Eurock '96, Barla (ed.) © 1996 Balkema, Rotterdam. ISBN 90 5410 843 6 Analysis of rock fragmentation with the use of the theory of fuzzy sets L'analyse de la destruction de la rüche a la base de la theorie des ensembles flous Analyse der Gesteinzerstörung mit Verwendung der Theorie der Fuzzy Mengen L. L. Mishnaevsky Jr & S. Schmauder - MPA, University of Stuttgart, Germany ABSTRACT: The damage evolution in loaded rock is modeled on the basis of the methods of the theory of fuzzy sets. The heterogeneity of rock as well as damage are characterized by the membership functions of the rock into fuzzy sets of materials with given properties and failed materials, respectively. On the basis of developed concept of the fuzzy damage parameter, the damage evolution in heterogeneous rock is simulated. The influence of initial rock heterogeneity on its strength is studied numerically, and it is concluded that the more heterogeneous is the rock, the less its strength. RESUME: L'evolution des endommagements dans la roche heterogene est etudie. Un modele mathematique de la evolution des endommagements dans la roche , que s' etablit a la base de la theorie des ensembles flous , est elabore. L'influence de la heterogenite de la roche a la resistance limite est analyse. On montre, que plus heterogene est une roche, moins la resistance limite de la roche. ZUSAMENFASSUNG: Die Schädigungsevolution in Gestein wird mit Methoden der Theorie von Fuzzy Mengen analysiert. Das Schädigungsniveau im Gestein wird mit Fuzzy Mitgliedschaftsfunktion characterisiert. Auf Grund des entwickelten Modells des Fuzzy Schädigungsparameter, wird die Schädigungsevolution in heterogenen Gestein simuliert. Der Einfluß der anfänglichen Heterogenität des Gesteins auf seine Festigkeit wird numerisch untersucht. Es wird gezeigt, daß mit steigender Heterogenität des Gesteins die Festigkeit abnimmt. 1 INTRODUCTION The efficiency of rock drilling is determined by the processes of rock fragmentation to a large extent. The fragmentation of rock depends on the conditions of loading as weHas on the rock properties (in particular, strength). Any data about the strength of rock are averaged and uncertain; it is a typical case, when the difference between the measured and predicted properties of rock is more than 40% , or when the strength of two specimens from the same material differ sufficiently. It is caused by the fact that rocks are usually highly heterogeneous materials, consist on different minerals and their formation during many thousands of years was determined by random processes. In order to improve the efficiency of rock frag- 735 mentation, mathematical models of this process' which describe all peculiarities of the rock destruction, are needed. Yet, only the models of single crack propagation (fracture mechanics) or the evolution of distributed microcracks (the continuum damage mechanics, see Lemaitre, 1992) can be applied efficiently to describe the rock destruction. Yet, such processes as rock crushing or cracks system evolution are of great importance in drilling (see review by Mishnaevsky Jr 1995b). For exampIe, the energy consumption in drilling is determined by the rock crushing to a large extentj about 80 % of the loading energy is spent just for it . One of the ways to improve the efficiency of drilling is to use the interaction between large cracks under cuts (Mishnaevsky Jr 1994). So, the models of rock destruction which take into account not only single objects (like the fracture mechanies), and are applicable not only in the region up to crack formation (like the continuum damage mechanics) are needed to describe the rock fragmentation in drilling. Here, it is suggested to use the concept of the theory of fuzzy sets to model the destruction of heterogeneous rocks, and to investigate interrelations between rock strength and heterogeneity. Traditionally, the methods of the theory of fuzzy sets are applied to operate quantitatively with some linguistic, inexact or uncertain values (see Yao, 1979, Brown, 1980, etc). For example, Brown (1979) has obtained theoretically the safety measure (probability of failure), which includes both subjective and objective information about considered system, and the subjective information is determined with the use of the fuzzy sets theory on the basis of the linguistic data. Yao (1980) has suggested to use the descriptive words of specialists to 0 btain the membership function of a structure in a corresponding fuzzy set . Blockley and Baldwin (1987) have considered an applicability of the fuzzy logic models in computer