You are on page 1of 22

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y.

Yu

Motion Sensors for Scour Monitoring:


Laboratory Experiments and Numerical Simulations
By J.-L. Briaud1, C. Yao2, C. Darby3, H. Sharma4, S. Hurlebaus5, G.R. Price6, K.-A. Chang7, B.E.
Hunt8, O.-Y. Yu9
Word count: 5166 (text) + 2250 (8 figures + 1 table) = 7416
Submission Date: November 4, 2009
Authors:
1. Jean-Louis Briaud (Corresponding Author)
Professor and Holder of the Buchanan Chair
Zachry Dpt. of Civil Engineering, Texas A&M University, College Station, Texas, 77843-3136
Phone: 1-979-845-3795
Fax: 1-979-845-6554
Email: briaud@tamu.edu
2. Congpu Yao
PhD Candidate, Research Assistant
Zachry Dpt. of Civil Engineering, Texas A&M University, College Station, Texas, 77843-3136
Phone: 1-979-739-2044
Fax: 1-979-845-6554
Email: congpu.yao@gmail.com
3. Colin Darby
PhD Candidate, Research Assistant
Zachry Dpt. of Civil Engineering, Texas A&M University, College Station, Texas, 77843-3136,
Phone: 1-281-743-1639
Fax: 1-979-845-6554
Email: c0d5579@neo.tamu.edu
4. Hrishikesh Sharma
PhD Candidate, Research Assistant
Zachry Dpt. of Civil Engineering, Texas A&M University, College Station, Texas, 77843-3136
Phone: 1-979-739-8265
Fax: 1-979-845-6554
Email: shrishikeshsharma@gmail.com
5. Stefan Hurlebaus
Assistant Professor
Zachry Dpt. of Civil Engineering, Texas A&M University, College Station, Texas, 77843-3136
Phone: 1-979-845-9570
Fax: 1-979-845-6554
Email: shurlebaus@civil.tamu.edu
6. Gerald R Price

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

ETI Instrument Systems, Inc.


1317 Webster Ave, Fort Collins, Colorado, 80524
Phone: 1-970-484-9393
Fax: 1-970-484-9397
Email: ETI@FRII.com
7. Kuang-An Chang
Associate Professor
Zachry Dpt. of Civil Engineering, Texas A&M University, College Station, Texas, 77843-3136
Phone: 1-979-845-4504
Fax: 979-862-8162
Email: kchang@civil.tamu.edu
8. Beatrice E. Hunt
Principal Hydraulic Engineer, P.E.
STV Incorporated, 225 Park Avenue South, New York, NY 10003-1604
Phone: 1-212-777-4400
Fax: 1-646-654-1861
Email: beatrice.hunt@stvinc.com
9. Ok-Youn Yu
Assistant Professor
Department of Technology, Appalachian State University, Boone, NC, 28608-2122
Phone: 1-828-262-2527
Fax: 1-828-265-8696
Email: yuo@appstate.edu

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

ABSTRACT
Bridge scour monitoring using fixed instrumentation is a good way for the owner to be warned of
imminent failure and take appropriate action before exposing the public to undue risk. This
paper shows an easy way to predict imminent failure of bridges by installing a motion sensor on
the bridge. The idea is that when the scour hole develops, the stiffness of the foundation
decreases and the natural frequency of the bridge-soil system also decreases. Two laboratory
experiments are described in the paper, and a relationship between the natural frequency of the
bridge and the scour depth is presented. Numerical models for the two experiments are
established and analyzed. Both the lab data analysis and numerical simulation show that the
natural frequency of the simulated bridge-foundation system changes with the depth of the scour
hole. Therefore motion sensors can be used for scour monitoring.

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

INTRODUCTION
With 58% of all bridge failures, bridge scour is the leading cause of bridge failures in the United
States (1). The Schoharie Creek Bridge in New York impacted the country when it collapsed on
April 5, 1987 and killed 10 people. Bridge scour monitoring using fixed instrumentation is a
good way for the owner to be warned of imminent failure and take appropriate action before
exposing the public to undue risk.
The Japan Railway Technical Research Institute (RTRI) published a study in 2008, which
provides a new method to judge the integrity of railway bridge columnImpact Vibration
Method (2). It is based on the fact that the foundation stiffness tends to decrease when a scour
hole forms around it. A lower stiffness of the foundation results in a decrease in natural
frequency of the column. Thus the integrity of the column can be judged by comparing the
natural frequency measured when it is known that the foundation is in good order (e.g.: water at
low level) with the natural frequency during a big flood.
Inspired by this idea, the authors are considering using a motion sensor, which is a
triaxial accelerometer, to monitor scour sensitive bridge columns. Such a sensor can give an
integrated response of the column regardless of where the scour hole is located; this is an
advantage over other instruments which give the depth of the scour hole at a chosen location; this
location is not always the deepest point in the scour hole. The objective of the experiment
described in this paper is to show that the natural frequency of the simulated bridge-foundation
system changes as the scour deepens.
LABORATORY TESTS
Two laboratory experiments, one with a shallow foundation and the other with a deep
foundation, were performed in a 2D flume which has a length, width and depth of 45 m (150 ft),
3.6 m(12 ft), and 3.3 m(11 ft), respectively, in the Haynes Coastal Engineering Laboratory at
Texas A&M University. In the bottom of the flume, there is a sediment pit which is 9 m (30 ft)
long, 3.6 m (12 ft) wide, and 1.5 m (5 ft) deep and is located about 2/3 of the way down the
flume from the flow inlet end.
Soil Properties
Before starting the laboratory experiments, fine clean and uniform silica sand was placed and
lightly compacted by hand tamping it in layers in the pit of the flume. The D50 of this fine sand
was 0.26 mm, the percent passing #200 sieve was 0.5% and the largest particles were less than
0.6 mm in diameter. The critical velocity as measured in the Erosion Function Apparatus was
0.2 m/s (3). A mold was used to obtain the unit weight of the sand which was found to be 18.9
kN/m3. A standard determination of the water content for the sand gave the value of 23.9%. In
order to get the Youngs Modulus of the sand, a Briaud Compaction Device (BCD) test was
performed (4). The BCD is a new instrument used to obtain a soil modulus in only a few
seconds; it consists of increasing the load on a rod equipped with a thin circular metal plate at the
bottom end and recording the bending of that plate under a standard load. The load is applied by
leaning on the rod. The BCD modulus of the sand averaged 12 MPa.
This soil was selected because it is very erodible and could generate a large scour hole in little

