Effect of Turbulence Model in Numerical Simulation of Single Round Jet at Low Reynolds Number

ISSN (e): 2250 – 3005 || Volume, 07 || Issue, 03|| March – 2017 ||
International Journal of Computational Engineering Research (IJCER)
www.ijceronline.com Open Access Journal Page 29
Effect of Turbulence Model in Numerical Simulation of Single
Round Jet at Low Reynolds Number
Raj Narayan Gopalakrishnan1
, Peter J. Disimile 2
1
Department of Aerospace Engineering, University of Cincinnati, Cincinnati, OH, USA 45221
2
Department of Aerospace Engineering, University of Cincinnati, Cincinnati, OH, USA 45221
I. INTRODUCTION
Round jet flow is a form of free shear flow that has multitude of industrial applications, from water jet exiting
the nozzles used by firemen for fire suppression to fuel jet in aircraft combustion chambers. Jet flow observed in
nature, such as thermal plumes and volcanic exhausts also exhibit a circular profile and has been source of
investigation for Bejan et al. [1]. The wide range of applications has ensured that a large number of research
work was done to establish some of its underlying physics.
Round jet flows can exist in either a laminar or turbulent state; and for our current study we are studying
turbulent axisymmetric round jets. An exhaustive review on turbulent round jets has been conducted by Ball et
al. [2] in 2012, who examined the work of Tollmein (done in 1926) to round jet analysis done as recent as 2010.
This work provided immense and valuable insight into the experimental and computational work done in the
field of round jets. Their work briefly discusses the effect of initial conditions of the jet on similarity profile;
which has been established by other authors. They also discuss the coherent structures observed in round jets
and the length scales associated with the flow. At the end of their work, they have pointed out the parameters
and physics of the round jet that still remain unknown; mainly the nature of interaction of the coherent structures
and the mixing transition. It is indeed interesting to notice that even in 2012, the complete physics of such a
ubiquitous fluid dynamic phenomenon is not completely understood.
Meanwhile, Kaushik et al. [3] and Dewan et al. [4] had performed a detailed survey on CFD (Computational
Fluid Dynamics) treatment of round jet, which provided excellent insight into the recent computational works
done in the field of research on round jets. While Kaushik et al. [3] discuss both laminar and turbulent jets,
Dewan et al. [4] has concentrated on turbulent plane and round jets thereby providing a generalized view on jets
and the recent research done.
One of the important aspect of round jets is the similarity nature of its mean velocity profile. It was believed
from data obtained from experimental analysis that the round jet velocity profile exhibited self-similar nature,
and that the profile was similar for all jets irrespective of the inlet flow conditions. It was George W. K. [5], who
in his pivotal work established the dependence of turbulent scalar properties on inlet flow conditions using
analytical methods. This work proved instrumental in initiating new generation of research where the initial
conditions of jet flow and its impact were analyzed. Mi et al. [6] examined round jets with two different initial
conditions (a top hat profile and a fully developed pipe flow) and were able to establish the dependence of
turbulence scalar properties on initial conditions similar to that presented by George W. K. [5]. Similar study
was conducted by Ferdman et al. [7] who studied the effect of initial conditions on velocity and turbulence
characteristics. Their study displayed a dependence of decay rates of jets on initial flow condition. From these
studies, it was concluded that the round jet velocity and turbulence profile has direct association with initial
ABSTRACT
Single axi-symmetric round jet flow was analyzed using computational techniques and validated with
experimental results to establish the suitable turbulence model for simulation of low Reynolds number
jets exiting from fully developed pipe. This work is performed as an initial study before
computationally simulating multiple impinging jets. To this end a single round jet at Reynolds number
of 7500 exiting from a fully developed pipe and entering into stationary air was modeled. Velocity
and turbulence profiles were extracted from the simulation and validated with in-house experimental
results. It was observed that although all the four turbulence models studied were able to closely
predict the mean velocity field, they were not able to accurately predict the turbulence intensity
distributions. From the models studied, it was concluded that SST k- ω model was the best turbulence
model for simulating low Reynolds number jet flow exiting from fully developed pipe.
Keywords: CFD analysis, Low Reynolds number, Round Jets, Realizable k-ε model, SST k- ω
model, Turbulence Intensity, Turbulence Model

