.* A A <* •?.»• a g* 5©e O u -G 7 x> bV .jQf^* **G V^ ♦7*C^* A <» -f. »* .6^ ^ ♦7X7^* A O 'i-.-i** .G* ^b, ~< > ^ , * -.4- & ;• *F o- y n o ? ^ - j9 ^ *>V: c,"'^ -^l ^ 4,*"^$. «' ^/™/ V^V %.*TE->\# % .♦%at...% w :. ^ ^ .*«* v^ /, ^^' VA 1 C> «. , •^^-^.• *Pj. A"* *■- IC 9137 Bureau of Mines Information Circular/1987 Eastern Coal Mine Geomechanics Proceedings: Bureau of Mines Technology Transfer Seminar, Pittsburgh, PA, November 19, 1986 Compiled by Staff, Bureau of Mines UNITED STATES DEPARTMENT OF THE INTERIOR Information Circular 9137 Eastern Coal Mine Geomechanics Proceedings: Bureau of Mines Technology Transfer Seminar, Pittsburgh, PA, November 19, 1986 Compiled by Staff, Bureau of Mines UNITED STATES DEPARTMENT OF THE INTERIOR Donald Paul Hodel, Secretary BUREAU OF MINES Robert C. Horton, Director 1 Library of Congress Cataloging in Publication Data: Bureau of Mines Technology Transfer Seminars (1986 : Pittsburgh, Pa.) Eastern coal mine geomechanics. (Bureau of Mines information circular ; 9137) Bibliographies. Supt. of Docs, no.: I 28.27: 9137. 1. Coal mines and mining- United States - Congresses. 2. Mine roof control -Congresses. 3. Rock mechanics -Congresses. I. United States. Bureau of Mines. II. Title. III. Series: In- formation circular (United States. Bureau of Mines) ; 9137. TN295.U4 [TN805.A3] 622 s [622'.334] 86-607950 PREFACE The papers contained in this Information Circular address the results of several research studies conducted by the Bureau of Mines in an ef- fort to improve mine roof support systems and mine design. Through eastern coal mine geomechanics research, new technology is being devel- oped that will enhance mine productivity, improve resource recovery, and increase mine workers' health and safety. The seven papers were presented at a technology transfer seminar on eastern coal mine geomechanics in November 1986. Technology transfer seminars are used by the Bureau of Mines to keep the minerals industry apprised of new technology and developments resulting from its research endeavors. Further information about current research efforts in other areas of concern may be obtained by writing the Bureau of Mines, Branch of Technology Transfer, 2401 E St., NW. , Washington, DC 20241. iii CONTENTS Page Preface i Abstract 1 Introduction 2 State-of-the-Art Testing and Analysis of Mine Roof Support Systems, by Thomas M. Barczak 3 Characterization and Measurement of Longwall Rock Mass Movement, by Jeffrey M. Listak, John L. Hill III, and Joseph C. Zelanko 12 Multiple-Seam Mining Problems in the Eastern United States, by Gregory J. Chekan, Rudy J. Matetic, and James A. Galek 27 Integrated Design for Stability in Multiple-Seam Mining, by Chris Haycocks, Wei Wu, and Yingxin Zhou 44 The Bureau of Mines Subsidence Research Program, by Michael A. Trevits, Roger L. King, and Bradley V. Johnson 57 Subsidence Over Chain Pillars, by P. W. Jeran and V. Adamek 65 Study of Dewatering Effects at an Underground Longwall Mine Site in the Pittsburgh Seam of the Northern Appalachian Coalfield, by Gregory E. Tieman and Henry W. Rauch 72 UNIT OF MEASURE ABBREVIATIONS USED IN THESE PAPERS deg degree lb pound ft foot lb/cu ft pound per cubic foot ft 2 square foot m meter f t/d foot per day min minute ft/ft foot per foot mm millimeter gal gallon ys/ft microsecond per foot gal/min gallon per minute pet percent gal/(min* acre) gallon per minute per acre psi pound per square inch gal/yr gallon per year psi/ft pound per square inch per foot gal/(acre •yr) gallon per acre per year psig pound per square inch, gauge h hour St short ton in inch yr year in/yr inch per year EASTERN COAL MINE GEOMECHANICS Proceedings: Bureau of Mines Technology Transfer Seminar, Pittsburgh, PA, November 19, 1986 Compiled by Staff, Bureau of Mines ABSTRACT The Bureau of Mines is establishing design criteria for effective roof support systems and is developing technology to improve mine planning and design. Geologic studies, surface and underground mine monitoring, and laboratory evaluations were conducted. The seven papers in this proceedings Information Circular present results of several investiga- tions. Addressed are problems existing in both longwall and multiple- seam mines. Discussions include the effect of dewatering at a longwall mine site in the Pittsburgh coal seam and the effect of chain pillars on subsidence. Other results are measurement and characterization of longwall rock mass movement and an integrated design for stability in multiple-seam mining. INTRODUCTION The Bureau of Mines conducted a re- search study of coal mine geomechanics in the Eastern United States. The goal is to improve mine planning and design by establishing better roof and ground sup- port systems. Longwall mine roof fail- ures often occur because the support sys- tem provides inadequate ground control at the face. Better design, selection, and operation of such support systems may help decrease the capital risk involved in equipping a longwall face. Proper supports will also enhance productivity, maximize coal recovery, and minimize health and safety hazards. Problems in controlling ground are also prevalent in multiple-seam mining, which is practiced almost exclusively in the East. Past disregard for the sequence of mining multiple seams has led to major problems. Entry and pillar instability are caused by subsidence, pressure con- centrations, and stress zones. If mining proceeds without concern for adjacent coal seams, our coal resources will be- come depleted at a much faster rate. New technology will provide high recovery of multiple-seam coal resources in an economic and safe manner. To help alle- viate some of these problems, the Bureau conducted a series of studies. The ob- jective was to evaluate not only the ef- fects of the mining method, but also how geology relates to support requirements. A rock mechanics investigation moni- tored deformation of near-seam strata above a longwall panel in the Pittsburgh Coalbed. The study provided data con- cerning the caving mechanism associated with longwall extractions. Analysis of the data provided a better understanding of the interaction of strata behavior with longwall face supports. Another Bu- reau research study analyzed subsidence over raining operations. The purpose was to provide mine operators with a means of predicting movement of the surface ground and its effect on ground water. This re- search led to a computer model capable of predicting subsidence over longwall pan- els in the Northern Appalachian coal region. The report presents further details about the results of these research ef- forts and of other studies. STATE-OF-THE-ART TESTING AND ANALYSIS OF MINE ROOF SUPPORT SYSTEMS By Thomas M. Barczak^ ABSTRACT The Bureau of Mines is conducting re- search to evaluate the interaction of mine roof supports with the surrounding strata to provide for more effective de- sign and utilization of these structures. Scientific testing of a system requires consistent definition of pertinent param- eters and concepts. Various interpreta- tions of support capacity, as they exist in the mining community, are briefly dis- cussed in this paper, as are the defi- ciencies in historical uniaxial testing and analysis of mine roof support sys- tems. Critical load analyses show that structural failure of the support can occur at loads well below those of rated support capacity. The significance of support stiffness and post yield charac- teristics demonstrates the need to eval- uate supports and the strata as a system rather than independently. Efforts to develop supports as load-sensing devices and strata activity monitors are de- scribed. The benefits and limitations of finite element modeling as an alternative to physical testing are discussed. A de- scription of the Bureau's Mine Roof Simu- lator Facility and an overview of future research efforts are also provided. INTRODUCTION The development of more effective roof support systems and strata control strat- egies requires an accurate understanding of the interaction of roof support sys- tems with the geological environment in which they are employed. Improvements in roof support design and strata control strategies can be realized only if the response of the support system and the conditions under which it is loaded are understood. To achieve this understand- ing, the Bureau of Mines has constructed facilities to evaluate full-scale roof support systems under controlled loading conditions. Modern roof support systems, such as longwall sheilds, represent so- phisticated structures, which require more indepth analysis than the simple prop supports of the past. State-of-the- art testing and analyses of mine roof support systems at the Bureau of Mines include laboratory, field, and theo- retical studies using the most sophisti- cated equipment and analysis techniques available. This paper describes the Bu- reau's mine roof support facilities and research. SUPPORT CAPACITY AND WHAT IT MEANS One of the most fundamental measures of support performance is its load-bearing capability, which in the most elementary analysis is often ambiguously referred to as the capacity of the support. Several measures have been used to describe sup- port capacity, and it is necessary to un- derstand the differences and deviations of these measures in order to properly 1 Physicist, Pittsburgh Research Center, Bureau of Mines, Pittsburgh, PA. interpret the load-bearing capability of the support. One method of describing capacity for active supports, such as longwall shields, is the yield load of the hydrau- lic leg cylinders. In the case of shield sypports, the effectiveness of the legs to support a roof load is diminished by inclination of the leg cylinder from the normal to the plane of the canopy; only the vertical component of the leg force should be used to determine the capacity of the support. Capacity in this context is more precisely defined as roof-to- floor or vertical support capacity, being premised by two inherent assumptions: (1) The support is generally assumed to be uniaxially loaded with no face-to- waste (horizontal) or parallel-to-the- face (lateral) loading, and (2) vertical load application is assumed to occur at the location of the leg reaction. Kine- matically, there are components other than the leg cylinders that contribute to support capacity in structures such as longwall supports, and the actual capac- ity of the support should be determined from the total contribution of all compo- nents. Load applications at locations other than that of the leg reaction re- duce the capacity of the support by in- troducing moments. The presence of off- axis (horizontal or lateral) loading can either increase or decrease vertical sup- port capacity, depending upon the geom- etry and kinematics of the support structure. Another method used to describe support capacity is termed "yield load density," which is defined as the yield load of the support per unit area of supported roof. This measure should be considered only as an average of load density reacted by the support and not in the context of a uni- form pressure distribution of applied roof loading. It is computed simply by dividing the yield capacity by the area of supported roof. Since the yield ca- pacity is determined from a concen- trated load (or the equivalent stress distribution) acting at the resultant support reaction, assuming a uniform load distribution acting on the canopy surface of magnitude equal to the yield load den- sity will produce a support reaction in excess of the yield capacity. This means that support requirements should not be predicted simply from the weight of rock masses, assuming uniform load distribu- tion on the support elements or load con- centrations consistent with the center of gravity of the rock mass. The loading mechanism and interaction of the supports with the strata environment are discussed in a later section. Finally, it must be recognized that supports are rarely loaded uniaxially. While the primary function of the sup- port is to control roof-to-floor strata convergence, supports such as longwall shields are designed to control face-to- waste strata activity as well. Face-to- waste capacity of shield-type supports is generally rated at 30 pet of the rated vertical capacity, derived from an as- sumed 0.30 coefficient of the friction between steel and coal measure strata. The 30 pet of vertical capacity rating simply assumes that horizontal loads can- not be generated in excess of 30 pet of vertical loading since strata would then slip on the canopy interface. Actual ca- pacity of the support to resist horizon- tal loading may be substantially higher. Unlike vertical loading, which is re- lieved by hydraulic yielding, horizontal load capacity is determined by the struc- tural strength of suport components. CRITICAL LOAD ANALYSES As indicated in the previous section, support capacity is generally determined from load concentration at the canopy re- sultant reaction. Support capacity will be reached prematurely by application of loads of substantially smaller magnitude at locations other than the resultant support reaction. For example, while a 500-st support could sustain a 500-st load concentrated at the leg (resultant) reaction, a load of less than 175 st ap- plied at the tip of the canopy would cause a typical two-legged shield support to yield. Hydraulic yield capacity is not the only measure of a support's load-bearing capability. Loading conditions exist that could structurally damage the sup- port with loads of magnitude well below the rated yield capacity. Critical load analyses are used to identify load con- ditions that cause maximum stress con- centrations in the support structure. Figure 1 illustrates symmetrical load Base Canopy Caving shiela 1 I i i t t t t t t 1 J 1 i k \\\Y\V I t t 1 I 1 Lemniscate links FIGURE 1.— Critical load test configurations. conditions that would cause maximum stresses in each major component of a generic, lemniscate shield. Several par- tial-contact conditions are illustrated in the figure by arrows, which represent load applications or restraint of the structure. Full-contact conditions are illustrated by a rectangle on the support member. Table 1 provides a description of the loading mechanism cases as well as a physical interpretation of the mine environment that could cause these load conditions. The results of these critical load tests for one particular shield, repre- sentative of two-legged shields in gen- eral, are described in the following paragraphs. CANOPY The canopy is the least stiff of all shield components and is therefore sub- ject to critical bending stresses at rel- atively small loads. Critical contact configurations suggest shield capacities of approximately one-half of rated hy- draulic yield capacity. In other words, loads of less than half of the rated hy- draulic yield capacity will cause stress concentrations approaching the yield strength of the steel under the contact conditions identified in table 1. TABLE 1. - Critical load tests Shield component Loading mechanism Physical interpretation Base. Canopy, Caving shield. Component link, Tension link. Create maximum bending stress in base by supporting base as a simple beam with canopy inducing concentrated force at center span of base. Same mechanism as for base with bending created in canopy. Simulate horizontal displacement of canopy relative to base by creating horizontal couple as shown in figure 1. Same strategy as with caving shield. Maximum loading occurs when canopy rotation is prevented, causing shield reactions at hinge pin and in caving shield. Same strategy as witb caving shield. Maximum tension link loading occurs when links are in horizontal con- figuration (low shield heights). Base supported on toe and rear with leg force causing bending of base structure. Same physical description as for base with canopy simply supported at tip and hinge. Canopy and base promoting horizontal reactions due to vertical shield convergence. Do. Do. BASE The base is the stiffest of all shield components, and no contact configurations were found to induce critical stresses for this particular shield. CAVING SHIELD Roof-to-floor convergence of the shield, which is thought to be the pre- dominant loading mechanism in an under- ground environment, produces relatively little loading in the caving shield. However, horizontal displacement of the canopy coupled with a horizontal re- straint of the base to cause relative mo- tion between the canopy and base cre- ates stresses approaching yield strength of the caving shield material with ver- tical loads below the hydraulic yield capacity of the support. The effect of gob loading directly on the caving shield study. was not investigated in this LEMNISCATE LINKS No contact condition was found to in- duce more than 6,000 psi of stress, which is less than 20 pet of the material yield strength. Apparently much of the energy is absorbed by the leg cylinders and in caving shield bending, with little en- ergy being transferred to the lemniscate links. The primary function of the links under these test conditions is simply as a guidance system to control canopy canopy motion. It should be noted that these test re- sults are for one shield only. Other shields may behave differently depending on their structural characteristics, but the critical load conditions should be applicable to most shields of this basic configuration. INTERACTION OF THE SUPPORT WITH THE STRATA In the evaluation of roof support sys- tems it is important to understand that the roof support elements and the sur- rounding strata act as a system, respond- ing to changes in the physical mine envi- ronment due to extraction of coal and associated redistribution of stresses. The strength (capacity) of a roof support element alone is not a meaningful measure of support performance. The loading of a roof support element can be significantly dependent upon the stiffness of the sup- port structure, as a stiffer structure will receive a higher load from converg- ing mine strata than will a softer struc- ture. This concept is illustrated in figure 2, where two hypothetical struc- tures of different stiffness, Kj and K.2 , with &i < K 2 , are displaced an equal amount. As shown, the resultant load on the stiffer structure would be 1.5 times that on the softer structure under the same load condition. The significance of support stiffness for behavior of the support is most cri- tical in passive roof support systems, such as posts and cribbing. Recently, concrete and steel-fiber-reinforced (SFR) concrete have been used as substitute materials for wood in crib applications. The higher compressive strength of con- crete compared to wood (about eight times) results in a much higher yield strength (capacity) for cribs constructed of concrete. However, as seen from the load-displacement characteristic of wood and concrete illustrated in figure 3, concrete produces a much stiffer crib structure than does wood. These results indicate that a solid SFR concrete crib DISPLACEMENT^) FIGURE 2.— Significance of support stiffness. 1,200 900 600 300 -SFR concrete Nonreinforced concrete -Wood ^--wooa 4 6 8 10 DISPLACEMENT, in FIGURE 3.