In most fluid flow problems fractures are subject to some level of effective normal stress and hence the fracture walls are in intimate contact. Considerable efforts (Iwai, 1976; Witherspoon et al., 1980; Engelder and Scholz, 1981; Gale, 1982; Gale, 1984) have been devoted to determining if flow through fractures that are pressed together can be described by the parallel plate law, corrected for roughness, with conflicting conclusions. Contact between the adjoining fracture walls should represent the maximum relative roughness. The initial contact area is a function of the fracture surface roughness and it increases with increase in effective normal stress restricting flow to those areas of the fracture plane where the fracture walls are not in contact. Hence, as a first step in understanding flow in fractures, as a function of normal stress, one must be able to characterize the role of small and large scale roughness in determining fracture closure and energy losses within the fracture plane.
The different scales of surface roughness first received attention in terms of their role in defining shear strength or friction between two rough contacting surfaces. Variations in height and spacing of asperities have received considerable attention from workers who have attempted to develop models for crack or fracture closure under normal stress conditions. Studies in this area include those by Bowden and Tabor (1954), Greenwood and Williamson (1966), Walsh and Grosenbaugh (1979), Swan (1983) and Brown and Scholz (1984) that are based on fundamental mechanics. Attempts to link fracture surface roughness to permeability include the phenomenological models of Gangi (1978) and Tsang and Witherspoon (1981) and the empirical models by Barton et al (1984). While the above studies represent significant contributions, nearly all of the proposed models have dealt with the roughness of one side of the fracture plane. For fractures in intimate contact over the range of normal stresses of interest it is necessary to measure the roughness of both sides of the fracture in order to describe the distribution of pore space and hence the variation in fracture aperture over the fracture plane if one is to derive appropriate relationships between surface roughness and fluid flow. In this paper I report the results of several fracture plane resin-impregnation experiments in which I have been able to measure surface roughness of both sides of the fracture plane and the structure of the pore space at the same effective normal stress conditions for which the fluid flow properties were measured. The data from these and similar experiments will provide the basis for assessing the relevance of existing coupled fracture deformation and fluid flow models to predicting fracture permeability for specific fractures at different levels of normal stress.
2 EXPERIMENTAL DESIGN AND PROCEDURES
Two granite core samples, Hi and STR2, each containing a single natural fracture oriented perpendicular to the core axis, were cored; HI from a surface outcrop at the Underground Research Laboratory, Pinawa, Manitoba, and STR2 from the walls of an experimental test site, 360 m below surface, at Stripa, Sweden.
Wireline hydraulic fracturing (hydrofrac) equipment and procedures were developed for testing in fluid-filled holes in salt formations. Tests were performed in the Rayburns dome in north Louisiana over an interval of depth from 1410-2455 ft (430-748 m). Results indicated that the initial state of stress was isotropic, however stress values were not quantified. Maximum pressures and fracture propagation pressures were essentially identical for the stable pressure-time records of tests. Testing indicated a larger formation resistance to fracture propagation than currently used for design of storage caverns in Gulf Coast salt domes.
Gulf Coast salt domes currently are used extensively for underground storage of hydrocarbons and production of brine. Projects also are well along for disposal of solidified wastes and compressed air energy storage (CAES) in caverns in U.S. salt domes. Containment considerations imply a need for field test data on fluid pressures that initiate and propagate fractures into salt dome formations at different depths, i.e., the "hydrofrac gradient". The primary purpose of the study described herein was to determine the hydrofrac gradient in a typical Gulf Coast salt dome. A secondary purpose was to gain information on the state of stress existing in the dome. As noted previously, the hydrofrac gradient is of considerable interest for storage of pressurized fluids in salt dome caverns, and any-information that can be gleaned on the "undisturbed" or initial-state of stress also is useful for design of underground openings in salt domes. Related similar work on hydraulic fracturing of salt has been per- formed by a number of workers, although apparently no fracturing has been performed at depth in an undisturbed Gulf Coast salt dome (Borjeson and Lamb, 1984; Nelson, et al, 1982; Flach, et al, 1985; and Doe, 1986). Frequent reference will be made to hydraulic fracturing studies in bedded salt performed in the Waste Isolation Pilot Plant (WIPP) by Wawersik and Stone (1985). This work is baseline in character, and thus is convenient for comparison of similarities existing between field studies of hydraulic fracturing in salt.
