See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/279502749

Duplicate testing conducted on the input parameters for the estimation of

potential expansiveness of clay.

Conference Paper · April 2015

CITATIONS READS

4 6,759

3 authors:

Elizabeth Theron Wessel Badenhorst

Central University of Technology Central University of Technology
51 PUBLICATIONS 139 CITATIONS 1 PUBLICATION 4 CITATIONS

SEE PROFILE SEE PROFILE

Pr Stott
Central University of Technology
14 PUBLICATIONS 29 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Soil Suction Measurement - A Review View project

A Review of Expansive Soils Stabilization Using Chemical & Mechanical View project

All content following this page was uploaded by Elizabeth Theron on 01 July 2015.

The user has requested enhancement of the downloaded file.

 Innovative Geotechnics for Africa - Bouassida, Khemakhem & Haffoudhi (Eds.) 2015 © ISBN : 978-9938-12-936-6

Duplicate testing conducted on the input parameters for the estimation of

potential expansiveness of clay.

Essais dupliqués sur les paramètres d’entrée pour l’estimation du potentiel de gonflement des
argiles.

Wessel BADENHORST1, Elizabeth THERON2, Philip STOTT3

1,2,3Central University of Technology, Bloemfontein, South Africa

ABSTRACT – This paper presents a study conducted on three active clay soils analysed at seven commercial
laboratories in South Africa. Commercial test results are often used as input parameters for prediction models
used to estimate potential heave expected from active clays, especially when designing foundations for light
structures. This paper briefly looks at the typical results obtained from such laboratories and comments on the
correlation achieved.

RÉSUMÉ – Cet article présente une étude menée sur trois argiles actives testées dans sept laboratoires privés en
Afrique du Sud. Les résultats de ces essais sont d’habitude utilisés comme paramètres d’entrée de modèles de
prédiction en vue d’estimer le potentiel de soulèvement d’argiles actives, en particulier lors du dimensionnement
des fondations de structures légères. Le présent article évalue les résultats typiques de ces essais de laboratoire et
discute les corrélations proposées.

1 INTRODUCTION

The American Society of Civil Engineers estimates that of all homes in the United States, a quarter has had some
damage caused by expansive soils. The financial loss to property exceeds that of earthquakes, floods, hurricanes
and tornadoes combined.
In Africa there exists a widespread problem of providing economic housing for lower income communities.
In South Africa the government is attempting to provide small subsidised houses for the very poor. Most of
South Africa has semi-arid and sub-humid conditions (Weinert, 1980) which lead to generally shallow residual
soils subject to seasonal saturation and aridity. Such conditions are known for giving expansive foundation
problems.
Advanced mathematical solutions are available for heave prediction, but unfortunately they rely on the input
Of the soil water characteristic curve (swcc) for which the typical cost is US $10 000 (Fredlund 2009) - more
Than the cost of a low cost house. Besides the cost, the time needed to develop the swcc takes in excess of 6
weeks.
In practice almost all low-cost housing relies on empirical methods – estimation of heave potential based on
quick, simple and economical tests. The most popular method in South Africa (Van der Merwe 1964), uses
Atterberg limits and particle size distribution These tests are also known as “Foundation Indicators”. Predictions
of heave are only as reliable as the input parameters. This paper focuses on the reliability of these input
parameters in the form of tes t results received from seven premier commercial laboratories.

2 Particle size analysis of soil samples

Particle size analysis of soil samples is a combination of grading analysis using sieves with different size
apertures and hydrometer analysis which uses settlement and Stoke’s law to determine the finer fractions of a
soil sample.

2.1 Grading Analysis

Figure 1 shows results from a sample that was divided and sent to seven leading commercial laboratories. The
results show an acceptable variance. Results are closely grouped from 13.2mm to 0.425mm. Sample preparation
might have played a role in divergence of results from 0.425mm to 0.075mm. The authors have noted that
details of the preparation of fines samples are critical and have a profound influence on the end result obtained.

