Effects of Bridge Pier Friction on Flow Reduction in a

Navigation Channel
Abstract- The Rambler Channel is an 8-kilometer long navigation waterway. A number of
bridges are built on piers over the channel; the presence of bridge piers changes the nearby
flow regime and induces friction in the ambient tidal currents. This paper presents a
hydrodynamic model to study the effects of bridge pier friction on flow reduction in the
Rambler Channel. Energy losses due to the bridge piers are estimated by introducing additional
quadratic friction terms into momentum equations. Comparisons of the tidal
flow patterns, changes in flow rate through the channel, and flow paths in the surface water
layer by drogue tracking simulations for situations with and without bridges are presented.
Keywords- Pier Friction; Flow Reduction; Hydrodynamic Model; Drogue Tracking
I. INTRODUCTION
The Rambler Channel is located to the northwest of Hong Kong Island, separating the Kowloon
Peninsula and Tsing Yi Island. The channel is a busy waterway for use by various commercial
vessels. In recent decades, reclamation was conducted along the southern section of the channel
to form new land for container port facilities. The width of the channel now varies from 1 km to
less than 300 m, and the length of the channel is approximately 8 km. There are ten water control
zones in Hong Kong; the Rambler Channel is located in the Western Buffer Water ontrol Zone. A
set of water quality objectives specifying permissible concentrations of pollutants in marine
water is applied in this water control zone. The water near the southern end of the channel is
influenced by discharge from Stonecutters Island Sewage Treatment Works, which employs a
chemically enhanced primary treatment process with relatively high E. col levels (110 3,000
count/100mL in 2013). Polluted discharge from two box culverts near the middle of the channel
causes water pollution. Dispersion of pollutants is highly dependent on the prevailing tidal
movement in the channel.
II. THE BRIDGES
There are seven bridges which cross the channel: Ting Kau Bridge, Tsing Yi North Bridge, Tsing Lai
Bridge, Cheung Tsing Bridge, Tsing Yi South Bridge, Kwai Tsing Bridge and Stonecutters Bridge.
Figure 1 shows the locations of the seven bridges. Amongst all these bridges, Tsing Lai Bridge is
the only rail bridge for Mass Transit Railway and Airport Express; all of the others are road
bridges. The two most recently completed bridges, including Stonecutters Bridge (2009) and Ting
Kau Bridge (1998) are cable-stayed bridges. Stonecutters Bridge is 1,600 m in length, and Ting
Kau Bridge is 1,177 m in length. Tsing Yi North Bridge opened in 1987 and is a pre-stressed
ntilever bridge, which spans 160 m across the channel. Tsing Yi South Bridge opened in 1974 and
is approximately 610 m wide. Kwai Tsing Bridge is a duplication of Tsing Yi South Bridge and was
built in parallel, reducing the traffic load on the Tsing Yi South Bridge. Cheung Tsing Bridge is
located to the immediate north of Tsing Yi South Bridge, connecting Cheung Tsing Tunnel and
Tsing Kwai Highway. Tsing Lai Bridge was completed in 1997 and has a length of 1.1 km, providing
a railway link to Tsing Yi Island and the Hong Kong international airport at Chek Lap Kok. All the
bridges are supported by bridge piers; both caissons and reclaimed islands were used as
foundations for the bridge piers. Stonecutters Bridge is one of the longest cable-stayed bridges
in the world and its bridge piers, which support the deck over Rambler Channel, are constructed
on both sides of the reclaimed land. The bridge itself does not cause disturbance to the tidal flow

in the channel. However, the bridge piers of the other bridges are constructed in the channel,
blocking the movement of tidal flow. In this study, a hydrodynamic model is used to investigate
the influence of the bridge piers of individual and groups of bridges on the tidal exchange in the
Rambler Channel. Energy losses due to bridge piers are estimated by introducing quadratic loss
terms into momentum equations. This paper presents a prediction model to investigate the
changes in tidal flow rate, flow pattern and the movement of surface water in the channel.
III. MODEL SETUP

