Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers

ARTICLE IN PRESS

JID: JTICE

[m5G;September 14, 2016;4:40]

Journal of the Taiwan Institute of Chemical Engineers 0 0 0 (2016) 1–11

Contents lists available at ScienceDirect

Journal of the Taiwan Institute of Chemical Engineers journal homepage: www.elsevier.com/locate/jtice

Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers ´ Svetlana S. Popovic, ´ Miodrag N. Tekic, ´ Nataša Lj. Lukic´ ∗, Ivana M. Šijacˇ ki, Predrag S. Kojic, Dragan Lj. Petrovic´ Department of Chemical Engineering, Faculty of Technology, University of Novi Sad, Bulevar cara Lazara 1 21000 Novi Sad, Serbia

a r t i c l e

i n f o

Article history: Received 21 March 2016 Revised 31 August 2016 Accepted 5 September 2016 Available online xxx Keywords: Impellers Self-agitated External-loop airlift Hydrodynamics Non-Newtonian Alcohol

a b s t r a c t In the present work, self-agitated impellers, as a novel type of internals, have been proposed for improving hydrodynamic and mass transfer characteristics of external-loop airlift reactors. The influence of inserted self-agitated impellers, in the riser section, in various liquid phases and with different sparger types, on main hydrodynamic parameters, was studied. The results show that the insertion of impellers led to significant bubble breakage and decrease in mean bubble size, particularly in pseudoplastic liquid. Obtained riser gas holdup values were up to 47% higher, in comparison to the configuration without impellers. Higher improvements were obtained with single orifice, given that this is the least effective gas distributor. Even though impellers represent an obstacle to the flow, relatively low reduction (about 10%) in downcomer liquid velocity was observed, for all investigated cases. Having this in mind, the benefit of inserting self-agitated impellers for improving the performance of external-loop airlift reactors was apparent. © 2016 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction As a result of simple construction without moving parts, low shear rate, good mixing and low energy requirements, externalloop airlift reactors (EL-ALR) are extensively employed in many biochemical and pharmaceutical applications. Their performance is notably affected by complex interrelationships between hydrodynamic parameters, transport phenomena, design and operating variables, microbial survival and production kinetics [1,2]. One of the key factors in determining their productivity is the gas-liquid mass transfer. Improved mass transfer and thereby higher productivity are achieved by increasing either the specific gas-liquid interfacial area, a, or mass transfer coefficient, kL . Smaller bubble diameter, higher and more uniform radial holdup profiles initiate an increase in the interfacial area and hence the more intimate contact between phases is achieved [3]. Intensified turbulence promotes higher mass transfer by increasing mass transfer coefficient, destabilizing large bubbles and increasing surface renewal frequency of bubbles [4]. Nevertheless, with an increase in liquid velocity rates, the residence time of the gas phase shortens and thus a decrease in the gas holdup is obtained. Because of the important influence of hydrodynamics on mass transfer rate, detailed characterization of hydrodynamic parameters, such as gas

∗

Corresponding author: Fax +381 21 450413. ´ E-mail address: [email protected] (N.Lj. Lukic).

holdup, liquid velocity rate and bubble behavior, is fundamental for the assessment of EL-ALR operation. Hydrodynamics and mass transfer in EL-ALRs are largely affected by liquid phase properties, such as rheological behavior and surface tension, and sparger design. Most of the commercial fermentation processes involve viscous non-Newtonian media thus instigating numerous studies, regarding their behavior in EL-ALRs. Carboxymethylcellulose (CMC) is mainly employed as a model fluid because it has properties that highly resemble fermentation media. The effect of CMC on hydrodynamics is greatly dependent upon the apparent viscosity of the solution. Wu et al. [5] observed that in CMC solutions with lower viscosities bubble coalescence was prevented. As a result, higher riser gas holdup values were achieved. However, in highly viscous CMC solutions very large irregularshaped and spherical-cap bubbles, with high rise velocities, are accompanied by very small bubbles [6]. Since larger bubbles have higher rise velocities, lower gas holdup values are obtained. Considering that an increase in mean diameter of the bubbles reduces a, whereas kL decreases because of lower diffusivity, mass transfer is highly diminished in viscous non-Newtonian liquids [7]. In some fermentation applications non-coalescing media are involved. Alcohols, as surface active liquids, supress bubble coalescence and thus have a huge impact on hydrodynamics and mass transfer characteristics and hence, are used to simulate the behavior of non-coalescing media. The addition of small amounts of aliphatic alcohols decreases the mean size of bubbles and reduces bubble

http://dx.doi.org/10.1016/j.jtice.2016.09.003 1876-1070/© 2016 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

ARTICLE IN PRESS

JID: JTICE

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

2

– 54–76% – 9–24% 17–29% – 23–37% – – – – 3–18% ‡

†

Relative improvement of riser gas holdup (water as a liquid phase). Relative reduction of liquid velocity (water as a liquid phase).

