SSP set
Implement MEMS Accelerometers
for slope stability monitoring
Po Shan Road Landslide, Hong Kong 1972
Commercial MEMS - Analog Devices ADXL Used in the project:
ADXL335
ADXL345
This project demonstrates MEMS
performance through:
Real-time field monitoring site
Laboratory flume test of landslide event
Description
ADXL345 MEMS accelerometer is packaged in a sensing
module named 1st Generation Field Smart Soil Particle (SSP) -
800 Hz sampling rate
3 units of SSP were installed in Lu Shan, Taiwan and currently
available for real-time monitoring.
Data are transferred from Taiwan to database in Hong Kong
through 3G mobile service.
Results of the Field Monitoring
Several events affecting slope movement occurred within this
period of study – there has been no failure.
Data shown reflects the real occurrence of the earthquakes
(dated 16/07/2013).
Earthquakes subject slopes to horizontal and vertical
accelerations that result in cyclic fluctuations in stresses within the slope, potentially causing landslide to occur.
Note that both of the graphs are not plotted in the same
coordinate system and they are located differently.
MEMS noise is larger.
The most important point of this result is that MEMS captured
the short-lived earthquakes, thus could be reliably used to monitor the slope.
Comparison of data: Taiwan Government Central Weather Bureau (left) and SSP (right)
SSP on-site
Landslides
Count for 17% fatalities due to natural
hazards
300 events in Hong Kong every year
Real-time monitoring is crucial
But the instruments used currently are
expensive and inflexible
MEMS Sensors - Accelerometer
Reliable sensors (used in phones)
Miniature in size
Cheap – US$10 a piece
Triaxial – measures acceleration in
three directions (labelled X, Y and Z)
Not popular in geotechnical engineering
Location of the field monitoring site: Lu Shan, Taiwan. Three SSPs were installed:
SSP001, SSP003 and SSP004 In-place inclinometers used for slope
Sequence of the Flume Test
After all the preparation, the soil sample is saturated via
the saturation (white) box. Physically we can see the sample consolidating (see M5 data).
Total 2 L/min water is pumped into the flume.
Question: How do the MEMS sensors perform in reflecting
the physical events?
M2 sensor is used in the discussion below.
Results and Discussions (see Figure below)
At 1330 seconds onwards (MEMS time, Part 1), the slope
is more or less in static equilibrium. Correspondingly, the
M2 the resultant plot is almost 1 g.
Sudden failure occurred at 1500 seconds. The data of all
three axes showed sudden jumps in the reading,
corresponding to the sudden mass movement. Again, the MEMS data reflected the physical phenomena happening during the experiment.
The slope then started to move very slowly as shown in
Part 2. This is reflected in the decreasing trend in all axes.
Slightly after 1730 seconds, the slope failed once again.
It was not until 1800 seconds that M2 started to be washed
away by the landslide. (see the JVC Top pictures)
The data recorded during this time (Part 3) shows more
dynamic movements through the fluctuations. This clearly reflects the physical phenomena of the washing of M2 by the landslide.
As M2 was deposited at its final destination, the resultant
data was close to 1 g, indicating once again static
equilibrium.
The laboratory flume test has shown how reliable the
data from MEMS are. They reflect the physical events.
Set-up of the flume laboratory test.
The flume is inclined at 30⁰. Water will be supplied from the bottom of the flume to imitate groundwater inflow. Two cameras capture the video of the whole
experiment. Picture and plot of M5 MEMS sensor during consolidation Set-up of the MEMS
sensors in the flume. There are 10 ADXL335
MEMS accelerometers named M1 to M10. X Y Z Resultant Wedge 2 cm
CONCLUSIONS
The field monitoring has shown real movement that potentially
destabilizes slope can be captured by MEMS accelerometers even for a short period of time.
The laboratory flume tests show how each phenomenon that
happened physically can be captured and interpreted back.
Therefore, MEMS accelerometers are reliable in capturing
real-time physical movements, and able to improve landslide early warning system.
IMPACTS
With this monitoring system, coupled with early warning systems, human lives in landslide-prone areas can be better protected against landslide hazards. Due to its flexibility, this technology can also be applied to:
Other natural disaster prevention like tsunami and volcanic
activities monitoring,
Structural health monitoring system, and
Field monitoring of geotechnical construction work such as deep
excavation and tunnelling.
Pictures of the flume (top: JVC Top, bottom: JVC Side) at 1330 s (left), 1500 s (middle) and 2010 s (right) MEMS time and M2 data for discussion