EA 422 COMPUTATIONAL FLUID DYNAMICS

The Copperbelt University

School of Engineering
BEng Aeronautical Engineering

2020/2021: Assignment 1
Mr. E. Zingapeta
Prediction of Unsteady Flow around A Square Cylinder Using ANSYS

S/N Full Names Signature

17109802 TREVOR KAPAMBWE
17110571 JAMES NYIRENDA
17113008 LEONARD NYIRONGO
17109739 MBUMBA CHINYAMA
Contents
1.0 LIST OF FIGURE ……………………………………………………………………………3
2.0 LIST OF TABLES …………………………………………………………………………...5
3.0 ABSTRACT .............................................................................................................................6
4.0 INTRODUCTION ....................................................................................................................7
2.1 BACKGROUND ................................................................................................................7
2.2 LITERATURE REVIEW ..................................................................................................7
5.0 NOMENCLATURE ………………………………………………………………………….8
6.0 GOVERNING EQUATIONS AND THEORIES………………………………….................9
7.0 METHODOLOGY .................................................................................................................12
8.0 RESULTS ...............................................................................................................................22
8.1 PLOT OF RESIDUALS …………………………………………………………….22
8.2 CONTOURS OF STREAM FUNCTION …………………………………………..24
8.3 CONTOURS OF VELOCITY MAGNITUDE ……………………………………..25
8.4 CONTOURS OF TURBULENT KINETIC ENERGY……………………………...27
8.5 CONTOURS OF STATIC PRESSURE …………………………………………….29
8.6 DRAG AND LIFT PLOTS ………………………………………………………….31
9.0 DISCUSSION..........................................................................................................................35
10.0 CONCLUSION ....................................................................................................................36
11.0 REFERENCES .................................................................................................................... 37
LIST OF FIGURES