ULR , UG,min , kL a Local ε GR , ULR , bubble rise velocity, db ε GR , ULc kL a ULR , ε GR ε GR , WLD

ε GR

kL a, tm, tc kL a, ε GR , ULR , bubble size kL a ε GR , ULR , axial dispersion, bubble size distribution tm , ULR

DR = 150, DD = 50, H= 30 0 0 DR = 89, DD = 47, HR = 1810 DR = 89, DD = 47, HR = 1810 DR = 89, DD = 47, HR = 1810 DR = 248, DD = 102, H= 1996 DR = 248, DD = 102, H= 1996 DR = 27, DD = 61, HR = 585, HD = 390 DR = 230, DD = 190, H= 4800 DR = 219.9, H= 1820 DR = 50, DD = 75 AD /AR = 0.1225 DR = 93, DD = 54, H=2360

– 20–130% – 21–35% – 5–15% – – – – – 11–36%

LVR Investigated parameters

Slanted baffles Woven stainless steel mesh packing Woven nylon mesh packing Woven nylon packing Two rolls of fiberglass packing Two rolls of fiberglass packing Static mixer Perforated baffles (45° angle between baffle and vertical axis) 7 internals (4 contraction and 3 expansion discs) SMV-12 static mixer elements Sulzer type static mixer 9 self-agitated impellers

rise velocities which lead to increased gas holdup [8,9]. In EL-ALRs, the type of gas sparger influences hydrodynamics only through initial bubble size [10]. The impact of sparger type is more pronounced at lower gas throughputs, i.e. bubbly flow or transition flow, in which the size of bubbles in the dispersion is determined by the bubble size at formation [11]. In the case of heterogeneous flow, the influence of sparger type is lessened due to strong bubble coalescence. Hence, sparger type influence is even more emphasized in systems with inhibited coalescence, like alcohol solutions. Various modifications of both internal- and external-loop airlift reactors have been developed as a result of increasing demand for improved yield and productivity. Some of them have inserted internals like baffles [12,13], perforated plates [3,7,14], static mixers [15– 17], mechanically driven impellers [18–20], packed beds [21–23] or custom designed internals [4,24,25], which obstruct fluid flow and intensify mixing and mass transfer. Summarized in Table 1 are listed characteristics of modified EL-ALR with the details of used internals.

Lin et al. [12] Nikakhtari and Hill [22] Nikakhtari and Hill [26] Meng et al. [23] Hamood-ur-Rehman et al. [21] Hamood-ur-Rehman et al. [27] Goto and Gaspillo [15] Zhang et al. [4] Mohanty et al. [24] Chisti et al. [16] Gavrilescu et al. [17] This paper

Abbreviations CMC carboxymethylcellulose DT-ALR draft tube airlift reactor EL-ALR external-loop airlift reactor EL-ALRI external-loop airlift reactor with self-agitated impellers EL-ALRoX external-loop airlift reactor with restriction orifice (X denotes orifice free area) SO single orifice SP sinter plate

Table 1 Review of investigated types of internals in external-loop airlift reactors.

Subscripts c circulation D downcomer DT draft tube i impeller m mixing R riser

Internal type

Greek letters ε GR,v riser gas holdup measured with volume expansion technique ε GR riser gas holdup ρ density, kg/m3 σ surface tension, N/m

EL-ALR characteristics (mm)

HI

†

cross-sectional area, m2 specific gas-liquid interfacial area, 1/m diameter, m bubble diameter, m gravitational acceleration, m/s2 height, m improvement of riser gas holdup flow consistency index, Pa·sn overall friction coefficient mass transfer coefficient, m/s volumetric mass transfer coefficient, 1/s distance between two conductivity electrodes, m reduction of downcomer liquid velocity flow behavior index time, s superficial gas velocity, m/s liquid velocity, m/s

Reference

Notation A a D db g H HI K Kf kL kL a L12 LVR n t UG WL

‡

Nomenclature

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

JID: JTICE

ARTICLE IN PRESS

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

The research studies of EL-ALRs have investigated the insertion of motionless internals mainly in order to change the gas phase dispersion. Chisti et al. [16] reported that two SMV-12 static mixer elements, mounted in the riser, significantly enhanced volumetric mass transfer coefficient mainly due to the breakup of larger bubbles. Since in highly viscous fluids large spherical bubbles are formed, bubble breakup was even more pronounced in pseudoplastic media. Zhang et al. [4] mounted a novel internal composed of baffles with semi-circular holes with a tongue-like plate, in the riser of an EL-ALR to break the bubbles and promote flow redistribution. They determined a decrease in the bubble Sauter diameter after passing through the internal. The insertion of packing materials in the riser could also improve characteristics of EL-ALRs. Stainless steel mesh packing mounted in the riser of an EL-ALR led to higher gas holdup values due to lower liquid velocity caused by increased resistance to the flow and due to bubble breakup [22]. As a result higher mass transfer coefficient was achieved compared to the reactor without the packing material (up to a factor of 2.45). The hydrodynamics of a novel multi-stage EL-ALR mounted with seven internals (four contraction discs and three expansion discs) was studied by Mohanty et al. [24]. By alternating contraction and expansion discs they successfully induced continuous bubble generation and breakup through rupture and regeneration. On the other hand, the instalment of mechanically driven impellers, already widely used in stirred tanks, so far has been investigated only in internal-loop airlift reactors [18,19,28]. This integration was studied with an objective to overcome limitations of stirred tanks and airlift reactors in highly viscous media like poor mixing pattern and low mass transfer, respectively. Mechanically agitated impellers definitely led to an enhancement in gas-liquid transfer rate, mixing and liquid circulation rate in comparison to the reactor without agitation. The drawback of this configuration was that improvements were made at the expense of a substantial increase in energy demands [18]. Having this in mind, as well as that resistance to the liquid flow inevitably increases with motionless internals, research on self-agitated impellers in an internal-loop airlift reactor was initiated in a previous work [29]. Agitation in draft tube airlift reactor (DT-ALR) was introduced by using the energy of the gas throughput and induced liquid circulation. Significant changes in the hydrodynamics occurred: riser gas holdup and mixing time increased while downcomer gas holdup and liquid velocity decreased. However, the addition of impellers showed almost no effect on the overall gas holdup because of the impellers’ opposite influence on downcomer and riser gas holdup. This was attributed to the conditions employed in that work, predominantly to the higher crosssectional area of the downcomer. Still, it was concluded that selfagitated impellers have potential in enhancing the hydrodynamics and mass transfer, especially in an airlift reactor configuration with lower downcomer to riser ratio, such as external-loop airlift reactor. The main goal of this experimental work was to develop a novel external-loop airlift reactor by installing self-agitated impellers in the riser in order to improve its hydrodynamic characteristics. The effect of inserted impellers on riser gas holdup and downcomer liquid velocity is investigated through comparative analysis. In addition, the study was extended on the effects of liquid phase properties (surface tension and viscosity) and sparger type (single orifice and sinter plate) on the efficiency of impellers. 2. Material and methods 2.1. Experimental setup A 36.6 dm3 external-loop airlift reactor made of stainless steel and Plexiglas was used in this study. The riser and downcomer

