The Second International Symposium on Rockfill Dams

Transcription

1 EVALUATION OF ROCKFILL PROPERTIES BASED ON INDEX TESTS A. Veiga Pinto 1, M. Quinta-Ferreira 2 1, Executive Director Mecasolos, Rua Xavier Araújo, 11, Núcleo 8-6ºB, Lisboa, Portugal, vpinto@mecasolos.com 2, Dep. Earth Sciences, University of Coimbra, Coimbra, Portugal, mqf@dct.uc.pt Abstract: A rockfill material can be idealized as a continuous model whose properties depend on individual particles. The characteristics of strength and durability of the granular matrix can be idealized as a discrete model. Correlations between the components of the discrete model and the continuous model are presented. A methodology to characterize the collapse of rockfill materials based on strength loss and increased fragmentation from air dried to saturated samples is presented. Types of stress-strain behaviour of rockfill materials and its correlation with the structural behaviour of prototypes are discussed. The long term observation of the rockfill behaviour, also allowed the establishment of some correlations with the results of the index property tests for the characterization of the sensitivity of the rockfill to the water, i.e., the collapse. The Portuguese experience obtained with the construction of rockfills with unweathered and high strength materials, allowed concluding that the best mechanical characteristics were obtained, by decreasing order, with the following materials: basalt, greywacke, limestone, granite and schist. When using weathered rocks in the construction of the rockfill, materials like greywacke, marly limestone and schist, have presented a defective behaviour, as a result of the weakening of rock fragments in the presence of water. It can be considered that granite can also be sensitive to the effects of collapse, but to a lesser extent than the other previously mentioned materials. As basalts present a low weathering grade they are almost insensitive to collapse, allowing concluding that they are a remarkable material for rockfill construction. Concerning limestones, greywackes and schists materials it is advisable to determine their ability to lead to collapse settlements. The advantage of testing expeditious and inexpensive procedures for the characterization of physical and mechanical properties of rockfill material is presented. Keywords: Rockfill, index-properties, mechanical properties, collapse, dam structural behaviour 1 Introduction In the late 50s, Paradela Dam was the tallest dam in the world with an upstream concrete curtain. Due to the accident at the first stage of filling, caused by large displacement of the dam, this led their designers to conclude for the need to develop research in order to characterize this type of construction material. In reality, the reliability of the design of rockfill structures depends on having a representative estimate of the mechanical properties of such material. By this reason intense research on the physical and mechanical properties of rockfill was developed in Portugal. From the analysis of the tested materials, it was possible to conclude that there is a significant variation in the mechanical properties of rockfill materials, depending on the lithological nature of the rock fragments. One indirect way to estimate the mechanical properties of rockfill materials, is relating the values of the index properties of rock fragments with the physical and mechanical properties of the granular material, considering it as a continuous material. The methodology that has been followed in the structural design of rockfills, including dams, is based on index tests on rock samples, of fast and simple execution. 2 Discrete and continuous models Rockfill materials consist of rock blocks that can reach large sizes of about 1.5 m. Therefore, particles are the basic elements of a particulate medium and, as such, the latter can be represented by a discrete model. Furthermore, they are expected to be part of an integrity criterion, i.e., the particles are to be linked by certain connections, the contact forces, and the voids between particles are also to be considered.

