Documente Academic
Documente Profesional
Documente Cultură
Integrated Coursework
Location: Soils (shaking table) laboratory
Inglis Building Ground Floor (ISG-64)
Experiment A3
(Short)
Soil Mechanics
Objectives
Introduction
This lab is part of the Integrated Coursework looking at buildings in earthquakes. This
experiment examines the influence of the ground underneath the structure. Structures are
primarily designed to carry vertical loads imposed by people, objects and their own self-weight.
Typically, each floor slab transfers the load to beams which in turn transfer the load to columns
at every floor level. All the vertical loads are then transferred by the columns to the foundations.
The job of the foundations is to transfer these vertical loads safely to the soil below.
Earthquakes, however, impose additional horizontal loads onto the structural elements. It is
therefore necessary to understand the behaviour of a structure under the action of these
horizontal loads.
There are many examples of structures that failed to
perform this task adequately. Figure 1 shows a collapsed
multi-storey building in Ahmadabad city following the
2001 Bhuj earthquake in India. The shaking amplitude
was low in Ahmadabad, which is some 400 km from the
epicentre, but it was the soil conditions at this building
site that contributed to the collapse. There are many other
instances in recent earthquakes where the ground
conditions caused failure of civil engineering structures,
such as buildings, bridges, embankments, dams etc. In
this lab we will concentrate on the response of the
buildings and their foundations to earthquake loading.
Although soil is made up of solid particles and voids we can treat it, at a macroscopic level, as a
continuum, and characterise its behaviour by means of bulk properties. For example, we can
weigh a volume of sand and find out its average bulk unit weight. Similarly we can define the
amount of void space in a soil sample as the void ratio e, equal to the ratio of the volume of
voids to the volume of solid particles. Even the bulk stiffness of the soil can be measured as the
settlement it suffers when subjected to an axial load (similar to the Youngs modulus for a solid
material). Using these we can estimate the behaviour of the soil around the foundation.
Apparatus
This experiment consists of two parts. Everyone will do both parts, with a change about halfway
through. In the first part, you will shake the same structure used throughout the Integrated
Coursework and in IA Experiment 7, except this time it is on a box of soil. This part of the
experiment looks at the dynamic behaviour of soil. The second part comprises an oedometer test
to measure the static stiffness of the soil, which can then be related to the dynamic properties.
Shaking Table
In the other IA/IB experiments on this
structure, it was clamped to a rigid base.
Here the structure is placed loosely on a
300mm deep sand bed which is then shaken
with a shaking table (Figure 2). The
structure is allowed to rock on the sand
along with the raft (base) foundation. There
are also additional accelerometers on this
structure. One has been added to measure
horizontal acceleration at the base (the
input to the structure), and 2 accelerometers
have been added to measure vertical
accelerations of the base plate (Figure 3) to
measure the rocking amplitude and
frequency. The accelerometers on the 3
upper floors work as before.
Oedometer
The oedometer exerts an axial force on a soil sample of known dimensions, with axial
displacement measured by an LVDT (linear variable displacement transformer), outputting a
calibrated voltage. Hence a stress-strain curve can be generated, and the stiffness and shear
modulus of the soil can be calculated. A view of the loading arrangement in the oedometer is
presented in Figure 4.
The oedometer is balanced such that with the
sample in place, the lever arm is horizontal but
there is no force being applied to the sample.
This is done by adjusting the position at which
the sample sits and also moving a
counterweight on the lever arm. The oedometer
will be set in the correct position for you.
Rather than applying the load to the sample
directly, the oedometer uses a lever to increase
the load by a factor of 11.16. When loaded, the
oedometer lifts the sample base, with the top
remaining at a fixed position, to apply an axial
stress. The soil stresses are relatively small, but
note that a typical sample stress of 1.6MPa
requires an oedometer load of 64.6kg; applied
directly to the soil, this would be 720kg.
Theory
The oedometer test shows that sand (and, in fact, all soil) is a non-linear material. This is not
simply due to the individual grains being non-linear. In general, soil compression tests show 3
stages. Firstly, there is an apparent strain of perhaps 0.5% due to the machine bedding in. With
no load applied, the sample may not be in fully in contact with the fixed part of the oedometer at
its upper end. A small load of around 200g puts the sample into the correct position. The second
stage of the test involves settling of the grains as the compressive force increases, increasing the
relative density of the sample. Finally, individual grains are crushed under high local stresses of
the order of 20 MPa, leading to further sample strain. The end of bedding in should be fairly
evident, but the transition from compaction to grain crushing is gradual. For the loads applied in
this test, there will be little, if any, grain crushing.
When the sample is unloaded,
individual grains unload as you
would expect, but the strain caused
by settlement or crushing of the
grains is not recovered. This leads to
significant permanent strain at the
end of a test.