3

Table 2 Properties of liquid phases used at 20 °C. Liquid

ρ [kg/m3 ]

σ [mN/m]

K [Pa·sn ]

n

Tap water 0.011 wt.% n-butanol 0.3 wt.% CMC

999 998 1002

73.7 71.8 73.1

0.0010 0.0010 0.1004

1.0 1.0 0.684

(Plexiglas) had an inner diameter of 93 and 54 mm, respectively, and height of 2 m. The distance between the central axes of the riser and the downcomer was 35 cm. Rectangular gas-liquid separator was 52 cm in length and 20 cm in width. The static liquid height in the reactor was 2.36 m in all runs. The schematic diagram of the experimental apparatus is shown in Fig. 1. A conventional EL-ALR was modified by mounting a shaft with 9 impellers in the centre of the riser. In this novel configuration, abbreviated as ELALRI, gas throughput and induced liquid circulation were driving the impellers to rotate. Seven-bladed axial flow impellers, made of plastic material, with 40 ° blade angle of attack and diameter of 73 mm were used. Friction between the shaft and impellers was reduced by installing Teflon rings in the impeller’s housing. The ratio of a distance between two adjacent impellers to the riser diameter, hi /DR =2, was chosen based on the optimal ratio hi /DDT determined in a previous work [29]. All experiments were carried out at 20 ± 1 °C and atmospheric pressure. The liquid height in the separator allowed complete deaeration and, therefore, no gas bubbles were entrained in the downcomer. The air, used as the gas phase, was sparged either through a single orifice (4 mm i.d.) or sinter plate (38 mm in diameter, average pore size 115 μm). The gas flow rate was regulated and measured by a mass flow controller (Bronkhorst High Tech F-201AV). It was varied from 4 to 65 LN /min, so the corresponding superficial gas velocities were in the range 0.0096-0.1552 m/s, based on the cross-sectional area of the riser. Tap water and following aqueous solutions: 0.011 wt.% nbutanol and 0.3 wt.% carboxymethylcellulose (CEKOL 10,0 0 0, CP Kelco, Aanekoski, Finland) were used as liquid phases in all experiments. The main physical properties of the liquid phases are summarized in Table 2. Rheological behavior was obtained by a controlled-stress rheometer HAAKE Rheostress RS600 (Thermo Electron Corporation, Karlsruhe, Germany) at a constant temperature of 20 ± 0.05 °C using the cone-and-plate geometry C60/1Ti (diameter = 60 mm and angle = 1°). Surface tensions of the liquid phases were obtained at 20 ± 0.5 °C using tensiometer Sigma 703D (KSV Instruments, Finland) applying the Du Noüy ring method. All measurements were performed in duplicate and average values are presented. In order to change the resistance to liquid circulation and obtain lower liquid recirculation rates for the same superficial gas velocity, restriction orifice was placed at the top of the downcomer in EL-ALR. Four different restriction orifices, having 70, 50, 30 or 10% free orifice area, have been used. These modified configurations are abbreviated as EL-ALRoX (X denotes 70, 50, 30 or 10). 2.2. Measuring methods In a conventional EL-ALR and EL-ALRoX riser gas holdup ε GR was determined by the manometric method, measuring pressure differences at the top and bottom of the riser using piezometric tubes, with a relative error less than 3%. Excess fluctuations in the liquid height readings, observed in nonviscous liquids, were suppressed by installing glass capillaries of 0.7 mm i.d. and 50 mm in length at the entrance of the piezometric tubes. On the other hand, in an EL-ALRI, the presence of self-agitated impellers caused significant pressure drop and consequently manometric method was

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

JID: JTICE

ARTICLE IN PRESS

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

4

Fig. 1. The schematic diagram of the experimental equipment.

inadequate for measuring ε GR . Therefore, an alternative procedure for determining ε GR had to be adopted. A series of experiments were performed in a configuration without impellers to construct nomograms which would be used to determine riser gas holdup in a configuration with impellers. To prepare nomograms, it was necessary to measure riser gas holdup by manometric method (ε GR ) and by volume expansion technique (ε GR,v ), and downcomer liquid velocity (WLD ), in EL-ALR. ε GR,v was determined from the ratio of the difference between the aerated and static liquid volumes to the volume of the riser. Piezometric tube, placed in the separator (see Fig. 1), was used to determine differences in the liquid volume by measuring liquid height before and during gas sparging. Glass capillary was inserted at the entrance of the tube to supress fluctuations in the readings of the liquid height. In this way, absolute error was less than 1 mm and, thus, maximum relative error of ε GR,v measurement was 11%. It was noticed that volume expansion technique slightly underestimated riser gas holdup despite installing flow stabilizer in the separator. This was mostly due to uneven liquid height in the separator because the swelling of the gas-liquid surface above the riser formed a wavelike liquid flow [30]. Nevertheless, linear relationship between measured values of ε GR and ε GR,v was found in EL-ALR. This relationship changes by altering the resistance to liquid circulation. For this reason, relations between ε GR and ε GR,v were determined at different recirculation rates, which was achieved by using various restriction orifices in EL-ALR. Fig. 2 illustrates obtained relationships between ε GR and ε GR,v for water and single orifice in EL-ALR, EL-ALRo70 and ELALRo10. A connection between riser gas holdup and liquid circulation rates for each liquid phase and sparger type proposed by Verlaan [31] was used:

