Characteristics of Tlemcen’s Clay


Bekkouche Abdelmalek

Department. of Civil Engineering, University of Tlemcen, Algeria


Aissa Mamoune Sidi Mohammed

Department. of Civil Engineering, University of Tlemcen, Algeria


Numerous cases of structural damage have been observed in the area as well as in the neighborhood of Tlemcen (North-West part of Algeria) and are regarded as a consequence or our less knowledge of the subsurface soil. Whereas some destroyed facilities have been lost other was restored with in turn lost of billion of Dinars (Algerian currency).

In order to identify the main cause of these damages a geotechnical survey has been conducted in the area of Tlemcen. Hence many soil investigations have been performed and the location of the holes was defined such as these will be close to the damaged structure. Through this soil investigation the soil profile of Tlemcen is now well understood. A geological map of Tlemcen has been established using a 1/2.000.000 scale. Many laboratories test such as physic and chemical identification process, compressibility test, swelling test following numerous standards, and shear test using Casagrande procedure have been conducted. The results of these laboratories test combined with geological map have been useful to establish a detailed geotechnical map of Tlemcen. This map is undoubtedly of great importance for urbanisms to avoid past errors. Indeed, this map highlights great heterogeneity of Tlemcen’s soil with emphasis on rock and clay soil on which many facilities have been built. For rock soil, no problem is expected in terms of soil bearing capacity but additional money is necessary for dewatering. On the other hand for clay soil, swelling phenomena could cause severe damages to structures due to unstable level of moisture. This instability is mainly due to drought which has been observed during many years in Tlemcen. This situation was unfortunately not undertaken by previous survey for some new parts of the town. Based on the availability of soils samples some statistical models are proposed to make prevision about the soil behavior. These models allow rapid evaluation of the amplitude and strength of swelling and could serve as a guideline for next implementations of structures.

Computational models and experimental results are presented in this paper.

Keywords: finite element method; permeability; soil-water; unsaturated soil.


Many structures founded on clay showed signs of degradations in the form of cracks on the superstructures in the town of Tlemcen (North-West part of Algeria) because of the dryness during the two last decades. These degradations led to total collapse of some facilities. The principal cause of these disasters is the shrinkage, phenomenon not taken into account at the time of the design of these strructures. This damage also observed on road embankments, roadways and foundations (Bekkouche and al., 1997). The awakening of this problem is new and the technical solutions are not always efficient, and not easy to use. If this problem is not taken into account for new constructions, swelling phenomenon could in turn occur (Bekkouche et al., 1999).

This work aims to study the variability of the swelling parameters in the area of Tlemcen, which compsed of of Tlemcen-Downtown, Mansourah and Chetouane. On other words, we have to understand the swelling potential in this area using Geographical Information System "G.I.S."

First, a geological outline of the area of Tlemcen is presented. The superimposing of this aspect with the data geotechnics available allowed us to establish a geotechnical map of the area of Tlemcen. This map emphasizes great sections of clay which could be subjected to swelling phenomenon. In order to measure swelling parameters of theses sections, a qualitative and quantitative study was carried out. This study starts with the identification of the potential of swelling of these formations using classifications available (Djedid and Al, 2001). Once this step accomplished, a series of direct measurements was then carried out.


Geological Characteristics

The entire site of Tlemcen covers an area of 11220 hectares et and is at an average altitude of 800m. We have found that this area is characterized by two great parts : soft soil in the north part composed of clay issued from miocene bassin et rock soil in the south part represented by limestones and Jurassic dolomites of the mounts of Tlemcen. The contact between these two formations is precisely at the level of the grouping, which partly explains the complexity of its geology

The mounts of Tlemcen (southern of the area) consist of mainly carbonated soil formed during the Jurassic superior. The latter are in contact with clay sediments with sandy soil from the miocene on the level of the catchments area of Tlemcen. This area is characterized by a breakable tectonics for which the rigid grounds issued from the Jurassic are subdivided in various formations according to the following nomenclatures:

One also distinguishes tertiary grounds which are composed of both dominant marls and more or less thick sandy levels allotted to miocene observable thanks to oyster presence in the west part of the town (District of Imama).

Geotechnical Characteristics

The town of Tlemcen experienced geotechnical damage noted on a significant number of construction (Institut de technologie de l’éducation ITE, rocade Est, Lycée Commandant Ferradj, etc.). Knowing that the majority of these disasters are caused by the behaviour of soil and its inadequacy with the nature of the structures, it becomes imperative to establish a geotechnical map of the town. This map will serve as a guideline for the town planners on future installations and will inform them on the difficulties to which they will be possibly exposed if they do not take into account geotechnical information in their projections.

