Professional Documents
Culture Documents
ScienceDirect
w w w . i i fi i r . o r g
Article history: Airflow maldistribution in heat exchangers and air channels is a common problem in
Received 22 December 2014 HVAC&R systems. Furthermore, in the refrigeration field, temperature and velocity mal-
Received in revised form distribution downstream the evaporator can reach the discharge air grille, negatively
27 May 2015 influencing the heat removal from the exposed goods. In the present paper, the effect of
Accepted 31 May 2015 the flow maldistribution in the air channel of an open refrigerated display cabinet is
Available online 9 June 2015 studied by means of computational fluid dynamics simulations. An equivalent thermal
approach is presented: in combination with the equivalent resistance approach, it allows a
Keywords: drastic reduction of computational effort. This thermal CFD model allows to evaluate the
Air distribution 3D effects of the maldistribution in terms of temperature, heat transfer coefficients, and
Heat exchanger cooling power in the full channel model. Discussion of the results is reported.
Simulation © 2015 Elsevier Ltd and IIR. All rights reserved.
Display cabinet
Computational fluid dynamics
Heat transfer
* Corresponding author. Tel.: þ39 049 829 5731; fax: þ39 049 829 5728.
E-mail address: antonio.rossetti@itc.cnr.it (A. Rossetti).
http://dx.doi.org/10.1016/j.ijrefrig.2015.05.014
0140-7007/© 2015 Elsevier Ltd and IIR. All rights reserved.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 7 ( 2 0 1 5 ) 2 0 8 e2 1 5 209
Fig. 3 e U velocity component contour plot on plane 2: a) experimental results; b) numerical results.
Firstly, a detailed model of the fin, accounting for heat con- external surface was modeled as a model boundary. A struc-
vection and conduction was set up. Afterwards, an equivalent tured hexahedral grid with proper inflation layer was used,
thermal model of the fin was developed and validated by counting 650,000 nodes.
comparison with the full fin model. The air velocity and temperature are input conditions to
The use of CFD simulation to derive the equivalent model the model. The inlet temperature of 4 C was assumed ac-
allows the entire procedure, once validated, to be used to test cording to the typical average temperature of the Return Air
non-standard, or complex fins geometry. The proposed pro- Grille (RAG) of an ORDC. Relative pressure was set at the outlet
cedure can then be easily generalized without the need of any of the model.
previous information on the fin performance. SST turbulence model coupled to the two equation tran-
sition model (Menter et al., 2004) was used, in order to describe
4.1. Detailed single fin model the turbulentelaminar transition induced by the fins.
Constant temperature Tev ¼ 10 C (corresponding to the
The volume considered by the single fin model is sketched in evaporation temperature of the refrigerant) was imposed on
Fig. 5, and consists in half fin thickness and half-fin pitch air the surface corresponding to tube wall. Phase changes inside
channel, assuming symmetry conditions on both sides. The the tubes are not considered at this analysis stage. No thermal
height of the volume corresponds to the tube-pitch. Tubes resistance was assumed between the fin and the tubes, thus
modeling an ideal continuous surface including both fin and
tubes. Conduction was solved on the solid domain of the fin.
Results are presented in Fig. 6, were the temperature of the
fin and the flow streamlines are showed. The fin temperature
increases close to the leading edge, thanks to the convective
exchange with the hot air, while in the rest of the surface the
temperature is very close to the evaporation temperature. The
air flow is divided by the presence of the tube rows, and a low
speed volume develops in the wake of each tube. This model
will be used as a reference to discuss the accuracy of the
equivalent model developed in the next sections.
Q_
h¼ (1)
A T Tev
Fig. 9 e Nusselt Number as a function of time (a) and of the average fluid temperature (b).
the time the fluid takes to reach the entrance length is not
2
dependent on the velocity. Taking advantage of this relation, _
qðV; TÞ ¼ hðTÞ ðT Tev Þ (5)
b
entrance-related phenomena can be easily normalized using
time instead of length. When the permanence time inside the
4.3. Validation of the thermal equivalence
evaporator t ¼ x=V is used as reference, the heat exchanger
coefficients for different velocities collapse on a single curve,
A new model was built to validate the thermal equivalence.
as reported in Fig. 9a.
