- Email: [email protected]
Hydrodynamics and pressure loss of concurrent gas–liquid downward flow through sieve plate packing
Chemical Engineering Science 143 (2016) 206–215
Contents lists available at ScienceDirect
Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces
Hydrodynamics and pressure loss of concurrent gas–liquid downward flow through sieve plate packing Wangde Shi, Weixing Huang, Yuhan Zhou, Huan Chen, Dawei Pan School of Chemical Engineering, Sichuan University, Chengdu 610065, China
H I G H L I G H T S
Four flow patterns (trickling, continuous, semi- and perfect-dispersed) were found. Pulsing flow regime was observed and analyzed, and the flow maps were presented. Mechanism of liquid dispersion and its effect on flow resistance were described. New correlations for the pressure drop of gas–liquid two phase flow were developed.
art ic l e i nf o
a b s t r a c t
Article history: Received 15 September 2015 Received in revised form 14 December 2015 Accepted 5 January 2016 Available online 12 January 2016
Gas–liquid concurrent downward flow through a new structured packing was investigated by experiment systematically. The experimental packing consisted of 21 sieve plates. Each plate was 190 mm 190 mm in side length and 1 mm in thickness. Three sets of packing with different sizes of sieve holes were tested. During experiment, four flow states have been observed, i.e., the trickling flow at low gas and liquid flux, the continuous flow at low gas but high liquid flux, the semi-dispersed flow at high gas flux and the completely dispersed flow as gas flux increases further. Another important phenomenon observed is the occurrence of pulsing flow, i.e., in some range of gas and liquid flow fluxes, both two phases will no longer flow smoothly through the sieve plates but flow downward in pulse regularly. Through extensive experiment, the rough boundaries for flow regime transition were obtained. Then, the pressure losses of gas flow and gas–liquid two phase flow in non-pulsing flow regime were systematically measured. The analysis for pressure differences measured at different locations show that for a given packing, the pressure loss through each plate is nearly the same. Its magnitude depends on the gas and liquid flow rates. The comparisons between pressure losses of different packings indicate that the pressure loss is associated with the hole diameter, hole pitch, free area ratio of sieve plate and the mounting distance between two adjacent plates. Finally, considering these factors, the correlations for pressure drops were developed, which approximates the experimental values with an average deviation less than 10%. & 2016 Elsevier Ltd. All rights reserved.
Keywords: Sieve plate packing Gas–liquid concurrent downflow Flow patterns Pulsing flow Pressure loss
1. Introduction Packed columns are widely used as separation equipment for many years and numerous kinds of packing (both random and structured) have been developed. Of the flow modes identified in packed columns, the countercurrent and concurrent flow are most commonly used in the operation of commercial separation units. Nowadays, a new structured packing consisting of multiple sieve plates has been developed in packed columns operated with gas– liquid concurrent downward flow mode, e.g., the gas stripper in nuclear plants. This new packing with concurrent flow can also be preferred in a wide range of other industrial situations, such as chemical, petrochemical and biochemical industries, etc, especially for mass transfer process when there is no significant difference in http://dx.doi.org/10.1016/j.ces.2016.01.005 0009-2509/& 2016 Elsevier Ltd. All rights reserved.
the mean concentration driving force offered by two modes (Beg et al., 1996). In such a condition, this new packing with concurrent mode has some marked advantages in two aspects. The first is in the excellent hydrodynamic performance as compared with countercurrent mode, i.e., high throughputs, relatively low pressure loss and absence of flooding (Beg et al., 1996; Saroha and Khera, 2006; Babu et al., 2006; He et al., 2012), and the second is in the simple geometry of the sieve plate packing for manufacture and installment. However, the design of this new structured packed column still relies on experience or simulation so far and the design result is dubious. To provide reliable guidelines for the design, the understanding to the hydrodynamic behaviors of gas–liquid flow in this new packing is badly needed. At present, most of the relevant
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
207
studies about concurrent flow are focused on random-packed columns, for example, those of Hutton and Leung (1974), Rao and Drinkengurg (1985), Benkrid et al. (1997), Nemec et al. (2001) and Babu et al. (2006) for hydrodynamics, Shende and Sharma (1974) and Mahajani and Sharma (1980) for mass transfer, and Taulamet et al. (2014) for heat transfer. A few previous studies are on structured packed columns, for example, those of Raynal et al. (2004) for metal corrugated sheet structured packing and Schildhauer et al. (2012) for foams and knitted wire packing. Whether for random or structured packing, knowing the hydrodynamic performance is the basis for the investigation of mass or heat transfer. Although those previous efforts have made some contributions to understand the hydrodynamic performance of gas– liquid concurrent flow in packed columns, a lot of research work is still needed to design this new kind of columns precisely and to develop more efficient sieve plate packing. The aim of the present study is to investigate the flow mechanism and interaction between gas and liquid when they flow concurrently downward through the sieve plate packing, as well as the effect of packing structures, based on systematical experiments, and then to develop practical correlations for the calculation of pressure loss accordingly, so that the guidelines can be provided for the hydrodynamic design and the practical operation of commercial columns.
2. Experimental apparatus and process The experimental arrangement is shown in Fig. 1. The experimental column consists of a sieve plate packing and a rectangular Perspex shell with dimensions of 200 mm 200 mm 600 mm, so the flow development in the packing can be observed visually and recorded by a camera. The packing is composed by 21 pieces of stainless steel square plate with 190 mm 190 mm in side length and 1mm in thickness as shown in Fig. 2. Three sets of packing are tested and their geometrical parameters, as listed in Table 1, are designed by reference to those of commercial unit. Four pressure taps are amounted along the height of the test packings to measure the pressure losses of gas–liquid flow from the top plate to the 7th, the 14th and the 21st plate, respectively, by three U-tube manometers. The minimum scale of U-tube manometer is 1 mmH2O. In order to insure the precision, the pressure loss through 21 plates is averaged as the pressure loss of T 7
T 1
11 10 10
10
10
2 9 4 T
9
9
one plate. The range of the measured pressure difference in the experiment is from 4 to 470 mmH2O, and the mean error for manometer readings is about 0.84%. All tests were carried out by using air and water at room temperature (about 20 °C) and common pressure (1 atm). The range of air flow rate for pure gas phase experiment was 20– 250 m3/h. During two phase experiment, the range of air flow rate for packing A was 20–230 m3/h while the range of air flow rate for both packing B and packing C was 20–210 m3/h. The range of water flow rate for each test packing was 0.4–1.6 m3/h. The minimum scale of gas flow-meter is 2 m3/h and in the range of gas flowrate tested, the mean error for gas flow-meter readings is about 1.5%. The minimum scale of liquid flow-meter is 40 L/h and in the range of liquid flowrate tested, the mean error for liquid flow-meter readings is about 4%. In the experiment, two phases flow concurrently downward through the test packing. Air was sent into the top of the column by a blower and its flow rate was controlled by regulating valves and monitored by rotameters. Water was sent to the liquid distributor mounted at the top of the column under the static pressure provided by an overhead surge tank, and then flowed downward through the test packing by gravity.
9
3 9
Fig. 2. Packing structure.
9
9
3. Experimental result and discussion 3.1. Flow state of gas/liquid through the packing
9 8 5
1.Gas-liquid distributor 2.Test section 3.Gas-liquid separator 4.U-tube manometer
5.Water storage tank 6.Pump 7.Water surge tank 8.Blower
6
9.Regulating valve 10.Rotameter 11.Silencer
Fig. 1. Sketch of experimental setup and air–water flow loop.
For concurrent downward flow of gas and liquid through the sieve plate packing, four typical states are observed, i.e., liquid trickling, continuous flow, semi-dispersed flow and completely dispersed flow. Experiment indicates that the occurrence of these four states depends on the magnitude of gas and liquid flux through the plate hole, the interaction between two phases, and the interfacial effect between liquid and plate.
208
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
Table 1 Geometrical characteristics of three test packings. Packing label
packing A
packing B
packing C
d (mm) Hole quantity m Plate quantity Free area ratio α l (mm) b (mm) e (mm)
14 30 21 0.1154 30 25 1
10 56 21 0.1099 20 18 1
6 132 21 0.0933 14 11 1
Fig. 3. Trickling flow at low gas and liquid flux (packing A, WL ¼ 24.0 kg/m2s). (a) WG ¼ 0 kg/m2s. (b) WG ¼2.9 kg/m2s. (c) corresponding diagram.
Fig. 4. Continuous flow at low gas flux but high liquid flux (packing A, WL ¼84.3 kg/m2s). (a) WG ¼3.6 kg/m2s. (b) corresponding diagram.
At low gas and liquid flux, the flow of liquid will be in a form of trickling, as shown in Fig. 3. In this case, the interaction between two phases is weak and the momentum of liquid is small, so that the interfacial effect between liquid and plate becomes relatively important, leading to liquid flowing along the back surface of the plate and finally dropping off to the next plate. This flow pattern is so called trickling flow. When the liquid flux is increasingly enhanced under low gas flux, the momentum of liquid will become strong enough compared to gas–liquid interaction and liquid-plate interfacial effect. In this case, the flow of liquid through the plate hole will evolve into a continuous state, i.e., the liquid from each hole will flow downward in a nearly continuous annular film, as shown in Fig. 4. Contrary to the case of low gas flux, when the gas flux becomes high enough so that the interaction between two phases becomes
a dominating factor, the flow of liquid through the plate hole will evolve into a semi-dispersed state, as shown in Fig. 5. The semidispersed state always occurs at relatively high gas flux, whether the liquid flux is high or low. In this flow state, one part of the liquid will be dispersed into droplets due to the dragging of gas stream, while the other part of the liquid will move along the back surface of the plate because of the wrapping of gas stream and then drop into the holes on the next plate. Of those dispersed droplets, most will be deposited on the surface of the plate downstream, while the rest will be carried away with gas stream or brought to the back surface of plate. In the semi-dispersed state, the liquid holdup on the space and the gas–liquid interaction increase apparently. From the view of mass transfer enhancement, the semi-dispersed flow pattern is preferred because it offers high interfacial area and frequent renewal of gas–liquid contacting
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
209
Fig. 5. Semi-dispersed flow at high gas flux (packing A, WG ¼10.1 kg/m2s). (a) WL ¼ 24.0 kg/m2s. (b) WL ¼ 84.3 kg/m2s. (c) corresponding diagram.
