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

net/publication/268448631

Evaluation of the Collapsibility of Soils in the Semiarid Region of Pernambuco,

Brazil

Article in Journal of Civil Engineering and Architecture · November 2014

DOI: 10.17265/1934-7359/2014.10.010

CITATIONS READS

7 168

1 author:

Silvio Ferreira
Federal University of Pernambuco
66 PUBLICATIONS 172 CITATIONS

SEE PROFILE

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

Behavior of saturated and unsaturated soils and their effects on the environment View project

Hydraulic conductivity and soil interaction with percolating fluid View project

All content following this page was uploaded by Silvio Ferreira on 16 May 2018.

The user has requested enhancement of the downloaded file.

 Geotechnical and Geophysical Site Characterization 4 – Coutinho & Mayne (eds)
© 2013 Taylor & Francis Group, London, ISBN 978-0-415-62136-6

Evaluation of the collapsibility of soils in the semiarid region

of Pernambuco, Brazil

S.R.M. Ferreira
Federal University of Pernambuco, Catholic University of Pernambuco and University of Pernambuco,
Recife, Brazil

S.P. Fucale
University of Pernambuco, Recife—PE, Brazil

ABSTRACT: Some unsaturated soils may undergo volumetric changes when submitted to an increase
in its water content or are inundated under applied loads. This behavior is related to the volumetric insta-
bility when the water content is changed. Natural collapsible soils in Brazil are generally found in alluvial,
colluvial and residual soils. There are known occurrences of natural collapsible soils in the States of
Amazonas, Piaui, Pernambuco, Bahia, Minas Gerais, Tocantins, Mato Grosso, São Paulo, Paraná and
Rio Grande do Sul. In the last two decades, many public projects have been developed in areas where
the occurrence of collapsible soils has been associated to geotechnical problems. This paper study of col-
lapsible soils in the State of Pernambuco, it has been associated with large engineering projects, such as,
housing and irrigation canals. The geotechnical investigation program included test with a field apparatus,
called Expanso-colapsometer, which allows the measurement of the field settlements of a small (0,10 m)
plate inserted at any depth inside a auger boring hole. Reconnaissance borings with SPT, investigation pits
with undisturbed block sampling and disturbed samples for laboratory tests where also made in order to
assess the type and characteristics of the soil. Field tests used the expanso-colapsometer to measure the
settlement of the soil in selected depths under controlled flooding. Laboratory work included double and
standard oedometric tests with a controlled rate of water inflow of 1.0 ml/s. It was found that the volume
change of the soils when flooded depends on their natural stress state (vertical consolidation stress, suc-
tion and structure) and the conditions that they are submitted to.

1 INTRODUCTION In Petrolina there are fluvial deposits with pre-

dominance of sandy soils (Dystrophic Quartz
In nature there are structurally unstable soils, Sands), with a slightly undulating to planar relief.
unsaturated, in which there are sudden changes in The climate is hot, with a mean temperature above
the stress-deformation behavior when the degree 18 ºC, with the rainy season in the summer. The
of saturation increases. These changes occur either mean annual precipitation lies between 250 and
under the in situ vertical total stress (self weight), or 500 mm, with peaks of 800 mm.
with increased vertical loading due to engineering In Petrolândia there is an eluvial surficial soil
constructions done before the degree of saturation from the Tacaratu Formation, which is character-
increase. They are called collapsible, and occur in ized by a medium to coarse sandstone, occasionally
several regions of the world. Collapse may mani- conglomeratic and ill sorted. The relief is planar;
fest in residual, colluvial, alluvial, aeolian, soporific the climate is hot and semi-arid. The mean annual
and compacted soils. In Brazil they occur in sev- temperature is above 18 °C, and the maximum and
eral States: Amazonas, Bahia, Ceará, Goiás, Mato minimum rainfall is 930 and 80 mm respectively,
Grosso, Minas Gerais, Pernambuco, Piaui, Rio with a mean of 437 m.
Grande do Sul, São Paulo, Tocantins and the Fed- In Salgueiro there are fluvial deposits with pre-
eral District. This paper presents characteristics of dominance of sandy soils (Dystrophic Quartz
volume change due to inundation of collapsible soils Sands). The climate is tropical semi-arid and its
of the States of Pernambuco in cities of Petrolina, average annual temperature is 25 °C. Rainfall
Petrolândia and Salgueiro. The soils from these three in the city ranges from 450 to 600 mm per year,
sites were chosen because they are involved in irriga- the wettest months of December to March, with
tion projects and in large housing construction. scarce and poorly distributed rains, Xerophytic

1833
Scrub vegetation and temporary rivers. With easy
access to and equidistant from the most northeast-
ern capital is only 518 km from Recife—PE.