— Crib loading characteristics. would experience a load 18 times that of a wood crib subjected to the same displacement prior to reaching yield strength of the crib. The displacement at which the yield strength of the sup- port is reached is also a critical mea- sure of support performance. The stiff concrete supports, despite having a much greater yield strength, reach yield load at much smaller displacements than cribs constructed of wood. Tests indicate con- crete cribs fail at less than 0.5 in of displacement, whereas wood can displace nearly 3 in before reaching yield load and over 10 in before failing. There- fore, if the convergence of the roof is irresistible (displacement loading), the high yield strength of concrete cribs will be of no advantage if the roof con- vergence exceeds 0.5 in. Equally important to the loading char- acteristics of passive roof support sys- tems is postyield behavior. As seen in figure 3, the stiff concrete supports lose load-carrying capability almost im- mediately after reaching yield load, whereas wood continues to provide support resistance for several inches of dis- placement. In terms of energy absorp- tion, the wood crib can absorb more en- ergy than a concrete crib owing to its ability to deform and maintain resistance to load. The idealized roof support should be initially stiff to be rapidly load bearing, should have yield strength sufficient to support the deadweight re- sponse of the immediate fractured strata, and should have sufficient postyield characteristics to be compatible with the displacement behavior of the overburden. The stiffness of an active roof sup- port, such as a longwall shield, is also critical to the interaction of the sup- port with the strata. A stiff structure that generates high reactive forces can cause unnecessary fracturing of incom- petent roof or floor strata or cause further instability of unstable strata. While the hydraulic support is self- relieving, the yield load of modern pow- ered supports ranges from 250 to 800 st; this is sufficient to cause pressure dis- tributions on the canopy and base that could damage incompetent roof or floor strata. SUPPORTS AS MONITORS Roof support elements, both passive and active structures, can be used as roof load monitors by proper instrumentation of the support structures. Past efforts to use roof supports as load-sensing devices have generally been limited to one-dimensional analysis by summation of measured forces to obtain a verti- cal (roof-to-floor) reaction. As previ- ously indicated, supports are usually not loaded uniaxially. Modern longwall sup- ports, such as the shield, are designed to resist both roof-to-floor (vertical) and face-to-waste (horizontal) loading, requiring two-dimensional analyses to determine vertical and horizontal load reactions. The Bureau has demonstrated that vertical and horizontal support re- sistance of a longwall shield can be rea- sonably determined from static rigid body analyses by measurement of leg, canopy capsule, and compression lemniscate link forces. The two-dimensional model as- sumes no gob loading, nor does it recog- nize any out-of-plane loading or the ef- fect of moment loading due to imbalances in the leg and lemniscate link forces. From controlled tests in the Bureau's Mine Roof Simulator, vertical shield loading could be predicted to within 3 to 5 pet and horizontal loading to within 25 pet. 2 In theory, the limitations of the two- dimensional model can be overcome by three-dimensional modeling of the support structure to account for out-of-plane forces and moments. Unfortunately, the advancement of a three-dimensional model is difficult since the accumulation of unknowns far outpaces the available force and moment equilibrium equations, making the system statistically indeterminate. While elimination of unknowns with rea- sonable engineering judgments as to non- participating forces produces a solu- tion, the shield support is primarily designed to resist loading in two dimen- sions (roof-to-floor and face-to-waste), making the two-dimensional model adequate for most analyses. It is also recognized from tests with shield supports that vertical conver- gence produces both a vertical and a hor- izontal load reaction; likewise, a horizontal displacement produces both a horizontal and a vertical support reac- tion. In other words, the shield support not only reacts axial loads in direct response to strata displacements (shield convergence), but also induces offaxis load reactions as a result of the mechan- ics of the shield structure. Only load reactions resulting from strata activity need to be resisted for successful appli- cation of the longwall method; support- induced loads are indications of ineffi- cient designs if they are of no benefit to strata control. To put this in per- spective, in situ tests indicate that over 75 pet of horizontal load experi- enced by shield supports is the result of setting the support against the roof. There is also some evidence that a sig- nificant portion of the remaining 25 pet is due to vertical roof convergence and not face-to-waste strata activity. To resolve the issue of the source of horizontal shield loading and to develop methods whereby the supports can truly be used as monitors of strata activity, the Bureau is investigating the utilization of a linear elastic model with two de- grees of freedom. Mathematically, the model is expressed as follows: F h " K l + K 2 + K, (1) (2) where F v = vertical support resultant load reaction, F h = horizontal support resultant load reaction, v = vertical shield displacement, h = horizontal shield displacement, and I , K 2 , K3, K4 = stiffness coefficients, K By controlled uniaxial displacement tests in the Bureau's Mine Roof Simula- tor, the stiffness coefficients are de- termined for a particular shield. Using numerical values for the stiffness param- eters, the inverse of equations 1 and 2 can be used to determine shield displace- ments, which are indicative of mine roof convergence and strata activity, if re- sultant shield loading is accurately known. This also enables horizontal loading produced by face-to-waste strata activity to be distinguished from shield- induced horizontal loading due to verti- cal roof convergence. MATHEMATICAL MODELING— AN ALTERNATIVE TO PHYSICAL TESTING Mathematical models provide a simple, cost-effective method to evaluate load responses of structures without the need 2 Barczak, T. M., and R. C. Garson. Technique To Measure Resultant Loading on Shield Supports. Paper in Rock Mechanics for full-scale physical testing. While simple rigid body statics may be used to determine the elementary behavior of a in Productivity and Protection (Proc. 25th Symp. on Rock Mech.). Soc. Min. Eng.,1984, pp. 667-679. support in relation to an applied load, more sophisticated analyses are required to analyze the load-displacement rela- tionship of a support structure. The linear elastic model presented in the previous section is a first step in eval- uating roof supports as elastic bodies. One of the most powerful techniques used to analyze structures is finite ele- ment modeling. According to its basic principle, a structure is idealized as a composition of a number of finite pieces rather than continuous elements. This concept enables the step-by-step buildup of the load-displacement relationship of a structure as a whole from those basic elements of which the structure is com- posed. The accuracy of the model is de- pendent upon the number of elements as the stresses are averaged over the area of the elements. Proper selection of elements and boundary conditions is re- quired because different types of ele- ments behave differently. Knowing the geometry and elastic properties of the elements allows computation of structural deformations for each of the finite ele- ments from which areas of critical stress concentrations can be identified. The Bureau has used finite element analyses in the following ways in mine roof support research: Identification of load cases for test- ing . - The critical load tests presented in the section on critical load analyses were determined from a simple two-dimen- sional finite element model. Location of strain gauges for physi- cal testing . - It is generally desirable to locate strain-measuring instrumen- tation at areas of critical stress concentrations. Identification of critical stress con- centrations . - Three-dimensional models have been used to identify critical stress concentrations in major support components. Evaluation of canopy pressure distribu- tion from strain contours on the canopy structure . - Measurement of canopy pres- sure distribution by analysis of under- side strain contours proved difficult owing to multiple contact conditions that produce similar strain profiles and to difficulty in modeling partial contact conditions. Finite element modeling of supports has proved more challenging than originally anticipated. The general behavior of the structure in terms of load distribution can be determined from finite element modeling, but an accurate picture of load transfer in the shield structure is some- times difficult to obtain. Mathematical modeling requires an accurate definition of the problem, including both the prop- erties of the structure and the load and boundary conditions. Discrepancies in finite element model predictions are of- ten caused by improper modeling of load conditions. This is particularly diffi- cult in modeling support behavior because of the multitude of contact configura- tions that exist with rock strata inter- faces. Longwall supports are basically crude pin-jointed structures, and model- ing pin frictions in joints can be diffi- cult. Factors such as friction, ability to accurately model internal platework, and the effect of stress concentrations at discontinuities can have a significant impact on results. The Bureau's experience with finite element modeling of longwall shields is that simple beam models constructed to fit a known shield response from physi- cal testing are more successful and much cheaper than complex models, which may geometrically look more like the support structure, but are very difficult to con- struct to properly simulate the struc- tural fabrication of the shield. A further implication of the Bureau's research in finite element modeling of roof supports is the discovery that com- plicated shield component constructions are sensitive to some unique load appli- cations. While current supports are suf- ficiently overdesigned that these unique load conditions do not represent prob- lems, efforts to improve designs from a stress optimization viewpoint will re- quire more refined load analyses than the relatively simple criteria presently used in support design. 10 BUREAU OF MINES FACILITIES The heart of the Bureau's roof support test facilities is the Mine Roof Simula- tor (MRS) illustrated in figure 4. The MRS is unique in the world in its abil- ities to apply both a vertical and a hor- izontal load simultaneously. Both the vertical and horizontal axis can be independently programmed to oper- ate in either force or displacement con- trol. This capability permits tests such as true friction-free controlled loading of shields, which cannot be accomplished in uniaxial test machines since the shield reacts a horizontal load to ver- tical roof convergence. Friction-free tests of this nature can be accomplished in the MRS by allowing the platen to float in the horizontal axis by command- ing a zero horizontal load condition. Likewise, the MRS can apply controlled horizontal loading to a shield support, whereas uniaxial test machines can apply only vertical loading with no control over horizontal load reactions or capa- bility to provide a specified horizontal load to the structure. The machine incorporates 20-ft-square platens with a 16-ft vertical opening, enabling full-scale testing of longwall suport structures. Capacity of the simu- lator is 1,500 st of vertical force and 800 st of horizontal force and controlled displacement ranges of 24 in vertically and 16 in horizontally. Load and dis- placement control is provided in four ranges operating under a 12-bit analog- to-digital closed-loop control network, providing a load control capability of better than 0.1 kip (100 lb) and dis- placement control capability of better than 0.001 kip in the smallest load- displacement range. Machine control and data acquisition are achieved by a DEC 11/34 computer. Eighty-eight channels of test article FIGURE 4.— Mine roof simulator. transducer conditioning are provided. Data acquisition is interfaced with the control network so that machine behavior can be controlled by response of the test article instrumentation. For example, tests can be terminated or held when strain values reach a designated level in specified areas of the support structure. High-speed data acquisition is available with a separate DEC 11/23 computer at a rate of 300 samples per second. An X-Y-Y recorder provides real-time plotting of three data channels, and all data are stored on computer disks for subsequent processing and analysis. FUTURE DEVELOPMENTS AND CONCLUSIONS The Bureau of Mines plans continued re- search on mine roof support systems in an effort to develop more cost effective and safer systems for the mining industry. Several aspects of ground control, sup- port design, and strata interaction are not yet well understood. The Bureau hopes to advance the state-of-the-art of 11 ground control and support testing with the following goals: 1. Determine the effectiveness and limitations of finite element analysis in support design and behavior. Recommend guidelines for advanced analysis of mine roof support systems using finite element analysis. 2. Continue efforts to develop tech- niques to effectively use supports as strata monitors to develop a more basic understanding of strata activity so that support requirements can be more closely engineered to strata behavior. 3. Continue to explore and propose concepts of support behavior, such as the significance of support stiffness in load behavior, which may not be apparent to mine operators, support manufacturers, or other researchers. 4. Continue to explore advanced tech- nology applications to current and novel mine roof support concepts or support designs. In conclusion, ground control is cri- tical to successful mining operations. Nearly all of the financial risk associ- ated with a mining venture can be attrib- uted to ground control. Current mining operations require capital-intensive sys- tems costing several million dollars for each installation, and roof support fa- talities continue to be the number one killer. Testing and analysis of roof support systems is necessary to provide better utilization of support systems and the development of improved designs for the future. 11 CHARACTERIZATION AND MEASUREMENT OF LONGWALL ROCK MASS MOVEMENT By Jeffrey M. Listak, 1 John L. Hill III, 1 and Joseph C. Zelanko 1 ABSTRACT The Bureau of Mines has conducted a rock mechanics study to monitor deforma- tion of near-seam strata above a longwall panel in the Pittsburgh Coalbed. The primary goal of this research was to de- termine the height of caving immediately behind advancing longwall face supports. This study, although site specific, pro- vides information on the caving mechanism associated with longwall extractions so that strata behavior and its interaction with longwall face supports can be better understood. Two vertical boreholes, positioned 550 ft apart along the centerline of a long- wall panel, were drilled from the surface to intercept the coalbed at a depth of approximately 650 ft. Various downhole geotechnical instruments were used to monitor strata deformation. In addition, surface elevation surveys were conducted to differentiate between surface and sub- surface activity. This report discusses the caving char- acteristics of the strata as the longwall panel approached and passed beneath the boreholes. Physical property data are also presented to demonstrate the rela- tionship between caving behavior and lo- cal geology. Data show that immediate caving of strata above the longwall face occurred at a height of less than 23.5 ft and that strata behavior above longwall extractions is highly dependent upon lithology, with major disturbances occur- ring at weak lithologic zones. INTRODUCTION For a safe and productive longwall op- eration, the optimization of both roof control and operational efficiency is es- sential. However, these two major con- tributors to longwall success are depen- dent upon the accurate prediction of roof support capacity requirements. Longwall failures of the past have been attributed to lack of understanding of roof and floor behavior and poor caving character- istics in the gob area. Historically, roof caving in the gob has been evaluated by several methods. One method is a theoretical approach based on the cantilever beam theory of structural mechanics. This approach as- sumes that the stratified roof rock lay- ers act as cantilever beams above the ex- tracted coal. By knowing the material properties and the thickness of the lay- ers, the extent of the interlayer loading 'Mining engineer, Pittsburgh Research Center, Bureau of Mines, Pittsburgh, PA. can be determined (1_). 2 As the roof lay- ers fail in stepwise fashion, caved mate- rial swells or bulks to fill the void and the caving height can be calculated. Rock classification is another method used to characterize mine roof for cave- ability prediction. Much of this work has been done in the European coalfields by Pawlowicz, Kidybinski, and Kostyk of Poland and by Proyavkin and Davidyanc of the U.S.S.R. as cited by Kidybinski (2_). In addition, Ghose (3_) used geotechnical logs to classify mine roof for cave pre- diction in coal mines in India. Ghose also cited the notable classification systems proposed by Vasiliev et al. and Nenasheva et al. of the U.S.S.R. and the Roof Quality Index of Bilinski and Konopko of Poland. Bieniawski (4_) devel- oped an engineering classification of ^Underlined numbers in parentheses re- fer to items in the list of references at the end of this paper. 