This model was defined using the same volume showed in
As a consequence, the convective exchange coefficient can
Fig. 5a, except for the fin, whose thermal effect was described
be expressed as function of the time h ¼ hðtÞ, not depending
adding Eq. (5) as a source term into the energy equation, while
directly upon velocity or position.
the fluid dynamic resistance was reproduced using the
Despite its apparent simplicity, the use of the permanence
directional losses of Table 1. The high value of the resistance
time in a CFD run is quite complex as it belong to a Lagrangian
along the y direction will assure the removal of the transversal
description, while CFD software is inherently Eulerian.
component of the velocity, deflecting the flow in the first
In order to bypass this problem, an available CFD variable,
millimeters of the fin, as in the actual geometry.
monotonically related to time, could be used instead. The fluid
As the tubes were described in the model their contribution
average temperature T was found to be eligible for this as it is
to the overall heat transfer was accounted by the CFD model
related to time by the power equilibrium equation, being the
relying on the boundary temperature Tev on the corresponding
heat flux always leaving the fluid. Heat exchange coefficient as
surfaces.
a function of the average temperature of the fluid is reported
Results are reported in Fig. 10, where the data obtained by
in Fig. 9b, where the interpolating curve h ¼ hðTÞ is reported by
the simplified model are compared with those of the 3D fin
a dashed line.
model presented in Section 4.1.
Based on this interpolation the heat exchange inside each
Results are in good agreement, especially for low velocities.
control volume can be described as:
The maximum error, defined as e ¼ Tsymplified T for velocities
lower than 1.1 ms1 is 0.25 C corresponding to 3% of the
Q V; T ¼ h T A T Tev (2)
minimum and is located in the first point of the highest ve-
Equation (2) should be then rearranged to obtain a locity curves. The average error is instead 0.12 C.
distributed model of the heat exchange. This term will be used The error derives from simplifications introduced in the
a source in the energy equation, accounting for the heat flux distributed equivalent method: the simplified model does not
induced by the fins. The expression of this term can be ob- account for fin conduction, for the boundary layers develop-
tained dividing Eq. (2) by the reference volume between the ment towards the fin length and for the interaction between
fins Vol, obtaining: the tubes and the fins.
Nevertheless, the performance of the proposed equiva-
A
q_ V; T ¼ h T T Tev (3) lence was satisfactory for the present study, considering
Vol
which can be simplified in
2
q_ V; T ¼ h T T Tev (4)
b
as A ¼ dL dz and Vol ¼ 0:5b dL dz,where dz is a fictional
thickness of the 2D model and dL is the considered element
length.
When Eq. (4) is used in the full channel model the average
temperature T of the detailed model can be replaced by the
local temperature T, as no significant gradient on the length
scale of b should reasonably arise, obtaining the punctual
relation: Fig. 10 e Validation of the thermal equivalence model.
214 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 7 ( 2 0 1 5 ) 2 0 8 e2 1 5
the relevant simplification introduced for the evaporator opposite is true for low velocity areas. It is worth reminding
model. that the heat transfer coefficient, as described in Section 5, for
the specific design is not very much dependent on velocity,
due to the fact that laminar flow takes place.
The effect of the maldistribution can be clearly seen on the
5. Channel thermo-fluidynamic model
cross sections of the channel downstream the evaporator. On
the evaporator outlet section (Fig. 12a) the temperature map
The isothermal model previously presented was used as
allows to clearly define three different areas: the tubes wakes,
reference to develop the thermo-fluid dynamic model. Air was
characterized by the lower temperature; the low speed area,
modeled as a dry ideal gas. The heat source expressed in Eq. (5)
on the left side of the channel; and the high speed areas, be-
was applied to the evaporator domain. Uniform temperature
tween the tubes and between the tubes and the walls. Mal-
(4 C) was imposed at the fans outlet surface.
distribution causes a temperature difference up to 2 C at the
The temperature and the velocity distribution on Plane 2
evaporator exit. Then mixing process between the high tem-
resulting from the analysis are reported in Fig. 11. The com-
perature jet and the tube wakes at lower temperature reduces
parison between the two plots highlight a strong relationship
this difference as the channel outlet is approached. The
between temperature and velocity patterns. In details, higher
average difference in temperature between the two sides of
velocities lead to higher outlet temperature, while the
the channel is approximately of 1 C at the outlet of the
channel.
