Professional Documents
Culture Documents
Basınc Düşümü PDF
Basınc Düşümü PDF
a r t i c l e i n f o a b s t r a c t
Article history: Pressure drop predictions on the shell side of a shell and tube heat exchanger (STHX) are investigated
Received 19 September 2014 using the concept of Finite Element Method (FEM). In this model the shell side region is discretised into
Received in revised form 12 December 2014 a number of elements and by taking into account the effect of flow pattern, the pressure drop on the shell
Accepted 9 January 2015
side of a STHX is determined. The present method is simple to apply and the predictions agree reasonably
well with a large number of experimental data available in the literature. The range of applicability of the
present method extends beyond that used by others in the literature. The earlier predictions were
Keywords:
restricted to tubes in the window region, however, the predictions of the present method are extended
Shell and tube heat exchanger
Pressure drop
to the cases of no tubes in the window (NTIW) region also.
FEM model Ó 2015 Elsevier Ltd. All rights reserved.
Friction factors
No tube in window section
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2015.01.068
0017-9310/Ó 2015 Elsevier Ltd. All rights reserved.
B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712 701
Nomenclature
times the calculated values. This clearly indicates that the method a number of variables involved in the calculations. Further these
of Gaddis and Gnielinski [7] cannot be applied safely in the form models use different correlations for window section and cross
suggested by them. Kapale et al. [10] have proposed a theoretical flow section. In all the above mentioned references, the methodol-
model to calculate the shell side pressure drop. Their model incor- ogy to find pressure drop coefficient involves tedious calculations
porates the effect of pressure drop in inlet and out let nozzles along which include various geometrical parameters and is time con-
with the losses in the segments created by baffles. For the range of suming. These pressure drop coefficients have been changed time
Reynolds number between 103 and 105, they found that their and again, yet no coefficient has been found which works satisfac-
results match more closely (deviation between +2.4% and 4%) torily for all cases.
with the available experimental results. But they have not shown Friction factors for flow over rectangular tube banks have been
the validity of their model to predict pressure drop in HXs with given by Zukauskas [11] and Gunter and Haw [12]. A Finite ele-
NTIW. The calculation adopted by Kapale et al. [10] is complex. ment model of STHX for determining amount of heat transfer has
They have not predicted pressure drop for all the cases for which been developed by Ravikumaur et al. [13] in 1988 but application
experimental data is available. Thus, there is a need to develop a of such a model to determine pressure drop in STHX has not been
simple model to calculate pressure drop on the shell side of STHXs. carried out so far.
All the theoretical models reported in literature to calculate the Yonghua et al. [14], experimentally investigated the shell-side
shell-side pressure drop in a STHX require a lot of calculations with thermo-hydraulic performance of a shell and tube HX with trefoil
hole baffles under turbulent flow regime. Based on the experimental
results, empirical correlations of the Nusselt number and pressure
loss as a function of the Reynolds number are obtained. To analyze
the mechanisms of these thermo-hydraulic characteristics, numer-
ical computation is carried out. Ender and Ilker [15], investigated
the baffle spacing, baffle cut and shell diameter dependencies of
the heat transfer coefficient and the pressure drop by numerically
modeling a small HX. The authors refer to the Bell-Delaware [5]
method as a very detailed and an accurate method to estimate the
outlet parameters and have compared their results to that method,
but Bell-Delaware method itself does not predict pressure drop val-
ues close to the experimental values. The authors have also com-
pared the pressure drop results to Kapale’s [10] model and have
found a deviation of up to 34%. Results obtained from the CFD sim-
ulations show that the existing analytical methods under predict
the pressure drop in many cases. Vera et al. [16], present a model
to determine the outlet conditions of a shell and tube HX working
Fig. 1. Flow through shell of shell and tube heat exchanger with segmental baffle in a refrigeration cycle either as a condenser or evaporator only.
with leakage streams. [7]. The model does not take the internal geometrical information into
702 B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712
Upper
Window
1
2
Mid-
Secon 3
Lower
Window 4
Transverse direcon
(Longitudinal direcon)
Flow direcon
account and as a result requires special types of correlations to but is found to be deviating up to 85% from the experimental works
estimate the heat transfer coefficients and the pressure drop. The of Halle et al. [20]. In the proposed model, the correlation for fric-
pressure drop model of Vera et al. [16] is much simpler than the tion factor as suggested by Zukauskas [11] is used for each ele-
earlier published works on pressure drop models, but this model ment. The overall pressure drop is computed after assembly of
is restricted to refrigerator systems. This model uses Zukauskas all elements. The novelty of the present model lies in the fact that
friction factor [11] to determine the influence of geometrical factors the pressure drop on the shell side of the STHX can be determined
on pressure drop (according to the expression proposed by Hewitt in a simple way with limited data of HX geometry provided by
et al. [17]). But these correlations are not explicitly used as a method manufacturers. The predicted pressure drop is compared to exper-
by itself elsewhere to find pressure drop in shell and tube HXs. imental work for 240 cases. In order to limit the number of pages,
Rajagapal and Srikanth [18] have studied the effect of baffle inclina- 130 cases are presented here. The configuration codes used
tion angles of 0°, 10° and 20°. The pressure drop decreases by 4% for throughout this paper for an abbreviated description of the config-
HXs with a 10° baffle inclination angle and by 16% for HX with 20° urations of HXs tested are as presented in Table 1.
