and space as the feed moves through a fixed-bed, so it is quite difficult to develop a model which accurately describes the dynamic behavior of adsorption in

a fixed-bed system Therefore, this study is starting from the mass balance between the solid and gas The variation

of this balance during the reaction can be illustrated

by Fig.2b The equation of mass balance material can

be stated as: input flow = output flow ? flow inside pore ? matter adsorbed onto the bed (Taty-Costodes et al

1.0

C t /C 0

0.1

C 0

C t = C 0

C 0 C 0

Mass transfer zone

C 0

Breakthrough curve

Saturation point

C t = 0 C t = 0 C t = C b

Break point

FC t

FC 0

dh

h

V p

ε : bed porosity

(a) (b)

Fig 2 Representation of a

typical breakthrough curve

(a) and schema of bed depth for

modeling (b) (Taty-Costodes

et al 2005 )

2005; Kundu and Gupta2007) For this system, the balance

can be expressed mathematically as:

FC0¼ FCtþ Vp

dC

dt þ mdq

where FC0 is the inlet flow of H2S in the column

(mg min-1), FCt is the outlet flow of H2S leaving the

column (mg min-1), Vp the porous volume (Vp= (1/

(1-e))V where V is the bulk volume and e is the void

fraction in the bed), Vp(dC/dt) is the flow rate through the

column bed depth (mg min-1) and m(dq/dt) is the amount

of H2S adsorbed onto Fe2O3(mg min-1) where m is the

mass of Fe2O3in the bed (g) and dq/dt is the adsorption

rate (mg g-1min-1)

According to Eq (14), it is obvious that the linear flow

rate (u = F/Sc, where Sc is the column section, m2), the

initial concentration, the adsorption potential and the

por-ous volume are the determining factors of the balance for a

given column bed depth Thus, in order to optimize the

fixed-bed column adsorption process, it is necessary to

examine the parameters and to estimate their influence

(Taty-Costodes et al 2005) However, these equations

derived to model the fixed-bed adsorption system with

theoretical power are differential in natural and usually

require complex numerical methods to solve them

Therefore, various simple numerical models have been

developed to predict the dynamic behavior of the hydrogen

sulfide on Fe2O3, the mathematical models must include

the adsorption isotherm, the mass-energy balance inside the

adsorbent particle and the gaseous phase in the bed Some

of these models have been discussed here

2.3.3 The Thomas model

The Thomas model is one of the most general and widely

used models to describe the behavior of adsorption process

in fixed-bed column The model was based on the

hypothesis that (1) the process follows Langmuir isotherms

and second-order kinetics of sorption–desorption with no

axial dispersion, (2) the adsorption is not limited by the

chemical reaction, but controlled by the mass transfer at the

interface (Kundu and Gupta2007; Thomas1944; Bharathi

and Ramesh 2013) The linearized form of the model is

given as:

ln C0

Ct

 1

¼kThq0m

where kThis the Thomas rate constant (L mg-1min-1), q0

is the maximum adsorption capacity (mg g-1), m is the

mass of adsorbent in the column (g)

The kinetic coefficient kTh and the adsorption capacity

of the column q0can be determined from a plot of ln((C0/

Ct)-1) against t (=Veff/F) at a given flow rate

2.3.4 The bed depth service time (BDST) model The BDST model (Hutchins1973), proposed by Hutchins, describes a relation between the service time and the packed-bed depth of column This model was derived based on the assumption that the diffusion steps (external and internal) are very fast, and the surface reaction step is rate-controlling The model has the following form:

t¼N0h

uC0

KC0

ln C0

Ct

 1

ð16Þ

where C0is the influent concentration of H2S (ppm), Ctis the effluent concentration of H2S at time t (ppm), K is the adsorption rate constant (L mg-1 min-1), N0 is the adsorption capacity (mg L-1), h is the bed depth (cm), u is the linear flow rate (cm min-1) and t is the service time to breakthrough (min) Experimental data obtained are used to plot BDST curves and estimate the characteristic parame-ters K and N0from the slope and intercept of the plots