knowledge bases. Lu Ping and Hudson (1993) have used the fuzzy sets in order to evaluate the inexact data about the stability of underground excavation. Shiraishi and Furuta (1983) developed a methodology for the evaluation of subjective uncertain factors (like mistakes,etc) on the basis of the fuzzy sets theory. So, the main direction of the application of the fuzzy sets theory in the engineering science is to use the descriptive, subjective or inexact data in order to obtain the quantitative assessments of safety, reliability or other characteristics of objects. Yet, the theory of fuzzy sets can be used not only to relate some qualitative and quantitative values, but also to describe uncertain relations or characteristics which are available in nature. One of the examples of an application of the fuzzy sets concept to describe a natural uncertainty. is the model of cyclic plasticity developed by Klicinski (1988). This paper seeks to develop a model of rock damage, which allows for the natural uncertainty of local rock properties as well as the microfracture processes, and to study the influence of the rock heterogeneity on the efficiency of drilling. 2 HETEROGENEITY AND UNCERTAINTY OF ROCK PROPERTlES Rocks are highly heterogeneous materials, and their heterogeneity has a pronounced effect on the efficiency of drilling. To study the destruction of such material, one needs to develop a method of description of the heterogeneity. Consider a homogeneous material with strength u. It can be taken as an element of a set of such materials with the membership degree 1. Yet, if one considers a rock, this set presents a fuzzy set: the strength of roc.k depends on size, conditions of loading, components, history of rock formation, origin, etc. So, the membership degree of a given rock specimen into the set of the rocks with given properties is less than 1. One can define the membership function X(u), which characterizes the closeness of behaviour (strength) of the rock to a homogeneous material with given properties. Usually, rocks consist on a number of components (minerals): for example, granite consists on grains of quartz, feldspar, etc. If one designates the strengths of each mineral as U1, U2, etc, the properties of rock depend on all these values, and are similar to a homogeneous materials with the strength U1 or U2, etc to a some (most often, very small) degree. So, one can define a membership nmction X(Ui) , which characterizes the role of each component in the rock behaviour. The heterogeneity of a material can be characterized as well by the statistical entropy of local properties (strength) of the material. Based on the results by Kauffmann (1977), one can write the relation between the function X( u) and this statistical entropy: H = -(l/N) Li Pi (Ui) In Pi (ai) (1) where H - the statistical entropy of local properties (strength) of the material, Ui - the strength of each mineral, Pi(Ui) = X(Ui)/L:i(X(Ui). It is evident, that the more heterogeneous is the rock, the less is the value X(u), and the greater the statistical entropy of the material properties. Thus, the function X(Ui) and the statistical entropy of local strength of the rock characterizes the heterogeneity of the rock, and the role of each constituent of the rock in its behaviour under loading. 736 3 DAMAGE PARAMETER AS A MEMBERSHIP FUNCTION OF MATERIAL INTO FUZZY SET OF FAILED STATE The damage mechanies, as differentiated from other, more traditional methods of modelling of destruction (the fracture mechanics or statistical theory of strength, for example) allows to describe the dependence of rock destruction on time, and thermodynamical aspects of destruction. Yet, the damage evolution is described in the framework of the damage mechanies only up to the formation of a macrocrack (Lemaitre 1992). As said above, the efficiency of rock fragmentation in drilling is influenced mainly by the rock crushing or cracks system evolution, what is outside the region of the applicability of the damage mechanics. Here, it is suggested to formulate the concept of damage, which is based on the fuzzy sets approach, and is applicable not only to the evolution of distributed microcracks. Consider the region of behaviour of rock, which is limited by the condition of local fracture. It is clear, that the condition of local fracture of heterogeneous rock is uncertain. So, this region presents a fuzzy set. The complement of this set (Le. a region of failed state of rock) is a fuzzy set as weIl. One can define the membership function of the rock into the fuzzy set of failed