In 1975 Congress passed the Energy Conservation Act to establish a U.S. Strategic Petroleum Reserve (SPR) with a capacity of 750 million barrels of crude oil. The most economic storage medium was determined to be salt caverns leached in salt domes in Louisiana and Texas. Salt caverns existed at several sites when the reserve was created. These were obtained by the U.S. Department of Energy (DOE) and used to initiate SPR oil storage. In order to meet the storage capacity approved by Congress, new caverns also had to be leached. To support the resulting design effort, finite element computer programs have been used to determine the creep closure and structural stability of salt caverns. Using site specific material properties including creep models, elastic moduli and fracture data, the finite element analyses have replaced earlier empirical approaches to cavern design. This report presents results of such finite element analyses to determine the best cavern roof shape and the minimum pillar to diameter ratio, P/D. These numerical predictions indicate that the current cavern design is safe.
FINITE ELEMENT AND MATERIAL MODELING
The finite element program used in this study has existed for approximately four years. Its theoritical basis (Key, et al, 1980) and application are well documented. It was originally developed to predict the creep around mined openings in bedded rock salt for the Waste Isolation Pilot Plant (WIPP). The program was tested by modeling the creep of salt around a drift and by comparing the results with the predictions of eight other structural computer codes (Morgan, et al, 1981). Since then the program has had continual use for calculations to support the-WIPP (Miller, et al, 1982). For about two years the program has also been used to calculate the creep and structural stability of SPR salt caverns (Preece and Stone, 1982). Other applications include the analysis of laboratory creep tests that are used to develop creep models for salt (Branstetter and Preece, 1983). The program numerically simulates the creep of rock salt and the resulting stresses, strains and displacements as functions of time. The constitutive model used has elastic and creep components where strain rate is a power function of effective stress as given below. (mathematical equation) (available in full paper)
The elastic constants and the creep parameters used are listed in Table I. They are based on extensive creep data on rock salt from the Salado formation in New Mexico and on limited but site specific measurements on salt samples from West Hackberry and Bryan Mound (Wawersik, et al, 1980a and 1980b). The parameters in Table I either match or overpredict the creep rates that are recorded in laboratory tests. Within the context of this study, overpredictions are considered conservative and acceptable.
Ceramic projectiles made from 94% to 99% alumina with a nominal mass of 3.1 gms were fired into granite blocks. The average compressive strength of the ceramic was 800 MPa and that of the rock was 135 MPa. Flat, ogive, conical and pyramidal nose shapes were evaluated at velocities ranging from 450 to 1400 m/sec. It was found that the conical and pyramidal nose shapes at the higher velocities did a significant amount of damage to the rock. On the average, they removed 10 mm of rock and were six and ten times better than the flat and ogive nose shapes. It was also observed that increases in volume occur in discrete steps. This process is highly dependent on the ability of the projectile to penetrate the rock.
It is a recognized fact that rock drilling is more efficient in a fractured rock than in a homogeneous rock. Several methods to preconditioning the rock or to drill exclusively by novel fracturing methods have been tried, and many of the early attempts have been summarized by Maurer (1968). The technique under study in this work is one where three ceramic projectiles are fired between the rollers of a tri-cone roller bit. All three projectiles are fired from a common chamber, and arrive at the rock face within microseconds of one another. The drilling and firing operations are continuous, and the rate of fire is dictated by the rock type and the thrust and rotation of the drill. Ceramic was chosen for a projectile because it is highly frangible, and thus will not interfere with or destroy the carbide buttons on the roller bits. In addition, the remaining ceramic fragments will be easily circulated out of the drill hole. Fundamental to the success of this hybrid drill is an understanding of how the ceramic projectiles penetrate and fracture the rock, and then to determine what improvements can be made in projectile geometry and properties to enhance the fracture process. Several investigators have examined the relationship between rock and a dynamic penetrator. A study using military projectiles with rock types ranging from granite to sandstone was made by Tolch and Bushkovitch (1947). Vanzant (1962) examined concrete and marble impacted with steel and armor-piercing projectiles. Bauer and Calder (1969) used steel, aluminum and tungsten carbide high velocity projectiles of different geometries with limestone and granite specimens. Kumano and Goldsmith (1982) impacted diorite with variously-shaped steel projectiles. In all of the above studies the projectiles behaved as elastic-plastic members; none of them could be classified as brittle. Newsom (1976) did some preliminary work with metallic and ceramic projectiles in a two step shoot and drill series of tests, and found that the ceramic material was a good penetrator and that drilling rates were improved. The present work examines the behavior and interaction of a brittle projectile with rock.
