Comparing Circular and Rectangular Pipes in HT
Comparing Circular and Rectangular Pipes in HT
Research Article
Comparison of Heat Transfer between a Circular and Rectangular Tube Heat
Exchanger by using Ansys Fluent
Priyanka BishtȦ*, Manish JoshiȦ and Anirudh GuptaȦ
Ȧ
Department of Mechanical Engineering, Bipin Tripathi Kumaon Institute of Technology, Dwarahat, Almora, Uttarakhand (India) 263653
Accepted 10 June 2014, Available online 30 June 2014, Vol.4, No.2 (June 2014)
Abstract
Heat exchangers are the important equipments with a variety of industrial applications including power plants, chemical,
refrigeration and air conditioning industries. Circular tube heat exchangers are used in order to obtain a large heat
transfer area per unit volume and to enhance the heat transfer coefficient on the inside surface. This paper deals with the
CFD simulation of circular tube heat exchanger used for cooling water under constant wall temperature conditions.
CFD results are compared with the results obtained by the simulation of rectangular tube heat exchanger of the same
length under identical operating conditions. The results are validated by the results obtained by the numerical
correlations used by different researchers. Results indicated that circular pipe heat exchangers showed 2.5% increase in
the heat transfer rate over the rectangular tube. Simulation results also showed 8.5% increase in Nusselt number for the
circular tube whereas pressure drop in case of circular tube is higher when compared to the rectangular tube.
while shorter reviewer has appeared in handbooks (John commonly used mathematical equations to describe flow.
wiley and sons, 1987). (Gessner and Jones, 1976) The simulation is done based on the NS equations and
examined the turbulent flow in square cross section they then K-Epsilon model.
conducted a series of experiments using hotwire
anemometry to analyze fully developed turbulent flow in a
v x v y v z 0
square duct at a Reynolds number of 150,000. They also t x y z
carried out computations by a finite difference method
with an algebraic stress model to predict qualitatively the 4.2. Kappa-Epsilon Model
major feature of the flow field namely, eight-vortex
secondary flow structure. (Melling and Whitelaw, 1976) The K-epsilon model is most commonly used to describe
performed detailed experiments for fully-developed flow the behavior of turbulent flows. It was proposed by A.N
using laser- droppler anemometry and where the first to Kolmogrov in 1942, then modified by Harlow and
describe the axial velocity field and the Reynolds stress Nakayama and produced K- model for turbulence. The
distribution in detail. (Nakayama et al, 1983) on the other Transport Equations for K- model are for k, Realizable
k- model and RNG k- model are some other variants
hand, analyzed the fully-developed flow field in ducts of
of K- model. K- model has solution in some special
rectangular and trapezoidal cross-sections computationally
using a finite-difference method based on the algebraic
turbulence stress model of (Launder and Ying, 1972). cases. K- model is only useful in regions with turbulent,
They were able to obtain a flow field in good agreement high Reynolds number flow.
with the available experimental measurements for a
number of selected cross-sections. Improved calculations K Equation
were conducted by (Gessner and Po, 1977) and (De Muren
and Rodi, 1984) using the nonlinear algebraic stress model [u kx v kr ] x [( l t ) kx ] 1r r [r( l t ) kr ] g
of Rodi.
k k -
3. Mathematical Formulation
Where, G is the production term and is given by
The system consists of water flow moving through a
G = t [2{( r ) ( x ) ( r ) } ( r x ) ]
v 2 u 2 v 2 u v 2
circular and rectangular channel. The geometric model of
the circular tube and rectangular tube were constructed
using workbench in ANSYS 14 environment. In order to The production term represents the transfer of kinetic
numerically establish the heat transfer coefficient of energy from the mean flow to the turbulent motion
circular tube heat exchanger the parameters were assumed through the interaction between the turbulent fluctuations
to be same that of rectangular tube. Tube diameter was and the mean flow velocity gradients.
considered to be 0.015 m and length considered was 3 m.
The three dimensional computational domain modeled Equation
using quad mesh for both models are as shown in fig. 3.1.
