46 soil dynamics and earthquake engineering - wit press · soil dynamics and earthquake engineering...

Dynamic behaviour of partially saturated sand

H. MeiBner, A. Becker

Institute for Soil Mechanics and Foundation Engineering,

University ofKaiserslautern, D-67653 Kaisers lantern, Germany

Abstract

The main object of this investigation is to identify the significance of capillaryeffects on the dynamic behaviour of partially saturated sand. This isaccomplished through an experimental investigation of the factors affecting theshear modulus ratio. The degree of saturation on partially saturated sand showsthe influence of the capillarity.

The dynamic behaviour of the unsaturated sand is derived from dynamictriaxial tests subjected to sinusoidal loading as far as earthquake-frequencies areconcerned. The samples with diameters 10 cm are subjected to different stressstates and various degrees of saturation. A further parameter to be varied is thevoid ratio.

1 Introduction

One of the most important soil properties governing the behaviour of sand is thedensity. Both settlement and shear strength are dependent on this factor.Compacting sands to increase their density is a widely used method forimproving their stability.

For static cases it is well known that small degrees of saturation lead tohigher shear parameters caused by capillary effects. The behaviour ofunsaturated soils is quite different of that of nearly saturated or completely drysoils.

In analysing seismic behaviour of soil deposits and studying soil-structureinteraction problems, it is essential to evaluate dynamic properties of soils. Forthis purpose, in situ measurements of elastic waves and various laboratory testsusing triaxial apparatus, simple shear or resonant column apparatus have beenconducted by many investigators. It is well known that dynamic properties ofsoil such as shear modulus and damping characteristics depend on parameterssuch as stress condition, void ratio, ambient stress and vibration history, degree

Transactions on the Built Environment vol 14, © 1995 WIT Press, www.witpress.com, ISSN 1743-3509

46 Soil Dynamics and Earthquake Engineering

of saturation, strain amplitude, frequency of vibration, soil structure,temperature, grain characteristics etc.

In the past years numerous experimental studies have been reported, butmost of the studies were directed to static behaviour of partially saturated soil(e.g. Weckbecker [1J). Only a few papers in the literature document dynamicbehaviour of partially saturated soil. Holzlohner [2] shows by means ofexperimental studies the effect of water content on the settlement ofdynamically loaded foundations. Numerous examinations concentrate ondifferent behaviour of dry and saturated soil (e.g. Chang/Chae f3J). Theinfluence of degree of saturation on dynamic shear modulus is shown by Wu etal. [4] using resonant column tests.

To clarify the relationship between dynamic properties of soil and the degreeof saturation, the authors have conducted extensive stress controlled dynamictriaxial tests on partially saturated sand.

2 Test Material

A medium-grained cohesionless sand was used in this investigations. The graindistribution curve is shown in Fig. 1 and material properties such as specificgravity, capillarity and range of void ratio are summarized in Table 1.

Table 1: Material properties

100 „_90

•— * 80&mc«J 60 .f -0)C 40 ..

30 -20 -10 _ .00«

sand

fin*""

J1

17/

medium

^/J

> 01 0.2 0

COWM'

e 2 d [mm]

Parameter

Particle density y* [t/m ]Maximum void ratio e [-]Minimum void ratio e [-]Passive capillarity [cm]Active capillarity [cm]Permeability k [m/s]

Angle of internal friction cp[°]

Value2,635

0.920.5943.573318

MO"*35

Figure 1: Grain size distributionof sand testet

3 Test Equipment

A special triaxial test device was developed at the Institute for Soil Mechanicsand Foundation Engineering Kaiserslautern, based on servohydraulic closedloop systems both for the axial and the radial loading. Fig. 2 shows a principledraft of the test device. A view of the device with control unit as well as dataacquisition is given in Fig. 3.


Soil Dynamics and Earthquake Engineering 47

Due to the fact that axial and radial stresses can be varied independently manyvariations of dynamic stress paths can be applied. In addition to static stressconditions dynamic components were added for the presented investigation. Theapparatus is constructed for cylindrical soil and rock samples with heights from10 cm to 30 cm and diameters up to 10 cm. Depending on the material to betested, stress- and strain-amplitudes a highest possible frequency of about 50 Hzcan act in axial direction with a maximum load of 100 kN. The lateral workingfrequency is limited to about 10 Hz up to now at maximum 25 bar. A specialfeature of the apparatus is the bearing at the top of piston. It is possible toseparate axial loading from torsional loading. Later studies will include this kindof separate loading.