2 GEOTECHNICAL INVESTIGATION
PROGRAM

The investigation program developed at the

Laboratory of Soils and Instrumentation of the
UFPE in order to analyze the behavior of collapsi-
ble soils has two phases: field and laboratory work.
The field investigations consisted of geotechni-
cal characterization of the subsoil, extraction of
representative soil samples and undisturbed block
samples, and evaluate the volumetric changes in
situ due to inundation at different depths with
the Expanso-colapsometer. The laboratory tests
consisted of grain size analysis and the evaluation
of the influence of the stress state on the collapse
behavior due to inundation, by means of single
and double oedometric tests.
In the single oedometric tests the applied verti-
cal stresses were increased with the ratio Δσ/σ = 1,
with an initial stress of 10 kPa up to 1280 kPa. The
duration of each stage was such that the defor-
mation between two consecutive time intervals
(Δt/t = 1) were less than 5% of the total deforma-
tion up to the previous recorded time. The verti-
cal deformations due to inundation were measured
at 0; 0,10; 0.25 1, 2, 4, 8, 15, 30, 60, 120, 480 and
1440 minutes.
Some tests of the Petrolina soils a specific meth-
odology was employed in order to compare field
results with the Expanso-colapsometer and labora-
tory results with single oedometric tests.
In double oedometric tests a specimen was
loaded at the natural water content, while the other
was previously inundated under a vertical stress of
1.35 kPa with an inundation rate of 1.0 ml/s. In
other aspects the procedure was similar to that of
single oedometric tests. The vertical stresses were Figure 1. Expanso-colapsometer—equipment for the
the same as those used in the single specimen tests. measurement of volume change with time in situ under
The unloading stresses were 640; 160; 40 e 10 kPa. controlled water flow.
Each stage (loading and unloading) was main-
tained during 24 hours.
In the field, the Expanso-colapsometer was used, with control of the discharge rate. This equipment
shown in Figure 1. Ferreira (1993), Ferreira & identifies potentially collapsible or expansive soils.
Lacerda (1993 e 1995), Ferreira et al (1996) and The equipment is composed of two parts. The
Ferreira et al (2002) describe in detail its principles. first simulates a plate load test (plate diameter of
It is a simple equipment that permits the measure- 100 mm), and the second is the controlled inunda-
ment of vertical movements of unsaturated soils. tion kit. These tests were done at the three loca-
It permits to follow in the field the same stages tions. In Petrolândia tests were made at a depth
of single and double oedometric laboratory tests, of 0.5 m, and the loading stages followed the ratio
preserving the natural water content and struc- Δq/q = 1. The vertical stresses at which inundation
ture of the soil. The curve of stress applied at was allowed were 10; 20; 40; 80 and 160 kPa. The
the plate against vertical deformation is obtained duration of each stage was established such that
at the desired depths. Inundation can be applied the difference in vertical movements between two

1834
consecutive readings was less than 5% of the previ- cases the soil is about the same, fine sand with lit-
ous movement, with time intervals Δt/t = 1. The soil tle medium sand, yellow, medium to very dense,
was inundated with a discharge rate of 1.0 ml/s. Figure 2B. At a depth of 5.5 m there is a highly
In Petrolina the tests were performed at different resistant layer (45 blows/0.1 m). No water level
locations, depths and inundation at different verti- was found up to 5.6 m, when the boring was inter-
cal stresses: 50, 130, 140 and 180 kPa, with inunda- rupted (impenetrable to the SPT sampler). NSPT
tion rates between 0.8 and 1.0 ml/s. In Salgueiro values ranged from 10 to 56 at a depth of 5.3 m
the tests were performed at different locations and the number of blows increase almost linearly with
depths and inundation at different vertical stresses: depth. These high values are due to the high suc-
50, 100 kPa, with inundation rates 1.0 ml/s. After tion of the soil that had a natural water content
the tests the water content of the soil under the varying from 1.6 to 3.3%. The second boring, with
plate was performed each 0,10 m under the plate pre-wetted soil, showed NSPT values ranging from
depth, until the water content was the same of 7 blows/ 0.30 m at 0.35 m to 19 at 5.3 m, a reduc-
the original soil. This permitted the evaluation of tion of 30% at shallow depths to 70% at the end
the wetting bulb at all sites. of the collapsible layer, showing that the reduction
in suction is the governing factor in the Standard
Penetration Test results. This sounding was contin-
3 RESULTS ued up to 9.4 m, as shown in the Figure 2.
Figure 3 shows the Activity and Plasticity chart
3.1 Geotechnical soil sections of these soils. The points are located slightly above
line A and the liquidity limits lie between 15 and
The geotechnical section at Petrolina consists of 65%. The soil from Salgueiro is poorly graded
two layers before the impenetrable layer from SPT sand, not liquid and not plastic (SP). The major-
borings was reached, Figure 2A. The first layer ity of the soils from three sites are classified as SP
consists of clayey sandy silt, loose to medium and SM.
dense, yellow brownish in color, with a thickness
of 3.5 m. The second is silty fine sand, with gravel,
mica, very dense, grayish yellow, with a thick- 3.2 Oedometric tests
ness of 0.15 m. The impenetrable layer (NSPT) is a The variation of collapse (or expansion) poten-
schist of equigranular texture, moderately weath- tial obtained from double oedometric tests (CP or
ered with a slight dip, and small quartz veins. The SP ((%) = (ΔH/Hi) × 100, where ΔH is the varia-
phreatic level was found at a depth of 2.9 m. tion of the specimen thickness due to saturation
In Petrolândia two SPT soundings were and Hi is the initial thickness) with the vertical
performed. The first was without water circula- consolidation stress of the soils at the Petrolina
tion, with the soil in its natural water content. The and Petrolândia is shown in Figure 4. Results of
second was in a previously inundated soil. In both tests performed in the city of Palmas—To, Brazil
(Ferreira et al. 2002) are also shown. The col-
lapse potential of the Petrolândia soil increase
with vertical stress, reaching a maximum at
320 kPa. The same behavior was observed with