13 rock masses called the Geomechanics Clas- sification or rock mass rating (RMR). This classification system utilizes six parameters, which are measurable in the field or which can be obtained from bore- hole data, to determine the "stand-up time" for an unsupported roof span. More recently, numerical modeling has been used to provide good estimates for cave prediction (e.g. , the finite ele- ment method, distinct element method, and boundary element method). Although these methods are of value since they offer guidance, precise iden- tification of the caving behavior of the strata is still unresolved. Very few systematic studies that actually moni- tored the caving mechanism associated with longwall mining have been conducted. Field investigations (_5~J7) performed to directly measure bed separation over longwall extractions have had limited success. However, one such study con- ducted recently (_5) used a borehole cali- per tool to measure bed separation and had conclusive results that are very sim- ilar to the findings stated later in this report. To develop a better understanding of longwall strata behavior and its inter- action with longwall face supports, the Bureau of Mines installed various geo- technical instruments in two boreholes drilled over a longwall extraction lo- cated in southwestern Pennsylvania. This study is intended to lay the groundwork for additional research in order to de- velop a data base for strata behavior above longwall extractions. The informa- tion from these studies could effectively improve the method for selecting longwall roof support capacities. Several analytical methods (8-10) have been developed to predict longwall strata behavior and associated face support loading. These methods for selecting the proper load capacity of longwall roof supports are based on the assumption that some finite volume of roof material, of- ten assumed to be cubic or parallelpiped in geometry, is being held by the sup- port. The boundaries of this volume of material are defined by the spacing of the supports (width), the supporting dis- tance from face to gob, and the height and angle of caving. The height of cav- ing is generally estimated as a factor times the height of extraction, the fac- tor being determined by estimating the bulking factor of caved material. Formu- lae using assumed caving heights for pre- dicting load density vary considerably. Wilson (8) assumes that caved rock oc- cupies 1.5 times the volume of the same rock in situ and therefore maintains that the height of caving (the height of caving above the level of the roof) is twice the extraction height. Wade ( _9) , however, assumes that caved material will occupy 1.25 times the volume of rock in situ, yielding a caving height equal to four times the extraction height. To find the caving horizons above a longwall panel extraction, two instrumen- tation stations were installed in verti- cal boreholes located along the center- line of one longwall panel. Each station utilized instruments that measured both horizontal and vertical displacements as a function of longwall face advance. In addition, surface evaluation surveys were performed to differentiate between sur- face and subsurface displacements. ACKNOWLEDGMENTS The authors thank Roctest, Inc. ,- Plattsburgh, NY, for cooperation in — _ _ -"Reference to specific manufacturers does not imply endorsement by the Bureau of Mines. designing, manufacturing, and installing the borehole instrumentation. Special thanks is extended to Girard Theroue and David Prentice of Roctest, Inc., for their assistance at the field site. 14 DESCRIPTION OF TEST SITE The longwall panel under investigation is located within the Appalachian Plateau Province of western Pennsylvania. Struc- tural relief in the region does not ex- ceed 350 ft, and dips are generally less than 4°. Mining takes place in the Pittsburgh Coalbed, which lies strati- graphically within the Pennsylvania age coal-bearing strata of the Monongahela Group (fig. 1). The longwall panel is mined to a height of 5.8 ft; overburden depths vary throughout the length of the panel. Figure 2 illustrates the lateral continuity of this interval over the pan- el under study. The panel (panel 3) di- mensions are 630 ft wide and 5,570 ft long. This longwall panel utilizes a four-entry headgate and tailgate system with square pillars on 90-ft centers and entries 15 to 18 ft wide. Roof support along the face was main- tained by Dowty 460-st two-leg shield supports with setting pressures of 3,600 psi. The rate of advance for the long- wall face was approximately 35 ft per three-production-shift day. 500 510 520 530 540 «- 550 H— I CL LlI o 560 570 580 590 600 610 620 630 Sewickley Coalbed Redstone Coalbed Pittsburgh Coalbed LEGEND I Shaly sandstone F^^ Shale ■■Coal EJ*ii Shaly limestone [gig Limestone FIGURE 1.— Generalized stratigraphic column of the Monongahela Group. Direction of mining »- Panel 3 1,000 Scale, ft 2,000 LEGEND Sandstone Shale Coal ||3jShaly limestone K^ Limestone FIGURE 2.— Fence diagram of strata overlying the study panel. 15 SITE PREPARATION — BOREHOLE DRILLING Factors considered during site selec- tion included topography and terrain, surface rights (private land ownership), and environmental restrictions. From a technical standpoint, however, the most important consideration was to choose a test site that would yield representative results over the length of the panel. Therefore, in addition to the nontech- nical considerations, borehole locations were chosen toward the center and along the length of the centerline of the panel to minimize the effects of the panel boundaries on the caving process. The two monitoring stations were located 2,600 and 3,150 ft from the start of the panel (fig. 3). Two 6-in-diam boreholes were drilled through the coalbed. With the exception of a 15-ft standpipe at the surface, the holes were not cased. This allowed the borehole anchors to be set directly in distinct stratigraphic members. A combi- nation of rotary and core drilling was used to drill the first borehole (BH1). One hundred feet of core was extracted from BH1 for descriptive geologic logging and laboratory testing. This interval included the entire interburden between the Sewickley and Pittsburgh Coalbeds. The surface at BH1 had a mean sea level elevation of 1,004.68 ft, and the bore- hole depth reached 630 ft. This final depth was approximately 3 ft below the base of the Pittsburgh Coalbed. Geophys- ical logging was performed on each hole to determine lithology and the location of water-bearing strata. These logs were also used to calculate a rock strength index. The quality of the borehole is of ut- most importance. It is imperative to have a straight hole that is free of ob- structions so that problems during in- strument installation can be kept to a minimum. However, in this study, bore- hole conditions were not ideal. This prevented placement of borehole anchors at desired depths (5 ft above the coal- bed), so caving of the immediate roof behind the longwall supports could not be detected. Figure 4 shows a borehole di- rectional survey of BH1. Although the horizontal difference between the top and bottom of the borehole was only 3.5 ft, sharp deviations of the borehole, due to the spiraled drill path, prevented place- ment of a borehole anchor at the desired depth of 615 ft, which is 5 ft above the coalbed. o 81* Anchor _jih locations! I h Anchor locations 1 Pittsburgh Coalbed SECTION A- A' aooooaaooaaomaaaoaooooaoooaoDOQoao ODDOODDOOOaaDOQOOODQpOOOOoaoODoaDD BH1 Direction of mining BH2 o A' Panel 3 aooaooaoaoCDaoaoooooaaoaoDaaaooaa^ oooooaaaoaooooooooooaaaoaaoaaoaaaa oocjaoQDaooooooooooDoaaaaooQoaoDoaa Scale, ft PLAN VIEW FIGURE 3.— Study panel and cross section of borehole an- chor locations. 16 ROCK MASS CHARACTERIZATION An assessment of geologic and mechani- cal rock properties of the entire unit of strata overlying the longwall panel at the mine site was conducted to relate these characteristics to caving height and lateral overburden movement. To ac- complish this, geophysical logging was carried out over the entire length of each of the two boreholes. Figure 5 is a summary of the character- istics obtained for the Pittsburgh and Sewickley interburden at the longwall site. The characteristics include geo- logic description, strength index from well logs (dynamic elastic modulus of deformation) , uniaxial compressive strength, indirect tensile strength, and rock quality designation (RQD) (11). As the figure shows, the immediate roof rock (approximately 28 ft) above the Pitts- burgh Coalbed is mainly composed of weak shales with a low RQD. Since the RQD values shown were calculated over each 10-ft core run, the three clayey shale units (indicated as "disintegration zones" in the figure) do not specifically stand out as weak zones. However, each of these units disintegrated upon removal from the core barrel. As will be dis- cussed later, it is probable that caving height was coincident with one of these horizons. 90° 270° 100° 260° American Society for Testing and Mate- rials (ASTM) standards were used for strength evaluations of the NX core from BHl , including uniaxial compressive strength and Brazilian tensile strength (diametral compression) tests. All of the core used was ample in size to meet specimen standards, and a statistically significant number of specimens were tested for most of the rock types recov- ered. Mechanical property data could not be obtained for several areas in the borehole because either the core disinte- grated during recovery or the lengths of recovered core were inadequate for the preparation of test specimens. These areas are indicated in figure 5. After the hole was cored, the bottom 100 ft was reamed to 6 in for geophysical logging and subsequent installation of the borehole instruments. Unfortunately, by the time geophysical logging was con- ducted, the tools were unable to proceed beyond 16 ft above the coalbed. The suite of geophysical tools used at the site included caliper, natural gamma, density, resistivity, spontaneous poten- tial, temperature, fluid conductivity, and sonic logs. The sonic and density logs were used to calculate a strength index for the rock above the coal seam where the extensometer anchors were positioned. The following equation was used to give a relative strength value for those sec- tions of the overburden that could not be tested in the laboratory: Eh = Pb x 3.36 x 10 9 Scale, ft FIGURE 4.— Borehole directional survey of station 1. ' d (6t)2 where pb = bulk density, g/cm 3 , 6t = interval transit time, us/ft, and Ed = dynamic elastic modulus of deformation, psi. This equation was developed by Schlum- berger Well Services (12) and relates the sonic and density logs to the dynamic elastic modulus of deformation. The 17 500 r 510 520 530 540 - 550 560 0. LJ Q 570 580 590 600 610 620 - 630 L GE0L0GIC DESCRIPTION Stratigraphic Anchor column locations DYNAMIC ELASTIC MODULUS OF DEFORMATION, I0 6 psi I 2 UNIAXIAL BRAZILIAN COMPRESSIVE TENSILE STRENGTH, STRENGTH, I0 4 psi I0 2 psi 30 I 2 30 3 ROCK QUALITY DESIGNATION, pet 6 9 12 15 25 50 75 100 Cross-bedded sandstone ."•'.-■■ .'-f.V" : Sewickley Coalbed Green clayey shale Dark-gray shale - Shaly limestone Massive limestone Interbedded black shale and limestone Dark-gray shale Massive limestone Interbedded green shale and limestone Massive limestone Green claystone Black shale Fine-grained massive limestone Clayey shale Green shale Clayey shale Green shale Clayey shale Black shale with interbedded coal Pittsburgh Coalbed Green-sandy shale FIGURE 5.— Geologic and rock strength characterization of borehole 1. value is not to be regarded as an abso- lute strength value but rather as an up- per limit of the possible strength of the rock. As compared with actual laboratory tests of specimens, the strength index showed a good correlation between low in- dex values and corresponding low strength test values. For the higher strength test values, the strength index often indicated a relatively higher strength than did the strength value that resulted from laboratory tests. Bond (12) offers an explanation for this relationship: ... a competent appearing formation could be fractured enough to weak- en the rock structurally but not enough to create an observed effect on the logs. On the other hand, formations appearing weak on the strength index curve could not be considered stronger than the cal- culated index. Therefore, the strength index should be considered an indication of the upper limit of bed competence. 18 In addition to physical property test- ing of the immediate roof rock, in-mine geologic mapping was conducted in the gate roads of the monitored panel. All clastic dikes, slips, and roof falls were recorded as shown in figure 6. A high frequency of clastic dikes (often referred to as clay veins) is found throughout the study area. These dikes are characterized as normal fault-type fractures infilled with a clay matrix and inclusions of coal, sandstone, and shale. The dip of the normal fault-type frac- tures ranges from the vertical to 45°, with as much as 3.3 ft of vertical dis- placement along the fault plane. At this site, the dikes had no obvious preferred orientation, although their relatively close spacing may have facilitated cav- ing of the gob. Otherwise, observations showed that the dikes adversely affected ground control only in isolated areas. Coal cleat measurements were also taken at the site. The mean orientations of the butt and face cleat were determined to be N 25° E and N 65° W, respectively. The direction of mining was subparallel with the face cleat at N 60° W. Although no joints were found within the roof rock, other studies have shown that coal cleat orientations often mirror the ori- entations of joints in overlying strata oco ODOl OQDCD DQDD DODQ DDDCD DODC OOCO ocxr ODD .OODD DQQCDDDDO. QOQCDODDDa Panel 3 March line SOD lODODQaCJO ClScx] GATE ROAD 9- RIGHT GATE ROAD 10-RIGHT \J FIGURE 6.— Roof falls and geologic anomalies of gate road entries adjacent to study panel. 19 ( 11 , 13-14), Thus, since the face line of the panel was subparallel with one of the major orientations of the coal cleat, it is possible that jointing in the over- burden may have contributed to the caving characteristics of the gob. Overall, the rock mass characterization supports the favorable longwall mining conditions that are evident at the mine. The immediate roof is strong enough to remain stable between the tip of the shield line and the face, and weak enough to allow immediate collapse directly be- hind the shield line. The weak, clayey zones of the immediate roof appear to al- low for a consistent caving height, and, thus, consistent loading on the shields. INSTRUMENTATION Each instrumentation site was comprised of a surface monitoring station and a subsurface instrument installation as shown in figure 7. A premining surface elevation survey was performed at each site to establish an elevation datum, and successive surveys took place at inter- vals during panel extraction. These sur- veys allowed differentiation between surface subsidence and subsurface strata activity. Each hole was instrumented with a multiple-point borehole extensom- eter and an inclinometer casing, which allowed measurement of vertical and hori- zontal displacements, respectively. Ver- tical measurements were made by direct readout of extensometer scales, continu- ous recording units connected to the ex- tensometer system, and direct readout of a magnetic settlement probe system. Lat- eral displacements were calculated from inclinometer probe measurements. MONITORING OF VERTICAL STRATA DEFORMATION To measure vertical displacement of substrata, each borehole was equipped with a multiple-point borehole extensom- eter; this device detects vertical strata movement through the use of mechanical spring anchors. The anchors were posi- tioned at specific intervals within the strata and connected to the surface by stainless steel wires. Depth intervals for anchor placement were selected fol- lowing an analysis of the recovered core and the geophysical logs. Each anchor was positioned within a distinct strat- igraphic member. Interfaces between stratigraphic units were avoided on the assumption that caving would be most likely to occur along these planes. Pulley for lowering inclinometer or magnetic settlement probe Optional cover Rotary potentiometers and pulley assembly Inclinometer or magnetic settlement probe Spring anchor Inclinometer, 1.9-inOD Centering guide FIGURE 7.— Complete extensometer-inclinometer system. (Courtesy Roctest, Inc., Pittsburgh, NY) 20 Each of the two 6-in boreholes accommo- dated eight anchors, the maximum number that could be used in this borehole diam- eter. Eight sections of 1.9-in-OD poly- vinyl chloride (PVC) inclinometer casing were prepared at the factory to accept the eight anchors. The remaining sec- tions of inclinometer casing were stan- dard 5-ft sections. The anchor springs were compressed and held closed during installation by nylon strings, which passed through the casing and were at- tached to opposite pairs of anchor springs (fig. 8). The anchors were posi- tioned on the casing at the predetermined depths shown in figure 5. The 5-ft sec- tions of casing were glued together and lowered down the hole. A grout tube was fastened to the first section of casing and lowered as the system was being as- sembled (fig. 9). Grouting of the bore- hole was necessary to seal any water- bearing zones, which could have caused water inflow into the mine when the bore- hole was undermined. The casing remained centered in the hole by means of two sets of PVC centering blades installed on the casing, 5 ft above and below each anchor (fig. 10). A 1/16-in-diam stainless steel wire surrounded by 1/4-in-diam oil- filled nylon tubing was attached to each anchor. The tubing was necessary to al- low free movement of the wire after the hole was grouted. A wire-tubing assembly was attached to each of the eight anchors and lowered with the anchor and casing assembly. The grout tubing and each of the wire-tubing assemblies were posi- tioned on scaffolding and lowered into the borehole as the 5-ft sections of cas- ing were added. When the entire assembly had been lowered into the borehole, the anchors were set in place by dropping a weighted knife down through the casing to cut the nylon strings. A reference head was placed at the top of each borehole. The head consisted of a 6-in-OD steel pipe with a welded circu- lar steel plate used to seat eight poten- tiometer-pulley assemblies (one for each anchor). Each anchor wire passed through the center of the instrument reference head,, over its own pulley, and was fixed with a 50-lb tensioning weight. Grad- uated scales were fastened to the outside circumference of the head to allow direct 9m . m FIGURE 8.— Multiple-point borehole extensometer anchor positioned on Inclinometer casing. FIGURE 9.— Grout tube attached to lead section of in- clinometer casing. 21 FIGURE 10.— Borehole anchor and PVC centering blades. FIGURE 11.— Multiple-point borehole extensometer refer- ence head. readout of displacements. For remote readout, the potentiometer leads were soldered onto a terminal panel to which a continuous recorder was connected. The head also incorporated a large pulley for lowering the inclinometer and magnetic settlement probe (fig. 11). The magnetic settlement probe, which works by magnetic inductance, was used to verify anchor locations in the strata (fig. 12). Magnetic rings were incorpo- rated into each of the eight borehole an- chors, creating a magnetic field inside the inclinometer casing. A reed switch probe was connected to a graduated cable mounted on a cable reel. An audible buz- zer housed inside the cable reel was ac- tivated by entry of the probe into the localized magnetic field produced by the anchor. This device provided excellent results for verification of anchor loca- tions. However, after mining progressed beneath the borehole, high concentrations of methane began propagating up the bore- hole, preventing further use of the non- permissible magnetic probe and the con- tinuous recording unit. MONITORING OF HORIZONTAL STRATA DEFORMATION An inclinometer probe was used to mea- sure the progressive changes in the angle of inclination of the inclinometer cas- ing. These measurements provided an evaluation of lateral movement as mining approached the station. The probe was supported laterally in the casing by guide wheels and suspended vertically by a cable connected to a reel and readout unit. The guide wheels traversed oppos- ing longitudinal grooves spaced equally 90° around the inside circumference of the casing for directional control. Two 22 servoacceleroraeters , mounted with sensi- tive axes 90° apart, simultaneously moni- tored inclination both parallel and per- pendicular to the direction of mining. Recording of data was accomplished by the use of a digital indicator equipped with a magnetic tape cassette recorder. Although the inclinometer probe output is recorded in terms of the angle of in- clination, lateral deflections can be calculated easily from these data. Fig- ure 7 illustrates the complete instrument arrangement. FIELD DATA ANALYSIS Base reference data were established at each station approximately 500 ft in ad- vance of mining, and readings were taken relative to the longwall face position. Surface elevation datums were also estab- lished for each borehole prior to mining. Instrument readings were taken weekly while the face was more than 200 ft from the stations, and daily when the face was within 200 ft of the instruments. Face advance was obtained from the mine daily, and surface elevation surveys were made periodically after the boreholes. mining passed beneath FIGURE 12.