ICFD14-EG-7052 Numerical investigation of cooling single wall mounted electronic component by jet impingement in cross flow configuration

Abstract

Computational study of cooling single wall mounted electronic component by jet impingement in cross flow configuration has been investigated. Five different turbulence models; standard k - ԑ, Realizable k - ԑ, RNG k - ԑ, Standard k–ω, and SST k–ω models were investigated. The results of flow structure as well as the local heat transfer coefficient have been investigated at Reynolds number ratio between the jet and cross flow, Rej / Rec = 2. Both standard and Realizable k - ԑ models predicted the main flow pattern while it failed to predict the LHV however, the RNG k - ԑ model predicted small LHV. Standard k–ω model predicted the same flow pattern while it captured a significant UHV and LHV. The SST k–ω model predicted approximately the same flow structure captured by standard k–ω model except the WV which is similar in shape similar to these captured by RNG k - ԑ model. The local heat transfer coefficient has been investigated on front, side, and rear faces. The effect of SV, WV, and LHV were clearly appear on the heat transfer distribution and the rear face had the lowest heat transfer coefficient. The SST k–ω model gave the more accurate prediction to the flow structure and heat transfer distribution.

Full text

1 Copyright © 2020 by ICFD14
Proceedings of ICFD14:
Fourteenth International Conference of Fluid Dynamics
18-19 December 2020, Le Méridien Pyramids Hotel, Cairo, EGYPT
ICFD14-EG-7052
Numerical investigation of cooling single wall mounted electronic component by
jet impingement in cross flow configuration
M. Khalil1*
1Department of Mechanical Engineering, Faculty of
Engineering, Sohag University, Sohag,82524, Egypt
M. Attalla2
2Department of Mechanical Engineering, Faculty of
Engineering, South Valley University, Qena 83521,
Egypt
Hussein M. Maghrabie2
2Department of Mechanical Engineering, Faculty of
Engineering, South Valley University, Qena 83521,
Egypt
H. E. Fawaz3
3Department of Mechanical Engineering, National Research
Centre, Dokki, Cairo12311, Egypt
ABSTRACT
Computational study of cooling single wall mounted electronic
component by jet impingement in cross flow configuration has
been investigated. Five different turbulence models; standard k -
ԑ, Realizable k - ԑ, RNG k - ԑ, Standard k–ω, and SST k–ω
models were investigated. The results of flow structure as well
as the local heat transfer coefficient have been investigated at
Reynolds number ratio between the jet and cross flow, Rej / Rec
= 2. Both standard and Realizable k - ԑ models predicted the
main flow pattern while it failed to predict the LHV however,
the RNG k - ԑ model predicted small LHV. Standard k–ω model
predicted the same flow pattern while it captured a significant
UHV and LHV. The SST k–ω model predicted approximately
the same flow structure captured by standard k–ω model except
the WV which is similar in shape similar to these captured by
RNG k - ԑ model. The local heat transfer coefficient has been
investigated on front, side, and rear faces. The effect of SV,
WV, and LHV were clearly appear on the heat transfer
distribution and the rear face had the lowest heat transfer
coefficient. The SST k–ω model gave the more accurate
prediction to the flow structure and heat transfer distribution.
KEYWORDS:
Jet Impingement, Cross Flow, Turbulence Models, Electronic
Component.
INTRODUCTION
Many Turbulence models are applied to solve the cooling
process of heated electronic components. Some use one-
equation model to characterize the turbulence statistics, another
solve two- equation model of mass, momentum and energy
equations to characterize turbulence statistics, others use more
than two equations in turbulent discretization. One-equation
models falls to predict the main turbulent flow structure. The
multiple turbulence equations models are used widely due to its
capability to calculate the small eddies statistics and whole flow
features. The two-equation turbulence models are the most
common models used in heat transfer problems. These models
are effective to solve the kinetic energy and dissipation rate
equations for turbulent flow with the conservation equations.
Moreover, it takes into account the turbulent flow
characteristics and the variation in fluid properties that
influence the performance of heat transfer.
The differences between turbulence models to capture the
flow structure around electronic components have been studied
by many researchers. Rodi [1] studied the flow structure around
a single cubical obstacle that is subjected to cross flow at Rec =
40,000 experimentally and numerically using k-ԑ model. The
results showed that the k-ԑ model failed to predict the detailed
flow structure around the component. Lakehal & Rodi [2]
conducted another numerical simulation using modified k-ɛ
model [3]. The numerical results were validated using the
experimental results of Martinuzzi et al. [4]. The modified k-ɛ
model showed the same main flow structure when compared
with the experimental results. Yaghoubi and Velayati [5] studied
the effect of flow Reynolds number on cooling process of
electronic component by cross air flow numerically using RNG
k-ɛ model. The flow structure and the local heat transfer
coefficient were matched with those obtained by the
experimental data. Chomdee and Kiatsiriroat [6].
Ratnam and Vengadesan [7] conducted a computational
study to investigate the performance of two-equation turbulence
models on the flow structure over a single cubical component
cooled by cross flow. Low Reynolds number k-ɛ, standard k-ɛ,
non-linear k-ɛ, standard k–ω, modified k–ω models were
studied. The results showed that the modified k–ω gave the

