AJAMC, Vol 2, Issue 1, 2022 19

Simulation of flows with different shaped cylinders

using CFD
Anjan Samanta
Department of Basic Science and Humanities,
University of Engineering and Management, Kolkata, India
We are going to discuss how turbulent flows are being affected in presence of a suitable
bluff body which is considered as a cylinder. Simulation has been performed using
FEAtool Multiphysics to analyse the flow numerically. Subsequently two post simulation
analysis have been done namely pressure difference and drag coefficient. Lastly one
simulation has been performed for a turbulent flow past square cylinder.

1 Introduction tend to be turbulent. It is defined as:

ρuL uL
In fluid dynamics, a bluff body is defined as that Re = =
µ ν
body whose surface is not aligned with the stream-
lines, when placed in the flow. For this reason flow where L is a characteristic linear dimension (m)
separation occurs at the leading edge of the body and µ is 2the dynamic viscosity of the fluid (P as
and it experiences huge pressure drag; and minimal or N s/m or kg/(ms))
friction drag. Due to the presence of bluff body
in flow we are having two phenomenon - boundary 2.3 Drag Coefficient
layer and wake region. Because of flow separation Drag coefficient (c ) is a dimensionless quantity
d
a disturbed region can be observed behind the bluff that represents how much a body is being dragged
body- known as wake region. These phenomenon in a turbulant flow. It is defined as:
can be analysed experimentally; but with the help
2Fd
of software we can predict these stuffs almost accu- cd = 2
rately. That’s what CFD is all about - Simulation ρu A
of fluid flow and numerical analysis of the proper- where:
ties related to fluid. I find this field very interesting Fd is the drag force, which is the force applied
and we can explore a lot of things here. on the body due to pressure along the flow velocity
diretion.
A is the reference area, which is usually the 2D
2 Review of Literature projection of the frontal area.

2.1 Navier Stokes equation 2.4 CFD - a Brief Introduction

The description of incompressible fluid flow can be CFD (Computational Fluid Dynamics) is a numeri-
represented in terms of set of partial differential cal and simulation technique by which one can anal-
equations; known as Navier Stokes equation. They yse different aspects of fluid flow which are diffi-
are as follows: cult or expensive to derive from experimental setup.
Computers are used to perform the calculations re-
∂ui ∂ui 1 ∂p ∂ ∂ui quired to simulate the free-stream flow of the fluid,
+ uj =− + (ν )
∂t ∂xj ρ ∂xi ∂xj ∂xj and the interaction of the fluid with surfaces de-
fined by boundary conditions. With high-speed su-
where u is the velocity in the streamwise direction, percomputers, better solutions can be achieved, and
p is the pressure, ρ is the fluid density and ν is the are often required to solve the largest and most com-
kinematic viscosity of the flow. plex problems. In present day we have some very
advanced softwares like ANSYS FLUENT, open-
2.2 Reynolds Number FOAM, FEATool multiphysics to analyse flow spe-
cially turbulent flow. But it is very much required
The Reynolds number (Re) helps predict flow to validate the softwares; this can be done by per-
patterns in different fluid flow situations. At low forming experiment. In addition, previously per-
Reynolds numbers, flows tend to be dominated by formed analysis of a particular problem can be used
laminar flow, while at high Reynolds numbers flows for comparison.
anjan.samanta@uem.edu.in
American Journal of Applied Mathematics and Computing
US ISSN: 2689-9957 ©2019 Society for Makers, Artist,Researchers and Technologists
website:http://thesmartsociety.org/ajamc/
 AJAMC, Vol 2, Issue 1, 2022 20