3 Results and discussion 3.1 Characterization of the Fe2O3-based extruded adsorbent

The results of XRD analysis as shown in the Fig.3 indi-cates that the crystalline structure of the Fe2O3-based extruded adsorbent consists of Fe2O3and bentonite struc-ture This means that the calcination temperature (500°C) was successfully converted ferric hydroxide (Fe(OH)3) into

Fe2O3and no other new crystalline structure was formed during the heat treatment Figure4reports also SEM image

of the inner part of the extruded adsorbent which was obtained by breaking the sample before SEM analysis Moreover, the specific surface area, which was analyzed by

N2adsorption at 77 K using BET method, of the Fe2O3 -based extruded adsorbent was 60 m2/g The average pore

Fig 3 XRD pattern of the Fe2O3-based extruded adsorbent

size was 99 nm which was large enough for the gas

dif-fusion Thus, it can be concluded that the adsorbent was

relatively porous and potential for sorption application

(Table1)

3.2 Experimental results of H2S removal

The shape of the breakthrough curve and the time for the

breakthrough appearance are the predominant factors for

determining the operation and the dynamic response of an

adsorption column (Kundu and Gupta 2007; Mahmoud

2016) The general position of the breakthrough curve

along the volume/time axis depends on the capacity of the

adsorbent with respect to bed height, the feed concentration

and flow rate The dynamic behaviors for hydrogen sulfide

removal of the Fe2O3-based extruded adsorbent at different

operating conditions are reported in the Fig.5 The detail

adsorption data are given in the Table2 It can be seen

from Fig.5a and Table2 that when the bed depth of the

adsorbent increased from 3 cm to 6 cm the breakthrough

time at 10 %C0(t0.1) was increased from 228 to 438 min

The adsorption capacities at t0.1, however, were almost

unchanged, in the range 17.1–17.4 mg/g The maximum

adsorption capacities (when C = C0) were from 22.2 to

24.5 mg/g It means that about 70–80 % of the adsorption

sites had been used before t0.1 On the other hand, the

change of volume flow-rate led to significant changes of

the maximum adsorption capacity and slightly changes of the adsorption capacity at t0.1 as revealed from Table2 The adsorption capacities at t0.1 and the maximum adsorption capacities were decreased from 17.1 to 16.0 and 25.1 to 18.5 mg/g when the volume flow-rate was raised from 3.3 9 10-2L/min to 8.3 9 10-2L/min These results can be explained by the shorter contact time at the higher flow-rate Figure5c and Table2 show that the ini-tial H2S concentration slightly affected the adsorption capacities at t0.1and the maximum adsorption capacities The adsorption capacity at t0.1was 16.0 and 17.1 mg/g at

C0= 600 and 1200 ppm, respectively while the maximum adsorption capacity was 22.4 and 25.1 mg/g at C0= 600 and 1200 ppm, respectively

Additionally, according to the adsorption mechanism (Eqs.1 3) capacity of the Fe2O3-based extruded adsorbent could be about 226 mgS/g if all Fe3?ions were exposed to the surface and the reaction time was long enough to obtain equilibrium state Since the estimation of total O2- on

Fe2O3material is (1–2) 9 1019atoms/m2(Davydov et al

1998), total Fe3?ions present on surface of the Fe2O3-based extruded adsorbent is approximately (24–48) 9 1019 atoms/g As a result, the H2S capacity of the adsorbent may reach (77–154) mgS/g The capacity of the Fe2O3-based extruded adsorbent at Cb= C0as shown in Table2varies

in the range 18.5–25.1 mgS/g Therefore, the interaction is possibly in monolayer Even though the capacity in the dynamic conditions of the Fe2O3-based extruded adsorbent

is lower than the equilibrium capacity due to the short reaction time and mass transfer limitation, the capacity of this material is much larger than those of the granule Fe2