WLD 2 =

2gH εGR Kf

(1)

This enabled us to construct nomograms of ε GR versus WLD for various ε GR,v values. As an example, Fig. 3 depicts nomogram constructed for water and single orifice. Finally, riser gas holdup (ε GR ) was read off from nomograms using ε GR,v and WLD measured in EL-ALRI, as illustrated in Fig. 3. Tracer response technique was employed to determine downcomer liquid velocity. A pulse of 30 ml of 4 M NaCl aqueous solution was used as a tracer. Tracer was added instantaneously into the liquid at the top of the downcomer at H= 1.85 m. The response of the impulse input was measured simultaneously with two conductivity probes (Microelectrodes ET915, eDAQ , Australia) that were mounted downstream, in the upper (just below the tracer injection point) and lower section of the downcomer at H= 1.7 and 0.9 m, respectively. The signals from the probes were converted and transferred to a computer using Conductivity USB isoPodTM (model EPU357, eDAQ , Australia) at a sampling frequency of 50 Hz. Downcomer liquid velocity was determined by applying the following equation:

WLD =

L12 t2 − t1

(2)

where L12 is the length between two electrodes and t2 and t1 are the times of the first peak corresponding to the moment when the tracer passes over the lower and upper conductivity probe, respectively. Maximum relative error for the determination of WLD was 10% at the highest UG . For Newtonian fluids, chosen volume of the tracer allowed six runs without disturbing the hydrodynamics of the column [32], while for non-Newtonian fluids only two runs were performed to avoid changes in the rheological behavior of CMC solution, after which the column was drained and fresh liquid phase was used. The efficiency of self-agitated impellers was determined by evaluating improvement of riser gas holdup (HI) and the reduction

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

ARTICLE IN PRESS

JID: JTICE

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

5

Fig. 2. Linear relationship between ε GR and ε GR,v for water and single orifice.

Fig. 3. Nomogram constructed for the determination of riser gas holdup for water and single orifice in EL-ALRI.

of downcomer liquid velocity (LVR). Riser gas holdup improvement was defined as relative improvement of ε GR in the presence of impellers, compared to ε GR without them:

HI =

(εGR )with

impellers

− (εGR )without

(εGR )without

impellers

· 100%

(3)

impellers

while reduction of downcomer liquid velocity was defined as relative reduction of WLD due to the insertion of impellers:

LV R =

(WLD )without impellers − (WLD )with (WLD )without impellers

impellers

· 100%

(4)

The impeller rotation speed was recorded on a high speed video camera (Canon Ixus 500HS) that records up to 240 fps. Still image frames were extracted using Matlab R2015b. Impeller blade was marked with bright yellow colour and by counting the number of full rotations during a period of 10 s the impeller speed was calculated. By using high speed camera maximum error between two measurements was less than 4%. All measurements were carried out in duplicate and average values were calculated.

Fig. 4. Effect of the gas superficial velocity on impeller rotational speed.

3. Results and discussion 3.1. Hydrodynamic observations Within the range of gas superficial velocities investigated in EL-ALRI, all impellers developed almost uniform rotational speed with the maximum deviation less than 6%, indicating that impeller self-agitation ability is independent of axial position in the riser. By examining Fig. 4, which demonstrates the average rotational speed of impellers at different superficial gas velocities, it can be seen that with the increase of UG , the speed of rotation also increased. The speed of impellers constantly increased with a more moderate increase at higher gas velocities unlike the results previously obtained in DT-ALR [29], where impeller rotational speed tended to be constant above UG = 0.05 m/s. Accumulation of very small bubbles in the downcomer of DT-ALR formed a resistive layer to circulation flow and led to a steady circulation velocity despite the increase in UG . This shortcoming was avoided in EL-ALRI used in this work which operated under total disengagement of the gas phase with no additional energy losses due to the presence of bubbles in the downcomer. The effect of liquid phase on rotational speed was also investigated. From

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

JID: JTICE 6

ARTICLE IN PRESS

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

Fig. 5. Pressure of the gas entering the reactor in air-water system.