— Magnetic settlement probe. EXTENSOMETER DATA ANALYSIS To establish the progress of caving, borehole anchors were positioned in the mine roof strata as shown in figure 5. The deepest achor in BH1 was located 23.5 ft above the coal seam. A distance of 10 ft separated anchors 8 through 3, with anchors 2 and 1 separated by 15 and 30 ft, respectively. It is important to note that the anchor displacements shown are with reference to the extensometer head located at the sur- face. The movements of the extensometer head were determined by surface elevation surveys. The displacements shown in fig- ures 13 and 15 have not been corrected to include the measured movements of the surface. Initial anchor movement was detected in all anchors when the face had approached to within 500 ft of station 1 (fig. 13). Since it cannot be associated with cav- ing, this movement has been attributed to lateral displacements of strata and to a rising surface elevation in advance of the face (fig. 14). As the face drew near and passed beneath the station, an- chor positions were recorded hourly based on the assumption that large movements would be seen immediately after the long- wall supports passed beneath the bore- hole. However, anchor movement was not detected at that time. Figure 13 shows that significant movement (caving re- lated) in anchors 2 to 8 began when the face was 35 ft past the station. Anchor 1 (the farthest from the extraction) be- gan moving when the face was 75 ft past the station, at the same time that sur- face subsidence began (figs. 13-14). The abrupt failure of the immediate roof 23 Ll.1 o < _l Q. CO -400 -200 DISTANCE, ft 600 FIGURE 13.— Station 1 extensometer displacements. associated with the advance of the long- wall supports was not detected by anchor 8, which was positioned 23.5 ft above the extraction. This indicates that 23.5 ft is above the upper limit of the first strata separation. Descriptive geologic logging of the immediate strata revealed three very weak bands (approximately 6 in thick) of soft clayey shale in the imme- diate 25 ft of roof. These weak zones occurred at heights of 8, 17, and 25 ft, and it is assumed that immediate caving behind the supports occurred up to one of these zones. Using these zones as possi- ble caving horizons and relating each to the following equation yields three dif- ferent bulking factors for the extraction height of 5.8 ft. H = c + h H = ck \J.O 1 ' 1 ' 1 .4 - JK^/\ - 0< — **- - .4 - \ - UJ ~ o - ,R k — •z. UJ \ n in' -1.2 — \ — 00 3 CD -1.6 •k^^^ — -2.0 - \ -2.4 - * • - -? 8 1 1 i 1 > -200 200 400 600 FACE POSITION, ft FIGURE 14.— Surface elevation survey data for station 1. where h = extracted height, ft, c = height of caved material from roof level of extracted height, ft, H = distance from floor level to caving horizon, ft, and k = bulking factor, unitless. Substitution yields k - - + 1. c The caving horizons of 8, 17, and 25 ft yield bulking factors of 1.72, 1.34, and 1.23, respectively. The fact that anchor 8 did not begin to detect small movements until the face had passed 35 ft beyond station 1 suggests that immediate caving behind the supports occurred at some height less than 23.5 ft. The bulking factor of 1.34 for the 17-ft horizon closely corresponds to the commonly used value of 1.33 for bulking of shale, which comprises 30 ft of the immediate roof. Therefore, caving is assumed to have oc- curred up to the 17-ft horizon. Surface subsidence and anchor movement occurred concurrently when the face was between 75 ft and 290 ft past the borehole (figs. 13-14). Maximum surface subsidence dur- ing this period was 2 ft, and maximum ex- tensometer movement of anchor 8 was 4.25 in. The fact that both surface and small subsurface movements occurred at the same time and at a distance of 75 ft beyond station 1 suggests that the entire over- burden member above the assumed caved height of 17 ft began to sag and compact the gob material. Although the most subsidence occurred between 75 and 290 ft, surface movement did not cease until the face was 530 ft past the station. Anchor movement ceased at 290 ft. This difference is attributed to the closure of fractures in the strata above the anchors. Closure is seen as an apparent upward movement of anchors. INCLINOMETER SURVEY DATA ANALYSIS The inclinometer was used to detect lateral deflections in advance of the longwall face. The inclinometer probe measured the angle of tilt of the casing within the borehole in two directions, parallel and perpendicular to the center- line of the panel. The following analy- sis will refer to an A and B direction. The A direction is parallel to the direc- tion of mining. Positive A deflections are movements toward the direction of mining, and negative deflections are in the opposite direction. The B direction refers to deflections perpendicular to the direction of mining. Positive de- flections are movements toward the previ- ously mined panel, and negative deflec- tions show movement toward the adjacent unmined panel. The inclinometer casing in borehole 1 was installed to a depth of 604 ft, which was 16 ft above the mined height of the Pittsburgh Coalbed. Initial readings were established 290 ft in advance of the approaching longwall face. Although the inclinometer casing was lowered to a depth of 604 ft, the initial readings Readings were approached and toring station, could only be taken to a depth of 489 ft. This blockage was attributed to the sharp deviation in direction of the borehole at a depth of 500 ft, as shown on the di- rectional survey that was performed im- mediately after the hole was drilled (fig. 4). taken daily as the face passed beneath the moni- but they had to be dis- continued after the face had passed 5 ft beyond the borehole because high concen- trations of methane were vented up the inclinometer casing. Inclinometer sur- vey data show that the vertical strata movements discussed in the previous sec- tion were accompanied by lateral deflec- tions of the strata. Figure 15 shows the variation of lateral deflection of bore- hole 1 relative to face advance. Figure 15A shows movement in the A direction (paralle to the direction of face ad- vance), and figure 15B shows movement in the B direction (perpendicular to the di- rection of face advance). Figure 15 re- veals that shear zones begin forming at various depths in the borehole 263 ft in advance of the face. Large deviations from the initial reading began to occur I A, Parallel KEY Face positions 263 ft 100 200 300 DEPTH, ft 400 500 FIGURE 15.— Lateral displacements of borehole 1 parallel and perpendicular to the direction of face advance. 25 when the face was 157 ft from the bore^- hole. In both the A and B directions, movement began to develop at a depth of 112 ft. Geophysical logs show a soft fire clay at this depth. Two other areas of activity at this face position occur at depths of 325 and 370 ft, where the strata again are composed of fire clay. When the face advanced to within 35 ft of the borehole, greater movements were apparent. Three distinct areas are dis- cernible. The aforementioned depths of 112, 325, and 370 ft show relatively large displacements (fig. 15). At 325 ft, a 15-ft member of fire clay exists. A maximum deflection of 4.2 in can be seen at 360 and 480 ft. The shear zone begins to form at 460 ft in a 5-ft layer of carbonaceous shale that is situated between an upper member of sandstone and a lower member of limestone. When the face passed beneath the station and was 5 ft beyond the borehole, a shear zone developed at a depth of 179 ft, prohibit- ing the probe from progressing past this point. A final reading taken after face advance had progressed more than 195 ft past the station revealed that the high- est progression of shear had extended to within 46 ft of the surface. RESULTS Significant anchor movement and surface subsidence was detected when the face had progressed 35 ft past station 1. Surface and subsurface movements continued to in- crease until the face had advanced to 290 ft past the station. At this point posi- tive anchor deflection ceased. Surface subsidence continued slightly to a face position of 530 ft. Differential surface and subsurface displacement is attributed to the closing of fractures in the strata above the anchor positions. This ap- peared on the reference head as upward anchor movement. Anchor 8 (the deepest) in station 1 did not detect an abrupt failure of the roof immediately after passage of the supports beneath the borehole but moved a total of 4.25 in as the face moved 310 ft past the station. This indicates that caving of the immediate roof occurred less than 23.5 ft above the top of the coalbed. Thus, the actual bulking factor must be greater than 1.25. Three very weak bands of clayey shale are present between the top of the coalbed and anchor 8. It is assumed that immediate caving occurred up to one of these weak horizons. Calculation of a bulking factor based on caving to the 17- ft horizon yields an estimated bulking factor of 1.34. Inclinometer surveys show the formation of a number of shear zones throughout the length of the borehole. Shear zones were detected in station 1 263 ft in advance of the face. As the face drew nearer to the station, lateral displacements in the borehole became more apparent. A compar- ison of inclinometer survey data and geo- physical logs revealed that shear zones are associated with weak strata horizons. These weak horizons occurred in fire clay material at depths of 46, 112, 325, and 370 ft. Another displacement of 480 ft occurred at a 5-ft layer of carbonaceous shale. As the face progressed beneath the borehole, a shear zone developed at a depth of 179 ft and prevented further inclinometer readings. The final shear zone detected was at a depth of 46 ft. CONCLUSIONS The primary goal of this investigation was to define the height of caving imme- diately behind the advancing longwall supports. Previous estimates of caving height, predict the height of caving to be two times (%) and four times (9) the extraction height. This investigation revealed that caving occurred less than 23.5 ft above the coalbed, most likely at a height coincident with one of three clayey shale zones. These zones are lo- cated 8, 17, and 25 ft above the coalbed, and calculation of bulking factors based on caving to each of these horizons yields values of 1.72, 1.34, and 1.23, respectively. The fact that 1.34 closely 26 corresponds to the commonly used bulking factor for shale (1.33) suggests that the caving horizon occurred at 17 ft, or three times the extraction height. Re- sults of a field study by Matthews (_5) revealed similar results. Inclinometer data revealed another characteristic of longwall strata behav- ior. Comparison of inclinometer data with geophysical logs showed that the major lateral deflections occurred in weak strata (i.e. , fire clay and carbona- ceous shale). Based on these observations, the behav- ior of strata over longwall panels ap- pears to be largely dependent upon lith- ology. Future studies should allow the caving behavior of various lithologies to be characterized. REFERENCES 1. Virginia Polytechnic Institute & State University. Design Optimization in Underground Coal Systems, Volume 10: Un- derground Longwall Ground Control Simu- lator. U.S. DOE AC01-76ET10722, Final Rep. , 1981, 235 pp. 2. Kidybinski, A. Classification of Rocks for Longwall Caveability. Paper in State-of-the-Art of Ground Control in Longwall Mining and Mining Subsidence, ed. by Y. P. Chugh and M. Karmis. Soc. Min. Eng. AIME, 1982, pp. 31-37. 3. Ghose, A. K. Assessment of Caving Characteristics and Support Resistance for Longwall Mining Under Massive Coal- measure Roof Rocks. Paper in Stability in Underground Mining II, ed. by A. B. Szwilski and C. 0. Brawner. Soc. Min. Eng. AIME, 1984, pp. 460-471. 4. Bieniawski, Z. T. Rock Mechanics Design in Mining and Tunneling. A. A. Balkema, 1984, 272 pp. 5. Matthews, J. Specifying and Ac- quiring Longwall Shield Supports. Paper in Rock Mechanics: Key to Energy Produc- tion, ed. by H. L. Hartraan (Proc. 27th U.S. Symp. Rock. Mech.). Soc. Min. Eng. AIME, 1986, pp. 360-366. 6. Barla, G. B. , and S. Boshkov. In- vestigations of Differential Strata Move- ments and Water Table Fluctuations During Longwall Operations at the Somerset Mine No. 60. U.S. DOE contract ET-76-C-01- 9041, Oct. 1978, 51 pp.; NTIS FE-9041-1. 7. Schaller, S., and B. K. Hebble- white. Rock Mechanics Design Criteria for Longwall Mining at Angus Place Colli- ery. Australian Coal Industry Research Laboratories Ltd. , May 1981, 84 pp. 8. Wilson, A. H. Support Load Re- quirements on Longwall Faces. Min. Eng. (London), v. 134, No. 173, 1975, pp. 479- 491. 9. Wade, L. V. Longwall Support Load Prediction From Geologic Information. Trans. Soc. Min. Eng. AIME, v. 262, 1977, pp. 209-213. 10. Barry, A. J., 0. B. Nair, and J. S. Miller. Specifications for Se- lected Hydraulic-Powered Roof Supports. BuMines IC 8424, 1969, 15 pp. 11. Deere, D. V. Technical Descrip- tion of Rock Cores for Engineering Pur- poses. Felsraech. und Ingenieurgeol. , v. 1, 1963, pp. 18-22. 12. Bond, L. 0., R. P. Alger, and A. W. Schmidt. Well Log Applications in Coal Mining and Rock Mechanics. Trans. Soc. Min. Eng. AIME, v. 250, Dec. 1971, pp. 354-362. 13. Parker, J. M. Regional Systematic Jointing in Slightly Deformed Sedimentary Rocks. Geol. Soc. America Bull. , v. 53, 1942, pp. 381-408. 14. Ver Steeg, K. Jointing in the Coalbeds of Ohio. Econ. Geol. , v. 37, 1942, pp. 503-509. 27 MULTIPLE-SEAM MINING PROBLEMS IN THE EASTERN UNITED STATES By Gregory J. Chekan, 1 Rudy J. Matetic, 1 and James A. Galek^ ABSTRACT The Bureau of Mines, in an effort to improve planning and development in coal mining, is currently investigating strata interactions associated with mining of multiple coalbeds. Strata interactions between adjacent coalbeds occur frequent- ly in the Appalachian coalfields and can have unfavorable effects on both product cost and worker safety. Two common in- teractions that occur between adjacent coalbeds are subsidence and pillar load transfer. At two mine sites where such ground interactions were present, the Bu- reau conducted geologic studies and gath- ered various geotechnical Information on pillar and entry stability using rock mechanics instrumentation. At the mine affected by pillar load transfer, the results of the study show: (1) overbur- den depth changed dramatically in the study area and reached a maximum at the study site, (2) innerburden thickness was less than a pillar width (40 to 45 ft), (3) heaving was experienced in both mines where sandstone and/or shale floor units were observed, (4) overlays of the mine layout show pillars were not totally superpositioned, (5) a maximum of 5 in of roof-to-floor convergence was measured, and (6) monitoring of pillar pressures showed that only the pillar core was loading, an indication of a stiff pillar approaching failure. At the mine affected by subsidence, measurements show that the undermining had little effect on upper mine pillar stability, but had a more severe effect on the development and maintenance of entries. Roof-to-floor measurements re- corded over four times more convergence in entries developed over gob than in en- tries developed over support pillars in the lower mine. This has led to major roof falls in the active panels, result- ing In production delays and supplemental roof support costs. INTRODUCTION Simultaneous mining of adjacent coal- beds or mining over or under a previously mined-out coalbed occurs frequently in the Appalachian Region of the Eastern United States. In West Virginia, Penn- sylvania, and Ohio, it is estimated that 57 billion st of coal exists in a mul- tiple-seam configuration. West Virginia alone has over 50 minable seams (J_). In the past, mining sequence was based pri- marily on availability and economics with little regard to the effects mining would have on other coalbeds both above and below the one being mined. This has strong implications for resource con- servation, especially if these practices 1 • • Mining engineer. Engineering technician. Pittsburgh Research Center, Mines, Pittsburgh, PA. Bureau of continue. Problems that result from strata interactions could render these resources unminable unless methodologies and techniques are developed that allow for economical extraction. To verify these interaction mechanisms in the field and their influence on mine ground stability, the Bureau conducted geological and geotechnical investiga- tions. The purpose of these studies is to develop a better understanding of sub- sidence and pillar load transfer and its effects on current workings. Eventually, this knowledge will lead to improvements in mine planning and development. •^Underlined numbers in parentheses re- fer to items in the list of references at the end of this paper. 28 PILLAR LOAD TRANSFER Pillar load transfer, as the name im- plies, involves the transfer of load pil- lars in overlying workings to pillars in underlying workings. This interaction occurs particularly when coalbeds are in close proximity, less than 110 ft apart (2— j4) , and either isolated, remant pil- lars (barriers) or many strong, competent pillars are present in the upper work- ings. This condition may serve to con- centrate stresses in the innerburden, causing ground instability in the lower workings. The mechanics of load transfer and distribution from overlying opera- tions have been analyzed extensively by other researchers through the use of mathematical and photoelastic models (3_- Two theories have been developed to ex- plain ground disturbances due to load transfer from adjacent workings: pres- sure bulb theory and arching theory. Pressure bulb theory (3_-4) assumes that the pillar is the major structural ele- ment in the transfer of load. It is use- ful in analyzing pillar load transfer when a "passive" interaction occurs. This condition is met when pillars are columnized and lower seam pillars are sufficiently large to prevent them from yielding. Arching theory (3-_4) assumes that the mine opening is the major struc- tural element in the transfer of load. Load transfer is the result of the pres- sure arch that forms around the mine opening upon excavation. This theory is useful in analyzing pillar load transfer when a "reactive" interaction occurs. This condition is met when lower seam pillars yield and redistribute their load to larger barriers or abutments. CASE STUDY Mine Location and Geology The study mines are located in Raleigh County, WV, as shown in figure 1. The company is operating in two superimposed coalbeds. The upper mine is located in the Peerless Coalbed, which is approxi- mately 72 in thick. The lower mine is located in the No. 2 Gas Coalbed, which is approximately 48 in thick. The mines are separated by approximately 40 to 45 ft of innerburden. The overburden consists predominantly of sandstone with interbedded shale units of varying thickness. The innerburden is a predominant sandstone with some inter- bedded shale units. Depth Depth in relation to all room-and- pillar mining operations is critical as overburden increases (3_-_4). The overbur- den above the lower mine at the study site is approximately 1,000 ft, the in- nerburden is approximately 40 to 45 ft thick, and approximately 960 ft of cover is located above the study site in the upper mine. Overburden depth changes dramatically in the area and reaches a topographic high over the study site. I nnerburden Thickness and Physical Characteristics Interval thickness between two coalbeds is also a critical variable. Figure 2 is an innerburden isopach map constructed from available corehole information. As shown on the figure, approximately 43 ft separates the upper and lower mines at the study site. Core logs show that sandstone comprises 77 pet of the innerburden. KEY MAP Scde ' mileS FIGURE 1.