2. TEST SITE
Tests were performed in the Rayburns salt dome in north Louisiana. The Rayburns dome incorporates a diapiric salt stock that flowed plastically upward from a parent bedded formation, the Louann Salt. The buoyancy driven genesis of the Rayburns dome in a relatively quiet geologic setting, coupled with the small shearing strength of rock salt, furnishes a basis for anticipating that the virgin state of stress will be lithostatic in character, i.e., isotropic and of a magnitude due only to the overburden. The Rayburns dome is approximately 1 mi (1.6 km) in diameter, with top of salt at a depth of 127 ft (39 m) at the test site. The test site consisted of a "corehole" with diameter of approximately 8½ in. (22 cm) and depth of 5002 ft (1525 m) near the center of the dome. Drilling was performed for the U.S. Department of Energy (DOE) in 1978 by Louisiana State University (LSU) as part of a Study to investigate Gulf Coast salt domes for possible nuclear wastes disposal (Hawkins, 1978). Salt cores of 4 in. (10 cm) diameter were retrieved from the upper 2650 ft (808 m) of the corehole and their lithology documented (Nance and Wilcox, 1979).
Turk, N. (Department of Geotechnical Engineering, University of Newcastle upon Tyne) | Greig, M.J. (Department of Geotechnical Engineering, University of Newcastle upon Tyne) | Dearman, W.R. (Department of Geotechnical Engineering, University of Newcastle upon Tyne) | Amin, F.F. (Department of Geotechnical Engineering, University of Newcastle upon Tyne)
The roughness of rock joint surfaces has a marked influence on the shear strength of rock masses. The joint roughness angle is generally estimated either from direct asperity angle measurements or analysis of joint surface profiles. Fractal is shown to be applicable to the characterization of irregular joint surface profiles. According to this theory, rock joint surface roughness is represented by a fractional number called a fractal dimension (D). Additionally, the self similarity concept of fractals suggests that as measuring step length (E) decreases, the total joint profile length (L) increases, as shown by the fundamental relation: Log (L) = (l-D) Log (E). The theoretical relation established between the fractal dimension is given. Fractal dimensions of the ISRM suggested roughness profiles and Silurian sandstone joint profiles were determined and the measured roughness of angles of sandstone surfaces are compared with those estimated. Finally, the limitations of the fractal method of joint roughness determination are discussed.
Rock mass is a jointed medium. The engineering properties of rock masses are influenced by joints which break up the rock mass into small rock blocks. While the joints increase permeability and deformation of a rock mass, they tend to decrease its strength and bearing capacity. However, the influence of joints on rock mass behaviour is controlled by joint properties such as roughness, weathering grade, dip and dip direction, presence of infilling, openings. Of these properties, rock joint roughness has attracted the attention of several researchers over the last 20 years (Patton, 1966; Barton,1973; Fecher and Renger, 1971; Mogilevsakya, 1974; Schneider, 1976; Barton and Chouby, 1977; Wu and Ali, 1978; Bandis et al, 1983; Lamb and Johnston 1985; Turk and Dearman, 1985; Reeves, 1985, and Gerrard, 1986) because of its important influence on the shear strength of rock joints.
Studies have shown that roughness angle increases the apparent shear strength of rock joints (Patton, 1966).
(mathematical equation)(available in full paper)
The effect of the roughness angle of a rock joint surface is dependent upon the normal stress acting across the joint surface. As the normal stress increases, the effect of the roughness angle on the shear strength decreases, but the apparent cohesive strength of the rock joint increases. Schneider (1976) proposed the following relation to estimate the effective roughness angle, under any normal stress, based on experimental findings:
(mathematical equation)(available in full paper)
io can be determined either by direct measurements of the rock joint surface asperity inclinations in the field or by analysing joint profile traces.
Barton and Chouby (1977) proposed an alternative relation for estimating shear strength of rock joints:
(mathematical equation)(available in full paper)
While rock joint compressive strength is determined by using a Schmidt hammer on the rock joint surfaces, the rock joint roughness coefficient is determined by comparing the rock surface roughness with 10 sets of standardized rock roughness profiles ranging from 0 to 20, in steps of 2. This method of roughness determination has also been supported by ISRM (Brown, 1981).
The variation with pressure and temperature of the thermal-expansion coefficient of cracked and porous rocks can be explained by using an asperity-deformation model. This model explains the low and, some- times, negative expansion coefficients measured under confining pressure. These low values are due to the negative temperature variation of the material modulus and not, as has been explained in the past, due to the thermal expansion of the material into the cracks or pores. The measured results for igneous rocks, granites in particular, are explained by the theoretical model.