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

time during the experiment. Different soils would likely have different stiffnesses thereby
influencing the natural frequency of the system. However since the experiment was to only
prove the relative impact of the deepening of the scour hole, any erodible soil would have
worked. The soil properties were also used in the numerical simulations.
The Instruments
During the experiments, a motion sensor, a tilt sensor, a sonar sensor, a float-out device, a
tethered buried switch (TBS), a water stage and an acoustic Doppler velocimetry device (ADV)
were installed to monitor data. All instruments except the ADV were provided by ETI
Instrument Systems, Inc. The motion sensor measures the acceleration in three directions and
sends it through an RS 232 connection directly to a data logger. The tilt sensor measures the tilt
of the structural member to which it is attached. The sonar sensor operates under water and
measures the distance between the location of the head of the sensor and the soil surface it is
aimed at. The float-out device floats out when the scour hole reaches the depth where the float
out is located; when it floats out it gives a signal indicating that this scour depth has been
reached. The TBS is a float-out device which is hardwired to the data acquisition system. The
wire is a drawback in itself but also an advantage as it can be powered over time. The classical
float out is battery operated and therefore has a finite life. The TBS consists of a hollow
aluminum rod containing an electrical switch which triggers when the rod is horizontal or near
horizontal. The rod is not buoyant and instead relies on hydraulic drag to rotate it from vertical
to horizontal. The water stage sensor is fixed to the bridge and measures the distance from the
instrument to the water surface. The ADV measures the velocity of the water. All instruments
can be either hardwired or wireless except for the float out which is always wireless and the TBS
which is always hardwired.
Experiment with the Shallow Foundation
Experimental Set up and Instrumentation
In the first experiment, the concrete column, 0.45 m (1.5 ft) in diameter and 4 m (13 ft 1 in.)
long, was embedded to a depth of 0.3 m (1 ft) in the sand, then two prefabricated concrete decks
each 0.53 m (1.75 ft) wide, 2.03 m (6.75 ft) long, and 0.1 m (4 in.) thick were placed end-to-end
on top of the column to simulate a bridge with a shallow spread footing foundation. The
instruments are shown in Figures 1 (a) and (b). The motion sensor was mounted towards the top
of the column. The tilt sensor was fixed on one of the decks to indicate the inclination of the
deck in two directions: the flow direction and the traffic direction. The sonar sensor was
mounted on the column 0.75 m (2.5 ft) above the top of sand because its head needs to be under
water to work. The float-out device was buried just beneath the sand. The water stage sensor
was tied on the steel I-beam, which was not connected to the deck or column but was there to
provide a safety harness for the system. An ADV was installed in the flow ahead of the bridge to
monitor the water velocity at the height of 0.4 m (1.3 ft) above the flume bottom.
Experimental Procedure
The experiment lasted 6 hours and 45 minutes. First, and with no water in the flume, the deck
was struck with a 4.4 N (1 lb) rubber hammer. The flume was then filled until the water depth