— Location of study mines. 29 LEGEND X Instrumented pillars Contour interval = 5ft O Drill holes FIGURE 2.— Innerburden isopach map. l_ 800 Scale, ft According to an equation developed by Haycocks and Karmis (3j-_4) , approximately 78 ft would be the innerburden spacing above which no interaction damage may re- sult from room-and-pillar mining. The actual innerburden with respect to the study site is approximately 40 ft, con- siderably less than the calculated mini- mum value. However, it should be noted that the equation is independent of pil- lar and entry design and is derived from a rather limited data set and therefore may not represent all stable and/or un- stable mining conditions. Results of In-Mine Mapping A geologic investigation was performed within both the upper and lower mines. The areas of floor heave in both mines are underlain by a rooted, fine-grained, well-cemented, hard sandstone and/or shale, and the roof consists of either a hard, banded siltstone or sandstone. Roof structure was generally uniform ex- cept for some slight undulations in the shale-sandstone contact. The immediate roof rock in the heave zones of the lower mine appears to be sandstone. In areas where a shale roof was observed, no floor heave was detected. The heave zones in the upper mine could not be correlated to roof rock lithology since sandstone is present throughout. Figures 3 and 4 rep- resent the results of in-mine mapping. Figure 3 shows the location of floor heave around the study site in the upper mine. Note that the mine roof is chiefly sandstone near the study area and the floor is comprised of hard, fine-grained, 30 Face Preferred cleat Trends a LEGEND Sandstone in immediate roof Siltshaie roof^ hard, banded with ironstone (siderite) streaks and nodules Instrumented pillar wvw Floor heave or buckle 1_ 200 Scale, ft FIGURE 3.— Results of in-mine mapping of upper mine. Buff sandstone Gray sandstone LEGEND ^] Sandstone in immediate roof ] Siltshaie roof-, hard, banded with ironstone (siderite) streaks and nodules J Instrumented pillar Floor heave or buckle Mine track Silt- shale 200 I Scale, ft FIGURE 4.— Results of in-mine mapping of lower mine. 31 rooted sandstone. Figure 5 was con- structed from core logs and displays floor lithology in the area of the study site. Note the lateral changes in the immediate floor lithology. The immediate floor is comprised of both a high-modulus material, a sandstone, and a low-modulus material, a shale. In the heaving experienced near the study site, there was a sandstone floor. Inby and outby the study site, major floor heaving was experienced where a shale floor was present. Floor heaving experienced in these areas was observed to be humplike structures characteristic of a low-modulus material. A buckling type of failure was also observed where a sandstone floor was present. It may be assumed that major concentrations of load can be transferred from a shale (low- modulus material) to a sandstone (high- modulus material), which may result in movement within the sandstone floor. Figure 4 represents the results of the in-mine mapping performed within the lower mine. The immediate floor located within the heave zones is also comprised of sandstone. Heave zones located in shale floor were found and also demon- strated humplike structures. The trans- fer of load from a shale to a sandstone unit could also be related to the ground conditions located within the lower mine. Mining Engineering Design Parameters Seam Sequencing The upper mine area was driven in June 1980, and the same section located in the lower mine was driven during December 1982. Major floor heaving and excessive pillar loading were observed in October 1984 within the lower mine. Approximate- ly 3 to 4 months later the upper mine experienced excessive entry convergence and pillar loading. Affected areas are shown in figures 3 and 4. 13 20 a. g o 10- 20 Shale • Sandstone \ 15 Upper mine floor- Shale \ Coal Shale Shale Coal Sandstone \ /■\ Sandstone Lower mine floor 588 Shale '■'i Sandstone v Sandstone ^L. s Shale Shale Sandstone 'f LEGEND 15 Drill hole 588 Depth, ft 629 Shale Sandstone Shale FIGURE 5.— Floor lithology In study site area. 32 Superpositioning of Pillars The use of coluranlzed pillars is stan- dard practice in multiple-seara mine design. Columnization of the pillars lessens the effect of interaction that may be transferred from overlying work- ings. Figure 6 represents the superposi- tioning of pillars at the study site. Although very difficult to achieve, this practice requires alignment of pillars of similar size for both seams. Both the upper and lower mines were driven with pillars on 70-ft centers. Note from fig- ure 6 that pillars and entries are near- ly, but not totally, superpositioned. stations. The pillars selected for the study (fig. 6) are nearly superpositioned and have equivalent dimensions. Four BPF's and 12 convergence stations were installed in the upper mine (fig. 7), and 5 BPF's and 12 convergence sta- tions were installed in the lower mine (fig. 8). The estimated setting pressures for the BPF's were calculated using the tributary area method. 4 At the time of installa- tion, depth and extraction ratio were estimated at 720 ft and 0.35 respective- ly. This gives setting pressures of 1,200 psig for the upper mine and 1,300 psig for the lower mine. It was later Instrumentation and Results Instrumentation 4 (d) Estimated pressure, psi = 1.1 psi/ft 1 (1-R) Instrumentation installed in both mines included borehole platened flatjacks (BPF) (_5) and removable convergence where d - depth, ft, and R = extraction ratio. LODDQD D G a a LEGEND Instrumented pillar Upper mine Lower mine 1 00 Scale, ft FIGURE 6.— Superpositioning of instrumented pillar. LEGEND o Borehole platened flatjack • Convergence station 40 Scale, ft FIGURE 7.— Instrument location in upper mine. 33 LEGEND o Borehole platened flatjack • Convergence station 40 _l Scale, ft FIGURE 8.— Instrument location in lower mine. determined that depth and extraction ra- tio were underestimated and were actually 960 ft and 0.50. As a result, setting pressures for the upper and lower mines should have been 2,100 psig and 2,200 psig respectively. Although the original setting pressures were low, these pres- sures do not directly affect the recorded results. Any increase in pillar pressure MMJMt Ml j 1 ! 1 20 40 60 80 100 120 140 160 180 TIME, days FIGURE 9.— Pressure increase versus time in upper mine. above 2,100 psig and 2,200 psig would be a result of relative increases in pillar pressure. Results The monitoring of the instrumentation continued for a total of 177 days. The instruments were monitored at least once a week. Tables 1 and 2 provide results at 49 days (approximately 25 pet of the study period), 92 days (approximately 50 pet of the study period), and 177 days. All increases in pillar pressure for the upper mine were from BPF's installed within the core of the instrumented pil- lars. The BPF's located at 10-ft depths showed no increase in pressure. Figure 9 shows pressure increases versus time for TABLE 1. - Flatjack (BPF) pressure during 177-day monitoring period, pounds per square inch BPF Initial (instal- Day 49 (25 pet Day 92 Day 177 lation date) of total period) (50 pet) (final) UPPER MINE 1,100 8,100 8,100 8,100 1,225 950 900 950 3 1,200 3,175 4,050 5,100 4 1,275 1,200 1,250 1,250 LOWER MINE 5 1,300 1,090 1,050 1,050 6 1,000 880 850 850 1,300 1,050 1,000 1,050 8 1,300 900 900 900 10 1,200 1,000 1,000 950 34 TABLE 2. - Results of convergence monitoring in upper mine, inches Station Day 49 Day 92 Day 17 7 (final) Station Day 49 Day 92 Day 177 (final) 1 1.40 .84 ND .60 1.24 1.32 1.95 1.25 ND 1.29 1.88 2.03 3.54 2.13 ND U.so 2.71 3.37 7 1.15 .81 2.23 .94 .80 1.40 1.85 1.26 3.26 1.31 1.13 1.45 3.00 2 8 2.08 3 9 5.00 4 10 2. 12 5 11 1.92 6 12 2.24 ND No data because station was destroyed. Monitoring discontinued on day 134 owing to bad roof conditions. all BPF's installed in the upper mine. No increases in pillar pressure were re- corded from BPF's installed in the lower mine. This was due to height restric- tions which limited instrument installa- tion to more stable mine areas. Maximum convergence of 5 in occurred at station 9, which is located in the track entry within the upper mine. Roof-to- floor convergence within the upper mine increased very rapidly, averaging 0.5 in of closure every month. Figures 10 and 11 show total convergence within the up- per mine after 49 days (25 pet of moni- toring period) and 177 days (100 pet of monitoring period). Note the trend of movement with respect to figure 11. Movement occurred outby and in a south- west direction in relation to the instru- ment array. Figure 12 displays total convergence versus time for the four con- vergence stations located along this LEGEND 40 -— -1.5 — Convergence contour line; Sca | e fr contour interval 0.5 in • Convergence station FIGURE 10.— Upper mine convergence contours after 49 days of monitoring. LEGEND 40 —-I-5 — Convergence contour line; Scale, ft contour interval 0.5 in • Convergence station FIGURE 11.— Upper mine convergence contours after 177 days of monitoring. 35 trend of major roof-to-floor movement. Roof-to-floor convergence monitored in the lower mine was limited. The most movement that occurred was 0.13 in at station 11. Figure 13 represents total convergence (117 days) monitored within the lower mine. UJ o z UJ o cr ui > z o o 3 - 2 - I - KEY ^^ Station: ^^"^ • 10 -^ A j^**^ 20 40 60 80 100 120 140 160 180 TIME, days FIGURE 12.— Total convergence versus time in upper mine. LEGEND -O.i — Convergence contour line; contour interval 0.5 in • Convergence station 40 I Scale, ft FIGURE 13.— Lower mine convergence contours after 177 days of monitoring. CONCLUSIONS ON PILLAR LOAD TRANSFER Based on the information received and collected throughout the study, the fol- lowing conclusions can be made: 1. Overburden depth above the study site was approximately 1,000 ft. Prior research and other case studies (_2-4) have shown that excessive overburden depths can lead to unstable ground conditions. 2. Innerburden thickness in the study area was approximately 40 to 45 ft, less than one pillar width. Prior research has shown (3-4) that workings in close proximity, less than two pillar widths, may create ground control problems in, above, and below workings. 3. Prior research has shown (3-_4) that innerburden material comprised mostly of sandstone dampens the effects of pillar load transfer. Sandstone at the study area comprises 77 pet of the innerburden. According to Haycocks (3) this percentage requires 78 ft of innerburden for stable conditions. 4. Heaving was experienced in both sandstone and shale floor units. The shale floor was a low-modulus material, resulting in humplike floor heaving. Whereas the sandstone floor was a high- modulus material, resulting in a buckling type of floor heave. 5. Average convergence in the upper mine entries was 2.50 in, compared with 0.04 in. in the lower mine entries. This difference is due to height restrictions in the lower mine, which limited instru- ment installation to a more stable mine area. 6. To minimize interaction effects, optimum seam sequencing would be to mine the upper seam first to total extrac- tion and then continue downward. In this case, the upper seam pillars were devel- oped first, and the lower seam pillars were developed approximately 2 yr later. Pillar columnization was practiced, but mine overlays show that pillars and en- tries were not totally superpositioned. It is difficult to determine whether columnized pillars cause a "passive" in- teraction through pressure bulb inter- ference or if a "reactive" interaction 36 occurred owing to arching effects. BPF pressure readings in the upper mine show a core loading that is characteristic of a stiff pillar approaching failure. Sim- ilar loading characteristics were not ob- served in lower seam pillars, mainly be- cause height restrictions limited BPF installation to more stable areas. But assuming lower seam pillars also exhibit stiff pillar characteristics, the arch- ing concept could be applicable in this situation. Photoelastic and mathemati- cal models of multiple openings in close proximity, less than two pillar widths, have shown that pressure arch interaction can create zones of excessive pressure in the innerburden (3-4). This is especial- ly the case for strong, competent pil- lars or stiff pillars that do not yield readily, allowing independent arches to form from pillar to pillar. When high abutment pressures associated with arch interaction exceed the in situ strength of the rock, ground failure results. In this case, a soft floor stratum is the weakest member in the mine structure. Strong competent pillars punching into the floor cause considerable floor heave, which is then followed by eventual pillar failure or yielding. When this occurs, their load is transferred to neighboring pillars, forming a secondary arch. This cycle of failure, load transfer, and arch formation continues until sufficient sup- port is encountered to stabilize the load transfer process, such as barrier or abutment pillars. SUBSIDENCE Strata interactions due to subsidence result when an underlying coalbed is ex- tracted first. Undermining subjects the superjacent strata to a mining-induced stress field (6_). The distribution of this stress within the strata is a func- tion of the subsidence process and is most damaging to overlying coalbeds after the critical to supercritical subsidence phase has been reached (4_, _6_-^7). Depend- ing upon the uniformity of lower coalbed extraction, there exists a relatively de- stressed zone toward the middle of the subsidence area. Most ground distur- bances in overlying coalbeds occur toward _ Outer limit of subsidence ,\ § j<> Tension zone (roof cracking and failure) / Roof .■>"">*/ ^compression zone Dead ground Unsubsided ground Subsidence zone FIGURE 14.— Strata flexure in upper coalbed due to sub- sidence. Adapted from Haycocks, Karmis, and Topuz (7). the boundaries of the subsidence trough. Within the trough, strata flexure creates zones of tensile and compressive stress, as shown in figure 14 (17). The extent of this zone is defined by the angle of draw, which is dependent upon the geo- logic and physical characteristics of the strata. As mining develops through this trough, these stresses have a severe ef- fect on entry stability, particularly on the integrity of the roof. Other types of failure in upper coal- beds that are attributed to undermining include interseam shearing and the ef- fects of arching. In interseam shearing, highly inclined shear or tensile-shear failures develop and result in displace- ment of the coalbed into lower excava- tions (4_, 7). Recent studies (4_, 7) in- dicate that the elastic modulus of the superincumbent strata is a major factor influencing this type of failure. Stud- ies have demonstrated that a high-modulus strata such as sandstone is more prone to shear failure. Arching is actually a subcritical phase of subsidence, and its effects on upper coalbeds are dependent upon the opening width-to-depth ratio and the height of the resulting pressure arch. Arching effects can produce a zone of high compressive stress that may cause pillar and roof control problems. 37 CASE STUDY Mine Location and Geology The study mine is located in Greene County, PA, as shown in figure 15 and is operating in areas of the Sewickley Coal- bed that were subsided by undermining of the Pittsburgh Coalbed. The overburden above the Sewickley Coalbed ranges from 425 to 580 ft and consists predominantly of interbedded shale with a competent sandstone unit that varies in thickness. The innerburden ranges from 90 to 100 ft and consists of interbedded shale and limestone. The average height of the Sewickley Coalbed is 5 ft. The immediate roof is composed of a highly jointed dark sandy shale that ranges from 10 to 15 ft thick, overlain by a competent limy shale. The immediate floor is composed of a 3-ft- thick, dark, limy shale underlain by a competent limestone unit. An underground geologic investigation found no geologic anomalies (clay veins, discontinuities, etc. ) in the study area. Additional site-specific information is given in table 3. Instrumentation and Results 1 Left Panel 1 Left was a short 400-ft panel that was developed through a subsided area of the coalbed created by pillar retreat ac- tivities in the lower mine. The boundary TABLE 3. - Site-specific information, 1 Greene County, PA, mine Upper mine Lower mine 3 Av mining height.. Pillar centers.... Av entry width. . . Extraction, pet: • in. . .ft. . , .ft. . 60-64 100x100 18-20 36 100 72 100x100 20 36 100 ^oom-and-pillar continuous mining at both levels. Sewickley Coalbed. Pittsburgh Coalbed. GREENE COUNTY Pennsylvania key map Pittsburgh Umontown PENNSYLVANIA MARYLAND Morgantown Scale, mi FIGURE 15.— Study mine location, Greene County, PA, mine. of this retreat raining in the Pittsburgh Coalbed is shown as the gob line in fig- ure 16. The 1 Left panel started in sub- sidence (over gob) and developed across the gob line and onto pillars located in the lower mine. It was developed and re- treated in less than 50 days. Several years earlier, the 1 East panel crossed the gob line (from over support pillars and onto gob in the lower mine); it ex- perienced two roof falls as shown in fig- ure 16 but encountered no displacements in the coalbed. These ground conditions indicated that the strata directly super- jacent to the gob line were flexed due to subsidence, and mine personnel were an- ticipating these same conditions in the 1 Left panel. To predict the location of the subsidence trough within the coalbed is a complex problem. Although current theories and models are based on surface subsidence, their application for pre- dicting in-mine subsidence could prove useful. Based on a model recently devel- oped by the Bureau (8-9) , the edges of the subsidence trough at the Sewickley Coalbed level were calculated to be ap- proximately 102 ft inby to 300 ft outby the gob line of the lower mine for this case. The edges of the predicted trough and pillar arrangements in 1 Left before retreat mining are shown in figure 17. 38 DDDDDD □DQDDD DDDDnC DDDQDD □DODOD DDDDnDc DDDDDDSmDr I East Panel ODD □DsnnDnrag! DDD nan DDD DDD DDI □□DM DaQDOTa □□□DDD DDI DDD DDDDBa aDDaai nan □CL1DD LEGEND ■ Convergence station cOd Roof falls occurring in I East when mining crossed the gob line GZg Roof falls occurring in I East since mining was discontinued in September 1984 •» Roof falls occurring in 2 East during development Gob line - lower mine t 2 East Panel 200 Scale, ft FIGURE 16.