baffle inclination angle. Prithiviraj and Andrews [19] have devel-
oped a three-dimensional CFD numerical code, HEATX, to simulate 2. New pressure drop model
flow and heat transfer in shell and tube heat exchangers. The model
presented by Prithiviraj and Andrews [19] uses ideal tube bank Fluid flows when there is a pressure drop between 2 regions.
correlations of Zukauskas [11] to find the distributed resistances The pressure drop encountered between the 2 sections for a given
for shell side cross flow pressure drop and heat transfer. The CFD flow quantity is connected in the form of an element (as in the case
analysis is carried out only for a 30° tube bundle and the validation 1 1 Pi Qi
of FEM) 1R ¼ , where R is the flow resistance
for other cases are not presented. Problem set up and calculation 1 1 Pj Qj
time for executing the Bell [5] and Kern [2] method is about encountered during the flow between the 2 sections. Thus, the ele-
20 min. The typical computational time taken to run the HEATX ment relates the pressure drop encountered to the flow quantity
[19] three-dimensional simulations are about 20 h on a Pentium through the resistance encountered by the fluid. When the friction
133. However, the HEATX results are fairly accurate and a deviation factor varies along the length of the fluid flow as in the developing
up to 10% is observed when verified against the experimental work section of a pipe, this method helps to incorporate such variations
of Halle [20]. in the model. This method of representation allows us to determine
In the present work, an attempt has been made to formulate a the total pressure drop by summing up the pressure drop in each of
new and simple method to predict the pressure drop on the shell the sub regions.
side of a shell and tube HX using the concept of FEM. Initially the The steps in the FEM involve geometric representation of the
model is tested with friction factors given by Gunter et al. [12], domain, discretization of the domain into elements and nodes,
B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712 703
Position Symbols Definition In most of the investigations so far reported in literature, the
1st letter F Full tube bundle actual flow direction of fluid has not been taken into account for
N No-tube-in-window bundle calculation of pressure drop. Most of the authors have assumed a
2nd letter P Plain tube model wherein, fluid flow is perpendicular to tube bundles. How-
1st number 6 or 8 Number of cross passes ever Kapale et al. [10] have proposed a model which takes into
2nd number 10 or 14 in Nominal size of nozzles
10 in. size: 0.241 m (9.50 in.)
account direction of fluid flow. They consider fluid flow in the win-
14 in. size: 0.337 m (13.25 in.) dow region to be parallel to the tubes and that in the mid-section
3rd number 30 to 90 deg Tube layout pattern to be flowing at an angle. In the present investigation the fluid flow
Last digit 15.5 to 29.8 percent Baffle cut as percentage of inside direction adopted is different from that of the model of Kapale
diameter
et al. [10].
The flow pattern for this model for inter-baffle region and inlet
and outlet sections are shown in Figs. 3–5. Figs. 3 and 4 represent
the flow in inter-baffle region, while Fig. 5 represents flow in both
inlet and outlet sections. The flow patterns in inter-baffle region
obtaining the characteristics of an element, assembly of all differ from that for inlet and outlet section as depicted in Figs. 3–5.
elements, incorporating boundary conditions and solving the These flow patterns decide the angle w (Appendix A) at which
system of equations. the fluid crosses the tube bundle. The correction factor given by
In FEM, to obtain the characteristics of the entire domain for Zukauskas [11], which is in the form of a graph is correlated by
any variables like temperature, pressure etc., a small region is Schlunder [22], Eqs. (3a) and (3b) give correlations to obtain yaw
selected which has typical characteristics of the entire domain. correction factors.
Using variational or weighted residual method, characteristic of Yaw correction factor for inline tube arrangement:
the sub-region is obtained. After obtaining the characteristic of
K w ¼ 1:107 expð0:301w2:412 Þ ð3aÞ
the element, all the elements are assembled. Nodal values like
temperature, pressure are then determined by inserting the Yaw correction factor for staggered tube arrangement:
boundary conditions and solving the system of equations.
K w ¼ 1:245 expð0:478w1:733 Þ ð3bÞ
2.1. Geometrical model
The Euler number (Eu) as obtained from [11] is multiplied by
correction factor K w and the corrected Euler number (Euc) is
The model requires the shell side of STHX to be discretised into
calculated:
many elements, with elements in window section, mid-section,
and inlet–outlet section as shown in Fig. 2. In the present model, Euc ¼ K w Eu ð4Þ
Zukauskas friction factor has been used. The discretization and
the number of elements taken should be such that, it should reflect The shell-side fluid flow is found to vary depending on the Shell
the situation (in rectangular tube bundles) Zukauskas [11] has diameter, baffle spacing and tubes in window section.
used for determining the friction factor. In this way 4 elements
gives satisfactory results. These 4 elements in one cross flow 2.3. FEM model
section are distributed with 1 element in each window section
and 2 elements in the mid-section to find pressure drop across that In the case of a STHX, the fluid flow on its shell side is complex.
section. The flow path of a shell side fluid is first determined as it decides
The longitudinal and transverse pitches for various tube arrays the characteristics of the element (shown in Section 2.2). The
are as given by Shah et al. [21] and are illustrated in Table 2. friction factor required to find the pressure drop is selected (in
For inline tube arrangement, the minimum cross flow area used the present case Zukauskas correlations) for each of the sections.
is given by Stiffness matrix for each of the element is then determined. These
stiffness matrices are assembled to obtain the global stiffness
ðxt do ÞSxðiÞ
Amin ¼ þ Sxe ð1Þ matrix. On the load vector, known boundary conditions are
xt inserted and pressure at each of the nodes is determined (see
whereas for staggered arrangement, the minimum cross flow area is Fig. 6). Typical example for pressure drop in a pipe network is
given by available in Lewis et al. [23].