Figure 1: Computational Domain

Figure 2: Grid generation

American Journal of Applied Mathematics and Computing

 AJAMC, Vol 2, Issue 1, 2022 21

CFD has a wide range of application in re-

search and engineering problems in many fields
-------------------------------------------------------------------------
of study and industries, including aerodynamics it | def-u def-v def-p | rel-u rel-v rel-p |
-------------------------------------------------------------------------
and aerospace analysis, weather simulation, natural 0 | 0.0071 0.00014 0.02 | |
1 | 0.0071 0.00014 0.02 | 1 1 1 |
science and environmental engineering, industrial 2 | 0.0035 0.0036 0.0028 | 0.15 0.49 0.26 |
3 | 0.0014 0.00064 0.0011 | 0.058 0.14 0.057 |
system design and analysis, biological engineering, 4 | 0.00047 0.0002 0.00034 | 0.017 0.052 0.024 |
5 | 0.00014 4.9e-05 9.2e-05 | 0.0055 0.017 0.006 |
fluid flows and heat transfer, and engine and com- 6 | 3.9e-05 1.4e-05 2.3e-05 | 0.0016 0.0047 0.0014 |
7 | 1e-05 4.4e-06 6.1e-06 | 0.00044 0.0013 0.00033 |
bustion analysis. 8 | 2.5e-06 1.2e-06 1.6e-06 | 0.00012 0.00032 8.7e-05 |
9 | 6.4e-07 3e-07 3.9e-07 | 3.1e-05 8.2e-05 2.3e-05 |
10 | 1.7e-07 7.7e-08 9.6e-08 | 8.6e-06 2.2e-05 6.2e-06 |
11 | 4.7e-08 2e-08 2.5e-08 | 2.5e-06 6.2e-06 1.7e-06 |
12 | 1.4e-08 5.5e-09 7e-09 | 7.4e-07 1.9e-06 4.9e-07 |
3 Simulation of flow past 13 | 4.2e-09 1.6e-09 2.1e-09 | 2.2e-07 5.6e-07 1.5e-07 |
-------------------------------------------------------------------------

cylinder --------------------------------------------------
Solver Statistics (solvestat)
--------------------------------------------------
Number of dependent variables
3.1 Numerical Details --------------------------------------------------
u : 3183
v : 3183
3.1.1 Computational Domain p : 3183
--------------------------------------------------
Simulation timings
The computational region we have taken for flow --------------------------------------------------
t_asm(A) : 2.9s \ 39%
simulation is the following diagram: t_asm(f) :
t_sparse :
0.0s \
0.6s \
0%
8%
a rectangular tunnel with a cylinder is being t_bdr
t_lc/mv :
: 2.0s \ 26%
0.0s \ 0%
taken for simulation. Dimension of rectangle is t_solve :
t_tot :
1.3s \ 17%
7.3
2.2 × 0.5. Circle of center (0.25, 0.25) and radius 0.1 --------------------------------------------------

in the diagram represents the cylinder (bluff body).

3.1.2 Grid generation where

A grid is being generated in the computational do-

main of size 0.02 unit as follows: • Surface plot represents the velocity field;

3.1.3 Fluid density and viscosity

• Contour plot represents the pressure;
Fluid density is taken as 1 unit and viscosity as
0.002 unit.
• Arrow plot represents the direction of fluid flow
3.1.4 Boundary conditions i.e. velocity field.

• The inlet velocity profile (Boundary 4) for the

flow is described to be parabolic i.e.

u0 = 4 umax y(h − y)/h2

3.3 Post Simulation Analysis

Where umax = Maximum velocity at the center From the simulation of flow field we can conclude
line and h is the width of the channel/region. that a recirculation zone has been formed behind
Here h = 0.5 and umax = 0.35. the cylinder and the flow gradually becomes steady
laminar downstream.
• At outlet boundary 2 the pressure outflow is
taken to be zero.
• At cylinder surface and top - bottom bound-
3.3.1 Pressure difference
aries 1 and 3, we are assuming No-slip condi-
tion.
Using FEATool we can calculate the pressure differ-
ence between any two points (preferably front and
rear of cylinder) which would help us to validate the
3.2 Simulation simulation.
Simulation is being done in FEAtool Multi-
physics, a matlab software:
• Pressure at the point (0.15, 0.4) = 0.20525
Grid Statistics:
Grid Generation Done.
number of grid points: 3183
number of grid cells: 6064
grid cell min area:
grid cell mean area:
1.0351e-04
1.7625e-04 • Pressure at the point (0.4, 0.2) = 0.02383
grid cell max area: 3.5934e-04
grid cell min quality: 0.7769
grid cell mean quality: 0.9932
number of boundaries: 8
number of subdomains: 1 Hence pressure difference = 0.20525 − 0.02383 =
time for grid generation: 0.9591
0.18142

