Professional Documents
Culture Documents
4 RESULTS
4.1 Sectional static aerodynamic force coefficients
Sectional aerodynamic forces of the twin-box girder include drag force (Fd), lift force (Fl) and
pitching moment (M), which are calculated by integrating the surface pressure around twin-box girder.
These aerodynamic forces and moment are then normalized to form:
𝐹𝑑 𝐹𝑙 𝑀
𝐶𝑑 = , 𝐶𝑙 = , 𝐶𝑚 =
1 2 1 2 1 2 2
2 𝜌𝑈 𝐷 𝑑𝑧 2 𝜌𝑈 𝐶 𝑑𝑧 2 𝜌𝑈 𝐶 𝑑𝑧
where, Fd is drag force, Fl is lift force, M is pitching moment, ρ is air density, U is mean wind speed, C is
chord length, D is deck height and dz is the length of the cross-section of the deck.
Throughout this paper, positive lift and moment are considered in the upward direction and
clockwise rotation respectively. In addition, positive drag is taken in the downstream wind direction, which
is along the direction of the x-axis. This is shown clearly in Fig. 2 below.
Fig. 2. Sign convention.
Fig. 3a, 3b and 3c below displays the mean pressure distribution for a deck configuration only, a
deck under rod-induced turbulence and a deck installed with guide vanes under rod-induced turbulence
respectively. Regions of flow separation and reattachment can be distinguished by fluctuations in pressure
distribution.
Fig. 3. Mean pressure distribution: (a) bare deck; (b) deck + cylinder; (c) deck + cylinder + guide vanes
a
Fig. 4. Pressure standard deviation distribution: (a) deck; (b) deck + cylinder; (c) deck + cylinder + guide vanes
For the twin-deck only scenario, the flow separation at the leading edge of the upstream deck is
highlighted by the large negative pressures, which in turn will induce a positive mean pitching moment.
The presence of cylindrical rod causes the growing region of constant pressure underneath that would
generate a negative pitching moment. Furthermore, the presence of a small circular cylinder placed
upstream of the deck resulted in the alteration of flow fields at the wake region elongated in the direction
of the flow. Consequently, the velocity outside the boundary layer at the separation point is reduced, which
would in turn affecting the pressure distribution at the leading edge of the upstream deck and to some extent
the downstream deck. Indeed, highly concentrated vorticity structures imply the presence of high
perturbation energy in the wake, which is directly proportional with both the instantaneous and mean drag
forces. The flow field around the deck with due to the presence of upstream cylinder is noticeably different
from that of without cylinder. Hence, the pressure distribution at the leading edge of the upstream deck
(forebody) is slightly different, and the pressure returns to values not very far from those of deck-only in
the afterbody. Note that there is a slight reduction in drag forces due to the presence of upstream flow
disturbance which would introduce different distribution, thereby the introduction of cylindrical rod acts as
a buffer which would reduce pressure drag on the deck.
Conversely, the force component of the deck due to circular cylinder in the across-wind direction
(lift force) is substantially quite different than for the case without. This can be explained due to a non-
symmetrical potential flow induced by the cylinder, which present a rapid and impulsive vortex
development. The positive vorticity induces a stronger vortex passing over its lower surface, which in turn
would cause the flow to accelerate and the resulting velocity difference produces a pressure difference and
a resultant negative downward lift force.
Guide vanes act as effective vortex mitigation devices, which are installed along the underside of
the twin-deck. Mean pressure distribution and its standard deviation depicted in Fig. 3 and 4 for both cases
with and without the vanes are closely interlinked with flow separation and reattachment. When examined
further, the negative pressure distribution of the upstream deck fitted with guide vanes is fairly constant and
low in magnitude. The vanes ensure that flow remains attached to the bottom non-streamlined surface of
the upstream deck and hence, lift forces is expected to be positive while moment coefficient remains more
negative. In addition, drag force is seen to increase for decks fitted with vanes, which can be explained by
the larger positive pressures at the downside of the downstream deck. However, the upstream deck without
fitted vanes has large negative pressures peaks at the upper corner of its streamlined face, explaining for
the greater negative lift. The significant decrement in standard deviation of the drag and lift coefficient for
deck-fitted vanes is explained by the suppression of vortices and vortex-induced oscillations. Supporting
this statement is the lower magnitudes of standard deviation pressure in Fig. 4 for decks fitted with guide
vanes and those without.
Following this, standard deviations pressure distribution of the deck under laminar flow, deck under
rod-generated turbulent flow and deck-fitted vanes are also displayed in Fig. 4a, 4b and 4c accordingly.
Standard deviation of the aerodynamic forces (Cʹd and Cʹl) and moment coefficient (Cʹm) are also recorded
to examine fluctuations on the forces and moment of the twin-deck girder. These values are recorded in
Table 2 below.