To reach this goal, we have established an approach which is not limited to a qualitative description of the soil in it various aspects (geomorphology, geology, climatology, geotechnics, etc). Indeed, a number of information such as the geological maps, soil charts and wells were collected from the Laboratory of Public works of West (LTPO-Unit of Tlemcen) in addition to those collected from the university. We constituted a bank of geotechnical data (more than 345 set of data points) related to the soft soil which surround the town of Tlemcen (Fig. 1).


Figure 1. Geotechnical map of the area of Tlemcen (Tlemcen-Mansourah-Chetouane)


Table 1. Results of the statistical survey conducted in the area of Tlemcen.
%<2µm %<80µm IP
Number of Observations344 344 344 344 344 344 344 344
Minimum0 0 0 0 20 0 4 20
Maximum16 40 50 80 100 100 61 100
Mean6.044 0.477 0.84 16.015 87.677 67.061 20.34 45.265
Standard deviation 3.787 3.052 4.276 12.547 15.084 18.673 10.065 13.824


Table 1 collects the principal parameters taken into account in the establishment of the geotechnical map with some statistical properties of these parameters. According to this bank of data, and by superimposing this information with the geological map, six categories of ground are highlighted:

Category 1: The first category of ground highlighted is the marly formations of which the depth often exceeds the 30 meters. For these formations, we expect:

Category 2: The second principal category highlighted is made up of alluvium dissolved from which starts marly or sandy formation. Because of the very heterogeneous band of these formations, the foundations must be anchored beyond the alluvium layers.

Category 3: The falls constitute the third formation. Knowing that the latter are prone to important settlement if charged, it will thus be necessary to put foundation beyond this layer.

Category 4: Formations of likings tortoniens were also highlighted. These formations do not pose problem of bearing capacity and settlement, but locally contain marl part of which the hydro-geological behaviour would have to be identified.

Category 5: The data as well as the visual observations made possible the detection of thalwegs hidden under constructions. The thalwegs are filled with alluviums and thus served as natural drains. They are the heart of flows convergence. This can lead to headwater erosion involving the fine particles and there will be thus risk of modification of the structure of the ground. Consequently For for certain projects the drains filters are to be envisaged.

Category 6: The geology of the area of Tlemcen shows the presence of rock formations in some parts. These families of ground present neither of the problems of bearing capacity nor of settlement. However, earthworks in these parts are very painful and cost much. Furthermore, use of heavy tools can disturb the structures already existing


The geotechnical map clearly lets appear that the town of Tlemcen is built on heterogeneous formations surrounded by marly formations. However, it is on these last formations that the wide extension of the city happened. This extension was carried out without previous study as a preliminary step to include/understand the behavior of these marls; what partly explains the disorders noted on certain structures.

These marls, principal formations of all the area of Tlemcen, caused numerous disorders noted in some places and localities more or less far away from the grouping: The road 7A to Bab El Assa (78 km), buildings in Sidi Abdelli (35km), school in Oum El Allou (Nédroma, 60 km), etc.

In the area of Tlemcen, the marls are in contact with the likings and between them we have noticed at a flow in the underground which generates dissolution of the limestone of the rock formations and its deposit in the marls. This led to the very heterogeneous formation of an area for which the laws and models of the soil mechanics are not easily applicable. This area covers a band which connects the industrial zone to the area known as Champ de tir (District El Kiffane). Moreover the disorders that occurs in this district were attributed to the ground heterogeneity

In what follows, we will focus to the swelling character of these marls, with their localizations and finally with the management of these information for the purpose of spatial variability of the swelling parameters.


Though a statistical study, which is applied on data obtained, statistical parameters of the set of data are given. This stage enables us to define and establish models of probabilities laws for each variable. In addition to the statistical data, an analysis in principal component is carried out in order to test the reliability of the observations to estimate the pressure and amplitude of swelling.

Numerous statistical parameters (Table 2) were determined per parameters in order to test the representativeness. A careful examination of the average of the variables, one notices that the presence of argillaceous minerals is important owing to the fact that the average of the elements lower than 2µm is higher than 50% (53.083%) and the average of the value of blue is higher than 2 (6.415) with very low coefficients of variation of about 0.3 (0.34 and 0.295 respectively).