Fig. 4 it can be also observed that rheological properties of the liquid phase had a strong impact on impeller speed. In nonviscous Newtonian fluids (water and 0.011 wt.% n-butanol) values of impeller speeds are similar and in agreement. On the other hand, impeller rotational speeds in viscous fluids (0.3 wt.% CMC) are up to a factor of 2.7 smaller than in nonviscous fluids. Dissipation of energy necessary to overcome frictional forces in viscous fluids hindered impeller rotation and eventually caused lower rotational speeds in CMC. Nevertheless, obtained impeller rotational speeds in both nonviscous and viscous systems were sufficient enough for the blades of the impeller to break up and disperse bubbles across the riser. Influence of sparger type was noticed only in viscous CMC at UGR = 0.0096 m/s. Specifically, sinter plate (SP) had induced 34% lower rotational speed in comparison to the single orifice (SO) due to the existence of bubbly flow regime. It should be noted that results obtained for nonviscous liquids with sinter plate are omitted since the differences were within 6%. Since the insertion of self-agitated impellers caused additional resistance to liquid circulation, it was important to determine whether the added impellers substantially increased the power required for air to flow through EL-ALRI. Fig. 5 depicts the effect of inserted impellers on the pressure of the gas at inlet with water used as the liquid phase and single orifice. As it can be seen, until UG ≈ 0.09 m/s, the pressure of the gas at inlet was unaffected by insertion of impellers, thus suggesting no additional power requirements. At UG > 0.09 m/s impellers gave only about 5% higher values of inlet pressure which is considerably lower compared to DT-ALR thus justifying the suitability of installing impellers in ELALR over DT-ALR. Similar behavior was observed in other liquids with up to 10% higher values of pressure in both configurations when viscous CMC was used. Separate hydrodynamic flow regimes (bubbly, transition, churnturbulent or unstable slug flow) were identified in the riser of ELALR. In the range of investigated UG , the existence of a specific regime was dictated by liquid phase properties and sparger type, as reported by other authors [33]. When single orifice was used, bubbly flow was absent. In nonviscous liquids, transition flow fully developed into churn-turbulent flow for UG in the range 0.0215– 0.0343 m/s. As expected, the addition of alcohol delayed the appearance of churn-turbulent flow. In viscous CMC solution, liquid turbulence intensity was diminished and newly formed bubbles were more stable in terms of bubble breakup. Because of a relatively lower viscosity of CMC used in this work, coalescence was present in the riser of EL-ALR. As a result, the frequency of bubble coalescence exceeded bubble breakup and very large spherical-cap bubbles were present in the riser, even at the lowest UG . Therefore, when single orifice was used unstable slug flow was noticed at low gas throughputs in CMC. This flow is characterized with

slug bubbles which demonstrate some level of inconsistency, as defined by Bajón Fernández et al. [34]. Besides spherical-cap bubbles, very small bubbles generated during the coalescence process when a tail of a bubble breaks down were also observed [35]. With an increase in UG churn-turbulent flow was developed, while at higher UG (> 0.1194 m/s), flow with mixed characteristics of slug and churn-turbulent flow was present. Sinter plate had significant influence on regimes considering that bubbly flow was identified for all liquid phases. In water and n-butanol, transition and churnturbulent regime were developed at UG in the range 0.0215–0.032 and 0.0756–0.091 m/s, respectively. Besides the appearance of bubbly flow, sinter plate delayed the existence of churn-turbulent flow. The viscosity of CMC solution used in this work allowed the appearance of bubbly flow only at the lowest UG , while at UG – 0.04 m/s transition flow developed into churn-turbulent flow. Inserted impellers altered bubble behavior in all liquid phases. The notable impact is the appearance of bubbly flow in n-butanol at lower gas throughputs when single orifice was used. This could be attributed to the combined effect of impellers and liquid phase properties: impellers initiated bubble breakup and afterward coalescence of generated small bubbles was prevented due to the inhibiting nature of n-butanol. Influence of impellers on regime transition was further confirmed by prolonging the appearance of transition points, for both spargers. Moreover, uniform radial distribution of the bubbles was visually observed in the configuration with impellers. In CMC, impellers initiated the existence of transition regime instead of unstable slug flow by breakage of large spherical-cap bubbles. In order to capture bubble breakup mechanism by self-agitated impellers in 0.3 wt.% CMC solution, a visualization technique was employed. Images were captured in both configurations, EL-ALRI (Fig. 6b) and EL-ALR (Fig. 6a), at the location of the seventh impeller (at H= 1.4 m from the gas sparger). The sequence taken in EL-ALRI shows a bubble before contact with the impeller, interaction between bubble and impeller that occurs a few moments later and a short time after generation of small bubbles. Images confirmed that for the same UG , rotation of impellers led to substantial changes in bubble behavior compared to the configuration without impellers. Specifically, efficient bubble breakage occurred when large bubbles came in contact with impellers. Afterward, generated small bubbles remained rising and some of them collided as their distance from impeller increased, resulting in coalescence; nevertheless bubble breakage promptly prevailed over coalescence due to the vicinity of the following impeller. This behavior is beneficial for mass transfer [36]. Also, it can be seen that the bubble approaching the seventh impeller had smaller diameter and lower rise velocity in comparison to the spherical-cap bubbles present in EL-ALR at the same height. This confirms the significant influence of added impellers on the hydrodynamics in external-loop airlift reactors. 3.2. Riser gas holdup Fig. 7 shows the influence of sparger type and superficial gas velocity on riser gas holdup in a configuration without impellers, for all liquid phases investigated in this work. It can be seen that in all the cases studied, riser gas holdup increases with the increase in UG , which is an expected result. At small gas throughputs, the effect of sparger type on ε GR , by influencing the size of the bubbles leaving the sparger, can be seen since higher values were obtained with sinter plate. Clearly, considering a 67% increase in ε GR , sparger type influence was most pronounced when viscous CMC solution was used. Since the viscosity of 0.3% CMC used in this work is relatively low, there is no coalescence present at the lowest gas velocities when sinter plate was used. Bubbles generated at the distributor were very stable and bubbly flow was present. Due to the

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

JID: JTICE

ARTICLE IN PRESS

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

7

Fig. 6. Sequence of images portraying bubble behavior at the location of the seventh impeller in 0.3 wt.% CMC solution at UG = 0.096 m/s with single orifice. (a) without impellers (EL-ALR); (b) with impellers (EL-ALRI).