2 Copyright © 2020 by ICFD14
better prediction among all the two-equation models, and the
non-linear k-ɛ model gave better results compared with the
other k-ɛ models. Tominaga and Stathopoulos [8] used four
different k-ɛ turbulent models; Standard k-ɛ, RNG k-ɛ,
Realizable k-ɛ, and modified k-ɛ model that were proposed by
[9]. The results showed that, a little variation in capturing the
flow field is achieved because it cannot reproduce the reverse
flow near the walls. The RNG, Realizable, and modified k-ɛ,
relatively showed better agreement with the experimental
results of Saathoff et al. [10].
Different k-ɛ models have been investigated to study the
cooling process of electronic components using jet
impingement as well as jet impingement in cross flow. Dutta et
al. [11] conducted a numerical experiment using various RANS
models to study cooling process of a flat plate using jet
impingement. Standard k-ɛ, RNG k-ɛ, realizable k-ɛ, Launder
and Sharma low-Re k-ɛ [12], Chang–Hsieh–Chen low-Re k-ɛ
[13], standard k–ω, SST k–ω, and v2-f models have been used.
The results compared with the experimental data presented by
Ashforth-Frost et al. [14]. All turbulence models were validated
with the experimental data and both k-w models (standard and
SST) show better agreement with the experimental values of
Nusselt number at small values of nozzle to plate planes
distance. However, for higher value of nozzle to plate distance,
the standard k–ω and k-ɛ models accurately predicted the flow
structure. Another numerical study has been performed by
Kucinskas and Keith [15] to investigate the flow structure and
heat transfer characteristics of a cubical component cooled by
jet impingement in cross flow. The k-ɛ turbulence model of
COMSOL software has been used and validated with the
experimental results of Tummers et al. [16].
Cooling process using cross flow or impinging jet has a
significant effect on cooling characteristics of flat surface or
electronic components. The effect of cooling using single [17-
19], multiple jets [20, 21], or using jet impingement in cross
flow [JICF] [16] have been experimentally investigated widely.
In the present study five different two equations turbulence
models are used to predict the flow structure and its signature
on the local heat transfer distribution around an electronic
component subjected to JICF. For the same grid size, same
design, and same operating parameters the models are tested.
The accuracy of the solution and the computational details are
presented for the all models. The results are compared with the
experimental results of Masip et al. [22].
FLOW AND COMPUTATIONAL DETAILS
Geometrical description
Cooling an electronic components by two streams of air is
sketched in Fig. 1. The first fluid stream is a cross flow and the
second stream is an impinging jet directed normally onto the top
face of the component. The Reynolds number of cross flow
(Rec) based on the hydraulic diameter and flow velocity in the
channel, Uc = 3 m/s, equals to 2800. The Reynolds number of
jet flow (Rej) based on the jet velocity, Uj = 4 m/s, equals to
5560. The principal geometry dimensions of the square
component: the length (l), and height (b) are 25 and 3.5 mm,
respectively. The channel height (H = 2b), and channel width
(W = 8l). Moreover, the orifice diameter (Do= 0.8l), and the
orifice height (Ho = 0.4l) equals to 20 and 10 mm, respectively.
These dimensions satisfy the jet length-to-diameter ratio (Ho/Do
= 1.25) and Reynolds number ratio Rej/Rec = 2 [23, 24].
Governing Equations and Boundary Conditions
Steady-state Reynolds-averaged Navier–Stokes in the three-
dimensional and the energy equations formulate the governing
equations. The equations can be expressed using the Cartesian
tensor notation as follows:
Continuity equation, Momentum equation, and energy equations
The inflow conditions for both jet and channel flow are
assumed turbulent with uniform flow velocity. The channel
entrance is enough to avoid the entrance region. The inlet air
temperature ( for both cross and jet flow is constant at
27°C. Non-slip boundary conditions are applied at the all faces
of component with constant surface temperature ( of 70°C.
Moreover, all faces of the channel are adiabatic and non-slip
boundary conditions are applied at all solid surfaces. The flow
outlet is assumed to be in atmospheric pressure.
Numerical Method and Mesh Independence
A commercial CFD package ANSYS FLUENT 18.0 is
used to construct the computational grid, to discretize and to
solve the governing equations of cooling process of a single
wall mounted component with JICF. The second order upwind
differencing scheme is used in equations of momentum and
energy. SIMPLE (Semi Implicit Method for Pressure Linked
Equation) algorithm is used for coupling the pressure and
velocity terms.
For capturing the flow structure, the mesh pattern was
chosen to cover the flow structure details by keeping special
attention for mesh generation to optimize the number of cell and
minimize the running time. The shear layer regions, flow
interactions regions, and boundary layers regions near the walls
were carefully have sufficient mesh cells to guarantee better
simulation and larger cells size were further away the
interaction regions for economic purposes as shown in Fig. 2.
Four different grid sizes with statistics as presented in Table .1
are tested using standard k-ɛ turbulence model as shown in Fig.
3. The local Nusselt number on the component front face is