state " which depends on the applied stress, and characterizes the degree of c!oseness of the body to failure (Mishnaevsky Jr 1995a). In the continuum damage mechanics, the damage parameter is considered often as a measure of closeness of loaded body to failure. This meaning of the damage parameter corresponds to this membership function, which we just defined, and that is why we prefer to use the short term "fuzzy damage parameter" instead of "membership function of the material into the fuzzy set of failed state" . The function , can be considered as some kind of generalization of the probability of failure, but it characterizes not the possible, but present state or kind of behaviour of rock, and this state of rock can lie not only between elastic and failed one, but between cracked and crushed rock as weIl. In general, the function , can be related with any continuously growing from monolithic to failed state value (Mishnaevsky Jr 1995a), or determined with the use of people experience and knowledge (Yao 1980, Brown 1979, etc). Yet, it follows from the physical nature of fracture, that the change of 737 the nmction , proceeds only in the direction from ,=0 to 1. So, this value characterizes the degree of irreversibility of the evolution of the material, and can be related with the accumulated entropy per unit volume of material (Mishnaevsky Jr 1995a). 4 DAMAGE EVOLUTION AND HETEROGENIZATION OF LOADED VOLUME Suppose that the fuzzy damage parameter of a rock volume is equal to ,I. Then, if this volume of rock is loaded, and this load can produce in nondamaged rock the fuzzy damage ,2, the summary damage in the rock is calculated by the following formula (Kauffmann 1977): (2) 'I where MAX is the maximal value from and ,2' As differentiated from the probabilities offailure or reliabilities of elements, membership functions are not added. It means that if the rock specimen contains some number of microcracks and some number of microcracks is added, the state of rock does not changed. This conclusion (which is evidently not correct) is drawn, since eq.(2) does not allow for the change of rock properties due to the damage accumulation and the fact that the function , depends on the local strength of material. Taking into account the dependence of , on local strength, one can present this membership nmction as a conditional one, depending on the strength of rock. So, the fuzzy damage parameter in heterogeneous rock is determined by the formula of conditional membership function (Kauffmann 1977): ,= 1- MIN[l-,(la),x(a)] (3) where ,(la) is a conditional fuzzy damage parameter for given strength of rock, x(a) is the membership function, which characterizes the heterogeneity of the strength of rock. Eq.(2) gives the relation between the heterogeneity of material properties (x(a)) and the damaged state in the m,aterial. 1.0 initial (Xo) and damage-induced (1 - ,) heterogeneity, the function x(ai) is determined as an algebraic sum of these two functions. On rearrangements, one can obtain: ~ 0.8 cu ~ 0.6 Lt 0.2 "0 >N 0.4 N Eq.(4) relates the heterogeneity of rock and the damage distribution. Substituting eq.(4) into eq.(3), one can obtain the damage evolution law in recursive form. In so doing, one should note that the values , in right and left sides of eq.(4) correspond to different instants of time. So, the damage evolution law shoud be rewritten as folIows: 0.0 0.0 2.0 4.0 6.0 8.0 10.0 Time (timesteps) Figure 1. Fuzzy damage versus time Consider the damage evolution in rock under loading. Let the initial heterogeneity of the rock be characterized by the function Xo (o"i). Due to the damage evolution, the local strength of rock is changed as weil (Lemaitre 1992). -- C (j) Q) :c äi •... g>- 0.139 0.149 0.144 0.134 -0.51.5 3.5 5.5 7.5 9.5 Time (timesteps) where t and t + At mean two successive timesteps. Fig.l shows the fuzzy damage parameter versus time (in timesteps). The value, was calculated by the formula (5); the rock was supposed to consists on 15 different component. The values X and , for one of these components are equal to 0.5 and 0.002, respectively (it means, that this constituent is rather strength and determines the rock properties to a large extent). For all other minerals, , and X are equal to 0.04 and 0.06 (Le., these minerals are relatively soft). For each timestep, the statistical entropy of local strength of rock was calculated by the formula (1). Fig.2 shows the statistical entropy versus time dependence. One can see that the heterogeneity of rock increases with increasing damage parameter. 