605
 Figure 1. Grading analysis of Brandwag sample.

Six of the laboratories used in this study are accredited with SANAS and conform to the international ISO:17025
standards, suggesting that the quality is well managed and ensured.
For each of the three samples analysed, the grading analyses were reasonably consistent.

2.2 Hydrometer Analysis

Hydrometer analysis is used to determine the various fractions of fines in a sample. The method is based on
Stoke’s law and is flawed due to several of its assumption being dubious. Clay particles may be flakey and may
have very large specific surface areas (Whitlow, 2001). Hydrometer accuracy is doubtful for four reasons
(Savage 2007),: Stoke’s law assumes all particles are spherical, de-flocculation may not be complete at the time
of testing, clay particles are partially carried down by larger particles and a relative density of 2.65 is assumed
for all particles, which may not be true. More on this in section 2.3.
Figures 2a and 2b indicate the variance obtained from different commercial laboratories. Note that the values
vary between 63% and 18% on the Steelpoort sample (Figure 2b), a difference of 45%.

Figure 2a. Hydrometer analysis of Botshabelo Sample.

606
 Figure 2b. Hydrometer analysis of Steelpoort Sample.

2.3 Savage’s method to determine the 0.002mm fraction

Savage suggested using Skempton’s activity formula to relate activity to the ratio (R) of the liquid limit (LL) to
the plastic limit (PL) of a soil sample. Savage’s analysis is as follows:

Activity = PI / P0.002 (1)

Where P0.002 refers to the percentage of material smaller than 0.002mm

Savage found an exponential relationship between the ratio LL / PL (R) from a table of the activity values for
Montmorillonite , Illite and Kaolinite published by Cornell University in 1951.

Activity = 0.16R2.13 (2)

Using formulas (1) and (2) the clay content can be established empirically as:

P0.002 = PI / Activity = PI / 0.16R2.13 = 6.25 PI.R-2.13 (3)

The P0.002 value obtained is based on the PI that was tested, typically at P 0.425, and does not represent the
whole sample. The equation is adjusted to reflect the entire sample (PI Gross):

PI0.425 x P0.425 = PIGross (4)

P0.002 = 6.25 PIGross.R-2.13 (5)

Table 1 draws a comparison between the hydrometer values for clay fraction and those derived from Savage’s
formula (5).

Table 1. Hydrometer P0.002 compared to Savage P0.002.

Sample LAB1 LAB2 LAB3 LAB4 LAB5 LAB6 LAB7

Steelpoort
Hydrometer 17 56 24 26
Savage 68 44 11 49
Brandwag
Hydrometer 49 58 33 44 56 50 47
Savage 51 27 18 44 40 43 36
Botchabelo
Hydrometer 35 56 29 43 44 40 35
Savage 31 26 22 30 31 37 34
In Table 1 it can be seen that some of the results compare favourably, while those in italics are unacceptable.
Savage’s method does not consistently predict lower or higher values and the values do not correlate well. Figure
5 shows a graphical representation of the Steelpoort sample comparison.

607
 Figure 3. Steelpoort sample Hydrometers compared to Savage’s values.

2.4 Atterberg limits

Atterberg limits include the Liquid Limit (LL), Plastic Limit (PL), Plasticity Index (PI = LL-PL), the Shrinkage
Limit (SL) and Linear Shrinkage (LS).
Commercial laboratories in South Africa make use of the Casagrande –cup apparatus and typically perform
the LL test according to TMH1:1986 Method A2, which uses the same apparatus as the British Standard BS
1377-2. The authors have found that sample preparation has a significant bearing on the results, which may be
responsible for the variance between the different commercial laboratories. Figure 4 illustrates typical results
obtained for one of the three different active clay samples.

Figure 4. Liquid limit, plastic limit, plasticity index and linear shrinkage on Brandwag sample.

Based on figure 4, it is clear that the LS values are relatively consistent, with most laboratories within two per-
cent of the average. The LL, PL and PI values show a large variance, although there is a grouping of LAB2,
LAB3, LAB5 and LAB6 which seem to compare well.
Casagrande’s plasticity chart, as used by the Unified Soils Classification System can be derived from the
relationship between the LL and PI of a soil sample.
Figure 5 shows the relationship between the liquid limit and plasticity index for the Steelpoort sample.