O3/montmorillonite which were (0.57–9.65) mgS/g (Nguyen-Thanh et al 2005) Furthermore, the capacity of the Fe2O3-based extruded adsorbent (per gram of Fe2O3) is much higher than that of the powder pure Fe2O3 which capacity was (7.9–13.0) mgS/g (Davydov et al.1998) The difference is possibly from the specific surface area

Fig 4 SEM images of the Fe2O3-based extruded adsorbent

Table 1 Properties of the Fe2O3-based extruded adsorbent by N2

physical adsorption at 77 K

difference The specific surface area of the extruded

adsorbent was 60.6 m2/g which was much higher than that

of the powder pure Fe2O3 (10 m2/g) It is because the

powder pure Fe2O3was prepared by precipitation method

while the extruded one was prepared from the Fe2O3which

synthesized by hydrothermal-precipitation method The capacity of the Fe2O3-based extruded adsorbent is (30.8–41.8) mgS/g_Fe2O3which is comparable to that of the powder pure Fe(OH)3 which capacity was about (34.5–42.5) mgS/g_Fe2O3(Davydov et al.1998) Since the

(a)

0

0.2

0.4

0.6

0.8

1

t (min)

3 cm

4 cm

5 cm

6 cm

0 0.2 0.4 0.6 0.8 1

t (min)

0.033 L/min 0.050 L/min 0.067 L/min 0.083 L/min

(b)

0 0.2 0.4 0.6 0.8 1

t (min)

600 ppm

800 ppm

1000 ppm

1200 ppm

(c)

Fig 5 Breakthrough curves of H2S adsorption onto the Fe2O3-based

extruded adsorbent: a at different bed depths (F = 0.05 L/min, C0 =

1000 ppm), b at different flow rates (h = 4 cm, C0 = 1000 ppm) and c at different initial H2S concentrations (F = 0.05 L/min, h = 4 cm)

Table 2 Adsorption data of H2S removal by the Fe2O3-based extruded adsorbent at different process parameters

Process

parameters

Breakthrough time (Cb= 0.1 C0) (min)a

Treated volume (L)b

H2S adsorption capacity (Cb= 0.1 C0) (mg/g)c

Total H2S removal (%)d

H2S maximum adsorption capacity (Cb= C0) (mg/g)c

Bed depth, h (cm)e

Flow rate, F (L/min)f

Initial H2S concentration, C0(ppm)g

a Obtained from Fig 5

b,c,d Calculated according to Eqs ( 7 ), ( 11 ) and ( 13 ), respectively

e at C0= 1000 ppm, F = 3 L/h

f at C0= 1000 ppm, m = 1.13 g

g at F = 3 L/h, m = 1.13 g

surface area of the powder pure Fe(OH)3 was 100 m2/g

which was higher than that of the extruded adsorbent, the

surface (–OH) groups may limit the H2S to access to the

Fe3? ions reducing the H2S capacity according to the

adsorption mechanism (Eqs.1 3)

