Relationship between RMR b and GSI based on in situ data

Transcription

1 Relationship between RMR b and GSI based on in situ data F. Ceballos Civil Engineer, Técnicas y Proyectos S.A. (TYPSA), Madrid, Spain. C. Olalla, R. Jiménez Professors, Technical University of Madrid, Spain. ABSTRACT: Due to the large experience accumulated in the use of the RMR b (Rock Mass Rating), as well as to the simplicity of the estimation of the GSI (Geological Strength Index) and to the importance of GSI as input data in the Hoek&Brown failure criteria, both the RMR b and GSI are widely used in geotechnical engineering practice. This article analyzes the relationship between both classifications using in situ data corresponding to different types of rocks collected from different outcrops in Spain. Currently available correlations between RMR b and GSI have been compiled and analyzed in order to compare them with the results of the analysis conducted in this study. Finally, the best (most suitable) statistical relations between RMR b and GSI, depending on the type and quality of rock media, are shown and they are used to establish general correlations. To conclude recommendations are presented, suggesting the use of a particular expression and its limits of applicability. 1 INTRODUCTION Rock masses are generally an inhomogeneous, inelastic, discontinuous, and anisotropic medium; such properties make its characterization rather complex and difficult. Geomechanical classifications, such as the Q index (Barton, 1974), the Rock Mass Rating (Bieniawski, 1973/1979), and the Geological Strength Index (Hoek, 1994) are a common way of defining rock mass behavior, especially in the early stages of a project and for tunnels. These ratings are computed using several observable and measurable characteristics of the rock mass. The most common geomechanical classifications are the one mentioned before (Q index, the Rock Mass Rating and the Geological Strength Index). They use several parameters (type and spacing of joints, unconfined compressive strength, etc.) to provide a single value that serves as a measure of the quality of the rock mass. To be able to use them in practice, it seemed interesting to correlate them using the same rock mass as reference. In this sense, the relation between Q index and the Rock Mass Rating has been widely discussed in the literature, based on extensive experimental campaigns (including as well those proposed by their authors). However, the relationships between GSI and RMR or Q are still not so widely accepted and, even more importantly, they have not been tested by extensive experimental campaigns that confirm their applicability. Therefore, and despite the fact that the goal of RMR and GSI in rock engineering design is significantly different, a campaign of field data collection has been carried out, in several locations within Spain, to develop recommendations for the correlation of these two parameters. This will allow the use of experimental data obtained with the RMR b in the contrast of the results of the numerical calculations based on the GSI, and vice versa. 2 DESCRIPTION OF THE GEOMECHANICAL CLASSIFICATIONS EMPLOYED 2.1 Rock Mass Rating (RMR) This index of rock mass quality was originally developed between 1972 and 1973 by Z.T. Bieniawski; it was based on 8 measurable parameters. Then, in a subsequent review in1979, the number of input data was reduced to 5. In this work, the 1989 version ---which introduced slight changes with respect to the 1979 alternative (see Table 1)--- has been employed. Although the RMR was originally intended for its application in tunnel design only, its use has become common in civil engineering applications, such as the characterization of rock mass strength and deformability and the design of foundations and slopes.