3 Copyright © 2020 by ICFD14
presented for independency test. Based on the criteria of
computational time consumed and the accuracy of local Nusselt
number, the selected mesh was G3 with total number of element
4.894 x 1006 hexahedral cells. The convergence of the
computational solution is determined based on the scaled
residuals for the continuity, momentum and energy equations.
Scaled residuals of 10-05 are set for all governing equations
while 10E-07 is set for energy equation.
Table. 1: Computational mesh details
Grid Size
Number of
Elements
Computational Time
[Hrs]
G1
0,957,320
7
G2
2,791,180
11
G3
4,894,608
14
G4
6,0 12,790
23
For the selected grid of Hexa8 elements, the statistics of
the mesh were acceptable. The skewness has a very small range,
up to 0.082, throughout the whole domain except the region of
the interference between the round jet body and the channel
domain where it is about 0.825. By resizing the elements on this
region, the skewness fall to 0.573 as shown if Fig. 4. The
element quality ranged between 41% and 99% and the average
was 73%.
Parameter Description
The Reynolds number based on the hydraulic diameter and
mean velocity is expressed as follow:
Where the hydraulic diameter is calculated from:
(5)
For the channel, and for the jet orifice,
The heat flux through the heated obstacles is presented as [25]:
However the heat flux ( can be calculated
during the simulations by solving the energy equation, hence the
average heat transfer coefficient ( ) can be presented as:
Results and Discussion
The flow structure around single component subjected to JICF
is investigated by Masip et al. [22] as shown in Fig. 5. It is
characterized as; Lower horseshoe vortex (LHV) is generated
before the front face of the obstacle as a result of interaction
between the flow and obstacle. Upper horseshoe vortex (UHV)
is produced due to the interaction between the jet and the cross
flow. Vortex ring (VR) is produced surrounding the jet core as
well. Arch-shaped wake vortex (WV) is produced after the rear
face as a result of flow separation at the trailing edges. Very thin
side vortex (SV) is formulated on both sides due to flow
separation at the leading edges.
Flow structure
The flow structure are presented on two perpendicular planes
are chosen to display the velocity vectors combined. A
horizontal plane (x-y plane) that is constructed above the
channel bottom wall slightly at z/b = 0.5, and vertical plane (x-
z plane) that passes through the middle of the obstacle at y/W =
0.5
The velocity vectors around the component on both x-z and x-y
planes when using Standard k-ɛ are presented in Fig. 6. It is
shown from x-z plane that the component is preceded by a thin
UHV and followed by a significant WV. The jet is bounded by a
VR. And there is no evidence on LHV before the front face.
Two cell of SV on both component sides are produced and the
WV has a two circular cells. It shown from Fig. 7. That the flow
structure on both x-z and x-y planes by using the Realizable k-ɛ
is similar to those captured by Standard k-ɛ except that the SV
and UHV are became larger.
The RNG k-ɛ model could predicted more significant UHV than
the other two k-ɛ models while it captured only half ring of VR
on the upstream direction as shown in Fig. 8. The LHV is
appeared by using this model than the other k-ɛ models with
small recirculation cell. Moreover, the WV captured by this
model has very small recirculation cell compared with the two
other k-ɛ model. The downstream half of the VR is joined with
a new back recirculation vortex that generated due to sweeping
the top separated shear flow downwash to the channel bottom.
The sweeping of the top flow downwash causing an unoccupied
area upwash it with is filled by the recirculation vortex that
absorb the VR with it.
The standard k-ω models could predicted the SV, VR, WV, SV,
UHV, and LHV as well. The UHV and LHV vortices have a
greater recirculation cells than those predicted by k-ɛ models as
shown in Fig. 9. The WV is significant similar to these
proposed by the standard k-ɛ model. Furthermore, the two cells
of the WV is stretched than those predicted by k-ɛ models. The
SST k-ω has a consistent prediction to the standard k-ω model
about the UHV, LHV, and SV while the VR and the WV is
similar to those predicted by RNG k-ɛ model as shown in Fig.
10. The SST k-ω model totally predicted flow structure similar
to these investigated by the experimental study of Masip et al.
[22].
Heat transfer
The distribution of local heat transfer coefficient at z/b =
0.5 along path ABCD on the; front (AB), side (BC), and rear
(CD) faces is shown in Figs. 11-13. It is observed that the
distribution of heat transfer coefficient has a concave profile on

