230 103 28MB
English Pages 450 pages [452] Year 2020
ENGINEERING CHARACTERISTICS OF ARID SOILS
Taylor & Francis
Taylor & Francis Group http:/taylorandfrancis.com
PROCEEDINGS OF THE 1ST INTERNATIONAL SYMPOSIUM ON ENGINEERING CHARACTERISTICS OF ARID SOILS LONDON I UK I 6-1 JULY 1993
Engineering Characteristics of Arid Soils Edited by
P.G.FOOKES
Winchester, Hampshire, UK
R.H.G.PARRY
ISSMFE, Cambridge, UK
C\ Taylor&Francis ~
Taylor & Francis Group
LONDON AND NEW YORK
Published by Taylor & Francis 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN 52 Vanderbilt Avenue, New York, NY 10017 Transferred to Digital Printing 2006 © 1994 Taylor & Francis
Organising Committee P.G. Fookes (Chairman), Consultant lH. Atkinson, City University London R.H.G. Parry, ISSMFE S.l Wheeler, University of Oxford Sponsors Geotechnical Engineering Research Centre, City University, London International Society for Soil Mechanics and Foundation Engineering TC3 on Arid Soils The texts of the various papers in this volume were set individually by typists under the supervision of each of the authors concerned.
ISBN 90 5410 365 5
Publisher's Note The publisher has gone to great lengths to ensure the quality of this reprint but points out that some imperfections in the original may be apparent Printed and bound by CPI Antony Rowe, Eastbourne
Engineering Characteristics of Arid Soils, Fookes & Parry (eds) © 1994 Taylor & Francis, ISBN 90 5410 365 5
Table of contents
Preface
IX
1 Arid environments and descriptions of arid soils General report: Arid environments and description of arid soils A. Warren
3
Overview of geotechnical problems in Basilicata, Italy, resulting from extensive desiccation of clay strata E.N.Bromhead, M.Del Prete, H.M.Rendell & L.Coppola
5
Keynote lecture: Salt attack on buildings and other structures in arid lands A.S.Goudie
15
Corrosiveness of the arid saline soil in China Hong Naifeng
29
Playas: New ideas on hostile environments A. Millington
35
Rivers and their deposits- Cinderella's ofthe arid realm LReid
41
Sand dunes: Highly mobile and unstable surfaces A. Warren
47
2 Classification of arid soils General report: Classification of arid soils for engineering purposes J H.Atkinson
57
The FAO/UNESCO outline soil map for Africa- A possible basis for geotechnical classification AI Bashir Assallay & 1.1Smalley
65
v
Geotechnical soils mapping for construction purposes in Central Saudi Arabia JS.Griffiths, P.G.Fookes, A.D.Hardingham & R.D.Barsby
69
Technical note: Development of an engineering description of cemented arid soils and calcrete duricrusts A.D.Hardingham
87
Engineering characteristicsof arid soils Izhar-ul-Haq & Sohail Kibria
91
Classification of expansive soils in arid regions: Thermogravimetric investigation of smectite content /.Jefferson & USmalley
95
Keynote lecture:Classification of arid soils for engineering purposes An engineering approach C.D.F.Rogers, T.A.Dijkstra & USmalley
99
Classification of arid soils for engineering purposes: A pedological approach USmalley, T.A.Dijkstra & C.D.F.Rogers
135
Classification of arid soils for specific purposes USmalley, T.A.Dijkstra & C. D. F. Rogers
145
Engineering soil classification ofthe loess of Gansu, China- Based on sedimentary properties: Relationships to geotechnical behaviour Wang Jingtai & E. Derbyshire
153
3 Engineering behaviour and properties of arid soils General report: Engineering behaviour and properties of arid soils S.JWheeler
161
Keynote lecture: On the mechanical behaviour of arid soils E.E.Alonso & A.Gens
173
Deposition and the behaviour of partially saturated silt T.Amirsoleymani
207
An alternative approach to the understanding of the collapse mechanism in desert sands, loess and other collapsing soils M.E.Barton
215
Keynote lecture: The geotechnical behaviour of arid and semi-arid zone soils Southern African experience G. E. Blight
221
Collapse of soil structure JFeda
237
Mechanical behaviour of an unsaturated loam on the oedometric path JM.Fleureau & S.Taibi
241
VI
Expansive soils: TRL's research strategy C.S.Gourley, D. Newill & H. D. Schreiner
247
An investigation into the use of dune sand in concrete E.A.Kay & JP.H.Frearson
261
Keynote lecture: Some engineering behaviour and properties of arid soils A.Komornik
273
Mechanical properties of gypsum soils V. P. Petrukhin
285
A new device for the direct measurement of soil suction over a wide range A.M. Ridley & lB. Burland
289
Soil suction measurement with the transistor psychrometer !A. Woodburn
297
4 Case histories of construction and .field investigation in arid soils General report S. L. Houston
305
Drilling expansive mudstone in an undeveloped arid area A.WCook
307
A foundation code for problematic arid soils A.M.Elleboudy
309
Geotechnical characteristics of some carbonate sands in the Fezzan Sahara P.G.Fookes & JS.Gahir
313
The unique Lisan soils as engineering materials D.! Knight
331
Field wetting tests on a collapsible soil fill A. Kropp, S.Houston & D. McMahon
343
TRL research on road construction in arid areas D. Newill & M.JO'Connell
353
Experience of arid soils on the Great Man-made River Project in Libya Z.M.Nyirenda & H.R.Samuel
361
Expansive clay road embankments in arid areas: Moisture-suction conditions M.JO'Connell & C.S.Gourley
389
Collapse of an aeolian sandy clay caused by seepage R.H.G.Parry
405
Technical note on the investigation of a building failure on swelling soils A. D. Robinshaw
411
VII
The Teton Dam failure (Idaho, USA, 1970) C.D.F.Rogers, T.A.Dijkstra & lJSmalley
415
Wind-blown fine sand in sub-Saharan Northern Nigeria and its influence on dam design and performance C.JSammons
419
In-situ investigations of horizontal subsidence deformations A.Y.Shmuelyan
425
Expansive clay soils and their effects on a factory building in the Sudan N.A.Trenter
429
Problems on saline soil foundations in arid zone of China Xu Youzai
435
Author index
441
VIII
Engineering Characteristics of Arid Soils, Fookes & Parry (eds) © 1994 Taylor & Francis, ISBN 90 5410 365 5
Preface
The suggestion for a symposium exclusively devoted to arid soils stemmed from a letter to members of the ISSMFE TC3 Committee on Arid Soils by its Chairman,Dr V. P. Petrukhin of Moscow, 1992. He appealed for an offer to run a symposium at which he could get together the members of his committee, before the Thirteenth International Conference on Soil Mechanics and Foundation Engineering due to take place in New Delhi in January 1994. At this Conference the TC3 Committee was to report. The United Kingdom responded to his appeal and a small symposium committee was set up by the UK representative on the TC3 Committee, Professor P.G.Fookes. This Committee was composed of geotechnical engineers in UK who were researching or were known to have a positive interest in the subject. Fortunately, Professor J.H.Atkinson, who became the Secretary to the Symposium Committee, was able to offer the facilities of the City University, London, and the Symposium was quickly brought into being. It was organised from the late autumn of 1992 onwards and took place on 6th and 7th July 1993. The TC3 Committee met immediately following the Symposium, on 8th July. The outcome of the Arid Soils Symposium was considered most successful, judged by the quality of the contributions (all prepared, as indicated above, in a very short time) with representatives from fifteen overseas countries, giving it a truly international flavour. The Symposium was organised in four Sessions, spread over the two days. On the first day, the Environment and Classification were addressed attracting contributions from geologists and geomorphologists as well as from engineers; these were thought to give a strong background to the engineering characteristics and case histories on arid soils that formed thefinal two Sessions. The Proceedings of the Symposium have therefore been laid out in this book following the Sessions to which papers were assigned. The sessions each had a Session Reporter and their work is also reproduced here. One of the many points of interest to emerge from the Symposium was a working definition of arid soils, viz. 'Arid soils are those conditioned by an arid climate'. This allows not only soils in the current arid climates to be included in the definition, but also ancient soils deposited or formed in situ in a former arid climate or subsequently modified by an arid climate. Broadly speaking, an arid climate is one with a limited IX
rainfall, where evaporation exceeds precipitation on an annual basis. Arid climates can be cold as well as hot. There was also a strong measure of support for using the word 'ground' instead of 'soil' in the title. The front cover of the book shows the location of 'arid', 'semi arid' and 'sub tropical dry summer' climates adapted from the Times ( 1967) and Trewartha ( 1968), which give some indication of the distribution of the world's modem arid soils and is modified from a paper in the Symposium by Rogers, et al. The Editors, on behalf of the Symposium Committee, would like to thank all the delegates who attended the Symposium, those who spoke, those who wrote papers and the several organisations and individuals who worked to make the Symposium a success. They would also specifically like to thank the Sponsors of the Symposium dinner- W. S.Atkins & Partners of Epsom, Surrey; Sir Alexander Gibb & Partners of Reading, Berks; and Price Brothers (UK) Limited ofWeybridge, Surrey. The Symposium was also warmly enthusiastic for the suggestion that a second international symposium be held in four years' time. We hopethis will happen. P.G. Fookes and R. H. G. Parry Editors August 1993