FIGURE 1…………………………………………………………………………… 12
FIGURE 2…………………………………………………………………………….. 13
FIGURE 3……………………………………………………………………………. 13
FIGURE 4……………………………………………………………………………. 14
FIGURE 5…………………………………………………………………………….. 15
FIGURE 6……………………………………………………………………………… 16
FIGURE 7………………………………………………………….………………….. 16
FIGURE 8……………………………………………………………………………… 17
FIGURE 9……………………………………………………………………………… 18
FIGURE 10…………………………………………………………………………….. 20
FIGURE 11……………………………………………………………………………. 20
FIGURE 12…………………………………………………………………………….. 22
FIGURE 13…………………………………………………………………………….. 23
FIGURE 14…………………………………………………………………………….. 23
FIGURE 15…………………………………………………………………………….. 24
FIGURE 16…………………………………………………………………………….. 24
FIGURE 17……………………………………………………………………………. 25
FIGURE 18……………………………………………………………………………. 25
FIGURE 19…………………………………………………………………………… 25
FIGURE 20…………………………………………………………………………… 26
FIGURE 21…………………………………………………………………………… 26
FIGURE 22…………………………………………………………………………… 27
FIGURE 23…………………………………………………………………………… 27
FIGURE 24…………………………………………………………………………… 27
FIGURE 25……………………………………………………………………………… 28
FIGURE 26…………………………………………………………………………… 28
FIGURE 27……………………………………………………………………………. 28
FIGURE 28…………………………………………………………………………… 29
FIGURE 29…………………………………………………………………………… 29
FIGURE 30…………………………………………………………………………… 30
FIGURE 31……………………………………………………………………………. 30
FIGURE 32…………………………………………………………………………… 31
FIGURE 33…………………………………………………………………………….. 31
FIGURE 34…………………………………………………………………………..... 32
FIGURE 35…………………………………………………………………………….. 32
FIGURE 36………………………………………………………………………….. 32
FIGURE 37…………………………………………………………………………….. 33
FIGURE 38…………………………………………………………………………….. 33
FIGURE 39…………………………………………………………………………….. 34
LIST OF TABLES
TABLE 1……………………………………………………………………………………. 22
 ABSTRACT
In this present paper, ANSYS fluent simulation software is used to study backward-facing step
by using three different turbulence models, namely the Standard model, the Realizable and the
Shear Stress Transport model (SST). This paper includes the presentation of fluid dynamic
measurements in a separated and reattaching boundary laver, in which a lot of focus is dedicated
on the near-wall region/area. The flow is investigated under a constant Reynolds number equal to
28000, a density of 1kg/m3, viscosity of 3.5714x10-5Pa.s and velocity of 99.9992m/s. A
comparison of the results of the present study with existing experimental and numerical data
(investigated by vogel and Eaton- 1985) is made in which good agreement is achieved. For this
investigation the Realizable Model is used where for the near wall treatment the Standard Wall
Function and Enhanced Wall Treatment is utilized. Two different meshes are created in which
one is appropriate for utilizing wall function and the other mesh is suitable for using enhanced
wall treatment which reflects on the fact that uses a well refined boundary layer mesh in order to
capture flow close to the wall. In addition, the effect, the differences and similarities between the
different spatial discretization scheme are investigated.
KEY WORDS: unsteady flow, turbulent flow, standard wall function, enhanced wall function,
spatial discretization, the Standard model, the Realizable and the Shear Stress Transport model
(SST).
 INTRODUCTION/BACKGROUND
Backward-facing step (BFS) flow is a very important aspect/ issue in fundamental fluid
mechanics. The investigation of flow separation, flow reattachment and recirculation has been
studied to a greater depth due to the flow geometry of such particular models that provide a significant
prototype. The phenomenon of flow separation is common in engineering applications.
Application of flow separation in the engineering field varies from simple problems to intricate
problems such as the water flow past the hydrofoil, the airflow past the blades of compressor and
turbine, suddenly expanding pipes, combustors etc. (Howruz, Salman, & Ghasseni, 2018).
Flow separation and vortex formation are major considerations in most engineering fields including the
civil, aerospace and mechanical fields. Although numerous investigations have been carried out on this
topic, an insightful and complete understanding of the physical origin of flow separation and vortex
formation has not been clear. The major reason for this lies in the fact that an analytical treatment of
the flow is not available and hence experimental and numerical investigations are involved. (Biswas,
Breuer, & Durst, 2014).

Most of the previous work on reattachment involved study of the hydrodynamics of backward-facing
step flows. Several famous engineers such as Eaton, Johnston, Watkins and Gooray reviewed the
literature in this area. A large number of early studies served to delineate the basic characteristics of
reattaching flows, but could not provide quantitative data. More recently, experimenters have used the
pulsed-wire anemometer (e.g., Eaton and Johnston and the laser- Doppler anemometer (LDA) (e.g.,
Durst and Tropea to supply quantitative data in the highly turbulent reattaching flow. Near-wall
velocity data have only recently become available with the advent of the thermal tuft Eaton et al; the
pulsed wall probe Westphal et al; and specially configured I-DA systems.

There have been fewer studies of flow separation in in these flows. Fletcher et al. Aung and Watkins
[9], and Watkins and Gooray have all recently reviewed the literature in this area. Most of the
experiments cited in the reviews contained only mean heat transfer rates and very little fluid dynamic
data. The data sets show the same general features for a variety of geometries: a drop in the heat
transfer coefficient at separation followed by a sharp rise in the reattachment zone (Vogel & Eaton,
1985).
LITERATURE REVIEW
 NOMECLATURE

AR – area ratio, upstream height/ downstream height

1
Cf – instantaneous skin friction coefficient = τ0 / ρ∞U∞
2

Č – time mean skin friction coefficient

C’f – fluctuating component of Cf

C∞ - free- stream specific heat

S – step height

ReH - step-height Reynolds number = Uref S /ν

St – Stanton number = h / Uref ρ∞C∞

T – local mean temperature

T∞ - free stream temperature
TW – local wall temperature
Ū – mean stream wise velocity
u’ – fluctuating component of stream wise velocity
u+ - non-dimensional velocity = U/√ τ /¿ ρ ¿
Uref = Ub – reference free- stream velocity measured at x/S = -3
x – stream wise coordinate measured from step edge
xR – reattachment length
y – coordinate normal to wall
y+ - non-dimensional distance from wall = y√ T w /¿ ρ /ν ¿