4 Copyright © 2020 by ICFD14
the front face for all types of turbulent models as shown in Fig.
11. The variation of local heat transfer coefficient is a concave
shape with higher values near the leading edges and lowers at
the face mid span as a result of sweeping the flow towards the
leading edges. The values of local heat transfer coefficient for
both k-ω models is greater than these predicted by the different
k-ε models due to the LHV that produced significantly in case
of k-ω models than k-ε models. The distribution is
approximately the same for all k-ω models and for all different
k-ε models.
It is noticed from Fig. 12 that the distribution of local
heat transfer coefficient on the side face (BC) has its highest
value near the leading edge and then fall dramatically due to the
effect of the side vortex. Then it increases at the reattachment
point after the separation then fall gradually due to the
dissipation in flow momentum. All turbulence model could
predicted the heat transfer coefficient distribution while for the
Standard, Realizable, and RNG k-ε models, local heat transfer
coefficient increases rapidly at the prior of the path and then
decrease gradually until the face end. The local heat transfer
coefficient increases gradually until reaching the peak later for
both Standard and SST k-ω models.
The distribution of the local heat transfer coefficient on
the rear face is reported in Fig. 13. The effect of two cell WV in
significant in all k-ε models. The value of local heat transfer
coefficient is high at the trailing edge as a result of the separated
shear layer. Then the local heat transfer coefficient decreases
due to the recirculation cell then it increases gradually as a
result of high recirculation speed of the two recirculation cell
wake vortex. At the center of the rear face, the heat transfer
decreases due to the stagnation that is produced by the counter
rotation of the two cells. The adjacent and stretched two cell
WV that predicted by both k-ω models has a significant effect
on the local heat transfer coefficient as shown in the results. The
heat transfer coefficient decreases due to the circulation and
then increases to its highest value at the face center unlike the k-
ԑ models.
Conclusions
The performance of the two equation turbulence models in
simulating the cooling process of single electronic component
have been investigated at different Reynolsds number. The
following conclusions are the main results of the present study:
 Both standard and Realizable k - ԑ models predicted
VR, WV, SV, and very thin cell of UHV while it failed
to predict the LHV. RNG k - ԑ model predicted half
VR, SV, and significant UHV than two other k - ԑ
models, smaller WV, and small LHV.
 Standard k–ω model predicted VR, SV, WV, and
significant UHV and LHV. The SST k–ω model
predicted approximately the same flow structure
captured by standard k–ω model except the WV which
is similar in size and shape and similar to these
obtained by RNG k - ԑ model.
 The SST k–ω model gave the more accurate prediction
to the flow structure.
 All present models investigated the neat shape of the
local heat transfer distribution on the component front
face while both k–ω models predicted higher values
than k - ԑ models.
 Distribution of heat transfer on the side face has a
significant variations among the different models. The
SST k–ω model give heat distribution considering the
side vortex effect on heat transfer.
 The local heat transfer distribution on the rear face has
a little variation for different models and the
distribution taking in account the recirculation wake
vortex.
NOMENCLATURE
Latin letters
A
Area, m2
b
component height, m
Cp
specific heat, J/kg .K
D
diameter, m
g
gravitational acceleration, m/s2
average heat transfer coefficient, W/m2.K
H
height, m
k
thermal conductivity, W/m2.K
L
channel length, m
l
obstacle length, m
mass flow rate, kg/s
average Nusselt number
heat flux [ w/m2]
T
temperature, K
U
flow mean velocity, m/s
W
channel width, m
Greek letters
µ
dynamic viscosity, kg/m .s
ρ
density, kg/m3
Subscripts
Inlet
o
orifice
out
outlet
Acronyms
JICF
jet impingement in cross flow
LHV
lower horseshoe vortex
SV
side vortex
TV
top vortex
UHV
upper horseshoe vortex
VR
vortex ring
WV
wake vortex
x, y, z
Cartesian coordinate axes