WG=16.0 kg/m2s, WL=84.3kg/m2s, packing A
corresponding diagram
Fig. 6. Flow pattern at completely dispersed state. (a) WG ¼ 16.0 kg/m2s, WL ¼ 84.3 kg/m2s, packing A. (b) Corresponding diagram.
surface. For this reason, commercial units are commonly operated in this state. As expected, when gas flux increases continually, more and more liquid will be dispersed into droplets and the flow of liquid will finally reach to a completely dispersed state as shown in Fig. 6. This state may have higher gas–liquid contacting efficiency, but the pressure loss will be increased apparently. 3.2. Pulsing flow regime and flow maps Another important flow phenomenon observed is the occurrence of pulsing flow, i.e., in some range of gas and liquid flow fluxes, the gas–liquid will not flow smoothly but instead rush alternately through each plate. Meanwhile, the readings of U-tube manometers are pulsating regularly. Pulsing flow has also been found in other kinds of packing operated in gas–liquid concurrent downward flow mode, such as in the trickle bed (Benkrid et al., 1997; Tsochatzidis et al., 2002; Al-Naimi et al., 2011; Salgi et al., 2015). The cause leading to pulsing is analyzed as follows. As gas and liquid concurrently flow downward through the plate hole, each phase will occupy a corresponding flow area, in which the flow area of liquid relies on the liquid flux (defined as the ratio of flowrate to hole area), the drag of gas stream, and the resistance of hole rim as well. At relatively low gas flux, the drag of gas phase is weak and the liquid flow area under given liquid flux has no apparent change as gas flux increases. At this situation, once the gas flux reaches to a limitation where the necessary area for the smooth flow of gas phase exceeds its actual flow area, a part of gas
will be plugged and gas pressure will increase transiently until the gas pushes the liquid through the hole, leading to the occurrence of pulsing. On the other hand, when the gas flux becomes sufficiently high, the drag of gas stream will become strong enough to expand the flow area for itself and the pulsing flow will disappear. By combining the observation results of the pulsing regime and the flow states with the corresponding operation conditions, the flow maps for packings A, B and C are presented in Figs. 7–9. For packing A, as shown Fig. 7, the pulsing has not been observed in whole range of liquid flux examined, so that only flow state transition lines are presented. It should be pointed that in the experiments, including for packings B and C, the transition from trickling to continuous flow is observed to happen at WL ¼51– 63 kg/m2 s, so that the transition line at WL ¼60 kg/m2 s in Fig. 7, as well as in Figs. 8 and 9, is just a rough line. Further observation on flow phenomenon in packing A confirms that the transition boundary to the semi-dispersed flow varies from WG ¼4.33 kg/ m2 s to WG ¼ 7.94 kg/m2 s with the increase of liquid flux, giving a rough line of transition as shown in Fig. 7. Under the flow fluxes examined, the pulsing flow regime is observed in packings B and C, as shown in Figs. 8 and 9. From the flow maps of packings B and C, it can be seen that the variation of the lower or upper limit line of the pulsing region with liquid flux has the same trend for both packing B and packing C, i.e., the lower limit line goes down and the upper limit line goes up with increasing liquid flux. Obviously, the lower limit line represents the start points of pulsing flow. According to the mechanism for pulsing generation described above, the lower liquid flux will leave more flow area for gas phase and thus the corresponding gas
210
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 15
5000
14 13
completely dispersed flow
12
7 layers 14 layers 21 layers packing A
4000
11 3000
9
semi-dispersed flow
8
ΔP (Pa)
W (kg/m /s)
10
7 6
2000
5 4 3
continuous flow
trickling flow
2
1000
1 0
0 20
30
40
50
60
70
80
90
20
100
40
60
80
100
15
160
180
200
220
240
220
14 13
7 layers 14 layers 21 layers
200
completely dispersed flow
12
180
11
packing A
160
10
140
9 8
semi-dispersed flow
ΔP (Pa)
W (kg/m /s)
140
Fig. 10. Effect of plate quantity on pressure loss of single phase (gas) flow.
Fig. 7. Flow map for packing A.
7 6
120 100 80
5
pulsing flow
4
60
3
trikling flow
2
40
continuous flow
20
1 0 20
30
40
50
60
70
80
90
100
110
W (kg/m /s)
14
semi-dispersed flow 8
pulsing flow
4
trickling flow continuous flow 20
40
60
60
80
100
120
140
160
180
200
220
240
3.3. Pressure loss of single phase (gas) flow
10
0
40
flow smoothly through the hole, the pulsing will not occur, as the case in Fig. 7 for packing A and the case at low liquid flux region in Fig. 8 for packing B.
12
2
20
Fig. 11. Average pressure loss of single phase (gas) flow through 7, 14 and 21 plates.
completely dispersed flow
6
0
Q (m /h)
Fig. 8. Flow map for packing B.
W (kg/m /s)
120
Q (m /h)
W (kg/m /s)
80
100
W (kg/m /s) Fig. 9. Flow map for packing C.
flux to produce pulsing will become higher. The upper limit line represents the end points of pulsing flow. According to the mechanism for pulsing elimination, when liquid flux becomes higher, a larger gas flux will be required to expand enough flow area and eliminate pulsing, thus the upper limit line goes up with the increasing liquid flux. It should be pointed that when the hole size and the free area of the plate relative to liquid flux are large enough to allow the gas
As an extreme case of two phase flow, the pressure loss of single phase (gas) flow through the packing are measured. The pressure losses of gas flow through 7, 14 and 21 plates in packing A are presented in Fig. 10 and it is shown that the pressure losses vary with gas flow rate and plate quantity. For comparison, the average pressure losses per each plate are calculated by dividing the total pressure loss through 7, 14 and 21 plates with corresponding plate quantity and the results are given in Fig. 11. It shows that the average pressure loss through each plate is nearly the same, indicating that the pressure loss of single phase (gas) flow increases linearly with the quantity of sieve plates. Similar results are found for packings B and C. Therefore the average pressure loss for one plate can be used as a basis to investigate the resistance behaviors of single phase (gas) flow through different packings. For simplicity, the average pressure loss of single phase (gas) flow through one plate will be termed as the dry-plate pressure loss in the following discussion. The variation of dry-plate pressure loss with gas flow rate for packings A, B and C is shown in Fig. 12. From the figure it is not difficult to find that the geometrical parameters of the packing have a complex effect on the pressure loss. Generally, the hole diameter d and the free area α may be two important affecting
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 300
211
220
packing A, d=14mm,α =0.1154, d/b=0.560, l/b =1.20 packing B, d=10mm,α =0.1099, d/b=0.556, l/b=1.11 packing C, d= 6 mm,α =0.0933, d/b=0.545, l/b =1.27
250
180
7 layers 14 layers 21 layers
160
packing A, Q =0.8m /h
200
200
ΔP (Pa)
Δ P (Pa)
140 150
120 100 80
100
60 50
40 20
0 0
20
40
60
80
100
120
140
160
180
200
220
240
0
260
20
40
60
80
100
Q (m /h)
120
140
160
180
200
220
240
Q (m /h)
Fig. 12. Variation of dry-plate pressure loss with gas flow rate for different packings.
Fig. 14. Average pressure loss of two phase flow through 7, 14 and 21 plates.
5000
240 220
4000
200
7 layers 14 layers 21 layers
180 160
packing A, Q =0.8m /h
ΔP (Pa)
3000
ΔP (Pa)
140 2000
120 100 80
1000
60 40 20
0 20
40
60
80
100
120
140
160
180
200
220
240
Q (m /h) Fig. 13. Effect of plate quantity on pressure loss of two phase flow.
factors, which account for the effect of the cross-section area on flow resistance. For example, the d and α of packing C are apparently the smallest compared with those of packings A and B, thus packing C has the largest pressure loss at a given gas flow rate. However, when these two parameters become relatively large, the effects of the ratios d/b and l/b may become apparent, in which d/b accounts for the effect of energy loss caused by the impact of gas stream on the next plate, while l/b represents the effect of energy loss caused by the abrupt change of gas flow direction. For example, although the packing A has the largest d andα, its pressure loss is higher than that of packing B, out of common expectation. Comparing the geometrical parameters of packings A and B, it can be seen that both packings have nearly equal free area and relatively large holes, but the l/b of packing A is apparently larger than that of packings B. This means that the flow resistances through the hole for packings A and B are comparable, but the energy loss caused by abrupt change of gas flow direction in packing A is apparently larger, so that packing A has a higher pressure loss than packing B. 3.4. Pressure loss of two phase flow For two phase flow through the packings, the pressure loss and its variation with gas flow rate under different liquid flow rates have many similarities to the case of single phase (gas) flow. Fig. 13 shows the variation of pressure losses of two phase flow through 7, 14 and 21 plates with gas flowrate under QL ¼0.8 m3/h in packing
0 20
40
60
80
100
120
140
160
180
200
220
240
Q (m /h) Fig. 15. Variation of two phase pressure loss for one plate with gas/liquid flow rate (packing A).
A and Fig. 14 shows the average pressure losses per each plate calculated by dividing the total pressure loss through 7, 14 and 21 plates with corresponding plate quantity. As shown in Fig. 14, for different total pressure losses, their average pressure losses through each plate are also nearly the same, similar to the case of single gas flow, indicating that the pressure loss of two phase flow also increases linearly with the quantity of sieve plates. Similar results are found for packings B and C. For this reason, the following discussion will be focused on the average pressure loss for one plate. To compare the effect of liquid flowrate, the variations of pressure loss with gas flowrate under different liquid flowrates are presented in Figs. 15, 16 and 17 for packings A, B and C, respectively. These figures indicate that the higher liquid flowrate generally corresponds to the higher pressure loss under the given gas flowrate for a given packing. Examining the pressure loss through packing A as shown in Fig. 15, it is found that as gas flowrate increases, the effect of liquid flowrate becomes steadily increased, i.e., the interval of variation of the pressure loss over the range of liquid flowrate increases steadily with gas flowrate. Further, when the gas flow rate QG reaches 110 m3/h, the interval of pressure loss variation will reach a nearly constant size. Considering the fact that the transition boundary to the semi-dispersed flow for
212
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 200 180 160 140
ΔP (Pa)
120 100 80 60 40 20 80
100
120
140
160
180
200
220
Q (m /h) Fig. 16. Variation of two phase pressure loss for one plate with gas/liquid flow rate (packing B).