X
1 Arid environments and descriptions of arid soils
Taylor & Francis
Taylor & Francis Group http:/taylorandfrancis.com
Engineering Characteristics of Arid Soils, Fookes & Parry (eds) © 1994 Taylor & Francis, ISBN 90 5410 365 5
General report: Arid environments and description of arid soils A. Warren Department of Geography, University College, London, UK
The speakers in this Section all stressed six aspects of the distinctiveness of dry environments in terms of their soil mechanics and foundation engineering properties. The importance of what they said is underlined by the fact that arid and semi· arid environments cover 37% of the terrestrial glove. Above all, these environments are dry, and this has a number of environmental and engineering implications. These include: Deep desiccation, as in the Mio·Pliocene clays of Basilicata in Italy which are dry even at 80 meters beneath the surface, and which support much steeper slope angles than they would in wetter climates. The accumulation of salts, which are problematic in concrete production and damage foundations and other structures in crystallisation. High mobility of sediment in the wind. Soil moisture and vegetation in humid climates inhibit the wind, but it has much freer reign in dry environments. This in turn means sand and dune movement, high levels of dust, which may carry damaging materials, like salts. The removal of sediment by the wind can also be a problem. Despite aridity however, there are still problems with moisture, in restricted areas; indeed these have the added dimension that moisture may bring very marked changes to thoroughly desiccated materials. There may also still be problems with algal and lichen weathering, for these can withstand desiccation for long periods and then revive when it occurs. Second, dry environments suffer high degrees of variability. Even if the absolute degree of variability may not be any greater in arid climates (though it commonly is), variability occurs across very critical moisture thresholds, which bring about in turn distinct changes of phase for many materials. Third, in some spheres, arid landforms are very active indeed. This is particularly so for wind-formed environments. Dunes, even several meters high may advance at meters per year. Sand discharges can reach more than 150m3 (m Width)-2 yr·l. In some 3
circumstances, even fluvial landforms can experience very sudden major change, as on alluvial fans and in river channels. Fourth, in marked contrast, many other arid landforms and soils are so inactive that they have retained many of their characteristics from the past. Inheritance may be an even more significant feature of arid conditions because very low activity rates permit inherited features to survive for longer than they do in other conditions. Fifth, for vulnerable magnified feature to
all these reasons, arid landforms and soils are very to disturbance, for disturbance can be immediately by active processes, or may expose an inherited readjustment to its new environment.
Finally, arid environments, despite the visibility of their landforms and soils, and the apparent simplicity of the controls on their formation, show a high degree of complexity, and this has yet to ·be unravelled in the great majority of cases. Even simple landforms like sand dunes are far from being understood in their entirety. Complex landforms on alluvial fans and terraces are much further from understanding. The interplay of fluvial and aeolian forms, and inheritance adds to complexity. Furthermore, much less effort is expended on research into landforms and soils in these areas than in humid areas. There is clearly still a large amount of work to be done, much of it with very practical significance. New technology, as in the use of remote sensing, new instruments and modelling techniques, is often particularly appropriate to research in these environments, so that research, though still restricted, is now much more effective.
4
Engineering Characteristics of Arid Soils, Fookes & Parry (eds) © 1994 Taylor & Francis, ISBN 90 5410 365 5
Overview of geotechnical problems in Basilicata, Italy, resulting from extensive desiccation of clay strata E.N.Bromhead Kingston University, London, UK
M.DelPrete Universitii di Basilicata, Potenza, Italy
H. M. Rendell Sussex University, Brighton, UK
L.Coppola Universitii degli Studi di Sienna, Italy
ABSTRACT. The Plio-Pleistocene clay area in Basilicata is characterized by high relative relief - the product of relatively rapid uplift during the last 800,000 years, together with accompanying incision. This clay area is approximately 195,700 ha. Mass movements tend to be polarized between the relatively shallow and the more deep-seated. The latter are controlled by aspects of the structural geology. Many of the larger deep seated slides are overlain by permeable sands and gravels, and may date back to pluvial periods in the Quaternary. The behaviour of the clays is controlled by the climate, which is strongly desiccating, with an annual soil moisture deficit in excess of 600mm. Rainfall is highly variable, both in time and space, with annual values ranging from 350mm to more than lOOOmm. Infrequent heavy rainfall combined with the generally desiccated nature of the clays cause badlands type erosion, with the appearance of geomorphological features termed calanchi and biancane. The landforms are all extremely steep relative to the geotechnical properties of the soils in which they are formed. The paper gives an overview of the geographical extent and the geomorphological character of the area, and is illustrated with examples of geohazards ranging from shallow and deep-seated slides, clay swelling and ground heave, and rapid erosion. INTRODUCTION This paper is intended to give an overview of the nature of the arid climate desiccated clay soils of Basilicata, and the geotechnical conundrums they pose, especially to the geotechnical engineer with a Northern European perspective. The paper reflects the extensive experience of both the Italian and UK collaborators. An interesting area geotechnically, the South of Italy has been relatively neglected until recently. Lack of development led to many of the historic medieval hilltop settlements becoming depopulated. There are few, if any, industries outside agriculture. However, the Naples-Avellino earthquake of November 1980, and the need for extensive repair and rebuilding in the damaged areas, led the authorities to realize something of the extraordinary historic and artistic patrimony of the area. In turn, this demanded the sympathetic reconstruction of the area, and the injection
5
of funds to create new industries to halt or reverse the depopulation trends. Naturally, planning for reconstruction has led in many cases to extensive geotechnical investigations. Many of the results quoted below have arisen from such investigations. GEOGRAPHICAL SETTING Mountainous and hilly terrains dominate Italian land areas, of which only 23% might be classified as plains. In the south, the relative extent of plains is even smaller, for example in Basilicata and Calabria, where they occupy only 8-9% of the land area. Fig. 1 divides southern Italy into two zones on a crude climatic basis: the sub-humid zone (unshaded) which typically has annual soil moisture deficits of 400mm and the semi-arid
N
1' 0
;-·:':···:';Moisture deficit !".::.-.::."!greater than 600mm
km
ll!!!l!!!lm Plio-Pleistocene clay ISSS88II8 hills lopes - Baslllcata
tOO
Fig. 1 Location Map, showing extent of Plio-Pleistocene Clay area 6
zones (shown shaded) with moisture deficits of typically 600mm. Also shown on the Figure is the extent of the outcrop of Plio-Pleistocene clays which outcrop over 19.5% of the land area of the Basilicata Region, largely in the semi arid zone. In the Province of Matera, outcrops of these clays cover an area of approximately 195,700 ha, or 35% of the land area. The relative relief of up to approximately BOOm is developed in these soils, and it should be unsurprising that since the rainfall is both temporally and spatially variable, erosion can be intense and leads to the development of a characteristic land surface morphology. GEOTECHNICAL PROPERTIES The range of geotechnical properties of the clay soils measured over a number of studies is shown in Table 1 (Consistency Limits) and in a more restricted sense for Shear Strength parameters in Table 2. They are typical of medium to high plasticity clays with peak angles of shearing resistance upwards of about 20", and with a residual angle of shearing resistance between 10 and 20 ·, typical of illite-kaolinite clays (i.e. not exhibiting the sometimes spectacularly low residual angles of shearing resistance found in montmorillonite clays e.g. the scaley clays).