2 In fact, that behaviour mainly depends on four factors, which are listed below in decreasing importance [1]: 1. State of stress; 2. Void ratio; 3. Strength of rock fragments; 4. Grain size distribution. The first factor depends on the size of the structure to be built and, therefore, the mechanical characteristics of the granular medium will take into account the magnitude of the applied pressures. In fact, it has been observed that, with the increase in the state of stress, there is a significant decrease in the mechanical properties of the rockfill material. The compactness began to loose its importance in the behaviour of those materials when, as from the 60s, heavy vibrating rollers began to be used in compaction, which led to reduced void ratios, generating highly homogeneous materials with high shear strength and reduced deformability. It has also been observed that the materials resulting from excavations, in the case of road works, or from quarries of good quality rock formations, almost invariably led to rockfill materials with fairly extensive grain size curves. Consequently, the consideration about the distribution of rock fragments in rockfill materials also became an almost irrelevant factor in the study of their mechanical behaviour. Thus, it has been observed that the physical-mechanical characteristics of the rock fragments of rockfill materials are the factor that, as regards the particulate medium, has a higher relevance in the mechanical characterisation of rockfill materials. From among these characteristics, reference must be made to the crushing strength of the individual granular elements. The strength of rockfill blocks to crushing is one of the main factors that influence stress-strain behaviour of this particulate medium. The Elasticity and Plasticity theories assume the rockfill materials as a continuous medium, separating their existence from their component materials. An application of this theory is, for instance, the determination of the mechanical properties of rockfill materials in large test cells or the integration of constitutive laws in the finite element method, which has been increasingly used in the stress-strain mathematical modelling of embankments, namely in dams. 3 Index-properties The physical-mechanical characteristics of rockfill fragments are usually designated as indexproperties. Those properties, which are determined in a rather expedite and inexpensive way, have made possible to estimate the mechanical characteristics of rockfill materials [2], in a first magnitude, without having to use time-consuming tests in large test cells. The authors collaborated on studies of index properties performed at the National Laboratory of Civil Engineering (LNEC, Lisbon). In these studies tests have been performed on 110 samples of carbonate rocks; as well as on 30 samples of granite and 23 samples of metagreywacke [3], [4]. Table 1 presents the index-property tests carried out. Table 1. Index-property tests Characterisation of: Parameter Symbol References Texture Porosity n [5] Bulk density dg [5] Uniaxial compressive strength c [6] Compression strength Point Load Test PLS [7], [8] Crushing strength P a 50 [9], [10] Durability Slake durability test Id 2 [5] Los Angeles LA [11] Water sensitivity Swelling strain - [5] Non-soluble residue - [12] From the analysis of results obtained for the index-properties, it was considered that a classification of rockfill materials distributed by the 3 lithologic types that prevail in Portugal could be experimentally proposed: carbonate rocks (Table 2), greywacke rocks, which occurrence is usually associated with schist (Table 3), and granitic rocks (Table 4). The values of the parameters presented in Tables 2 to 4 concerns the average values obtained in the tests, in an air drying condition. An attempt was made to roughly distribute the same number of tests per each class mentioned above. Only a reduced number of basalts were tested and, therefore it was not possible to establish a distribution of index-properties by classes for this rock type. However, it was observed that this type of material was the one

3 exhibiting the best mechanical characteristics between all the rockfill materials tested. Based on LNEC experience, it is considered that rockfills consisting of rock fragments of the same class number even in different lithologic types, roughly presents identical structural behaviour of the rockfills. The materials consisting of class 1 rock blocks, which exhibit less weathering or less weatherability, higher strength, less abrasion and less susceptibility to water, are likely to produce rockfills with high mechanical characteristics. When placed in embankments, those materials have exhibited an excellent structural behaviour, even when they are used in the compaction of 1.60 m thick layers, such as the embankments built in expressway A1, in the stretch Torres Novas-Fátima (limestone) and the works for extending Ponta Delgada Airport in the Azores (basalt). On the other hand the materials of Class 3 would meet a low strength rockfill soil mixture. Both materials have shown a poor performance when used in the construction of embankments, which was not predictable, even when compaction was carried out in layers of only 0.60 m thick. In the light of experience, it should be considered mandatory testing individual samples of material of Class 3, both in dried and saturated states, to determine the sensitivity of those materials to the influence of water and subsequent collapse. Table 2 Index-properties of carbonate rocks Parameter Class 1 Class 2 Class 3 n (%) dg (kn/m 3 ) c (MPa) PLS (MPa) P a 50 (kn) LA (%) Id 2 (%) Swelling strain ( l/l.10-4 ) Non-soluble residue (%) Table 3 Index-properties of greywacke rocks Parameter Class 1 Class 2 Class 3 n (%) dg (kn/m 3 ) c (MPa) PLS (MPa) P a 50 (kn) LA (%) Id 2 (%) Swelling strain ( l/l.10-4 ) Table 4 Index-properties of granitic rocks Parameter Class 1 Class 2 Class 3 n (%) dg (kn/m 3 ) c (MPa) PLS (MPa) P a 50 (kn) LA (%) Id 7 (%) Swelling strain ( l/l.10-4 ) The analysis of results of Tables 2 to 4 allow to immediately demonstrate the difference in the limit values of some properties of rock fragments of a certain type of rock, comparatively with the others. Those limits should therefore be taken into account in the analysis of results of new tested materials. The correlations between index-properties and the mechanical characteristics led to a significant