— Location of study areas 1 Left and 2 East showing gobline in lower mine (Pittsburgh Coalbed) and roof fall activity in upper mine (Sewickley Coalbed). To determine if the trough affected pillar loading and stability, four BPF's (12) were installed during development to measure vertical changes in pillar pres- sure. Setting pressure was approximately 1,000 psig for each BPF 5 as overburden was 580 ft with an extraction ratio of 0.36. Figure 17 shows BPF locations in se- lected pillars. During development the instruments recorded no increases in pil- lar pressure. The major ground problem experienced within the trough during de- velopment was the occurrence of a large roof fall, 10 to 12 ft high, over gob as shown in figure 17. This fall started in a crosscut at the face and eventually propagated into the belt entry, causing considerable production delays. BPF pressure increased, as anticipated, when retreat raining approached the in- strumented pillars. Figure 18 shows the pressure changes recorded for the four BPF's during development and retreat 5 See footnote 4 (p. 32). mining. Pressure increases ranged from 400 to 700 psig but were not considered significant because they did not render a pillar or adjacent pillar unminable dur- ing retreat. Observations of roof behav- ior at the face showed the roof strata to fall tight against the pillar line, causing no excessive loading in adjacent pillars from roof cantilevering. The 1 Left panel was completely retreated, but ground water inflow and accumulation slowed retreat operations in the 1 East panel. Eventually, retreat mining was discontinued completely in the 1 East panel due to this condition. 2 East Panel 2 East was a long 2,200-ft panel devel- oped off the 1 South Mains. It started from over support pillars located in the lower mine and developed across the gob line into subsided ground as shown in figure 16. To better understand immedi- ate ground movement as entries were de- veloped through the subsidence trough, 39 I T* r □E ra I East Panel (direction of mining) 16 15 .? 14 w a. CVJ O |3 LU o LU CC CO CO UJ or 12 10 1 1 KEY - BPF No. A | 1 1 x x — ■ 2 • 3 A 4 / / x - x BPF lost / // ~~ to retreat _ mining / 1/ ~ /• BPF /installation x a y /V pressure i / ~ X^ \ Y^_ ^LX __^j ^r .--- Start of retreat 1 ^^ mining _ 1 1 10 20 30 TIME, days 40 50 FIGURE 18.— Pressure changes recorded for BPF locations in 1 Left before development and retreat mining. I oo 200 LEGEND o Borehole platened flatjack o eZZ) Roof fall ' i J^ ' — — Gob line- lower mine Edges of predicted sub- sidence trough FIGURE 17.— Pillar arrangements and BPF locations in 1 Left before retreat mining. roof-to-floor convergence measurements were used in the 2 East panel. A total of 25 convergence stations were installed within the trough area, 14 in entries developed over support pillars and 11 in entries developed over the gob of the lower mine. As shown in figure 16, large roof falls 10 to 12 ft high occurred over gob during pillar development, but with greater frequency than in other gob line crossings. Convergence measurements taken over 143 days showed that the aver- age total convergence measured in entries developed over the gob was over four times the amount measured in entries de- veloped over support pillars. Figure 19 shows cumulative convergence for sta- tions 1, 8, and 13 installed in entries 2.0 80 TIME, days 160 FIGURE 19.— Cumulative convergence for stations 2, 8, 13, 18, 22, and 25 over 143 days. developed over support pillars and sta- tions 18, 22, and 25 installed in entries developed over gob. This graph shows that the rate of convergence is nearly 40 uniform on both sides of the gob line but increases dramatically once the gob line is crossed. Table 4 lists cumulative convergence measured over 143 days for all stations. CONCLUSIONS ON SUBSIDENCE Throughout the subsidence area, the coalbed showed no vertical displacements that would indicate an interseam shear- ing, and no recurring problems where a result of strata flexure. Several obser- vations support this: First, as mining development crossed the gob line in the 2 East panel, roof- to-floor convergence increased, as did TABLE 4. - Cumulative convergence measured over 143 days for all stations Cumulative Station convergence, in 0.11 .15 .16 .13 .19 .12 .15 .24 1.39 .27 .16 .17 .31 .95 .33 .96 1.11 1.56 1.50 2.71 1.46 1.38 1.38 .73 1.33 Entries developed support pillars: 1 over 2 3 4 5 6 7 8 9 10 11..., 12 13 14. Entries gob: 15 developed over 16 17 18 19 20 21 22 23 24 the incidence of roof falls. In figure 20, graph 1 is the predicted subsidence profile (_8~9) for this case, graph 2 shows cumulative convergence measured over 143 days for selected stations 2, 8, 13, 18, 22, and 25 and their locations with respect to the gob line, graph 3 de- picts the cumulative length of roof falls with respect to the gob line, and graph 4 Is a plot of the overburden above the 2 East panel. Graph 1 shows that the sub- sidence trough begins 102 ft outby the gob line and subsidence reaches a maximum at 300 ft inby the gob line. Graphs 2 and 3, which depict ground movements, correlate well with this predicted subsi- dence profile as roof-to-floor conver- gence and frequency of roof falls both increase dramatically as the gob line is crossed. Graph 4 shows there were no dramatic fluctuations in overburden above the panel. Second, all roof falls in the study area occurred in entries developed 500 G *1 = H 450 CO q_ ° 400 1 ■ ' 1 • 1 4 . Direction ot mining ~^ in 2 East / 1 "^\ ■*" ^S ' y - ^^ 1 - 2 in 3 > o -z. o o Q - UJ bJ 1 i ■ ' i KEY 18 Station 1 - ■■► 1 2 8 Hh 18 I 22 I 1 25 I 1 I 2 or w o_ no I 2 3 4 -600 - 1 \ i i ' / - -► \ Gob line H \ 1 N - ' ' i , , i \_ s MAX i -300 300 DISTANCE FROM GOB LINE, ft 600 FIGURE 20.— All graphs in relation to gobiine of the lower mine. 1, Predictive subsidence; 2, cumulative convergence; 3, cumulative length of roof falls; 4, overburden depth. 41 over gob; no roof falls occurred in en- tries developed over support pillars. In addition, the roof fall activity contin- ued in the 1 East panel well after re- treat raining operations were discontinued in September 1984 (fig. 16). Finally, most roof falls displayed a similar type of roof failure, usually along a natural roof joint. As shown in figure 21, these joint surfaces were smooth, having no cohesive properties. This natural jointing system was present throughout the study area, yet as men- tioned earlier, all roof falls occurred in entries developed over gob. Presuming that the roof is in flexure, these joint surfaces would provide natural planes for tension failure. A major set of roof joints was oriented approximately N 55° W, subparallel to the direction of mining. This could further explain the high frequency of roof falls in intersec- tions and entries parallel with the face. The increased incidence of roof falls within the subsidence trough required the installation of additional roof support. Roof stability in entries developed over support pillars was readily maintained with 6-ft conventional bolts on 4-ft cen- ters. Cribs and posts were installed as dictated by the general roof support plan. Entries developed over the gob re- quired more comprehensive roof support consisting of 5-in by 7-in by 16-ft tim- bers bolted on 2-ft centers. In addi- tion, many intersections along the track and belt entries were supported with 6-in steel I-beams set on posts or cribbing. These support requirements, combined with downtime to clean and resupport fall areas, lowered production considerably. Figure 22 shows the average tonnage per shift as the 2 East panel mined through this subsidence trough versus tonnage for a similar panel, 2 North, located in the same mine but not affected by subsidence. Note that the high production values for the 2 East panel are less than the low values for the 2 North panel. As raining progressed farther into the subsided zone, roof conditions improved slightly and production increased, but this sup- plemental support was still required. Based on information collected at this particular study site, the following con- clusions can be made: 400 300 X CO W 200 O O Q 100 o _5 ) , there are considerable case studies available for research and analysis. To date, over 130 separate examples of mul- tiple-seam ground control problems have been collected to facilitate identifica- tion of major controlling factors and the magnitude and limits of interaction under a variety of conditions. Professor of mining engineering. ^Graduate research assistant. Virginia Polytechnic Institute and State University, Blacksburg, VA. Analysis of field experiences reveals four major classes of interactive ground control mechanisms: pillar load trans- fer, innerburden shearing, arching ef- fects, and upper seam subsidence (3, 6). Pillar load transfer mechanisms and in- nerburden shearing have been successfully used to explain interaction phenomena associated with undermining during overmining. 3 Underlined numbers in parentheses re- fer to items in the list of references at the end of this paper. 45 Factors that contribute to interaction may be classified into variables that are fixed by the geologic environment and those that depend on engineering design. These may be described 4 as follows (6^): Fixed: Depth, innerburden thickness, innerburden physical characteristics,' stress field, seam thickness, coal char- acteristics, completed mining operations either above or below the seam to be mined, and age of workings. Mining: Height, dimensions and geome- try, method, and spatial location of entries. Sorting of data revealed that, consid- ering all case studies, 46 pet reported relatively serious interaction problems. Among the mines experiencing interaction problems, simultaneous mining caused the most trouble and accounted for 48 pet ^Factors marked with an asterisk are regarded as being critical. 300 280 - 260 - 240 C220 $~200 UJ I 180 E 160 uj 140 Q cr CO uj !00 20 80 60 40 20 KEY + Unstable • Stable _L 10 20 30 40 50 60 70 80 90 100 110 INNERBURDEN SANDSTONE, pet FIGURE 1.— Influence of percentage of hard rock in the in- nerburden on stability in the lower seam. Adapted from Haycocks (3). of the problems; overmining and under- mining accounted for 30 and 22 pet, respectively. Initial analysis of field study data by plotting the data with respect to some related variables also demonstrated some simple but useful relationships among variables and some interesting trends (figs. 1-3). 320 _ 280- <+- $240 UJ g 200 x 2 160 ui cc 120 m uj 80 z z - 40 KEY + Unstable • Stable 1 J_ l 1 l 1 I 1 4 6 8 10 12 INNERBEDS 14 18 FIGURE 2.— Number of innerbeds versus innerburden thickness for stable and unstable lower seam conditions. Adapted from Haycocks (3). o a. 3 cc I- X UJ < UJ CO cc UJ Q. Q- z> 90 1 1 1 KEY - + Unstable 1 • Stable + + /. / 8U ' + + ♦ / ♦/ • 70 /. • - 60 - /.. - RD 1 1 u • 1 60 70 80 90 100 LOWER SEAM EXTRACTION, pet FIGURE 3.— Relationship between percent extraction in the upper and lower seams and lower seam stability. Adapted from Ehgartner (7). 46 Figure 1 establishes the innerburden limits for interaction problems due to pillar loading onto an underlying seam in terms of the percentage sandstone in the innerburden. Subsequent photoelas- tic analysis of this pillar load transfer phenomenon indicates that the degree of layering in the innerburden is as important as the modulus, as shown in figure 2 (_7)» It was found that the rel- ative extraction ratios in two adjacent seams are extremely important in deter- mining whether or not interaction will be experienced. Higher upper seam extrac- tion will increase the chance of unstable conditions in an adjacent seam (fig. 3). STATISTICAL ANALYSIS Since the field data collected are nu- merous and very complex, multivariate statistical procedures were used to gain further insight into the field data. Statistical analyses were also used to derive relationships between geological factors, which are generally difficult to incorporate into other analytic models (_3_, 8-10). For the proposed analysis, variable selection was primarily deter- mined by former research experience and the availability of data. The appendix shows a total of 24 variables selected and used. Certain qualitative variables are used for classification purposes, and some of them were found very useful in separating the local problems due to in situ geological disturbances from in- teraction problems caused by mining exca- vations in adjacent seams. Five multivariate analysis techniques were used in the analyses (11-12). Ex- cept for regression analysis, these pro- cedures are similar in that they all try to transfer the original data set into a So Q- - I -2- -3.0-2.5-2.0 -1.5 -1.0-0.5 0.0 0.5 1.0 1.5 2.0 PRINCIPAL 3 FIGURE 4.— Results from principal component analysis (overmlnlng). 1 1 1 1 i 1 1 1 1 s - y - KEY 'o • Interaction • / o No interaction • / - — /o y - - • • y o — • / o o — > - / o y y < o o - y - y i i ii i 1 1 °l 1 new data set. By data transformation, some specific goals can be achieved which may lead to better understanding of the original problem (13-14). RESULTS FROM PRINCIPAL COMPONENT ANALYSIS AND FACTOR ANALYSIS A principal component analysis was per- formed on eight quantitative variables. By Kaiser's rule (eigenvalue > 1), or proportional criterion, three principal components were extracted that explain 71 pet of the total variance of the original data. Here, the cases with interaction can be better separated from those with- out interaction (fig. 4). Also, when the first component is large or the third component is small, the chance of inter- action problems increases. The first principal component has relatively higher loadings on variables such as COVER, UST, USEP, and IBT while the third component has higher loadings on the variables IBSP and LSEP. (See appendix 1 for definition of variables.) This indicates that the interaction problem may be attributed to cover depth, upper seam thickness, ex- traction on both seams, and nature of innerburden. It is noted that innerbur- den (IBT) is negatively related to the first component. To decrease the score of the first component, and thus reduce the possibility of having interaction problems, the spacing between two seams must increase. The results from factor analysis showed a similar grouping of variables. RESULTS FROM DISCRIMINANT ANALYSIS AND CLUSTER ANALYSIS Discriminant analysis was applied to classify the multiple-seam mines into 47 two groups, one with interaction problems and the other without, based on observed geological and mining conditions. Al- though no general way of classifying the observations for all mining methods was found, the classification of mines using overmining methods was successful. Fig- ure 5 shows the scatter plot with two canonical scores of each observation as its coordinates. It can be seen that the group with interaction problems is well separated from the group without inter- action effects. This implies that for overmining mines it is possible to pre- dict interaction using the discriminant analysis. However, severity of inter- action cannot be estimated by such a classification scheme. When cluster analysis, like discrimi- nate analysis, was used in studying all case study data, no natural groups or clusters could be identified. However, when this procedure was independently used for data from mines using overmining or undermining methods, it was possible to classify the observations according to location of interaction problems. The location where the interaction oc- curred was originally classified into five categories: below or above remnant pillar or pillars (A or B), below or above the Interface between gob area and large section of solid coal (C or D), or not as above (E). The resulted cluster- ing indicates that most interaction prob- lems can be grouped into four types of locations (A-D). These results suggest 4 1 •1 • 1 1 A 1 1 /o 2 • • •• / KEY _ • Interaction _l "~- — . o No interaction < — o o z ~ o z -2 o o o < o o o - -4 o - _c 1 1 1 1 1 1 1 -2.0 -1.5 -1.0-0.5 0.0 0.5 1.0 1.5 2.0 2.5 CANONICAL 2 FIGURE 5.— Results from canonical discriminant analysis (overmining). that the initial classification is cor- rect and, in addition, the interaction problems occurring at the same location may have similar geological and mining conditions. In summary, of all reported severe problems, 59 pet occurred either below or above a remnant pillar or pillars (A and B type locations) and 22 pet occurred be- low or above the interface between a gob area and large section of solid coal (C and D type locations). This demonstrates the possibility of interaction due to stress anomalies caused by isolated pil- lars left on the upper or lower seams. RESULTS FROM REGRESSION ANALYSIS When data were sorted into different mining methods, and innerburden thickness between adjacent seams was limited to less than or equal to 300 ft, good linear models were obtained for close-seam mines using stepwise regression procedures. For example, for mines with close seams and using overmining methods, the follow- ing regression equation can be derived: INTERP = -2.13 + 0.32USEP - 0.38LSEP + 0.121BSP + 0.58LST. This model can explain 78 pet of total variance (R-square = 0.7799) and has a Cp value of 2.76, which is the best compared to other possible models. The error mean square of this model is also very small. In a physical sense, the degree of damage due to interaction in overmining can be related to, according to sequence of im- portance, upper and lower seam extraction percentages, Innerburden sandstone or strong rock percentage, and lower seam thickness. Through similar procedures, a model was developed for mines where the overlying seam has been previously mined. The re- gression model is given as follows: INTERP = 7.68 - 0.021BT - 0.06LSEP + 0.27LST. This is also a statistically signifi- cant model (F-value = 8.01) and has less 48 variance inflation (p = 1.12). These models demonstrate that interaction- controlling factors for mines using over- raining methods are different from those using undermining methods. In this latter model, the important variables are innerburden thickness, lower seam extrac- tion percentage, and lower seam thickness (15). MODEL ANALYSIS To extend the information gained from case studies, analysis of various inter- action phenomena has been carried out us- ing body-loaded photoelastic and finite element models. These two modeling meth- ods are concentrated in stress analysis in the innerburden for undermining conditions. PHOTOELASTIC STRESS ANALYSIS Photoelastic models were used to eval- uate stress phenomena in close proximity to the excavation since they could best incorporate the effects of layering and bed separation, which is extremely diffi- cult to accomplish using numerical meth- ods. Peng (2^) predicted that the load approaches the normal background value at a distance of approximately four times the pillar width below the floor. Eh- gartner (7_) found that the distance stress was transferred depended on the nature of the rock below the pillar. In particular, he found that low-modulus stratified materials tended to increase the distance through which stress is transferred, while stiff isotropic mate- rials had the opposite effect. Results of this model showed that the zero-influ- ence bulb could extend as deep as eight times the pillar width in a highly strat- ified material (fig. 6). FINITE ELEMENT ANALYSIS Finite element analysis was used to evaluate the influence of thickness and location of sandstone layers in the in- nerburden and overburden on the size and shape of the pressure arch around the lower opening. This information was, in turn, used to locate the area in which the upper seam arch intersected the lower workings. Hudock (16), modeling in Iso- tropic overburden, found that the modulus of the material had little influence on arch dimensions or abutment pressures. A high-modulus layer above an excavation, however, can affect subsidence throughout the formation (2^, 17). The effects of opening geometry and material properties on the stability of interactive excava- tions in stratified rock were also eval- uated using finite element methods. The influence of pillar location, Young's modulus, and Poisson's ratio were deter- mined. This system was modeled with four-node isoparametric, quadrilateral, linear elastic, plane strain elements. The standard finite element routine was modified to include the calculation of the factor of safety at the center of each element based on the Mohr-Coulomb failure criterion as outlined by Wang (18) and Haycocks (3). The results from this work showed that increasing Pois- son's ratio resulted in improved stabil- ity in the roof of both upper and lower openings for all innerburden spacings. The shear stress at the pillar edge in the roof of the lower opening and the floor of the upper opening is also 0.8 cr o I- o & hi O Z Ul _l in } R. J. Matetic, and J. A. Galek. A Case Study of Ground Con- trol Problems Related to Multiple Seam Mining in the Pittsburgh and Sewickley 55 Coalbeds. Soc. Min. Eng. AIME Preprint 85-325, 1985, 9 pp. 21. Grenoble, A., C. Haycocks, and W. Wu. Computerized Evaluation of Near Seam Interaction. Paper in Proceedings of 2nd Annual Conference on the Applica- tion of Computers to the Coal Industry. Soc. Min. Eng., Littleton, CO, 1985, pp. 317-322. 