Fig. 2
Landslide damage at Craco
7
Table 1
Typical Consistency Limits
Site
Montescaglioso Miglionico Pomarico Timmari Ferrandina Grassano Monte Finese Montalbano Pisticci
Table 2
Blue Clays (Argille Azzurre)
No. of samples
LL PL IP (mean values
34 21 40 25 5 6 26 45 41
41 52 50 58 49 50 47 47 44
20 24 23 26 21 23 21 22 20
Strength Characteristics
20 27 27 32 28 27 27 24 24
< 2 microns (% by weight) 34 44 41 46 56 46 37 30 32
Blue Clays (Argille Azzurre)
Site
No. of Samples
c' $' (peak) kPa
Montescaglioso Montescagioso Miglionico Pomarico Timmari Ferrandina Ferrandina Grassano Montalbano Pisticci Pisticci
1 6 3 1 4 3 4 6 12 36 5
6.9
tx =triaxial test(s), ds
%)
32.1 22.8
c' $' (residual) kPa 0
22.1 17.9 14.0 16.3
17.7 56.0 20.0 33.3 22.3 22.8 23.5 25.2 48.022.5
0 0
10.0 13.3
2.0
15.3
mode
ds ds ds ds ds tx ds ds ds ds tx
direct shear test(s).
LANDSLIDES Table 3 gives data on 130 shallow landslides observed by Rendell in a survey in the years immediately before 1976. This data set shows quite clearly the shallow nature of the landslides, and the steepness of the terrain on which they have developed. Taking this data in conjunction with the soil shear strength parameters (Table 2) it is readily seen that all but the shallowest slopes are steeper than all but the strongest angles of shearing resistance. Application of the infinite slope theory (Haefeli, 1948: Skempton & Delory, 1957) brings one readily to the conclusion that only strong soil suctions keep the slides in place. The loss of suction during periods of rainfall or snow-melt permits the slide to reactivate, only to be re arrested when suctions are restored.
8
Table 3 : Data on 130 shallow landslides in blue clays in the area between Grassano Scale and Ferrandino Scale in the Middle Basento Valley :
Length (m) Width (m) Depth (m) Slope (deg) D/L ratio
mean
st. dev.
max.
min
13.08 5.15 0.67 34.5 0.058
7.78 3.65 0.49 5.0 0.030
51.4 27.5 3.88 58.0 0.194
2.3 1.4 0.14 25.0 O.Oll
Note : depth measured normal to slope. Source: Rendell, 1976 (unpublished PhD thesis.) Where the landforms dictate, accumulations of mudslide debris may form and develope into mudslide chutes. Six examples were surveyed by Alexander, 1982. These had slope angles between 22.1 · (min) and 53.3 · (max) with a mean slope angle of 40. 9 ·. Again, the terrain slopes are significantly greater than$' for the materials involved in sliding. Whereas the shallow landslides are commonplace, and prove troublesome to agriculture, road and rail links etc, the deeper landslides are rather less common. Many of them are overlain by permeable sands and gravels, and may date back to pluvial periods in the Quaternary. Hillslopes with this sort of capping are favoured as sites for settlements, and the activation or reactivation of landsliding, either due to rainfall, toe erosion, or seismic activity, can be a serious problem. Fig. 2. shows the effects of such a landslide at Crace. GROUND HEAVE (CLAY SWELLING) As might be expected with clays of medium to high plasticity presently in equilibrium with a climate which is strongly desiccating, the clays of Basilicata have a strong tendency to swell when presented with an appropriate source of free water. A series of brick-built arch bridge structures on the road between Irsina and Grassano Scala have all suffered severe dislocation as a result of ground heave, causing a rise at the crown of each arch, buckling of the parapet wall, and fracturing of both the structural brickwork and the road surfacing. In these cases the damage is severe enough for the structures to be abandoned. The source of the water is thought to be run-off trapped inside the abutments. Modern structures may have extensive post-tensioned ground-anchor systems to control ground movement. Elsewhere, the effects of ground heave resulting from leaky water supplies or soakaway drainage have been seen. Modern house construction takes the form of reinforced concrete frames with clay block or brick infill, a mode of construction which is intrinsically less susceptible to the problems of settlement or clay heave than the brick I block load bearing walls
9
adopted for dwellings in the UK. The modern indigenous construction in S. Italy also realistically copes with the earthquake risk. CALANCHI & BIANCANE Two distinct types of badland landform may be identified in the clay areas. The term 'calanchi' is applied to the type of knife-edge erosional features (see Fig. 3) that occur on 40-60 · hillslopes. The steep side slopes within calanchi are unvegetated and intensely dissected, often exhibiting 'inverted' drainage networks. 'Biancane' (Fig. 4) are erosional landforms which form typically hummocky terrain. Individual biancana may be up to 4m high. It is possible that biancane are the next stage after calanchi in landform development. Areas of calanchi are found in zones of Miocene and Pliocene 'scaley clays' (argille scaglioso) but they are most spectacularly developed in the Plio Pleistocene clay areas. Rates of erosion in this terrain are dramatic. Sediment loads in rivers
Fig. 3 Example of 'calanchi'
10
Table 4. Slope data for calanchi and biancane
Mean angle ( ·)
Max angle ( ·)
Min angle ( ·)
Standard deviatn ( ·)
No of observns
Calanchi interfluves
37.7
51.0
16.0
6.7
108
Calanchi flow lines
36.5
54.0
21.0
6.1
57
Calanchi minor flow lines
41.6
55.0
28.0
6.2
32
Biancane N-facing
33.4
63.0
7.0
13.3
125
Biancane S-f acing
35.7
68.0
1.0
13.1
198
Biancane E-facing
35.5
70.0
1.0
14.2
184
Biancane W-facing
34.3
62.0
2.0
13.5
133
Fig. 4.