In the present model the whole HX is discretised into a number
ðxt do ÞSxðiÞ
Amin ¼ þ Sxe for xt do 6 2ðxd do Þ ð2aÞ of elements depending on the number of baffles. Between the baf-
xt fles, it is discretised into 4 elements in the direction of flow as illus-
or trated in Fig. 2(a) and the evaluation for pressure drop is performed
element by element using Zukauskas correlation for rectangular
2ðxd do ÞSxðiÞ
Amin ¼ þ Sxe for xt do > 2ðxd do Þ ð2bÞ tube bundles [11]. The formula to be used for calculating pressure
xt drop is discussed in reference [24]. Using Euc, the pressure drop for
Table 2
Properties of tube banks [21].
30° Triangular staggered array 60° Rotated triangular staggered array 90° Square inline array 45° Rotated square staggered array
pffiffiffi pffiffiffi
Transverse tube pitch (xt) pt 3pt pt 2pt
pffiffi pt ptffiffi
Longitudinal tube pitch (xl) 3 pt p
2 pt 2 2
704 B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712
ψw
ψm
ψw
ψm = ψw ψm = ψw
ψw ψw
ψm ψm
ψw
ψw
Table 3
Correction factors for 60° and 45° tube arrangement.
ψm
pressure drop due to sudden expansion at inlet nozzle and sudden
contraction at the exit nozzle as discussed by Gaddis et al. [7]:
ψw
2A2nozzle
k¼ ð5bÞ
kn q Q
Table 4
Prediction of pressure drop with tubes in window region and comparison with experimental results.
Heat exchanger configuration Volume flow rate Q (m3/s) Experimental pressure drop, DPexp (kPa) FEM model pressure drop, DPfem (kPa) Deviation (in%)
(a)
F P8 1400 30 25.5 0.090 74.240 75.385 1.542
0.133 157.750 153.131 2.928
0.150 198.980 191.204 3.908
0.180 282.890 268.533 5.075
0.200 346.690 327.276 5.600
F P8 1000 30 25.5 0.050 26.550 27.844 4.874
0.080 65.160 63.383 2.728
0.100 99.790 94.630 5.171
0.120 141.360 131.916 6.681
0.130 157.730 152.823 3.111
0.160 244.870 224.600 8.278
0.188 333.210 303.845 8.813
F P8 1400 90 25.5 0.063 28.900 30.150 4.324
0.080 45.830 46.608 1.697
0.100 70.497 70.355 0.202
0.120 100.230 98.838 1.389
0.140 134.960 132.058 2.150
0.164 183.160 178.172 2.723
F P8 1000 90 25.5 0.063 31.900 31.554 1.085
0.080 50.590 48.872 3.395
0.100 77.820 73.893 5.047
0.130 129.114 120.836 6.412
0.150 170.184 158.404 6.922
0.180 241.960 224.167 7.354
0.215 339.990 315.159 7.303
Heat exchanger configuration Volume flow rate Q (m3/s) Experimental pressure drop, DPexp (kPa) FEM model pressure drop, DPfem (kPa) Deviation (in%)
(b)
F P6 1400 30 28.9 0.070 28.300 28.627 1.154
0.085 40.660 40.198 1.136
0.090 45.280 44.466 1.797
0.100 55.140 53.620 2.757
0.120 77.540 74.392 4.060
0.135 96.640 92.127 4.670
F P6 1000 30 28.9 0.075 37.740 34.268 9.199
0.085 47.450 42.754 9.896
0.100 63.890 57.158 10.537
0.120 89.190 79.487 10.879
0.150 134.170 119.672 10.806
0.180 187.310 167.886 10.370
0.205 237.640 214.196 9.865
F P6 1400 90 29.6 0.050 11.290 12.249 8.496
0.070 21.190 22.431 5.856
0.090 33.900 35.460 4.600
0.100 41.284 43.040 4.253
0.130 67.430 70.044 3.876
0.160 99.420 103.439 4.043
0.189 135.760 141.796 4.446
F P6 1000 90 29.6 0.078 28.970 29.454 1.670
0.090 38.290 38.326 0.093
0.100 47.020 46.578 0.939
0.120 67.100 65.427 2.493
0.140 90.630 87.401 3.563
0.160 117.590 112.497 4.331
0.176 141.600 134.823 4.786
Heat exchanger Volume Experimental FEM model pressure drop using Deviation using FEM model pressure drop Deviation using
configuration flow rate Q pressure drop, Zukauskas correction factor, Zukauskas correction using new correction factor, new correction
(m3/s) DPexp (kPa) DPfem (kPa) factor (in%) DPfem (kPa) factor (in%)
(c)
F P6 1400 45 29.8 0.050 10.230 9.756 4.630 10.306 0.741
0.070 19.445 16.740 13.912 19.869 2.183
0.100 38.430 30.928 19.522 38.887 1.189
0.120 54.440 42.378 22.156 54.724 0.522
0.140 73.074 55.445 24.124 73.132 0.080
0.174 110.690 81.432 26.432 110.456 0.211
F P6 1400 60 29.6 0.066 22.285 30.270 35.833 23.903 7.262
0.080 32.118 41.840 30.270 33.156 3.230
0.100 49.078 60.778 23.839 49.662 1.189
Table 4 (continued)
Heat exchanger Volume Experimental FEM model pressure drop using Deviation using FEM model pressure drop Deviation using
configuration flow rate Q pressure drop, Zukauskas correction factor, Zukauskas correction using new correction factor, new correction
(m3/s) DPexp (kPa) DPfem (kPa) factor (in%) DPfem (kPa) factor (in%)
0.120 69.395 82.600 19.028 69.530 0.195
0.140 93.010 107.370 15.439 94.542 1.647
0.157 115.640 130.778 13.090 115.249 0.338
Heat exchanger Volume flow rate Q (m3/s) Experimental pressure drop, DPexp (kPa) FEM model pressure drop, DPfem (kPa) Deviation (in%)
configuration
(d)
F P6 1400 45 15.5 0.050 24.240 24.928 2.838
0.070 48.170 48.332 0.336
0.090 79.230 79.091 0.176
0.100 97.610 97.154 0.468
0.130 164.090 162.096 1.215
0.150 217.840 214.411 1.574
0.174 292.260 286.789 1.872
F P6 1400 60 15.5 z0.050 28.996 28.898 0.336
0.070 55.696 56.178 0.865
0.090 90.690 89.388 1.436
0.100 111.260 109.782 1.329
0.125 171.530 172.778 0.728
0.140 213.710 213.297 0.193
1 1 Pi 0 drop for STHXs with tubes in window agree well with experimen-
¼ ð7Þ
1 1 Pj 0 tal values with a deviation of 0.5% to +7.3% as seen from Table 4c.