22. Grenoble, A., and C. Haycocks. Design Factors in Near Seam Interaction. Paper in Proceedings of 4th Conference on Ground Control in Mining. WV Univ. , Mor- gantown, WV, 1985, pp. 166-177. 56 1. COVER 2. UST 3. LST 4. USEP 5. LSEP 6. IBT 7. IBSP 8. IBNL 9. TIME 10. INTERP APPENDIX. —VARIABLES SELECTED FOR DATA COLLECTION Quantitative Variables Cover depth, ft. Upper seam thickness, ft. Lower seam thickness, ft. Upper seam extraction percentage. Lower seam extraction percentage. Innerburden thickness, ft. Innerburden sandstone percentage. Number of layers in the innerburden. Delay between operations in two seams, yr. Degree of damage due to interaction: INTERP = - no damge. INTERP = 5 - serious damage. Qualitative Variables Location of mine. Names of two seams. Cover rock type. Immediate roof for upper seam. Immediate roof for lower seam. Immediate floor for upper seam. Immediate floor for lower seam. Locations and damages in upper seam. Locations and damages in lower seam. Innerburden rock type. Existence of water problems. Mining sequence. Surface subsidence problem. Location of interaction problems. 11. LOCATION 12. BEDNAME 13. CRT 14. USIR 15. LSIR 16. USIF 17. LSIF 18. USLP 19. LSLP 20. IBRT 21. WP 22. MM 23. SSUBP 24. LITERP 57 THE BUREAU OF MINES SUBSIDENCE RESEARCH PROGRAM By Michael A. Trevits, 1 Roger L. King, 2 and Bradley V. Johnson 3 ABSTRACT The Bureau of Mines, through its Subsi- dence Research Program, is focusing on providing the mine operator with the ability to predict surface movements and effects on ground water as a function of mining method and geologic context. The program is designed for coal basins where high mining activity may impact land use requirements. In the long term, all coal basins and mining methods will be addressed. Data sets from several sub- sidence monitoring sites have been or are being collected. Data sets are now available from the Eastern, Interior, and Rocky Mountain Coal Provinces for full- extraction mining methods (longwall and/ or room-and-pillar retreat mining). At select sites, shallow-aquifer monitoring wells have also been installed to observe the effects of subsidence on the ground water system. To date, an empirical mod- el for subsidence prediction has been generated for the Northern Appalachian Coal Region. This paper discusses the status of the Bureau of Mines Subsidence Research Program. INTRODUCTION The involvement of the Bureau of Mines, U.S. Department of the Interior, in mining-related subsidence probably dates back to the Bureau's inception in 1910, when a mining engineer was assigned to a study of mine filling in the northern Anthracite Field of Pennsylvania (_0. 4 The purpose of this work was to reduce the amount of mine waste resulting from imperfect mining methods, create a safe work environment, and reduce the settle- ment of the overlying ground surface. Results of the investigation were subse- quently reported by Griffith and Conner (2). Perhaps the longest sustained subsi- dence investigation ever made in the United States involved a cooperative ef- fort between the Bureau, the Illinois State Geological Survey, and the Univer- sity of Illinois during 1916-24. Reports covering the results of these studies 'Geologist, Pittsburgh Research Center, Bureau of Mines, Pittsburgh, PA. 2 Research supervisor, Pittsburgh Re- search Center. 3 Staff engineer, Bureau of Mines, Wash- ington, DC. ^Underlined numbers in parentheses re- fer to items in the list of references at the end of this paper. became classics of mining subsidence literature. Between 1926 and 1932, subsidence stud- ies were pursued under various sections of the Bureau's Mining Division: Ground Movement Investigations Section, Subsi- dence Section, and Subsidence and Ground Movement Section. During this period, subsidence was studied in connection with iron mining in Alabama and Oklahoma, and iron and copper mining in Michigan. Although subsidence research did not constitute a distinct organizational ele- ment in subsequent years, concepts and data fundamental to an understanding of the mechanics of subsidence were devel- oped in the course of many other Bureau investigations. Included among these were mine roof control, pillar extrac- tion, longwall and caving methods of min- ing, physical properties of mine rocks, in situ stress measurements, and physi- cal and mathematical modeling of mine structures. The Bureau's subsidence research was focused in the Minerals Environmental Technology Program in 1978. In 1982 cer- tain activities of this program were melded with work in mining technology to form the current Mining Technology Pro- gram under which subsidence research is presently conducted. 58 THE PROCESS OF SUBSIDENCE During the subsidence process, the ground surface is changed as a result of the excavation of underlying coal. The final surface profile is determined by a multitude of components: thickness of the coal extracted, width of the opening, depth of coal, lithology of the strata between the coal and the surface, mine design, mining method, and other factors. An analysis of the typical subsidence profile reveals that almost every point of the surface will be displaced both vertically and horizontally, causing zones of tensile and compressive strain along the surface. These can result in damage to surface structures. One must also consider the effect on ground water movement in the vicinity of the affected area. Historically, mining operations have been accused of causing water wells to become dry and/or pol- luted. However, such local effects may be minor compared with the potential problems that can occur in the Western and Midwestern United States. Western underground aquifers are a major source of water for populated areas; in the Mid- west, surface drainage patterns are im- portant to agriculture. It is, there- fore, easily understood how changes in the ground water system could have a major impact. SUBSIDENCE MODELS Three types of models have gained ac- ceptance by subsidence researchers — phys- ical, analytical, and empirical. Physi- cal models have been used in an effort to develop basic understandings of the sub- sidence process. Examples of physical models include centrifuges, sand models, and scale models of the overburden. Var- ious researchers have claimed different findings, but the drawbacks to such mod- els (accuracy and expense) have precluded widespread acceptance of this approach. However, basic insight into the subsi- dence process may be obtained from physi- cal models. With the advent of computers , analyti- cal models have become more practical to use in subsidence research. This tool, in conjunction with sophisticated soft- ware (finite-element programs), permitted researchers to analyze many more cases in a given time than they could using physical modeling. The major drawback to analytical models is that the researcher must mathematically describe a physical process for which very little is known. Another major disadvantage is that the analytical model requires detailed geotechnical information about the over- burden that may not be readily available to accurately predict the subsidence pro- file. Such unknowns result in predic- tions that do not give the required de- gree of accuracy. The empirical model has gained the most worldwide acceptance as the approach for subsidence prediction. Empirical models are based upon the fact that given enough field data (final subsidence profiles), a mathematical function can be written to describe a characteristic profile. This approach, of course, assumes that within a given area the important functional parameters (angle of draw, subsidence factor, etc.) are reasonably constant. If this does not happen to be the case, adjustments to the empirical model must be made to make it more or less univer- sal. The major drawback to empirical models is that they require a substantial amount of field data before an accurate mathematical function can be derived. However, for the mine operator, this ap- proach should be the easiest to utilize in predicting mine subsidence. STATUS OF BUREAU OF MINES RESEARCH The short-term strategy is to focus the research program into the coal basins where high mining activity is coupled with present land use requirements. These areas are delineated via consul- tations with industry and the Office of 59 Surface Mining. In the long term, all coal basins and mining methods will be addressed. Data sets from several subsidence-moni- toring sites have been or are being col- lected. The purpose of these data sets will be to develop and/or validate the predictive models being proposed by Bu- reau researchers. Initial data sets are now available from the Eastern, Interior, and Rocky Mountain Coal Provinces for full-extraction methods (longwall and/or room-and-pillar with retreat mining). ALTERNATIVE SURVEY SYSTEMS Since the Bureau's Subsidence Program is concentrating on the development of empirical model(s), the collection of data sets from a variety of locations is required. A study at a mine site normal- ly requires the installation of hundreds of survey monuments, followed by numerous conventional ground surveys over a period of time. If these tasks are being car- ried out at a number of mine locations, the data collection scheme can become prohibitively expensive. To compare al- ternate surveying capabilities, a test grid was constructed at the Bureau's Pittsburgh Research Center, Pittsburgh, PA. The grid of monumemts was estab- lished in an area that was not being un- dermined to eliminate the variable of surface movement (_3). Conventional and high-technology sur- veying systems including an electronic distance meter-theodolite-level (EDM- theodolite), an automatic recording in- frared laser tacheometer, a global posi- tioning system satellite surveyor (GPS), aerial photogrammetry , and prototype in- ertial surveyor were deployed over the grid during a 1-month period. Highlights of the systems used follow: The inertial system is extremely porta- ble and can be placed in surface vehicles and aircraft. Sources of survey error include accelerometer measurement caused by thermal effects, platform drift rate caused by vibration variations and ther- mal transients, and environmental effects such as variations in the earth's gravi- tational field and temperature varia- tions. Survey data for the Bureau study could not be used owing to a computer malfunction. The GPS system, when completed, will be able to determine receiver position in- stantaneously. Constraints are the high amperage required to power the field system, sensitivity to temperature variations, and site selection. This system was used to accurately determine the position of the control monuments, but the expense and time to survey the grid of monuments at the Bureau's study site precluded its use over the entire grid. The EDM-theodolite system is portable, rugged, and relatively simple to operate. The system is sensitive to windy and/or rainy conditions and is subject to errors due to hand tabulations and computations. This system was used as the base for the Bureau's study because it is the most commonly used method for making subsi- dence measurements in the field. The tacheometer system is completely automated for rapid recording, comput- ing, and data generation. The system is subject to the same environmental con- straints as the theodolite system. The aerial photogrammetry system can be used for inaccessible areas. All data are gathered simultaneously, providing a permanent record that can be reevaluated. Surveys cannot be conducted during incle- ment weather, and data for control points must be visible on the photographs. The results of the Bureau's study showed that the three-dimensional dis- placements from the base (EDM-theodolite) were almost identical for the tacheometer (0.25 ft) and photogrammetry (0.24 ft). GROUND SURFACE AND STRUCTURE RESPONSE In 1985, the Bureau of Mines initiated a cooperative mine subsidence research program with the State of Illinois (4-_6)« The purpose of this effort is to jointly fund research to develop guidelines for underground mining methods that could maximize coal recovery while preserving farmland productivity. 60 In 1985, the research effort evaluated and selected sites at operating mines that best fit the program's goals. Ini- tial surface and subsurface studies in- clude characterizing the response of overburden to subsidence, assessing the impact of subsidence on ground water, monitoring and characterizing subsidence profiles, collecting available data to build data bases on subsidence and rock mechanics, testing floor materials in mines, and evaluating crop production in subsided areas. Because much of the farmland in Illi- nois is flat with water table depths of 2 to 5 ft, differential surface ground movements could alter subsurface hydro- logical patterns and affect surface drainage. The Bureau of Mines is cur- rently studying the effects of subsidence EFFECTS ON To properly characterize the effects of mining-induced subsidence on the local hydrological system, a coordinated data collection scheme must be constructed. All monitoring points (weirs, wells, etc. ) should be installed in advance of mine development in the area of study. It should be noted that these data will represent base conditions, and enough preliminary data should be collected to allow for seasonal fluctuations and well stabilization if applicable. One such study was recently completed by the Bureau of Mines at a mine site in western Pennsylvania (_7_). Th e purpose of this work was to observe the effects of longwall mining in five water wells which typified local domestic water systems (fig. 1). Two very small streams and a spring located in the vicinity were also monitored to establish flow rate charac- teristics. The overburden above the coal unit being mined (Pittsburgh Coalbed) ranged from 750 to 1,000 ft thick. Results of the study showed that there was no pronounced change in the overall quality between premining and 1 yr after on surface drainage patterns to determine critical slope changes that may cause drainage problems detrimental to crop production. Such differential movements also can cause damage to structures and utilities, not only in Illinois but po- tentially in any area affected by under- ground mining. By the end of 1984, more than 1,500 claims for structural damage (of which 25 pet were attributed to mine subsidence) had been filed with the Illinois Mine Subsidence Insurance Fund. To better un- derstand how surface ground movements from high-extraction coal mining affect a residential foundation, the Bureau of Mines built two foundations which are currently being monitored for vertical, horizontal, and differential movements. HYDROLOGY the longwall face passed the water well profile (fig. 2). Water levels in the well located near the centerline of the panel began to decline when the longwall face had approached within 500 ft of the well. The water level continued to fall, and the well went dry about 2 months af- ter the face had passed beneath it, at which time the face was approximately 500 ft beyond the well location. The water levels in the two wells located 100 and 300 ft outside the rib line of the panel declined some 15 to 30 ft as a result of mining but recovered to near premining levels about 10 months after the longwall face passed by. The well located more than 500 ft outside the rib line of the panel showed no detectable change in flu- id level as a result of mining. No evi- dence of mining effects on the small streams or springs located within 1,200 ft of the panel could be detected. It should be noted that the results of this study apply to the local conditions of this mine site, which may differ from conditions at other sites. SUBSIDENCE MODEL The specific lithology over coalbeds in the Northern Appalachian Coal Region, created by highly resistive limestone and sandstone units with relatively shallow overburden, has precluded the use of predictive methods developed for European mining conditions. Bureau of Mines work to date has resulted in the 61 1,000 j I LEGEND (^ ) Surface contour Water well I Outline of longwall panel — _ J Outline of proposed panel ▼ V- notch weir /• Spring FIGURE 1.— Surface features and longwall panel. development of a model that is suitable to the geological and mining conditions in the northern Appalachian area (8). Because of these conditions, the subsi- dence coefficient varies within the area of the subsidence, trough. The effects of lithology in the form of a variable subsidence coefficient were separated for each site studied by introducing a corre- lation between hypothetically homogeneous overburden and existing lithological con- ditions, while providing for different mining situations, e.g., underground geometry and thickness of the overburden. To develop the model, data from 11 longwall test sites were used in a re- gression analysis. The width of the in- dividual panels studied ranged from 400 to 900 ft, and the overburden thickness was 345 to 910 ft. For each panel, the characteristics of the variability of the subsidence coefficient along individual profilelines were defined. The regres- sion analysis of the subsidence coeffi- cient from all of the test sites relative to the edge of the panel yielded a third- degree polynomial equation with a corre- lation coefficient of 0.99. 62 DISTANCE OF FACE FROM WATER WELL SECTION, ft 2,000 - 3,000 15° angle to limit of major surface deformation- / L Longwall paneh / 1/ l// 7/ I I I I I I // / 25° angle to limit of detectable surface deformation / Surface I i i_ 500 Scale, ft V FIGURE 2.— Relation of water levels to longwall face advance. 