608
 Figure 5. Casagrande’s plasticity chart, Steelpoort sample.

3 Van der Merwe’s empirical method for the estimation of potential heave

Van der Merwe’s method relies on the PI of the gross sample and the P 0.002 fines fraction, which according to the
British Classification is deemed to be the clay fraction of a soil sample. For Van der Merwe’s method to give
Valuable output, valuable input is required. This paper does not focus on the prediction model itself, but rather
the input parameters.
For better understanding of the method, a simplified example of the process follows:
PIGross is plotted on the y-axis and P0.002 on the x-axis as the “classification of heave potential” curve, (after
Van der Merwe 1964). Figure 6 shows an example of such a plot with values obtained from various laboratories
using Hydrometer analysis to obtain the P0.002 fraction.
Table 2 and Figure 6 compare the resultant heave potential to that obtained using Savage’s formula to
determine the P0.002 fraction.
Skempton suggested using the relationship between the P0.002 fraction and the Plasticity Index to give an
indication of the heave potential of soils (Skempton 1953). He suggested using slopes of less than 0.75 to refer to
inactive clays, slopes of more than 1.4 would suggest active clays and everything in between would be referred
to as normal clays. In figure 6 those designations have been adjusted to reflect slopes of 2.0, 1.0, 0.7, 0.6 and 0.5,
with anything less than 0.5 considered inactive.

Table 2. Heave Potential: Hydrometer Analysis compared to Savage’s Method

Method LAB1 LAB2 LAB3 LAB4 LAB5 LAB6 LAB7
Hydrometer Medium Medium Medium High Low Very High Low
Savage Medium High Medium Medium Medium Very High Low

609
Figure 6. The classification of heave potential after Van der Merwe, 1964, using Hydrometer values and Savage’s
values for the P0.002 fraction with Skempton’s Activity lines.

4 CONCLUSION

Seven leading commercial laboratories were tasked with performing “Foundation Indicators”, which refer to the
Atterberg Limits and Particle Size Analysis. Although the physical sieving provided comparable results down to
the 0.425mm sieve, anything finer proved troublesome. The authors concluded that the problem probably lies
with the preparation of the samples, as some details of preparation were found to have a major impact on testing
done “in-house”.
The Atterberg Limits were performed on fractions passing the 0.425mm sieve, and preparation might have
played a role there also. A grouping of laboratories obtained results that compared well, while two laboratories
found substantially different results.
The finer fractions, those passing the 0.075mm sieve, proved problematic as not all of the laboratories use the
same method, although theoretically they should yield similar results. The results varied substantially and the
range between the highest and lowest P0.002 is alarming.
It can be concluded that using “foundation indicators” alone as the basis for empirical heave prediction
methods is a very risky approach and that other approaches need to be identified.

REFERENCES

Fredlund D.G., (2009) Paradigm Shifts to Facilitate the Practice of Unsaturated Soil Mechanics. 8th C.W. Lovel
Lecture. Perdue Geotechnical Engineering. 26 October 2009.
Nelson J.D., Miller D.J., (1992) Expansive Soils – Problems and Practice in Foundation and Pavement
Engineering. John Wiley and Sons, New York, Chicester, Brisbane, Toronto, Singapore.
Skempton, A.W., (1953) The colloidal “Activity” of clays. Proceedings of the 3rd International Conference of
Soil Mechanics and Foundation Engineering.
Van der Merwe, D.H., (1964) The prediction of heave from the plasticity index and the percentage clay fraction
of soils. The Civil Engineer in South Africa. June 1964, pp 103-107.
Weinert, H.H., (1980) The natural road construction materials of southern Africa. Pretoria: Academica.
Whitlow, R., (2001) Basic Soil Mechanics. Fourth Edition, first edition published in 1983. Prentice Hall,
Pearson Education.

610
View publication stats