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

reached 0.9 m (3 ft). A set of impact tests followed under hydrostatic condition. The series of
impact tests were repeated with a water velocity of 0.15 m/s (0.5 ft/s), 0.3 m/s (1 ft/s), and 0.45
m/s (1.5 ft/s). At 0.45 m/s (1.5 ft/s) the scour hole started to develop. When the water velocity
reached 0.6 m/s (2ft/s), the scour hole reached the foundation level, the foundation started to be
undermined, the column began to settle, and the tilt sensor indicated a change in deck inclination.
The final depth of the scour hole was about 0.42 m (1.4 ft) deep. It had a conical shape with a
top diameter of 1.91 m (6.35 ft). The depth of the hole was measured at four points and the
width of the hole in two directions:
(1) Trailing edge of column, 0.36 m (1.2 ft) deep;
(2) Right side of the column, 0.41 m (1.36 ft) deep;
(3) Left side of column, 0.41 m (1.36 ft) deep;
(4) Leading edge of column, 0.42 m (1.4 ft) deep.
Experiment with Deep Foundation
Experimental Set up and Instrumentation
In the second experiment, the column was reconstructed to form a pile foundation. The bottom
0.3 m (1 ft) of the concrete column was removed. This exposed eight #5 rebars over a length of
0.19 m (7.5 inch). Sleeves were placed over the rebars and grouted in place to increase the pile
diameter to 38 mm (1.5 in.) and the pile length to 0.3 m (1 ft).
The 0.45 m (1.5 ft) diameter and 4 m (13 ft 1 in.) long reconstructed column was
embedded to 0.45 m (1.5ft) in the sand with 0.15 m (0.5 ft) of column and 0.3 m (1 ft) of pile
foundation. The two prefabricated concrete decks were placed end-to-end on top of the column
to simulate a bridge with a deep foundation (Figure 1 (c)).
A TBS was used instead of the float-out device in this experiment. The TBS was buried
just beneath the sand and hardwired to the data acquisition system. The sonar sensor was
mounted on the column 0.85 m (2.8 ft) above the top of the sand. The motion sensor, water stage
sensor, tilt sensor and ADV were installed in the same location as in the shallow foundation
experiment. Figure 1 (c) shows the experimental set up including the instrument locations.
Experimental Procedure
The experiment lasted 4 hours and 20 minutes. First, and with no water in the flume, the deck
was struck with a 4.4 N (1 lb) rubber hammer. The flume was then filled until the water depth
reached 1 m (3.3 ft). A set of impact tests followed under hydrostatic condition. The series of
impact tests were repeated with a water velocity of 0.2 m/s (0.67 ft/s), 0.36 m/s (1.2 ft/s), 0.45
m/s (1.5 ft/s), and 0.6 m/s (2 ft/s). At 0.45 m/s (1.5 ft/s) water velocity the scour hole started to
develop. When the water velocity reached 0.6 m/s (2 ft/s), the depth of the scour hole reached
the bottom of the column; the column began to settle as the scour hole continued to deepen. At
0.8 m/s (2.6 ft/s), the tilt sensor indicated a change in deck inclination, and the column leaned at
about 30 degree off the vertical direction (failure). The final depth of the scour hole was about
0.32 m (1.08 ft) and had a conical shape with a top diameter equal to 1.47 m (4.9 ft).

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

(a)
Side view of system and sensors set-up illustration
bridge decks
hammer
impact

tilt sensor
motion sensor
bridge
pier

0.45m
(1.5ft)

3.3m
(11 ft)

water stage
float out

0.9m
(3 ft)
0.3m
(1 ft)
1.2m
(4 ft)

sonar

water

scour
hole

ADV
1.5m
(5 ft)

0.26 mm
fine sand

3.6m (12 ft)

(b)

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

Side view of system and sensors set-up illustration


bridge decks
hammer
impact

tilt sensor
motion sensor
0.45m
(1.5ft)
1m
(3.3 ft)
0.15m
concrete+0.3m
pile (0.5ft
concrete+1ft
pile)

1.05m
(3.5 ft)

bridge
pier

3.3m
(11 ft)

water stage
Tethered
Buried Switch
sonar

water

scour
hole

ADV
v

1.5m
(5 ft)

0.26 mm
fine sand

3.6m (12 ft)

(c)
FIGURE 1 Experimental set-up illustration: (a) photo of the experiment, (b) illustration of
sensors in shallow foundation experiment, and (c) illustration of sensors in deep foundation
experiment.
Data Analysis
Shallow Foundation
Figure 2 (a) shows the acceleration of the column in three directions at the location of the
instrument as a function of time. The impact tests can be seen clearly as groups of peaks. At 4.5
hr after the start, the acceleration signal in the flow direction clearly indicates that significant
movement of the column occurred. This corresponds to the time when the scour hole reached
the bottom of the embedded portion of the column; beyond that time the column settled as the
scour hole continued to deepen. Note that the acceleration in the other two directions does not
show any significant change.
To study the signal in the frequency domain in detail, the acceleration trace was broken
into small time intervals. Eleven time intervals were selected to correspond to the following
impact tests: 0-0.25 hr (dry impact test), 0.25-2 hr (fill the tank, no test), 2-2.25 hr (wet impact
test, no flow), 2.25-2.4 hr (v = 0.15 m/s (0.5 ft/s), no impact test), 2.4-2.7 hr (v = 0.15 m/s (0.5
ft/s), impact test), 2.7-2.87 hr (v = 0.3 m/s (1 ft/s), no impact test), 2.87-3.1 hr (v = 0.3 m/s (1
ft/s), impact test), 3.1-3.2 hr (v = 0.45 m/s (1.5 ft/s), no impact test), 3.2-3.6 hr (v = 0.45 m/s
(1.5 ft/s), impact test), 3.6-4.5 hr(v = 0.45 m/s (1.5 ft/s), no impact test), 4.5-6.75 hr (v = 0.6 m/s
(2 ft/s), no impact test). The Fast Fourier Transform (FFT) of the acceleration vs. time plot for
each time interval can be generated separately for the X, Y, and Z axes. Figure 2 (b) shows as an

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

example the sequence of FFT plots corresponding to the time intervals mentioned above and for
the flow direction. From the FFT plots, we see that the peak frequencies in the flow direction are
about 10 Hz, 35 Hz, and 40 Hz. These frequencies changed during the last two periods, i.e. t =
3.6- 4.5 hr, no impact test; and t = 4.5hr to 6.75 hr, no impact test.