Ceramic projectiles ranging from 94% to 99% alumina obtained from different manufacturers and produced by a variety of extrusion or grinding methods were used in the tests. They were nominally 31.7 mm long by 6.3 mm diameter and had a mass of 3.1 gms.
Fracture tests were conducted on subsize specimens of Charcoal granite to demonstrate that the extent of the nonlinear region at the crack tip can be significant. The model of the fracture process considers an effective crack length as a traction free portion plus process zone where closing stresses act; this zone is identified by microcracking and interlocking. A singularity is allowed at the effective crack tip. The effective stress intensity factor is the superposition of the stress intensities due to the effective crack and the process zone. By assuming that the closing stress is a linear function, it is shown that a process zone length of approximately one-third the effective crack is needed to match the compliance-determined crack length.
A number of specimen geometries are employed to measure the fracture resistance of rock. The two most common are the notched- beam under three point loading and the compact tension specimen, both standardized by ASTM-E399, the specification for toughness testing of metals. A configuration that is quickly gaining acceptance in the rock mechanics community is the core-type variety, either a bend (Ouchterlony, 1980) or a short rod specimen (Barker, 1977). However, a correction due to size effects may be required for these relatively small specimens (Barker, 1979; Wecharatana and Shah, 1982). Other geometries utilized are the double cantilever beam (Hoagland, et al, 1973), the double torsion specimen (Atkinson, 1979), and the double- edge-notched plate (Ingraffea and Schmidt, 1978; Labuz, et al, 1984).
One problem in measuring a fracture parameter with different specimen configurations has to do with the formation of the process zone and its influence on determining a valid toughness value. If linear elastic behavior dominates (the nonlinear region is negligible), then the critical stress intensity factor (Kic) may be identified as a material property and specimen geometry should be irrelevant. However, depending upon dimensions and grain size of the rock, the process zone can be substantial. To obtain a realistic estimate of the fracture resistance, the effects of the process zone should be considered. This paper demonstrates that the extent of the nonlinear region in subsize specimens of Charcoal granite is significant. Consequently, linear elastic fracture mechanics (LEFM) or a J-integral approach may be inadequate for small specimens of relatively large-grained rock. An analysis that includes the fracture process zone would be better suited to estimate the crack resistance of these types of rock. Linear Fracture Mechanics: Fracture Toughness, Kic An apparent fracture toughness (KO) value can be computed from the stress intensity factor for a particular geometry by knowing the crack length, which is usually found through a compliance calibration, and the appropriate load, which is based on the crack extending a significant measurable amount under linear elastic conditions. Determining a crack length by the compliance technique procures an accurate measurement only when the nonlinear region is small. In addition, materials such as rock exhibit inelastic deformation along with slow crack growth so that the point of significant crack extension may be difficult to determine.
Gob pressure measurements were made in a Western U.S. coal mine as part of a long-term program to evaluate cave progress and to determine the influence of geological discontinuities on caving conditions, load transfer, and resulting instabilities. Gloetzl cells were selected for the measurements due to their simple, robust construction and a history of being able to monitor pressures in a broken medium. A methodology was developed for the successful installation and protection of the cells and hydraulic lines in the gob as the face retreated. The measurements indicated a cave progress controlled by the frequency of major faults. The pressure-mining progress profile was compared to those observed in other parts of the world. It was concluded that the significant differences in profiles was caused by the thick-bedded strata, the existence of high lateral stresses, and the spacing between faults. Recommendations are given for future applications of Gloetzl cells for gob pressure measurements.