2 2.1.1 Definition The method requires five input data. Four of them are intrinsic to the rock mass, and the fifth is related to the hydraulic conditions. These are: unconfined compressive strength, the RQD, spacing of discontinuities, their state, and presence of water. For its application, the rock mass is divided into several regions with similar characteristics, and these five parameters are quantified separately in all of them (where each parameter has a different weight). Then, the RMR is computed as the sum of their marks, obtaining a final value that ranges between 0 and 100. A correction, which depends on the orientation of the joints with respect to the construction, is needed in certain applications. In this study, however, the uncorrected value was used, because it is independent of the actual construction, hence making it easier to correlate with the GSI. Following the common rock mechanics nomenclature, this uncorrected value is herein referred to as the basic RMR, or RMR b. Table 1. Rating of parameters in different versions RMR 79 RMR 89 Min Max Min Max RQD and spacing of discontinuities Condition of the discontinuities Presence of water 2.2 Geological Strength Index (GSI) The Geological Strength Index was developed by Evert Hoek in 1994, and was slightly but continuously modified throughout the years. This index, which was mainly developed to rate the rock mass in a quick and simple way, and to estimate its strength in the context of the Hoek & Brown failure criterion, (1980, 2002 and 2007) is based on the qualitative description of two elements: the structure of the rock (i.e., the number and the position of discontinuities), and the condition of the surface of the discontinuities. From these descriptions, and using visual charts for reference, the GSI score for the rock mass can be obtained. Hoek and Marinos (2001) have proposed GSI charts for heterogeneous rocks. They also indicate that GSI must only be used with isotropic media. (In rock masses with few discontinuities, whose magnitude is similar to the size of the construction, the rock mass behavior would heavily depend on the tridimensional behavior of individual blocks, and then estimations based on GSI are not valid.) 3 PARAMETERS OF THE ROCK MASSES Fieldwork was conducted to develop a representative database of rock mass parameters. In this campaign two kinds of data were measured for each rock mass: those related to the discontinuities and those related to the (intact) rock matrix. 3.1 Properties of the discontinuities For discontinuities, the following parameters were measured: -Spacing -Roughness -Opening -Persistence -Filler -Water (leaks) All these measurements were made according to the recommendations given by the International Society for Rock Mechanics (ISRM; Brown, 1981) 3.2 Properties of the rock matrix For the rock matrix, the following parameters were measured or estimated: -Uniaxial compressive strength: It was estimated using the point load test (PLT). Because of the dispersion of the values associated to this test, the adjustment ratio (σ c /I s ), was selected according to the type of rock, as listed in Table 2. Table 2. (σ c /I s ) ratios for different rock types Interval Adopted value Igneous Metamorphic (high strength) Metamorphic (low strength) Calcareous Sedimentary (Good cemented) Sedimentary (Good cemented) - Surface Hardness: It was measured using the Schmidt hammer (esclerometer), following the instructions provided by the manufacturer. - Weathering: The recommendations given by the ISRM (Brown 1981) were followed.

3 4 BACKGROUND Before correlating RMRb and GSI, both classifications were compared with the intention of obtaining conceptual similarities and differences between them Parameters In the Rock Mass Rating system, there are five rock mass parameters, which define the final rating of the rock (including the presence of water), each one of them with a certain weight. On the other hand, the GSI is based on the observation and qualitative description of only two aspects: the structure of the rock mass and the condition of its discontinuities. It can be considered that, for this classification, these two parameters have the same weight. Table 3 shows the aspects that influence the final grade in both systems: Table 3. Influence of different parameters RMR b ( 89) GSI UCS 15 0 Structure of rock mass Condition of discontinuities Presence of water 15 0 In this comparison, the parameters of RMR b, RQD and spacing have been included, as sum, in the so called structure of the rock mass Goal One of the main differences between these two geomechanical classifications is their objective. The Rock Mass Rating was designed to give an indicative index of the quality of the rock mass dedicated to the design of tunnels, so that a comparison with a series of historical cases could be possible. Thus, some recommendations for tunnel's lining and reinforcements could be proposed, based on past experience. As it is a design method based on experience, without any further calculations, all external factors such as the orientation of the joints with respect to the construction must be taken into account through correction parameters. The Geological Strength Index was created with the intention of giving a numerical value to define the quality of the rock mass in a very simple way. This quality index is used as input data for subsequent analysis, like other classic geotechnical parameters such as shear strength and deformability. Extrinsic conditions to the work, such as water effects or the orientation of the joints, are taken into account later on by modeling and calculation with numerical methods. 4.2 Existing correlations The possibility of benefiting from the great amount of RMRb data available from real projects, as well as taking advantage of the simplicity of the methods to estimate GSI, makes it worthwhile to try to obtain a correlation between both indexes. It would allow developing numerical models based on the GSI and contrasting the experimental designs based on the RMRb by numerical calculations. There have been several previous efforts to correlate both magnitudes. One of the most used correlations, based on the experience of different authors, is the following (Hoek et al, 1995): RMR >23 GSI=RMR 5 RMR <23 The RMR should not be used to directly estimate the GSI. It should be noted that, for these relationships, the RMR' expression is used. This implies that a dry condition has been assumed or, equivalently, that a value of 15 was assigned to the parameter related to water. In this way, the action of water is not duplicated if (or when) considered in the numerical model. It's noteworthy to mention the application's limit of this relation for very fractured and weathered rocks, in accordance with the problems related to the use of the RMR b for this kind of outcrops. 4.3 Developing a field report model As a preliminary step to the fieldwork, a specific field datasheet was developed for this work. Its purpose was to allow the user to record, in an easy way, all the necessary parameters to estimate RMR b and GSI. The field datasheet employed by the Geotechnical Department of CEDEX (Centro de Estudio y Experimentación de Obras Públicas, Ministerio de Fomento Spain,) was taken as a reference. 5 USED ROCKY OUTCROPS 5.1 Criterion for selection The main criterion to select the geomechanical stations employed was that they should be representative of different rock masses and, due to budget limitations, located in a nearby area. Additionally, to be able to develop accurate correlations for all rock mass qualities, it was intended that they should cover a wide range of rock mass qualities, from rock masses of very high quality (with low weathering and few fractures), to outcrops representative of the boundary between rock (V degree of weathering) and residual soil. Experience suggested that the uniaxial compressive strength (UCS) is the most difficult parameter to obtain. As discussed in Section 3.2, estimates