Fig. 7. Influence of liquid phase properties and distributor type on ε GR in EL-ALR.

higher drag force in viscous CMC, bubble rise velocity was diminished which led to longer residence time. Therefore, the highest values of ε GR were achieved with viscous CMC at the lowest UG with sinter plate as sparger, in comparison to nonviscous liquids. On the other hand, at higher gas throughputs, corresponding to the inertial forces domination, riser gas holdup values achieved with both spargers were similar indicating that the influence of sparger type on ε GR was negligible. These findings concur with observations from other authors [37,38]. Fig. 7 also shows that liquid phase properties have a modest influence on riser gas holdup in the heterogeneous regime. By comparing nonviscous liquids with different surface tension properties, n-butanol gave slightly higher ε GR values in comparison to water. This is due to the much stronger tendency of bubbles to coalescence at highly turbulent conditions, i.e. heterogeneous regime, in which inhibiting nature of n-butanol is

emphasized. Opposite to this, shear-thinning behavior of CMC led to extensive bubble coalescence and as a result, lower values of ε GR were obtained in comparison to water. The decrease in ε GR did not manifest to this extent at lower gas velocities because, besides the negative effect of viscosity through higher coalescence rate, viscosity had a positive effect on ε GR , as mentioned above, by increasing drag force which reduced bubble rise velocity and increased bubble residence times. Influence of liquid phase properties and superficial gas velocity on the improvement of riser gas holdup (HI) achieved with impellers is depicted in Fig. 8. The presence of self-agitated impellers caused an increase in riser gas holdup for all investigated gas-liquid systems and sparger types, in comparison to conventional EL-ALR. Riser gas holdup was enhanced partially due to a decrease in bubble size caused by bubble breakup and partially

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

JID: JTICE 8

ARTICLE IN PRESS

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

Fig. 8. Influence of self-agitated impellers on the improvement of riser gas holdup.

due to the longer residence time of the gas phase in the riser [27]. Fig. 8a illustrates that when single orifice was used as a gas distributor, the largest differences for any liquid phase were obtained at the lowest superficial gas velocity. Impellers main objective, breakage of bubbles, is highly frequent when single orifice is used. Single orifice generates large bubbles which are then widely exposed to the effect of impellers. n-Butanol, as a coalescence inhibitor, has stabilized smaller bubbles generated by impellers and thereby the largest HI of 47% was obtained at the lowest UG . On the other hand, in CMC solution self-agitated impellers were less effective. Large spherical-cap bubbles, which are present in CMC, definitely have the highest preference for breakage by impellers. Nevertheless, subsided impact of impellers in viscous CMC occurred, in part because impeller rotation and consequently the rate of bubble breakage were hindered. Also, observed coalescence of generated small bubbles shortly after passing the impellers resulted in a lower increase in ε GR , in comparison to nonviscous liquids. Despite these shortcomings, self-agitated impellers increased riser gas holdup by 30% in CMC at the lowest UG . With increasing gas velocity, up to a value of 0.0478 m/s, HI decreased steeply in all investigated cases with single orifice as a sparger. Next, as the inertial forces became dominant over the surface forces the decrease became less steep and HI curves tended to be relatively constant regardless of UG . This probably occurred due to enhanced turbulence by impellers, which is more pronounced at higher gas velocities. In this way, impellers were not only directly responsible for bubble breakage, but also indirectly, since enhanced turbulence led to bubble breakage too. Opposite to single orifice, when sinter plate was used, the influence of impellers on the breakage of bubbles is weak at low gas velocities (see Fig. 8b). Small bubbles were already formed by sinter plate and consequently, impellers were less efficient. Also, it can be seen that when CMC was used, no improvements were obtained at lower UG . As previously discussed, this was also visually observed because very small and stable bubbles passed through the impellers unaffected. As was the case for configuration without impellers, in which sparger type influence was insignificant at higher gas velocities, i.e. in the range of 0.0478-0.1552 m/s, due to the domination of inertial

forces over surface forces, analogous conclusion could be derived for configuration with impellers. An average ε GR improvement of 19 and 22% was achieved for nonviscous water and n-butanol, respectively, and 13% for viscous CMC, with both sparger types. 3.3. Downcomer liquid velocity The effect of sparger type and liquid phase properties on downcomer liquid velocity in EL-ALR at various superficial gas velocities is depicted in Fig. 9. Obviously, with the increase in UG , downcomer liquid velocity increased for all investigated cases. It can be seen that the addition of n-butanol had no impact on WLD regardless of sparger type used. This type of behavior was also reported by Miyahara and Nagatani [8] in their study investigating the influence of alcohol addition on WLD . However, liquid phase viscosity had strong impact on WLD , evidenced by a decrease in WLD by 16-37% when viscous CMC was used. This was due to the stronger internal friction forces in viscous liquids, in comparison to water. Influence of sparger type was only noticed at low superficial gas velocities. The values of downcomer liquid velocity obtained with sinter plate were around 10% higher in water and n-butanol; and about 25% higher in CMC than those obtained with single orifice for UG in the range 0.0096-0.0143 m/s. Being an obstacle to the flow, self-agitated impellers increased frictional resistance causing energy dissipation; thereby a decrease in WLD was expected. The decrease in downcomer liquid velocity has a negative effect on kL a through the decrease of kL . So, it is important not to exceed the positive effects of impellers on a (higher residence time and smaller size of bubbles), which were previously determined, by having a limited reduction in WLD . Fig. 10 demonstrates the effect of self-agitated impellers on WLD reduction at different superficial gas velocities. Evidently, insertion of impellers had a modest impact on liquid velocity. Over the range of UG investigated in this study, LVR ranged from 3 to 18% when impellers were used. The results were highly scattered for all investigated cases and no conclusion could be drawn probably because of higher error in the determination of WLD . Still, it could be said that the instalment of impellers produced an average decrease in downcomer liquid velocity of 10% regardless of sparger type,

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

JID: JTICE

ARTICLE IN PRESS

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

9

Fig. 9. Influence of liquid phase properties and distributor type on WLD in EL-ALR.

