Professional Documents
Culture Documents
Abstract. In this present study, the round jet impingement on flat plate is ana-
lyzed numerically using four different RANS turbulence models. The four tur-
bulence models considered are k- SST, Realizable k-ε, RNG k-ε and 2f. The
numerical results in terms of Turbulent kinetic energy have been compared with
the previously published experimental results for the purpose of validation. It
has been observed that 2f model predicted the variation in turbulent kinetic en-
ergy accurately in the wall jet region above the near wall as compared with
other models.
1 Introduction
The predictions of flow by a turbulence model depend upon how well these models
are able to capture the Reynolds stresses. The turbulent kinetic energy (k) calculated
by various RANS model uses only the normal Reynolds stresses [2]. The closure
method approximates the Turbulent Kinetic Energy (k) with only normal Reynolds
stresses. Various RANS model uses turbulent kinetic energy as a parameter to predict
the turbulent flow. The Reynolds’ stresses were experimentally analyzed by many re-
searchers so as to obtain the nature of turbulence flow [6 - 8]. Cooper et al. [6] have
conducted experimental analysis of Reynolds stresses on round jet impinging on a flat
plate and presented their results. Turbulence flow in the near field of an axisymmetric
jet was experimentally investigated by Tummers et al. [7] and found instantaneous
flow reversal in the near wall region, attributed to the formation of small secondary
vortices. Yao et al. [8] have conducted studies on turbulent jet impingement on a flat
surface, investigated on Reynolds’ stress and concluded that, the mixing region in the
free jet has the maximum kinetic energy. Numerical investigation of normal
Reynolds’ stresses with RANS model is also done to get more insight into various as-
pects of turbulent flow. The comparative study done in this present work, with turbu -
lent kinetic energy was not previously done by many researchers.
2 Numerical Modelling
The problem of circular jet impinging on a flat plate was analysed numerically
assuming the flow to be steady, incompressible, turbulent and axisymmetric. In this
present study, the relative performance of different RANS turbulence models with
varying nozzle to plate spacing (h/d) was determined.
(a)
3
(b)
(c)
Fig.1. (a) Cut section view of plat, (b) 3 D view of domain geometry and (c) Mesh pattern of
computational domain
4
The time averaged continuity, momentum and energy equations solved during the
simulation are as follows,
(1)
(2)
(3)
5
is the turbulence viscosity and k is the square of length scale. For details
discussion of RANS turbulence models the readers are advised to review studies of
Singh et al. [11-12], and Dutta et al. [13].
3 Solution Methodology
The open source CFD solver, OpenFOAM 5 was used to solve the continuity,
momentum and energy equations. Finite volume method is used by this solver to
discretize the governing equations. Semi-Implicit Pressure Linked Equation
(SIMPLE) algorithm is used to couple velocity and pressure [9]. Four RANS
turbulence models namely k- SST, Realizable k-ε, RNG k-ε and 2f were used to
model turbulence. OpenFOAM supports variety of wall functions for different
parameters [10]. Corresponding wall functions were used according to the input
boundary conditions in order to capture the near wall effects. When the residuals of
continuity, velocity components and turbulence quantities fall below 10 -6 the solution
is deemed to be converged. The values of density and dynamic viscosity were taken
as 1.225 kg/m3, 1.7894×10-5 kg/ms respectively.
4 Boundary Conditions
A long channel was modelled so as to get a fully developed flow at the nozzle exit.
Uniform velocity distribution was given as input at nozzle inlet while no slip condi-
tion was given at both the plate and nozzle walls. The sufficient grid refinement was
near the target surface to obtain Y+ value less than 1. Both pressure inlet and outlet are
open sides of the domain with zero initial gauge pressure. Wedge boundary condition
was applied at both front and back sides of the geometry as openFOAM demands the
use of a wedge shaped geometry for axisymmetric cases. Remaining boundary condi-
tions were set using the following equations,
(4)
6
At pressure boundaries and inlet the flux of K, ε, ω, 2, f is set to zero. The values of k,
ε, ω and 2 at inlet were calculated from the above equations based on the values of
turbulent intensity and eddy viscosity ratios.
In this present study, the turbulent kinetic energy was numerically simulated and com-
pared with the experimental results of Reynolds stresses [7]. In the quest for an appro-
priate turbulence model, the simulations were carried out using k- SST, Realizable
k-ε, RNG k-ε and 2f turbulence models at Red = 23000, for various non-dimensional
distance of the jet exit to flat plate, h/d ratio = 2. To predict turbulent Kinetic energy
k, RANS turbulence models uses only the normal Reynolds stresses. With the help of
instantaneous velocity fluctuations perpendicular to each coordinate axis, at a particu-
lar point in the domain, the turbulent kinetic energy can be calculated. Turbulent ki-
netic energy is one of the parameter by which the turbulence flow is determined by
these turbulence models. Figure 2 shows the turbulent kinetic energy predicted by the
four turbulence models at various locations of the fluid flow domain, for nozzle to
plate spacing, h/d = 2, compared with the experimental results obtained by Tummers
et al [7]. At the stagnation point, Y/D = 0, the near wall turbulence kinetic energy pre-
dicted by both Realizable k-ε and 2f models are near to the experimental result
whereas RNG k-ε and k- SST models over predicted the results as depicted in Fig. 2
(a). As explained, from the fluid flow results, both RNG k-ε and k- SST models
over predicted at the stagnation point while the predictions of Realizable k-ε and 2f
models were close to the experimental results. However Realizable k-ε model over
predicted the turbulence kinetic energy after the near wall region. Same trend in re-
sults were followed by all the turbulence models at Y/D = 0.25 as shown in Fig. 2 (b),
however the predictions by RNG k-ε and k- SST models almost matched with the
experimental results further away from the near wall region. From Fig. 2 (c) it is evi-
dent that 2f model predicts close results with experimental values in the near wall re -
gion, but all other models over predicts the result.