The friction factors with their correction factors when applied
Apart from the elements defined as shown in Fig. 2 STHXs with low to a low baffle cut of 15.5% gave a deviation of more than 50% from
baffle cuts will also have spring elements just at the baffle space to experimental values. The reason for this deviation being, the minor
take into account the minor losses which will be discussed in Sec- losses encountered at the baffle end in the 15.5% baffle cut due to
tion 3.2. The coefficient k for the stiffness matrix of this spring ele- orifice effect. The effect of minor losses increases for lower baffle
ment is given by cuts and this is also evident from the pressure drop distribution
illustrated by Halle et.al [20], where the pressure drop at 15.5% baf-
fle cut is 60% higher than other baffle cuts. The minor loss coeffi-
A2bmin
k¼ ð8Þ cients (kl) for 15.5% baffle cut are represented in Eqs. (9) and (10):
kl qQ
K l ¼ 0:315 For 60 tube bundle ð9Þ
The minimum cross flow area at the baffle is calculated using Eq.
3.1. Comparison with experimental value and new correction factors (11), and the pressure drop due to minor loss is determined by
Eqs. (6) and (8):
Zukauskas [11] suggested friction factors are used for staggered
2 2
(30°) with a/b ratio as 1.155 and inline (90°) tube bundles with 1 Ds do
(a 1)/(b 1). For any other tube configuration Zukauskas [11]
Abmin ¼ ðhb sinhb Þ Ntw p ð11Þ
2 2 4
have proposed correction factors to be multiplied with correspond-
ing friction factor of staggered (30°) and inline (90°) configuration. The pressure drop in HX with tubes in window section (Table 4)
But when Zukauskas friction factor, along with his correction factor mostly deviated by less than ±6%. However, the maximum deviation
was applied to 45° and 60° tube bundle, this model gave a devia- (10.9%) is found in cases with nozzle diameter of 10 inches, prob-
tion of +27% and 36% respectively (Table 4c). Additional correc- ably, by the use of the nozzle pressure drop coefficient as suggested
tion factors k45 and k60 are proposed respectively for 45°and 60° by Gaddis et al. [7] which is 2 (for both the nozzles combined). HX
tube bundles as given in Table 3, on lines similar to Kapale et al. with NTIW (Table 4) deviated by less than ±7% in most of the cases.
who modified the friction factors presented by Kern [2].
The present method with 4 elements in a cross flow path, is 3.2. Present prediction of DP and its comparison with available values
applied to most of the cases presented by Halle et al. [20]. The
results are shown in Tables 4 and 5. The experimental results given The present predictions of DP are compared with those avail-
by Halle [20], has innate error of about 5% as discussed by Kistler able in literature (Table 6). The present method predicts the pres-
and Chenoweth [25]. It is found that, for 30° and 90° tube bundles sure drop with greater accuracy when compared to most of the
as shown in Tables 4a,b and Tables 5a,b, the pressure drops are other models. Zukauskas [11] has already provided well estab-
compared with values reported by Halle et al. [20] and the devia- lished results for friction factors for Reynolds number ranging 10
tion is less than ±5% for most of the cases. Therefore, the use of fric- to 106. The earlier authors have applied their models for Reynolds
tion factors proposed by Zukauskas [11] in the present model number 103 to 105. However, Gaddis et al. [7], have used the fric-
yields good results. tion factors for Reynolds number range from 10 to 105 with devi-
The correction factors are to be multiplied with the standard ations in predicted pressure drops up to ±35% and in some cases
friction factors given by Zukauskas [11] for staggered (30°) tube beyond that. Thus, in the present model, the predictions are
bundles with a/b ratio of 1.155 and inline (90°) tube bundles with extended beyond the Reynolds number 105 upto 106, in view of
(a 1)/(b 1) as 1. Using these correction factors, the pressure the friction factors of Zukauskas used in the present method.
B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712 707
Table 5
Prediction of pressure drop with no NTIW region and comparison with experimental results.