4 dispersion. Therefore, there is the need to increase the size of samples, i.e., to carry out a higher number of tests to be able to define the mentioned correlations. Other authors have also observed the difficulty in defining good correlations between the strength and the durability of stone fragments, namely in weathered materials [13]. 4 Characterisation of rockfills as regards to collapse Collapse in rockfills occurs when there are volumetric deformations by compression, only due to the decrease in the strength of rock fragments when in contact with water. Collapse is related with the crushing of particles, even for low confining pressures [1] and [14], when the crushing strength decreases to a value lower than the contact forces among rock fragments. In the limit it can be assumed that all embankment materials are sensitive to water. However, by the observed phenomena it was found that the more serious situations (excessive settlement) have been observed when the decrease of strength or the decrease in the wear resistance exceeds certain values. Class 3 materials of classifications presented in Tables 2 to 4 should correspond to low strength rockfills or to rockfill soil mixtures, which already present a significant quantity of fine soils. Both materials have presented a defective behaviour when used in the construction of embankments, with worse behaviour than expected, even when compaction was performed in layers of only 0.60 m thickness. In view of the experience gained, tests on samples of class 3 material must be performed also in dry and saturated conditions, in order to evaluate the possible decrease in the strength of rock fragments due to the increase in the water content. From the statistical analysis of the embankments materials that collapsed and of the results of the tested materials in the dry and saturated states, the threshold values beyond which it is anticipated the rockfill materials exhibit collapse are presented in Table 5. The limit relationships of potential collapse are variable, depending on the lithological type and the test procedure considered. Table 5 Proposed limit relation of potential collapse of rockfill materials Carbonate Greywacke rocks rocks Granitic rocks c sat/ c dry (%) PLS sat/pls dry (%) Pa50 sat/pa50 dry (%) Id 2 dry/ Id 2 sat (%) sat - saturated; dry - dry samples Therefore, it must be considered that an unsaturated rock sample is water sensitive when the values of the relations of index-properties in tested fragments, both in saturated and in dry states, are less than those indicated in Table 5 and, therefore, preventive measures should be adopted in the dimensioning of embankments, so as to reduce the effects of collapse. 5 Correlation between index tests and mechanical properties Rockfill materials exhibit relatively high mechanical characteristics, when compared with other materials of soil mechanics. For this reason, in rockfill embankments the inclination of slopes is usually significantly more pronounced than the one adopted in soils. It can be assumed that the rockfill materials, when properly placed in fills, exhibit Young s Modulus up to 200 MPa and internal friction angle up to 60 degrees. Using the current rockfill construction techniques that lead to relative densities close to 100%, they exhibit reduced deformation, consistent with the functional limitations of the structures. Even so, in the design it has been considered appropriate, that the modules of elasticity of these materials have values above 50 MPa. Despite the work developed we consider that research is still necessary to assure a reliable relation between the index-properties of rock fragments used in rockfills with the mechanical properties of the particulate medium. Nonetheless, Figure 2 presents a correlation between the oedometric modulus (E oed, secant obtained from no-lateral-strain compression tests, for a vertical stress of 1 MPa) and the values of the crushing strength for an average diameter of particles of 50 mm, in an air dryed condition.