Cm Cd Cl Cʹm Cʹd Cʹl St
Bare deck 0.028 0.146 -0.234 0.109 0.058 0.524 0.301
Deck + cylinder -0.031 0.116 -0.304 0.104 0.053 0.479 0.291
Deck + cylinder -0.106 0.138 0.073 0.133 0.033 0.214 0.134
+ guide vanes
Table 2. Drag, lift and moment coefficient and its standard deviation for deck, deck + cylinder and deck + cylinder + guide
vanes.
ca
Fig. 5. Pressure around computation domain: (a) deck; (b) deck + cylinder; (c) deck + cylinder + guide vanes.
The deck in smooth, laminar flow serves as a control for deck under rod-generated turbulent flow.
The mean pressure distribution for the bare deck and deck and cylinder are displayed in Fig. 2a and Fig.
2b, has several noticeable differences amongst one another which can be explained by closer examination
of Fig. 7 and 8.
Under smooth laminar flow, streamlines in Fig. 8a shows flow separation occurring at the top
surface of the streamlined section of the upstream deck. This phenomenon is resulted by the large adverse
pressure gradient recorded at the specified location, as shown in Fig 3a. Consequently, the increase in the
fluid pressure downstream has resulted in decreasing velocity (i.e. kinetic energy) being recorded in Fig.
6a and a zero-velocity region starting at the far-left corner of the upstream deck, suggesting flow separation.
On the other hand, no flow separation is observed upon addition of the rod. Turbulence formed by the
laminar separation of flow in the wake region of the cylinder, possesses increased transport of momentum
and the turbulent boundary layer which is formed has an increased wall shear stress. Separation is therefore,
prevented as shown in Fig. 8b, when compared to the bare deck in Fig. 8a. Pressure differences between
the leading and trailing edge of the upstream deck caused by flow separation explains the higher drag (Cd)
coefficient compared to deck under rod-generated turbulent flow recorded in Table 2. Due to the increase
drag force, the boundary layer observed in Fig. 8a is thicker than that of Fig. 8b.
Reduced drag and lift forces indicated in Table 2 can also be deduced by examining the mean
pressure distribution around the deck under smooth, laminar flow and rod-induced turbulent flow.
Differences in pressure distribution is caused by differences in flow separation as mentioned above. Under
rod-induced turbulent flow, magnitudes of the negative pressure on the bottom surface of the upstream deck
is greater compared to that of its top surface. As a result, flow is more accelerated over the more streamlined
bottom surface of the deck, leading to a lower drag force and an increased negative lift, in turn inducing a
negative pitching moment.
Further analysis of the vorticity field of the two scenarios depicted in Fig. 7a and 7b provides the
same conclusion. Under rod-generated turbulent flow, eddies are shed as soon as the flow hits the stagnation
point of the cylinder. As the flow progresses downstream, fewer eddies are continuously shed at the top
surface of the upstream deck, while eddies are no longer generated as the streamline from the cylinder
reattaches to the bottom surface of the upstream deck (Fig. 8b) after circulation. Fewer eddy formation
emphasizes the notion that flow from the cylinder remains continuous and that separation does not take
place upstream deck. Unlike the deck under turbulent flow, the formation of eddies for the bare deck
occurred immediately at the leading edge of the upstream deck as the smooth, laminar flow strikes the sharp
corner of the deck.
The placement of the rod causes flow to strike the cylinder at its stagnation point directly, inducing
high pressures on its free side, as shown in Fig. 5b. Since the inflow is smooth and laminar with low
Reynolds number (Re), vortices are shed alternatively from the lower and upper surface of the cylinder and
into the wake region of the cylinder, forming what is known as the von Kármán vortex street. Shown in
Fig. 7b, the upstream deck is immersed in these unsteady vortices. Hence, being sheltered by the cylinder
forms a region of low pressure and velocity between the cylinder and upstream deck as shown in Figure
Fig. 5b and Fig. 6b.
By closer examination of the streamlines around the downstream deck in Fig. 8b, a delay in flow
separation is observed, when compared to Fig. 8a. Again, this might be a result of the turbulent flow formed
by the cylinder, which has a Reynolds number (Re) greater than the critical Re. This causes an earlier
transition from laminar to turbulent boundary layer flow in the downstream deck. Thus, flow separation is
delayed with the addition of the cylinder. Moreover, a shorter separation bubble is observed for the deck
under smooth flow compared to that under turbulent flow. Streamlines are seen to reattach downstream the
leeward deck in Fig. 8b, while this phenomenon is not identified in Fig. 8a. No flow reattachment is
identified for the deck under smooth flow, since pressure distribution at the top surface of the downstream
deck remains negative and large in magnitude.
Results obtained in for the downstream deck is in agreement to the conclusions drawn by Kwok et
al. (1986) [4], except for the upstream deck. The study concluded that an increase in free-stream turbulence
caused by the placement of the cylinder promotes an earlier transition of laminar to turbulent boundary
layer flow; and hence, a delayed flow separation. For the upstream deck, no flow separation is inferred from
Fig. 8a, which is slightly different to the results obtained by previously mentioned study.
The deck without the cylinder has large negative pressures, which are induced by flow separation
are located further away from the rotational center, particularly near the leading edge of the upstream deck.