Heat exchanger configuration Volume flow rate Q (m3/s) Experimental pressure drop, DPexp (kPa) FEM model pressure drop, DPfem (kPa) Deviation (in%)
(a)
N P8 1400 30 25.5 0.100 46.070 45.123 2.056
0.130 73.690 72.159 2.078
0.160 106.860 105.307 1.454
0.200 159.320 159.008 0.196
0.240 220.800 223.571 1.255
0.280 290.960 298.994 2.761
0.316 361.295 376.161 4.115
N P8 1000 30 25.5 0.075 29.110 29.254 0.494
0.090 40.280 40.335 0.137
0.120 67.210 67.563 0.525
0.150 99.980 101.540 1.560
0.180 138.320 142.264 2.851
0.210 181.985 189.735 4.258
0.251 249.980 265.521 6.217
Heat exchanger configuration Volume flow rate Q (m3/s) Experimental pressure drop, DPexp (kPa) FEM model pressure drop, DPfem (kPa) Deviation (in%)
(b)
N P8 1400 90 25.5 0.064 18.540 17.031 8.137
0.090 35.320 31.576 10.600
0.120 60.840 53.624 11.861
0.150 92.750 81.310 12.335
0.170 117.510 102.898 12.435
0.200 159.760 139.978 12.382
0.230 208.058 182.694 12.191
0.251 245.420 215.948 12.009
N P6 1000 90 29.6 0.037 3.426 3.105 9.359
0.060 8.378 7.489 10.610
0.090 17.740 15.733 11.315
0.120 30.200 26.822 11.186
0.160 51.430 46.028 10.504
0.200 77.710 70.283 9.557
0.230 100.640 91.788 8.795
0.251 118.300 108.532 8.257
Heat exchanger configuration Volume flow rate Q (m3/s) Experimental pressure drop, DPexp (kPa) FEM model pressure drop, DPfem (kPa) Deviation (in%)
(c)
N P6 1000 30 28.9 0.102 24.281 23.966 1.298
0.130 37.574 37.058 1.373
0.160 54.601 54.121 0.879
0.200 81.591 81.755 0.202
0.230 104.929 106.143 1.157
0.261 131.747 134.640 2.196
N P6 1400 45 15.5 0.050 14.465 14.298 1.154
0.080 36.170 36.337 0.460
0.100 55.890 56.188 0.534
0.130 93.220 93.344 0.133
0.150 123.220 123.067 0.124
0.177 170.160 169.550 0.359
Heat exchanger Volume flow rate Q Experimental pressure drop, DPexp FEM model pressure drop using new correction factor, Deviation
configuration (m3/s) (kPa) DPfem (kPa) (in%)
(d)
N P6 1400 60 29.6 0.051 5.498 4.955 9.881
0.080 12.420 11.222 9.643
0.100 18.600 16.448 11.571
0.130 29.910 29.665 0.821
0.170 48.600 46.502 4.317
0.203 67.000 64.377 3.915
N P6 1400 60 15.5 0.054 17.130 18.034 5.277
0.070 28.130 29.268 4.047
0.090 45.460 46.150 1.518
0.130 91.750 98.218 7.049
0.160 136.410 142.498 4.463
0.189 187.510 198.329 5.770
Pressure drop predicted for the case of Reynolds number of 106, The methods provided by Bell [5] and Kern [2] show deviations
deviates from the experimental value by the order of 9%. Thus up to 40% in calculating pressure drop, as mentioned in the work
extending the validity of the present model beyond the Reynolds by Prithviraj and Andrews [19]. This deviation is observed because
number of 105 to the case of Reynolds number of 106. their methods are unable to predict nozzle pressure drop. Also,
708 B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712
It can also be seen that the method by Gaddis et al. [7], which is
Determine geometrical variables
currently the widely accepted pressure drop model, predicts pres-
for each element:
xl, xt, x(i), Amin(i), necf, new sure drop with lower accuracy compared to the present model.
HEATX results are promising, but since it’s a CFD simulation,
and time consuming it cannot be applied to new cases in the
Find Reynolds number at each element and absence of the availability of CFD tools and complete geometric
corresponding Euler number (Eu)
details.
Although Kapale et al. [10] model gives better results, the pro-
For window section elements For mid-section elements
cess of calculation using [10] is time consuming and tedious. Also,
Kapale et al. [10] have not shown the validity of their model for all
the configurations presented by Halle et.al [20]. Just like Bell [5]
Determine Yaw angle at window Determine Yaw angle at mid
section (ψw) using Fig 3a, Fig 4a section (ψm) using Fig 3b, Fig 4b and Gaddis [7], Kapale’s [10] method involves cumbersome calcu-
and Fig 5 Appendix 1. and Fig 5 lations to arrive at pressure drop coefficient. Kapale’s method also
involves referring to works of Gaddis et al. [7] to find various fac-
tors and hence, involves quite a lot of work in obtaining various
parameters required to calculate the pressure drop. Friction factors
Determine yaw correction factor (Kψ)
proposed by Kern [2] have been modified by Kapale et al. [10] by
and corrected Euler number (Euc ) introducing a correction factor to develop a new empirical formula.
But their correction factors to Kern’s friction factor are applicable
only for 3 tube pitches, 1.25, 1.33 and 1.50.