Acceleration (g)

Acceleration on X,Y,Z Axis-Shallow Foundation


0.08

Acceleration on X axis(flow direction)

0.04
0
-0.04
-0.08

Acceleration (g)

1.5

2.5

4
3.5
Time (hr)

0.08

4.5

5.5

6.5

Acceleration on Y axis(traffic direction)

0.04
0
-0.04
-0.08
0

Acceleration (g)

0.5

0.5

1.5

2.5

4
3.5
Time (hr)

0.08

4.5

5.5

6.5

Acceleration on Z axis(vertical direction)

0.04
0
-0.04
-0.08
0

0.5

1.5

2.5

4
3.5
Time (hr)

(a)

4.5

5.5

6.5

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

10

FFT on X( flow direction)-Shallow


Foundation
4
10000

3
FFT Magnitude

FFT Magnitude

X-0~0.25hr,dry WAK

5000

20
40
Frequency (Hz)

10000
FFT Magnitude

FFT Magnitude

5000

20
40
Frequency (Hz)

20
40
Frequency (Hz)

60

5000

20
40
Frequency (Hz)

60

10000
FFT Magnitude

X-2.4~2.7hr,v=0.15m/s,WAK

FFT Magnitude

X-2.25~2.4hr,v=0.15m/s,no test

60

10000

5000

10000
X-2~2.25hr,v=0, WAK

X-0.25~2hr,filling the tank,no test


2

60

x 10

20
40
Frequency (Hz)

X-2.7~2.87hr,v=0.3m/s,no test

5000

60

20
40
Frequency (Hz)

60

10000
X-2.87~3.1hr,v=0.3m/s,WAK

1.5
1
0.5
0
2

0
4
x 10

20
40
Frequency (Hz)

1
0.5

FFT Magnitude

x 10

0
5

1.5

5000

60

X-3.2~3.6hr,v=0.45m/s,WAK

X-3.1~3.2hr,v=0.45m/s, no test

FFT Magnitude

FFT Magnitude

x 10

FFT Magnitude

FFT Magnitude

20
40
Frequency (Hz)

60

0
4
x 10

20
40
Frequency (Hz)

60

X-3.6~4.5hr,v=0.45m/s,no test

4
3
2
1
0

20
40
Frequency (Hz)

60

X-4.5~6.75hr,v=0.6m/s,no test
3
2
1
0

20
40
Frequency (Hz)

60

(b)
FIGURE 2 Data analysis on shallow foundation experiment: (a) time domain plots and (b)
FFT in the flow direction (X axis).
Deep Foundation

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

11

Figure 3 (a) shows the acceleration of the column in three directions at the location of the
instrument as a function of time. The impact tests can be seen clearly as groups of peaks. At
about 3 hr after the start, the scour hole started to develop. At 4 hr after the start, the acceleration
signal in the flow direction indicates a slight movement of the column. This corresponds to the
time when the scour hole reached the bottom of the column; beyond that time the column settled
as the scour hole continued to deepen.
Figure 3 (b) shows the FFT for the whole 4 hr and 20 minutes of data; a clear peak in the
flow direction is observed. In order to study the signal in the frequency domain in more detail,
the acceleration trace was broken into twelve small time intervals: 0.08-0.14 hr (dry impact test),
0.25-1.6 hr (fill the tank, no test), 1.8-1.85 hr (wet impact test, no flow), 2h-2.5 hr (v=0, no test),
2.55-2.65 hr (v=0.2 m/s (0.67 ft/s), impact test), 2.8h-2.9 hr (v=0.2 m/s (0.67 ft/s), no test), 3-3.1
hr (v=0.36 m/s (1.2 ft/s), impact test), 3.3h-3.5 hr (v=0.36 m/s (1.2 ft/s), no test), 3.6-3.7 hr
(v=0.45 m/s (1.5 ft/s), impact test), 3.75-3.9 hr (v=0.45 m/s (1.5 ft/s), no test), 3.95-4.05 hr
(v=0.6 m/s (2 ft/s), impact test), 4.15-4.3 hr (v=0.8 m/s (2.6 ft/s), no test). The FFT plots for
each time increment and for the flow direction are shown in Figure 3 (c). The peak for the
impact tests on the left side of Figure 3 (c) is very clear and remains consistently clear. The
reason is that the plots on the left side of Figure 3 (c) correspond to impact tests in the same
direction as the measured acceleration, while the plots on the right side of Figure 3 (c)
correspond to no impact tests.