Effect of Turbulence Model in Numerical Simulation of Single Round Jet at Low Reynolds Number
conditions. Fellouah et al. [8] established that there was direct correlation between the mean velocity condition
at inlet to the turbulence and Reynolds stress profiles downstream. Their observation was similar to those made
by Dimotakis [9] who established a Reynolds number range of 1-2 x 104 based on outer scale for the flow to
become sustainably turbulent. These studies helped to establish the dependence of flow physics on the Reynolds
number. From these studies, it was concluded that when validating computational results of axisymmetric round
jet, unique data set data was acquired from specific experimental setup, where both initial conditions and
Reynolds number matched. Any mismatch in boundary or flow conditions would affect data set quality and lead
to a possible mismatch. Hence, work done by Landers [10] was used for validation since the experimental
conditions including the initial flow profile generation and Reynolds number matched with the current study.
II. COMPUTATIONAL METHODOLOGY
In the present work, the flow exiting a fully developed round pipe was used as the inlet condition for the single
axi-symmetric round jet. Gopalakrishnan and Disimile [11] previously noted that the purpose for an autonomous
pipe and jet simulation was to control the grid points in both the studies so that the simulations could be
performed independently with a high level of mesh refinement and accuracy. The velocity profile obtained from
the pipe flow simulation was validated with published results, whereby a considerable agreement was observed.
This provided high confidence in the pipe flow solution, which was then used as the inlet boundary condition of
the single jet simulation.
In the current study, a single round axi-symmetric jet was analyzed using Realizable k-ε model initially at
Reynolds number = 7,500. Multiple meshes were generated to identify the smallest mesh setup by which the
solution becomes independent of grid resolution. Once a grid independent solution was obtained, the result from
the simulation was compared with in-house experimental results which indicated discrepancies in turbulence
intensity data obtained between computational and experimental approach. This was further analyzed by
performing the single jet analysis using other turbulence models and comparing the results.
III. GEOMETRY AND MESH GENERATION
The reason for performing single jet CFD analysis was to gain a better understanding of the computational
capabilities of current turbulence models in capturing the jet’s free shear layer physics. Once adequate
understanding was obtained with single jet, the most suitable turbulence model was selected to perform the
analysis of two jets impinging at an angle. Work performed by Disimile et al. [12] on impinging jets has been
taken as the validation source for impinging jet analysis, wherein a Reynolds number of 7500 was used. Hence
for the single jet study, and the pipe flow study [11] which preceded this work; Reynolds number used was
7500. Disimile et al. [12] had used pipe internal diameter of 20 mm in their study, which was maintained in the
present work and in Landers [10].
Since in the experimental conditions, the jet exiting the pipe mixes with a still ambient fluid (air); the geometry
designed for the current study had to represent a similar setup. This was attained by keeping the boundaries very
far from the jet path, so that the presence of any form of computational boundary would not manipulate the jet.
With this consideration, the computational geometry was generated in form of frustum, with the pipe flow exit
(and jet inlet) in the smaller section and the domain exit at the larger section. All the surfaces, including the
curved periphery was modelled to allow flow of mass in either direction and was set as pressure outlet boundary
conditions where reverse flow was permitted.
A pictorial representation of the geometry is shown in Figure 1. All surfaces other than inlet was modelled as
pressure outlet. In order to maintain solution reliability, the domain was constructed with the smaller side of
frustum with radius of 15D while the larger side had 25D as shown in Figure 1 and Figure 2.
Figure 1: Geometry for single axi-symmetric jet simulation