A. Hydrodynamic Model
The dredging of marine sediments to deepen the seabed level to provide adequate water depth
for container vessels and reclamation to provide new land were conducted on the southern
section of Rambler Channel. There was no identified area of high ecological value within the
channel. The existing beneficial uses of the channel are the bathing beaches on the northern
part, and cooling water intakes along its length.
The Civil Engineering and Development Department [1] studied the feasible dredging scenarios
which would not cause adverse impact to water quality in order to increase the water depth for
Kwai Tsing container basin and its approach channel. A numerical model was used to conduct
both hydrodynamic and water quality simulations of the study area, which encompassed the
Rambler Channel and adjacent water bodies. Morelissen et al. [2] generated a tidal stream atlas
of the stratified tidal flows near Stonecutters Bridge using a three-dimensional numerical model
with horizontal large eddy simulation. Complex flow patterns such as tidally driven eddies and
recirculations were included to predict water levels and currents in the channel. Pun et al. [3]
established a three-dimensional hydrodynamic regional model to study water movement and
dispersion characteristics of pollutants in the Rambler Channel. The regional model is used to
provide open boundary conditions for the present study to simulate tidal exchange through the
channel for situations with and without the effects of bridge piers. The present model has been
developed using the Delft3D-FLOW module of the Delft3D model suite. The model grid is refined
to increase the resolution in the channel, and there are a total of 3,003 grid cells. The sizes of grid
cells within the channel vary from approximately 60 m in the inner portion of the channel to 330
m at the channel entrances. The water column is divided into ten layers to allow for vertical
variations in tidal speed, direction, temperature and salinity. The modelling period is between
June and July of 2012 and covering a total of 22 days, including a complete spring-neap tidal cycle
during the wet season. The first several days of the simulation are used for model spin up. A oneminute time step is adopted to avoid instability in the model simulation.
B. Bridge Pier Friction
The general schematization of the geometry in a numerical model generates various grid sizes in
order to cover the entire model domain. The dimension of a bridge pier is generally much smaller
than the smallest grid size of the model. One method to model the local resistance of bridge piers
is to add quadratic friction terms in both the x- and y-directions of the momentum equations to
approximate the energy losses generated by the bridge piers. Farraday and Charlton [4]
established the relationship between flow resistance and the presence of bridge piers. A
quadratic friction term at the bridge pier location can be added to the momentum equation to
estimate the energy loss. This method requires the determination of a friction loss coefficient.

With reference to the Delft3D-Flow User Manual [5], the quadratic friction terms in the u- and vcoordinate directions (Mu and Mv) of the momentum equations (in m/s2) can be expressed as:

where cLu and cLv are the energy loss coefficients in the u- and v-coordinate directions,
respectively; u and v are the velocity components (m/s); and x and y are the grid sizes (m) in
the u- and v-coordinate directions, respectively. Multiplication of Eq.
(1) and Eq. (2) by the mass of a control volume provides the corresponding drag forces in the uand v-coordinate directions.
The loss coefficient for piles or piers of different dimensions within a grid cell is calculated using
the sum of all of the pile diameters. The following mathematical expressions represent the loss
coefficients in the x- and y-directions, respectively:

where n is the number of piers in the grid cell; Cd is the drag coefficient, which is approximately
equal to one for a cylinder in the tidal flow; D is the diameter of the bridge pier (m); x and y
are the grid distances (m) in the u- and v-directions, respectively; and the parameter a (= At /
Ae = At / (At nD)) is the ratio of the total cross-sectional area (At) to the effective crosssectional area Ae (= At nD).
Fig. 2 presents the configurations and approximate dimensions of the bridge piers of different
bridges. Table 1 summarizes the calculated loss coefficients corresponding to the dominant
movement direction of the tidal flows. Tsing Yi South Bridge and Kwai Tsing Bridge are located
directly next to one another. The bridge piers are considered as a single unit in calculating the
loss coefficient. The bridge piers of Stonecutters Bridge are not located in the channel. Effects
due to the bridge piers of this bridge are not considered in the hydrodynamic simulation of this
study.

A. Model Verification
Verification of the model in the present study was accomplished by comparing the model results
of tide levels, current speeds, directions, salinities and temperatures at locations inside and
outside of the channel with those generated by the wellverified regional model established by
Pun et al. [3]. The performance of the present model with a higher grid resolution in Rambler
Channel was comparable to that of the regional model.
In addition, the predicted tide levels were also compared to the tide data recorded at Kwai Chung
tide station. Figure 3 shows that the model prediction matches reasonably well with the field
data, but that the predicted high tide levels immediately after the preceding ebb tides are slightly
lower at spring tide. Tidal current measurements were conducted at a location adjacent to Tsing
Yi North Bridge in June 2012. A propeller current meter was used to measure tidal current speeds
at a depth of three meters below the water surface. The duration of measurement was
approximately two hours. As demonstrated in Figure 4, the model was able to accurately
reproduce the current magnitudes during the period of flow measurement. B. Hydrodynamic
Results Figures 5 and 6 present velocity vector plots of tidal flows for both flood and ebb tides.
During flood tide, tidal flows enter the channel from the southern entrance, where Stonecutters
Bridge is located, and leave the channel through the northern entrance near Ting Kau Bridge; the
tidal movement reverses during ebb tide. By analyzing the accumulated flow through the
channel, there is a net volumetric flux from north to south. Therefore, it can be concluded that a
higher volumetric flux associated with ebb tides dominates the tidal movement in the channel.
The figures also show comparisons between the velocity vectors for situations with and without
the effects of the bridge piers. The magnitudes of tidal current are generally higher in the absence
of bridges, particularly at the locations where the bridge piers are present. This phenomenon was
also observed at the inner corner of the channel. The bridge piers appear to alter the local flow
directions, as can be observed in the region near Tsing Yi North Bridge.
Drogue tracking was conducted as part of the hydrodynamic simulations to examine how the
bridge piers affect the path of the surface layer tidal flow. Figures 7 and 8 graphically present the
predicted flow paths of a drogue, which was released during flood tide in situations with and
without bridges. The recovery time was 29 hours after release of the drogue. The time interval
between two consecutive dots is one hour, as shown in the figures. The point of release of the
drogue is at the outlet of a storm water box culvert. For many years, the storm water discharged
from this box culvert contained organic and inorganic contaminants. The selection of this point
for release of the drogue provides an indication of whether such contaminants would be trapped
or carried out of the channel by tidal flows. For the situation in which all bridges are present, the
drogue circulated in the inner part of the channel and was eventually trapped in the channel. In