Acceleration (g)

Acceleration on X,Y,Z Axis-Deep Foundation

0.02
0.01
0
-0.01
-0.02

Acceleration (g)

Acceleration on X(flow direction)

0.03

1.25 1.5 1.75

2 2.25 2.5 2.75


Time (hr)

0.08
0.06
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08

3.25 3.5 3.75

4.25 4.5

Acceleration on Y(traffic direction)

Acceleration (g)

0.25 0.5 0.75

0.25 0.5 0.75

1.25 1.5 1.75

2 2.25 2.5 2.75


Time (hr)

0.08
0.06
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08

3.25 3.5 3.75

4.25 4.5

Acceleration on Z(vertical direction)

0.25 0.5 0.75

1.25 1.5 1.75

2 2.25 2.5 2.75


Time (hr)

(a)

3.25 3.5 3.75

4.25 4.5

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

FFT Magnitude

FFT on X(flow direction)

4
3
2
1
0
5

FFT Magnitude

FFT on X,Y,Z Axis-Deep Foundation

x 10

0
4
x 10

10

20

30
40
Frequency (Hz)

60

70

FFT on Y(traffic direction

4
3
2
1
0

10

20

30
40
Frequency (Hz)

10
FFT Magnitude

50

x 10

50

60

70

FFT on Z(vertical direction)

10

20

30
40
Frequency (Hz)

50

60

70

(b)
FFT on X(flow direction)-Deep Foundation
4

10
X-0.08~0.14hr,Dry WAK

FFT Magnitude

FFT Magnitude

10000

5000

20
40
Frequency (Hz)

5000

20
40
Frequency (Hz)

20
40
Frequency (Hz)

60

5000

20
40
Frequency (Hz)

60

10000
X-2.55~2.65hr,WAK,v=0.2m/s

5000

20
40
Frequency (Hz)

60

FFT Magnitude

FFT Magnitude

X-2~2.5hr,No test,v=0

60

10000

10000
X-1.8~1.85hr,WAK,v=0

FFT Magnitude

FFT Magnitude

10000

X-0.25~1.6hr,No test,filling the tank

60

x 10

X-2.8~2.9hr,No test,v=0.2m/s

5000

20
40
Frequency (Hz)

60

12

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

10000
X-3~3.1hr,WAK,v=0.36m/s

5000

20
40
Frequency (Hz)

FFT Magnitude

FFT Magnitude

10000

20
40
Frequency (Hz)

FFT Magnitude

FFT Magnitude

5000

20
40
Frequency (Hz)

60

X-3.75~3.9hr,No test,v=0.45m/s

5000

60

20
40
Frequency (Hz)

60

10000
X-3.95~4.05hr,WAK,v=0.6m/s

5000

20
40
Frequency (Hz)

60

FFT Magnitude

10000
FFT Magnitude

5000

10000
X-3.6~3.7hr,WAK,v=0.45m/s

X-3.3~3.5hr,No test,v=0.36m/s

60

10000

13

X-4.15~4.3hr,No test,v=0.8m/s

5000

20
40
Frequency (Hz)

60

(c)
FIGURE 3 Data analysis on deep foundation experiment: (a) time domain plots, (b) FFT in
the whole time period, and (c) FFT in the flow direction (X axis).
Discussion
Shallow Foundation
Frequency vs. Time Plots Using the frequencies identified by the FFT plots for various time
periods and plotting these frequencies as a function of time gives Figure 4 (a). On each graph,
there are three lines referring to the evolution of the first, second, and third frequencies of the
system as the experiment progressed. This plot shows that the frequencies started to decrease
around 3.6 hours after the start. This corresponds to the time when the scour hole started to
develop. At 4.5 hours after the start, the scour hole reached the bottom of the foundation and the
column started to settle. Therefore the frequency domain analysis allows the sensor to detect the
start of the scour hole and its progression, not just when the foundation starts to move. This is
because the stiffness of the foundation decreases as the scour hole develops and so does its
natural frequency.
Other Data Figure 4 (b) shows the relationship between the first observed frequency of the
system, the tilt angle, and the scour depth. The figure indicates that the tilt sensor reported
movement at 4.5 hours after the start when the scour hole became deep enough and the column
started to settle. By comparison, the frequency vs. time plots gives earlier detection (3.5 hours
after the start) as seen in Figure 4 (a). This is when the scour hole started to develop as shown by
the sonar measurements. Therefore the tilt sensor shows when the column moves but the motion

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

14

sensor shows earlier detection since it indicates when the scour hole starts. Note that the sonar
data shows an increase in scour depth at first followed by a decrease in scour depth. The scour
depth is not actually decreasing but the column is settling into the sand; because the sonar is tied
to the column, the true scour depth is not measured. Note also that the sonar operated at its limit
of depth detection in the small scale lab experiment. Such limitations are rarely the case for field
scale applications.
Natural frequency on X(flow direction)-Shallow Foundation
50
45

1st frequency
2nd frequency
3rd frequency

Frequency(Hz)

40
35
30
25
20
15
10
5
0

0.5

1.5

2.5

3.5
Time(hr)

4.5

5.5

6.5

Natural frequency on Y(traffic direction)-Shallow Foundation


50
1st frequency
2nd frequency
3rd frequency

Frequency(Hz)

45
40
35
30
25
20

0.5

1.5

2.5

3.5
Time(hr)

(a)

4.5

5.5

6.5

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

15

1st Natural frequency in X,Y,Z axis-Shallow Foundation

Frequency(Hz)