Hence, a large positive mean pitching moment is expected. Nevertheless, since no flow separation is
identified in Fig. 8b, negative pressures are not distinguished at the leading edge of the upstream deck of
Fig. 5b. This is demonstrated in the positive Cm for the configuration under smooth flow and a negative Cm
for the other under turbulent flow.
No significant differences were identified in the mean pressure distribution and the standard
pressure distribution of the downstream deck, except for their magnitudes. As shown in Fig. 4a and 4b, the
magnitudes of the standard deviation pressure distribution of the rod-induced turbulent flow is examined to
be much higher than that of the smooth, laminar flow. Vortex shedding impeding the downstream deck in
the bare deck is mostly sourced from the upstream deck and the gap, however with the implementation of
the cylinder, the additional vortex shedding caused by the placement of the cylinder accounts for the higher
magnitudes of standard deviation pressure recorded in Fig. 5b. Further, since both cases have the
downstream deck being immersed in the wake of the vortices shed by the upstream deck, positive pressures
are recorded at the non-streamlined windward surface of the downstream deck.
Periodic shedding of vortices is associated with oscillating lift and drag forces, and is represented
by Cʹl and Cʹd. The relatively unchanged Strouhal number, St and the pattern of vortices being shed, shown
in Fig. 7a and 7b suggest that the frequency of vortices being shed remains approximately the same in both
cases. Hence, it is expected that standard deviations of lift, drag and moment coefficients remain relatively
constant.
Fig. 6. Velocity around computational domain: (a) deck; (b) deck + cylinder; (c) deck + cylinder + guide vanes.
Fig. 7. Vorticity field around computational domain: (a) deck; (b) deck + cylinder; (c) deck + cylinder + guide vanes.
5 CONCLUSION
Computational Fluid Dynamics (CFD) was used in this project to analyze the effect of a rod-generated
turbulence and guide vanes on the static aerodynamic force coefficients and vortex shedding of a twin-deck
configuration bridge. Three tests setups, with constant gap-width to total chord ration were modelled and
meshed accordingly.
It was shown that mean pressure distribution around the decks under rod-induced turbulence is relatively
similar to that under smooth, laminar flow. This is perhaps due to the placement and size of the cylinder
being relatively inadequate in simulating high turbulent flow. However, under rod-induced turbulent flow,
a number of critical conclusions can be made, which includes no flow separation around the upstream deck
and a delayed flow separation for the downstream deck. Further, flow is observed to be accelerated along
the bottom surface of the streamlined upstream deck. Hence, under rod-induced turbulent flow, there is a
reduction in moment, drag and lift forces.
Following this, the implementation of the guide vanes in the deck mitigates vortex-induced vibrations
generated by upstream deck in the gap spacing between the two decks. They introduce a region of low
velocity fluid in the wake region of the upstream deck, suppressing the formation of vortices. However, this
paper fails to study the effect guide vanes has on the flutter performance of the twin-deck configuration,
which can be included in future pursuit of this research.
The study conducted in this paper lacks further examination on the aerodynamic performance of the twin-
deck girders at different turbulence characteristics. Therefore, future studies must include the installation
of different cylinders of various diameters placed at various distances upstream of the deck. This will allow
for the examination of the decks’ performance under different turbulence intensity and turbulence integral
scale. Furthermore, experimentally obtained data through wind tunnel testing is required to validate the
results produced by the CFD in this study, since the use of CFD in estimating forces of an oscillating bridge
remains limited, especially for turbulent flows.
6 REFERENCES
[1] KWOK, K. C. S., QIN, X. R., FOK, C. H. & HITCHCOCK, P. A. 2012. Wind-induced pressures around
a sectional twin-deck bridge model: Effects of gap-width on the aerodynamic forces and vortex
shedding mechanisms. Journal of Wind Engineering and Industrial Aerodynamics, 110, 50-61.
[2] LAIMA, S., LI, H., CHEN, W. & OU, J. 2018. Effects of attachments on aerodynamic characteristics
and vortex-induced vibration of twin-box girder. Journal of Fluids and Structures, 77, 115-133.
[3] SÁNCHEZ, R., NIETO, F., KWOK, K. C. & HERNÁNDEZ, S. CFD analysis of the aerodynamic
response of a twin-box deck considering different gap widths. Proceedings of the 2015 Congress
on Numerical Methods in Engineering (CMN 2015), Lisbon, Portugal, 29 June to 2 July 2015,
2015.
[4] KWOK, K. C. S. & KWOK, K. C. S. 1986. TURBULENCE EFFECT ON FLOW AROUND
CIRCULAR CYLINDER. Journal of Engineering Mechanics, 112, 1181-1197.
[5] NIETO, F., KUSANO, I., HERNÁNDEZ, S. & JURADO, J. Á. CFD analysis of the vortex-shedding
response of a twin-box deck cable-stayed bridge. Proceedings of the 5th International Symposium
on Computational Wind Engineering, Chapel Hill, NC, 2010.
[6] LARSEN, A., ESDAHL, S., ANDERSEN, J. E. & VEJRUM, T. 2000. Storebælt suspension bridge –
vortex shedding excitation and mitigation by guide vanes. Journal of Wind Engineering &
Industrial Aerodynamics, 88, 283-296.