The thermomechanical properties of cracked and porous rocks are of increasing interest in applications such as nuclear-waste disposal and/or isolation, geothermal-energy extraction, magmatic and/or volcanic activity, and temperature-dependent tectonic behavior. These thermomechanical properties and their variations with temperature and pressure have a strong influence on the mechanical and transport properties of rocks. One of the thermomechanical properties, the thermal-expansion coefficient, has an interesting reported variation with temperature and pressure. At zero confining pressure and high temperatures, it is larger than expected while at high confining pressures it can become smaller than expected, or even negative. The thermal-expansion coefficients of cracked and porous rocks have been measured for many years (Somerton and Selim, 1961; Richter and Simmons, 1974; Cooper and Simmons, 1977; Simmons and Cooper, 1978. See Richter and Simmons, 1974, for references to some of the very early work). These measurements were made on laboratory samples at elevated temperatures (25 to 800°C) and room pressure (1 atmosphere). These measurements indicate the expansion coefficients of (microcracked and porous) rocks are very close to the expansion coefficients of the constituent minerals (materials) at low temperature (<100°C) and they tend to increase at higher temperatures more rapidly than expected on the basis of the known temperature behavior of the constituent materials. The cause of the latter effect is the generation of new cracks in the rocks due to thermal gradients or the anisotropy and mismatch of the expansion coefficients of the constituent minerals. The generation of INTRODUCTION The thermomechanical properties of cracked and porous rocks are Of increasing interest in applications such as nuclear-waste disposal and/or isolation, geothermal-energy extraction, magmatic and/or volcanic activity, and temperature-dependent tectonic behavior. These thermomechanical properties and their variations with temperature and pressure have a strong influence on the mechanical and transport properties of rocks. One of the thermomechanical properties, the thermal-expansion coefficient, has an interesting reported variation with temperature and pressure. At zero confining pressure and high temperatures, it is larger than expected while at high confining pressures it can become smaller than expected, or even negative. The thermal-expansion coefficients of cracked and porous rocks have been measured for many years (Somerton and Selim, 1961; Richter and Simmons, 1974; Cooper and Simmons, 1977; Simmons and Cooper, 1978. See Richter and Simmons, 1974, for references to some of the very early work). These measurements were made on laboratory samples at elevated temperatures (25 to 800°C) and room pressure (1 atmosphere). These measurements indicate the expansion coefficients of (microcracked and porous) rocks are very close to the expansion coefficients of the constituent minerals (materials) at low temperature (<100°C) and they tend to increase at higher temperatures more rapidly than expected on the basis of the
ABSTRACT: A functional relationship between values for the elastic constants and porosity has been developed for tuffs with various degrees of welding. This correlation defines the change in elastic constants with inelastic compaction for wave propagation analysis and also can be used to estimate elastic properties of tuffs with different initial porosity. The ability to relate stress induced changes to unstressed states with different porosity values is new and eliminates the need for a large, costly testing program every time the site changes.
Specification of the elastic constants of porous materials for stress and wave propagation analyses is often more complex than for other types of materials, because of the strong influence of porosity. For example, a change of over a factor of 10 in elastic constant values was observed for volcanic tuffs from Nevada, with porosities ranging from less than 10% to over 40% in their natural states (Butcher & Jones 1987, Price & Bauer 1985). Furthermore, a large change in porosity can occur in these rocks over vertical and horizontal distances of meters, complicating characterization of underground test and construction sites. Site characterization is costly, because large numbers of laboratory tests are required to measure the variation of mechanical properties. Characterization of the elastic constants of porous materials is also required for wave propagation analyses. Porosity in materials makes them effective absorbers of mechanical energy when irreversible deformation occurs during void collapse. Both the stress or shock wave velocities and the unloading wave velocities depend on the elastic constants, which can change by almost an order of magnitude as the porosity decreases. Laboratory characterization of porous materials for wave propagation analyses is also expensive because of the large ranges of porosity that must be investigated. The objective of this study was to reduce the cost of laboratory testing, by developing a method for predicting the variation in the elastic constants of tuff with porosity. We constructed a correlation which includes an unexpectedly large range of data and loading conditions. The ability to relate stress induced changes to unstressed rock with different porosity values is new. The paper proceeds by describing compaction results for tuff and the basis for comparing them with published results, followed by the results of the correlation.
2 EXPERIMENTAL DETAILS
The tuff for the compaction tests came from the Halfton test site in Bed 4G adjacent to G-tunnel at the Nevada test site (Smith 1985). The bed is approximately 6 m thick and is described as "white ashfall, abundant lithic fragments, massive". It is bounded on the top by "red-gray mottled ashfall", and on the bottom by "reworked, bedded ashfalls". Details of testing procedures and equipment are given in Butcher & Jones (1987). The unconfined compression data sets were taken from test results for various geological units under examination for an underground nuclear waste repository in volcanic rock (Nimick et al. 1987, Olsson 1981, Price et al. 1982, Price & Nimick 1982, Price & Jones 1982,Price et al. 1982, 1984).