Fig. 10. Influence of self-agitated impellers on the reduction of downcomer liquid velocity.

liquid phase or gas throughput. Therefore, instalment of selfagitated impellers will probably lead to improved mass transfer in external-loop airlift reactors through the increase in volumetric mass transfer coefficient. The comparison of our results with the results of other studies with internals in external-loop airlift reactors using water as liquid phase was performed. In Fig. 11 are shown the results for riser gas holdup improvement and downcomer liquid velocity reduction obtained with self-agitated impellers in this study along with other internals available in the literature. Nikakhtari and Hill [22] achieved highest HI of 130% at very low gas throughput. However, in order to obtain such high improvements, extreme reduction in WLD happened, as illustrated by LVR of 65%. Other authors mostly obtained improvements of ε GR and reduction of WLD in the range of 10 to 35% and 8 to 30%, respectively. Self-agitated impellers in our study gave similar results for all investigated cases. It is interesting to note that, for the range of superficial gas ve-

locities studied in this paper, self-agitated impellers gave the lowest decrease in downcomer liquid velocity by having the highest riser gas holdup increase when single orifice was used. As for sinter plate, self-agitated impellers produced lower HI values at lower UG . Still, when compared to other research studies, the results obtained with sinter plate are fairly respectable. Thus, based on this, a conclusion can be drawn that self-agitated impellers are quite successful in enhancing hydrodynamics not only with single orifice, but also with sinter plate used as sparger. 4. Conclusions This paper reports the effect of a novel type of internal, selfagitated impellers, on the main hydrodynamic characteristics (riser gas holdup and downcomer liquid velocity) in an external-loop airlift reactor. The insertion of self-agitated impellers in the riser section of EL-ALR caused considerable enhancement of riser gas

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

ARTICLE IN PRESS

JID: JTICE

[m5G;September 14, 2016;4:40]

N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

10

Fig. 11. Comparison of our results with previous research studies regarding internals in external-loop airlift reactors with water as a liquid phase.

holdup due to bubble breakage on the one hand, and increased turbulence, which led to a decrease in coalescence, on the other hand. The highest improvements, in the range 20–47%, were obtained with single orifice at lower superficial gas velocities. At higher gas velocities, i.e. heterogeneous regime, the type of distributor had no influence on ε GR improvement in all liquid phases. Besides accomplishment of obvious goals, smaller bubble size and higher riser gas holdup, the employment of self-agitated impellers was also motivated by a demand to limit the resistance to the flow. Reduction of downcomer liquid velocity was only about 10% in all investigated cases when impellers were used. Therefore, this novel type of impellers can be advantageously used to enhance hydrodynamics and ultimately mass transfer, as essential factors for EL-ALR performance, since they simultaneously caused bubble size decrease, higher riser gas holdup and modest reduction in liquid velocity, while requiring negligible energy for operation. Acknowledgments This research was financially supported by the Ministry of Education, Science and Technological Development of the Republic of Serbia (Project no. 172025). References [1] Chisti Y. Pneumatically agitated bioreactors in industrial and environmental bioprocessing: hydrodynamics, hydraulics, and transport phenomena. Appl Mech Rev 1998;51:33–112. [2] Merchuk JC, Glutz M. Bioreactors: air-lift reactors. In: Flickinger MC, Drew SW, editors. Encyclopedia of bioprocess technology: fermentation, biocatalysis, and bioseperations. New York City, New York: John Wiley & Sons, Inc.; 1999. p. 320–53. [3] Luo L, Yuan J, Xie P, Sun J, Guo W. Hydrodynamics and mass transfer characteristics in an internal loop airlift reactor with sieve plates. Chem Eng Res Des 2013;91:2377–88. [4] Zhang T, Wang J, Wang T, Lin J, Jin Y. Effect of internal on the hydrodynamics in external-loop airlift reactors. Chem Eng Process Process Intensif 2005;44:81–7. [5] Wu Q , Wang X, Wang T, Han M, Sha Z, Wang J. Effect of liquid viscosity on hydrodynamics and bubble behaviour of an external-loop airlift reactor. Can J Chem Eng 2013;91:1957–63. [6] Al-Masry WA. Gas holdup in circulating bubble columns with pseudoplastic liquids. Chem Eng Technol 2001;24:71–6. [7] Zhao M, Niranjan K, Davidson JF. Mass transfer to viscous liquids in bubble columns and air-lift reactors: influence of baffles. Chem Eng Sci 1994;49:2359–69.