300 280 260 240
ΔP (Pa)
220 200 180 160 140 120 100 80 140
160
180
200
220
Q (m /h)
Fig. 17. Variation of two phase pressure loss for one plate with gas/liquid flow rate (packing C).
packing A varies from QG ¼60 m3/h to QG ¼110 m3/h, it can be concluded that when the flow patterns at different liquid flow rates all have reached a semi-dispersed state, the interval of variation of the pressure loss over the range of liquid flow rate will reach a limiting size. This conclusion also can be obtained from Figs. 16 and 17 for packings B and C, in which all pressure loss data are measured in the region of the semi-dispersed flow. Fig. 18(a)–(c) presents the effects of geometrical parameters on the pressure loss of two phase flow at typical liquid flowrates, i.e., QL ¼ 0.4 m3/h, QL ¼0.8 m3/h and QL ¼1.2 m3/h. These three Figures show that under a given gas flowrate, the packing C has the largest pressure loss, the packing B has the lowest pressure loss, and the pressure loss of packing A is in between, no matter the liquid flow rate is low or high. This result, similar to the case of single gas flow, further confirms that the hole diameter d and the free area α are two important affecting factors on pressure loss, but when these two parameters become relatively large, the effects of the ratios d/ b and l/b may become apparent. This result also implies that the packing geometrical characteristics have some similar influence on both single phase (gas) and two phase pressure loss.
Fig. 18. (a). Effect of packing geometrical characteristics on two phase pressure loss at QL ¼0.4 m3/h. (b). Effect of packing geometrical characteristics on two phase pressure loss at QL ¼0.8 m3/h. (c). Effect of packing geometrical characteristics on two phase pressure loss at QL ¼ 1.2 m3/h.
4. Resistance coefficients for sieve plate packing 4.1. Resistance coefficient for dry plate For the fluid flow through a sieve plate, the flow mode through every hole can be regarded as the same and, therefore, the pressure loss through one hole will represent that through the sieve plate. Not like the pressure loss of fluid flowing in a pipe where the pressure loss is caused by wall friction, the pressure loss through the hole belongs to local energy loss, mainly caused by the
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
contraction and expansion of fluid stream, the friction of hole rim to the fluid, the abrupt change of flow direction and the impact of fluid stream on the plate downstream. According to this, the pressure loss for single gas flow through the sieve plate (termed as dry plate in the following discussion) can be expressed as
ΔP ¼ ξG
W 2G W 2G ¼ ξ0 þ ξ1 2 ρG 2 ρG
ð1Þ
where WG is the mass flux of gas flow through the hole, ρG is the density of gas, ξ0 is the resistance coefficient for impacting pressure loss, and ξ1 is the resistance coefficient through the dry plate corresponding to the effects of other factors, i.e., the fluid stream contraction and expansion, the friction of hole rim and the abrupt change of flow direction. For the calculation of ξ0, Minghui et al. (2014) has proposed a semi-empirical equation as follows
ξ0 ¼ 1:17
2
d
4b
ð2Þ
2
where d is the diameter of the hole, and b is the distance between two adjacent plates. For ξ1, many similar researches are referenced and analyzed in this work. The studies for the resistance coefficient through an orifice (Alimonti et al., 2010) or a perforated plate (Kolodzie and Winkle, 1957; Fried. and Idelchik., 1989; Gan and Riffat, 1997; Malavasi et al., 2012) show that the effect of geometrical parameters on resistance coefficient is associated with Reynolds number defined as Q ρ Re ¼ 4 G G mπ dμG
ð3Þ
where, QG is the total volume flow rate, m is the hole number in one plate and μG is the viscosity of gas. According to Eqs. (1) and (3), the Re and corresponding ξ1 are calculated based on experimental data, as shown in Fig. 19. From Fig. 19, it can be seen that when the Reynolds number Re is relatively low, about in the range of Rer 4500, the ξ1 is in a descending trend with increasing Re, while for Re44500, the ξ1 seems independent of Re and the magnitude of ξ1 is only related to the geometrical parameters of packings. For this reason, the data of resistance coefficient ξ1 are correlated, respectively, for two regions divided by Re¼4500. Next, the comparison between the flow mechanisms of fluid through a single perforated plate with that through multiple sieve plates is carried out in order to determine the relevant geometrical parameters affecting ξ1. As shown in Fig. 20, for the flow through multiple sieve plates, the subsequent plate will lead to abrupt
turning of flow stream and generating more vortexes, so that the abrupt change of flow direction must be included in ξ1 (Given that the impact has been considered in ξ0). Since the turning angle β, see Fig. 21, is related with the ratio of l/b, the l/b will be included in ξ1. According to the considerations above and fitting all experimental data, the correlations for resistance coefficient ξ1 are developed as follows qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4:851 2 4:851 1 þ l=2b secβ ξ1 ¼ 0:292 ¼ 0:292 ðRe 4 4500Þ ð1 αÞ7:2765 ð1 αÞ7:2765 ð4Þ
ξ1 ¼ 1:898
2.0
packing A packing B packing C
ξ
1.8
1.6
1.4
1.2 0
2000
4000
6000
8000
10000
12000
14000
Re Fig. 19. The variation of dry-plate resistance coefficient with Reynolds number based on one hole.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4:851 2 1 þ l=2b ð1 αÞ7:2765
Re 0:22
ðRe r 4500Þ
ð5Þ
The tested ranges of the parameters in Eqs. (4) and (5) are 1156o Re o13913, 0.0933o α o0.1154 and 1.1 ol/bo1.27, respectively. Both Eqs. (4) and (5) approximate most of the experimental values with the deviation of less than 10%, as shown in Fig. 22. 4.2. Resistance coefficient for two phase flow in semi-dispersed region The pressure loss ΔP of two phase flow through the sieve plate still belongs to local energy loss, so that
ΔP ¼ ξ m
W 2m 2 ρm
ð6Þ
In Eq. (6), ξm is the resistance coefficient of two phase flow through the sieve plate, Wm is the total mass flux of the gas–liquid mixture through the hole, and ρm is a nominal density of the gas– liquid two phase fluid when it is treated as homogenous mixture, (Schmidt and Friedel, 1997) defined as
ρm ¼
LþG L=ρL þ G=ρG
ð7Þ
where G ¼ ρGQG and L ¼ ρLQL are the mass flow rate of gas and liquid phase, respectively. As mentioned in Section 3.4, for two phase flow under given liquid flow rate, the variation of pressure loss with gas flow rate and the effect of geometrical parameters are similar to the case of single gas flow. For this reason, the resistance coefficient ξm for two phase flow will be treated as the product of dry-plate coefficient ξG times a modifying factor C, i.e.,
ξm ¼ C ξG
2.2
213
ð8Þ
As a modifying factor for resistance coefficient of two phase flow, the C should be associated with the dispersing process of liquid, the relative magnitude of gas and liquid flow rates and the geometrical parameters of packings. Since the dispersing of liquid is a result of the interaction between flowing gas and liquid, i.e., to tear liquid into fragments and droplets the shear stress acting on the liquid surface by gas must overcome the surface tension of liquid. Thus, the Weber number should be an important parameter affecting C, as suggested by numerous open literatures on liquid dispersion (Fischer, 1995; Nigmatulin et al., 1996; King and Piar, 1999; Berna et al., 2015). Also, lots of prior studies on concurrent gas–liquid core/annular flow in pipes indicate that interfacial shear stress relates to gas and liquid Reynolds numbers (Chung and Mills, 1974; Tishkoff et al., 1979) and many skin friction coefficients have been successfully correlated by superficial gas and liquid Reynolds numbers (Chien and Ibele, 1964; Henstock and Hanraty, 1976; Fedotkin et al., 1979; Asali et al., 1985; Hajiloo et al., 2001).
214
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
a single perforated plate
multiple sieve plates
Fig. 20. Flow through a thin perforated plate and multiple sieve plates. (a) A single perforated plate. (b) Multiple sieve plates.
l
Table 2 The empirical correlations for two-phase pressure loss in semi-dispersed flow region.
β
l /2
b
Pressure loss Dry-plate resistance coefficient
Fig. 21. The turning angle of flow path.
! 2 W 2m d W 2m ¼ C ξ1 þ 1:17 2 2ρm 4b 2ρm
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4:851 2 1 þ l=2b
For ReG r 4500
ξ1 ¼ 1:898
For ReG 4 4500
2.2
Modifying factor for two phase flow 2.0
Dimensionless parameter
1.8
ξ Calculated
ΔP ¼ ξm
7:2765
(17) ReG 0:22
ð1 α Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2ffi 4:851 1 þ l=2b
ξ1 ¼ 0:292 7:2765 ð1 !αÞ1 2 0:128 d Re0:72 L þ1 C ¼ 1 þ 1:1 l α ðReG WeÞ0:36 4L ReL ¼ mπdμL 4G ReG ¼ mπdμG 16G2 We ¼ 3 m2 π 2 ρ G d σ
(4)
(5) (16) (18) (19) (20)
1.6
packing A Re<4500 packing A Re>4500 packing B Re<4500 packing B Re>4500 packing C Re<4500 packing C Re>4500
1.4
1.2 1.2
1.4
1.6
1.8
2.0
2.2
ξ Experimental
Fig. 22. Parity plot for experimental and calculated dry-plate resistance coefficient.