the absence of all bridges, the drogue moved to the southern entrance of the channel after 26
hours and was eventually carried out of the channel by the flows during ebb tide. The model
results clearly demonstrate that the bridge piers reduce the tidal flow speeds in the channel and
may limit the removal of contaminants through tidal exchanges. The model runs address the
bridge pier effects from all bridges together, as well as the effects of individual bridges. Figure 9
presents the percentage reduction in tidal flow rate of all examined situations. As shown in the
figure, the cumulative effects of Tsing Yi North Bridge, Tsing Lai Bridge, Cheung Tsing Bridge, Tsing
Yi South Bridge (and Kwai Tsing Bridge), and Ting Kau Bridge contribute to a 40.8% reduction in
tidal flow rate through the channel over a spring-neap cycle. Individually, Tsing Yi North Bridge
causes the highest reduction in tidal flow rate at 28.2%. The bridge is located at the narrowest
part of Rambler Channel with a width of approximately 284 m. The two bridge piers with a width
of 16 m each reduce the channel width by 11.3%; together with two structures (each 8 m wide x
28 m long) for the protection of bridge piers against ship collision, the results in a higher reduction
of the tidal flow rate.
Cheung Tsing Bridge alone generates a tidal flow percentage reduction of 19.3%. This situation is
similar to that of Tsing Yi North Bridge as the width of the channel section is only approximately
292 m. There are three bridge piers in the channel to support Cheung Tsing Bridge. The width of
each pier is approximately 10 m, leading to a channel width reduction of 10% at this location.
Tsing Yi South Bridge is supported by three circular piles, each with a diameter of 14 m. Kwai
Tsing Bridge is supported by three rectangular-shaped piers with dimensions of 16 m wide x 28
m long. Together, the piers of these two bridges contribute to a 16.1% reduction in tidal flow
rate. Tsing Lai Bridge is supported by three irregular hexagon-shaped bridge pier foundations in
the channel. There are another three bridge piers: one on the breakwater of the Rambler Channel
Typhoon Shelter, and two inside the typhoon shelter. Therefore, only three piers in the channel
are included in the model simulation. Different from the other bridges, the alignment of Tsing Lai
Bridge makes an angle of approximately 45 to the center line of the channel. The total
percentage reduction in the tidal flow rate of Tsing Lai Bridge is 11.3%. The reclaimed island to
support the central tower of Ting Kau Bridge is the largest single obstacle in the channel when
compared to the sizes of other bridge piers. This island alone, however, causes a decreased
reduction rate of 0.2%. This may be related to the location of the reclaimed island, which is
situated at the northern entrance of the channel with a width of 820 m.
The ratio of the width of the reclaimed island (~ 80 m) to the channel width is approximately 10%.
However, water depths in this region vary from approximately 10 m to 19 m, and are much
deeper than the other regions with bridges. For this situation, the presence of a physical barrier
in a large water body appears to not cause a significant reduction in tidal flow rate.
V. CONCLUSIONS
A hydrodynamic model was applied to study the influence of the bridge piers of six bridges on
the tidal exchange in
Rambler Channel. The energy losses due to the bridge pier resistance are estimated by
introducing additional quadratic friction terms in the x- and y-directions of the momentum
equations. The model prediction has shown that all the bridges combined contribute to a
significant reduction in tidal flow rate, limiting the tidal exchange in the channel. Amongst all the
bridges, Tsing Yi North Bridge is the greatest contributor to the reduction in tidal flow rate. The
tidal flow path at the surface water layer may also be modified by the bridge piers, causing

negative impact to the removal of contaminants from the channel by tidal flows. In the present
study, the sizes of model grid cells are much larger than the sizes of the actual bridge piers.
Quadratic friction terms are introduced into momentum equations of the model to produce a
reasonable estimate of the reduction in flow rate through the channel. Refinement of the model
grid cells to more comparable sizes may improve the accuracy of the model prediction. However,
a refined grid model must resolve the problems of complex model grid layout configuration and
the
increase in computational time for model simulation. Further study in this area is required to
address the computational problems.