Example of
11
'Biancane'
can be used to form the basis of estimates of 1000 to 2500 tonnes I square km per annum, but individual hillslopes may be subject to intensities far in excess of this, rising to the order of 28000 tonnes per square km per annum. (Rendell, 1986) . Both the calanchi and biancane type badlands terrain are sometimes treated by simply bulldozing the landforms into a new slope forms suitable for growing wheat. This practice is more common in other areas of Italy where yield from agriculture are perhaps rather better. It is not known how effective this technique is in the long term. Other erosion control measures include check dams, afforestation and contour ploughing. Check dams filled with sediment may then suffer piping and be undermined - an example of this was seen in the Basento valley. These problems are not entirely confined to southern Italy. Indeed, the largest area of calanchi is in the north, in the Region of Emilia-Romagna. PIPING Flow down desiccation cracks accompanied by migration of wetting fronts parallel to the crack faces, swelling and slaking, developes pipes in the essentially dispersive (Rendell, 1986) inorganic clays. A surveyed cross section through a pipe is shown in Fig. 5, and its general appearance in the field is shown in Fig. 6.
Surveyed March 1979 Scale in metres
Fig. 5. Survey of a pipe developed in Plio-Pleistocene clays
12
Fig. 6.
Appearance of a pipe in the field
REFERENCES Alexander, D. (1982) . Difference between 'calanchi' and 'biancane' badlands in Italy. In Bryan, R. and Yair, A. Badland Geomorphology and piping, 71-87. Haefeli, R. (1948) . The stability of slopes acted on by parallel seepage. Proc. 2nd Int . Conf. on Soil Mechs & Fndn Engng, Rotterdam, 57-62. Rendell, H. M. (1976). Clay hillsope erosion in a semi-arid environment: the Basento valley southern Italy. Unpublished PhD Thesis, University of London.
13
Rendell, H. M. (1986). Soil erosion and land degradation in Southern Italy, Desrtification in Europe, eds Fantechi, R. and Margaris, N. S., Commission European Communities. Skempton, A. W. & Delory, F. A. (1957). Stability of natural slopes in London Clay. Proc. 4th Int. Conf. on Soil Mechs & Fndn Engng, London, 2, 378-381.
14
Engineering Characteristics of Arid Soils, Fookes & Parry (eds) © 1994 Taylor & Francis, ISBN 90 5410 365 5
Keynote lecture: Salt attack on buildings and other structures in arid lands A.S.Goudie University of Oxford, UK
ABSTRACf The weathering of building materials and structures by salt is a major problem for engineers in arid lands. The weathering results from chemical and physical (hydration, crystallisation and thermal expansion) processes. Sodium carbonate and sodium sulphate are especially aggressive, particularly where surface temperatures cycle across 32•c and where high nocturnal atmospheric humidity levels occur.
IN1RODUCTION The importance of salt as a cause of rock disintegration has been known since the time of Herodotus, and even in the nineteenth century numerous field and laboratory observations were made of its power. However, it was in the 1950s and 1960s that there was a re awakening of interest in this theme and the work of various French geomorphologists in the field and in the laboratory greatly extended available knowledge. Tricart (1960), for example, was concerned about the reasons for the decay of breakwaters in west Africa. Interest was further awakened by two notable papers that explored some of the theory behind salt weathering (Wellman and Wilson. 1965; and Evans, 1970). As Goudie (1985) has explained it has now become clear that salt weathering, whatever the mechanism or mechanisms involved, is a highly effective cause of rock and building material decay in a whole range of environments. The literature has more recently been summarily reviewed by Doornkamp and Ibrahim (1990) and some useful data on rates of salt decay provided from Israel by Amit et al (1993).
FIELD EVIDENCE FOR THE POWER OF SALT WEATHERING (a)
Introduction
Field work over a period of years has given various examples of the speed with which salt weathering can operate. In some cases the power of salt is indicated by the spatial propinquity of salt efflorescences and features produced by weathering (e.g Death Valley and the Rift Valley lakes of East Africa), in other cases it is demonstrated by the breakdown of geomorphological features of recent age (e.g. Karakorams), in yetother cases man-made objects of known age have broken down (e.g. the salt pans of South
15
Africa), while in two cases experiments have been conducted in the field to monitor the speed of change (e.g. Bahrain and southern Tunisia). (b)
Death Valley, California
Death Valley is a major fault trough enclosed by mountains. From these mountains large fans of debris are generated, reaching the salty zones of the playa, notably near Badwater (-86 m below sea-level). The fans are composed primarily of large boulders and cobbles made up of a wide range of lithologies, including Pre-Cambrian crystalline basement rocks, such as gneiss, and older Tertiary volcanic rocks, especially rhyolite. When these clasts reach the saline zone (Goudie and Day, 1980) they undergo rapid splitting and diminution insize. Near parallel changes in salinity (expressed by the conductivity of the soil paste), extent of splitting and size of particle with distance from the flat surface of the salt pan. occurred. When some of the cobbles were placed in water, disintegration took place as a result of the leaching of the salt which impregnated them and held them together, and the debris thus formed was found to contain appreciable quantities of silt sized material. It is clear that in Death Valley, and in the neighbouring Panamint Valley, size diminution of fan debris in the lower parts of the fans, is brought about by salt weathering as much as by attrition, a finding which confirms the observations of Beaumont (1968) on the andesiteand tuff pebbles entering the saline environment of the Great Kavir in Iran. (c)
The salt pans of South Africa
Large parts of the interior of southern Africa are characterised by the presence of closed depressions, called pans. Some of these pans are exploited for commercial salt production, andthis necessitates the construction of walls and other structures on the salt surfaces of the pans themselves. In particular the evaporation basins are constructed of brick, rock or concrete. Likewise, tracks across the pans have a surface of hard core. Observations at two of the major salt producing pans- Hayfield and Florisbad- showed that all such structures were severely disintegrated, and were thus being continually replaced by the salt companies. The effective salts are either sodium chloride and/or sodium sulphate. The importance of such palpably rapid salt action in helping to create pans was noted by Du Toit (1906, p. 257), while de Bruiyn (1971, p.123) has explained the retarded development of pans on the Beaufort Sandstones compared to the Ecca and Dwyka Shales by the observations that "Salts are relatively scarce in the Beaufort sediments due to their original environment in shallow fresh-water swamps", whereas weathering of exposed Dwyka and Ecca shales releases considerable quantities of highly effective sodium sulphate (Hugo, 1974). (d)
The Rift Valley lakes of East Africa
The closed rift valley lakes of East Africa are largely characterised by brines which contain a substantial component of sodium carbonate. Wherever bedrock exposures have been examined in close proximity to such lakesclear signs of weathering have been identified. At Lake Stefanie (Chew Bahir) in southern Ethiopia a transition can be witnessed as one 16
moves towards the ~ floor from well rounded volcanic masses, to mushroom-shaped pedestals, to planed off structures. By Lake Magadi in Kenya veins and blisters of trona (sodium carbonate) appear to be disrupting lavas, and stone-floored tracks across the playa show signs of substantial decay of their surfaces as a result of aggregate disintegration (Smith and McAllister, 1986). Exposed floors around the Galla lakes to the south of Addis Ababa show a whole array of micro-weathering forms including ~aromas. (e)
The Karakoram Mountains, Pakistan
The Karakoram Mountains in Pakistan are cut into a series of deep valleys which are tributaries to the Indus. These valleys have a low precipitation level (w direction
~
'
KEY
Bajada covered pediment with downslope. Pediment developed in
increasing alluvial thickness
It became apparent from the initial reconnaissance of the site that bedrock was relatively close
to the ground surface and that the area, rather than being a depositional environment, was primarily an eroding landscape. This allowed a relatively simple three fold framework of subsurface conditions to be established: a) b) c)
bedrock expected close or at the ground surface bedrock expected between the ground surface and 2.0 metres bedrock expected to be deeper than 2.0 metres below the ground surface.