Investigations by the authors during surface and underground excavations indicated that the success of presplitting and smooth blasting was largely depended on the nature and orientation of joints and bedding planes in the rock mass. The paper describes the results obtained when the above two techniques were experimented at various stages of excavation in a large underground power house cavern in highly jointed basaltic rock mass. Smooth walls were obtained with both presplitting and smooth blasting along the roof of the cavern during driving of a 9 m high gallery parallel to the axis of the cavern and widening it to 23 m. The over breaks were reduced to within ± 0.15 m from the earlier ± 1 m. Excellent results were obtained with smooth blasting for excavation of side walls and the crown of the cavern when compared to presplitting. Bench blasting was adopted for deepening the cavern up to 50 m. Two stage bench blasting gave better results as compared to single stage bench blasting. Presplitting yielded better results compared to smooth blasting though the difference was inappreciable. With both the techniques the over and underbreaks were reduced to within ± 0.15 m. The efficiency of the above two controlled blasting techniques was assessed by counting the number of drill hole markings visible after blasting, measuring the total area of over and underbreaks, and the magnitude of vibrations transmitted to the side walls. The paper describes and compares the results obtained with the presplitting and smooth. blasting during various stages of excavation of a power house cavern.
Presplitting in rock mass creates a continuous fracture all along the lightly charged and appropriately spaced bore holes away from the free face. The stable fracture resulting after presplitting reflects the elastic stress waves and utilises the explosive energy from being wasted beyond the fracture zone. The newly created fracture zone reduces spalling and gives a desired broken area. The technique is widely in use for stabilization of pit slopes (Hoover 1972 and Talbot 1977), reduction in blasting cost (Teller 1972b), protection of surface structures near open pit mines (Stenhouse 1967), tunnelling (Forsthoff 1973), excavation of ore from waste (Snith 1965), roam and pillar mining in linestone (Bjorn 1969) and destruction of iceland (Melior 1977). Better results were reported with smooth blasting when compared to presplitting during trial blasts in jointed rock formations (Langefors 1978). Joint orientation considerably influenced the success of presplitting when they were at an angle of I to 30 degrees to the presplit axis. In case of jointed rock mass the explosive energy facilitated the opening along the joint planes instead of formation of the presplit fracture (worsey 1981). Investigations by the authorsduring dismantling of cement concrete blocks in a hydel project confirmed findings of Worsey. It was observed that when holes for presplitting were drilled at a distance of 5 to 20 an from the joint in the concrete block fracture did not develop along the axis of holes.
This paper presents the case history of a monitored opening 4.5 m high by 42 m wide in a deep potash mine. Deformations and stresses were measured in the roof and a wall. From the monitoring data, the geometry of the active opening and viscoplastic zone was determined and compared with predicted values. Reasonable agreement between measured and predicted values was obtained. Using the data, the projected mining layout of relatively thick pillars with viscoelastic zones was evaluated and found to be satisfactory.
As part of a rock mechanics program to obtain data for design of a mining system, a test room measuring 42 m long by 42 m wide by 4.5 m high was driven in a deep potash mine in New Brunswick, Canada. The room is situated 50 m below the top of the salt strata at a depth of 840 m and is instrumented with closure meters, borehole extensometers, and borehole pressure cells. When driven in November 1983, the room was in an otherwise virgin area of the mine. Between February 1985 and June 1985 (days 441 through 560 after excavation), some mining (mainly back trimming) took place within a few hundred meters. From June 1986 (approximately day 920) onward, active stopping approached to within 50 m of the room. Presented in this paper are:
Overstressed rock fracturing was monitored with multipoint borehole extensometers during excavation of the Caladay shaft in the Coeur d'Alene district of northern Idaho. Rock deformations showed considerable time dependency both in the short-term response of the rock mass to a blast and the eventual stabilization of the shaft over several months. The estimated in situ stress and rock strength, rock noise during excavation, and the magnitudes of displacement measurements all point to a time-dependent fracturing mechanism. These results are in line with numerous laboratory uniaxial time-to-failure tests which give strength reductions from 10% to over 50% of standard test results, depending on rock type and test environment. The time- dependent fracturing of rock, or "relaxation", is an important mechanism for safely dissipating dangerous concentrations of strain energy that pose a rock burst hazard. Additionally, a fuller under- standing of fracture dynamics may prove valuable in predicting whether and when a rock burst will occur. Finally, recognition of a specific rock strength for the service life of an underground opening (which will vary widely with rock type and environment) removes a significant parameter from the "safety factor". However, further test development and standardization is needed to incorporate the time dependency of rock strength into standard design methods.