Vertical and radial displacements are monitored by inductive displacementtransducers inside the cell. A view of the radial transducer is given in Fig. 4. Theconfining pressure and the vertical load are measured by pressure transducers.

All datas necessary for adjustment and measurement of the testing processare controlled by computer permanently.

Figure 2: Cross-section of the apparatus and circuit diagram



Figure 3: Triaxial test device with control units and data acquisition

Figure 4: Radial displacement transducer mounted on a rock sample



4 Sample Preparation

Several methods of sample preparation have been examined to find an easy wayof producing homogeneous and reproducible samples. The well-known methodof pouring failed on the low achievable densities at different saturation states.Also the application of vibrations did not improve results. Therefore, it wasdecided to prepare a soil-ice mixture and stamp this material into the test box.The preparation runs as follows:• determination of the masses of soil and fine grained ice for the necessary

density and degree of saturation• stamping the mixture into a membrane premounted in a mould (height 10

cm, inside diameter 10 cm)• positioning the specimen in the triaxial cell• closing the pore pressure system and the triaxial cell• applying confining pressure in drained condition• starting the test program

Because of the larger volume of ice compared with water the samples wereprepared with heights of more than 10 cm. After thawing the desired height wasobtained.

5 Experimental Procedure

The tests were carried out in drained conditions with stress controlled loadhistories as shown in Fig. 5. After isotropic static loading ranging from 150 kPa

Compression failure envelope

.01+03

Extension failure envelope

o, ± Aa,

10cm

Figure 5: Variation of dynamic stress path, a^ = const.



to 600 kPa a deviatoric stress o^ = a, - a^ was applied by changing the axialload at constant confining pressure 03. The axial load changed in sinusoidalform and was progressively raised in steps Ac, as shown in Fig. 6. Herewith thefrequency of the loading varied in the range of 1 Hz to 10 Hz, but was keptconstant during every single test Successively the deviatoric stress wasenlarged to failure in such a way that the stress path reached the compressionfailure envelope before reaching the extension failure envelope. After eachincrease of the deviatoric stress a^ the stress state was kept constant for sometime (Fig. 6). "Failure" is defined in terms of displacements when the specimenbecame very distorted or bulged out.

The samples with void ratios of 0.65, 0.70, 0.75 and 0.80 were tested withdifferent water contents of (0%,) 2%, 4%, 6%, 8% and 10%.

Axial and radial stresses and strains of the samples were measuredcontinuously during the tests by pressure and displacement transducers. Thedatas were digitized by a data acquisition unit and recorded on harddisc forfurther evaluation in two ways. On the one hand sequences of 256 datas with atotal scan rate of 4 kHz were stored in one file every minute, on the other handmean values of these sequences were stored in a second file.

Before preparing the specimen and at the end of the tests small samples forcontrolling water content were taken.

Strain e

ep

CT, ± Aa,

Time t

Figure 6: Load history with respect to time



6 Test Results and Discussion

Fig. 7 shows a series of three dynamic triaxial tests with initial values CQ = 0.70,w = 4% and confining pressure ranging from 150 kPa to 600 kPa. In the way

described above the stress path in the (QJ + c,)/2-(C] - a^)/2-diagram wasincreased until failure.

Test series

A 150/375/600

eo-0.70w = 4%

Figure 7: Results of three dynamic triaxial tests

Table 2 summarizes the evaluated dynamic shear strength parameters.However, the angle of internal friction varies from 24.3° to 28.3° and acohesion of maximum 9.5 kPa is obtained.

Table 2: Dynamic shear strength parameters

eo [-]w[%]

0.02.04.06.08.010.0

0.70

V,4vn n-

27628.228.325.6

-

Cdv,, [kPa]-

4.89.58.04.8-

0.75

(PHvn H26.327027828.225.025.0

Cdvn M0.03.05.75.12.81.3

0.80

9^ M-

25.927224.324.4-

Cd,, [kP%]-1.73.83.02.0-



It was found that material behaviour was independent from loading historyduring the investigations. There was hardly any difference in reaching the failurepoint, either starting with low confining pressures and increasing to a higherlevel or starting with the higher level.