Figure 3. Plasticity and activity charts (Vargas 1985) of

Figure 2. Typical soil profiles at the three sites. the soils.

1835
Figure 4. Collapse/expansion potential as a function of
vertical consolidation tests—double oedometric tests.
Figure 5. Correlation of collapse potential in double
and single oedometric tests. CPED—Collapse potential,
double oedometric test; CPES—Collapse potential, sin-
the Petrolina soil, which showed maximum col- gle oedometric test; a—Linear coefficient; b—Angular
lapse under a stress of 630 kPa. In Palmas this coefficient; |r| – Correlation coefficient.
was not observed. This is a typical behavior
of collapsible soils, Vargas (1973), Aragão &
Melo (1982), Alonso et al. (1987), Ferreira
(1995). In Petrolina a small expansion was
measured for vertical stress smaller than 40 kPa.
A comparison of the collapse potential
obtained from double oedometric tests and sin-
gle oedometric tests is presented in Figure 5. The
values of collapse potential determined through
single oedometric tests are greater than those
obtained in double oedometric tests. The wetting
path followed in these tests (loading—inundation,
and inundation—loading) has an influence on the
values of the collapse potentials. Similar results
were found by Benvenuto (1982) and Ferreira
(1995).

3.3 Comparison between field and laboratory

values
Typical curves of vertical collapse deformation (%)
with the logarithm of time (min) and the variation
of the volumetric strain variation with the applied
vertical stress obtained from single oedometric tests
(laboratory) and with the Expanso-colapsometer
(field) are shown in Figure 6.
In the laboratory the time necessary for 98% of
the total deformation due to collapse after inunda- Figure 6. Collapse vs. log time—A) Field; B) Laboratory
tion varies from 1 minute (specimens 5 and 7) to (Petrolina soil).
30 minutes (specimen 1). The values of collapse
deformation with time determined in the field are
smaller than those of the laboratory tests. The dif- Further it was found that the collapse poten-
ference can be attributed to the smaller percola- tial measured in the field is smaller than those
tion path in the laboratory specimen, about 2 cm measured in the laboratory. The relationship
in thickness, thus reaching the final equilibrium between these two values is approximately linear,
w.c. in less time, as the curves of the Figure 6 show. CPFIELD = 0.83 CPLAB, Figure 7. A similar value,
This behavior has been verified before (Ferreira, 0.85, was also found in another study, Ferreira
1995). (1995), in a collapseble soil of Petrolândia.

1836
 Figure 8.

Figure 7. Relationship between collapse potential

obtained in field and laboratory tests.