When the overburden is undercut through removal of the coal seam, a redistribution of stresses takes place, transferring the weight of the undermined rocks to the panel boundaries. As the undercut dimensions increase, bed separation and roof falls develop. The falls (or the cave) progress up into the roof until the mined space is filled with broken rock. At this time, compaction of the caved rocks occurs, permitting a partial transfer of overburden load to the floor. The amount of load transferred through the gob and the distance beyond the face where full recompaction of caved rocks occurs are of great importance for determination of the stress redistribution to panel boundaries. The load transfer through the gob is a function of large scale geology (i.e., existence of major faults, lateral stress fields, and roof rock formations), as well as small scale structures (i.e., fractures) which affect the size distribution and stiffness of the caved rocks. Thus, the nature and degree of load transfer could vary from site to site. Evaluation of cave progress by in-panel measurements (in the gob, under the panel, or over the gob) is believed to be essential for evaluation of caving conditions and load transfer mechanisms. Such measurements, however, have been very limited due to the difficulties and costs of obtaining in-panel measurements on a large scale. Alternatively, researchers have preferred to perform measurements at panel boundaries, and then speculate as to what is happening in the gob. Difficulties with the latter approach are that the measurements should be obtained on a very large scale since the caving methods influence the strata much beyond the panel boundaries. Problems exist with the reliability of most instruments and the meaning of measurements because of the nonlinear behavior of coal at the face.
Review of In-Panel Measurements
In-panel measurements have been performed for evaluation of cave progress. Generally, they can be grouped into measurements obtained above the gob in the roof, in the gob, and under the panel. A brief description of these measurements follows. Measurements above the gob generally have been made in boreholes drilled from the surface or other openings located above the panel.
An understanding of the nature of yield zones surrounding coal mine excavations is fundamental to the design of mine support systems. Various methods have been designed for investigation of underground rock masses and an extensometric technique, the Magnetic Rod Extensometer, has been developed to overcome inherent design limitations in existing equipment of this type. This paper refers to the prototype form of the extensometer and details changes designed to overcome a variety of drawbacks indicated by routine use of the equipment. The principal change has been one of manual to computerized data acquisition and presentation in the laboratory phase of the system. This has dramatically reduced the time taken to process underground measurements and produce numerical and graphical results. The revised Magnetic Rod Extensometer represents an integrated instrumentation system that combines speed and accuracy in all areas of data handling.
The closure of coal mine excavations results from the development of a failed or yield zone in the surrounding rocks due to redistribution of the primitive stress field. This failed or yield zone presents problems to the mining engineer who is required to design a support system which balances economic and operational factors and allows mining to continue in the presence of this yield zone. An understanding of the nature of the yield zone is therefore a fundamental prerequisite for the design of mine support systems. Various methods have been developed for investigation of underground rock masses using stress measuring instruments, modulus of deformation instruments, and displacement measuring instruments. A careful appraisal of available in-situ instrumentation indicated that displacement or strain measuring instruments provided the most realistic approach to in-situ measurement in the 'soft rock' coal measures environment (Aziz, 1978). Such instrumentation allows identification of the origin of the pressure zone created by mining activity, and the duration and frequency of strata movements that induce excavation deformation. Hence characteristic deformation processes can be established for support system design in representative geological environments. Due to inherent design limitations in extensometric techniques, indicated by a market survey in 1976, a new system termed the Magnetic Rod Extensometer was developed for in-situ measurement of fracture zones surrounding coalmine gateroads (Smart et al, 1977).
Principle of Magnetic Rod Extensometer System
The technique consisted of monitoring the positions of magnetic reference points, i.e. small rectangular magnets installed at predetermined intervals along the axis of a borehole, typically 55mm diameter. The relative positions of the magnets were periodically recorded on a sensing medium which was temporarily inserted into the borehole. The sensing medium was then removed to the laboratory where a scanning arrangement measured the distances between the magnetic signals, these distances corresponding to the distances between reference points in the borehole. The reference points moved with the strata, and thus changing distances between magnetic signals recorded on the sensing medium measured differential movements of the rock adjacent to the borehole. Since 1977, the Strata Mechanics Research Group, Cardiff, U.K., has used the technique on a routine basis in a study of yield zones around arch and rectangular profile gateroads, and characteristic deformation processes have been identified (Isaac and Smart, 1983).