The effect of the uneven air flow distribution on local heat
flux is illustrated in Fig. 13. The heat flux over the whole
evaporator length reflects velocity distribution as represented
in Fig. 11a, with higher velocity corresponding to higher
thermal fluxes.
The detailed description of the heat flux inside the evap- maldistribution in fin-and-tube evaporators. Int. J.
orator allows to quantify the effect of the inlet flow maldis- Refrigeration 34 (3), 696e706.
tribution on the evaporator performance. The overall heat Koern, M.R., Elmegaard, B., Larsen, L.F.S., 2013. Comparison of fin-
and-tube interlaced and face split evaporators with flow
transfer was integrated over the entire coil. The half of the
maldistribution and compensation. Int. J. Refrigeration 36 (1),
evaporator characterized by higher velocity, which is the top 203e214.
half in Fig. 13, was responsible for the 55% of the overall Mao, J.N., Chen, H.X., Jia, H., Wang, Y.Z., Hu, H.M., 2013. Effect of
cooling power, while the remaining 45% was exchanged by the air-side flow maldistribution on thermalehydraulic
lower velocity side. performance of the multi-louvered fin and tube heat
exchanger. Int. J. Therm. Sci. 73, 46e57.
Marinetti, S., Cavazzini, G., Fedele, L., De Zan, F., Schiesaro, P.,
6. Conclusions 2012. Air velocity distribution analysis in the air duct of a
display cabinet by PIV technique. Int. J. Refrigeration 35 (8),
2321e2331.
A thermo fluid dynamic model of an open refrigerated display Marinetti, S., Cavazzini, Lauri I., Testa, F., Minetto, S., 2013.
cabinet evaporator channel was proposed, in order to study Numerical and experimental analysis of the air flow
the effects of the air mal distribution inside and downstream distribution in the cooling duct of a display cabinet. In: 2nd IIR
the evaporator. A thermal equivalence model of the evapo- Conference on Sustainability and the Cold Chain. Paris (April
rator fins were proposed, in order to reduce the computational 02.04.13).
cost of the model, preserving the overall accuracy. The pro- Menter, F.R., Langtry, R.B., Likki, S.R., Suzen, Y.B., Huang, P.G.,
Volker, S., 2004, January). A correlation-based transition
posed approach allowed to use just 1.2 million nodes to
model using local variables: part Idmodel formulation. In:
describe the evaporator, saving at least a factor 10 in term of ASME Turbo Expo 2004: Power for Land, Sea, and Air.
nodes in respect to similar models where the complete ge- American Society of Mechanical Engineers, pp. 57e67.
ometry is detailed (Moukalled et al., 2011). Moukalled, F., Verma, S., Darwish, M., 2011. The use of CFD for
The model highlighted the consequences of the uneven predicting and optimizing the performance of air conditioning
airflow at the evaporator inlet, in terms of air temperature and equipment. Int. J. Heat Mass Transf. 54 (1), 549e563.
Rossetti, A., Minetto, S., Marinetti, S., 2015. Numerical Modelling
heat flux inside the evaporator. In particular the air maldis-
and Validation of the Air Flow Maldistribution in the Cooling
tribution was responsible for a 10% difference in terms of
Duct of a Horizontal Display Cabinet. Applied Thermal
thermal load between the right and left side of the evaporator. Engineering. http://dx.doi.org/10.1016/
j.applthermaleng.2015.04.032.
Singh, V., Abdelaziz, O., Aute, V., Radermacher, R., 2011.
references
Simulation of air-to-refrigerant fin-and-tube heat exchanger
with CFD-based air propagation. Int. J. Refrigeration 34 (8),
1883e1897.
Aslam Bhutta, M.M., Hayat, N., Bashir, M.H., Khan, A.R., Yashar, D.A., Domanski, P.A., Cho, H.H., 2011. An experimental
Ahmad, K.N., Khan, S., 2012. CFD applications in various heat and computational study of approach air distribution for a
exchangers design: a review. Appl. Therm. Eng. 32, 1e12. finned-tube heat exchanger. HVAC&R Res. 17 (1), 76e85.
Koern, M.R., Brix, W., Elmegaard, B., Larsen, L.F.S., 2011.
Performance of residential air-conditioning systems with flow