5 INFLUENCE OF ROCK HETEROGENEITY ON STRENGTH Figure 2. Stat. entropy of strength versus time So, the heterogeneity of damaged rock is determined both by the initial heterogeneity and by the heterogeneity caused by the damage distribution. Ir the rock is homogeneous initially, and only the heterogeneity of damage distribution determines the heterogeneity of loaded rock, the fuzzy set of given properties of rock presents a complement of the fuzzy set of the damaged state of rock, and function X can be determined as folIows: x(ai) = 1- ,(lai). Ir the rock heterogeneity depends both on the 738 The developed model allows to investigate numerically the influence of rock heterogeneity on strength. We considered about 250 types of rock, with different properties and distributions of constituents in them. Each rock consists on 15 components; each component is characterized by the values X (Le. the degree of variation of the strength of the component from the rock strength) and , (i.e. the fuzzy damage parameter for this mineral at given load). The values , for each constituent were taken as random numbers; in order to ensure varying the entropy of rock strength, the values X were taken to be proportional to im, where i - the number of considered constituent of the rock, i =1,2, ... 15, and m - the number of considered type of rock, m = 1,2,...250. The statistical entropy of strength distribution and the fuzzy damage parameter were calculated by formulas (1) and (3). Fig.3 shows the fuzzy damage parameter of rock (at given load) versus the statistical entropy H of local rock strengths. 1.0 ~ 0.8 CU ~ 0.6 Lt 0.2 "0 >- 0.4 N N 0.0 0.00 0.05 0.10 0.15 0.20 Rock strength entropy Figure 3. Fuzzy damage versus rock strength entropy One can see that the more heterogeneous is the rock, the eloser it to failure, at the same loads. It is of interest to consider the scale efffect on rock fracture in the context of the fuzzy concept of damage. One can see that this effect can not be explained directly, like it can be done in the statistical theory of strength: the membership function of a system from independently loaded elements with the same membership functions does not depend on the number of elements (Kauffmann 1977) (in contrast to the reliability or probability of failure of this system). Yet, taking into account that the fuzzy damage parameter increases with increasing the heterogeneity of rock, one can suppose that the increase of size of a specimen leads most often to the increase in the rock heterogeneity : rocks have heterogeneous structure at severallevels (for example, anisotropy, granularity, texture, layers, etc), and the greater the size of used specimen, the more levels of structure influence on the strength of material and form its heterogeneity. Thus, the greater is the specimen, the greater is its heterogeneity and, consequently, the less its strength. 739 6 HETEROGENEITY OF ROCK AND EFFICIENCY OF ROCK FRAGMENTATION It follows from Fig.I-2, that the damage growth rate decreases when the fuzzy damage parameter is rather large and exceeds approximately 0.85. The region in which the damage growth rate is maximal, corresponds to the maximal heterogenization of loaded volume of rock due to the damage accumulation. It can be compared with the description of the stages of rock fragmentation, given in the review by Mishnaevsky Jr (1995b): the heterogenization of rock in damage accumulation, which is proved here theoretically, was observed in experimental investigations as weIl. The loaded volume of rock at last stages of rock deformation contains the zone of crushed rock, the system of large cracks and distributed microcracks. About 70-80 of the energy of loading goes into the zone of crushed rock . It means that the heterogenization of rock volume due to the damage localization (Le. the dividing of loaded volume into the zones with different degree of destruction) leads to the increase of the energy consumption in drilling, and, consequently, decreases of the efficiency of drilling. One can see from Fig.l, that the damage growth rate decreases at large damage parameter. This effeet can not be predicted on the basis of Lemaitre's or Kachanov's damage evolution laws (Lemaitre 1992), but the analogous result was obtained by Rossmanith et al (1995) on the basis of the computer simulation of the impact destruction of rock by the method of discrete element. The same dependence between the degree of rock destruction and the rate of damage growth was obtained experimentally (see review, Mishnaevsky Jr 1995b): at last stages of rock destruction, the zone