Professional Documents
Culture Documents
Study of Flow Over Dimpled Cylinder For Drag Reduction
Study of Flow Over Dimpled Cylinder For Drag Reduction
Study of Flow Over Dimpled Cylinder For Drag Reduction
Chapter 8
271
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 272
8.1. Introduction
Over the years, there have been many attempts made in rationalizing
the definitions and terminology behind drag. Total drag is the force
corresponding to the rate of decrease in momentum. It is unfavourable
as it causes structural vibrations, and hence it is extremely detrimental
to many engineering structures. Therefore, drag reduction is of par-
ticular interest to many industries as it leads to higher durability of
Bio-Inspired Surfaces and Applications Downloaded from www.worldscientific.com
equipment.1−5
Drag reduction is done through extensive research on controlling
by MONASH UNIVERSITY on 09/26/16. For personal use only.
the fluid flow over bluff bodies. It can be achieved through two
main approaches: passive and active flow controls. Passive flow control
relies on the structural components of the bodies such as dimples and
riblets,3−5 and active flow control makes use of additional installation of
devices to direct wake flow structures such as an upstream installation
of a small control rod.
The focus of this chapter is to investigate the effects of dimples
on the surface of a circular cylinder. It is to induce transition and thus
create a turbulent boundary layer to delay the point of separation in
order to reduce drag.
Due to the viscosity of real fluid, a thin layer called the boundary
layer will form. When the upstream flow reaches the leading edge of
a flat surface, fluid starts to slow down and a laminar boundary layer
is created. However, the fluid climbs up over smooth rounded surface
of the cylinder due to favorable pressure gradient and may accelerate
if it is a suction surface. Due to the no-slip condition imposed by
the viscosity of fluid, the fluid slows down as it moves downstream.
Hence, the boundary layer begins to grow. As the laminar boundary
layer continues to grow, it will become unstable and transition to a
turbulent boundary layer. Turbulent flow is highly irregular and eddies
occur. When a viscous fluid flows over a cylinder, there will be vortices
in a thin boundary layer adjacent to the cylinder where occurrence
of boundary layer separation and a trailing wake will form eventually
behind the cylinder.1,2
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 273
Fig. 8.1. Flow comparison between laminar and turbulent boundary layer.6
Separation of flow from the surface will occur, which is the result of
a rising external pressure along the surface (adverse pressure gradient)
that resists the flow. In order to oppose this rising pressure, the fluid
particles lose their kinetic energy. Separation of flow is able to occur
for both laminar and turbulent boundary layers. However, for the
turbulent boundary layer, the greater momentum transfer due to eddies
will be able to raise the velocity near the surface. Therefore, it has more
kinetic energy to oppose the adverse pressure gradient. With this, the
point of separation is delayed (refer to Fig. 8.1 for laminar and turbulent
flow comparison), and hence drag is reduced.6
Therefore, some methods of drag reduction are to induce the
transition to turbulent boundary layer. In this chapter, having dimples
on a circular cylinder produces turbulence in the slow-moving air, and
the more energetic layer separates much later, thus leading to a smaller
wake and drag. This method was inspired by the case of golf balls, where
further distance is achieved, encouraging the incorporation of dimples
on circular cylinders.
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 274
cylinder, one on each side, and they were aligned along the length
of the cylinder, as shown in Fig. 8.5. On the other hand, there were
four pressure tappings on the dimpled cylinder, two on each side (one
will be inside a dimple, another outside a dimple) and aligned along
the length, as shown in Fig. 8.6. The extra two tappings as compared
to the smooth cylinder were used to investigate if there are pressure
differences inside a dimple and outside a dimple. Due to this drilling
operation, imperfections such as scratches and chafing were left on the
surfaces, affecting the relative surface roughness, which may lead to
higher drag force acting on the cylinder.
The tappings were connected to the thin-walled stainless steel pipe
of 20 mm in length and to the pressure transducer via flexible tubes as
shown in Fig. 8.7.
The pressure transducer which was used to measure the pressure
readings around the cylinder’s surface was with a digital indicator as
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 277
shown in Fig. 8.8. The model is Flotech Setra 230 with a pressure
range of ±17,000 Pa.
The wind tunnel experiment was carried out in the Nanyang
Technological University (NTU) closed loop wind tunnel facility. It
is the AF6407 Wind Tunnel with a test section 720 mm height ×
2000 mm length × 780 mm width. The wind tunnel experimentation
set up is shown in Fig. 8.9. The wind tunnel is able to provide an air
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 278
Fig. 8.12. Comparison of CD for dimpled cylinder with Bearman and Harvey.8
As for the analysis of the pressure readings, they were compiled and
calculated into their respective pressure coefficients, CP . The pressure
by MONASH UNIVERSITY on 09/26/16. For personal use only.
readings were taken from 0◦ to 180◦ with an interval of 15◦ . For the
smooth cylinder, two readings were taken for each angle and averaged.
For the dimpled cylinder, two readings were taken from inside the
dimples and two from outside the dimples. The CP is then plotted
against the angle θ (for the case of Re 60,000) as presented in Fig. 8.13.
From the graph, it can be noticed that the curve corresponding
to the smooth cylinder is rather far away from the theoretical curve
calculated using potential flow theory. On the other hand, both inside
and outside dimple curves do not differ much from each other and
have a lower-pressure region between 45◦ and 120◦ . This shows that
the instantaneous ones. This may cause the vortex shedding to be less
observable.