Figure 2: Side view of geometry for single axi-symmetric jet simulation
The axial flow length of 25D was maintained for the flow to develop inside the geometry. Further downstream
was not modelled or analyzed since the primary interest of this study was to simulate the near field physics of
round jets. Once the geometry was designed and initial boundary conditions established, the next step involved
in the analysis was grid generation. Two separate O-grids were used to capture the outer and inner periphery of
the frustum as shown in Figure 3. It shows the mesh with node count of 4.3 million nodes.
Figure 3: Front view and close-ups showing the mesh with 4.3 million nodes in detail
It can be observed in Figure 3 that the mesh close to the center (which is the pipe exit domain) is very fine when
compared to the mesh at the periphery. This has been maintained for the accurate transfer of resulting flow data
from the pipe simulation, since using a coarse mesh leads to data averaging.
The dark ring seen in the right most snapshot of the mesh indicates the boundary layer mesh within the pipe
flow. In order to accurately capture of boundary layer physics, the first node height had to be reduced so that the
distance from the wall in wall coordinates, y+ is closer to 1. This condition yielded a first node height of 0.036
mm in the pipe domain. Since the pipe domain exits into the jet domain, the same first node height was initially
maintained into the jet domain. Figure 4 shows the side view of the mesh. As can be clearly observed, the
central core region was captured with very fine mesh while mesh was allowed to grow coarser as it reached the
periphery of the domain.
Various mesh quality checks using different parameters were performed to ensure that the mesh obtained was of
adequate quality. Detailed description of the parameters and allowable limits for the same have been presented
in [11]. For the current study, it was established that the aspect ratio never exceeded 100 (the maximum aspect
ratio allowed by Ansys Fluent). Once a mesh with the required quality was generated, it was saved as *.msh file
from ICEM-CFD which was then input into the commercial finite volume solver, Ansys Fluent.

Figure 4: Side view of mesh for single axi-symmetric round jet
IV. CHOICE OF TURBULENCE MODELS
One of the primary concerns in this analysis was the ability of the turbulence models to reliably capture the
physics of free shear flows. It was noted during literature review that certain models performed better for round
jet flow when compared to others. Also, there were articles that implied the dependence of the flow field on
initial conditions [7] and proposed that round jets cannot be accurately simulated easily with the turbulence
models and parameters currently in use [13]. For the purpose of this study, four turbulence models were selected
for examination; Realizable k-ε model, SST k-ω model, Standard k-ε model and Standard k-ω model. Studies
comparing the efficacy of Realizable k-ε model with Standard k-ε model and Standard k-ω model were found
and so was studies comparing SST k-ω model with Standard k-ε model and Standard k-ω model. But, no study
comparing all these four models for turbulent round jets was found by the authors. Hence, this study provides
insight into the area of choice of best turbulence model when simulating round jet flow exiting fully developed
pipe at low Reynolds number. The turbulence models used in the current study is described in some details
below.
4.1 Realizable k- εmodel
Realizable k-ε model has been said to be modified to accurately predict the spread rate of planar and round jets.
T.-H. Shih et al. [14] asserts that effect of rotation on both turbulent kinetic energy (k) and eddy dissipation rate
(ε) has been well captured in Realizable k-ε model. This is claimed to be a reason for the superior performance
of Realizable k-ε model when compared with Standard k-ε model. The realizability condition implies: a)
Turbulent shear stress satisfy Schwarz inequality and b) Non-negative values for normal stress. Also, Realizable
k-ε model uses a variable for model coefficient Cμ instead of using constant like Standard k-ε model.
4.2 SST k-ε model
SST k-ω model by Menter [15] has been found to perform better than Standard k-ε model for capturing round
jet physics during literature review. The model, being a combination of Standard k-ω model in the boundary
layer flow and Standard k-ε model for outer layer has been one of the most favorite go-to turbulence models in
the CFD community. The near wall capability of Standard k-ω model combined with blending function which
transforms it into Standard k-ε model in case of free shear flows makes it valid for wide range of applications,
even in areas with adverse pressure gradients and separation regions.