American Journal of Applied Mathematics and Computing

 AJAMC, Vol 2, Issue 1, 2022 22

Figure 3: Boundary Conditions

Figure 4: Simulation Obtained

American Journal of Applied Mathematics and Computing

 AJAMC, Vol 2, Issue 1, 2022
23

number,
ρUmean D
Re =
µ
where D represents the pipe diameter. For the
previous case Re = 100. Let us now increase the
Reynolds number and try to see how it affects the
fluid flow.
We assume D = 1, Umean = 0.5 so that
Umax ≈ 0.75, µ = 0.001 and ρ = 1. Then
Re = 500 and simulation in FEATool gives us
the following result. Clearly the wake region has
been increased behind the 1st cylinder because of
increment in Reynolds number, pipe diameter and
decrement in viscosity.

3.3.2 Drag coefficient

Drag coefficient can be determined by the follow- 6 Corner Modification - Pro-
2Fd
ing formula: cd = ρu2 A where Fd is the drag force posed work
defined as follows:
Z   We have seen how fluid behaves for circular and
∂uτ (t) square cylinder. Because of the presence of corner,
Fd = µ ny − pnx dS
S ∂n the flow structure is affected enough for the case of
square cylinder. So it will be interesting to see how
we can evaluate the above integral and find the drag corners are influencing the fluid flow. If we gradu-
coefficient. Here Cd = 11.334 ally convert the square to circle; i.e. corner radius
from zero to half of side length (r/D = 0.5) and
perform simulation that would be worth to investi-
gate.
In paper [3] research has been done on corner modi-
fication for corner ratio, r/D = 0, 0.167, 0.247, 0.333
and compared it with the properties of flow past
circular cylinder. We get conclusion from the paper
that the drag coefficient cd is inversely proportional
to the corner ratio values, except for r = 0.33. I
am proposing my work as follows: We will consider
corner ratio values from 0.25 to 0.5 and we will see
if we are getting same kind of anomaly or not; if so
we will try to understand why this kind of deviation
is happening. The reason could be also associated
with Reynolds number; hence we will perform our
simulations for different Reynolds numbers.
4 Flow past a square cylinder
References
Square cylinder means rectangular or almost rect-
angular sharp edged cylinder such as beams, fences [1] HP Kavya, Banjara Kotresha, and K Naik.
or buildings. Let us assume the same experimental “CFD Analysis of 2-D unsteady flow past a
set up like the previous one and perform the sim- square cylinder at an angle of incidence”. In:
ulation of flow past square cylinders. The diagram International Journal of Advanced Research in
obtained from FEATool given in Fig 6. Mechanical and Production Engineering and
Development 1.2 (2014), pp. 117–125.
[2] Veeralkumar Thakur, Tarun Yadav, and B Ra-
5 Simulation when we have jiv. “Drag optimization of bluff bodies using
two square cylinders in the CFD for aerodynamic applications”. In: Int J
Comput Eng Res 7.4 (2017), pp. 25–32.
flow
[3] Rujun Liu. “Flow Around Bluff Bodies with
The following simulation we are getting for pres- Corner Modifications on Cross-sections”. In:
ence of two square cylinders taking all the initial (2019).
conditions same as previous. We know Reynolds

American Journal of Applied Mathematics and Computing

 AJAMC, Vol 2, Issue 1, 2022 24

Figure 5: Flow past a square cylinder

Figure 6: Flow past two square cylinders under certain conditions

Figure 7: Flow past two square cylinders

American Journal of Applied Mathematics and Computing