γ – forward- flow fraction

ν – freestream kinematic viscosity
ρ ∞ - freestream density
τ0 – local wall shear stress
h - channel height upstream of the step
GOVERNING EQUATION
In relation with the available data and the investigation being carried out, an incompressible fluid
with constant fluid properties is assumed. The equations presented below include dimensionless
governing equations expressing the conservation of mass and momentum in Cartesian
coordinates. These are expressed in the following form:

∂ ui
= 0 …………………………………………………………………………………..[ 1 ]
∂x j

∂u i ∂(ui u j ) ∂ P 1 ∂ τi j
+ =- - ………………………………………………………….[ 2 ]
∂t ∂ xj ∂ xi ℜ ∂ x j

In the above equations ui∧u j dictate the Cartesian components of the velocity and the pressure is
given by P. the molecular momentum transport is given by τ i j. For a Newtonian fluid, the value
for τ i j is given by:
∂ ui ∂ u j
τ i j = -μ [ + ]………………………………………………………………………[ 3 ]
∂ x j ∂ xi

One of the key non- dimensional parameters in this study is the Reynolds equation. In such an
investigation the Reynolds number might be defined is several ways depending on the available
dimensional parameters.
Figure 1. Sketch of the flow configuration and definition of length scales

ρU b D
RE = RED = …………………………………………………………………………[ 4 ]
μ

In equation [4], the Reynolds number is dictated by the hydraulic diameter. In which the
hydraulic diameter, D = 2h.

ρU b D
RE = REh = …………………………………………………………………………[ 5 ]
μ

In equation [5], the Reynolds number is dictated by the channel height upstream of the step, h. In
which the value of D = h.
ρU b D
RE = RES= …………………………………………………………………………[6]
μ
In equation [6], the Reynolds number is dictated by the step height, S. In which the value of D =
S.
The standard parabolic velocity profile with a maximum velocity is expressed as follows below,
in equation [7]
3
Umax = Ub……………………………………………………………………………….....[7]
2
In the investigation the Reynolds number given is with respect to the step height, thus we employ
the Reynolds equation under equation [6].
 METHODOLOGY
In the investigation of the flow separation and reattachment, ANSYS FLUENT was used to
generate the desired information. This generated numerical information was compared to the
numerical and experimental data generated by Vogel and Eaton (1985). ANSYS- fluent was used
by employing the 3 turbulent simulation models, thus the k-ɛ standard, k-ɛ realizable and k-ω
SST turbulence models. The finite volume method (FVM) is used to solve the governing
equation within ANSYS FLUENT.

Several parameters and factors such as the boundary layer, computational domain and co-
ordinate definition are defined to enable a smooth analysis. The flow analysis in the backward
facing simulation is considered incompressible. The numerical method used in the simulation is
explained below.

BOUNDARY CONDITIONS

The wall boundary conditions used in this steady are those of impermeability and non-slip
condition, i.e., u = 0, v = 0
 To solve the governing equations numerically, uniform free stream condition with velocity U∞
=99.9992 m/s are applied at the inlet boundary. The periodic condition is considered at the lateral
boundaries. Also the flow exit is treated as a pressure outlet.

COMPUTATIONAL DOMAIN
The computational domain is defined as in the figure below. The computational domain defines
the surrounding and the system, giving the boundaries about which the simulation can take place
thus H defining the dimension of the Step height in both 2D and 3D arrangements, while the
distances of the wall domain are defined in terms of H as shown below.

Figure 2: Dimensions of the computational domain for the BFS.