The superficial deposits that overlay bedrock were divided into Recent (ie. Holocene) or Quaternary alluvium. Other geomorphological surface features that were included on the map were escarpments, duricrust development and areas of lag gravel. The map was extensively annotated with descriptions of the materials, together with estimates of excavatability, borrow potential, estimated foundations and dust hazard. In most instances boundaries between areas were conjectural and often gradational. The map was, therefore, only applicable to feasibility studies and preliminary design. The 1:25,000 map was found to be particularly valuable in the planning of a later borrow materials survey of the whole site. Figure 2 presents part of the final map as an example of the output produced. 1:10,000 Scale Thematic Maps.
The aim of the 1:10,000 field mapping was to provide the data required for designing and interpreting the subsequent detailed site investigation, as well as giving preliminary information about the extant ground conditions and the occurrence of natural hazards to construction. In the field, maps showing the distribution of surface materials were compiled using the classification system presented in Table 3, to produce an initial superficial materials map. The mapping was carried out over a four week period by a two man team driving transects between survey markers and using the vehicle oedometer to fiX the position of changes in the surface materials. As with all the thematic maps produced during the mapping programme, the boundaries of the various categories shown on the superficial materials map were gradational and simplified and, therefore, could be considered as only being accurate to the scale at which the maps were reproduced (i.e. 1:10,000).
TABLE 3 -
KEY TO SURFACE MATERIALS MAPS
Gc
Coarse
Coarse gravel to cobble size
Gm
Medium
Medium gravel size
Gf
Fine
Sand and fine gravel size
Gm/Gf
Alternating, interdigitating combination of Gm & Gf
NOTES:
1. 50:50 Combinations are indicated by for example Rrn/Gm. 2. Where one matter dominates buttherels a substentllll quanuty or the subslduary matenals. This is described as Rm subslduary Gm.
75
~ al
NOTES:
...
Moo. rate Moderate
Poor Poor
3
4
FAJR.Y FAIR...Y cot.t.1ON COr.t.IION
COMvlON COr.t.IION
00Ivf0M)N COr.t.IION
FAJR.Y FAIR...Y CO""-'ON COr.t.IION
FAIRLY FAIRLY fW\E RARE
TypiCilI FroQJII'1CY'
tOO 800
T)1)lCilI ,A/lowatJe 8ear-ng C8pad1y tn'allan
Dominant clay component
Turbid waters of strong yellow-brown to red-brown color
Montmorillomtes. illite plus sml salinit)
Clear waters
Calcium, magnesium or iron-ncb soil, highly acid soil, sands
Clear waters w1th a bluish cast
Non-saline kaolins
Erosion gullies and/or
Saline clays, usually montmorillonites
field tunnelling in the natural soil
As above, mild
Kaolinites
Landslips
Kaolinites. chlorites
Surface microrelief (gilgai)
Montmorillonites
Country rock type granitic
Kaolinites, micas
Country rock type basaJtic. poorly drained topography
Montmorillomtes
Country rock type basaJtic, well drained topography
Kaolinites
Country rock type sandstones
Kaoliniuc
Country rock type mudstones and shales
Montmorillonites or illite. often soil salinity
Country rock type limestone
Alkaline montmorillonltes and chlorites of very variable propenies
Country rock type recent pyroclastiCS
AUopbanes
Kaolinites
Mottled clays. yellow-orange-grey mottle
Montmorillonites
Medium to dark gray and black clays
Montmorillonites
Brown and red-brown clays
Appreciable illite, some montmorillonite
White and light gray clays
Kaolinites and bauxites
Discrete microparticles of high light reflectance (micas)
M1C8Ceous soils
Discrete microcrystals, easily crushed
Gypsum-rich soils, or (rarer) zeolites
Soft nodules, acid-soluble, disseminated
Carbonates
Hard nodules. red-brown
Ironstones. laterite
Extensive cracking. wide. deep and closely spaced at S to 6 ems or less
Calcium-rich illites and montmoriUonites
Up to intervals of 30 ems
Illites
iv)
Open-textured friable loamy soils with appreciable clay content, black
Organic soils. peats
Open-textured friable loamy soils of low clay
Carbonate, silts and sands
Wormy appearance on
Montmorillonites. plus sod salinny
Relatively thin, strongly bleached horizon near the soil surface (up to 60 ems from the top)
Above the bleach, fine silt: below the bleach, dispersive clay. Probably a seasonal perched water table at the bleach level
exposed pre-existins weathered profile
Texture Oburvation
/tltJt~rial B~htlvior
Sandy soils-hish sand content
Good engineering properties. but if monosized. will need mechanical stabilization, and if whoUy sand, may need some clay. cement. or bituminous binder
Silty soils-hip silt
Good when dry, but will lose all bearina capacily and tnflicability when wet; no economic treatment except good water shedding
Silty soils-high silt content. little or no clay
Dusty and cohesionless when dry, no bearing capacity when wet; no economic treatment: avoid
Clay soils-high clay content
Very troublesome eDJineering propenies unless protected from moisture c:Jlanae. Montmorillonites and illites responsive to lime stabilization
Clay soils-hiah clay
Reasonably good engineerina propenies when heavily compacted
Clay soils-clay softens rapidly with water, clay areasiness develops rapidly
Saline clays, poor trafficability when wet
content, some clay
Dominant clay component
Mottled clays. red-orange-white mottle
and more
Usually associated with carbonate, allophane or kaoline. but n~v~r montmorillonite and seldom illite
content
Profile ObservatiOn
Open-textured friable loamy soils with appreciable clay content
content (either non-crackina or open-textured)
Clay soils-clay softens only slowly with water. greasmess develops slowly
Calcium·rich mon1morillonites of poor trafficability and high adhesiveness (sticky) when wet
Particle shape and roughness These parameters are perhaps only relevant for particles of silt size or greater and can be 120
determined by observation, either with or without magnification, or indirectly by strength tests. This is a further case in which the use of standard descriptions or reference to standard diagrams would help reduce subjectivity. v)
Particle orientation, packing structure and cementation
and
fabric,
micro
These parameters are generally assessed only by visual appearance, again either with or without magnification. While these parameters are of importance in traditional northern hemisphere soils, they are of greater importance still to arid soils since they will often dominate behaviour. A standard description of structure is provided, for example, by BSI (1981) and, while this will provide no more than a basis, it does permit an expansion to cover the structures likely to be encountered with arid soils. The part of the soil description table that relates to structure is reproduced in Table 3. Aspects such as the strength of cementation and the volume change potential of the structure are dealt with later. vi) Particle Specific Gravity This can be measured relatively simply in the laboratory and, in any case, is unlikely to vary across a wide range. For this reason values, typically of 2.65 or 2.70, can often be assumed for preliminary design and analysis purposes. For detailed analysis, however, the true value should be measured since calculations of other parameters can be significantly affected by the resulting value. For details of laboratory test methods, both for this parameter and other standard test parameters, reference can be made to BSI (1991). vii) Joints and Fissures These parameters, while distinct in some senses from those reported in subsection (v) above, can be considered together with them using the same method. These two groups of parameters are described in BSI (1981) as mass characteristics, as opposed to material characteristics, and for description require large samples or, where possible, complete exposed faces. This is more relevant within this category than the previous category, since discontinuity spacing can be large and yet still important in terms of soil properties. This completes the discussion of first order properties and tests therefor. It is apparent that methods can be adapted and/or derived for each of these parameters in respect of arid soils. The second order properties of void ratio, bulk density, and water content are straightforward to 121
Table 3.
Descriptions of soils (after BS5930: 1981, and Atkinson 1993, p.48). Structure Term
Field identification
Interval scales
Homo
Deposit consesu essenteally of one type.