4.3 Standard k- εmodel
Proposed in 1974 by Launder and Sharma [16], it is one of the most common and oldest turbulence model in
use. It’s a semi-empirical model with the turbulent kinetic energy (k) and dissipation rate (ε) based on model
transport equations. It has been observed that the Standard k-ε model performs poorly in case of adverse
pressure gradients. As further research provided conclusive evidence on the strength and weakness of the
Standard k-ε model; modifications were made to it to improve its performance. Such modifications have led to
the creation of few variations, of which two models are RNG k-ε model and Realizable k-ε model.
4.4 Standard k-ε model
Standard k-ω model by Wilcox [17] is another popular model among CFD users. It is an empirical model based
on model transport equations for the turbulence kinetic energy (k) and the specific dissipation rate (ω). The
Standard k-ω model has better wall physics capturing capability than Standard k-ε model and can handle meshes
with y+ closer to 1. Similar to Standard k-ε model, it has many weaknesses due to which modifications were
made, which eventually lead to the creation of SST k-ω model.
V. BOUNDARY CONDITIONS AND SOLVER SETUP
After choosing a turbulence model, the simulation requires the proper boundary conditions that will impart the
real world environment to the simulation. The flow from fully developed pipe was to be used as the inlet
boundary condition to the jet flow. Velocity and turbulence kinetic energy profiles as shown in Figure 5; were
extracted from the outlet of pipe flow simulation and were provided as the input boundary condition. In order to
simulate that the jet escaping into ambient fluid at atmospheric conditions, the outlet condition for the round jet
simulation was set as pressure outlet at ambient pressure. The periphery of the domain was also set as pressure
outlet condition at ambient pressure in order to replicate natural conditions. This allowed for entrainment of the
ambient fluid from all the directions. Air at ambient temperature and pressure was used as the working fluid.
Solver settings used by authors can be found in previous pipe study [11]. SIMPLE algorithm was utilized for
pressure-velocity coupling with the 2nd order Upwind scheme used for spatial discretization. A convergence
criteria of 1e-6 was used to confirm that the solution had fully converged with minimal possible error. Once the
residuals for the convergence criterions were satisfied, the run was stopped and the results file extracted to be
post processed.
Figure 5: a) Velocity and b) Turbulence Kinetic energy used as inlet condition for the round jet
VI. GRID INDEPENDENCE
As shown in the earlier pipe study performed by the authors, Gopalakrishnan and Disimile [11]; it is imperative
that the solutions obtained from any CFD analysis be resolved over multiple meshes in order to ascertain that
the solution achieved is not dependent on the grid size used. Assessing grid independence is carried out by
running simulations using multiple mesh configurations for the same geometry and boundary conditions and
observing the change in target parameters as the mesh is altered. Ideally, a very small mesh is generated initially
which satisfies the bare minimum mesh quality requirements (as discussed in [11]). Based on this mesh, further
meshes are typically generated by refining the mesh size over every iteration. The results from all these meshes
are compared, and the mesh size beyond which no significant change in target parameter occurs for any
variation in mesh size is considered as the point of grid independence. This mesh at the point of grid
independence is then taken as the minimal mesh required to obtain a node independent solution for that
particular geometry and simulation setup.
For the current study, six (6) different mesh geometries were generated starting from 0.25 million nodes. All the
analysis for the grid independence study was performed using Realizable k-ε turbulence model. The mesh at the
inlet of the jet domain was matched with the mesh from the outlet of the pipe used in [11] and the initial mesh
height of 0.036 mm into the jet domain was initially maintained. This was required for a smooth continuation of

mesh from the pipe outlet and for the accurate transfer of flow variables without any interpolation or averaging.
However, starting with a very low mesh node count and maintaining such small grid spacing at center yielded
regions with coarse mesh and high aspect ratio. In the current study those regions were successfully maintained
at the periphery of the domain, ie away from the area of interest.
This can be clearly observed in Figure 6 where the mesh close to the center of the domain is maintained at a
height of 0.036mm whereas the peripheral mesh was on the order of millimeters. This meshing philosophy was
applied in order to keep the starting node count low. As meshing iterations proceeded with the generation of
finer meshes, the mesh along the periphery was refined while adding nodes in the central region as shown in
Figure 6.
Figure 6:Front view of grid with a) 0.25 million nodes and b) 4.3 million nodes
A similar mesh refinement tactic was applied along the axial flow direction thereby reducing the aspect ratio
with every iteration of mesh enhancement. Impact of this refinement strategy can be clearly observed in Figure
7.
Figure 7 : Side view of central plane for grid a) 0.25 million nodes and b) 4.3 million nodes
Once all the test cases were successfully converged, the results from the analyses were extracted for post
processing. Mean velocity and turbulence kinetic energy (TKE) were obtained from all the mesh configurations
and plotted systematically to identify the point of grid independence as shown in Figure 8 and Figure 9.