5 Copyright © 2020 by ICFD14
REFERENCES
[1] W. Rodi, Comparison of LES and RANS calculations of the
flow around bluff bodies, Journal of Wind Engineering and
Industerial Aerodynamics 69 (1997) 55–75.
[2] D. Lakehal and W. Rodi, Calculation of the flow past a
surface-mounted cube with two-layer turbulence models,
Journal of Wind Engineering and Industerial Aerodynamics
67 (1997) 65–78.
[3] M. Kato and B. Launder, The modeling of turbulent flow
around stationary and vibrating square cylinders, in
proceding of Ninth Symposium on Turbulent Shear Flows
(1993) 1– 6.
[4] R. Martinuzzi and C. Tropea, The Flow Around Surface-
Mounted, Prismatic Obstacles Placed in a Fully Developed
Channel Flow , Journal of Fluids Engineering 115 (1993)
85-92.
[5] M. Yaghoubi, E. Velayati, Undeveloped convective heat
transfer from an array of cubes in cross-stream direction,
International Journal of Thermal Sciences 44 (2005) 756–
765.
[6] S. Chomdee and T. Kiatsiriroat, Enhancement of air cooling
in staggered array of electronic modules by integrating delta
winglet vortex generators, International Communications in
Heat and Mass Transfer 33 (2006) 618–626.
[7] G. Seeta Ratnam and S. Vengadesan, Performance of two
equation turbulence models for prediction of flow and heat
transfer over a wall mounted cube, International Journal of
Heat and Mass Transfer 51 (2008) 2834–2846.
[8] Y. Tominaga and T. Stathopoulos, Numerical simulation of
dispersion around an isolated cubic building: Comparison of
various types of k-Ԑ models, Atmospheric Environment
journal 43 (2009) 3200–3210.
[9] M. Kato and B. Launder, The modeling of turbulent flow
around stationary and vibrating square cylinders, in
proceding of Ninth Symposium on Turbulent Shear Flows
(1993) 10.4.1–10.4.6.
[10]M. Saathoff, P.J., Stathopoulos, T., Dobrescu, Effects of
model scale in esti- mating pollutant dispersion near
buildings, Journal of Wind Engineering and Industerial
Aerodynamics 1 (1995) 549–559.
[11]R. Dutta, A. Dewan, and B. Srinivasan, Comparison of
various integration to wall (ITW) RANS models for
predicting turbulent slot jet impingement heat transfer,
International Journal of Heat and Mass Transfer 65 (2013)
750–764.
[12]B. E. Launder. and B. I. Sharma, Application of the energy-
dissipation model of turbulence to the calculation of flow
near a spinning disc, Heat and Mass Transfer 1 (1974) 131–
138.
[13]C. S. Chen, K.C. Chang, and W.D. Hsieh, A modified low-
Reynolds-number turbulence model applicable to
recirculating flow in pipe expansion, journal of Fluids
Engineering 117 (1995) 417–423.
[14]S. Ashforth-Frost, K. Jambunathan, Effect of nozzle
geometry and semi- confinement on the potential core of a
turbulent axisymmetric free jet, International
Communications in Heat and Mass Transfer 23(1996) 155–
162.
[15]K. J. Kucinskas, Vortical Structures of an Impinging Jet in
Cross-flow, COSMOL Conference in Boston (2013) 1–7.
[16]M.J. Tummers, M.A. Flikweert, K. Hanjalic, R. Rodink and
B. Moshfegh, Impinging jet cooling of wall mounted cubes,
Engineering Turbulence Modelling and Experiments 5
(2005) 789–798.
[17]M. A. Teamah and S. Farahat, Experimental and numerical
heat transfer from impinging of single free liquid jet,
Alexandria Eng. J. 42 (2003) 559-575.
[18]M.A. Teamah, S. Farahat, Experimental heat transfer due to
impinging of water from multiple jets on a heated surface,
Alexandria Eng. J. 45 (2006) 1–13.
[19]M. Bedrouni and A. Khelil, International Journal of Heat
and Mass Transfer Numerical study on the performance of
rounded corners on the top of electronic components on
cooling effectiveness, Int. J. Heat Mass Transf., vol. 150,
2020.
[20]Y. Masip, A. Campo, and S. M. Nuñez, Experimental
analysis of the thermal performance on electronic cooling by
a combination of cross- fl ow and an impinging air jet, Appl.
Therm. Eng., vol. 167, no. December 2019, p. 114779,
2020.
[21]H. M. Maghrabie, M. Attalla, H. E. Fawaz, and M. Khalil,
Effect of Jet Position on Cooling an Array of Heated
Obstacles, J. Therm. Sci. Eng. Appl., vol. 10, no. February,
pp. 1–10, 2018.
[22]Y. Masip, A. Rivas, G.S. Larraona, R. Anton, J.C. Ramos,
B. Moshfegh, Experimental study of the turbulent flow
around a single wall-mounted cube exposed to a cross-flow
and an impinging jet, International Journal of Heat and Fluid
Flow 38 (2012) 50–71.
[23]H. M. Maghrabie, M. Attalla, H. E. Fawaz, and M. Khalil,
Impingement / effusion cooling of electronic components
with cross-flow, Appl. Therm. Eng. J., vol. 151, no.
February, pp. 199–213, 2019.
[24]H. M. Maghrabie, M. Attalla, H. E. Fawaz, and M. Khalil,
Numerical investigation of heat transfer and pressure drop
of in-line array of heated obstacles cooled by jet
impingement in cross-flow, Alexandria Eng. J., vol. 56, no.
3, pp. 285–296, 2017.
[25]T. L. Bergman, A.S. Lavigne, F.P. Incropera, D.P. Dewitt,
Fundamentals of Heat and Mass Transfer, seventh ed. Wiley,
2011.