50
flow direction
traffic direction
vertical direction

40
30
20
10
0

2
3
4
5
TiltX(traffic direction)-Shallow Foundation

TiltX(degree)

6
TiltX

5.5
5
4.5
4
3.5
3

0.5

1.5

2.5
3.5
4.5
Scour depth-Shallow Foundation

5.5

6.5

Scour depth (m)

2
Scour depth
1.5
1
0.5
0

Time (hr)

(b)
FIGURE 4 Discussion on the plots for shallow foundation experiment: (a) time-frequency
plots in three directions, and (b) plots of tilt sensor, sonar sensor and motion sensor.
Deep Foundation
Frequency vs. Time Plots Using the frequencies identified by the FFT plots for various time
periods (Figure 3 (c)) and plotting these frequencies as a function of time gives Figure 5 (a). In
the traffic direction 3 frequencies were identified but only one was found in the flow direction.
Figure 5 (a) indicates that the frequencies remain steady during the experiment up until the
column settled and rotated. Unfortunately because failure was quite sudden, no data could be
obtained during the later part of the test.
Other Data Figure 5 (b) shows the relationship between the first observed frequency of the
system, the tilt angle, and the scour depth. The figure indicates that the tilt sensor recorded tilt of
the deck at 3.6 hours after the start when the scour hole became deep enough and the column
started to settle and rotate. Again the sonar data shows results which are erratic towards the end
of the test; this is because the sonar operated at its limit of depth detection in the small scale lab
experiment. Such limitations are rarely the case for field scale applications.

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

Natural frequency on X(flow direction)-Deep Foundation


15

Frequency(Hz)

1st frequency

10

0.5

1.5

2.5
Time(hr)

3.5

4.5

Natural frequency on Y(traffic direction)-Deep Foundation


50
1st frequency
2nd frequency
3rd frequency

Frequency(Hz)

40

30

20

10

0.5

1.5

2.5
Time(hr)

3.5

4.5

(a)
1st Natural frequency in X,Y axis-Deep Foundation
Frequency(Hz)

40
flow direction
traffic direction

30
20
10

TiltX(degree)

0.5

1.5
2
2.5
3
TiltX(traffic direction)-Deep Foundation

3.5

4.5

TiltX

4
2
0
0

0.5

1.5
2
2.5
3
Scour depth-Deep Foundation

3.5

4.5

Scour depth (m)

5
Scour depth

4
3
2
1
0

0.5

1.5

2
2.5
Time (hr)

3.5

4.5

(b)
FIGURE 5 Discussion on the plots for deep foundation experiment: (a) time-frequency
plots in three directions, and (b) plots of tilt sensor, sonar sensor and motion sensor.

16

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

17

NUMERICAL SIMULATION
The two scour experiments were simulated using the finite element method and a parametric
study was undertaken. An Eigenvalue analysis was carried out to identify the predominant mode
shapes and frequencies and their variation as the scour hole deepened. The results are compared
to the experimental data.
Finite Element Model
In this study, the commercial finite element program LS-DYNA was used for the analysis (5).
The finite element models (FEMs) considered in this study were generated by HyperMesh (6).
The mesh is shown in Figure 6 and matched the dimensions of the experiment. The material
properties were obtained by a combination of field testing and manufacturer specifications. The
material properties used were as follows:
(1) Concrete :-Modulus of Elasticity(E)=24.6 GPa, =2400 kg/m3, =0.2,
(2) Rebar :-Modulus of Elasticity(E)=200 GPa, =7850 kg/m3, =0.3
(3) Deck Support :-Modulus of Elasticity(E)=200 GPa, =7850 kg/m3, =0.3
(4) Soil
:-Modulus of Elasticity(E)=0.012 GPa, =1928 kg/m3, =0.35
The soil and the concrete were considered elastic. A strain rate dependent elasto-plastic
model was used as the material model for the rebars which were represented explicitly using one
dimensional element. The contact between the concrete and reinforcements made use of the
Lagrangian coupling method. The column, deck, supports and soil were modeled by a fully
integrated quadratic eight node element with nodal rotations. Mesh refinement was done to
achieve convergence. The eigenvalue analysis was performed using the lanczos solver.
The penalty method was used to model the contacts between the different elements. In
this method normal interface springs are placed between all penetrating nodes and the contact
surfaces. The method is stable and it does not excite mesh hourglassing. It is capable of
handling contacts between dissimilar materials, which is well suited for this study.
Shallow Foundation
In the shallow foundation model, the column was embedded in the soil for 0.3 m (1 ft). The
reinforced concrete deck rested on the column and on the rail supports. The deck supports were
modeled as rigid elements with fixed end condition at the base. No constraint was applied to the
deck which was free to displace. The boundary condition for the soil block was fixed on the four
faces and at the base. This was done to simulate the conditions in the laboratory where the soil
was surrounded by concrete walls. In this study, the water was not included except that the soil
unit weight was the saturated unit weight. The presence of the scour hole was simulated by
changing the contour of the mesh along the soil surface (Figure 6). The scour depth is varied in
increments of one-third of the total embedment of the column which was 0.3 m (1 ft). So, four
conditions are simulated:
(1) No scour, column embedded 0.3 m (1 ft) in the soil (Figure 6)
(2) Scour, column embedded 0.2 m (0.66 ft)
(3) Scour, column embedded 0.1 m (0.33 ft)
(4) Scour, column embedded 0 m (0 ft) (figure 6) (the system started to settle at that
point)