4 based on both the Schmidt hammer and the point load test (PLT) were obtained. This allowed to identify and discard invalid values (in that case, the value recorded is the one that provides closer results to the strength values estimated with the geological hammer); and, if both values do not differ significantly, it allows to take their average as the representative strength value. Regarding the presence of water, the specific status at the time of observation has been considered, therefore not necessarily taking into account the worst case scenario. That is because, although the worst scenario should probably often be assumed for design, it should not be considered when establishing the correlation between the two classifications. 5.2 Description of the Database Geomechanical stations were classified by type of rock: Sedimentary, metamorphic and igneous, respectively. This classification was made, not only through direct observation, but with the support of maps published by the Geological Institute of Spain (IGME) about the geology of Spain (mainly the MAGNA50 series, with 1: scale). Similarly, each geomechanical station was classified based on its type, following these four criteria: Natural, quarries, slopes and others (tunnels, excavation of foundations, etc.). This classification allows the development of different correlations between RMR b and GSI, and the assessment of whether they are different depending on the origin of the data. 5.3 Data used in this study 59 data pairs (RMR b and GSI) from the following types of geomechanical stations were employed: - NATURAL: In this category the naturally rocky outcrops, without any man-made influence, were included. Five stations have been made in this category, with four of them on sedimentary rock, and only one on metamorphic rock. - QUARRIES: Stations in the vicinity of quarries are included in this category (for both active and abandoned quarries). These outcrops have been heavily affected by human action, so that many of their existing discontinuities can be due to such human effects. This group has six stations, four of them on igneous rock and two on sedimentary rock. - SLOPES: It is the largest study group. These outcrops have been created by human action, normally in association with excavations for road works. Thirty one stations are included in this group: nine on metamorphic rocks, six on igneous rocks and sixteen on sedimentary rocks. - OTHERS: This group was considered in order to include other sources of data. However, in this particular work, it was not possible to take data from tunnels, excavations for foundations, or other types of outcrops different to the previously mentioned categories. In addition to these directly acquired data, RMR and GSI values from other published studies, as well as from geotechnical studies of different projects, have been used. This comprises a total of 17 pairs of new data, distributed as follows: four on natural outcrops and thirteen on other outcrops (excavations, tunnels, etc.). Four stations correspond to igneous rocks, eight to metamorphic rocks and five to sedimentary rocks. Tables 4 and 5 summarize all the data used and their distribution by material and type of outcrop. (Note that, although there are more sedimentary rocks, it can be assumed that there are enough values of all rock types. Also note that samples from slopes clearly dominate over the other sources.) Table 4. Data used by material Nº pairs % Igneous Metamorphic Sedimentary Table 5. Data used by source Nº pairs % Quarries 6 10 Natural 9 15 Slopes Others CORRELATIONS BETWEEN RMR AND GSI After the data was filtered and classified, a statistical analysis was carried out. This analysis was performed from three points of view: - Considering all samples together, without differentiating type of rock or source. This was made with the intention to define a generally valid formula for correlation between both indexes. - Considering each rock type separately, for a better fit of the correlation, obtaining formulae with validity limited to each material. - Distinguishing qualities, to test the limits of validity, depending on the range of values of RMR b or GSI in which the rock masses were found. 6.1 Main results The most significant results from the statistical analysis of the data are shown below.