Figure 5 shows a typical set of
stress-strain results for a range of
initial relative densities (a measure
of how densely the sand grains are
packed see the website for a more
detailed explanation). As expected,
the looser sample shows higher
strains as grain settlement and
rearrangement makes a larger
contribution to the compression.
The soil stiffness is simply the gradient of the loading curve once the sample is in position
starting at around 0.45% strain in the example above. However, it is the soil shear modulus
which is important in our earthquake simulation. Recall from IA materials that the shear
modulus G is related to the Youngs modulus E by the formula
G =
E
2 1+
(1)
However, in our experiment the sample is not free to expand in the horizontal direction, and so a
modified formula needs to be used:
G =
E 1
(2)
2 1 + 2 1 2
In this experiment you will record the time-acceleration data from a number of accelerometers.
However, the shaking table motion is not sinusoidal and so further analysis is limited. You will
learn in the IB Signals course that any waveform can be represented by the sum of a series of
sine waves of varying amplitudes and frequencies. A Fourier Transform (FT) converts the data
from the time domain to the frequency domain. The signals we are analysing are assumed to be
periodic (they repeat themselves at the shaking frequency, as you would expect). We do not lose
any valuable data by working in the frequency domain from these transient vibrations, provided
that we record the signals for a long enough period of time. The FT plot produced by Matlab
should show a large peak at the shaking table frequency for all of the accelerometers, with
smaller peaks at higher frequencies due either to higher frequency resonances in the structure at
its natural frequencies, or components of the shaking motion at frequencies other than the base
frequency.
As an example, in Figure 6 the Fourier Transforms obtained from the structure were used to
construct a response plot. In this case, the structure was fixed rigidly to the base of the shaking
table with no soil present. The x-axis of these plots is the shaking frequency of each test, while
the y-axis is the amplification factor. This is the Fourier transform amplitude, at the shaking
frequency of the floor in question, divided by that of the base.
Figure 6: Experimental data (solid line with markers) and best-fit databook
(dashed line) single degree of freedom curves (case C) for each of the 3
floors. Damping factors are 3%, 1.6% and 1.3% for modes 2, 3 and 4
respectively, with resonant frequency of 3.2Hz.
Operation of Equipment
Shaking table
Figure 7 shows a view of the control panel. To operate the shaking table, first set the desired
frequency with the control marked Speed 2. Set the control marked Speed 1 to be 0.10.2Hz
higher (this sets the frequency that the motor will initially spin at, before the clutch is engaged).
Finally, set the shake time to around 10s. The final readout displays the actual frequency of the
motor during a shake it should be equal (or very close) to the desired frequency.
Data from the shaking table will be logged into Matlab in the same way that it is in the signals
lab (A4), though in this experiment Matlab will not drive the structure.
To initialise Matlab type initialise and press enter. This command runs a script that tells Matlab
the data logger properties. Next, type setiorates(40,1000). This sets the card to log data for 40s
at a rate of 1kHz. Now set up the shaking table, as described above. Finally, when the structure
begins to shake (typically 1213s after pressing start on the control panel), execute the
command run to record the data. This also saves the data: the saved data files are named
testa.mat & testa.csv, testb.mat & testb.csv, etc.; the mat files can be loaded back into Matlab
(using the import wizard), the csv files can be loaded into a spreadsheet for processing.
To analyse the response of the structure, we will calculate the Fourier Transform of the motion at
each floor. (The Fourier Transform is similar to the Fourier Series you met in IA; you will meet
it again in the Signal Processing lab A4 and in the IB lecture course. Essentially, it is a method of
calculating the frequency components of a signal). The function analyse1 displays the recorded
signals in the time domain; the function amp calculates the amplification factors for each
channel.
6
Oedometer
An oedometer sample is contained within a metal cylinder 19.2mm deep, and squeezed between
2 porous discs 75mm in diameter. These allow water to escape from a wet sample as the load is
increased, though with the dry samples here this feature is superfluous. To prepare the sample,
hold the steel ring on top of the bottom disc and pour sand in at a constant rate. A loose sample
is achieved by pouring sand fairly quickly from a very low height (a few mm above the base),
while a dense sample needs a slow pour from a height of 30 or 40cm. Your demonstrator will
give you some advice on how best to pour to achieve your target density. The sample is then
fastened into the carrier. Each group will carry out a test at one of the following relative
densities: 40%, 50%, 60% or 80%.
As this is a dry sand sample, there is no need to wait for more than a couple of seconds for the
reading to stabilise (a wet clay sample might take several days to fully consolidate when loaded).
Compression is measured by an LVDT. This gives a voltage output in the range 4V over a
10mm displacement range. The calibration constant (in V/mm) is required to calculate the total
displacement this will be provided in the lab session.