Request to Mine Beneath Bridge The Jakes Run Bridge is a reinforced portland concrete arch filled bridge located on WV Route 7 near Pentress, WV. The bridge has a span of 22.9 m (meters) and a height of 4 m above normal pool elevation to the underside of the center of the arch (See Fig 1). The bridge is a two lane structure with a width of 6.86 m. The structure is over 50 years old having been built in 1929 and is owned by the Department (WV Department of Highways). CONSOL (Consolidation Coal Company) is the owner of the mineral rights to the Pittsburgh Coal Seam under the subject structure, and had projected longwall mining under the bridge. They subsequently approached the Department with a proposal to mine beneath the bridge and determine the measures necessary to insure continuity of service and safety of the travelling public. The average height of mining in the coal seam was 2.47 m. The average depth from the surface to the bottom of the coal seam was approximately 140 m. The bridge structure was centrally located near the end of the longwall panel. The bridge was oriented on a skew of approximately 16 ° to the direction of mining. See Fig 2 for panel dimensions. Mining Method The long wall method was the system employed to mine beneath the structure and leaves no support for overlying strata. Because of this, rapid subsidence is normal immediately after mining. Longwall mining is highly desirable because of the high percentage of recovery, high productivity and conservation of resources.
EVALUATION OF THE SITE AND PROPOSAL
Agreement Conditions CONSOL performed an in-house engineering evaluation of the bridge. It was determined that the bridge could subside to the projected depths without critically damaging bridge or disrupting traffic. In May of 1982, CONSOL approached the Department concerning mining beneath the bridge. In June of 1982, the Department and CONSOL concluded an agreement which allowed miming under the structure. The agreement provided measures necessary to insure the safety and continuity of public travel on the bridge. The following are some of the significant conditions of the agreement: 1 - CONSOL was to install a support system comprised of beams and cribbing under the bridge that would subside with the structure and prevent any catastrophic failure. The contact points of the support were situated a few millimeters under the bridge so as not to impede the settlement of the structure. Approval by the Department was required for all protective measures. 2 - CONSOL was responsible for obtaining temporary detour easements and/or right of ways, and to provide a temporary creek crossing in the event the Department determined the bridge unsafe. The determination of the safety was to be the sole province of the Department. 3 - CONSOL was responsible for repair to the structure after subsidence was complete. 4 - CONSOL was to alleviate damage to the bridge by mining continuously so that the settlement process would be uninterrupted. CONSOL agreed to mine continuously within a distance of 21.34 m in advance of the nearest points of the structure to 33.5 m beyond the furthest point of the bridge.
In this paper the imminent relationship between rock burst and its phenomenon with normal rock pressure and its appearance, as well as the factors of one changing into another have been analyzed. Based on this, the necessary and sufficient appearing factors of rock bursts have been put forward, and the term "burst pressure" has been defined. According to the statistics data of tremendous rock bursts which appear in collieries in Hunan Province, the features of high frequency, great strength and shallow depth of the beginning burst have been found. There are some special structures in the coal seams, which are closely related with the burst phenomenon. A hypothesis of "star structure" has been suggested for the appearing mechanism of bursts. Inside and around this star structure the distribution rule of burst pressure and rock strength has been researched depending upon the theory of the limited stress field. A new description of the appearing causes of bursts has been raised.
THE ESSENTIAL FEATURES OF ROCK BURSTS IN HUNAN COLLIERIES
In Hunan Collieries, rushing phenomenons appear almost by way of bursting. Coal, rock, gas and water all rush out, especially the outburst of coal and gas at Ma-Tian Colliery in April 1955 to the end of 1980, 3,421 bursts had come into being during this period involving 30 mines. A great many of the bursts threatened the mine safety and production seriously as well as the measures taken to prevent them were not very efficient. Miners in China usually called them "mine cancers." The essential features of bursts in Huanan are as follows:
Experimental and theoretical analyses are presented to study the mechanics of growth and interaction for three inclined left-stepping, three inclined right-stepping and two inclined non-stepping cracks in a brittle material subjected to uniaxial compression. It was found that the non-stepping cracks did not interact and they propagated individually in a direction parallel to the uniaxial compressive stress. The propagating cracks formed in regions where tensile stresses developed. The left-stepping cracks interacted and formed branching cracks as a result of failure in tension of the material between them. Also, they propagated from their tips in a direction parallel to the uniaxial compressive stress. The right stepping cracks did not interact, but extended themselves from their tips following a direction parallel to the uniaxial compressive stress. The extended cracks formed in areas of compressive stresses.