5 Figure 1. Relation between the oedometric modulus (E oed ) of rockfill samples and the crushing strength of their rock fragments It is observed that in most materials, the deformability of the granular sample decreases with the increase in the strength of particles. Reference must also be made to the fact that the moduli of deformability may vary within a range, of about 10 times (Figure 1). The high variation of shear strength of rockfill materials is easily explained by the way these materials react when subjected to a distortional strain. For low strength particles, the fracturing of the blocks begins to be pronounced even at low average stresses. Thus, the behaviour of the material is not very different when the material is subjected to high stresses. However, when the rockfill consists of high strength particles, for low stresses, there is virtually no fracture, so that particulate materials tend to increase in volume in the shear zone [1]. However, when stress states increases, the law that defines the strength of the blocks decreases more sharply than the contact forces between them, occurring high fracturing of the blocks and hence a marked decrease in strength of the rockfill material. Thus, it justifies the sharp curvature of the Mohr-Coulomb envelope in the rockfill materials of better quality. This leads to the need, in rockfill materials, to consider two parameters to characterize the shear strength. The law that has generally been adopted is of type: = 0 - log ( 3 /p a ) (1) Being 0 the value of the internal friction angle at 3 of 1 atmosphere (p a ) and the reduction in the friction angle due to the increase of the confining stress from 1 to 10 atmospheres. Figure 2 presents the values of the shear strength parameters of rockfill materials tested in the cell of Figure 3 [1]. It is possible to observe that, in average, there is less shear strength in saturated samples, aspect that is related with the collapse of rockfill materials, which shall be addressed in a subsequent section. In most cases, the value of 0 is higher than or equal to 50º, but is also fairly high, reaching very often values higher than 10º. The high values of the shear strength of rockfill materials explains the fact that embankments containing those materials have high inclination faces, of about 45 º, and, in that type of works only a very few accidents occur due to sliding of embankment slopes. From among rockfill materials tested at LNEC, basalts were the ones presenting higher friction angles. The materials that were used in the expansion of Ponta Delgada Airport led to very high values, having reached values of 0 =63º and =15º. Large cells should be used to carry out the mechanical characterisation of rockfill materials. Figure 3 presents LNEC triaxial cell, which allows testing 0.30 m diameter cylinder specimens. Some authors consider that there are a few limitations in the laboratory determination of the mechanical properties of rockfill materials, by considering that those properties are usually overestimated [14] and [15]. It can be considered that the rockfill materials, compacted with powerful vibrating rollers, as is usual nowadays, make possible to construct embankments with high rigidity and strength, i.e., with a

6 modulus reaching 500 MPa and internal friction angles up to 60º. 30 ΔΦ (0) Limestone dry Limestone-dolomite dry Granite dry Greywacke dry Schist dry Limestone-dolomite sat Greywacke sat Schist sat Basalt sat (0) Φ o Figure 2. Values of the relation between the 0 and of rockfill samples Figure 3. Triaxial cell of rockfill materials From LNEC experience, it was concluded that the modulus of elasticity, in the rockfill materials, comparatively with soils, is, in average, about 4 times higher. It was also concluded that the rockfill materials, by adopting the current construction techniques, lead to relative densities close to 100% and exhibit reduced deformations, compatible with satisfactory behaviour of the structures. Therefore, and in view of the experience gained, LNEC has recommended a modulus of elasticity for rockfill materials, after placement on site, higher than 50 MPa, to ensure structures with proper serviceability levels. 6 Characterisation of rockfills as regards collapse Collapse in rockfills occurs when there are volumetric deformations by compression, only due to the decrease in the strength of rock fragments in contact with water. Collapse is related with the crushing of particles, even for low confining pressures [1] and [14], when the crushing strength decreases to such a value that becomes less than the forces of contact among them. Class 3 materials of classifications presented in Tables 2 to 4 should correspond to low strength rockfills or to soil-rockfill mixtures, which already present a significant quantity of fine soils. Both materials have presented a defective behaviour when used in the construction of embankments, with