Such a fact is associated to the following fac-

tors: non uniformity of the vertical stress under the
plate in the field, which diminishes with depth, and
the non-uniformity of water content, which also
decreases with depth. This behavior was presented
in previous papers (Ferreira & Lacerda 1993, 1995),
in studies of the collapsible soil of Petrolândia.
Are shown in Figure 8 the potential results
obtained in the field collapse in three sites inves-
tigated by testing with Expanso-colapsometer. The
values of the potential collapse grow with the addi-
tion of stress flooding. The lowest values of the Figure 9. Measured settlements and simulation results
potentials collapse was found in soils of Petrolina of the Expanso-colapsometer test, Costa et al. (2003).
in relation to those obtained in soils of Petrolândia
and Salgueiro.
The simulations carried out for Costa et al.
(2003) made showed the capability of the BBM
model and the computer code, CODE_BRIGHT,
to deal with hydro-mechanical behavior of collaps-
ible soils of Petrolândia. Figure 9 presents the set-
tlements measured in the field at the point located
under the plate axis and the results obtained in
the numerical simulation. It can be verified that
for both natural water content soil and previous
flooded one the settlements predicted are very
good not only in respect to the qualitative behavior
but also concerning the amount achieved. The pre-
dicted average values are about 88% of the experi-
mental data. The wetted front zone, at the end
of the previous flooded soil test can be analyses Figure 10. Water content variation with depth in the
in Figure 10. Comparison between in situ data previous flooded test, Costa et al. (2003).
and simulation results of water content variation
with depth indicate a similar behavior of the wet-
ted front. A large discrepancy is observed at the 4 CONCLUSIONS
surface; however average values are 25% higher
than the experimental data. These results also The volumetric strain of the studied is about 15%
demonstrated that the Expanso-colapsometer is a smaller in the field as compared to laboratory tests.
simple and flexible apparatus able to be used as in The collapse potential measured in the field
situ test. are, for all locations, smaller than those measured

1837
in the laboratory. The relationship between these Ferreira, S.R.M. (1993). Comportamento de Mudança
two values is approximately linear, CPFIELD = 0.83 de Volume em Solos Colapsíveis e Expansivos (In
CPLAB. Portuguese). 7º Congresso Brasileiro de Geologia de
The Expanso-colapsometer performed satisfac- Engenharia. V.1, pp. 283–299. Poços de Caldas-MG.
Ferreira, S.R.M. (1995). Colapso e Expansão de Solos
tory, its main advantage being that it is not neces- Naturais não Saturados Devidos à Inundação” Rio
sary to carve undisturbed blocks in a very unstable de Janeiro, COPPE/UFRJ. PhD dissertation (In
and dry soils, besides maintaining the original state Portuguese), 379p.
of stress in the field practically unchanged. Ferreira, S.R.M. & Lacerda, W.A. (1993). Variação de
Volume em Solo Colapsível Medidas Através de
Ensaios de Campo e de Laboratório (In Portuguese)
ACKNOWLEDGMENTS Revista Solos e Rochas. Vol 16, Nº 4 pp 245–253
Ferreira, S.R.M. & Lacerda, W.A. (1995). Volume Change
The authors are grateful the CNPq/Brazil and Measurements in Collapsible Soil by Laboratory and
Field Test. 1st International Conference On Unsatu-
FACEPE by support in the investigation described rated Soils. Vol. 2 pp. 847–854. Paris.
above were performed. Ferreira, S.R.M.; Lacerda, W.A. & Lopes, F.R. (1996).
Down-Hole Collapse System. Journal of Geotechni-
cal Engineering, 122, 10, 869–872, Discussion.
REFERENCES Ferreira, S.R.M., Fucale, S.P., Amorim, S.F. & Lacerda,
W. (2002). Comportamento de variação de volume
Alonso, E.E.; Gens, A. & Hight, D.W. (1987). Special em solos colapsíveis da cidade de Palmas—Tocantins
Problems Soils. Proceedings of the 9Th European In: XII Congresso Brasileiro de Mecânica dos Solos e
Conference on Soil Mechanics and Foundation Engenharia Geotécnica, São Paulo. 2. 595–603.
Engineering. Dublin, General Report, Session 5, pp. Vargas, M. (1985). The Concept of Tropical Soils.
5.1–55. 1st International Conference on Geomechanics of
Aragão, C.J.G. & Melo, A.C. (1982). Fundações Rasas Tropical Lateritic and Saprolitic Soils, ISSMFE,
em Solos Colapsíveis no Semi-Árido em Pernambuco. Brasília—Brazil.
(In Portuguese), VII Cobramsef, Olinda, 2, 9–40. Vargas, M. (1973). Structurally Unstable Soils in Sour-
Benvenuto, C. (1982). Uma Avaliação Teórico-Experimental ther Brazil. Proc. Of The 8th I.C.S.M.F.E., Moscow.
do Comportamento de Alguns Solos Colapsíveis. 2, 239–246.
(In Portuguese) Tese de Mestrado, EPUSP, São Pau-
lo-SP, 123p.
Costa, L.M., Pontes Filho, I.D.S.P. & Ferreira, S.R.M.
(2003). Numerical modelling of hydro-mechanical
behaviour of plate tests with wettig in collapsibles
soil. Proceeedings of the Computational Plasticity VII
Fundamentals and Applications. Barcelona, Espanha.
(CIMNE), 1. 1–8.

1838

View publication stats