Figure 3. 2D Computational domain in simulation software

GRID GENERATION

The computational grid generated after meshing is as shown below. Due to the two different
meshes imposed on the geometry the following grid is expected to be generated within
simulation of such a geometry.

Figure 4: 2D Grid generated in simulation software

Figure 5:3D Grid generated by simulation software

NUMERICAL METHOD
The simulation involves several procedures. In order to generate the desired information, the
following procedures act as a guide for the operation of the ANSYS-FLUENT software.

Open Ansys fluent and double click fluid flow (fluent). Click geometry, and in the panel
(properties of project schematics) change analysis type to either 3D or 2D depending on
the type of simulation to be conducted. Right click geometry and Select design modeler.
Select the X-Y orientation. On sketch select rectangle and draw a square geometrically
about the origin. Select line and draw the simulation geometry. Click units and change
length units to centimeters. Click on dimensions and dimension the simulation geometry
as desired. Generate the geometry and rename it to “free domain”. Go back to geometry
and click update in order to update the geometry. Right click on geometry and select
“import geometry”. Double click on mesh. Select “edge” on the panel board then select
the edges.
 Select the edge of the geometry closest to the edge of the square, right click then
select “create named selection”; type “inlet” in the space provided.
 Select the edge of the geometry furthest to the edge of the square, right click then
select “create named selection”; type “outlet” in the space provided.
 Select the top and bottom edges of the geometry, right click then select “create
named selection”; type “walls” in the space provided.
Click mesh control, select “sizing” then select edge on the panel. Select the (inlet)
rectangle edge, click apply. Click mesh, select sizing then select inlet and outlet and
apply geometry selection. Click on edge on the panel to the left, then select the top and
 bottom edges of the body. Double click mesh, then select update. Click generate then
Right click mesh. Click insert, click sizing then Select top wall near the outlet. Click
apply.

Figure 6:Opening of Ansys

Figure 7: Beginning of meshing of the geometry.

Figure 7: Naming of boundaries.

Figure 9: face splitting boundaries

Click element size and insert number of divisions to the desired number, click bias type
and select bias of choice. Right click mesh insert sizing. Select edge pointer and select
bottom wall boundary near the inlet. Click apply and select number of divisions. Select
bias type and select bias factor required; Repeat previous step then right click mesh and
select sizing. Click edge pointer and apply. Change element size to number of division
then Select bias type and insert bias factor needed. Right click mesh and select sizing;
Generate mesh. Right click mesh and select sizing. Select all edges and generate mesh.
Right click mesh and insert face meshing. Click face pointer and click on the face and
apply. Click coordinate system then Right click mesh and generate mesh. Go back to
design model and right click surface sk1 then generate. Click concept and select line
three points. Click tools go on face split and click on the surface and apply. Click
geometry tools and edge pointer. Click the on the three-point line joining the front
boundary wall and the wall square and apply then generate. Click tool and select face
split then select the face square and click apply. Click edge pointer and select both line
and apply. Repeat the above process on all the line then generate. Click tools and select
face split and apply and do the same the all lines then generate.
Click on edge pointer select the line, and generate. Click concept and select line from
points. Select two points one on the geometry wall and the boundary wall and click apply
then Click generate. Repeat the previous process for all other lines splitting the geometry
faces into four parts and Click save. Click file and select export to Geom and save. Go on
mesh and right click model (A3) then click on named selection and go on inert. Click on
geometry and click edge pointer and select the 3 edge inlet lines then click apply. Click
face meshing and select all the split surfaces then click apply. Click edge size and go on
edge pointer, select edges of the geometry and then apply. Click edge sizing 2 select four
lines and apply then Click four edge then select the line and apply. Repeat the previous 2
processes on all edge sizing then click bias and select bias type. Click face meshing then
click on geometry and select the faces within the geometry. Right click mesh and insert
face meshing. Repeat the above process on two rear faces. Right click mesh and select
face meshing. Select three faces past the square wall and apply then Repeat for the
remaining internal face. Right click mesh and generate mesh.
Figure 10: Generated mesh.