Moreover, for two phase flow through the sieve plate hole, the additional effects of the flow area, the hole pitch and the distance between plates on the liquid dispersing and pressure loss must be considered. Therefore, given that the effect of plate distance b has been considered in ξG, the modifying factor C will be a function of Weber number, gas and liquid Reynolds numbers, α and d/l, i.e., ð9Þ C ¼ f 1 We; ReL ; ReG ; α; d=l where We, ReL and ReG are defined as We ¼ ReG ¼
ReL ¼
ρ
2 G uG d
ð10Þ
σ
duG ρG
μG
duL ρL
ð11Þ
ð12Þ
μL
In Eqs. (10)–(12), the uG and uL are superficial gas and liquid velocities defined as uL ¼
uG ¼
L mρL π d =4 2
G mρG π d =4 2
ð13Þ
ð14Þ
Before developing the correlation about the modifying factor C, it is necessary to have an analysis to its variation trend with the dimensionless variables based on Eq. (9). By the physical meanings of We, ReG and ReL, when We and ReG increase and ReL decreases, the resistance of two phase flow will approach to that of single gas flow, i.e., C-1, especially when liquid phase doesn’t exist, C ¼1. As α or flow area decreases, the velocity of gas–liquid mixture increases and hence the pressure loss will increase, but C will decrease, similar to the case of pipe flow where resistance coefficient decreases as pipe diameter decreases under given flow rate. When all other variables are given, the decreased l will lead to less energy loss due to smaller turning angle β and thus C will decrease. From the considerations above, the expression for C may be written in the form of C¼
1 1 þ f 2 We; ReL ; ReG ; α; d=l
ð15Þ
Based on this equation and the considerations for the variation trends of C with the relevant dimensionless variables, a correlation for modifying factor is developed as C¼
!1 2 0:128 d Re0:72 L þ 1 1 þ 1:1 l α ðReG WeÞ0:36
ð16Þ
This equation indicates that C is always less than or equal to 1, and when there is no liquid, i.e., ReL ¼0, the flow becomes the single gas flow, i.e., C ¼1; or when gas flow rate is much larger than liquid flow rate, i.e., ReG» ReL and We»1, the resistance coefficient C will approach to 1. Also, from Eq. (16), it is seen that when α or l decreases, the C will decrease. These variation trends are all in agreement with the previous analysis. Summarizing the above, all the empirical correlations to calculate pressure loss of concurrent gas–liquid downward flow through the sieve plate packing for semi-dispersed flow pattern are listed in Table 2.
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 300
250
packing A packing B packing C
ΔP Calculated (Pa)
200
+10%
150
−10% 100
50
0 0
50
100
150
200
250
300
ΔP Experimental (Pa)
Fig. 23. Parity plot for experimental and calculated two phase pressure loss.
In the above correlations for two phase flow, the tested ranges of the parameters are 136oReL o1032, 2ReL þ 2500oReG o13200, 2oWeo46, 0.0933r α r0.1154, 0.43rd/lr0.50 and 1.11rl/ br1.27, respectively. The correlations predict the two phase pressure loss with an average error of less than 10%, as shown in Fig. 23.
5. Conclusion For gas–liquid concurrent downward flow through sieve plate packing, four flow states of liquid phase (trickling, continuous, semidispersed and completely dispersed) have been found. Pulsing flow regime through packings has been observed and the mechanism of pulsing generation is presented. The flow regime transition boundaries are obtained visually and presented in flow maps. At given liquid flux, the pulsing flow appears once the gas flux reaches to a limitation where the necessary area for the smooth flow of gas phase exceeds its actual flow area. For packing with smaller holes it is easier to achieve pulsing flow regime at fixed free area ratio. The pressure losses for three packings are measured and compared. Based on the analysis of experimental results and packing structure, the resistance coefficient correlations for single phase (gas) flow are obtained for both developing flow regime (Rer4500) and developed flow regime (Re44500), i.e., Eqs. (4) and (5), respectively. Interaction between two phases in conjunction with the packing structure contributes to the liquid dispersion and the pressure loss of two phase flow. The enlargement of both free area ratio and hole pitch/diameter ratio promotes the liquid dispersing. Thus, the appropriate hole distribution may enhance the mass transfer between gas and liquid. Furthermore, a series of correlations predicting the two phase pressure loss through sieve plate packings has been developed, see Table 2, the calculated results by which are in good agreement with the experimental values.
Reference Alimonti, Claudio, Falcone, Gioia, Bello, Oladele, 2010. Two-phase flow characteristics in multiple orifice valves. Exp. Therm. Fluid Sci. 34, 1324–1333. Al-Naimi, Safa A., Al-Sudani, Farah Talib Jasim, Halabia, Essam K., 2011. Hydrodynamics and flow regime transition study of trickle bed reactor at elevated temperature and pressure. Chem. Eng. Res. Des. 89, 930–939. Asali, J.C., Hanratty, T.J., Andreussi, P., 1985. Interfacial drag and film height for vertical annular flow. AIChE J. 31, 895–902.
215
Beg, S.A., Hassan, M.M., Naqvi, M.S.M., 1996. Hydrodynamics and mass transfer in a cocurrent packed column: a theoretical study. Chem. Eng. J. 63, 93–103. Babu, Ponnan Ettiyappan Jagadeesh, Arunagiri, Appusamy, Murugesan, Thanapalan, 2006. Prediction of two-phase pressure drop and liquid holdup in co-current gas–liquid downflow of air–Newtonian systems through packed beds. J. Chem. Technol. Biotechnol. 81, 70–81. Benkrid, K., Rode, S., Midoux, N., 1997. Prediction of pressure drop and liquid saturation in trickle-bed reactors operated in high interaction regimes. Chem. Eng. Sci. 52 (21,22), 4021–4032. Berna, C., Escrivá, A., Muñoz-Cobo, J.L., Herranz, L.E., 2015. Review of droplet entrainment in annular flow: Characterization of the entrained droplets. Prog. Nucl. Energy 79, 64–86. Chung, D.K., Mills, A.F., 1974. Effect of interfacial shear on gas absorption into a turbulent falling with co-current gas flow. Lett. Heat. Mass. Transf. 1, 43–48. Chien, S.F., Ibele, W., 1964. Pressure drop and liquid film thickness of two-phase annular and annular-mist flows. J. Heat. Transf. 86, 89–96. Fried., E., Idelchik., I.E., 1989. Flow Resistance: A Design Guide for Engineers. Taylor &Francis, Philadelphia. Fischer, R., 1995. Calculation of the discharge characteristic of an orifice for gas– liquid annular-mist flow. Int. J. Multiph. Flow. 21 (5), 817–835. Fedotkin, I.M., Ivanov, V.S., Lisman, V.S., Chepurnoy, M.N., Shnayder, V.E., 1979. Pressure drop in downtake turbulent film and gas flows. Fluid Mech. Sov. Res. 8, 85–90. Gan, Guohui, Riffat, Saffa B., 1997. Pressure loss characteristics of orifice and perforated plates. Exp. Therm. Fluid Sci. 14, 160–165. Hutton, B.E.T., Leung, L.S., 1974. Cocurrent gas–liquid flow in packed columns. Chem. Eng. Sci. 29 (8), 1681–1685. Henstock, W.H., Hanraty, T.J., 1976. The interfacial drag and the height of the wall layer in annular flows. AIChE J. 22, 990–1000. Hajiloo, M., Chang, B.H., Mills, A.F., 2001. Interfacial shear in downward two-phase annular co-current flow. Int. J. Multiph. Flow. 27, 1095–1108. He, Jie, Pengfei, Shan, Changbin, Tang, Lili, Zhang, 2012. Pressure drop of packed column in gas–liquid co-current flow. Chem. Ind. Eng. (China) 29 (4), 45–49. Kolodzie JR, P.A., Winkle, Matthew Van, 1957. Discharge coefficients through perforated plates. AIChE J. 3 (3), 305–312. King, S. Ho. Kee, Piar, G., 1999. Effects of entrainment and deposition mechanisms on annular dispersed two-phase flow in a converging nozzle. Int. J. Multiph. Flow. 25, 321–347. Malavasi, Stefano, Messa, Gianandrea, Fratino, Umberto, Pagano, Alessandro, 2012. On the pressure losses through perforated plates. Flow. Meas. Instrum. 28, 57–66. Mahajani, V.V., Sharma, M.M., 1980. Mass transfer in packed columns: co-current (down flow) operation: 1 in. and 1.5 in. Metal pall rings and ceramic intalox saddles: multifilament gauze packings in 20 cm and 38 cm i.d. columns. Chem. Eng. Sci. 35 (4), 941–947. Minghui, Yang, Ke, Deng, Yinlong, Wang, Weixing, Huang, 2014. Discussion on calculation of flow resistance in U-tube heat exchangers for feed water heating in nuclear power plant. Process. Equip. Pip. (China) 51 (2), 17–21. Nemec, Damjan, Berčič, Gorazd, Levec, Janez, 2001. The hydrodynamics of trickling flow in packed beds operating at high pressures. The relative permeability concept. Chem. Eng. Sci. 56, 5955–5962. Nigmatulin, R.I., Nigmatulin, B.I., Khodzhaev, Y.A.D., Kroshilin, V.E., 1996. Entrainment and deposition rates in a dispersed-film flow. Int. J. Multiph. Flow. 22 (1), 19–30. Rao, V.G., Drinkengurg, A.A.H., 1985. A model for pressure drop in two-phase gas– liquid downflow through packed columns. AIChE J. 31 (6), 1010–1018. Raynal, Ludovic, Ballaguet, Jean-Pierre, Cécile, Barrere-Tricca, 2004. Determination of mass transfer characteristics of co-current two-phase flow within structured packing. Chem. Eng. Sci. 59, 5395–5402. Saroha, Anil K., Khera, Ritesh, 2006. Hydrodynamic study of fixed beds with concurrent upflow and downflow. Chem. Eng. Process. 45, 455–460. Shende, B.W., Sharma, M.M., 1974. Mass transfer in packed columns: co-current operation. Chem. Eng. Sci. 29 (8), 1763–1772. Schildhauer, Tilman J., Pangarkar, Kalyani, van Ommen, J. Ruud, Nijenhuis, John, Moulijn, Jacob A., Kapteijn, Freek, 2012. Heat transport in structured packings with two-phase co-current downflow. Chem. Eng. J. 185–186, 250–266. Salgi, Paul, Balakotaiah, Vemuri, Ramé, Enrique, Motil, Brian J., 2015. Pulse properties in gas–liquid flow through randomly packed beds under microgravity conditions. Int. J. Multiph. Flow. 73, 11–16. Schmidt, J., Friedel, L., 1997. Two-phase pressure drop across sudden contractions in duct areas. Int. J. Multiph. Flow. 23 (2), 283–299. Taulamet, María J., Mariani, N.éstor J., Barreto, Guillermo F., Martínez, Osvaldo M., 2014. Estimation of overall heat transfer coefficients in packed beds with cocurrent downflow of gas and liquid. Fuel 138, 176–182. Tsochatzidis, N.A., Karabelas, A.J., Giakoumakis, D., Hu, G.A., 2002. An investigation of liquid maldistribution in trickle beds. Chem. Eng. Sci. 57, 3543–3555. Tishkoff, J.M., Pinchak, A.C., Ostrach, S., 1979. Turbulent co-current gas–liquid flow in a tube with and without swirl. J. Fluids Eng. 101, 61–68.