Calibration constant = ....................................................... V/mm
Results
Oedometer test
Calculation of sample properties
The voids ratio is defined as the ratio of the volume of voids to the volume of solid grains.
Total volume of sample =
d h =
(75 10 3 ) 2 19.2 10 3 = 8.482 10 5 m3
4
4
emax e sample
emax emin
Record your results in the table below. There are 3 additional strain columns for you to record
results from the other groups. When you have a full set of results, plot them on the 2 grids
provided. Calculate the stiffness and shear modulus of the model sand bed, which has a relative
density of 90%. The stiffness is simply the gradient of the loading line. The shear modulus is
calculated using equation 2 (p.4).
Sample Pressure
Load
LVDT output
(kPa) = Load x
(kg)
(V)
24.8
10
15
20
15
10
Displacement
(mm) = LVDT
output
/calibration
constant
Change in
displacement
from 0kPa
(mm)
ID =
Strain
(displacement
change / 19.2)
ID =
Strain
ID =
Strain
ID =
Strain
The shear modulus G can be used to give the soil spring stiffness, K. This can then be used to
refine the theoretical model of the system by incorporating a fourth degree of freedom.
Calculate K for each of the 3 samples using the equation:
K =
2.4
Gb 3
l
3.73 + 0.27
(2 )
b
Summary of results:
Relative density
( %)
Stiffness, E
(MPa)
Shear modulus, G
(MPa)
A3
A2
A1
A6
A5
Figure 10:
Accelerometer
numbering
Carry out the tests shown in the table below, noting the accelerometer
numbering in Figure 10. Use the Fourier transform produced by the
command amp for shakes 1-6 to obtain the frequency spectrum of the
signals, and record the amplification factor for accelerometers 2, 3 and
4 at each shake frequency. The amplification factor is the Fourier
transform peak from the floor in question divided by that of the base at
the shaking frequency.
Plot the amplification factors on the following page, and estimate the
values of natural frequency and damping ratio for each floor. While the
databook curves are shown at a resonant frequency of 2.6Hz, you may
suggest another value if you think your peak lies elsewhere. This
should not affect the curve amplitude (and hence your estimation of the
damping ratio).
If there is time left after the first 6 shakes have been processed,
investigate the rocking of the structure. Process these results in the
same way. You should observe (visually as well as in the analysis!) that
for a given displacement amplitude the structure visibly rocks only
above a certain frequency the 2 tests at each amplitude should be
either side of this critical frequency. A more detailed investigation can
be carried out as an Extended Exercise.
10
Frequency/Hz
Shake
amplitude/mm
Accelerometers
measured
1.5
2.2
2.5
2.8
3.2
4.0
File name
Now process these results. If there is time, continue with tests 7-10.
7
1.5
2.8
1.5
3.5
10
2.2
3.5
Shake
number
Freq/Hz
Shake
amp/mm
1.5
A2/A1
A3/A1
A5/A1
A6/A1
xxxxxx
xxxxxxx
2.2
xxxxxx
xxxxxxx
2.5
xxxxxx
xxxxxxx
2.8
xxxxxx
xxxxxxx
3.2
xxxxxx
xxxxxxx
4.0
xxxxxx
xxxxxxx
1.5
xxxxxxx
xxxxxx
2.8
xxxxxxx
xxxxxx
1.5
3.5
xxxxxxx
xxxxxx
10
2.2
3.5
xxxxxxx
xxxxxx
11
A4/A1
Frequency of maximum
amplification factor (Hz)
Damping ratio
Floor 2
Floor 3
Floor 4
Compare your values of frequency and damping ratio with the theoretical and experimental
values from Figure 6 (another 4-degree of freedom system, though this time without soil). The
IA experiment was a 3-degree of freedom system, and so direct comparison is not possible.
Conclusions
The response and natural frequencies of the structure change when it is placed on realistic
(non-fixed) foundations:
Comment on the change in damping:
Rocking frequency:
At low frequency/amplitude there is little or no rocking
At higher frequency/amplitude rocking frequency = shaking frequency
There is a threshold frequency above which amplification factors are significantly higher
_____________________________________________________________________________________
APPENDIX: Summary of Matlab code
initialise
setiorates(40,1000)
run
analyse1
amp(freq)
Filter
Creates software representation data acquisition card. Only needs to be run once.
Sets Matlab to record data for 40s at a rate of 1kHz.
Starts logging data. Run this command when the structure first shakes.
Displays the captured data in the time domain.
Calculates the Fourier transform and amplification factor for each floor.
The freq argument is optional if not given, the function will calculate the
shaking frequency from the data (overridden if the argument is given a value).
Filters data to reduce noise use this for data from accelerometers A5 and A6
(for rocking analysis). Note Filter not filter.
13