The volumetric strain behaviour for samples with the same confiningpressure o, = 150 kPa and initial values e% = 0.70 at different degrees ofsaturation ranging from (0%) to 10% is presented in Fig. 8. Additional to thewell-known influence of the void ratio the material behaviour depends also onthe water content.

Volume change behaviour

= 0.70

- w=0%--- w=2%...... w=4%---- w=6%---- w=8%

5 25

6, [%]

Figure 8: Volumetric strain as a function of the water contents.

As result of best fitting approximations the cohesion is shown in Fig. 9 withrespect to the void ratio and the water content respectively. The approximationsfor the cohesion CHvn depending on water content w [%] is given by:

with parameters a, 0 and y summarized in Table 3 for the investigated area Theapproximation of Cdyri shown in Fig. 10.

Table 3: Parameters a, P and y for the approximation

eo [-10.700.750.80

a24251.4740.995

33.03.03.0

Y-0.70-0.70-0.70



Capillary cohesion C (QQ, w)

10.0

Cl 4.0

2.0 --

0.0 2.0 4.0 6.0 8.0 100 12.0 14.0 16.0

W [%]

Test results(Table 2)

• Co = 0.70• eo = 0.75A Co = 0.80

Approximation:

Figure 9: Cohesion c^ and approximation

Capillary cohesion c<]yn(eo, w)

150

Figure 10: Cohesion c^ in dependence on water content and void ratio



To get an idea about the distribution of water content in the sandspecimensmall samples were taken at different heights after the tests. A description ofthese measurements is given in Fig. 11. The following results can besummarized:• water content varies throughout the specimen independently from the initial

confining pressure• water "moves" to the bottom plate• balancing, i.e. supposing that water did not come out the specimen, a kind

of equilibrium was reached• the process of seeping seems to be independent from duration of loading• with the exception of initial water content w = 2% the difference in water

content between top and bottom of the specimen ranges from 1% to 3%• the lower the initial void ratio the lower the above mentioned difference in

water content

Initial water content w = 2% Initial water content w = 6%n _ 2 "[%]

Initial water content w = 4% Initial water content w = 8%

z [cm]

Figure 11: Distribution of water content after testing

7 Conclusion and further investigations

Dynamic triaxial tests on partially saturated sand were carried out in drainedconditions to investigate the influence of water content on dynamic soilparameters and volumetric strain. Due to capillary effects a kind of "dynamiccohesion" can be defined. This effect mostly depends on void ratio and watercontent and shows its peak value at about 4% to 5% water content. Static



triaxial tests with the same material show the maximum influence of capillarity,i.e. peak values for the cohesion in the region of 6% water content. But incontrast to these tests, in which capillary influence measurable appears at initialvoid ratios of 0.70 to 0.65, the presented investigations established cohesionalready at initial void ratios of 0.80.

An empirical approximation for the cohesion is formulated with respect toboth the water content and the void ratio. Three parameters describe theapproximation in the course of which one is dependent on the void ratio and fortwo of them a physical interpretation is not found yet.

The value of internal friction shows a certain dependence on the previousparameters and is much smaller compared with the static value.

The influence of water content on the volume change behaviour is shown,whereas the influence of frequency in the range below 10 Hz seems to benegligible with regard to the results of the dynamic shear strength parameters.

Concerning the damping ratio no examinations had been done. This will beone of the next steps besides investigations of sands with smaller void ratios, i.e.higher densities.

References

[1] Weckbecker, W. (1993), "Beitrag zur Berechnung oberflachennaherTunnel", Veroffentlichung des Fachgebietes Bodenmechanik und Grund-bau der Universitat Kaiserslautern, Dissertation

[2] Holzlohner, U. (1977), "Setzung von Schnellbahnfundamenten infolgedynamischer Last", BAM-Bericht 48

[3] Chang, P.W., Chae, Y.S. (1987), "A parametric study of effect ofvibration on granular soils", Soil Dynamics and Liquefaction, ed. Cakmak,Elsevier, pp. 137-151

[4] Wu, Shi-ming, Gray, D.H. and Richard, F.E. Jr. (1984), "Capillary Effectson Dynamic Modulus of Sands and Silts", J of Geotechnical EngineeringDiv., ASCE, Vol. 110, No. 9, pp. 1188-1204


46 soil dynamics and earthquake engineering - wit press · soil dynamics and earthquake engineering...

Documents