Contents lists available at ScienceDirect
Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces
Hydrodynamics and pressure loss of concurrent gas–liquid downward flow through sieve plate packing Wangde Shi, Weixing Huang, Yuhan Zhou, Huan Chen, Dawei Pan School of Chemical Engineering, Sichuan University, Chengdu 610065, China
H I G H L I G H T S
Four flow patterns (trickling, continuous, semi- and perfect-dispersed) were found. Pulsing flow regime was observed and analyzed, and the flow maps were presented. Mechanism of liquid dispersion and its effect on flow resistance were described. New correlations for the pressure drop of gas–liquid two phase flow were developed.
art ic l e i nf o
a b s t r a c t
Article history: Received 15 September 2015 Received in revised form 14 December 2015 Accepted 5 January 2016 Available online 12 January 2016
Gas–liquid concurrent downward flow through a new structured packing was investigated by experiment systematically. The experimental packing consisted of 21 sieve plates. Each plate was 190 mm 190 mm in side length and 1 mm in thickness. Three sets of packing with different sizes of sieve holes were tested. During experiment, four flow states have been observed, i.e., the trickling flow at low gas and liquid flux, the continuous flow at low gas but high liquid flux, the semi-dispersed flow at high gas flux and the completely dispersed flow as gas flux increases further. Another important phenomenon observed is the occurrence of pulsing flow, i.e., in some range of gas and liquid flow fluxes, both two phases will no longer flow smoothly through the sieve plates but flow downward in pulse regularly. Through extensive experiment, the rough boundaries for flow regime transition were obtained. Then, the pressure losses of gas flow and gas–liquid two phase flow in non-pulsing flow regime were systematically measured. The analysis for pressure differences measured at different locations show that for a given packing, the pressure loss through each plate is nearly the same. Its magnitude depends on the gas and liquid flow rates. The comparisons between pressure losses of different packings indicate that the pressure loss is associated with the hole diameter, hole pitch, free area ratio of sieve plate and the mounting distance between two adjacent plates. Finally, considering these factors, the correlations for pressure drops were developed, which approximates the experimental values with an average deviation less than 10%. & 2016 Elsevier Ltd. All rights reserved.
Keywords: Sieve plate packing Gas–liquid concurrent downflow Flow patterns Pulsing flow Pressure loss
1. Introduction Packed columns are widely used as separation equipment for many years and numerous kinds of packing (both random and structured) have been developed. Of the flow modes identified in packed columns, the countercurrent and concurrent flow are most commonly used in the operation of commercial separation units. Nowadays, a new structured packing consisting of multiple sieve plates has been developed in packed columns operated with gas– liquid concurrent downward flow mode, e.g., the gas stripper in nuclear plants. This new packing with concurrent flow can also be preferred in a wide range of other industrial situations, such as chemical, petrochemical and biochemical industries, etc, especially for mass transfer process when there is no significant difference in http://dx.doi.org/10.1016/j.ces.2016.01.005 0009-2509/& 2016 Elsevier Ltd. All rights reserved.
the mean concentration driving force offered by two modes (Beg et al., 1996). In such a condition, this new packing with concurrent mode has some marked advantages in two aspects. The first is in the excellent hydrodynamic performance as compared with countercurrent mode, i.e., high throughputs, relatively low pressure loss and absence of flooding (Beg et al., 1996; Saroha and Khera, 2006; Babu et al., 2006; He et al., 2012), and the second is in the simple geometry of the sieve plate packing for manufacture and installment. However, the design of this new structured packed column still relies on experience or simulation so far and the design result is dubious. To provide reliable guidelines for the design, the understanding to the hydrodynamic behaviors of gas–liquid flow in this new packing is badly needed. At present, most of the relevant
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
207
studies about concurrent flow are focused on random-packed columns, for example, those of Hutton and Leung (1974), Rao and Drinkengurg (1985), Benkrid et al. (1997), Nemec et al. (2001) and Babu et al. (2006) for hydrodynamics, Shende and Sharma (1974) and Mahajani and Sharma (1980) for mass transfer, and Taulamet et al. (2014) for heat transfer. A few previous studies are on structured packed columns, for example, those of Raynal et al. (2004) for metal corrugated sheet structured packing and Schildhauer et al. (2012) for foams and knitted wire packing. Whether for random or structured packing, knowing the hydrodynamic performance is the basis for the investigation of mass or heat transfer. Although those previous efforts have made some contributions to understand the hydrodynamic performance of gas– liquid concurrent flow in packed columns, a lot of research work is still needed to design this new kind of columns precisely and to develop more efficient sieve plate packing. The aim of the present study is to investigate the flow mechanism and interaction between gas and liquid when they flow concurrently downward through the sieve plate packing, as well as the effect of packing structures, based on systematical experiments, and then to develop practical correlations for the calculation of pressure loss accordingly, so that the guidelines can be provided for the hydrodynamic design and the practical operation of commercial columns.
2. Experimental apparatus and process The experimental arrangement is shown in Fig. 1. The experimental column consists of a sieve plate packing and a rectangular Perspex shell with dimensions of 200 mm 200 mm 600 mm, so the flow development in the packing can be observed visually and recorded by a camera. The packing is composed by 21 pieces of stainless steel square plate with 190 mm 190 mm in side length and 1mm in thickness as shown in Fig. 2. Three sets of packing are tested and their geometrical parameters, as listed in Table 1, are designed by reference to those of commercial unit. Four pressure taps are amounted along the height of the test packings to measure the pressure losses of gas–liquid flow from the top plate to the 7th, the 14th and the 21st plate, respectively, by three U-tube manometers. The minimum scale of U-tube manometer is 1 mmH2O. In order to insure the precision, the pressure loss through 21 plates is averaged as the pressure loss of T 7
T 1
11 10 10
10
10
2 9 4 T
9
9
one plate. The range of the measured pressure difference in the experiment is from 4 to 470 mmH2O, and the mean error for manometer readings is about 0.84%. All tests were carried out by using air and water at room temperature (about 20 °C) and common pressure (1 atm). The range of air flow rate for pure gas phase experiment was 20– 250 m3/h. During two phase experiment, the range of air flow rate for packing A was 20–230 m3/h while the range of air flow rate for both packing B and packing C was 20–210 m3/h. The range of water flow rate for each test packing was 0.4–1.6 m3/h. The minimum scale of gas flow-meter is 2 m3/h and in the range of gas flowrate tested, the mean error for gas flow-meter readings is about 1.5%. The minimum scale of liquid flow-meter is 40 L/h and in the range of liquid flowrate tested, the mean error for liquid flow-meter readings is about 4%. In the experiment, two phases flow concurrently downward through the test packing. Air was sent into the top of the column by a blower and its flow rate was controlled by regulating valves and monitored by rotameters. Water was sent to the liquid distributor mounted at the top of the column under the static pressure provided by an overhead surge tank, and then flowed downward through the test packing by gravity.
9
3 9
Fig. 2. Packing structure.
9
9
3. Experimental result and discussion 3.1. Flow state of gas/liquid through the packing
9 8 5
1.Gas-liquid distributor 2.Test section 3.Gas-liquid separator 4.U-tube manometer
5.Water storage tank 6.Pump 7.Water surge tank 8.Blower
6
9.Regulating valve 10.Rotameter 11.Silencer
Fig. 1. Sketch of experimental setup and air–water flow loop.
For concurrent downward flow of gas and liquid through the sieve plate packing, four typical states are observed, i.e., liquid trickling, continuous flow, semi-dispersed flow and completely dispersed flow. Experiment indicates that the occurrence of these four states depends on the magnitude of gas and liquid flux through the plate hole, the interaction between two phases, and the interfacial effect between liquid and plate.
208
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
Table 1 Geometrical characteristics of three test packings. Packing label
packing A
packing B
packing C
d (mm) Hole quantity m Plate quantity Free area ratio α l (mm) b (mm) e (mm)
14 30 21 0.1154 30 25 1
10 56 21 0.1099 20 18 1
6 132 21 0.0933 14 11 1
Fig. 3. Trickling flow at low gas and liquid flux (packing A, WL ¼ 24.0 kg/m2s). (a) WG ¼ 0 kg/m2s. (b) WG ¼2.9 kg/m2s. (c) corresponding diagram.
Fig. 4. Continuous flow at low gas flux but high liquid flux (packing A, WL ¼84.3 kg/m2s). (a) WG ¼3.6 kg/m2s. (b) corresponding diagram.
At low gas and liquid flux, the flow of liquid will be in a form of trickling, as shown in Fig. 3. In this case, the interaction between two phases is weak and the momentum of liquid is small, so that the interfacial effect between liquid and plate becomes relatively important, leading to liquid flowing along the back surface of the plate and finally dropping off to the next plate. This flow pattern is so called trickling flow. When the liquid flux is increasingly enhanced under low gas flux, the momentum of liquid will become strong enough compared to gas–liquid interaction and liquid-plate interfacial effect. In this case, the flow of liquid through the plate hole will evolve into a continuous state, i.e., the liquid from each hole will flow downward in a nearly continuous annular film, as shown in Fig. 4. Contrary to the case of low gas flux, when the gas flux becomes high enough so that the interaction between two phases becomes
a dominating factor, the flow of liquid through the plate hole will evolve into a semi-dispersed state, as shown in Fig. 5. The semidispersed state always occurs at relatively high gas flux, whether the liquid flux is high or low. In this flow state, one part of the liquid will be dispersed into droplets due to the dragging of gas stream, while the other part of the liquid will move along the back surface of the plate because of the wrapping of gas stream and then drop into the holes on the next plate. Of those dispersed droplets, most will be deposited on the surface of the plate downstream, while the rest will be carried away with gas stream or brought to the back surface of plate. In the semi-dispersed state, the liquid holdup on the space and the gas–liquid interaction increase apparently. From the view of mass transfer enhancement, the semi-dispersed flow pattern is preferred because it offers high interfacial area and frequent renewal of gas–liquid contacting
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
209
Fig. 5. Semi-dispersed flow at high gas flux (packing A, WG ¼10.1 kg/m2s). (a) WL ¼ 24.0 kg/m2s. (b) WL ¼ 84.3 kg/m2s. (c) corresponding diagram.
