You are on page 1of 5

Soil Behaviour and Hyperbolic Model

IGC 2009, Guntur, INDIA

SOIL BEHAVIOUR AND HYPERBOLIC MODEL

R.B. Arun Murugan


P.G. Student, Division of Soil Mechanics and Foundation Engineering, Department of Civil Engineering, Anna University
Chennai, Chennai–600 025, India. E-mail: arunmurugan18@gmail.com
V.K. Stalin
Assistant Professor, Division of Soil Mechanics and Foundation Engineering, Department of Civil Engineering, Anna
University Chennai, Chennai–600 025, India. Email: staliniisc@yahoo.co.in

ABSTRACT: Many empirical correlations have been obtained to predict the Engineering properties with the help of Index
properties. These empirical equations are not only used to predict the engineering properties, but also can be used as a
measure of validating experimental values. Of the available mathematical correlations, hyperbolic model has been widely
used by many researchers to establish the stress-strain, time-settlement, time-swelling and time-shrinkage characteristics of
fine grained soils. In this investigation an attempt is made to re-examine the suitability of hyperbolic model for widely
varying plasticity characteristics, initial moisture content and initial density. Using hyperbolic model few equations are
proposed in this paper for predicting the stress-strain behaviour, suction-water content and time-shrinkage behaviour of soil
for any liquid limit, initial moisture content and density. The validity of the proposed equations have been attempted and
found to be in good agreement experimental values.

1. INTRODUCTION relationship especially during the primary and secondary


stages. In this paper, attempts are made to predict the stress–
Understanding and prediction of index and engineering
strain behaviour of soil, shrinkage and suction characteristics
characteristics of soils have their own significance in the
of soil based on hyperbolic model for varying initial
geotechnical engineering practices and need hardly be
conditions such as initial moisture content, density, clay
stressed. Empirical correlations are available to predict the
content etc.
properties like compression index swell potential,
permeability coefficient, shear strength etc., using index
properties. Kondner (1963) proposed a functional form based 2. MATERIALS
on hyperbolic stress–strain function, which developed later Natural soils were collected from nearby Chennai area at 1.5 m
by Duncan & Chang (1970) for static and qusai-static depth from ground level by making open trench. Commercial
behaviour of soil. Al-Shayea et al. (2002) used hyperbolic soil bentonite was collected from local market. The physical
model for simulating stress-strain response of soils by properties of natural soils and bentonite are shown in Table 1
varying confining pressure. Sridharan & Prakash (1985) (a) and (b). The soils were selected such that they exhibit
studied from the Terzahi’s theory of consolidation and wide range of plasticity characteristics.
showed that T/U and T relation was rectangular hyperbola
over a fairly wide range of degree of consolidation. In the Table 1: (a) Physical Properties of Soils
plot of T/U versus T, a straight line was fitted for the range
of 60% to 90% degree of consolidation with a high precision, Ben-
Soil- Soil Soil- Soil
Soil description tonite
for calculating the coefficient of consolidation. Stalin et al. (1) (2) (3) (4)
(1)
(2004) observed from the shrinkage test conducted on Specific Gravity 2.68 2.60 2.65 2.66 2.77
bentonite + sand mixtures that the time-shrinkage curves LL % 70 45 24 60 70
generally follow hyperbolic relationship irrespective of the PL % 34 20 18 21 41
soil type and amount of coarser fraction. Dakshinamurthy PI % 36 25 6 39 30
(1978) proposed a new empirical equation to predict the Free Swell Index % 76 40 10 150 294
swelling of expansive clays using hyperbolic relationship. Maximum
17.28 17.8 19.4 18.09 15.2
The swelling experiments were conducted by Sivapullaiah et Dry Density, kN/m3
al. (1996) on bentonite + sand mixtures at a constant density OMC,% 14.5 13.5 12.2 16.95 24.25
and IMC. They found that irrespective of percentage of sand Compressibility
CH CI CL CH CH
in bentonite, the time-swell relationship follow hyperbolic Classification

632
Soil Behaviour and Hyperbolic Model

Table 1: (b) Physical Properties of Soils linearization of the curve, initial portion of the curve
Ben- is ignored. The slopes and intercept of that plot was
Properties
Soil Soil Soil
tonite determined. A typical plot of (strain/stress) % Vs
(5) (6) (7) Strain hyperbolic relationship after ignoring initial
(2)
Specific Gravity 2.71 2.6 2.7 2.8 portion of stress–strain curves is shown in Figure 2.
Liquid Limit % 62.5 69 72 430
Linearised equation is
Plastic Limit % 27.5 31 40 48 Y = X / (c + mX)
Plasticity Index % 35 38 32 382
Shrinkage Limit % 9.3 12 10 7.5
Free Swell Index % 78.0 70 70 350
Maximum Dry
16.5 15.9 16 12.8