7 Embankment height (m) The Second International Symposium on Rockfill Dams behaviour worse than the expected, even when compaction has been performed in layers of only 0.60 m thickness. In view of the experience gained, tests on samples of class 3 material must be performed both in dry and saturated conditions, in order to determine the possibility of decrease in the strength of rock fragments due to the increase in the water content. 7 Structural behaviour of rockfill embankments Based on results of index-properties tests and on the mechanical characteristics obtained in laboratory tests done in large cells, it has been possible to estimate the behaviour of rockfills. Figure 4 presents the settlements observed, during construction in the maximum cross-section, on 3 Portuguese rockfill dams with an upstream membrane. Odeleite Dam is a 65 m maximum height dam and has been constructed with metagreywacke; Apartadura Dam is a 46 m high dam constructed with limestone-dolomitic rock and Arcossó Dam is a 40 m high granite dam. Further details on the characteristics of those dams are presented in another paper [16]. The higher settlements observed in Odeleite Dam are fairly significant. The maximum observed deformation at mid-height of the embankment was about 2.4%. From a back-analysis, for the bottom 15 m of embankment, the modulus of deformability (E) obtained was 35 MPa and for the upper 50 m the value of E obtained was 50 MPa. Those fairly low values were less than those determined in the stage of design, which are the values of the greywacke material presented in Figure 2. Probably, the fact that, in the construction stage, moderately weathered metragreywackes were used, which proved to be highly sensitive to compaction with abundant water, might justify the fact that higher settlements than those expected were obtained. Mention must be made of the fact that the moduli determined for the material of Odeleite Dam, presented in Figure 1, were the most reduced. The settlements observed in Odeleite Dam, during the first filling, were less than those expectable in view of the behaviour observed during construction (Figure 4). At the end of the first filling, the long-term settlements were estimated, by assuming the variation over time, using a semi-logarithmic law. Nonetheless, the settlements observed, in the first five years of the dam operation, were less than half of the expected ones [16] Apartadura Arcossó Odeleite ,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 Settlement (m) Figure 4. Settlements measured during construction in faced rockfill dams In Apartadura Dam, the settlements observed during construction were less than the predicted ones, even by taking into account the high moduli of deformability determined in the design stage and which are presented in Figure 1 (limestone-dolomite). Considering the results of Figure 4, the maximum creep settlements estimated in Apartadura Dam were about 0.3 to 0.4%, for the highest cross-section. Nevertheless, after 8 years, a maximum settlement of only 8 mm was observed, which is fairly less than the maximum long-term expectable value that was estimated to be around 15 to 20 cm. On the expansion work of Ponta Delgada Airport, the preliminary study for the embankments recommended the use of slopes with an inclination of 1V:2H. Nevertheless, during the construction of the basalt rockfill materials, it was necessary to extend the width of the runway, which led to the need

8 of carrying out triaxial compression tests in the cell shown in Figure 3. The results obtained allowed to assume as feasible to construct embankments with a greater inclination, of about 1V:1.5H. Actually, it was necessary to adopt, in some zones, slopes with an inclination of about 45 0 (Figure 5). The observation of the 20-year old embankment has demonstrated that the slopes are perfectly stable, even after the occurrence of small earthquakes. Figure 5. Rockfill built in the expansion work of Ponta Delgada Airport Beliche Dam was built in the 80s, a highly brittle rockfill having been used in the zone of the transition material [17]. The dam was instrumented ever since the beginning of construction with inclinometers that made it possible to obtain the vertical internal displacements under construction. The significant settlements observed led to the decision of constructing a berm in the upstream face, at level m, which initially had not been predicted, allowing to install surface marks and to measure vertical displacements by precision levelling, when the first filling of the reservoir was to take place. That procedure proved to be completely adequate because, in winter, when the construction of the dam was under way, the partial and almost instantaneous filling of the reservoir up to a level close to the berm was observed, which led to a very high collapse settlement that exceeded 0.40 m (Figure 6). The settlements were extended, even though with less variation rate, to the operation stage. a) b) Figure 6. Beliche Dam. a) Berm built on the upstream face. b) Settlements measured during the first filling