3.3 Modeling

Two mathematic models have been used for modeling the

breakthrough curves basing on the experimental data

reported in Fig.5 and Table2 Adsorption process by the

low-cost Fe2O3-based extruded adsorbent consists of

sev-eral steps However the adsorption rate usually determines

by the slowest step which is called the rate-determining

step The rate-determining step can be one of the

follow-ings: (1) surface reaction, and (2) mass transfer

Identifi-cation of the rate-determining step is important for

understanding the kinetic nature of the adsorption process,

designing the adsorption column, and providing crucial

information for adsorbent development The two selected

models were based on various rate-determining step

assumptions The Thomas model assumed that the

diffu-sion step is the rate-determining step while the surface

reaction was assumed to be the rate-determining step in the

BDST model Firstly, a set of experimental data with

C B 10 % C0 was used as fitting data for both models to

determine the model’s parameters The effects of the

pro-cess conditions on the model’s parameters were then

ana-lyzed Finally, both models were applied to predict the

breakthrough curve of a pilot experiment

The coefficients of the Thomas model were determined

from the slope and intercepts obtained from the linear

regression derived from the Eq (15) Table3presents the

values of kTh and q0 obtained from the model fitting and

Fig.6 shows the comparison of simulated breakthrough

curves and the experimental data The fact that values of coefficient of determination (R2) were closed to unity ([0.95) revealed good agreement between the obtained models and the experimental data The errors of the pre-dicted 10 %C0 breakthrough times were from -0.8 to 3.3 % Thus, the model can provide correct information of the t0.1 The errors of the predicted 50 %C0breakthrough times were relatively high, from 3.3 to 18.6 % The reason

is that the complicated changes of adsorption’s driving force, which is the concentration different between the concentration of H2S on the Fe2O3-based adsorbent and the

H2S in the gas, leads to complicated mechanisms as increasing reaction time As can be seen from the Fig.6the deviation became larger when the reaction time increased for all the tested conditions

Additionally, the kTh and q0 were varied when the process parameters changed While the adsorbent’s weight and the initial H2S concentration slightly affected the model’s parameters, the effect of volume flow-rate on them was significant In order to simplify the effects of process parameter and to possibly obtain the Thomas model’s coefficients for scale-up cases, the kThand q0are correlated

to (m/F), a contact-time related parameter The results are shown in Fig.7 together with the maximum adsorption capacity (q0 Exp) taken from Table2 It is seen that the relationships between the coefficients and the (m/F) were not clear at low (m/F) However, if the (m/F) is large enough ([23 g L-1 min), the constant values can be obtained for the Thomas model’s coefficients (kThand q0) The two constant values are obtained by taking the average

of the kThand q0with (m/F) [ 23 g L-1.min, respectively

It is known that if the contact time is long enough the equilibrium state is obtained At a fixed temperature, the adsorption capacity at equilibrium state is a constant For the Fe2O3-based extruded adsorbent in this study, when the

Table 3 Thomas model parameters for H2S adsorption by the Fe2O3-based extruded adsorbent at different bed depths, inlet H2S concentrations and flow rates

C0(ppm) m (g) F (L/min) m/F (g L-1min) kTh(L mg-1min-1) q0(mg/g) R2 e 0.1 e 0.5

e0.1= (t0.1_exp.- t0.1_cal)/t0.1_exp.9 100 (%), e0.5= (t0.5_exp.- t0.5_cal)/t0.5_exp.9 100 (%)

(m/F) is larger than 23 g L-1min, the equilibrium state is achieved as can be reflected from the curve for experi-mental maximum adsorption capacity (q0.exp) in Fig.7 Coefficients for BDST model were obtained from linear relationship plots between experimental data t and ln(C0/

Ct -1) and reported in Table5 This model also can be used to simulate the breakthrough curve with CtB 0.1C0 for the mini column since the values of R2 are close to unity The small errors of the predicted t0.1 by the BDST model from -9.5 to 0.5 % were obtained It means that the prediction by this model is acceptable However, similar to

0

0.1

0.2

0.3

0.4

0.5

0 200 400 600

C 0

Time (min)

3 cm

4 cm

5 cm

6 cm model (a)

0 0.1 0.2 0.3 0.4 0.5

0 200 400 600 800 1000

C 0

Time (min)

0.033 L/min 0.050 L/min 0.067 L/min 0.083 L/min

model (b)

0 0.1 0.2 0.3 0.4 0.5

0 200 400 600 800 1000

C 0

Time (min)

600 ppm

800 ppm

1000 ppm

1200 ppm

model (c)

Fig 6 Comparison between experimental data and simulated data by Thomas model: a at different bed depths, b at different flow-rates, and c at different initial H2S concentrations

0 0.05 0.1 0.15

0

10

20

30

k Th

-1 m

-1 )

q 0

m/F (g.L -1 min)

q0 (Exp.)