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

18

FIGURE 6 FEMs for shallow foundation without scour (left); fully developed scour
(right).
Deep Foundation
In the deep foundation model, the column was embedded 0.15 m (0.5 ft) in the soil. The pile
dimensions in the model matched the dimensions in the experiment (0.3 m or 1 ft length).
The other details of the modeling process were the same as for the shallow foundation
case. The scour depth was varied in increments of one-third of the total embedment of the
column and the pile which is 0.45 m (1.5 ft). So, four conditions were simulated:
(1) No scour, column embedded 0.15 m (0.5 ft) in soil, pile embedded length 0.3 m (1 ft)
(2) Bottom of scour hole at the bottom of the column, pile embedded length 0.3 m (1 ft)
(3) Bottom of scour hole halfway through the pile embedment length, (failure of the
system is observed near this point 0.32 m (1.08 ft) in the experiment)
(4) Bottom of scour hole at the bottom of the piles.
Mode Shape and Frequency Comparison
The eigenvalue analysis gives the frequencies and mode shapes of the entire system. The total
response of the system is the combination of these modes. The observed mode shapes and
frequencies are reported in Table 1 for both shallow and deep foundation.
Shallow Foundation
The lowest frequency obtained is the bending of the column (3 Hz). The soil provides restraint
at the bottom of the column. Due to its low stiffness, the end condition is between free (2.63 Hz)
and fixed (12.3 Hz) condition. The next higher modes are torsion and compression in the
vertical direction (6.6 Hz and 8.7 Hz). After that, the translational modes of the soil in the flow
and traffic directions (10.2 Hz and 11.1 Hz) appear. Both frequencies are very close, due to the

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

19

fact that the deck is not applying a significant restraint in any of the two directions. The bending
of the deck shows up near 20 Hz combined with translation of the soil. This behavior is contrary
to the behavior of normal bridges. In normal bridges, the deck is very flexible as compared to
the column, so the fundamental mode is normally the bending of the deck. In the shallow
foundation laboratory experiment, the deck is very stiff as compared to the long column; hence it
has a higher frequency than the column.
The frequencies observed in the experiment can be compared to the frequencies
calculated in the numerical simulation. The first observed frequency is 8.9 Hz; this frequency is
in the flow direction and is close to the first translational mode of the soil in the flow direction,
which corresponds to frequency of 10.2 Hz. This mode consists of an interaction between the
soil moving horizontally and the column being entrained by the soil around it. The bending
mode of the column (3 Hz) was not observed in the experiment. The bending of the deck
combined with the second mode of translation of soil appears as the first frequency in the traffic
direction. The simulated value of 20.2 Hz is very close to the experimental value of 21.59 Hz.
Deep Foundation
The deep foundation model follows the same trends as the shallow foundation model. The first
frequency drops to 0.6 Hz and corresponds to the bending mode of the piles. This value is
smaller than the value for the shallow foundation because the piles are more flexible than the
column. The first translational frequency of the soil is 9 Hz which drops from 10.2 Hz for the
shallow foundation due to a decrease in contact area leading to a decrease in effective stiffness.
This also affects the other higher modes.
Again the experimentally observed value of 8.89 Hz in the flow direction is very close to
the simulated value of 9 Hz in the flow direction for translational mode of the soil. The first
observed frequency in the traffic direction is 19.6 Hz. It is comparable to the simulated value of
22.39 Hz which again corresponds to the second translational mode of soil combined with the
bending of the deck.
TABLE 1 Identified Mode Shapes from FEM
Shallow Foundation

Bending(Column)

Flow
Direction (Hz)
3

Translation(Soil)

10.2

Bending(Soil)
Translation(Soil)

Mode

Bending(Column)

Traffic
Direction(Hz)
3.26

Torsion

Vertical
Direction (Hz)
6.6

Translation(Soil)

11.1

Compression

8.7

14.6

Bending(Soil)

15.3

Compression

16.2

22.1

Translation(Soil)

20.2

Compression

19.1

Mode

Mode

Pile Foundation
Bending(Piles)

0.6

Bending(Piles)

0.6

Torsion

1.2

Translation(Soil)

Translation(Soil)

8.3

Compression

7.7

Bending(Soil)

14

Bending(Soil)

13

Compression

12.2

Translation(Soil)

22

Translation(Soil)

19.6

Compression

17

Frequency Variation with Scour Hole

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

20

The frequency of the system depends on the boundary conditions, the material properties and the
geometric properties. The material and geometric properties remains the same during the
formation of the scour hole. The boundary conditions change due to the formation of the scour
hole. Initially, the column has the largest embedment in the soil which provides restraint to the
system. This condition is between a free and a fixed condition. As the scour progresses, the soil
is eroded, the effective embedment decreases and so does the restraint. This results in a shift in
the boundary condition closer to a free condition (lower frequency) due to a decrease in the
effective horizontal stiffness provided to the column by the soil. Figure 7 shows the evolution of
the first observed frequency in the flow direction and the traffic direction as the scour depth
increases for both the shallow and the deep foundation. This trend compares well with the
measurements as shown in Figure 7.