9 In view of the loss of crest level comparatively with the maximum operation level, it was necessary, in 1998, to increase the crest level by about 1.5 m (Figure 7). High collapse settlements were also observed in abutment embankments of a bridge in an expressway, in Algarve region, which was built with low strength marly limestone blocks, which were highly sensitive to collapse [18]. 8 Conclusions Figure 7 Heightening of the crest of Beliche Dam The results of index-properties tests make possible, on the basis of fairly simple tests, to estimate the mechanical properties of rockfill materials, using the experience of test results obtained on similar materials, as well as correlations and abacuses. A contribution to the physical-mechanical characterization of rockfill materials is also presented. The long term observation of the rockfill behaviour, also allowed the establishment of some correlations with the results of the index property tests for the characterization of the sensitivity of the rockfill to the water, i.e., the collapse. From the experience obtained with the construction of rockfills in Portugal, it is concluded that when the rock materials have a high strength and a reduced weathering grade, the rockfills with better mechanical characteristics are, by decreasing order, as follows: basalt, greywacke, limestone, granite and schist. Nevertheless, when the rock materials are weathered, rockfills, consisting of greywacke, marly limestone and schist, have led to a defective behaviour, as a result of the weakening of rock fragments in the presence of water. Furthermore, it can be considered that granite can also be sensitive to the effects of collapse, but to a lesser extent than the other previously mentioned materials. Basalts presented a reduced weathering grade and, as such, it is almost insensitive to collapse. In brief, it can be concluded that basalts are a remarkable material for rockfill construction, and particularly, as far as limestones, greywackes and schists materials are concerned, it is advisable to determine their ability to lead to collapse settlements. 9 References [1] Veiga Pinto, A. (1983) Structural behaviour forecast of rockfill dams, Laboratório Nacional de Engenharia Civil Thesis, Lisbon, (in Portuguese); [2] Quinta Ferreira, M. (1990) The use of engineering geology in the study of rockfill dams, University of Coimbra PhD Thesis, Coimbra, (in Portuguese). [3] Veiga Pinto, A., Monteiro, B., Quinta Ferreira, M. and Delgado Rodrigues, J. (1995) Properties and classifications on rockfill materials, LNEC Pub., Lisbon, 1-88 (in Portuguese); [4] Quinta Ferreira, M., Veiga Pinto, A., Monteiro, B. and Delgado Rodrigues, J. (2000) Contribution of index tests for the caracterization of rockfills materials, 7º Congresso Nacional de Geotecnia, Porto, Vol. I, (in Portuguese);

10 [5] ISRM (1979) Suggested methods for determining water content, porosity, density, absorption and related properties, and swelling and slake-durability index properties, ; [6] ISRM (1978) Suggested method for determining the uniaxial compressive strength and deformability of rock materials, Int. J. Rock Mech., Vol. 16, nº 2, ; [7] ISRM (1985) Suggested method for determining the point load strength, Int. J. Rock Mech., Vol. 22, nº 2, 51-60; [8] Guifu, X., Hong, L. (1986) On the statistical analysis of data and strength expression in the rock point load tests, Proc. 5 th Int. Cong. of the IAEG, 1.5.7, London, ; [9] Marsal, R. J. (1973) Mechanical properties of rockfill, in Embankment-Dam Engineering, Casagrande Volume, John Wiley & Sons Pub., ; [10] Quinta Ferreira, M., Delgado Rodrigues, J., Veiga Pinto, A. and Jeremias, F.T. (1990) Evaluation of strength of irregular rock lumps for characterization of rockfills, 6 th Int. Cong. of the IAEG, Vol. 4, Amsterdam, [11] LNEC (1971) Los Angeles test, Specification E 237, Lisbon, 1-14 (in Portuguese); [12] ASTM (1981) Chemical analysis of limestone, quick lime and hydrated lime, ASTM C25, Part 13; [13] Parish, D.W., Borden, R.H. (2001) Engineering properties and slake durability of weak Triassic Basin rock, Proc. of the 15 th Int. Conf. on Soil Mechanics and Found. Eng., Vol. 1, Istambul, ; [14] Reiffsteck, P., Blivet, J., Valle, N. and Khay, M. (2001) Écueils de la mesure en laboratoire du comportment mécanique des sols grossiers, Proc. of the 15th Int. Conf. on Soil Mechanics and Found. Eng., Vol. 1, Istambul, ; [15] Veiga Pinto, A. (1982) Rockfil modelling, MS Thesis at UNL, Lisbon, 1-76 (in Portuguese); [16] Pardo de Santayana, F., Fortunato, E. and Veiga Pinto, A. (2005) Behaviour of Portuguese rockfill dams with impervious membranes, Proc. of the 16 th Int. Conf. on Soil Mechanics and Found. Eng., Vol. 3, Osaka, [17] Naylor, D., Maranha das Neves, J., Maranha das Neves, E. and Veiga Pinto, A. (1997) A backanalysis of Beliche Dam, Géotechnique 47, Nº 2, ; [18] Veiga Pinto, A., Papadimitropoulos, I. and Prates, M. (1995) The failure of two approach embankments of the bridge over Rio Seco, Proc. of the 11 th Eur. Conf. on Soil Mechanics and Found. Eng., Vol. 8, Copenhagen,