Figure 8: Grid Independence based on Mean Velocity value
Figure 9: Grid Independence based on Turbulence Kinetic energy value
It is evident from Figure 8 and Figure 9 that the solution becomes independent of the grid size at around 1.38
million nodes since no significant change was observed in the velocity or turbulence kinetic energy value
beyond 1.38 million nodes, even with tripling the mesh size. The maximum variation from 1.38 million case is
of the order of 1.7% in terms of mean velocity and 5% in terms of turbulence kinetic energy for mesh size above
1.38 million nodes as seen in Table 1. These values are negligible when compared to the variation observed in
experimental data [10] which indicated a range of 10-18% fluctuation in the calculation of mean velocity values
and 22-41% fluctuation in the calculation of turbulence kinetic energy values for different experimental setup in
the case of single jet analysis.
Table 1 : Grid Independence data
Grid Independence Study Difference from 1.38 million
node case
% Difference from 1.38 million node
case
Vel (m/s) TKE (m2/s2) Vel (m/s) TKE (m2/s2) Vel (m/s) TKE (m2/s2)
0.25 Mill 0.1554 0.0436 0.0189 0.0125 13.8836 39.9865
0.44 Mill 0.1480 0.0391 0.0116 0.0079 8.4796 25.3214
0.78 Mill 0.1404 0.0338 0.0040 0.0026 2.9040 8.3026
1.38 Mill 0.1364 0.0312 0.0000 0.0000 0.0000 0.0000
2.44 Mill 0.1341 0.0296 -0.0023 -0.0016 -1.7191 -5.0892
4.3 Mill 0.1351 0.0303 -0.0013 -0.0009 -0.9762 -2.8290
Hence for further analyses, 1.38 million node mesh was considered as the baseline mesh. Using this mesh, a
single axisymmetric turbulent jet was simulated as exiting a long constant diameter round pipe with fully
developed turbulent velocity profile.

7. Effect of Turbulence Models
The single jet analysis was initially performed using Realizable k-ε model and was further simulated using SST
k-ω model, Standard k-ε model and Standard k-ω model. It has to be noted that the choice of Realizable k-ε
model as the first model used for analysis was based on the literature review, which claimed that the
performance of Realizable k-ε model was adapted specifically for accurate prediction of the spread rate of round
jets. During literature review, it was also noted that the SST k-ω model’s performance was claimed superior to
that of Standard k-ε model and Standard k-ω model [Menter [15]]. But a research work comparing the
performance of SST k-ω model and Realizable k-ε model for single round jets was not found during literature
survey.
8. Comparison of the velocity profiles from different turbulence models against in-house experimental
results
From Figure 10, it can be observed that there exists significant difference in the distribution of area of high
velocity (as shown within the demarcated region) when calculated by four turbulence models. It was imperative
to look closely into this zone to establish the accuracy of each turbulence model. Figure 11 gives the zoomed in
look at demarcated zone. A general trend observed was that the epsilon based turbulence models generated
smaller regions of higher velocity when compared to omega based models. Of the four turbulence models under
study; Standard k-ω model seems to have the largest zone with maximum velocity, denoting lower turbulence
dissipation.
Figure 10: Velocity contour along central plane for single axi-symmetric jet obtained using various turbulence
models