FIGURE 7 Variation of frequencies with depth of scour for shallow foundation (left), and
for deep foundation (right).
FIELD IMPLEMENTATION
The bridge on US Highway 59 southbound at the Guadalupe River, south of Victoria, was
selected for implementation. It was equipped with one wireless motion sensor, one hardwired
motion sensor, one tilt sensor, one water stage sensor, two float-out devices, two TBS for scour
monitoring. This bridge has been subjected to changing river conditions such that the Guadalupe
now approaches the bridge with an approximate angle of attack of 60 degrees. This created
increased foundation scour due to the greatly increased attack angle. The bridge consists of two
overwater bents each spanned by five steel beams overlaid with a 0.3 m (1 ft) concrete/asphalt
roadbed. Both columns are webwalls between two cylindrical columns with a diameter which
increases with depth. The foundation consists of battered piles to a depth of approximately 11 m
(35 ft) below the pile cap. The bridge is subject to debris flows, and the northern of the two
overwater bents has a large debris buildup immediately upstream of the column.
Prior to installation, we visited the site in order to select the best instrument package for
the bridge. Because of debris, instruments such as sonar or sliding collars were eliminated in
favor of a package consisting of two batteries of below-soil instruments, one immediately below

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

21

the south bent and another in the south bank, which was deemed at-risk for slope collapse, and a
battery of above-waterline instruments. The above-waterline battery consisted of one wired and
one wireless motion sensor collecting data at 80 Hz, one two-dimensional tiltmeter and one stage
sensor, the latter two collecting at 1 Hz. Figure 8 shows the instruments locations.

SB 0

SB 1 (debris)

SB 2

SB 3

LEGEND
Tilt Sensor

Motion Sensor
Master Station
Float-out

Water Stage Sensor


TBS

Solar Array

FIGURE 8 Schematic of instrument placement for US 59.


Initial installation of the buried instruments revealed a scour hole about 2.5 m (8 ft) deep
south of the south bent. At the bottom of the scour hole was a gravel layer approximately 1.5 m
(5 ft) deep. This gravel layer created some installation issues, including near-loss of one of the
float-outs. Installation of the above-water instruments went smoothly, in comparison. A
Campbell Scientific CR1000 data logger, which was hardwired into all of the instruments except
for one battery of buried devices and a remote motion sensor located at the debris column,
collected data on-site. The data was relayed by cellular modem to Texas A&M University, where
the data was processed as for the lab tests. The data logger transmits during one minute of every
hour. The wireless sensor transmitted in synch with the others, leading to confidence that the
system can be fully wireless if required.
CONCLUSION
This contribution is based on the fact that the foundation stiffness tends to decrease when a scour
hole forms around it. A lower stiffness of the foundation results in a decrease in natural
frequency of the column. Two laboratory experiments of a simulated bridge are presented: one
with a shallow foundation and the other with a deep foundation. The bridge is instrumented with
a motion sensor. The measured data as well as the numerical simulation results indicate that
natural frequencies of the bridge system are affected by the development of a scour hole around
the embedded portion of the pier. While further work is necessary, early results are encouraging

J.-L. Briaud, C. Yao, C. Darby, H. Sharma, S. Hurlebaus, G.R. Price, K.-A. Chang, B.E. Hunt, O.-Y. Yu

22

for the use of motion sensors as scour monitors. Field scale experiments are on-going.
ACKNOWLEDGEMENTS
This project is sponsored by The Texas Department of Transportation. Marcus Galvan and John
Delphia at TxDOT have been particularly helpful. The views expressed in this paper are those of
the authors and not necessarily those of TxDOT. We are also indebted to John Reed at Texas
A&M University for his help during the laboratory experiments.
REFERENCES
1. Hunt, B. E. Monitoring of Scour Critical Bridges. NCHRP Synthesis of Highway Practice
396, Transportation Research Board, Washington, D.C. 2009.
2. Shinoda, M., H. Haya, S. Murata. Nondestructive Evaluation of Railway Bridge
Substructures by Percussion Test. Fourth International Conference on Scour and Erosion.
CD-ROM. Tokyo, Japan. 2008.
3. Briaud, J.L., F.C.K. Ting, H.C. Chen, Y. Cao, S.W. Han, K.W. Kwak. Erosion Function
Apparatus for Scour Rate Predictions. Journal of Geotechnical and Geoenvironmental
Engineering, ASCE, Vol. 127, No. 2, 2001, pp. 105-113.
4. Briaud, J.-L., Y. Li, K. Rhee. BCD: A Soil Modulus Device for Compaction Control. Journal
of Geotechnical and Geoenvironmental Engineering, ASCE, Vol. 132, No. 1, January, 2006,
pp. 108-115.
5. LSDYNA Keyword Users Manual, Version 970, Livermore Software Technology
Corporation, Livermore, CA, 2003.
6. HyperMesh Ver. 6.0 Basic Tutorial, Altair Computing, Altair Engineering Inc. 1820 E. Big
Beaver, Troy, MI 48083, 2003.

You might also like