Figure 10. surface split meshed geometry.

Save the project in the backward facing flow files folder. Right click mesh and update
mesh then double click setup. Create a Microsoft excel worksheet file and rename it as
“calculations”. Write the Reynolds formula and standard values in the worksheet. Click
on material on the workbench panel, select fluids then create a fluid element with the
given viscosity under fluids. Copy the value of the generated viscosity and paste it in the
worksheet as a standard value. Generate the value for the velocity. Right click on models
and select k- omega (2 equation). Select SST under k- omega then Click on boundary
conditions and select inlet. Under inlet click components and put the calculated value of
the x- component of velocity then apply changes. Click on reference values and put
“inlet” in the compute from blank space. Click on methods and change scheme to couple;
also update the Pressure-Velocity Coupling to the required Spatial discretization. Click
on report definitions and select new.

Figure 11: Completed design

 Figure 12: Worksheet viscosity calculation
SETUP
Under setup the, various models are simulated in order to determine the model that is the
most agreeable as compared to the experimental data results. In setup, the 3 models are
assessed and the different values and plots of the skin friction coefficient compared among
the models and the experimental data generated. For the 3 turbulent models analysis is
conducted while keeping the convergence criteria at 10-6 in residuals within the report
definition. The minimum number of iterations is maintained at 500.
Results

For the flow study, in this report, the ANSYS models are compared with the experimental results
for a Re of 28, 000. This is shown in the table below-table 1.

Table 1 Summary of typical parameters from various studies

Authors Method Skin friction Pressure Velocity continuity

coefficient coefficient

Vogel and
Eaton (1985)

Group k - ω SST

Group k- ε Standard

Group k- ε
Realizable

The plot of Residuals for 2D simulations

For all models simulated, 500 Number of iterations was used and all three models become stable
at x and y velocity at approximately 1e-3
Figure 13: k - ω SST model, x velocity and y velocity components.

Figure 14: k- ε Standard scaled residuals.

Figure 15: k- ε Realizable, showing x and y velocity components.

Contours of stream function

Comparison of the stream functions for the solution models for velocity magnitude.
Figure 16: k - ω SST model, depicting stream function contour from 0 m/s to 104.3207 m/s.

Figure 17: k- ε Standard, depicting stream function contour from 0 m/s to 104.6388m/s.
Figure 18: k- ε Realizable, stream function contours 0 m/s to 104.202 m/s.

Contours of static pressure

Static pressure contours are compared for the solution models to observe the flow over the
geometry.
Figure 19: k - ω SST model, depicting static pressure from -1702.871pa to 133.3245pa.

Figure 20: k- ε Standard, depicting static pressure from -1695.383pa to 190.9123pa.

Figure 21: k- ε Realizable, depicting static pressure from -1638.545pa 167.389pa.

Contours of Turbulent kinetic energy

Range of study used was 3.00e-07(m2/s2) to 600 (m2/s2)

Figure 22: k- ω SST, showing turbulence kinetic energy (3.622011e-07 m2/s2 to 490.6652 m2/s2)
Figure 23: k- ε Standard, showing turbulence kinetic energy 0.1517325 m2/s2 to 561.6001 m2/s2

Figure 24: k- ε Realizable, showing turbulence kinetic energy (0.01189936 m2/s2 to 504.4446
m2/s2
Contours of the x-velocity component.

Figure 25: k- ω SST model, showing x- velocity component contours from -24.9744 m/s to
104.2507 m/s.

Figure 26: k- ε Standard model, showing x- velocity component contours from -22.24945m/s to
104.5776 m/s.
Figure 27: k- ε Realizable model, showing x- velocity component contours from -20.88272 m/s
to 104.1366 m/s

Vertex Average of the skin friction coefficient plots

Fig 32: k- ω SST model, Drag report

Fig 33: k- ω SST model, Lift report

Fig 34: k- ε Standard model, Drag report.