Scale of bedding sPacint
geneous
Term
Mean
Inter
Alternating layers of vary
sPacing.
stratified
ing types or with bands
mm
Hetero· geneous
or lenses of other materials. Interval scale for bedding sPacing may be used.
Verv thickly bedded
over 2000
A mixture of types.
Thickly bedded
2000 to 600
Medium bedded
600 to 200
Thinly bedded
200 to 60
Very thinly
60 to 20
Weathered j Particles may be
weake~ed
and may show concentnc layering.
bedded Thickly l~minatod
Thinly laminated
Fissured
20 to 6 under 6
Break into polyhedral
fragments along fissures. Interval scale for spacing of discontinuities may be used. Intact
No fissures.
Homo geneous
Deposit consists essentially of one type.
Inter
Alternating lavers of varying types. Interval scale for thickness of layers may be used.
strat•fied
I
Scale of spacing of other discontinuities
Term
Mean spacing. mm
Weathered Usually has crumb or columnar structure.
Fibrous
Plant remains recogmzable and retain some strength.
Am or· phous
Recognizable plant remains absent.
122
Verv widely spaced
over 2000
Widely spaced
200010600
Medium spaced
600to 200
Closely spaced
200to60
Verv closely spaced
60 to 20
Extremely closely spaced
under 20
measure and are amply covered by many existing texts, Discussion is thus not warranted herein. Plasticity and consistency limits are equally well reported and tests for these parameters, essentially via Atterberg limits and water content, are well defined. The amount of clay mineral (s) in arid soils can often be small and thus collection of enough material for these tests is generally a problem. If this is the case, other indications of mineralogy need to be sought. The properties that distinguish arid soils are typically compressibility and compactibility (including relative density), stability (collapsibility or swelling potential), and perhaps permeability, and as a result shear strength (including frictional, cohesive and cementing properties). These properties will be examined together herein. The strength and durability of the soil structure, including any influences of the fabric, are commonly the controlling influences on arid soil behaviour and should be assessed in some way. The soil response to both load and water are thus important in this case. What is required, therefore, appears to be a programme of empirical laboratory and field tests that load (both statically and dynamically) a sample to different levels in both the natural and the fully saturated states. In the laboratory the tests could be carried out in a number of different ways. If the sample has no cemented, or intact, structure then the soil can undergo material tests without recourse to specific tests for classification purposes. If the sample retains its structure, then the following programme is suggested. Since there are a number of tests to be conducted in different states, it is suggested that six intact samples of the soil, measuring say 50mm in diameter and 75mm in length, be created. This would permit repetition of suspect tests and/or tests that demonstrate critical results. i)
Subjection to static load The carefully weighed and measured, square-ended sample is confined in a 50mm diameter sample container with a rigid base and a load platen of slightly smaller diameter is placed onto the top of the sample. Loads of progressively increasing magnitude can then be added to the load platen, via a load hanger, and the deflection of the load platen from its original position can be monitored over a set period. The loads should be doubled in each case and loading should continue until relatively high stresses, in civil engineering terms, have been reached. From the data recorded, and supplementary information gained in above tests such as specific gravity of soil particles and water content, information on density, voids ratio, compressibility and compactibility under static load would be gained. 123
The process could be repeated without confinement to determine an unconfined compressive strength, although a steadily increasing load rather than incremental load should clearly be used. This would provide a measure of the strength of cementation, and suction (or cohesion) effects in a partially drained or fully drained condition, depending upon the rate of loading and the characteristics of the sample. A series of three, confined triaxial samples could be tested under increasing cell pressure if a better indication of the soil's strength characteristics was required, and samples could be more appropriately dimensioned (with an aspect ratio of 2, rather than 1.5). This latter suggestion deviates, however, from the simplicity of the suggested test programme. ii)
Subjection to dynamic load It is proposed that a similar sample and sample holder be used, but that a dynamic load be applied to the top platen by a falling mass. This idea is identical in principle to the Moisture Condition Value (MCV) test, the only differences here being the sample aspect ratio and reduction of sample size from 100mm diameter to 50mm diameter. A miniature MCV test has been propounded by Nogami and Villibor (1985) for the classification of tropical soils. Such a test has many attractions. It can be used to provide an MCV classification by being carried out at a standard drop height, although this is perhaps of least use. A test of greater value would be one in which a specified number of blows (say 10) were applied from a small drop height, measuring the settlement of the top platen after each, followed by a repetition of the process at three further, progressively increasing drop heights. The final drop height should impart a relatively high dynamic impact stress to the sample. The test could be criticised in terms of realism of compaction characteristics, perhaps, since the kneading action of the standard compaction tests is missing. However in the context of simplicity, repeatability and the types of materials tested this criticism can be discounted. The information can thus be used to determine compactibility, and, in association with additional tests and other measurements, relative density. (A caveat could be added that if platen settlement occurred with the final three blows at the greatest drop height, further blows should be applied until no settlement occurs over three consecutive blows.) It might also provide a loose indication of the soil response to earthquake loading, although a better test here would be to provide a vibration to the sample, perhaps by tapping with a hammer, under static load.
124
iii) Subjection to flooding A similarly prepared and measured sample could be placed into a similar mould, though with holes drilled into the base to permit water passage and moist filter paper placed over it. The sample could then be flooded with the top {perforated, filter paper covered) platen in place and observations of volume change with time made. The process could then be repeated with progressively increasing static loads applied to the top platen as above. Clearly evidence of water penetration through the sample would be necessary {an appropriate time must be allowed) and the whole sample arrangement could be submerged, by for example filling a water bath once the sample container was assembled and sample was under load, for simplicity. The ideas can be considered as parallel with those of a soaked Californian Bearing Ratio {CBR) test, although with very different aims and procedures. Sample expansion would be of interest, and value in classification, and contraction would yield information on the potential for structural collapse and hydroconsolidation. The tests under different static loads could warrant fresh samples prior to flooding, although a multi-stage test should be used for the first sample. These tests together would provide a relatively simple and rapid means of determining the effects of structure on soil properties, and have the considerable benefit of being equally applicable to, for example, the temperate quickclays and tropical soils. The samples required for this testing should be free from structural defects, such as fissures and other discontinuities, unless they occur regularly and sufficiently frequently to demand inclusion within the sample. The influence of such structural features would have to be determined from descriptions and engineering judgement in the general , i.e. untested case, as would be the case now for soil and rock testing. The tests can readily be transferred to the field by suggesting loading by simple equipment or hand. For the static load tests, hand pressure can be applied at, say, three levels to unconfined samples of approximately the correct size, or to confined samples trimmed to the correct side, depending on sophistication. A level of twisting could be incorporated to define shear resistance in the hand. The submerged tests could similarly be carried out by hand in a bucket of water. This would complete the test programme and add sufficient data to enable problem soils, such as arid soils, to be adequately classified. While the detail needs to be debated, it is believed by the authors that these tests would provide a sufficiently complete picture of mass behaviour for others to use in engineering interpretation.