Figure 11: Zoomed view of velocity contour along central plane for single axi-symmetric jet obtained using
various turbulence models
In order to obtain a holistic understanding of the general trend of velocity at central zone, velocity profiles were
extracted from central line and different locations downstream of pipe exit. These profiles were compared with
data from Landers [10] in order to establish the turbulence model which provided solution closest to the
experimental result.
Velocity data extracted from axial central line was normalized with maximum velocity along the center line
(which occurs at the exit of the pipe flow) and was plotted over axial distance measured in pipe diameters. From
Figure 12; we notice that the Standard k-ω model follows a more deviated path when compared to the three
other remaining models. This consistent deviation of results obtained for Standard k-ω model demands further
investigation on the performance of the model. Another key-point to be noted from the Figure 12 is that the
performance of SST k-ω model, Realizable k-ε model and Standard k-ε model are very similar except in the
region between 5D to 10D from the pipe exit. This may be due to the breakdown of potential core in that region
suggesting different models simulates this region differently.
Figure 12 : Profile of normalized velocity along centerline
Initially, velocity data extracted from different location downstream of the pipe exit were plotted against data
from Landers [10] as shown in Figure 13 to Figure 16. The general trend observed was that the performance of
Realizable k-ε model and SST k-ω model closely matched each other with Standard k-ε model being another
close contender. It has to be noted that, beyond 6D; data from experimental study also showed a wider spread.
Meanwhile, the Standard k-ω model showed the maximum deviation from the experimental data at 9D and
10.33D. As seen in Figure 15 and Figure 16, the Standard k-ω model starts to deviate from the general trend
while the 3 other turbulence models illustrate similar results.

Figure 13: Radial profile of normalized velocity over radius normalized by half width at 3D from pipe exit

Figure 16: Radial profile of normalized velocity over radius normalized by half width at 10.33D from pipe exit
VII. COMPARISON OF TURBULENCE PROFILE FROM DIFFERENT TURBULENCE
MODELS WITH IN-HOUSE EXPERIMENTAL RESULTS
It was concluded that the performance of the four turbulence models was comparable in accuracy when
considering the mean velocity. However, different parameters need to be examined, in order to distinguish the
most suitable model for the current single simulation and future impinging jet study. Turbulence intensity was
identified as a suitable parameter in previously established works and hence it was taken for consideration in the
present study also.
As done previously, central plane was considered for extraction of data for plotting contours and profiles.
Turbulence intensity data was extracted along the center line and at various locations downstream of pipe exit
and compared with experimental data obtained from [10]. The turbulence kinetic energy (TKE) contours along
the central plane is shown in Figure 17. It is observed that there exists significant difference in the simulation of
turbulence by the four turbulence models. While the SST k-ω model and the Standard k-ε model displayed
similar performance for TKE parameter, Realizable k-ε model and Standard k-ω model predicted lower values
of TKE, with Standard k-ω model predicting the lowest.

Figure 17: Turbulence kinetic energy contour along central plane for single axi-symmetric jet obtained using
various turbulence models
Figure 18 gives the zoomed in look at demarcated zone shown in Figure 17, so that a better judgement of the
profile can be obtained. The presence of low turbulence region in the middle of the jet for Standard k-ω model
indicates that the model predicts lower turbulence in the high velocity region.
Figure 18: Zoomed view of turbulence kinetic energy contour along central plane for single axi-symmetric jet
obtained using various turbulence models
For obtaining clear understanding of the nature of turbulence simulated by the different models, it was
imperative to look at the turbulence intensity (TI) data at different locations and compare it with experimental
results from [10]. Initially, TI for different models were plotted along the center line and compared as shown in
Figure 19. Again, the Standard k-ω model stands as the odd-one-out, predicting very low values of TI until 18D
downstream of pipe exit.

Figure 19 : Profile of turbulence intensity along center line
An interesting observation can be made regarding Realizable k-ε model was that beyond 18D, the TI dips the
most for that model. This can be observed in Figure 18 also, as the contour for Realizable k-ε model alone
indicates the presence of lower TKE zone near the right end of the zone. This needed to be further analyzed to
ensure that confidence can be assigned to that model.
Warda et al. [18] had used relative turbulence intensity (RTI) instead of normal turbulence intensity in their
study. This motivated the calculation of RTI profile for various turbulence models. It can be observed in Figure
20 that the RTI values obtained for all 3 models except Realizable k-ε model showed similar trend of initially
peaking to a maximum value and then become constant (similar to the trend shown by Warda et al. [18]) while
Realizable k-ε model predicted an initial cresting and then slowly decreasing. The trend of reduction in RTI
after reaching a maximum value seems to indicate underlying inaccuracy in capturing the turbulence physics by
the Realizable k-ε model.
Figure 20: Profile of relative turbulence intensity along center line
The need to further ascertain the performance of Realizable k-ε model in predicting TI parameter was
recognized and TI data profiles were extracted at locations 3D, 6D, 9D and 10.33D from pipe exit for
comparison and validation. Again, the performance of Realizable k-ε model and SST k-ω model were close to
one another, with Standard k-ε model being the next close model. Standard k-ω model, as observed in other
scenarios, performed differently.