q0 (Thomas)

k_Th

Fig 7 Effect of (m/F) on Thomas model parameters

0

0.1

0.2

0.3

0.4

0.5

0 200 400 600

C 0

Time (min)

3 cm

4 cm

5 cm

6 cm model (a)

0 0.1 0.2 0.3 0.4 0.5

0 200 400 600 800 1000

C 0

Time (min)

0.033 L/min 0.050 L/min 0.067 L/min 0.083 L/min

model (b)

0 0.1 0.2 0.3 0.4 0.5

0 200 400 600 800 1000

C 0

Time (min)

600 ppm

800 ppm

1000 ppm

1200 ppm

model (c)

Fig 8 Comparison between experimental data and simulated data by BDST model: a at different bed depths, b at different flow-rates, and c at different initial H2S concentrations

the Thomas model, this model showed deviation in the

breakthrough time (t0.5) which was obtained

experimen-tally at Ct= 50 %C0 The differences were from 4.4 to

17.1 % Figure8 shows the comparison of modeled

breakthrough curves and the experimental data It can be

seen that the model can be applicable for simulation of the

adsorption at low C/C0region

Moreover, the model’s coefficients (K and N0) varied

complicatedly with the column bed depth, the linear

flow-rate and the initial H2S concentration It is seen from the

Table4that similar to previous models the contact time (h/

u) was strongly influenced the BDST model’s adsorption

capacity (N0) and the adsorption rate constant (K) Unlike

the Thomas model, K was changed significantly with initial

H2S concentration (C0) The correlations are shown in

Fig.7 It can be seen from Fig.9a that if the contact time is

longer than 0.04 min, K and NB0 can be set at

0.0623 L mg-1min-1 and 20.4 mg g-1 (or 11,441

(mg L-1) at C0= 1000 ppm, respectively The

relation-ship of K and C0 is a linear relation as observed from

Fig.9b: K = 1.01 9 10-4C0 [ppm] - 3.03 9 10-2

[L mg-1min-1]

3.4 Scale-up assessment

Two models have been analyzed and both models can

simulate well the breakthrough curves with CtB 0.1C0of

the mini column Depending on the contact-time related

parameter (m/F or h/u) the model’s coefficients can be

obtained For checking the scale-up, the adsorption of H2S

onto the Fe2O3-based extruded adsorbent was carried out

on a pilot column to demonstrate the BDST, Thomas

models ability to be scale-up and used for design

Param-eters of the pilot column are presented in Table5 The

results are reported in Fig.10and Table6 It can be seen

that the two models can simulate the breakthrough curve of the pilot test However, the breakthrough times for 10 %C0 predicted by BDST model were shorter than the actual breakthrough time obtained by the experiment The Tho-mas model gave better estimation with about 4 % error There are two possible reasons for the results Firstly, even though both models predicted well the t0.1 in the mini column tests, the errors from the Thomas model were much lower than those from the BDST model as shown in Tables3, and4 Secondly, Additional axial mass transfer resistance of the pilot column, which cross section area was

10 times larger than that of the mini column, makes the mass transfer resistance become the actual

rate-Table 4 BDST model parameters for H2S adsorption onto Fe2O3at different bed depths, inlet H2S concentrations and flow rates

(mg/L) (mg/g)

e 0.1 = (t0.1_exp.- t0.1_cal)/t0.1_exp.9 100 (%), e0.5= (t0.5_exp.- t0.5_cal)/t0.5_exp.9 100 (%)

0 0.1 0.2 0.3 0.4

0 10 20 30

-1 m

-1 )

q 0

h/u (min)

q0 (Exp.) N0 (BDST) K

(a)

K = 1.01×10 -4 C 0 - 3.03×10 -2

R² = 0.99

0 0.05 0.1 0.15 0.2

-1 m

-1 )

C 0 (ppm) (b)

Fig 9 Effect of (h/u) on BDST model parameters at C0 = 1000 ppm (a), and the effect of initial H2S concentration on the rate constant K

at h/u = 0.04 min (b)