125
'LAND SYSTEMS' AND RELATED CLASSIFICATIONS It is not intended that this topic be discussed at depth herein, but certain comments are warranted. This is because there is a discipline developing between soil mechanics and geomorphology which must be relevant to arid regions (see Cooke and Doornkamp 1974, and later editions). A scheme for classifying the earth's surface (and engineering soils) based on an integrated assessment of relief form, materials, and processes is of value in environmental management and engineering construction (Cooke and Doornkamp 19 7 4, p. 3 26) . For this purpose a formalised scheme known as the 'land systems' approach already exists, and was used in Australia (Aitchison and Grant 1967) over 25 years ago. The 'land systems' mapping procedure involves subdividing the country into areas that have within them common physical attributes that are different from those of adjacent areas. Land systems may range in size from only tens of km2 up to some hundreds of km2. Within any one land system there is usually a recurring pattern of topography, soils, and vegetation (Christian and Stewart 1952). 'Land systems' mapping was developed by CSIRO in Australia as a rapid method of reconnaissance survey of otherwise unmapped or poorly mapped areas (Christian 1957, Mabbutt and Stewart 1963) . The primary aim was to establish a classification of the suitability of land for agricultural purposes, but this can be extended to engineering purposes. Note, however, that the ideas here relate to the strictly map-now-use-later (MNUL) approach used by CSIRO, a government agency. The Land system ideas (and related approaches) are interesting, but may be of little consequence to the engineering industry. Further details of geotechnical classification systems are given by Brand (1985), who discusses several forms of land use mapping in relation to tropical soils, and this should be referred to for development of ideas in relation to arid soils. A GENERAL CLASSIFICATION FOR SOIL The fundamental question that arises here is can engineering soil classification be compared with other soil classifications or with classification schemes in general? The question arises because there is a tradition in civil engineering of using words in unique ways, and thus in ways that do not relate to general usage. At the most basic level a problem must arise with the word 'soil': engineers classify soil, and pedologists classify soil. But the material that engineers classify is by no means the same as that which the pedologists classify. Notwithstanding this fact, it may be that some of the ideas and approaches used by the pedologists are useful and relevant in soil engineering, but care will have to be taken to redraft them 126
to suit the engineering version of 'soil'. One might consider the slow introduction of a linguistic revolution in which engineers change the terms in general usage. If this were the case then the first change would be soil to ground and the second would be to clarify the distinction between classification and description. In this respect engineers have created the problem for themselves: the rest of the world uses soil in a pedological sense, but engineers use it to mean ground. There are no problems in other languages: Dutch - grond, German - grund, Russian grunt Who was it who translated 'Erdbaumechanik' as soil mechanics? Whoever it was certainly did a disservice to those who struggle with the idea of soil (ground) classification for engineering purposes. 'Classification' also causes problems. In their study ot the classification of laterite soils, Queiroz de Carvalho et al (1985) quoted an observation by Kubiena: "Show me your (classification) system and I will tell you how far you have come in the perceptions of your research problems". That this quotation should appear in an ISSMFE publication is surprising (and encouraging), and suggests that a wider view is being taken. A classification, in normal speech, represents a way of ordering a subject, or a collection of things and observations. Between the classes in a classification there exists a relationship which might, in fact, emerge as the classification is produced. The periodic classification of the chemical elements cited earlier is a case in point. The idea revolutionised chemistry because the relationships between the elements became apparent: the similarities in the same group, and the change in nature in changing from group to group. there is a long way to go, and a large gulf to bridge, before a unified soil classification, cutting across engineering and science, could be even attempted and it is unclear, as yet, whether such a course is even desirable. CONCLUSIONS It has been demonstrated that the currently used systems of classification are inadequate for certain soils, among which lie arid soils, largely due to the influences of structure, mineralogy and cementation. Suggestions have been made on how the currently most popular systems can be adapted to take account of these influences, this being preferable to the attempted creation of a new classification system altogether. It has been noted that any adjustment to the classification systems should not be aimed at one set of soils, such as arid soils, but at all problem soils and thus the proposed additional tests have been composed to take this into consideration. The new tests consist of soil sample response to static load, dynamic load and flooding. The ideas of linkage with scientific classification and the production of 'land systems' classification have been introduced, but it was concluded that the value of this was 127
(X)
N
~
r--
10
9
8
20
30
40
0
VFI Sih
10
VF2 Loamy Sill
\
\ \
- -l..\
20
VF3 Silly Loam
\
M5
I
I
C5,
7
\
\
\
\
\
\
30
Loam
VF4
ISilly Clay
F5
Clay
Clay(%)
50
1
60
:;;~
I
Silly Clay
~
40
~
\
~?
f4 Clay Loam
Loa~=r---
80 70 I- F l - f-~ - F 3 60 Fine- Loamy t-Sand- Sand 50
90
100
r---
\
M4 -Sandy Clay-
\
"
70
80
90
Figure 7. An attempt to portray soil texture on a continuous representa tional field, the Marshall texture diagram, based on percentage of clay size ( tll)ll)
e·;n o e ..!:::::.. VI
-or- ; c
"' . ll).&l
•
II)
>
~
~U;::
•
~
1:!"' tlllc
~~to
~~~; ~~CQ
~ 0 ~N
o·:
0
~&!::
C2 crosses the current LC curve. Along this path the yield locus will be dragged towards position LC2 which is characterized by a larger saturated preconsolidation stress p~2 . This increase in preconsolidation stress 177
s
a)
... p
e
b)
8...---_,.o.. A v----~CI.
p
s c)
COMPRESSION
C2
EXPANSION
Fig.2. Conceptual model for an unsaturated soil. a) Yield curves and stress paths; b) Compression curves; c) Deformation in a wetting path
(P~l -+ p~ 2 ) is the result of a volumetric compaction of the sample (collapse). The LC yield locus explains therefore two basic features: the loading behaviour at different suctions and the collapse phenomena. This is the origin of the acronym LC (Loading-Collapse). Note that along the wetting path BC1C2 the model predicts a first stage (B -+ Cl) where swelling occurs. This tendency is, in the case depicted in Fig. 2 c, later reversed in C1 -+ C2 until a net collapse is finally reached.
An important consequence of the model is that, given an initial structure of the soil, a more compressed state may be reached either in a loading or in a wetting path. This is schematically indicated in Fig.3. The left part of the figure indicates a macrostructure of a soil in which granular sand-like particles and aggregates of clay platelets are in equilibrium with water under suction s and an isotropic stress state given by the total net mean stress Pl· Water partially fills the connected pore space. However, clay aggregates are almost certainly saturated. Total water suction inside 178
these aggregates must be in equilibrium with the capillary suctionwhich prevails in the larger pores and helps to maintain the rigidity of these aggregates. If suction is reduced to zero (lower path of the figure) two mechanisms contribute to volume change reduction: a rearrangement of the structure and the distorsion of the aggregates which are unable to resist the shear induced by contact forces, once the suction is reduced. Some bending of clay particles has also been sketched as a possible mechanism for partial elastic recovery if the external load Pl is eventually removed. Loading the specimen at constant suction (upper path in Fig. 3) leads essentially to the same mechanisms of deformation. The maintained suction is now unable to resist the increased values of concentrated shear forces between particles. Rearrangements and distorsion of aggregates will again result. The LC yield locus represents the limit arrangement of particles and shape of aggregates which is in equilibrium with a given combination of mean stress, p, and suction, s. Any increase in p or reduction in s beyond the frontier given by LC provokes an irreversible rearrangement and distorsion of particles and hence marks the begining of a collapse.