2 SITE DESCRIPTION
The Caladay shaft is located at the east end of the southern extension of the Coeur d'Alene Mining District on the south side of the Osburn Fault (fig. 1). The shaft is rectangular, approximately 7-m by 4-m with 30-cm x 30-cm timber sets installed with a vertical spacing of 2 m. Conventional drill-and-blast sinking with bench blasting was used. Timber support is carried 6 to 8 m above the bottom, and extended in three set intervals. Initial support at the face consists of 150-cm long split-set rock bolts and 180-cm long grouted rockbolts with wire mesh and matting. The shaft short dimension is oriented parallel to the strata, which strikes at N 35 W and dips 65 to 75 northeast. There are numerous gouge-filled joints, partings and minor faults. Figure 2 shows the overall plan with orientation of the shaft, bedding, and instruments.
Figure 1, Location of the Caladay Project in the Coeur d'Alene Mining District. (available in full paper)
Figure 2. Caladay shaft section showing geology, timber blocking scheme, and location of spray-painted E2 and E4 labels. (available in full paper)
The test level is at a depth of 1450 m, between the 1400-m and 1500-m stations. A recent estimate of the vertical (Sv) and principal horizontal (Shl and Sh2) components of the in situ stress field at this depth (Whyatt, 1986) is:
Sv 46 MPa
Shl 69 MPa @ N 20-40 W
Sh2 52 MPa
The rock is thinly bedded argillitic quartzite with interbedded argillaceous material of the Revett Geologic Formation. In situ rock behavior around underground openings exhibit extensive stable fracturing with audible "popping" and occasional rock bursts which pose a serious safety hazard throughout the Coeur d'Alene district. Rock specimens typically demonstrates violent (class II) failure in the laboratory.
ABSTRACT: Early-time mining sequence closure displacements show how effects of multipass excavations are accommodated in bedded salt.
The Waste Isolation Pilot Plant (WIPP) Facility [Matalucci et al., 1982] is now being constructed in southeastern New Mexico. The underground workings of the facility are about 659 m below the ground surface. A portion of the facility is devoted to large scale experiments for the Thermal/Structual Interactions (TSI) research and development program to study deformation of the openings in salt [Munson and Matalucci, 1984]. These experiments are used to determine the creep closure of excavations in bedded salt; they provide the data base necessary to develop and validate the predictive models of behavior required for performance assessment. The extensive early-time closure data collected during the multipass excavation of each room provide information on the response of salt as a function of the geometry of the openings. Historically, creep closure of excavations in salt have been framed in terms of displacements. However, it appears possible because of the relatively homogeneous nature of salt to introduce a pseudostrain representation, which calls attention to the fact that the displacements are related directly to the strains in the material around the opening. Application of this representation to extensive early-time closure data illustrates the nature of salt response. This paper presents a description of the experimental rooms and explains how early-time mining sequence data were obtained during the multipass excavation. Representative mining sequence displacement data for various size openings are compared and analyzed in terms of pseudostrains. In addition, numerical calculations with a simple back-fitted model of creep are presented. Implications of the results are discussed in the conclusions.
2 COMPARISONS OF IN SITU DATA (MINING SEQUENCE CLOSURES)
Mining sequence data were collected during excavation of seven rooms and one room entry drift. Procedures for the measurements were tightly controlled to assure accurate data. This paper compares three of these rooms, which are nearly identical in configuration, size, and stratigraphic setting. Both displacements and strains for these excavations can be selectively compared.
2.1 Mining sequence experiments
Three TSI test rooms were mined to specified final dimensions of 5.5 m x 5.5 m in cross-section by 93.3 m long. They were excavated in multiple passes by a continuous miner. The rooms D, B, and A2 were excavated in that order and were separated sufficiently to be considered isolated during mining. Figure 1 is a mine plan showing locations of the mining sequence stations. Later, the excavation of A1 and A3 on either side of A2 resulted in an interactive, three-room complex; discussion of the interactions with A2 is too extensive for presentation here. The rooms were excavated in four passes, with some variations. As shown in Figure 2, the first two passes mined an upper bench and the second two passes excavated the bench to final room height. In B, passes 3 and 4 were advanced simultaneously. In some rooms, a floor trim was used to assure tolerance control. Nominal pass and trim dimensions are given in Figure 2. The first pass produced a small, nearly square cross-section opening.