Under the RANS turbulence modeling, there are many models to
choose from in FLUENT. In industrial CFD, the two most widely used
models are k-epsilon (k-ε) and k-omega (k-ω) models.
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 284
solving. The PISO scheme is used for transient flow problems, which is
applicable in this simulation. The discretization schemes are of second
order as they provide better accuracy than first order ones. However,
when the PISO scheme was selected for the dimpled cylinder simu-
lation, numerous errors were detected and simulation was not able to
proceed smoothly. The errors might be caused by the poor mesh quality
of the dimpled cylinder with partially structured mesh being used.
Due to these errors, the pressure-velocity coupling scheme was
changed to Coupled and simulations were able to continue without
further errors. The FLUENT input parameters applied in the problem
setup were tabulated in Table 8.2.
Appropriate time step sizing is required to achieve the best results
possible without sacrificing valuable computational resources. The
smaller the time step size, the longer the computational time despite
its ability to provide more accurate results. The required time step
size can be calculated by using the Strouhal (St) number, which is
The simulations were allowed to run for 5s and the CD results were
averaged over the number of time steps. For both smooth and dimpled
by MONASH UNIVERSITY on 09/26/16. For personal use only.
Re number Velocity (ms−1 ) Time stepping size (s) No. of time steps
Fig. 8.17. Graph of CP against θ with simulated and experimental results for Re of
60,000.
From Fig. 8.17, it can be seen that the CP values are close for the
dimpled cylinder (blue, yellow and orange curves) before 60◦ . There is
a slight discrepancy at 60◦ and 75◦ for the smooth cylinder where the
experiment captured a higher CP . There are much larger discrepancies
at angles after 105◦ between the experimental and simulated CP values.
The experimental CP values are much lower than the simulated values.
This is because during the wind tunnel experiment, the recorded
pressure readings were observed to be extremely low from one side
of the cylinder as illustrated in Table 8.5 for the smooth cylinder at Re
140,000.
From Table 8.5, it can be seen that the measured pressure readings
at 75◦ differ by as much as −670 Pa. This resulted in a far less accurate
average pressure reading. This problem arised for the smooth cylinder
for all Re (60,000 to 140,000) from 30◦ onwards, but was not so
obvious for the dimpled cylinder. Initially, the authors suspected that
the extreme low pressure detected (on the right side from the direction
of air flow) may be due to the pressure tappings. Hence, the cylinder
was flipped over and the experiment repeated. However, the extreme
low pressure was again detected at the right side. Thus, it may be other
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 288
Smooth Dimpled
% %
Re Experiment Simulated Difference Experiment Simulated Difference
disturbances within the wind tunnel which affected both pressure and
drag force readings.
The overall trends for the graphs are noticeable for all Re and
the respective graphs were plotted from Figs. 8.A.1 to 8.A.8 in the
Appendix.
Both the current experimental and simulated results were compared
for validation of the accuracy of the simulation. Only comparisons
between Re 60,000 to 140,000 were made as this was the Re range
for the experiment, and the CD results are shown in Table 8.6.
The differences observed in all results were approximately 35 to
40%. These deviations can be attributed to the following reasons. The
high experimental CD results may be due to the interference in the
air flow in the wind tunnel set up due to the pressure transducer and
tubings as well as the scratches made on the cylinders. The simulations
conducted have many limitations such as element count restriction and
selection of the RANS turbulence modeling. The surface roughness of
the material was also not accounted for in the CFD simulations.
8.5. Conclusion
and improving mesh quality can also yield more representative results.
Using more computational intensive turbulence modeling such as LES
can also help in capturing and understanding the difference in the
vortex shedding of the cylinder.
Opportunities to expand this study are aplenty. With a reliable and
accurate CFD model, the work can be expanded to optimization of the
sizing, shape and arrangement of dimples on the cylinder. Studies on
the effects of dimpled cylinder on vortex shedding and vortex-induced-
vibration are also of great interest. Low-Re analysis on dimples on micro
tubes may be of interest to the biomedical industry.
While these challenges are addressed, the usage of dimpled cylinders
can be far ranging in many different industry and applications.
References
[1] Pang S, Ng EYK, Chiu WS. (2013) Comparison of Turbulence Models in Near
Wake of Transport Plane C-130H Fuselage. AIAA J 50(3): 847–852. [DOI:
10.2514/1.C032031]
[2] Ng EYK, Pang S, Chiu WS. (2013) Computational Aerodynamics of the C-130
Predicting Airflow Effect on Airdrop. International Journal of Computational
Methods 10(5): 1350025–1350033. [DOI: 10.1142/S0219876213500254]
[3] Luo Y, Liu YF, Zhang D, Ng EYK. (2014) Influence of Mor-
phology for Drag Reduction Effect of Sharkskin Surface, Journal of
Mechanics in Medicine and Biology 14(2): 1430001-1–14300016. [DOI:
10.1142/S0219519414500298]
[4] Luo Y, et al. (2015) Chemical, Mechanical and Hydrodynamic Properties
Research on Composite Drag Reduction Surface Based on Biological Sharkskin
Morphology and Mucus Nano-long Chain. Journal of Mechanics in Medicine
and Biology, accepted.
May 24, 2016 13:32 Bio-Inspired Surfaces and Applications 9in x 6in b2378-ch08 page 290
Appendix