The following parameters are part of the model: A (o): compressibility coefficient for the virgin loading curve for saturated conditions; k: elastic stiffness parameter (unloading-reloading); r: ratio of minimun value of compressibility coefficient (for high values of suction) to A (o); fJ: controls the rate of increase of stiffness with suction; (.X(s) = .X(o)[(l- r)exp(-fJs) + r]) and Pc: a reference stress. For triaxial stress states, yield conditions should be specified in the (p, q) plane where q is the deviatoric stress. The framework described should he consistent with the behaviour under saturated conditions (s = 0). For simplicity the limiting model adopted for saturated conditions was the modified Cam-Clay model, although more complex models can also he coupled to the unsaturated formulation if a better representation of some aspects of soil behaviour is required. Yield for a given suction
¢
Pz
{}Pz
¢
P,
{}P,
P,
¢
{}P, Fig.3. Microstructural interpretation of loading and collapse deformations 179
was therefore described by ellipses exhibiting an isotropic hardening, controlled by plastic volumetric strain. Figure 4 shows a threedimensional view if the yield surface in a (p, q, s) stress space. The equation for the ellipses in planes of constant suction, 2 q2 - M (p
+
Ps) (Po - P)
=
(1)
0
introduces the final set of parameters of the basic model: M is the constant slope of the critical state failure line and Ps = ks, where k is a parameter which controls the rate of increase of apparent cohesion with suction. It is finally necessary to specify the shear modulus, G, within the elastic domain. In order to describe the capabilities of the described framework to model unsaturated soil behaviour, a reference soil, characterized by a set of parameters, will be subjected to ideal stress paths commonly found in laboratory tests performed in unsaturated soils. Only a qualitative comparison between model results and some selected laboratory data will be made. The intention here is to show how the few basic hypotheses made in the development of the framework may capture relevant aspects of unsaturated soil behaviour. Most of the tests reported in the literature to characterize the volume change behaviour upon wetting involve a comparison of loading under constant natural water content with loading under saturated conditions. The condition of constant water content is equivalent to undrained loading in saturated specimens. It results in a change of negative pore water pressure and therefore in a change of suction, which generally decreases with applied load. In this way two 'types' of deformation coexist along a constant water content path: those induced by changes in external load and those associated with suction reduction (which may be either an expansion or a collapse if the LC yield curve has been reached). Modelling these "undrained" paths requires the specification of the relationship between water content (which should
q
s
p
Fig.4. Yield surface of LC model for triaxial stress states 180
be considered as an additional state parameter of the soil) and stress conditions (suction and net stresses). A simple version of this fundamental relationship is to specify the water retention characteristics of the soil. In fact, it seems that water content (or degree of saturation) is strongly controlled by applied suction and, to a lesser degree, by net stresses. Accordingly, in the examples presented below a relationship between degree of saturation and suction of the type
Sr
=
(2)
1- tanhs
will be selected. This equation, based on previous developments (Lloret and Alonso, 1985), implies that saturation reduces to zero if suction increases indefinitely and that a suction of 1Mpa brings the degree of saturation to zero. The remaining parameters specified (.\(0) = 0.09; K- = .\(0)/10; r = 0.4; (3 = 0.8MPa- 1 ; pc = 5kPa; K-s = 0.01, G = 5MPa, k = 0.8; M = 1.2 (r.p' = 30°)) correspond to a slightly expansive clayey sand, with a moderate increase in stiffness with applied suction and prone to collapse if confining load exceeds the value of the apparent preconsolidation stress for saturated conditions, taken as Po = 100kPa. The following illustrative examples have been solved:
a) Behaviour in compression Two oedometer loading tests were simulted (Fig. 5). In the first one load was increased at a constant water content. Note in Fig. 5a, which shows the stress path followed by the sample in a (p,s) plane that suction reduces along the loading path for the reason mentioned before. In the second simulated test the sample was first soaked when unloaded and then loaded in saturated conditions. These model results may be compared with the tests on a loose clayey silt reported by Blight (1993) to this Symposium and reproduced in Fig. 6. It is shown that the theoretical framework reproduces correctly the transition between swell at low normal stresses to a zone of collapse at higher stresses. The model predicts a continuous increase in collapse with applied external stress although changes can be introduced to simulate a maximum of collapse (Josa et al., 1992). 1.20
1.70 a)
'G 0.80
1.60
~
...
"
0.40
0.00
1.50
-f-rrrnrTT1TTT~TTrnT1rTT1TTTTTTTTTnT1rrn
0.0
0.5
1.0 1.5 o1 -u 0 (M.Pa)
2.0
1.40 +--,-,--,rrrTTr-,---r-1rT1TTTr--, 0.01 0.1
o,-ua (M.Pa)
Fig.5. Model predictions in oedometer compression tests. a) Stress paths; b) Specific volume - vertical stress relationships 181
LOG PRESSURE kPa 10
20
30
so
80 100
200
1000
2000
1.7
1.6
0
~
a:
1.5
0
6
> 1.4
BROWN AND ORANGE CLAYEY SILT FROM DEPTH OF 3.0m
NATURAL WATER CONTENT 12"/o
INITIAL DRY DENSITY 1030 kgim 3
PLASTICITY INDEX - 26 PERCENTAGE FINER THAN 0.002mm:13
1.3
1.2
Fig.6. Reaction of a loose clay silt to loading at natural water content and after saturation (Blight, 1993)
b} Wetting tests at different confining pressures Fig. 7 indicates the response of the model to wetting at different confining stresses. The stress paths are shown in Fig. 7a. The model predicts that the wetting paths end in the saturated compression line as many laboratory collapse tests indicate. An example is given in Fig. 8 which shows the results obtained by Erol and El-Ruwaih {1982) in wetting tests of a loess classified as ML-CL. This loess had a natural water content below the plastic limit, its initial degree of saturation did not exceed 25% and therefore its initial suction was presumably very high. Note also that in the wetting path at a vertical stress o-v = 0.2MPa an expansion is first computed followed by a final net collapse. This behaviour has been illustrated in Fig. 2c and has been measured by some authors in suction controlled experiments {Escario and Saez, 1973; Cox, 1978). c) Alternate application of drying and loading paths (stress path dependence) Consider in Fig. 9a two alternative stress paths to impose a given suction and net mean stress to a given sample (point Din Fig. 9a). Starting at a low stress and suction {point A), Path 1 involves a first drying followed by loading. In path 2 the sample is first loaded and finally dried. The computed specific volumes along these two paths are shown in Fig. 9b. Path 2 leads to a substantially larger deformation than path 1. This is actually the result observed in suction controlled tests which impose paths of this type. Unfortunately no tests of this kind, performed on natural soils, has been found in the literature. However, experimental data is available on compacted samples of kaolin (Karube, 1986; Josa et al, 1988) and low plasticity clay (Barden et al, 1969). In the first two cases tests were performed in suction controlled triaxial cells whereas in the latter case a suction controlled oedometer was used. The 182
1.70
al
-----~
0.80
1.60
"ii'
!
.. 0.40
1.50
0.00 0.00
0.50
1.00
o,-ua (MPa)
1.40 +-----,r-~-rrn-rrr--.---,.---rrrTn-r---. 0.01 0.1
1.50
o1-u 0 (MPa)
Fig. 7. Model predictions in wetting tests at different confining stresses (oedometric conditions). a) Stress paths; b) Specific volume- vertical stress relationships results obtained by Josa et al (1988) are given in Fig. 10. The measured specific volume in Josa's tests under the paths EJ and EG (Fig. lOa) have been indicated in Fig. lOb. The model shows a satisfactory qualitative agreement with these results which can also be found in the experiments reported by Barden et al (1969) and Karube (1986).
EXPANSIVE CLAYS The previous framework predicts reversible elastic swelling strains in the elastic zone. Experience with swelling clays indicates that these soils exhibit large volumetric changes, in part irreversible. In addition, strong path dependence is observed in wetting and loading paths (Justo et al, 1984). The framework described before is unable to reproduce these features of expansive soil behaviour.
-r-= 41
SATURATION AT :
Q
• 0.28 kg/cm2
0.56
t
0.80
075 01
0.2
0.5
0.7
1.0
2.0
5.0
7.0 10.0
LOG PRESSURE (IN kg/cm2)
Fig.8. Collapse test results on loess samples reported by Erol and El-Ruwaih (1982) 183
1.20
A
a)
1.00 0.80
?
B
1.65
0