Researches have observed that the rock adjacent to slopes, cuts and underground openings, and room and pillar underground mine systems exists in a fissured state, but nonetheless the rock structure comprising the surrounding rock and the overburden mass remains stable (Sture and Ko, 1978). The fissure system develops during the process of excavation when the rock reaches its peak strength. With time, due to tectonic stresses, fluid corrosion or creeping of the fissure system, the fissures may grow and interact, expanding the fracture regions. Therefore failure can occur. Excellent analytical and experimental techniques have been developed with respect to the propagation mechanics of a pre-existing single crack in brittle materials (Brace and Bombolakis, 1963; Hoek and Bieniawski, ' 1965; Ingraffea and Heuze, 1980). However, few experimental and theoretical studies have been conducted on brittle materials that contain more than one pre-existing fissure or crack with varying degrees of separation and overlap. The present study presents an experimental and theoretical investigation of the mechanics of secondary crack formation and growth in prismatic samples of brittle clay with pre-existing cracks subjected to uniaxial compressive stresses.
For the laboratory experiments on brittle materials with pre-existing cracks, prismatic block samples of kaolinire clay were used. Kaolinire clay is a very homogeneous material. This homogeneity is an important factor in that it facilitates the study of crack formation and growth in the samples without the adverse effect of microscale mechanical heterogeneities (Tapponier and Brace, 1976). .The laboratory samples for the testing were obtained from large block samples of kaolinire clay prepared by consolidating clay slurries in a cylindrical container, 30 cm in diameter. The clay samples used for the testing were prismatic block samples 7.2 x 7.2 x 2.8 cm that were cut from the large block samples prepared in the consolidometer. When the prismatic samples were still wet, inclined slits or cracks were artificially made in the samples by inserting and removing thin glass sheets 1 mm in thickness and 2.5 cm in with in a direction normal to the samples' face. After the artificial cracks were inscribed in the samples, they were allowed to dry so their degree of brittleness increased. After the clay samples dried, they were subjected to uniaxial compression.
Rock-mass fracturing was measured on oriented drill core and along scanlines in three mines extracting mineralized porphyry deposits. Fracture families identified in the core were consistent with those found along scanlines, but fracture frequency in the core was twice that measured along scanlines. The excess fractures, which occur within 45 ° of a right angle to the core, apparently are produce during drilling, which implies that fracture spacing in drill core is an ambiguous quantity dependent on the drilling procedures, on the orientation and cohesion of the planes of weakness, and on the actual fractures present in the rock mass.
One objective in logging and testing drill cores is to infer the structural characteristics of the rock mass. The structural stability of an excavation in rock presumably depends in part on the geometry of fracturing or jointing in the rock mass--the number of distinct fracture families, their orientations, the spacing between fractures, and the extent or size of the fractures. Within any fracture family the orientation, spacing, and extent are never fixed values; rather they vary over substantial ranges, just as do the strength and stiffness parameters. The variation is usually describable in terms of a frequency distribution, most commonly the negative exponential or the lognormal. Determining the geometry of rock-mass fracturing from drill core is not a simple task. Unless the core is oriented on recovering it from the drill hole, the fracture families cannot be established, and hence the fracture spacing can be determined only between successive fractures along the core, without regard for fracture family. Fracture extent cannot practically be determined from core that is only 30 to 50 mm in diameter. Nevertheless, many instances arise in which exploratory drill cores, ordinarily unoriented, are available for study. Fracturing measurements from oriented cores and scanline mapping, summarized herein, are used to explain some of the fracture distributions that may be expected in unoriented cores.
DATA ACQUISITION AND RESULTS
Jointing geometries were determined for three ore bodies being mined by undercut-cave methods (Panek and Melvin, 1984), part of an investigation aimed at calculating the cavability of a large mineralized rock mass as a function of the jointing and of the strength and stiffness parameters (Panek, 1981). The San Manuel is a typical porphyry-copper ore body, strong but well fractured; the Lakeshore mine copper-oxide deposit is highly altered and friable porphyry and metasediments; the Henderson mine molybdenite deposit is a competent, tight, granite rock mass. The three deposits provide a wide range of strength characteristics; representative unconfined compressive strengths are 90, 30, and 107 MPa, respectively. At each mine the rock mass was sampled by core drilling three or four NX-size holes in different directions at each of one or two sites, producing 180 to 300 m of core by a different pattern of holes at each mine. A split, double-tube core barrel was used, and hence core recovery was virtually 100%. The core was oriented by making an impression of the core stump remaining in the hole after the removal of each length of core and matching the impression to either side of the break surface.