5 In the analysis considering each rock type separately, a very good correlation was obtained for ig- After testing different types of formulae for fitting, it could be concluded that a linear model fits well to the available data. See Equation (1) and Figure 1. neous rocks (R²=0.954, n=14). The correlation is also good for sedimentary rocks (R²=0.936, n=24) and, with a larger deviation, for metamorphic rocks (R²=0.848, n=20). The corresponding formulae for GSI=1,13RMR 11,63; R²=0,89 (1) linear correlation are: (1) GSI=1,08RMR-10,44 ;Igneous (4) GSI=0,95RMR-10,44 ;Metamorphic (5) GSI=1,30RMR-20,19 ;Sedimentary (6) A third analysis separates the data according to the rock mass quality. For low values (RMR b <30), the RMR b is always bigger than the GSI (see Figure 3, left). For average RMR b values (30 <RMR b <60), most data values are in a range between RMR b =GSI+10 and RMR b =GSI-10 (see Figure 3, right). Finally, for rock masses with higher quality (RMR b > 60), the fit is fairly accurately for the line GSI=RMR b. Figure 1. Lineal correlation between RMR b and GSI, for all 59 pair of values. Likewise, a similar analysis was performed using RMR', in which the parameter of water was always assumed to be 15 (dry conditions). For this case, Equation (2) is obtained: GSI=1,17RMR -11,36 ; R²=0,864 (2) As observed, there is a slight increase in the dispersion, with a small decrease of the linear correlation coefficient. See Figure 2, where the line currently used to correlate both magnitudes has also been included, i.e.: Figure 3. Values of RMR b -GSI for poor quality (left) and medium quality (right) The results of the statistical analysis for these populations are not representative, since the values are concentrated in a certain range (for example, from 30 to 60), so the correlation coefficient values are always low. GSI=RMR -5 (3) (3) 7 RECOMMENDATIONS AND CORRELATIONS Figure 2. Lineal correlation between RMR and GSI, for all values. 7.1 Recommendations One of the main differences (which has great impact when attempting to correlate both indexes) is the goal: the RMR defines the quality of the rock mass for a given use (e.g., to propose a valid support in tunnels), while the GSI serves mainly as input data to estimate the strength and deformability of rock masses. This implies that, when correlating both indexes, extrinsic aspects to correct the basic RMR value, such as the orientation of discontinuities, should not be considered. Similarly, the water parameter in the RMR b is a conflictive point; for instance, when using the GSI, the hydraulic conditions should be taken into account by the numerical model.

6 7.2 Proposed correlations Based on the analysis above, recommendations for the correlation of both indexes can be proposed. The following two options are considered: using the value of RMR b, or RMR'. In both cases, most of the values are within the range defined by GSI=RMR+5 and GSI=RMR-15, as shown, for the case or RMR, in Figure 4. Therefore, the following formulae are proposed: GSI=RMR b -5 (7) GSI=RMR -5 (8) These formulae, with significance for practical application, are similar to the ones proposed by other authors. Figure 4. Proposed correlation between RMR and GSI From a statistical point of view (besides the simple correlations described before), the most approximate formula is the one shown in Equation (1). 8 CONCLUSSIONS AND DISCUSSION finally (3), subjectivity in the measurement of GSI makes difficult the correlation with other indexes. 9 REFERENCES Barton, N, & Lien R. & Lunde, J Engineering classification of rock masses for the design of tunnel support (eds). Oslo (Norway): Springer-Verlag. Bieniawski, Z.T Engineering rock mass classification (eds). Pennsylvania (USA): Wilry-interscience. Brown, E.T Rock characterization testing and monitoring (eds). London (England): Pergamon Press. Deisman, N. & Khajeh, M. & Chalaturnyk, R. J., Using geological strength index (GSI) to model uncertainty in rock mass properties of coal for CBM/ECBM reservoir geomechanics. I International Journal of Coal Geology 112, Hoek. E Practical rock engineering (ed.). Online version: Rocscience. Hoek. E, Marinos, P Estimating the geotechnical properties of heterogeneous rock masses such as flysch. Bulletin of Engineering Geology and the Environment Marinos, P. & Hoek, E.& Marinos, V., Variability of the engineering properties of rock masses quantified by the geological strength index: the case of ophiolites with special emphasis on tunneling. Bulletin of Engineering Geology and the Environment, vol. 65, Serrano, A Mecánica de rocas, Volumen I & II (ed.). Madrid (Spain): Escuela Técnica Superior de Ingeniero de Caminos Sonmez, H. & Ulusay, R. & Gokceoglu, C Indirect determination of the modulus of deformation of rock masses based on the GSI system. International journal of rock mechanics and mining sciences, Tzamos, S. & Sofianos, A.I., A correlation of four rock mass classification systems through their fabric indices. International Journal of Rock Mechanics and Mining Sciences, vol. 4, Zekai Sen & Bahaaeldin H. Sadagah, Modified rock mass classification system by continuous rating. Eng. Geol., 67, Based on a compiled database of GSI and RMR b values from different locations in Spain, correlation formulae are proposed to correlate GSI and RMR b. A linear fit is observed to be acceptable; the formula that provides the best fit is the one shown in equation (1). Other regression analyses conducted to consider different rock types, rock origin, or rock quality also seem to provide good fits. Although, as previously described, these correlations can be accepted in practice (at initial stages of a project), it is important, when using such formulae, to take into account the following aspects: (1) extrinsic parameters to rock masses should not be considered when evaluating RMR; (2) there are conceptual differences that exist at the origin of RMR and GSI that could affect the computed estimates; and