Strain/Stress, kN/m2
Density, kN/m3
OMC,% 23.36 25 28 35
Compressibility
CH CH CH CVH
Classification

3. METHODS
Unconfined compressive strength, suction and shrinkage
tests were conducted on natural and commercial soils for
varying initial moisture content and density. The stress–strain
Strain
curve, suction–water content curve and time–shrinkage
curves were analysed for linearization, so as to propose a Fig. 2: Strain Vs Strain /Stress Relationship for Soil–4
more generalized equation for predicting the same.
Where,
3.1 Procedure Involved in the Determination of Soil
Properties Using Hyperbolic Model c = Intercept of (X/Y) Vs X plot
1. The values of ‘X’ and ‘Y’ variable (For example m = Slope of hyperbolic plot of (X/Y) Vs X plot
strain (X)-stress (Y) or time (X)–shrinkage (Y)) is
3. The linear portion of the transformed plot (i.e. ‘X’ Vs
obtained from different engineering properties for
‘(X/Y)’) was critically analyzed using best fitting
varying initial conditions such as density, IMC, clay
curve technique by ignoring some initial points and
content etc., from experimental results. A typical
“X-Y” variables of strain–stress (kN/m2) is shown the same alone was considered for analysis and
in Figure 1. proposing a more generalized empirical equations.

4. RESULTS AND DISCUSSIONS


4.1 Stress–Strain Behaviour of Soil
Considering stress–strain curve as rectangular hyperbola, the
Stress, kN/m2

curves are re-plotted into (strain/stress) Vs strain curve with


high correlation coefficient of 0.95 to 0.91. The soil with low
liquid limit (LL) is not following a hyperbolic relationship.
Hence, such curves observed for low clay content or low
liquid limit or dry side of optimum were ignored. Further, it
is noticed from table 2 that the slope (m) and intercept (c) of
the linear plot ((ε/σ) = m (ε) + c), keeps decreasing with
initial moisture content, initial density and liquid limit.
Strain
Hence, the slope and intercept are independently related to
Fig. 1: Stress–Strain Relationship for Soil–4 ratio of liquid and initial moisture content. A generalized
equation is proposed to predict the stress–strain behaviour of
2. The X’ and ‘Y’ variables were transformed and soil as,
repotted as ‘X’ Vs ‘(X/Y)’. Because the initial
portion of the curve would not impart in the σ = ε/(2.518 (LL/IMC)–1.24 × ε + 0.544 (LL/IMC)–3.23) (1)

633
Soil Behaviour and Hyperbolic Model

Table 2: Best Fit Linear Equations of Transformed Plot of Table 3: Best fit Linear Equations of Transformed Plot of
Stress–Strain Curves of Soil for Varying IMC and Density Suction–Water Content Curves of Soils for Varying Initial
Density IMC Moisture Content (IMC)
Soil Equation
(kN/m3) (%) Soil IMC% Equation
16.0 11.0 (ε/σ) = 0.273 (ε) + 0.002 Soil - 1 70 (ψ/w) = 0.025 (ψ) – 0.904
16.5 12.0 (ε/σ) = 0.282 (ε) + 0.003 Soil - 4 40 (ψ/w) = 0.049 (ψ) – 2.176
Soil - 1 17.28 14.5 (ε/σ) = 0.221 (ε) + 0.004
67 (ψ/w) = 0.033 (ψ) – 4.195
16.5 18.0 (ε/σ) = 0.937 (ε) + 0.001
16 19.75 (ε/σ) = 0.892 (ε) + 0.006 50 (ψ/w) = 0.035 (ψ) – 3.566
Soil - 5
16.5 10.75 (ε/σ) = 0.349 (ε) + 0.003 40 (ψ/w) = 0.041 (ψ) – 3.476
17.84 13.5 (ε/σ) = 0.219 (ε) + 0.007 32 (ψ/w) = 0.042 (ψ) – 1.919
Soil - 2
17.0 17.90 (ε/σ) = 0.7 (ε) + 0.015 63 (ψ/w) = 0.026 (ψ) – 2.093
16.5 19.90 (ε/σ) = 0.950 (ε) + 0.019 53 (ψ/w) = 0.029 (ψ) – 2.114
17.95 16.6 (ε/σ) = 0.525 (ε) + 0.003
Soil - 6
43 (ψ/w) = 0.037 (ψ) – 2.308
18.095 16.95 (ε/σ) = 0.499 (ε) + 0.003 33 (ψ/w) = 0.050 (ψ) – 4.560
Soil - 4
17.95 17.5 (ε/σ) = 0.635 (ε) + 0.003 64 (ψ/w) = 0.022 (ψ) – 1.251
17.7 18.5 (ε/σ) = 0.622 (ε) + 0.003
52 (ψ/w) = 0.034 (ψ) – 2.579
15.2 24.25 (ε/σ) = 0.2163 (ε) + 0.011 Soil - 7
Bent (1) 43 (ψ/w) = 0.038 (ψ) – 2.152
14.5 29 (ε/σ) = 0.572 (ε) + 0.036
90 % Bent 15.4 19.5 (ε/σ) = 0.075 (ε) + 0.007 33 (ψ/w) = 0.047 (ψ) – 3.696
(1) + 10 % 16.0 23.75 (ε/σ) = 0.261 (ε) + 0.03
Sand 15.4 26.5 (ε/σ) = 0.577 (ε) + 0.045 It is observed that the slope is decreasing with increasing
70 % Bent 17.6 18 (ε/σ) = 0.312 (ε) + 0.023 IMC and intercept is increasing with increasing IMC.
(1) + 30 % 17.25 20 (ε/σ) = 0.288 (ε) + 0.025 Further, both slope and intercept are decreases with
Sand 16.5 22 (ε/σ) = 0.787 (ε) + 0.063 increasing clay content (LL). Hence the slope and intercept
18.5 12.5 (ε/σ) = 0.3075 (ε) + 0.001 are independently related to product of LL and IMC. Using
50 % Bent
18.95 15 (ε/σ) = 0.451 (ε) + 0.024 this correlation an empirical equation is proposed to predict
(1) + 50 %
18.5 16.5 (ε/σ) = 0.687 (ε) + 0.056 the suction–water content curve of soil.
Sand
17.5 18.5 (ε/σ) = 1.360 (ε) + 0.076
w = ψ/(90.94(LL × IMC)–0.97 × ψ – 20297 (LL × IMC)–1.4) (2)
The stress–strain values predicted based on above equation Based on above equation (2), the suction–water content
(1) are in close agreement with experimental values irrespective values were predicted and found to be in close agreement
of soil type, density and IMC as shown in Figure 3. with experimental values irrespective of soil type and IMC as
shown in Figure 4.
Stress, kN/m2