[8] Miyahara T, Nagatani N. Influence of alcohol addition on liquid-phase volumetric mass transfer coefficient in an external-loop airlift reactor with a porous plate. J Chem Eng Jpn 2009;42:713–19. [9] Kojic´ PS, Tokic´ MS, Šijacˇ ki IM, Lukic´ NL, Petrovic´ DL, Jovicˇ evic´ DZ, et al. Influence of the sparger type and added alcohol on the gas holdup of an external-loop airlift reactor. Chem Eng Technol 2015;38:701–8. [10] Snape JB, Zahradník J, Fialová M, Thomas NH. Liquid-phase properties and sparger design effects in an external-loop airlift reactor. Chem Eng Sci 1995;50:3175–86. [11] Chisti Y. Airlift reactors–design and diversity. Chem Eng 1989;457:41–3. [12] Lin CH, Fang BS, Wu CS, Fang HY, Kuo TF, Hu CY. Oxygen transfer and mixing in a tower cycling fermentor. Biotechnol Bioeng 1976;18:1557–72. [13] Vorapongsathorn T, Wongsuchoto P, Pavasant P. Performance of airlift contactors with baffles. Chem Eng J 2001;84:551–6. [14] Krichnavaruk S, Pavasant P. Analysis of gas–liquid mass transfer in an airlift contactor with perforated plates. Chem Eng J 2002;89:203–11. [15] Goto S, Gaspillo PD. The effect of static mixer on mass transfer in draft tube bubble column and in external loop column. Chem Eng Sci 1992;47:3533–9. [16] Chisti Y, Kasper M, Moo-Young M. Mass transfer in external-loop airlift bioreactors using static mixers. Can J Chem Eng 1990;68:45–50. [17] Gavrilescu M, Roman RV, Tudose RZ. Hydrodynamics in external-loop airlift bioreactors with static mixers. Bioprocess Biosyst Eng 1997;16:93–9. [18] Chisti Y, Jauregui-Haza UJ. Oxygen transfer and mixing in mechanically agitated airlift bioreactors. Biochem Eng J 2002;10:143–53. [19] de Jesus SS, Moreira Neto J, Santana A, Maciel Filho R. Influence of impeller type on hydrodynamics and gas-liquid mass-transfer in stirred airlift bioreactor. AlChE J 2015;61:3159–71. [20] Gibbs PA, Seviour RJ. The production of exopolysaccharides by Aureobasidium pullulans in fermenters with low-shear configurations. Appl Microbiol Biotechnol 1998;49:168–74. [21] Hamood-ur-Rehman M, Dahman Y, Ein-Mozaffari F. Investigation of mixing characteristics in a packed-bed external-loop airlift bioreactor using tomography images. Chem Eng J 2012;213:50–61. [22] Nikakhtari H, Hill GA. Hydrodynamic and oxygen mass transfer in an external-loop airlift bioreactor with a packed bed. Biochem Eng J 2005;27:138–45. [23] Meng AX, Hill GA, Dalai AK. Hydrodynamic characteristics in an external-loop airlift bioreactor containing a spinning sparger and a packed bed. Ind Eng Chem Res 2002;41:2124–8. [24] Mohanty K, Das D, Biswas MN. Hydrodynamics of a novel multi-stage external-loop airlift reactor. Chem Eng Sci 2006;61:4617–24. [25] Zhang T, Wei C, Feng C, Zhu J. A novel airlift reactor enhanced by funnel internals and hydrodynamics prediction by the CFD method. Bioresour Technol 2012;104:600–7. [26] Nikakhtari H, Hill GA. Enhanced oxygen mass transfer in an external-loop airlift bioreactor using a packed bed. Ind Eng Chem Res 2005;44:1067–72. [27] Hamood-ur-Rehman M, Ein-Mozaffari F, Dahman Y. Dynamic and local gas holdup studies in external-loop recirculating airlift reactor with two rolls of fiberglass packing using electrical resistance tomography. J Chem Technol Biotechnol 2013;88:887–96. [28] Pollard DJ, Ison AP, Shamlou PA, Lilly MD. Reactor heterogeneity with Saccharopolyspora erythraea airlift fermentations. Biotechnol Bioeng 1998;58:453–63.

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003

JID: JTICE

ARTICLE IN PRESS N.Lj. Luki´c et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–11

[29] Tekic´ MN, Šijacˇ ki IM, Tokic´ MS, Kojic´ PS, Petrovic´ DL, Lukic´ NL, et al. Hydrodynamics of a self-agitated draft tube airlift reactor. Chem Ind Chem Eng Q 2014;20:59–69. [30] Akita K, Nakanishi O, Tsuchiya K. Turn-around energy losses in an external-loop airlift reactor. Chem Eng Sci 1994;49:2521–33. [31] Verlaan P. Modelling and characterization of an airlift-loop bioreactor. Wageningen: Agricultural University; 1987. [32] Keitel G, Onken U. Errors in the determination of mass transfer in gas–liquid dispersions. Chem Eng Sci 1981;36:1927–32. [33] Kojic´ PS, Šijacˇ ki IM, Lukic´ NL, Jovicˇ evic´ DZ, Popovic´ SS, Petrovic´ DL. Volumetric gas–liquid mass transfer coefficient in an external-loop airlift reactor with inserted membrane. Chem Ind Chem Eng Q 2016;22:275–84. [34] Bajón Fernández Y, Cartmell E, Soares A, McAdam E, Vale P, Darche-Dugaret C, et al. Gas to liquid mass transfer in rheologically complex fluids. Chem Eng J 2015;273:656–67.

[m5G;September 14, 2016;4:40] 11

[35] Philip J, Proctor JM, Niranjan K, Davidson JF. Gas hold-up and liquid circulation in internal loop reactors containing highly viscous Newtonian and non-Newtonian liquids. Chem Eng Sci 1990;45:651–64. [36] Maretto C, Krishna R. Modelling of a bubble column slurry reactor for Fischer–Tropsch synthesis. Catal Today 1999;52:279–89. [37] Miyahara T, Hamanaka H, Umeda T, Akagi Y. Effect of plate geometry on characteristics of fluid flow and mass transfer in external-loop airlift bubble column. J Chem Eng Jpn 1999;32:689–95. [38] Cao C, Dong S, Geng Q, Guo Q. Hydrodynamics and axial dispersion in a gas−liquid−(solid) EL-ALR with different sparger designs. Ind Eng Chem Res 20 08;47:40 08–17.

Please cite this article as: N.Lj. Lukic´ et al., Enhanced hydrodynamics in a novel external-loop airlift reactor with self-agitated impellers, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2016.09.003