WG=16.0 kg/m2s, WL=84.3kg/m2s, packing A
corresponding diagram
Fig. 6. Flow pattern at completely dispersed state. (a) WG ¼ 16.0 kg/m2s, WL ¼ 84.3 kg/m2s, packing A. (b) Corresponding diagram.
surface. For this reason, commercial units are commonly operated in this state. As expected, when gas flux increases continually, more and more liquid will be dispersed into droplets and the flow of liquid will finally reach to a completely dispersed state as shown in Fig. 6. This state may have higher gas–liquid contacting efficiency, but the pressure loss will be increased apparently. 3.2. Pulsing flow regime and flow maps Another important flow phenomenon observed is the occurrence of pulsing flow, i.e., in some range of gas and liquid flow fluxes, the gas–liquid will not flow smoothly but instead rush alternately through each plate. Meanwhile, the readings of U-tube manometers are pulsating regularly. Pulsing flow has also been found in other kinds of packing operated in gas–liquid concurrent downward flow mode, such as in the trickle bed (Benkrid et al., 1997; Tsochatzidis et al., 2002; Al-Naimi et al., 2011; Salgi et al., 2015). The cause leading to pulsing is analyzed as follows. As gas and liquid concurrently flow downward through the plate hole, each phase will occupy a corresponding flow area, in which the flow area of liquid relies on the liquid flux (defined as the ratio of flowrate to hole area), the drag of gas stream, and the resistance of hole rim as well. At relatively low gas flux, the drag of gas phase is weak and the liquid flow area under given liquid flux has no apparent change as gas flux increases. At this situation, once the gas flux reaches to a limitation where the necessary area for the smooth flow of gas phase exceeds its actual flow area, a part of gas
will be plugged and gas pressure will increase transiently until the gas pushes the liquid through the hole, leading to the occurrence of pulsing. On the other hand, when the gas flux becomes sufficiently high, the drag of gas stream will become strong enough to expand the flow area for itself and the pulsing flow will disappear. By combining the observation results of the pulsing regime and the flow states with the corresponding operation conditions, the flow maps for packings A, B and C are presented in Figs. 7–9. For packing A, as shown Fig. 7, the pulsing has not been observed in whole range of liquid flux examined, so that only flow state transition lines are presented. It should be pointed that in the experiments, including for packings B and C, the transition from trickling to continuous flow is observed to happen at WL ¼51– 63 kg/m2 s, so that the transition line at WL ¼60 kg/m2 s in Fig. 7, as well as in Figs. 8 and 9, is just a rough line. Further observation on flow phenomenon in packing A confirms that the transition boundary to the semi-dispersed flow varies from WG ¼4.33 kg/ m2 s to WG ¼ 7.94 kg/m2 s with the increase of liquid flux, giving a rough line of transition as shown in Fig. 7. Under the flow fluxes examined, the pulsing flow regime is observed in packings B and C, as shown in Figs. 8 and 9. From the flow maps of packings B and C, it can be seen that the variation of the lower or upper limit line of the pulsing region with liquid flux has the same trend for both packing B and packing C, i.e., the lower limit line goes down and the upper limit line goes up with increasing liquid flux. Obviously, the lower limit line represents the start points of pulsing flow. According to the mechanism for pulsing generation described above, the lower liquid flux will leave more flow area for gas phase and thus the corresponding gas
210
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 15
5000
14 13
completely dispersed flow
12
7 layers 14 layers 21 layers packing A
4000
11 3000
9
semi-dispersed flow
8
ΔP (Pa)
W (kg/m /s)
10
7 6
2000
5 4 3
continuous flow
trickling flow
2
1000
1 0
0 20
30
40
50
60
70
80
90
20
100
40
60
80
100
15
160
180
200
220
240
220
14 13
7 layers 14 layers 21 layers
200
completely dispersed flow
12
180
11
packing A
160
10
140
9 8
semi-dispersed flow
ΔP (Pa)
W (kg/m /s)
140
Fig. 10. Effect of plate quantity on pressure loss of single phase (gas) flow.
Fig. 7. Flow map for packing A.
7 6
120 100 80
5
pulsing flow
4
60
3
trikling flow
2
40
continuous flow
20
1 0 20
30
40
50
60
70
80
90
100
110
W (kg/m /s)
14
semi-dispersed flow 8
pulsing flow
4
trickling flow continuous flow 20
40
60
60
80
100
120
140
160
180
200
220
240
3.3. Pressure loss of single phase (gas) flow
10
0
40
flow smoothly through the hole, the pulsing will not occur, as the case in Fig. 7 for packing A and the case at low liquid flux region in Fig. 8 for packing B.
12
2
20
Fig. 11. Average pressure loss of single phase (gas) flow through 7, 14 and 21 plates.
completely dispersed flow
6
0
Q (m /h)
Fig. 8. Flow map for packing B.
W (kg/m /s)
120
Q (m /h)
W (kg/m /s)
80
100
W (kg/m /s) Fig. 9. Flow map for packing C.
flux to produce pulsing will become higher. The upper limit line represents the end points of pulsing flow. According to the mechanism for pulsing elimination, when liquid flux becomes higher, a larger gas flux will be required to expand enough flow area and eliminate pulsing, thus the upper limit line goes up with the increasing liquid flux. It should be pointed that when the hole size and the free area of the plate relative to liquid flux are large enough to allow the gas
As an extreme case of two phase flow, the pressure loss of single phase (gas) flow through the packing are measured. The pressure losses of gas flow through 7, 14 and 21 plates in packing A are presented in Fig. 10 and it is shown that the pressure losses vary with gas flow rate and plate quantity. For comparison, the average pressure losses per each plate are calculated by dividing the total pressure loss through 7, 14 and 21 plates with corresponding plate quantity and the results are given in Fig. 11. It shows that the average pressure loss through each plate is nearly the same, indicating that the pressure loss of single phase (gas) flow increases linearly with the quantity of sieve plates. Similar results are found for packings B and C. Therefore the average pressure loss for one plate can be used as a basis to investigate the resistance behaviors of single phase (gas) flow through different packings. For simplicity, the average pressure loss of single phase (gas) flow through one plate will be termed as the dry-plate pressure loss in the following discussion. The variation of dry-plate pressure loss with gas flow rate for packings A, B and C is shown in Fig. 12. From the figure it is not difficult to find that the geometrical parameters of the packing have a complex effect on the pressure loss. Generally, the hole diameter d and the free area α may be two important affecting
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 300
211
220
packing A, d=14mm,α =0.1154, d/b=0.560, l/b =1.20 packing B, d=10mm,α =0.1099, d/b=0.556, l/b=1.11 packing C, d= 6 mm,α =0.0933, d/b=0.545, l/b =1.27
250
180
7 layers 14 layers 21 layers
160
packing A, Q =0.8m /h
200
200
ΔP (Pa)
Δ P (Pa)
140 150
120 100 80
100
60 50
40 20
0 0
20
40
60
80
100
120
140
160
180
200
220
240
0
260
20
40
60
80
100
Q (m /h)
120
140
160
180
200
220
240
Q (m /h)
Fig. 12. Variation of dry-plate pressure loss with gas flow rate for different packings.
Fig. 14. Average pressure loss of two phase flow through 7, 14 and 21 plates.
5000
240 220
4000
200
7 layers 14 layers 21 layers
180 160
packing A, Q =0.8m /h
ΔP (Pa)
3000
ΔP (Pa)
140 2000
120 100 80
1000
60 40 20
0 20
40
60
80
100
120
140
160
180
200
220
240
Q (m /h) Fig. 13. Effect of plate quantity on pressure loss of two phase flow.
factors, which account for the effect of the cross-section area on flow resistance. For example, the d and α of packing C are apparently the smallest compared with those of packings A and B, thus packing C has the largest pressure loss at a given gas flow rate. However, when these two parameters become relatively large, the effects of the ratios d/b and l/b may become apparent, in which d/b accounts for the effect of energy loss caused by the impact of gas stream on the next plate, while l/b represents the effect of energy loss caused by the abrupt change of gas flow direction. For example, although the packing A has the largest d andα, its pressure loss is higher than that of packing B, out of common expectation. Comparing the geometrical parameters of packings A and B, it can be seen that both packings have nearly equal free area and relatively large holes, but the l/b of packing A is apparently larger than that of packings B. This means that the flow resistances through the hole for packings A and B are comparable, but the energy loss caused by abrupt change of gas flow direction in packing A is apparently larger, so that packing A has a higher pressure loss than packing B. 3.4. Pressure loss of two phase flow For two phase flow through the packings, the pressure loss and its variation with gas flow rate under different liquid flow rates have many similarities to the case of single phase (gas) flow. Fig. 13 shows the variation of pressure losses of two phase flow through 7, 14 and 21 plates with gas flowrate under QL ¼0.8 m3/h in packing
0 20
40
60
80
100
120
140
160
180
200
220
240
Q (m /h) Fig. 15. Variation of two phase pressure loss for one plate with gas/liquid flow rate (packing A).
A and Fig. 14 shows the average pressure losses per each plate calculated by dividing the total pressure loss through 7, 14 and 21 plates with corresponding plate quantity. As shown in Fig. 14, for different total pressure losses, their average pressure losses through each plate are also nearly the same, similar to the case of single gas flow, indicating that the pressure loss of two phase flow also increases linearly with the quantity of sieve plates. Similar results are found for packings B and C. For this reason, the following discussion will be focused on the average pressure loss for one plate. To compare the effect of liquid flowrate, the variations of pressure loss with gas flowrate under different liquid flowrates are presented in Figs. 15, 16 and 17 for packings A, B and C, respectively. These figures indicate that the higher liquid flowrate generally corresponds to the higher pressure loss under the given gas flowrate for a given packing. Examining the pressure loss through packing A as shown in Fig. 15, it is found that as gas flowrate increases, the effect of liquid flowrate becomes steadily increased, i.e., the interval of variation of the pressure loss over the range of liquid flowrate increases steadily with gas flowrate. Further, when the gas flow rate QG reaches 110 m3/h, the interval of pressure loss variation will reach a nearly constant size. Considering the fact that the transition boundary to the semi-dispersed flow for
212
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 200 180 160 140
ΔP (Pa)
120 100 80 60 40 20 80
100
120
140
160
180
200
220
Q (m /h) Fig. 16. Variation of two phase pressure loss for one plate with gas/liquid flow rate (packing B).