Water Content, %

Strain
Fig. 3: Comparisons of Experimental and Predicted Values Suction, kN/m2
of Stress–Strain Curve for Soil – (2) Fig. 4: Comparison of Experimental Vs Predicted Values of
Suction–Water Content Curve for Soil – (1)
4.2 Suction–Water Content Behaviour of Soil
An empirical equation is proposed to predict the suction– 4.3 Time-Shrinkage Behaviour of Soil
water content curve of any soil irrespective of LL and IMC Earlier Stalin et al. (2004) have proposed an empirical
using hyperbolic relationship. Treating suction–water content equation to predict the time–shrinkage behaviour of soil
curve as rectangular hyperbola, the curve is re-plotted into irrespective of soil type. In that equation influence of IMC is
suction Vs (suction/water content) with high correlation not included. The existing equation could not be used
coefficient of 0.999. The slope (m) and intercept (c) of the successfully to predict the shrinkage of soil with IMC lower
linear plot ((ψ/w) = m (ψ) – c) is shown in Table 3. than LL water content. In order to overcome this difficulty,

634
Soil Behaviour and Hyperbolic Model

in this investigation an attempt is made bring the effect of


IMC in the existing equation. The hyperbolic shape of time–

Vertical Shrinkage, mm
vertical shrinkage curve is re-plotted into time–(time/vertical
shrinkage) with high correlation coefficient of 0.94 to 0.99.
The slope (m) and intercept (c) of linear plot ((t/Vsh) = m (t) +
c) with high correlation coefficient is shown in Table 4. It is
observed that the slope is decreasing with increase in IMC
and intercept is increasing with increase in IMC. From the
closer observation of graph, it is noticed that the slopes of
linear equations are decreasing with increase in liquid limit.
Hence, attempt is made to correlate the slope and intercepts Time, min
with liquid limit and IMC of respective soils. Using this
Fig. 5: Comparisons of Experimental and Predicted Values
correlation an empirical equation is proposed to predict time
of Time–Vertical Shrinkage Curve for Bentonite - (2)
–shrinkage behaviour of soil.