Figure 21 to Figure 24 shows the radial profile of TI at different locations downstream of pipe exit. It is noted
that the Realizable k-ε model and SST k-ω model were able to predict the TI value at 3D and 6D with some
success. Standard k-ε model was able to closely predict the TI value only at 3D, while Standard k-ω model
performed poorly in all locations. Beyond 6D, no model was able to accurately predict the TI for single round
jet even though they were able to predict the general trend (except Standard k-ω model). All models except
Standard k-ω model either correctly or over-predicts TI while Standard k-ω model under-predicts it at all the
locations.
Figure 21: Radial profile of turbulence intensity at 3D from pipe exit

Figure 24: Radial profile of turbulence intensity at 10.33D from pipe exit
From the current results, it was concluded that the performance of Standard k-ω model was not adequate enough
to properly model the physics of single axi-symmetric round jet. Hence, it was decided to forego the Standard k-
ω model from future consideration. Since the performance of Realizable k-ε model and SST k-ω model were
comparable (except for the inaccuracy of Realizable k-ε model in calculating relative turbulence intensity near
the end of the domain), it was deemed that SST k-ω model was the better of the two. Also, performance of
Standard k-ε model and SST k-ω model were quite comparable (except for Standard k-ε model over-predicting
turbulence intensity at 6D from pipe exit). From these observations, it was determined to use SST model alone
to analyze impinging round jets.