300 280 260 240
ΔP (Pa)
220 200 180 160 140 120 100 80 140
160
180
200
220
Q (m /h)
Fig. 17. Variation of two phase pressure loss for one plate with gas/liquid flow rate (packing C).
packing A varies from QG ¼60 m3/h to QG ¼110 m3/h, it can be concluded that when the flow patterns at different liquid flow rates all have reached a semi-dispersed state, the interval of variation of the pressure loss over the range of liquid flow rate will reach a limiting size. This conclusion also can be obtained from Figs. 16 and 17 for packings B and C, in which all pressure loss data are measured in the region of the semi-dispersed flow. Fig. 18(a)–(c) presents the effects of geometrical parameters on the pressure loss of two phase flow at typical liquid flowrates, i.e., QL ¼ 0.4 m3/h, QL ¼0.8 m3/h and QL ¼1.2 m3/h. These three Figures show that under a given gas flowrate, the packing C has the largest pressure loss, the packing B has the lowest pressure loss, and the pressure loss of packing A is in between, no matter the liquid flow rate is low or high. This result, similar to the case of single gas flow, further confirms that the hole diameter d and the free area α are two important affecting factors on pressure loss, but when these two parameters become relatively large, the effects of the ratios d/ b and l/b may become apparent. This result also implies that the packing geometrical characteristics have some similar influence on both single phase (gas) and two phase pressure loss.
Fig. 18. (a). Effect of packing geometrical characteristics on two phase pressure loss at QL ¼0.4 m3/h. (b). Effect of packing geometrical characteristics on two phase pressure loss at QL ¼0.8 m3/h. (c). Effect of packing geometrical characteristics on two phase pressure loss at QL ¼ 1.2 m3/h.
4. Resistance coefficients for sieve plate packing 4.1. Resistance coefficient for dry plate For the fluid flow through a sieve plate, the flow mode through every hole can be regarded as the same and, therefore, the pressure loss through one hole will represent that through the sieve plate. Not like the pressure loss of fluid flowing in a pipe where the pressure loss is caused by wall friction, the pressure loss through the hole belongs to local energy loss, mainly caused by the
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
contraction and expansion of fluid stream, the friction of hole rim to the fluid, the abrupt change of flow direction and the impact of fluid stream on the plate downstream. According to this, the pressure loss for single gas flow through the sieve plate (termed as dry plate in the following discussion) can be expressed as
ΔP ¼ ξG
W 2G W 2G ¼ ξ0 þ ξ1 2 ρG 2 ρG
ð1Þ
where WG is the mass flux of gas flow through the hole, ρG is the density of gas, ξ0 is the resistance coefficient for impacting pressure loss, and ξ1 is the resistance coefficient through the dry plate corresponding to the effects of other factors, i.e., the fluid stream contraction and expansion, the friction of hole rim and the abrupt change of flow direction. For the calculation of ξ0, Minghui et al. (2014) has proposed a semi-empirical equation as follows
ξ0 ¼ 1:17
2
d
4b
ð2Þ
2
where d is the diameter of the hole, and b is the distance between two adjacent plates. For ξ1, many similar researches are referenced and analyzed in this work. The studies for the resistance coefficient through an orifice (Alimonti et al., 2010) or a perforated plate (Kolodzie and Winkle, 1957; Fried. and Idelchik., 1989; Gan and Riffat, 1997; Malavasi et al., 2012) show that the effect of geometrical parameters on resistance coefficient is associated with Reynolds number defined as Q ρ Re ¼ 4 G G mπ dμG
ð3Þ
where, QG is the total volume flow rate, m is the hole number in one plate and μG is the viscosity of gas. According to Eqs. (1) and (3), the Re and corresponding ξ1 are calculated based on experimental data, as shown in Fig. 19. From Fig. 19, it can be seen that when the Reynolds number Re is relatively low, about in the range of Rer 4500, the ξ1 is in a descending trend with increasing Re, while for Re44500, the ξ1 seems independent of Re and the magnitude of ξ1 is only related to the geometrical parameters of packings. For this reason, the data of resistance coefficient ξ1 are correlated, respectively, for two regions divided by Re¼4500. Next, the comparison between the flow mechanisms of fluid through a single perforated plate with that through multiple sieve plates is carried out in order to determine the relevant geometrical parameters affecting ξ1. As shown in Fig. 20, for the flow through multiple sieve plates, the subsequent plate will lead to abrupt
turning of flow stream and generating more vortexes, so that the abrupt change of flow direction must be included in ξ1 (Given that the impact has been considered in ξ0). Since the turning angle β, see Fig. 21, is related with the ratio of l/b, the l/b will be included in ξ1. According to the considerations above and fitting all experimental data, the correlations for resistance coefficient ξ1 are developed as follows qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4:851 2 4:851 1 þ l=2b secβ ξ1 ¼ 0:292 ¼ 0:292 ðRe 4 4500Þ ð1 αÞ7:2765 ð1 αÞ7:2765 ð4Þ
ξ1 ¼ 1:898
2.0
packing A packing B packing C
ξ
1.8
1.6
1.4
1.2 0
2000
4000
6000
8000
10000
12000
14000
Re Fig. 19. The variation of dry-plate resistance coefficient with Reynolds number based on one hole.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4:851 2 1 þ l=2b ð1 αÞ7:2765
Re 0:22
ðRe r 4500Þ
ð5Þ
The tested ranges of the parameters in Eqs. (4) and (5) are 1156o Re o13913, 0.0933o α o0.1154 and 1.1 ol/bo1.27, respectively. Both Eqs. (4) and (5) approximate most of the experimental values with the deviation of less than 10%, as shown in Fig. 22. 4.2. Resistance coefficient for two phase flow in semi-dispersed region The pressure loss ΔP of two phase flow through the sieve plate still belongs to local energy loss, so that
ΔP ¼ ξ m
W 2m 2 ρm
ð6Þ
In Eq. (6), ξm is the resistance coefficient of two phase flow through the sieve plate, Wm is the total mass flux of the gas–liquid mixture through the hole, and ρm is a nominal density of the gas– liquid two phase fluid when it is treated as homogenous mixture, (Schmidt and Friedel, 1997) defined as
ρm ¼
LþG L=ρL þ G=ρG
ð7Þ
where G ¼ ρGQG and L ¼ ρLQL are the mass flow rate of gas and liquid phase, respectively. As mentioned in Section 3.4, for two phase flow under given liquid flow rate, the variation of pressure loss with gas flow rate and the effect of geometrical parameters are similar to the case of single gas flow. For this reason, the resistance coefficient ξm for two phase flow will be treated as the product of dry-plate coefficient ξG times a modifying factor C, i.e.,
ξm ¼ C ξG
2.2
213
ð8Þ
As a modifying factor for resistance coefficient of two phase flow, the C should be associated with the dispersing process of liquid, the relative magnitude of gas and liquid flow rates and the geometrical parameters of packings. Since the dispersing of liquid is a result of the interaction between flowing gas and liquid, i.e., to tear liquid into fragments and droplets the shear stress acting on the liquid surface by gas must overcome the surface tension of liquid. Thus, the Weber number should be an important parameter affecting C, as suggested by numerous open literatures on liquid dispersion (Fischer, 1995; Nigmatulin et al., 1996; King and Piar, 1999; Berna et al., 2015). Also, lots of prior studies on concurrent gas–liquid core/annular flow in pipes indicate that interfacial shear stress relates to gas and liquid Reynolds numbers (Chung and Mills, 1974; Tishkoff et al., 1979) and many skin friction coefficients have been successfully correlated by superficial gas and liquid Reynolds numbers (Chien and Ibele, 1964; Henstock and Hanraty, 1976; Fedotkin et al., 1979; Asali et al., 1985; Hajiloo et al., 2001).
214
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215
a single perforated plate
multiple sieve plates
Fig. 20. Flow through a thin perforated plate and multiple sieve plates. (a) A single perforated plate. (b) Multiple sieve plates.
l
Table 2 The empirical correlations for two-phase pressure loss in semi-dispersed flow region.
β
l /2
b
Pressure loss Dry-plate resistance coefficient
Fig. 21. The turning angle of flow path.
! 2 W 2m d W 2m ¼ C ξ1 þ 1:17 2 2ρm 4b 2ρm
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4:851 2 1 þ l=2b
For ReG r 4500
ξ1 ¼ 1:898
For ReG 4 4500
2.2
Modifying factor for two phase flow 2.0
Dimensionless parameter
1.8
ξ Calculated
ΔP ¼ ξm
7:2765
(17) ReG 0:22
ð1 α Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2ffi 4:851 1 þ l=2b
ξ1 ¼ 0:292 7:2765 ð1 !αÞ1 2 0:128 d Re0:72 L þ1 C ¼ 1 þ 1:1 l α ðReG WeÞ0:36 4L ReL ¼ mπdμL 4G ReG ¼ mπdμG 16G2 We ¼ 3 m2 π 2 ρ G d σ
(4)
(5) (16) (18) (19) (20)
1.6
packing A Re<4500 packing A Re>4500 packing B Re<4500 packing B Re>4500 packing C Re<4500 packing C Re>4500
1.4
1.2 1.2
1.4
1.6
1.8
2.0
2.2
ξ Experimental
Fig. 22. Parity plot for experimental and calculated dry-plate resistance coefficient.