Table 6b presents the prediction of DP and its comparison with
Low baffle
other works for the case of NTIW. For the 2 cases considered, it can
cut ?
be observed that FEM predictions are quite close to the experimen-
If no tal values. It can also be observed that the present method predicts
the pressure drop in the NTIW better than the predictions of Bell
If yes
and Gaddis. However, the predictions from HEATX are close to
Calculate Abmin and determine the spring the experimental values with a penalty of taking very large time
element stiffness matrix coefficient of the order of 20 h for the predictions.
The proposed model in this paper predicts the shell side pres-
sure drop of a shell and tube heat exchanger using the concept of
Construct all Element stiffness matrix Finite Element Method. This method is simple to apply and the
predictions are quite close to the experimental values. The model
has been successfully tested for shell and tube heat exchangers
Assemble element matrices to find with baffle cut in the range of 25% to 30%. And also, for a minimum
Global stiffness matrix and load vector baffle cut of 15.5% the predictions are quite good. From this inves-
tigation the following conclusions can be drawn:
Apply boundary condition The present model is simple and is able to predict pressure drop
with minimum geometrical details.
The model has been successfully tested for shell and tube Heat
Solve for the system of equations Exchangers with baffle cut from 25 to 30%, and can be applied
with confidence up to a minimum baffle cut of 15.5%.
This model takes considerably less computation time to predict
the pressure drop compared to all other available models.
Pressure distribution and pressure drop
The pressure drop can be predicted up to any point, along the
flow, on the shell side of a shell and tube Heat Exchanger.
Fig. 6. Algorithm for the present model. The present model is applicable to the case of no tubes in win-
dow section which many other models do not have a provision
to calculate. Flow directions predicted in this shell and tube
Kern’s [2] method as pointed out by Prithviraj and Andrew [19], is Heat Exchanger take into account the effects of baffle spacing,
specifically designed for units with a baffle cut of 25%. Kern’s baffle cut and no tubes in window section. By using these flow
method does not have a provision to account for NTIW bundles directions, the predictions of pressure drop are quite close to
and so cannot be used to predict the pressure drop for NTIW bun- the experimental values for 240 cases.
dles. Bell’s [5] method rightly predicts the trend of the number of It was found out that for 30° and 90° tube bundles, Zukauskas
baffle on the overall pressure drop. Although, Kern’s method has friction factor gave agreeable results but for 45° and 60° correc-
fewer calculation steps compared to that of Bell-Delaware, the tion factors had to be proposed. These correction factors were
complexity still persists. This can be seen in the published works verified after applying the same for 130 experimental points.
of Kapale et al. [10] and Prithviraj and Andrews [19], the values
of pressure drop calculated using Kern’s method yields different
results, for the same HX and with the same mass flow rates. This Conflict of interest
clearly shows that the analytical methods of Bell [5] and Kern [2]
are difficult to implement. None declared.
B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712 709
Table 6a
Prediction of DP and validation for shell and tube heat exchangers with tubes in window section.
Heat exchanger Volume flow rate Exp. pressure drop FEM model pressure drop, Percentage (%) deviation from experimental values
configuration Q (m3/s) DPexp (kPa) DPfem (kPa)
Present Kapale’s HEATX Taborek Gaddis Kern
model model [10] [19] [21] [7] [2]
F P6 1400 30 28.9% 0.08 36.32 36.14 0.5 3.1 6.3 4.1 14.8 26.5
F P6 1400 30 28.9% 0.1 55.14 53.62 2.8 1.4 5.6 5.8 15.4 25.2
F P6 1400 30 28.9% 0.12 77.54 74.39 4.1 0.0 7.1 7.2 16 17.6
F P6 1400 30 28.9% 0.133 93.97 89.66 4.6 0.8 7.4 8 28 20.5
F P8 1400 30 25.5% 0.133 157.75 153.13 2.9 2.4 3.1 20.1 9.2 59
F P6 1000 30 28.9% 0.133 107.65 95.91 10.9a 3.9 2.8 9.5 10.4 5.1
F P6 1400 45 29.8% 0.05 10.19 10.31 1.2 4.0 1.5 27.5 22.9
F P6 1000 90 25.5% 0.215 339.99 315.16 7.3 2.3 19.5 27.9 18.1
a
Low nozzle diameter case.
Table 6b
Prediction of DP and validation for shell and tube heat exchangers with no tubes in the window section and comparison of CFD simulation [19].
Heat exchanger Volume flow rate Q Exp. Pressure drop DPexp FEM model pressure drop, DPfem Percentage (%) deviation from experimental
configuration (m3/s) (kPa) (kPa) values
Present HEATX Bell Gaddis
model [19] [6] [7]
N P8 1400 30 25.5% 0.133 39.9 38.62 3.2 1.5 5.4 14.4
N P6 1000 30 28.9% 0.133 77.1 75.20 2.5 4.7 12 13.8
This work is supported in part by the TEQIP 1.2.1 research grant HX configuration – HX with Tubes in window section
(World Bank), for the Centre of Excellence in Knowledge Analytics Inner diameter of the shell (Ds) – 0.59 m
and Ontological Engineering (KAnOE) at PES Institute of Technol- Diameter of outer tube limit (Dotl) – 0.568 m
ogy, Bangalore 560085, India. Outer diameter of tubes (do) – 0.191 m
Tube pitch (p) – 1.25
Appendix A Tube bundle array angle – 30°
Baffle cut – 28.9%
Calculation of Ww and Wm. Number of baffles – 5
Length of each cross path (S) – 0.597 m
Heat Heat exchanger Ww wm (radians) Inlet and outlet nozzle diameter – 0.337 m
exchanger section type (radians) Volume flow rate (Q) – 0.085 m3/s
case q – 997.0479 kg/m3
D
D
l – 8.9 104 Pa s
Ds Tubes in window
S 61 tan1 S
s
tan1 s
S
section Here, the elements are labeled in each section according to
No tubes in window tan1 4lc
tan1 Ds 2lc
direction of fluid flow, i.e. when the fluid enters the section, the
3S S
section 4
element it encounters first is the first element and the element
p
Inlet and When tan1 2lc
2 from which it leaves that section is the 4th element.