6 Copyright © 2020 by ICFD14
List of Figures
Fig. 1. Schematic diagram of electronic component subjected to JICF.
Fig. 2. 3D view of the computational mesh around electronic component subjected to JICF.

7 Copyright © 2020 by ICFD14
Figure 3: Local heat transfer coefficient on the front face (AB) of the first obstacle cooled by CF at z/b = 0.5 for
different grid sizes.
Figure 4. Mesh statistics: skewness at the interface between the jet body and the channel domain

8 Copyright © 2020 by ICFD14
Fig. 5. Flow structure around component subjected to JICF by Masip et al. [11].
Fig. 6. Velocity vector on x-z (above) and x-y planes around an electronic component subjected to JICF at
Rej/Rec = 2 using standard k - ԑ model

9 Copyright © 2020 by ICFD14
Fig. 7. Velocity vector on x-z (above) and x-y planes around an electronic component subjected to JICF at
Rej/Rec = 2 using Realizable k - ԑ model
Fig. 8. Velocity vector on x-z (above) and x-y planes around an electronic component subjected to JICF at
Rej/Rec = 2 using RNG k - ԑ model.

10 Copyright © 2020 by ICFD14
Fig. 9. Velocity vector on x-z (above) and x-y planes around an electronic component subjected to JICF
at Rej/Rec = 2 using standard k - ω model
Fig. 10. Velocity vector on x-z (above) and x-y planes around an electronic component subjected to JICF at
Rej/Rec = 2 using SST k - ω model

11 Copyright © 2020 by ICFD14
Fig. 11. Local heat transfer coefficient on the front face of electronic component for different turbulent models.
Fig. 12. Local heat transfer coefficient on the side face of electronic component for different turbulent models.

12 Copyright © 2020 by ICFD14
Fig. 13. Local heat transfer coefficient on the rear face of electronic component for different turbulent models.