VIII. SUMMARY
Single jet simulations and analyses were performed at a Reynolds number of 7500 using various turbulence
models to evaluate the performance of each model. This was done in order to determine the best possible
turbulence model for a low Reynolds number round jet exiting from fully developed pipe flow. Based on the
results from this study, jets impinging at an angle will be modeled in the next phase of analysis using the
turbulence model that was found superior. Initially, grid independence was performed using Realizable k-ε
model to ensure that the solution obtained was independent of the grid size used. This was attained by
performing the simulation using a very small mesh (0.25 million nodes) and gradually increasing the node
count. It was found that beyond 1.38 million nodes, the solution did not show much dependence on the grid
count, with only 2% difference in velocity and 5% difference in turbulence kinetic energy (TKE) with almost
threefold increase in node count. Hence, for the purpose of single axi-symmetric round jets; 1.38 million node
mesh was taken as the grid independent mesh for the current Reynolds number and simulation setup.
It was observed that all the turbulence models performed well in case of mean velocity profile (with Standard k-
ω model showing slight fluctuation at 10.33D). But, while plotting TI; it was observed that none of the
turbulence models were able to accurately predict the TI values beyond 6D. Standard k-ω model was found to
under predict TI values at all the locations, while Realizable k-ε model and SST k-ω model were able to predict
TI values till 6D. The performance of Standard k-ε model was comparable with that of Realizable k-ε model and
SST k-ω model. It was concluded that for the simulation of round axi-symmetric jets originating from a fully
developed pipe flow at low Reynolds numbers, the SST k-ω model was the best model to be used and Standard
k-ω model may be the least appropriate model.
Nomenclature
Uc – Centreline Velocity (m/s)
U – Local Velocity (m/s)
D – Diameter
k – Turbulence kinetic energy (m2
/s2
)
ε – Dissipation rate (m2
/s3
)
ω – Specific Dissipation Rate (s-1
)
r1/2 – Jet half width (m)
REFERENCES
[1]. Bejan, A., Ziaei, S. and Lorente, S., "Evolution: Why all plumes and jets evolve to round cross sections," Scientific Reports, vol. 04,
no. 23, 2014.
[2]. C. Ball, H. Fellouah and A. Pollard, "The flow field in turbulent round free jets," Progress in Aerospace Sciences, vol. 50, pp. 1-26,
2012.
[3]. M. Kaushik, R. Kumar and G. Humrutha, "Review of Computational Fluid Dynamics Studies on Jets," American Journal of Fluid
Dynamics, pp. 1-11, 2015.
[4]. A. Dewan, M. Pathak and A. K. Dass, "A survey of selected literature on important flow properties and computational fluid
dynamics treatments of incompressible turbulent plane and round jets in quiescent ambient," Indian Journal of Engineering and
Materials Sciences (IJEMS), vol. 13, pp. 180-194, 2006.
[5]. W. K. George, "The self-preservation of turbulent flows and its relation to initial conditions and coherent structures," Advances in
turbulence, pp. 39-73, 1989.
[6]. MI, J., NOBES, D.S. and NATHAN, G.J., "Influence of jet exit conditions on the passive scalar field of an axisymmetric free jet,"
Journal of Fluid Mechanics, vol. 432, pp. 91-125, 2001.
[7]. E. Ferdman, M. V. Otugen and S. Kim, "Effect of Initial Velocity Profile on the Development of Round Jets," Journal of Propulsion
and Power, vol. 16, no. 4, pp. 676-686, 2000.
[8]. H. Fellouah, C. Ball and A. Pollard, "Reynolds number effects within the development region of a turbulent round free jet,"
International Journal of Heat and Mass Transfer, vol. 52, no. 17–18, pp. 3943-3954, 2009.
[9]. P. E. Dimotakis, "The mixing transition in turbulent flows," Journal of Fluid Mechanics, vol. 409, pp. 69-98, 2000.
[10]. B. Landers, "Mixing Characteristics of Turbulent Twin Impinging Axisymmetric Jets at Various Impingement Angles," Electronic
Thesis or Dissertation, p. https://etd.ohiolink.edu/, 2016.
[11]. R. N. Gopalakrishnan and P. Disimile, "Development of Inlet Boundary Condition for Accurate Multi-Jet Simulations,"
International Journal of Mechanical Engineering (IIJME), 2016.
[12]. P. J. Disimile, E. Savory and . N. Toy, "Mixing characteristics of twin impinging circular jets," Journal of Propulsion and Power,
1995.
[13]. S. B. Pope, "An explanation of the turbulent round-jet/plane-jet anomaly," AIAA Journal, vol. 16, no. 3, pp. 279-281, 1978.
[14]. T.-H. Shih, W.W. Liou, A. Shabbir, Z. Yang and J. Zhu, "A New k-e Eddy Viscosity Model for High Reynolds Number Turbulent
Flows-Model Development and Validation," NASA Technical Memorandum, Vols. ICOMP 94-21, 1994.
[15]. F. R. Menter, "Two-Equation Eddy-Viscosity Turbulence Models for Engineering Applications," AIAA Journal, vol. 32, no. 8, pp.
1598-1605, 1994.
[16]. B. E. Launder and B. I. Sharma, "Application of the Energy Dissipation Model of Turbulence to the Calculation of Flow Near a
Spinning Disc," Letters in Heat and Mass Transfer, vol. 1, no. 2, pp. 131-138, 1974.
[17]. D. Wilcox, "Re-assessment of the scale-determining equation for advanced turbulence models," AIAA Journal, vol. 26, no. 11, pp.
1299-1310, 1988.
[18]. H. Warda, S. Kassab, K. Elshorbagy and E. Elsaadawy, "An experimental investigation of the near-field region of free turbulent
round central and annular jets," Flow Measurement and Instrumentation, vol. 10, no. 1, pp. 1-14, 1999.
[19]. D. Yoder, J. DeBonis and N. Georgiadis, "Modeling of turbulent free shear flows," Computers & Fluids, vol. 117, pp. 212-232,
2015.

Effect of Turbulence Model in Numerical Simulation of Single Round Jet at Low Reynolds Number

More Related Content

What's hot (16)

Viewers also liked (13)

Similar to Effect of Turbulence Model in Numerical Simulation of Single Round Jet at Low Reynolds Number (20)

Recently uploaded (20)

Effect of Turbulence Model in Numerical Simulation of Single Round Jet at Low Reynolds Number