In addition, the variation of coefficients of kurtosis shows that the majority of the variables do not show a spread variation except for liquid limit and amplitude. Moreover, one notices that the coefficient of skewness is negative for 7 variables and positive for the others. However, it is worth noting that a preliminary a conclusion about a general tendency as for the position of the peak (more high frequency) compared to the average.

Finally, a detailed examination of these results shows that the majority of the variables present coefficients of variation ranging between 0.055 for the degree of saturation and 1.207 for amplitude. This result is in accordance with the factors causing of the errors of measurement per type of test at the laboratory. For instance, one notice that the coefficient of variance related to the percentage of fines (0.34) is definitely lower than that of plasticity index (0.646), which confirms the difficulty of the determination of the limits of Atterberg compared to that of granulometric analysis. Moreover, one notices the difficulty of the measurement of swelling amplitude compared to the measurement of the swelling pressure by the fact that the coefficient of variance of amplitude (1.207) is higher than that is obtained for the pressure (0.713), which suggest the use of a semi-logarithmic models in order to reduce the coefficient of variance (from 1.207 to 1.044 for amplitude). (Aissa Mamoune, 2002).

Table 2. Statistical Description of the Swelling Parameters
Aver. Depth %<2µm Wn Wr Ip A CaCO3 Vb gd s Logs e Loge
Observations82 78 82 78 78 78 78 78 78 82 82 47 47
Minimum1.480 24 6.8 7 15.5 0.18 3 0.9 1.29 0.36 -0.44 0.030 -1.47
Maximum26.75 100 34.07 23.84 100.16 1.35 45.16 8.60 1.80 11.13 1.05 51.30 1.71
Mean6.577 53.083 20.12 16.15 34.63 0.66 21.15 6.41 1.66 3.21 0.38 7.01 0.56
Variance21.91 325.99 57.41 20.05 501.2 0.08 153.29 3.58 0.02 5.22 0.13 71.65 0.35
Standard deviation4.68 18.055 7.577 4.478 22.38 0.28 12.38 1.89 0.12 2.28 0.36 8.46 0.59
Coefficient of skewness1.909 0.975 -0.215 -0.228 1.929 1.17 0.20 -1.16 -1.74 0.96 -0.47 3.33 -1.09
Coefficient of Kurtosis5.136 0.347 -0.341 -1.161 2.921 0.90 -0.92 0.598 2.388 0.548 -0.47 14.27 1.734
Coefficient of Variation0.712 0.340 0.377 0.277 0.646 0.43 0.585 0.295 0.074 0.713 0.947 1.207 1.044



Estimation of the Swelling Pressure

David and Komornik (Quoted by Kabbaj, 1989), based on a statistical study on 200 samples, deduced a relationship which allows the estimation of the swelling pressure. Adjusted with clays of the area of Tlemcen, one found that:

ps : swelling pressure (in kg/cm²), wL : limit liquid (in %), Wn: water content (in %) and gd : dry unit weight (in g/cm3).

On the other hand, Williams and Donaldson (Quoted by Mouroux, 1989) propose a model which gives the amplitude of swelling with respect to the load which acts in the ground. Hence:

Consequently if the amplitude is zero (Sp=0), the pressure of free swelling is given by:

e = 2.718..., Ip plasticity index (in %) ps: swelling pressure (in 0.1 MPa).

On the other hand, Vijayvergiya and Ghazzaly issued two models of swelling parameters by using the results obtained from 270 tests of swelling carried out on various soils. Adjusted with clays of the area of Tlemcen, these models could be written as follow

ps: swelling pressure (in tsf), wL : limit liquid (in %), wn: water content (in %) and gd: dry unit weight (in pcf).

Estimation of the Swelling Amplitude

It is generally accepted that models of the swelling amplitude are given for the free swelling. The model suggested by O’Neil and Ghazzaly is:

Whereas the model proposed by Jonhson et Snethen is written as follow

For both models, Sp is the amplitude for free swelling (in %) wL and wn (in %).

Again, based on the results obtained from the 270 tests of swelling, Vijayvergiya and Ghazzaly showed that:

where gd is in pcf and Sp, wL and wn are in %.

The free swelling, obtained from the above models, can be reduced if the ground would be subjected to a confining pressure sv through the formula suggested by Gogoll:

where sv is the confining pressure (in kPa).