Moreover, for two phase flow through the sieve plate hole, the additional effects of the flow area, the hole pitch and the distance between plates on the liquid dispersing and pressure loss must be considered. Therefore, given that the effect of plate distance b has been considered in ξG, the modifying factor C will be a function of Weber number, gas and liquid Reynolds numbers, α and d/l, i.e., ð9Þ C ¼ f 1 We; ReL ; ReG ; α; d=l where We, ReL and ReG are defined as We ¼ ReG ¼
ReL ¼
ρ
2 G uG d
ð10Þ
σ
duG ρG
μG
duL ρL
ð11Þ
ð12Þ
μL
In Eqs. (10)–(12), the uG and uL are superficial gas and liquid velocities defined as uL ¼
uG ¼
L mρL π d =4 2
G mρG π d =4 2
ð13Þ
ð14Þ
Before developing the correlation about the modifying factor C, it is necessary to have an analysis to its variation trend with the dimensionless variables based on Eq. (9). By the physical meanings of We, ReG and ReL, when We and ReG increase and ReL decreases, the resistance of two phase flow will approach to that of single gas flow, i.e., C-1, especially when liquid phase doesn’t exist, C ¼1. As α or flow area decreases, the velocity of gas–liquid mixture increases and hence the pressure loss will increase, but C will decrease, similar to the case of pipe flow where resistance coefficient decreases as pipe diameter decreases under given flow rate. When all other variables are given, the decreased l will lead to less energy loss due to smaller turning angle β and thus C will decrease. From the considerations above, the expression for C may be written in the form of C¼
1 1 þ f 2 We; ReL ; ReG ; α; d=l
ð15Þ
Based on this equation and the considerations for the variation trends of C with the relevant dimensionless variables, a correlation for modifying factor is developed as C¼
!1 2 0:128 d Re0:72 L þ 1 1 þ 1:1 l α ðReG WeÞ0:36
ð16Þ
This equation indicates that C is always less than or equal to 1, and when there is no liquid, i.e., ReL ¼0, the flow becomes the single gas flow, i.e., C ¼1; or when gas flow rate is much larger than liquid flow rate, i.e., ReG» ReL and We»1, the resistance coefficient C will approach to 1. Also, from Eq. (16), it is seen that when α or l decreases, the C will decrease. These variation trends are all in agreement with the previous analysis. Summarizing the above, all the empirical correlations to calculate pressure loss of concurrent gas–liquid downward flow through the sieve plate packing for semi-dispersed flow pattern are listed in Table 2.
W. Shi et al. / Chemical Engineering Science 143 (2016) 206–215 300
250
packing A packing B packing C
ΔP Calculated (Pa)
200
+10%
150
−10% 100
50
0 0
50
100
150
200
250
300
ΔP Experimental (Pa)
Fig. 23. Parity plot for experimental and calculated two phase pressure loss.
In the above correlations for two phase flow, the tested ranges of the parameters are 136oReL o1032, 2ReL þ 2500oReG o13200, 2oWeo46, 0.0933r α r0.1154, 0.43rd/lr0.50 and 1.11rl/ br1.27, respectively. The correlations predict the two phase pressure loss with an average error of less than 10%, as shown in Fig. 23.
5. Conclusion For gas–liquid concurrent downward flow through sieve plate packing, four flow states of liquid phase (trickling, continuous, semidispersed and completely dispersed) have been found. Pulsing flow regime through packings has been observed and the mechanism of pulsing generation is presented. The flow regime transition boundaries are obtained visually and presented in flow maps. At given liquid flux, the pulsing flow appears once the gas flux reaches to a limitation where the necessary area for the smooth flow of gas phase exceeds its actual flow area. For packing with smaller holes it is easier to achieve pulsing flow regime at fixed free area ratio. The pressure losses for three packings are measured and compared. Based on the analysis of experimental results and packing structure, the resistance coefficient correlations for single phase (gas) flow are obtained for both developing flow regime (Rer4500) and developed flow regime (Re44500), i.e., Eqs. (4) and (5), respectively. Interaction between two phases in conjunction with the packing structure contributes to the liquid dispersion and the pressure loss of two phase flow. The enlargement of both free area ratio and hole pitch/diameter ratio promotes the liquid dispersing. Thus, the appropriate hole distribution may enhance the mass transfer between gas and liquid. Furthermore, a series of correlations predicting the two phase pressure loss through sieve plate packings has been developed, see Table 2, the calculated results by which are in good agreement with the experimental values.
Reference Alimonti, Claudio, Falcone, Gioia, Bello, Oladele, 2010. Two-phase flow characteristics in multiple orifice valves. Exp. Therm. Fluid Sci. 34, 1324–1333. Al-Naimi, Safa A., Al-Sudani, Farah Talib Jasim, Halabia, Essam K., 2011. Hydrodynamics and flow regime transition study of trickle bed reactor at elevated temperature and pressure. Chem. Eng. Res. Des. 89, 930–939. Asali, J.C., Hanratty, T.J., Andreussi, P., 1985. Interfacial drag and film height for vertical annular flow. AIChE J. 31, 895–902.
215
Beg, S.A., Hassan, M.M., Naqvi, M.S.M., 1996. Hydrodynamics and mass transfer in a cocurrent packed column: a theoretical study. Chem. Eng. J. 63, 93–103. Babu, Ponnan Ettiyappan Jagadeesh, Arunagiri, Appusamy, Murugesan, Thanapalan, 2006. Prediction of two-phase pressure drop and liquid holdup in co-current gas–liquid downflow of air–Newtonian systems through packed beds. J. Chem. Technol. Biotechnol. 81, 70–81. Benkrid, K., Rode, S., Midoux, N., 1997. Prediction of pressure drop and liquid saturation in trickle-bed reactors operated in high interaction regimes. Chem. Eng. Sci. 52 (21,22), 4021–4032. Berna, C., Escrivá, A., Muñoz-Cobo, J.L., Herranz, L.E., 2015. Review of droplet entrainment in annular flow: Characterization of the entrained droplets. Prog. Nucl. Energy 79, 64–86. Chung, D.K., Mills, A.F., 1974. Effect of interfacial shear on gas absorption into a turbulent falling with co-current gas flow. Lett. Heat. Mass. Transf. 1, 43–48. Chien, S.F., Ibele, W., 1964. Pressure drop and liquid film thickness of two-phase annular and annular-mist flows. J. Heat. Transf. 86, 89–96. Fried., E., Idelchik., I.E., 1989. Flow Resistance: A Design Guide for Engineers. Taylor &Francis, Philadelphia. Fischer, R., 1995. Calculation of the discharge characteristic of an orifice for gas– liquid annular-mist flow. Int. J. Multiph. Flow. 21 (5), 817–835. Fedotkin, I.M., Ivanov, V.S., Lisman, V.S., Chepurnoy, M.N., Shnayder, V.E., 1979. Pressure drop in downtake turbulent film and gas flows. Fluid Mech. Sov. Res. 8, 85–90. Gan, Guohui, Riffat, Saffa B., 1997. Pressure loss characteristics of orifice and perforated plates. Exp. Therm. Fluid Sci. 14, 160–165. Hutton, B.E.T., Leung, L.S., 1974. Cocurrent gas–liquid flow in packed columns. Chem. Eng. Sci. 29 (8), 1681–1685. Henstock, W.H., Hanraty, T.J., 1976. The interfacial drag and the height of the wall layer in annular flows. AIChE J. 22, 990–1000. Hajiloo, M., Chang, B.H., Mills, A.F., 2001. Interfacial shear in downward two-phase annular co-current flow. Int. J. Multiph. Flow. 27, 1095–1108. He, Jie, Pengfei, Shan, Changbin, Tang, Lili, Zhang, 2012. Pressure drop of packed column in gas–liquid co-current flow. Chem. Ind. Eng. (China) 29 (4), 45–49. Kolodzie JR, P.A., Winkle, Matthew Van, 1957. Discharge coefficients through perforated plates. AIChE J. 3 (3), 305–312. King, S. Ho. Kee, Piar, G., 1999. Effects of entrainment and deposition mechanisms on annular dispersed two-phase flow in a converging nozzle. Int. J. Multiph. Flow. 25, 321–347. Malavasi, Stefano, Messa, Gianandrea, Fratino, Umberto, Pagano, Alessandro, 2012. On the pressure losses through perforated plates. Flow. Meas. Instrum. 28, 57–66. Mahajani, V.V., Sharma, M.M., 1980. Mass transfer in packed columns: co-current (down flow) operation: 1 in. and 1.5 in. Metal pall rings and ceramic intalox saddles: multifilament gauze packings in 20 cm and 38 cm i.d. columns. Chem. Eng. Sci. 35 (4), 941–947. Minghui, Yang, Ke, Deng, Yinlong, Wang, Weixing, Huang, 2014. Discussion on calculation of flow resistance in U-tube heat exchangers for feed water heating in nuclear power plant. Process. Equip. Pip. (China) 51 (2), 17–21. Nemec, Damjan, Berčič, Gorazd, Levec, Janez, 2001. The hydrodynamics of trickling flow in packed beds operating at high pressures. The relative permeability concept. Chem. Eng. Sci. 56, 5955–5962. Nigmatulin, R.I., Nigmatulin, B.I., Khodzhaev, Y.A.D., Kroshilin, V.E., 1996. Entrainment and deposition rates in a dispersed-film flow. Int. J. Multiph. Flow. 22 (1), 19–30. Rao, V.G., Drinkengurg, A.A.H., 1985. A model for pressure drop in two-phase gas– liquid downflow through packed columns. AIChE J. 31 (6), 1010–1018. Raynal, Ludovic, Ballaguet, Jean-Pierre, Cécile, Barrere-Tricca, 2004. Determination of mass transfer characteristics of co-current two-phase flow within structured packing. Chem. Eng. Sci. 59, 5395–5402. Saroha, Anil K., Khera, Ritesh, 2006. Hydrodynamic study of fixed beds with concurrent upflow and downflow. Chem. Eng. Process. 45, 455–460. Shende, B.W., Sharma, M.M., 1974. Mass transfer in packed columns: co-current operation. Chem. Eng. Sci. 29 (8), 1763–1772. Schildhauer, Tilman J., Pangarkar, Kalyani, van Ommen, J. Ruud, Nijenhuis, John, Moulijn, Jacob A., Kapteijn, Freek, 2012. Heat transport in structured packings with two-phase co-current downflow. Chem. Eng. J. 185–186, 250–266. Salgi, Paul, Balakotaiah, Vemuri, Ramé, Enrique, Motil, Brian J., 2015. Pulse properties in gas–liquid flow through randomly packed beds under microgravity conditions. Int. J. Multiph. Flow. 73, 11–16. Schmidt, J., Friedel, L., 1997. Two-phase pressure drop across sudden contractions in duct areas. Int. J. Multiph. Flow. 23 (2), 283–299. Taulamet, María J., Mariani, N.éstor J., Barreto, Guillermo F., Martínez, Osvaldo M., 2014. Estimation of overall heat transfer coefficients in packed beds with cocurrent downflow of gas and liquid. Fuel 138, 176–182. Tsochatzidis, N.A., Karabelas, A.J., Giakoumakis, D., Hu, G.A., 2002. An investigation of liquid maldistribution in trickle beds. Chem. Eng. Sci. 57, 3543–3555. Tishkoff, J.M., Pinchak, A.C., Ostrach, S., 1979. Turbulent co-current gas–liquid flow in a tube with and without swirl. J. Fluids Eng. 101, 61–68.