Se
outlet fluid
section enters Step 1. Determination of geometrical variables from the given
section geometrical data
When p
2
fluid
lc ¼ Baffle cut percentage Ds ¼ 0:289 0:59 ¼ 0:1705 m
exits
section
Ds
a ¼ 1:25; b ¼ 1:0825
>1 Tubes in window tan1 3lc
tan1 Ds 2lc
S 2S S
section 4
xt ¼ 0:0239 m; xl ¼ 0:0207 m
No tubes in window tan1 2lc
tan1 Ds 2lc
S S
section 2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
Inlet and When 1 2lc p Ds 2lc Dotl
tan Se
2 x1 ¼ 2 þ ¼ 0:5105 m
outlet fluid 2 2
section enters
section x2 ¼ 0:5680 m
When p
2
fluid xe ¼ Ds Dotl ¼ 0:0220 m
exits
section 0:5105 þ 0
xð1Þ ¼ ¼ 0:2553 m
2
710 B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712
Step 2. Determination of Reynolds number for the given volumetric Step 4(a) For inter-baffle region
flow and Euler number n(1) and n(4) correspond to number of rows of tubes in win-
dow sections (new)
Euler number (Eu) for the Reynolds number is either deter- n(1) = n(4) = 6.17
mined from the correlations presented in Schlunder [22] or directly Using Eq. (5a),
from the Zukauskas graphs [11,23]. k(1) = k(4) = 3.8014 105
Re(1) = A qQdð1Þ0 l = 41703.3108, for which Euler number Eu(1) is n(2) and n(3) correspond to number of rows of tubes in win-
min
P1
P2 ∆P1
P3 ∆P2
P4 ∆P3
∆P4 P
5
Fig. 7. Elements of inter-baffle region. P1, P2, P3, P4, P5 are the pressures at nodal points 1 to 5, and DP1, DP2, DP3, DP4 are the pressure drop in the elements of the section.
B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712 711
∆P4 P5
P4
∆P3
P3
∆P2 P2
P1 ∆P1
Fig. 8. Elements of inlet–outlet region. P1, P2, P3, P4, P5 are the pressures at nodal points 1 to 5, and DP1, DP2, DP3, DP4 are the pressure drop in the elements of the section.
P1
∆P1 P2
Fig. 9. Entry and exit Nozzle element. P1, P2 are the pressures at nodal points 1 and 2 respectively, and DP1 is the pressure drop in the element of one nozzle section.
Step 4(b) For Inlet and exit section k k Pi Q
¼
n(1) and n(4) correspond to number of rows of tubes in win- k k Pj Q
dow sections (new)
n(1) = n(4) = 6.17
Using Eq. (5a), 1:87756 1:87756 P1 0:085
104 ¼
k(1) = 8.0755 105 1:87756 1:87756 P2 0:085
k(4) = 2.2657 105
n(2) and n(3) correspond to number of rows of tubes in win-
P1 452:7146
dow sections (necf) ¼
n(2) = n(3) = 6.021 P2 0
Using Eq. (5a), Therefore, net pressure drop from inlet and outlet noz-
k(2) = k(3) = 6.3178 105 zles = (DPnz) = 2 0.4527 = 0.9054 kPa (see Fig. 9).
Step 4(d) Low baffle cut pressure drop
k k Pi Q
¼
k k Pj Q If the baffle cut is low, then the sudden-expansion and con-
traction losses should be taken into account by using Eq. (8)
Assembly of the elements give: in (6).
2 38 9 Here, baffle cut is not low, hence pressure drop due to baffle
0:80755 0:80755 0 0 > P1 > 0
> >
> >
6 7> > cut of 28.9%, (DPbl) = 5 0 = 0
6 0:80755 1:43933 0:63178 0 0 7>
> P > >
6 < 2>
7> =
6 7
104 6 0 0:63178 1:26356 0:63178 0 7 P 3 Net pressure drop determined by FEM model ðDPfem Þ
6 7>
> >
60 0:63178 0:85835 0:22657 7>P > >
4 0 > 4>
5>
> >
> ¼ DPibr þ DPior þ DPnz þ DPbl
: >
> ;
0 0 0 0:22657 0:22657 P5
8 9 ¼ 24:3032 þ 14:99 þ 0:9054 þ 0 ¼ 40:1986 kPa
> 0:085 >
>
> >
>
>
> >
>
> 0 > >
>
< = Experimental pressure drop ðDPexp Þ ¼ 40:66 kPa:
¼ 0
>
> >
>
>
> 0 > >
>
> >
>
>
: >
; References
0:085
Let P5 be 0. [1] D.A. Donohue, Heat transfer and pressure drop in heat exchangers, Ind. Eng.
Chem. Res. 41 (1949) 499–2511.