Due to the large difference between the values of the swelling parameters directly measured and those provided by the models above, we think that a better idea is to adjust the suggested models for the soil under study (Bekkouche and al., 2001). To this end, a statistical analysis was thus carried out which lead to the following results

The models of Seed and Col. do not take into account the water content natural believed as an important parameter in the swelling process. These models are based on interdependent parameters (content of clay, activity and plasticity index) only.

The model of Nayak and Christensen and the model of Vijayvergiya and Ghazzaly cannot be used for high values of the clay content, water content and liquid limit. The first model is based on interdependent parameters, whereas the second does not take into account water content.

The model of Johnson predict overall percentages of swelling within limits usually observed in practice

The results of adjustment performed in this study; show clearly that only the models of Johnson seem to be applicable. Thus even adjusted, the majority of the models remain imperfect in their forecasts (Bekkouche and al, 2000b). A statistical study on the data available of clays of Tlemcen was carried out to seek specific models to these grounds. One gives below the models considered to be satisfactory.

Proposed Model for the Clay in the Area of Tlemcen

For the amplitude of swelling, the models are as follow:

log Sp= –0.008 Z + 0.27 A – 0.02 TCa + 0.016 Sr – 0.16

log Sp=–0.1 Z + 1.06 A + 0.22 gd – 0.04 wn + 0.82

For the pressure of swelling, the models are as follow:

log ps = 0.01 Ip + 1.26 gd – 0.008 wn – 0.1 M – 2.179

log ps = – 0.001 wn Ip + 0.024 wL + 0.1 M – 0.713

log ps = 0.006 Ip + 1.21 gd – 0.013 wR + 0.11 M – 1.97


A, C, Z, wn, wR, wL, gd, Ip and TCa are respectively the activity, the content of clay (in %), the depth (in m), the water content (in %), the shrinkage limit (in %), the liquid limit (in %), the dry unit weight  (in kN/m3), the plasticity index and the content of carbonate calcium (in %). the parameter M is a factor which characterizes the weight of the procedure used to measure the swelling parameter in question (Balu, 1987). This parameter is equal to 1 for the procedure of free swelling, to 2 for the procedure of swelling with constant volume (DTU 11.1), to 3 for the procedure of swelling under constant constraint (AFNOR), to 4 for the procedure of swelling with constant volume (LCPC) and 5 for the procedure with variation of volume (LCPC).

For the pressure of swelling, the models suggested seem to be representative since the values obtained are rather close to those measured directly. They can thus be used to obtain approximate values of the pressure of swelling of sites whose physical properties are comparable.

For swelling amplitude of, the results of the models remain controversial because of the difference noted between direct measurements and calculations. These differences can have be the consequence of the type of loading (either the piston alone, or both the piston and the weight of the grounds).


The establishment of a map that shows high-expansion soils in the area of Tlemcen has a great interest in the comprehension of the many recorded damage and the projection of new projects. To this end, we used the technique Geographical Information System "G.I.S." This tools, largely used in the treatment and the visualization of physical information, allows on a graphic support to give information to decision-makers.

By superimposing the geological and geotechnical maps (fig. 1), we located the presence of the high-expansion soils in the area of Tlemcen (see fig. 2).


Figure 2. Localization of the swelling soil in the area of Tlemcen

In order to have a better view of the variability of the pressure and amplitude of swelling in the area of Tlemcen, they have been measured directly from the data collected from the multiple boreholes. Next these values are mixed in the GIS which allow us to establish maps that show the variation of pressure of swelling (Fig. 3) as well as of the amplitude of swelling (Fig. 4).


Figure 3. Variation of the pressure of swelling


Figure 4. Variation of the amplitude of swelling

Figs. 3 and 4 show that the pressure of swelling varies between 4 and 8.5 bars whereas the amplitude varies from 8 to 14 %. These two results confirm qualitative study presented above because we have now evidence that the area of Tlemcen is highly subjected to swelling phenomenon. For instance, the district of Mansourah has a pressure ranging from 6 to 8 bars with amplitude of 11% what means high risk for existing and future constructions. In addition, a detailed examination of these figures shows that extreme North of the area of Tlemcen has a pressure of swelling of 4 bars and an average amplitude of 13 %. This same amplitude was obtained in the east part of the area with an average pressure of 6 bars. This difference can be explained by:


The statistical analysis carried out on the data showed that the majority of the soil samples are of argillaceous formation. Furthermore, the probabilistic analysis has confirmed the complex behaviour of the soil and shed light on the difficulty to conduct comparable test on soil. The analysis in principal component, show that theses observations could be used to propose estimated models of the pressure and amplitude of swelling.