(a) (b)
7
(c) (d)
(e) (f)
Fig.2. Turbulent kinetic energy predicted by the four turbulence models at various locations of
the fluid flow domain, compared with the experimental results obtained by Tummers et al [7].
The turbulent kinetic energy predicted by other models at stagnation point is very
high compared with the prediction of 2f model; hence it is able to predict accurate re-
sult at stagnation point. From Fig. 2 (d) it is clear that the predictions of both RNG k-ε
and k- SST models matched the nature of the curve beyond the near wall region,
while only 2f model under predicted the results in the near wall region. This explains
why the Nusselt number was under predicted by 2f model in this region. The trend of
the experimental result was captured by k- SST, Realizable k-ε and RNG k-ε, how-
ever 2f model under predicted the result at the near wall region. At Y/D = 1.5, the
variation in turbulent kinetic energy at near wall region were captured by k- SST
while RNG k-ε and Realizable k-ε over predicted the results, however 2f model under
predicted the results as depicted in Fig. 2 (e).
The k- SST turbulence model’s prediction matched with the experimental result
accurately at Y/D = 2.5, in the near wall region and beyond, while 2f model followed
the same trend and accurately matched with the experiment results only beyond near
wall region as shown in Fig. 2 (f). RNG k-ε and Realizable k-ε also captured the trend
of the curve but their result varied slightly from experimental result. At the wall jet re-
gion k- SST captured the behavior of the curve accurately. This explains the reason
for the accurate prediction of Nusselt number in wall jet region by k-SST. In all the
8
cases Realizable k-ε model tend to over predict the result. The 2f model is better in
predicting the turbulent kinetic energy results at stagnation point, while both RNG k-ε
and k- SST turbulence models predicted the nature of the curve in the vicinity of
stagnation point.
6 Conclusion
In this present analysis, four different turbulence models were used in the analysis of
turbulent kinetic energy using Reynolds stresses for nozzle to plate spacing at h/d = 2
and Red = 10000. The 2f model predicted the variation in turbulent kinetic energy
accurately in the wall jet region above the near wall while it predicted with moderate
accuracy in the stagnation region.
References
1. Geers, L.F.G., Hanjalic, K., Tummers, M. J.: Wall imprint of turbulent structures and heat
transfer in multiple impinging jet arrays. J. Fluid Mech. 546, 255–284 (2006).
2. Zuckerman, N., Lion, N.: Jet impingement heat transfer: physics, correlations, and numeri-
cal modeling. Adv. Heat Transf. 39, 565-631 (2006).
3. Dewan, A., Dutta, R., Srinivasan, B.: Recent trends in computation of turbulent jet im -
pingement heat transfer. Heat Transfer Engineering, 33(4–5), 447–460 (2012).
4. Colucci, D.W., Viskanta, R.: Effect of nozzle geometry on local convective heat transfer to
a confined impinging air jet. Exp. Therm. Fluid Sci. 13, 71-80 (1996).
5. Brignoni, L.A., Garimella, S.V.: Effects of nozzle-inlet chamfering on pressure drop and
heat transfer in confined air jet impingement. Int. J. Heat Mass Transf. 43, 1133-1139
(2000).
6. Cooper, D., Jackson, D. C., Launder, B. E., Liao, G. X.: Impinging jet studies for turbu-
lence model assessment-l. Flow-field experiments. International Journal of Heat and Mass
Transfer, 36 (10), 2675-2684 (1993).
7. Tummers, M. J., Jacobse, J., Voorbrood, S. G.J.: Turbulent flow in the near field of a
round impinging jet. International Journal of Heat and Mass Transfer, 54, 4939-4948
(2011).
8. Yao, S., Guo, Y., Jiang, N., Liu, J., An experimental study of a turbulent jet impinging on
a flat surface. International Journal of Heat and Mass Transfer, 83, 820–832 (2015).
9. Patankar, S.V.: Numerical heat transfer and fluid flow. Hemisphere, Washington DC
(1990)
10. OpenFOAM 5 User guide
11. Singh, D., Premachandran, B. and Kohli, S., Numerical simulation of the jet impingement
cooling of a circular cylinder, Numerical Heat Transfer, Part A: Applications,vol. 64, no.
2, pp.153-185, 2013(a).
12. Singh, D., Premachandran, B. and Kohli, S., Experimental and numerical investigation of
jet impingement cooling of a circular cylinder, International Journal of Heat and Mass
Transfer,vol. 60, pp.672-688, 2013(b).
9
13. Dutta, R., Dewan, A. and Srinivasan, B., Comparison of various integration to wall , no.
ITW RANS models for predicting turbulent slot jet impingement heat transfer, Interna-
tional Journal of Heat and Mass Transfer, vol. 65, pp.750-764, 2013.
Nomenclature
Cµ - Model constant
d -Nozzle diameter
f - Elliptic relaxation function
h – Nozzle to plate spacing
I - Turbulent intensity
K - Turbulent kinetic energy
Kf - Thermal conductivity of the fluid
- Mass flow rate
Pr-Prandtl number
Tj – Jet exit air temperature
Tw - Local wall temperature
T∞ - Ambient temperature
2- Velocity variance scale
X and Y– Radial distance from stagnation point
Z – Vertical distance from the plate
Greek Letters
ε- Turbulent dissipation rate
εss - Emissivity of the Stainless Steel foil
µ - Dynamic viscosity of air
µt – Turbulent viscosity
ρ - Density of air
ω - Specific dissipation rate