Solving the above matrix, we obtain the pressure distribu- [2] D.Q. Kern, Process Heat Transfer, McGraw-Hill, New York, 1950.
tion in one Inlet/outlet region and the results are as follows: [3] T. Tinker, Shell-Side Characteristics of Shell and Tube Heat Exchangers, Parts I,
II, III, General Discussion on Heat Transfer, Institution of Mechanical Engineers,
8 9 8 9 London, 1951. pp. 97–116.
>
> P1 > > 7:4950 103 >
> >
> >
>
>
>
>
>
>
3>
[4] T. Tinker, Shell-side characteristics of shell and tube heat exchangers: a
> > > >
< P2 >
> simplified rating system for commercial heat exchangers, Trans. ASME 80
= >
< 6:4424 10 >= (1958) 36–52.
3
P3 ¼ 5:097 10 [5] K.J. Bell, Final report of the cooperative research program on shell and tube
>
> > >
> > >
>
> > >
> P4 > > 3:7516 103 >
> heat exchangers, University of Delaware, Engineering Experimental Station,
>
> >
> >
> >
> Bulletin No. 5, 1963.
: ; : ;
P5 0 103 [6] J.W. Palen, J. Taborek, Solution of shell side flow pressure drop and heat
transfer by stream analysis method, Chem. Eng. Prog. Symp. Ser. 65 (1969) 53–
The net pressure drop in inlet and outlet region (DPior) = 63.
2 7.495 103 Pa = 14.99 kPa (see Fig. 8). [7] E.S. Gaddis, V. Gnielinski, Pressure drop on the shell side of shell and-tube heat
exchangers with segmental baffles, Chem. Eng. Process. 36 (1997) 149–159.
Step 4(c) Nozzle pressure drop [8] E.S. Gaddis, V. Gnielinski, Druckverlust in querdurchstriimten Rohrbiindeln,
The nozzle pressure drop is determined using Eq. (5b) in (6), vt.verfahrenstechnik 17 (1983) 410–418.
712 B. Parikshit et al. / International Journal of Heat and Mass Transfer 84 (2015) 700–712
[9] E.S. Gaddis, V. Gnielinski, Pressure drop in cross flow across tube bundles, Ind. [18] Rajagapal Thundil Karuppa Raj, Srikanth Ganne, Shell side numerical analysis
Eng. Chem. Res. 25 (1985) 1–15. shell and tube heat exchanger considering the effects of baffle inclination
[10] Uday C. Kapale, Satish Chand, Modeling for shell-side pressure drop for liquid angle on fluid flow, Therm. Sci. 16 (4) (2012) 1165–1174.
flow in shell-and-tube heat exchanger, Int. J. Heat Mass Transfer 49 (2006) [19] M. Prithiviraj, M.J. Andrews, Comparison of a three-dimensional numerical
601–610. model with existing methods for prediction of flow in shell-and-tube heat
[11] A.A. Žukauskas, Heat transfer from tubes in cross flow, Adv. Heat Transfer 18 exchangers, Heat Transfer Eng. 20 (2) (1999) 15–19.
(1987) 87–159. [20] H. Halle, J.M. Chenoweth, M.W. Wambsganss, Shell-side water flow pressure
[12] A.Y. Gunter, W.A. Haw, A general correlation of friction factors for various drop distribution measurements in an industrial-sized test heat exchanger,
types of surfaces in cross flow, Trans. ASME 67 (1945) 643–660. ASME J. Heat Transfer 110 (1988) 60–67.
[13] S.G. Ravikumaur, K.N. Seetharamu, P.A. Aswatha Narayana, Finite element [21] R.K. Shah, Dusan P. Sekulic, Fundamentals of Heat Exchanger Design, Wiley,
analysis of shell and tube heat exchangers, Int. Commun. Heat Mass Transfer New York, 2003.
15 (2) (1988) 151–163. [22] E.U. Schlunder, Heat Exchanger Design Handbook, vols. 1–4, Hemisphere, New
[14] Yonghua You, Aiwu Fan, Xuejiang Lai, Suyi Huang, Wei Liu, Experimental and York, 1983, http://dx.doi.org/10.2514/3.60159.
numerical investigations of shell-side thermo-hydraulic performances for [23] Roland W. Lewis, Perumal Nithiarasu, Kankanhalli N. Seetharamu,
shell-and-tube heat exchanger with trefoil-hole baffles, Appl. Therm. Eng. 50 Fundamentals of Finite elements for Heat and Fluid flow, John Wiley & Sons,
(1) (2013) 950–956. UK, 2004.
[15] Ender. Ozden, Ilker. Tari, Shell side CFD analysis of a small shell-and-tube heat [24] Frank P. Incropera, David P. Dewitt, Theodore L. Bergman, Adrienne S. Lavine,
exchanger, Energy Convers. Manage. 51 (5) (2010) 1004–1014. K.N. Seetharamu, T.R. Seetharam, Fundamentals of Heat and Mass Transfer,
[16] F. Vera-García, J.R. García-Cascales, J. Gonzálvez-Maciá, R. Cabello, R. Llopis, D. Wiley India Private Ltd., 2013.
Sanchez, E. Torrella, A simplified model for shell-and-tubes heat exchangers: [25] R.S. Kistler, J.M. Chenoweth, Heat exchanger shellside pressure drop:
practical application, Appl. Therm. Eng. 30 (10) (2010) 1231–1241. comparison of predictions with experimental data, ASME J. Heat Transfer
[17] G. Hewitt, G. Shires, T. Bott, Process Heat Transfer, CRC Press Inc., 1994. 110 (1988) 68–76.