Of course, if we can suspect the potential of swelling in a particular area before the laboratories tests then this latter could efficiently conducted and would give better idea on the shape of foundation recommended in the same area. However, it is strongly advised to use classifications based on more than one parameter. The parameters determined from identification tests which appear as being most important in the expansive behaviour of clays are the plasticity index, the percentage of the argillaceous particles, the value of test to the methylene blue and the shrinkage limit.

Also, knowing that the measurement tests of the parameters of swelling are long and expensive, it would be very interesting to be able to quickly obtain a first estimate of these parameters. The use of the mathematical models offers this possibility. As it was specified in this work, the proposed models are not usable blindly since even adjusted with clays of the area of Tlemcen, these models provided values whose variations compared to the measured values are important. The solution would be to seek specific models to each family of clay but if sufficient data are available

Our present work showed that with only 80 observations on sample soil from the area of Tlemcen, the forecast was improved considerably. The values obtained through these models should never replace direct measurements while a structure is to implemented. However, the use of these models can reduce considerably the number of measurements to be carried out.

Geotechnical information and its visualization on map are of great importance for the establishment of geotechnical map related to each phenomenon. This map could be updated once additional materials are provided. This map could, by superimposing the information, give valuable idea on the potential of swelling of an area in the preliminary phase of investigation for the future projects.

We have presented an approach to analyse the variability of the swelling parameters for the Tlemcen-Mansourah-Chetouanne area known as area of Tlemcen by using the G.I.S. The results obtained primarily highlight the expansive behaviour of the marls in the area and especially in the new extension of the city. Consequently, the use of this map can help decision-makers to take into account the swelling phenomenon. Another technique allows us to estimate the swelling parameters using forecasting models. Those two materials, map and model, are to be used to avoid long and expensive tests generally used to determine the potential of swelling.


  1. Aissa Mamoune, S.M. (2002) “Contribution à la mesure, prévision et modélisation du comportement des sols expansifs,” Thèse magister, DGC – Faculté des Science de l’Ingénieur, Université Aboubakr Belkaïd, Tlemcen.
  2. Bekkouche, A., A. Djedid and N. Bousmaha (1997) “Construire sur les sols expansifs,” Séminaire National de Génie Civil. M’sila, Algérie.
  3. Bekkouche, A., A. Djedid and N. Bousmaha (1999) “Aspects expansifs des marnes de Tlemcen,” 1er Séminaire National de Géotechnique et Génie Civil, Alger.
  4. Bekkouche, A., A. Djedid and S. M. Aissa Mamoune (2000a) “An experimental investigation on the assessment of the swelling parameters,” IV. International Congress in Civil Engineering, Eastern Mediterranean University, GAZIMAGUSA, North Cyprus.
  5. Bekkouche, A., A. Djedid and . M. Aissa Mamoune S(2000b) “Prévision du gonflement des sols argileux,” Conférence Internationale de GEOENGINEERING, U.S.T.H.B, Alger.
  6. Bekkouche, A., Djedid A. and S. M. Aissa Mamoune (2000c) “Mesures directes des paramètres du gonflement,  Séminaire National de Génie Civil 2000, Mostaganem, Algérie.
  7. Chen, F. H. (1988) “Foundations on expansive soils,” Elsevier, U.S.A.
  8. Djedid, A., Bekkouche A. and S.M. Aissa Mamoune (2001) “Identification et prévision du gonflement de quelques sols de la région de Tlemcen (Algérie),” Bulletin des Laboratoires des Ponts et Chaussées, N°233, pp 67-75.
  9. Magnan, J. P. (1982) “Les méthodes statistiques et probabilistes en mécanique des sols,” Edition Presses de l’ENPC. France.
  10. Kabbaj, M. M. (1989) “Sols gonflants – mesures des propriétés au laboratoire,” Symposium International de Mécanique des Sols, Tiaret, Algérie.
  11. Mouroux P., P. Margon, J. C. Pinte (1989) “La construction économique sur les sols gonflants,” Manuels et méthodes, N°14, BRGM. France.
  12. Seed H. B., Woodward R. J., Lungreen R. (1962) “Predicting of swelling potential of compacted clay,” Journal of soil mech. and Found. div. ASCE Vol.


© 2005 ejge