Vsh = t/(4.784 (LL × IMC)–0.35 × t + 2693 (LL × IMC)-0.11) (3) 5. CONCLUSIONS


The following conclusions may be drawn from this study,
Table 4: Best Fit Linear Equations of Transformed Plot of
Time–Shrinkage Curves of Soils for Varying Liquid Limit 1. Stress–Strain characteristics of low plastic clays and
(LL) and Initial Moisture Content (IMC) clays whose initial moisture content is on the dry of
LL IMC optimum side are not following hyperbolic relationship.
Soil Equation
(%) (%) 2. The slope and intercepts of linearised transformed plot of
50 (t/Vsh) = 0.116 (t) + 1626 strain–(strain/stress) relationship are decreasing with
Soil - 5 62.5 40 (t/Vsh) = 0.24 (t) + 183.7 increasing liquid limit, IMC and density, irrespective of
30 (t/Vsh) = 0.296 (t) + 819.9 soil type and hence a generalized equation is proposed to
60 (t/Vsh) = 0.095 (t) + 1175 predict the stress – strain characteristics of clays.
50 (t/Vsh) = 0.087 (t) + 1754
Soil - 6 69
40 (t/Vsh) = 0.208 (t) + 556.9
σ = ε / (2.518 (LL / IMC)–1.24 × ε + 0.544 (LL / IMC)–3.23)
30 (t/Vsh) = 0.26 (t) + 1292 3. Suction–Water Content curves were re-plotted into
60 (t/Vsh) = 0.11 (t) + 544.3 suction Vs (suction/water content) and the slope and
50 (t/Vsh) = 0.133 (t) + 759.5 intercepts of the transformed plot is linearly decreasing
Soil - 7 72 with LL and IMC. Accordingly, a generalized equation is
40 (t/Vsh) = 0.149 (t) + 987.5
30 (t/Vsh) = 0.16 (t) + 1071 proposed to predict the suction–water content
70 (t/Vsh) = 0.169 (t) + 648.3 characteristics for any LL and IMC as
Bentonite (1) 70 49 (t/Vsh) = 0.243 (t) + 107.4 w = ψ/(90.94(LL × IMC)–0.97 × ψ – 20297 (LL × IMC)-1.4)
28 (t/Vsh) = 0.625 (t) + 355.7
430 (t/Vsh) = 0.093 (t) + 747.5 4. An empirical equation is proposed to predict time -
Bentonite (2) 430 301 (t/Vsh) = 0.082 (t) + 723.8 vertical shrinkage behaviour for any liquid limit and IMC
172 (t/Vsh) = 0.142 (t) + 722.6 as
70% Bent (1)
49 34 (t/Vsh) = 0.235 (t) + 442
Vsh = t/(4.784 (LL × IMC)–0.35 × t + 2693 (LL × IMC)–0.11)
+ 30 % Sand
50% Bent (1)
35 25 (t/Vsh) = 0.397 (t) + 604.7 REFERENCES
+ 50 % Sand
30% Bent (1) Al-Shayea, S. Abduljuwad, R. Bashir, Al-Ghamedy and I.
21 15 (t/Vsh) = 0.933 (t) + 140.6
+ 70 % Sand Asi (2002). “Determination of Parameters for a
70% Bent (2)
245 172 (t/Vsh) = 0.091 (t) + 772.3 Hyperbolic Model of Soils”, Proceedings of the institution
+ 30 % Sand
50% Bent (2)
of civil engineers, Geotechnical Engineering 156, April
170 119 (t/Vsh) = 0.116 (t) + 796.5 2003 issue GE 2. pp. 105–117.
+ 50 % Sand
30% Bent (2) Dakshanamurthy, V. (1978). “A New Method to Predict
85 60 (t/Vsh) = 0.293 (t) + 503.1
+ 70 % Sand Swelling using a Hyperbolic Equation”, Geotechnical
Engineering, 9, pp. 29–38.
The time–shrinkage values were predicted based on above Duncan J.M. and Chang C.Y. (1970). “Nonlinear Analysis of
equation (3) and found to be in close agreement with Sstress and Strain in Soils”, Journal of the Soil Mechanics
experimental values irrespective of soil type (LL) and IMC and Foundations Division, ASCE, 96, No. SM 5, pp.
as shown in Figure 5. 1629–1653.

635
Soil Behaviour and Hyperbolic Model

Kondner R.L. (1963). “Hyperbolic Stress–Strain Response: Sivapulliah, P.V., Sridharan, A. and Stalin, V.K. (1996).
Cohesive Soils”, Journal of the Soil Mechanics and “Swelling Behaviour of Soil Bentonite Mixtures”,
Foundations Division, ASCE, 89, No. SM1, pp. 115–143. Canadian Geotechnical Journal, Vol. 33, pp. 808–814.
Sridharan, A and Prakash, K. (1985). “Improved Rectangular Stalin, V.K., Anuradha, P. and Ambily A.P. (2004). “A
Hyperbolic Method for the Determination of Coefficient Study on the Control of Shrinkage Potential of Expansive
of Consolidation”, Geotechnical Testing Journal Clays and Their Predictions”, 15th South East Asian
GTJODJ, Vol. 8, No. 1, March 1985, pp. 37–40. Geotechnical Society Conference, November 2004,
Bangkok, Thailand, Vol. 1, pp. 867–870.

636