The flow is assumed to be steady and turbulent. In this
numerical investigation, the following hypotheses are [u x v r ] [(
x l t ) x ] 1r r ( rl t ) r ]
adopted.
(i) Physical properties of water are constant. C S1G k
Cs 2 2
k
+
(ii) A profile of velocity is uniform at the inlet.
(iii) The radiation heat transfer is negligible. here Cs1 , Cs 2 , k and are the empirical turbulent
(iv) The flow is assumed to be steady. constant. The values are considered according to the
Launder et al., 1974. The values of Cμ, Cs1 , Cs 2 , k and
4. Governing Equation are 0.09, 1.44, 1.92, 1.0 and 1.3 respectively.
4.1. Continuity Equation
4.3 Boundary Condition
Continuity Equation also called conservation of mass.
Consider fluid moves from point 1 to point 2. The overall A turbulent flow is considered. The quantities U, k, ε are
mass balance is Input – output = accumulation. Assuming obtained by using numerical calculations based on the k-ε
that there is no storage the Mass input = mass output. model for low Reynolds Number. The boundary
However, as long as the flow is steady (time-invariant), conditions are listed below:
within this tube, since, mass cannot be created or
destroyed then the above equation. According to 1) At the inlet of the channel:
continuity equation, the amount of fluid entering in certain
volume leaves that volume or remains there and according u U in, v 0
to momentum equation tells about the balance of the
momentum. The momentum equations are sometimes also kin 0.005 Uin2
Kin stands for the admission condition for turbulent kinetic obtained. The values in the table indicate the properties of
energy and εin is the inlet condition for dissipation. circular duct for different meshing size.
The whole analysis is carried out with the help of software Table 6.2 Properties of working fluid water
“ANSYS Fluent 14.0”. ANSYS Fluent 14.0 is
computational fluid dynamics (CFD) software package to Density (kg/m3) 990
stimulate fluid flow problems. It uses the finite volume Specific heat Cp (J/kg-k) 4184
method to solve the governing equations for a fluid Thermal conductivity (W/m-k) 0.65
Geometry and grid generation is done using GAMBIT Kinematic viscosity (m2/s) 0.516 e-06
which is the pre-processor bundled with FLUENT. The Prandtl number (Pr) 3.15
two dimensional computational domain modeled using
hex mesh for 2-D models. The complete domain of 2-D
circular tube in all three cases have element size is 0.005 7. Results and Disscussion
m have 59925 nodes and 47138 element, and rectangular
tube in all three cases have element size is 0.002 m, and CFD computations were done for three different mass
115577 nodes 90000 Elements. Grid independence test flow rate of water 0.2624, 0.3498 and 0.5248 kg/s
was performed to check the validity of the quality of the respectively for both circular and rectangular tube.
mesh on the solution. Further refinement did not change Performance parameters adopted for comparison are Heat
the result by more than 0.9% which is taken as the transfer coefficient, Nusselt number and pressure drop in
appropriate mesh quality for computation. both the cases. In order to validate the CFD results the
important parameters like Nu and heat transfer
5.1 Sensitivity Analysis of the Mesh coefficient were calculated by using the different
correlations both for circular and the rectangular tube.
A non-uniform mesh in both horizontal and vertical Fig. 7.1 and Fig. 7.3 show the CFD simulated nusselt
directions proved to be sufficient to model the system. number plot vs. correlation values for circular and
rectangular tubes for the different mass flow rates. The
Table 5.1 Mesh sensitivity analysis results have shown a good agreement between the
correlation values used by different researchers and fluent
Element size (m) Nodes Element Umax Total pressure (pa) results as the average error is within 3% and 4.5% for both
0.003 422073 387220 1.94 1.47 e03
cases. Similarly fig. 7.2 and Fig. 7.4 shows the plot of
0.004 88682 68760 1.96 1.49 e03
heat transfer coefficients plot vs. correlation for both
0.005 59925 47138 1.97 1.50 e03
cases.