Fig 35: k- ε Standard model, Lift report.

Fig 36: k- ε Realizable model, Lift report.

Fig 37: k- ε Realizable model, Drag report.

Fig 38: k- ε RNG model, Drag report.

Fig 39: Excel Calculations

DISCUSSION
The target goal was reached as the simulation of air flow around a square cylinder was
successful. It is shown that with comparison to the data generated over time by other researchers
that most of their experimental and numerical data coincides with simulated design numerical
data obtained.
In the simulations several parameters used to reach the target goal was; the Reynolds number of
20, 000, the air density used was 1.225 kg/m2 thus at standard sea level, and a flow over a region
of area 0.001m2. From the data generated graphs of the results are shown in the results above.
The variation of the air velocity about the body can be seen to be smoother at the side edges and
a turbulence flow at the sharp corner edges. The velocity is also seen to reduce towards the wake
region. This tends to create vortices that have significant effects on the aerodynamic and
structural force aspects of the body. The values of the velocity had also several impact on the
flow simulation.
The scaled residual values from the simulation were also accounted for, and thus graphs showing
various residuals were generated. These residuals show variation patterns with iterations, were
the number of iterations is 1000. These observations were very clear as the simulation was
investigated in 2D space planar analysis and a transient time.
 CONCLUSION
In the report, the 2D finite volume method is used in the study of the unsteady turbulent flow of
around a square cylinder with the use of a Reynolds number of 20,000. The Reynolds averaged
Navier-stokes equations methods was applied to the unsteady flow past a square cylinder.
Comparing the numerical results of studies conducted by other researchers with similar raw data
shows that the results obtained experimentally and numerical results obtained by using ANSYS-
Fluent software simulation coincide. The most agreeable results are obtained from the k-ω SST
turbulence model, among the different models used. The agreement between the experimental
data and the results obtained from the k-ω SST turbulence model show the degree of accuracy of
the simulation software. This is shown in the plot of aerodynamic parameters like CL and CD
against flow time. In addition, the behavior of results of different fluid properties such as
pressure, density, velocity etc., are presented and compared. The use of computational fluid
dynamics (CFD) technic for data analysis is faster than analyzing the raw experimental data
manually, thus providing a faster convergence of reliable results efficiently.

References
(n.d.). Retrieved 02 21, 2021, from prediction of unsteady flow around square cylinder using
RANS: www.researchgate.net >publication>
Alex , M. B., Luiz, A., & Miguel, H. H. (2009). simulation of viscous flow around a circular
cylinder near a moving ground. mechanic depertment science and engineering. ABCM.
Retrieved 03 12, 2021, from https:www.scielo.br/pdf/jbsmse/v31 n3/a10v31n3.pdf
Fuyou XU, Xuyong, Y., & Zhe, Z. (2011). Prediction of unsteady flow around a square cylinder
using RAN. applied mechanics and materials. Retrieved 03 06, 2021, from
www.scientific.net/AMM.52-54.1165
Lakshmipathy,S. (2004). PANS Methods for turbulence simulation of high and low Reynolds
number flow past a circular cylinder. Taxes A&M university, Aerospace engineerin.
Retrieved 03 12, 2021
Md. Mahbubar , R., Mashud, K., & Abdul, A. (2007, june). Munerical ingestigation of unteady
flow past a circular cylinder using 2 -D finite volume methods. journal of naval
architecture and marine engineering. Retrieved 03 12, 2021
Mehmet , I. y., & Dalshad , k. a. (2006). a numerical analysis of fluid flow around a circular and
square cylinder. university of Gaziantep, depertment of civil engineering, Turkey.
Retrieved 201
Rashko. A. (1959). Vortex Shedding from circular cylinder at low Reynolds numbers.
Versteeg H.K, & Malalasekera. W. (1995). An introduction to computational fluid dynamics.
England. Retrieved 03 12, 2021