of crushed rock is formed in the vicinity of the rock/tool contact surface and almost all energy of loading goes into this zone; the specific energy needed for rock crushing is much more than that for crack growth, and that is why the damage growth slows down. It is seen from Fig.3 that a heterogeneous rock is destructed more intensively than homogeneous or less heterogeneous one, at the same load. Since the heterogeneity of rock can be achieved through the heterogeneous damage distribution (or it can be heterogeneous initially, what can not be controlled), one can conelude that in order to achieve the maximal degree of rock fragmentation, one should form maximum heterogeneous stress field in loaded rock. This conclusion can be compared with the results of Mishnaevsky Jr (1996), who has shown that the high efficiency of drilling can be achieved by high heterogeneity of conditions of loading. Practically, in order to obtain the heterogeneous distribution of damage in rock volume, which is to be destructed and removed, one can load this valume before drilling (pre-Ioading) and form cracks in it (Mishnaevsky Jr 1994). So, the improvement of efficiency of rock fragmentation can be achieved by pre-cracking or pre-Ioading of the rock volume. REFERENCES Blockley, D.L, and Baldwin, J.F. 1987. Uncertain Inherence in Knowledge-Based Systems, Journal of the Engineering Mechanics Division, Vo1.113, NOA, ppA67-481 Brown, C.B. 1979. A Fuzzy Safety Measure, Journal of the Engineering Mechanics Division, Vol. 105, No. 5, pp.855-871 Kauffmann, A. 1977. Introduction a la theorie des sous- ensembles flous. Paris: Mason Klisinski, M. 1988. Plasticity Theory based on Fuzzy Sets. Journal of the Engineering Mechanics Division, Vo1.114, NoA, pp.563-582 7 CONCLUSIONS It is shown that the methods and concepts of the theory of fuzzy sets can be efficiently applied to study the damage evolution in heterogeneous rocks. The damage accumulation in loaded rock leads to the heterogenization of the rock, i.e. to the formation of zones with sufficiently different level of damage (zone of crushed rock, cracked zone, etc, see Mishnaevsky Jr 1995b). The heterogenization of rock, especially, the formation of the zone of crushed rock, gives rise to the energy consumption in drilling, and decreases the efficiency of rock fragmentation. It is shown that the more homogeneous is rock, the more its strength. Practically, it means that in order to increase the efficiency of rock fragmentation, the rock should be made more heterogeneous by formation of damage or cracks in the rock valume, which should be removed. Such pre-cracking allows to improve the efficiency of rock fragmentation. ACKNOWLEDGEMENTS The author (L.M.) is grateful to the Alexander von Humboldt Foundation for the possibility to carry out the research project in the University of Stuttgart, MPA (Germany). The author (L.M. also) was first introduced to the theory of fuzzy sets by Doz. Dr. H.P.Rossmanith, Institute of Mechanies, Technical University of Vienna (Austria). Interesting discussions with Dr.Rossmanith during my work in the Photo and Fracture Mechanics Laboratory, TU Vienna, and his valuable advices are gratefully acknowledged. 740 Lemaitre, J. 1992. A Course on Damage Mechanics. New York: Springer Lu Ping and Hudson, J.A. 1993. A fuzzy evaluation appreoach to the stability of underground excavation. In Ribeiro e Sousa and M.Grossman (eds), Proc. Int. Conf. Eurock-93. Rotterdam: Balkema. Mishnaevsky Jr, L. 1994. Investigation of the cutting of brittle materials. Int.J. Mach. Tools. Manufact., Vo1.34,No.5:499-505 Mishnaevsky Jr, L. 1995a. On the applicability of fuzzy sets theory for the description of the damage and fracture of solids. In: Proc. 32nd Ann. Mtg. Soc. Eng. Sei., pp.271-272, New Orleans Mishnaevsky Jr, L. 1995b. Physical Mechanisms of Hard Rock Fragmentation under Mechanical Loading: A Review. Int.J. Rock Mech. Min. Sei. Vo1.32,No.8: 763-766 Mishnaevsky Jr, L. 1996. A New Approach to the Design of Drilling Tool. Int.J. Rock Mech. Min. Sei. Vo1.33,No.1: 97-102 Rossmanith, H. P., R. E. Knasmillner and L. Mishnaevsky Jr. 1995. Impact induced damage in rock. In H. P. Rossmanith (ed), Proc. 2nd Int. Conf. on Mechanics of Jointed and Faulted Rocks. Rotterdam: Balkema Shiraishi, N. and Furuta, H. 1983. Reliability Analysis Based on Fuzzy Probabilities, Journal of the Engineering Mechanics Division, Vo1.109, No.6, pp.1445 - 1459 Yao, J. T. P. 1980. Damage Assessment of Existing Structures. Journal of the Engineering Mechanics Division. Vo1.106, NoA, pp.785-799
© Copyright 2025 Paperzz