0.006 138652 103938 2.00 1.515 e03
Fig. 7.5 shows the variation of Nusselt number for
The meshing size is comparatively small near the circular and rectangular tubes. Nusselt number
boundaries so a good estimate of the gradients can be corresponding to the circular is higher than the rectangular
90 | International Journal of Thermal Technologies, Vol.4, No.2 (June 2014)
Priyanka Bisht et al Comparison of Heat Transfer between a Circular and Rectangular Tube Heat Exchanger by using Ansys Fluent
tube for all mass flow rates. This is because of the shape in Correlation CFD
the circular tube which aids the heat transfer. For different
Fig. 7.5 Variation Nusselt number with mass flow rate for
6000 circular and rectangular tube
4000
2000
0 circular rectangular
0.2 0.3 0.5 16000
Heat transfer coefficient(w/m2-K)
250
Mass flow rate(kg/s)
200
150
100 Fig. 7.6 Variation heat transfer coefficient with mass flow
50
rate for circular and rectangular tube
0
0.2 0.3 0.5
Fig. 7.6 shows the variation of heat transfer coefficient for
different mass flow rates. From the graph it is revealed
Mass flow rate(kg/s)
that as the mass flow rate increases heat transfer
coefficient also increases as expected since heat transfer
rate is proportional to the mass flow rate. Further for
Fig. 7.3 Comparison of correlation and CFD values of circular tube heat transfer coefficient has increased by
Nusselt number for rectangular tube 10% when the mass flow rate is increased from 0.2624 to
91 | International Journal of Thermal Technologies, Vol.4, No.2 (June 2014)
Priyanka Bisht et al Comparison of Heat Transfer between a Circular and Rectangular Tube Heat Exchanger by using Ansys Fluent
0.5248 Kg/s. Fig. 7.7 show the comparison of pressure Burmeister,(1993) L.C, Convective heat transfer, john
drops for the circular and rectangular tubes. Pressure drop wiley and sons Inc., New York .
for the circular tube is found to be more than the Bejan, (1993) A., Convective heat transfer, john wiley and
rectangular tube for all mass flow rates. Presence of sons Inc., New York., 1995.
secondary flow dissipates kinetic energy, thus increasing Kays, W.M and craford, M.E., Convective heat transfer,
the resistance to flow. For lower mass flow rates pressure McGraw-Hill., New York.
drop varies linearly whereas on increasing the mass flow Shah, T.K and London. A.L, (1978) Laminar flow forced
rate pressure drop varies exponentially as seen in the convection in ducts, Academic press, New York
graph. Shah, R.K. and Bhatti, M.S, (1987) chapter 3: Laminar
convective heat transfer in ducts, in handbook of single
circular rectangular phase convective heat transfer, eds.S. Kakac, R. K Shah
and W. Aung, john wiley and sons Inc., New York.
4000
F.B. Gessner and A.F. Emery, (1976) A Reynolds stress
3500
model for turbulent corner flows – Part I: Development
Pressure drop(pa)
3000
of the model, Journal Fluids Eng. 98, 261-268.
2500
Melling, A. & Whitelaw, J.H. (1976) “Turbulent Flow in a
2000
Rectangular Duct,” Journal of Fluid mechanics, Vol. 78,
1500
289-315.
1000
Nakayama, A., Chow, W. L., & Sharma, D. (1983)
500
“Calculation of fully Developed Turbulent Flows in
0
0 0.2 0.4 0.6
Ducts of Arbitrary Cross-Section,” Journal of Fluid
Mass flow rate(kg/s)
Mechanics, Vol. 128,199-217.
B.E. Launder and W.M.Ying., (1973) Prediction of flow
and heat transfer in ducts of square cross section”, Proc.
Inst. Mech. Eng., 187, 455-461
Fig. 7.7 Variation of pressure drop with mass flow rate for F.B. Gessner, and J.K. Po, (1976) A Reynolds stress
circular and rectangular tube model for turbulent corner flows – Part II: Comparison
between theory and experiment, Journal Fluids Eng. 98,
Conclusions 269-277.