63 Although very good results for the mod- el have been obtained through sensitivity tests, the model must be considered pre- liminary. The model has been constructed using a limited amount of field data for relatively similar mining conditions. Its applicability in other mining areas must be tested. To facilitate the use of the predictive model, a program was written in BASIC language for use on personal computers. The program requires approximately 48K of memory and has been constructed so that hard copies of the subsidence predictions are possible. The inherent beauty of the predictive model is that it can be oper- ated by individuals with little or no previous knowledge of the theory of sub- sidence, it is simple and fast in compar- ison with existing predictive methods, and it eliminates the use of inaccurately estimated functional parameters, e.g., maximum subsidence, location of the in- flection point, etc. necessary with ex- isting predictive methods (fig. 3). FIGURE 3.— Comparison of field data, subsidence model prediction and data obtained using various European methods. SUMMARY The Bureau's Subsidence Research Pro- gram has been constructed to address nearly every aspect of the phenomenon of subsidence and the results of ground movement. The program is currently fo- cused on high-coal extraction systems; however, in the long term all coal basins and mining methods will be investigated. Work completed to date includes prelimi- nary observations of the effects of long- wall mining on shallow water sources, delineation of the impact of mining on primary and secondary land use, and de- velopment of a prediction model that is fast and simple to use. Future research is scheduled to quantify the effects of mining on the hydrological system in a variety of geologic environments, surface areas (including structures) will be in- strumented to observe movement as mining progresses beneath, and the prediction model will be modified and/or updated for coal basins outside the Northern Appala- chian Coal Region. REFERENCES 1. Quan, C. K. Overview of the Bureau of Mines Subsidence Research Program. Pres. at Soc. Min. Eng. AIME 1979 Annual Meeting (New Orleans, LA, Feb. 18-22, 1979). Soc. Min. Eng. AIME preprint 79- 84, 1979, 9 pp. 2. Griffith, W. , and E. T. Conner. Mining Conditions Under the City of Scranton, Pa., Reports and Maps. BuMines B 25, 1912, 89 pp. 3. Krantz , G. W. , and J. C. LaScola. Longwall Mine Subsidence Surveying — An Engineering Technology Comparison. Paper in Mine Subsidence Control. Proceedings: Bureau of Mines Technology Transfer Sem- inar, Pittsburgh, PA, September 19, 1985, comp. by Staff, Bureau of Mines. BuMines IC 9042, 1985, pp. 2-12. 4. Powell, L. R. Subsidence Studies in the Illinois Coal Basin. BuMines Re- search 85, 1986, pp. 36-37. 5. DuMontelle, P. B. IlHnois Mine Subsidence Research Program. IL Geol. Surv. , 1985, 2 pp. 64 6. Powell, L. R. , and P. B. DuMon- telle. The Illinois-Bureau of Mines Co- operative Mine Subsidence Research Pro- gram. Paper in Proceedings of the Second Conference on Ground Control Problems in the Illinois Coal Basin. Southern IL Univ., Carbondale, IL, 1985, pp. 13-17. 7. Moebs, N. M. , and T. M. Barton. Short-Term Effects of Longwall Mining on Shallow Water Sources. Paper in Mine Subsidence Control. Proceedings: Bureau of Mines Technology Transfer Seminar, Pittsburgh, PA, September 19, 1985, corap. by Staff, Bureau of Mines. BuMines IC 9042, 1985, pp. 13-24. 8. Adamek, V. , and P. W. Jeran. Pre- diction of Subsidence Over Longwall Pan- els in the Northern Appalachian Coal Re- gion. Pres. at Soc. Min. Eng. AIME 1985 Fall Meeting (Albuquerque, NM, Oct. 16- 18, 1985). Soc. Min. Eng. preprint 85- 404, 1985, 16 pp. 65 SUBSIDENCE OVER CHAIN PILLARS By P. W. Jeran 1 and V. Adamek 2 ABSTRACT Subsidence over two or more adjacent longwall panels and the intervening chain pillars was monitored by the Bureau of Mines at four mines in the Northen Appa- lachian Coal Basin. The magnitude of the subsidence over the chain pillars ranged from 0.06 to 1 ft. The width of the chain pillars affects the shape of the subsidence curve. Wider chain pillars yield a wider area of minimum subsi- dence. Comparison of the field-measured subsidence with precalculated subsidence over the chain pillars indicates a range of pillar deformation. The data show that at three of the sites additional subsidence was induced over the first panel by the mining of the second panel. Curves of the additional subsidence are similarly shaped for these sites. This indicates that with sufficient data a model to predict subsidence over chain pillars could be developed. INTRODUCTION Early monitoring of subsidence over room-and-pillar workings showed that some exhibited fairly regular subsidence, while others were highly irregular. The irregular subsidences were thought to re- sult from the incomplete extraction of the pillars where anything from stumps to entire pillars were left depending on lo- cal mining conditions. When the Bureau of Mines began a proj- ect to develop a method of subsidence prediction, it was reasoned that data ob- tained from monitoring over longwall op- erations would be the least complicated and therefore the easiest to interpret the process of subsidence in this coun- try. Longwall mining removes all the coal and any observed surface subsidence irregularities are therefore related to the response of the overburden to that removal. With room-and-pillar mining, there would always be the question of what portion of the subsidence was due to overburden movement and what was caused by incomplete mining. The monitoring of selected longwall panels has shown that subsidence above the centerline is not uniform (1). The 1 Geologist. 2 Mining engineer. Pittsburgh Research Center, Mines, Pittsburgh, PA. Bureau of lack of uniformity may be due to local stratigraphic inhomogeneities or varia- tion in the thickness of coal extracted. Sufficient field data are not available, at this time, to resolve this question. A predictive method has been developed (2) using data from half profiles extend- ing from the centerline of a panel out over the solid coal. The method has been proven valid for the Northern Appalachian Coal Basin (3_). Using the principle of superposition, the subsidence over the chain pillars separating two longwall panels can be predicted by adding the subsidence caused by the mining of each of the two panels illustrated in figure 1. This assumes that the chain pillars resist deformation and behave in the same way they would if the second panel had not been mined. If these assumptions are not met, then the actual subsidence should exceed the prediction. Chain pillars are used to maintain the integrity of the entries surrounding longwall panels, and there is wide vari- ation in the dimensions of chain pillars used in this country. Some are designed to yield, while others are sized to -^Underlined numbers in parentheses re- fer to items in the list of references at the end of this paper. 66 resist movement. To provide additional support at the working place, some mine operators backfill the entries between the chain pillars. Others install crib- bing or posts, and in some instances they may do nothing. Each of these support materials will behave differently when it becomes part of the gob; hence, effects on subsidence will also vary. This paper examines the surface subsidence over the chain pillars between longwall panels at four sites. The data were obtained using standard surveying methods. DISCUSSION The Bureau of Mines has monitored sub- sidence over two or more adjacent panels wherever possible. Since this is a mul- tiyear task, great care must be taken to assure long-term stability of survey con- trol points. Any undetected movement of the control points can yield erroneous data on ground movement. Data from four study sites were chosen based upon data reliability as well as difference in mining geometry and measured subsidence. While these data sets are insufficient to construct a model of the subsidence oc- curring over chain pillars, their study can provide some insight to the process. The four study sites are located in the Northern Appalachian Coal Basin, south- western Pennsylvania and north-central West Virginia. At each site subsidence was monitored over at least two longwall panels and the Intervening chain pillars. The main criterion for site selection was that monitoring begin sufficiently dis- tant from mining to preclude any prior subsidence movements. All surveying was conducted using EDM and precision level- ing equipment. Survey control points were established and maintained through- out the monitoring. Initialization of survey points occurred prior to any pos- sible movement. Profiles were located Predicted subsidence from panel 2 o | - 2 CO CD z> co -3 Predicted subsidence from panel I Subsidence from pane plus panel 2 — Panel I Panel 2 100 200 300 400 500 600 700 800 LOCATION, ft FIGURE 1.— Subsidence over chain pillars based on super- position. far enough from the beginning and end of each panel to eliminate any reduction in subsidence caused by these areas. The generalized stratigraphy over each study site as illustrated by columnar sections is shown in figure 2. The sec- tions are based on coreholes drilled at or very near each study site. The strata are typical of the Northern Appalachian Coal Basin, being composed of alternat- ing layers of varying thicknesses of resistant rocks (limestones and sand- stones) separated by shale. 900 r 800 700- 600 «= 500 L±J _l <-> 400 co 300 200 I00- L IT I If LEGEND EUSI Sandstone Limestone Shale Coal A B C D MINE FIGURE 2.— Generalized columnar sections. 67 Table 1 contains a summary of the min- ing geometries. The panel widths range from 625 to 1,000 ft. The coal extracted averaged 5.5 to 6 ft thick. Chain pillar sizes and numbers as well as overbur- den thicknesses varied among the sites. Lithologic composition of the overbur- den was similar in that the bulk of the resistant strata at each site is within the first 300 ft above the mine. Mine A is located in northern West Vir- ginia on the Pennsylvania border. Three panels were monitored at this site. Each was 5,300 ft long by 625 ft wide. The panels were developed using a four-entry system driven on 100-ft centers except for the fourth entry, which was driven on 60-ft centers to accommodate a yield pillar on the gob side of the panel. The average extracted height was 6 ft. The overburden ranged from 700 to 950 ft thick over the first two panels. It ex- ceeded 1,100 ft over the third panel. The lithologic composition of the over- burden is shown in figure 2. The first 250 ft above the mined coalbed contains most of the resistant strata in the column. Mine B is located in southwestern Penn- sylvania. The panels were 630 ft wide. Panel 1 was 4,700 ft long, and panel 2 was 5,650 ft long. The panels were de- veloped using four entries driven on 90-ft centers. The total width of the chain pillars and entries was 280 ft. The overburden ranged from 785 to 890 ft thick. The columnar section (fig. 2) shows resistant strata between 30 and 190 ft above the coalbed that was mined and a 45-ft-thick sandstone 135 ft above that. An average 6 ft of coal was extracted from these panels. Mine C is located in north-central West Virginia. Panel 1 was 1,000 ft wide, and panel 2 was 950 ft wide. They were sep- arated by a pair of chain pillars 62 ft wide. Total width of the chain pillars and entries was 160 ft. The overburden ranged from 665 to 690 ft thick. Most of the resistant strata are between 60 and 235 ft above the mined coalbed (fig. 2). The extraction thickness averaged 5.5 ft. Mine D is located in the panhandle of West Virginia between Ohio and Pennsyl- vania. The panels were 605 ft wide by 3,000 ft long. They were developed using a four-entry system. Two of the three chain pillars were 85 ft wide, and one is 65 ft wide. The total width of the chain pillars and entries was 280 ft. The overburden ranged between 510 and 660 ft thick; its composition is shown in figure 2. The first 250 ft above the coal contain most of the resistant strata. The extraction thickness aver- aged 5.5 ft. The Bureau subsidence prediction mod- el was used to determine the amount of subsidence caused by the mining of each panel. Applying the principle of super- position, the subsidence resulting from the mining of each panel was added to- gether to obtain the subsidence over the the chain pillars between the adjacent panels at each mine. Comparing these data to the field measurements (figs. 3- 6) showed only mine D in close agreement with the model. The other mines show that subsidence greater than predicted by the model had occurred. Given that the model predictions are reasonable, it must be concluded that at three sites the overburden reacted as if the chain pil- lars had deformed. TABLE 1. - Mining geometries Overburden thickness, ft Panel width, ft Coal height, ft Chain pillar Mine Centerline to centerline Total width, ft 680-950 745-910 660-710 510-660 625 630 1,000 605 6.0 6.0 5.5 5.5 f 2 at 100 ft 1 1 at 60 ft 3 at 90 ft 2 at 80 ft f 2 at 100 ft 1 1 at 80 ft | 260 280 160 | 280 68 CD in ' 3 - -i r~ oooooooo, o • KEY o Prediction • Field o° Panel I Chain pillars • Panel 2 ® 100 200 300 400 500 600 700 800 900 LOCATION, ft FIGURE 3.— Predicted and measured subsidence over chain pillars at mine A. LU O UJ -1 -2 0000000o o°° o o o .« • °o i i o o o •• ° • • ° Q CO m °» KEY o • o Prediction • o -3 o • • Field • •o • - -3 ! i i i •• • • • - • • • " Panel 1 i Chain pillarsy • Panel 2 i i 200 400 600 LOCATION, ft 800 1,000 FIGURE 8.— Difference in subsidence between mining of panels 1 and 2 at mine A. 400 600 LOCATION, ft 800 1,000 FIGURE 9.— Difference in subsidence between mining of panels 1 and 2 at mine B. uj -2 Q CO m 3 -3 1 Panel 1 Chain i pillars-? • • • • • Panel i • • • • # 2 ^ 200 400 600 800 LOCATION, ft 1,000 1,200 FIGURE 10.— Difference in subsidence between mining of panels 1 and 2 at mine C. 71 SUMMARY AND CONCLUSION Chain pillar left in place tween adjacent retreating of come part of subsidence is profile across rate troughs, over chain pil Northern Appal face movements ing between 0. troughs subsid The data indi width of the minimize the between that of the panels s are the blocks of coal to protect the entries be- longwall panels. With the the longwall face, they be- the gob. Their effect on to break up the subsidence several panels into sepa- Measurements of subsidence lars at four mines in the achian Coal Basin show sur- over chain pillars rang- 06 and 1 ft. The adjacent ed between 3.5 and 3.9 ft. cate that minimizing the chain pillars would also difference in subsidence occurring over the center and that over the chain pillars. The subsidence prediction model developed by the Bureau predicts subsi- dence over chain pillars only when the chain pillars behave as if the second panel had not been mined. The similarity of the shapes of the curves of difference in subsidence between mining of first and second panels indicates that with suf- ficient data the Bureau model could be expanded to include the prediction of subsidence over chain pillars. The long- term stability of chain pillars is un- known, and the materials used to support the intervening entries are subject to eventual deterioration. Therefore the potential exists for further subsidence using the present configurations of chain pillars. REFERENCES 1. Jeran, P. W. , and T. M. Barton. Comparison of the Subsidence Over Two Different Longwall Panels. Paper in Mine Subsidence Control. Proceedings of Tech- nology Transfer Seminar, Pittsburgh, PA, September 19, 1985, comp. by Staff, Bu- reau of Mines. BuMines IC 9042, 1985, pp. 25-33. 2. Adamek, V., and P. W. Jeran. Pre- calculation of Subsidence Over Longwall Panels in the Northern Appalachian Coal Region. Paper in Mine Subsidence Con- trol. Proceedings of Technology Transfer Seminar, Pittsburgh, PA, September 19, 1985, comp. by Staff, Bureau of Mines. BuMines IC 9042, 1985, pp. 34-56. 3. Jeran, P. W. , V. Adamek, and M. A. Trevits. A Subsidence Prediction Model for Longwall Mine Design. Paper in Proceedings of Longwall USA Conference (Pittsburgh, PA, June 17-19, 1986). In- dustrial Presentations West Inc. , 1986, pp. 101-112. 4. Adamek, V., and P. W. Jeran. Eval- uation of Existing Predictive Methods for Mine Subsidence in the U.S. Paper in Proceedings of First Conference on Ground Control in Mining. WV Univ. , Morgantown, WV, 1981, pp. 209-219. 72 STUDY OF DEWATERING EFFECTS AT AN UNDERGROUND LONGWALL MINE SITE IN THE PITTSBURGH SEAM OF THE NORTHERN APPALACHIAN COALFIELD By Gregory E. Tieman 1 and Henry W. Rauch 2 ABSTRACT Dewatering effects from longwall raining were studied for a mine site in south- western Pennsylvania as part of a re- search project funded by the Energy Re- search Center of West Virginia University with contributions by the U.S. Bureau of Mines. The mine showed evidence of dewa- tered streams and ground water supplies. Water sources located much above base level (major stream level) and over or adjacent to recently mined longwall pan- els were partly to completely dewatered, probably owing to downward leakage along subsidence fractures. These lost waters did not migrate to the deep mine because of its thick overburden (at least 500 ft), but instead flowed laterally over confining strata to discharge at nearby streams. Many affected water supplies recovered partially, and all streams re- covered fully within 1 to 3 yr following longwall mining. Spring and well recov- ery occurred most frequently near local stream level where newly formed springs were also common. INTRODUCTION BACKGROUND The coal produced from underground min- ing is an integral component of the econ- omies of the Northern Appalachian Coal Basin. Unfortunately, adverse environ- mental effects can be associated with underground coal mining. These effects include (1) surface damage resulting from land subsidence, (2) degradation of ground and surface water quality, and (3) stream and aquifer dewatering with associated fluctuations of the ground wa- ter levels. Federal regulations are in- tended to ensure the health and safety of the public and to minimize potential dam- age to the environment. The regulations require the mining permit applicant to "...identify the extent to which the pro- posed underground mining activities may proximately result in contamination, di- minution, or interruption of an under- ground or surface source of water within Hydrologist, Malcolm-Pirnie, Inc., White Plains, NY. 2 Professor of geology, Department of Geology and Geography, West Virginia Uni- versity, Morgantown, WV. the proposed mine plan or adjacent area for domestic, agricultural, or other le- gitimate use" (_j_). The partial or complete loss of ground water supplies and streams can be of sig- nificant impact in rural areas of the Northern Appalachian Coal Basin. Al- though mine subsidence damage to a sur- face structure may be significant, the loss of water is often a longer term problem. For example, it is inconvenient to haul replacement water for domestic household use, and replacing lost water supplies for livestock often is unecono- mical. Therefore, an assessment of the impacts of underground coal mining on nearby water supplies is important to help avoid such impacts and to ensure the optimum replacement of lost supplies. The assessment of mining impacts upon local water supplies, particularly im- pacts on the longwall mining method, have been difficult to determine because of a general lack of documented evidence. The ■^Underlined numbers in parentheses re- fer to items in the list of references at the end of this paper. 73 longwall mining method is a nearly total coal extraction system that is account- ing for an increasing percentage of un- derground coal production in the North- ern Appalachian Coal Basin (2^). In this basin there are numerous reports of negatively impacted domestic water wells, springs, ponds, and streams, but there have been few scientific attempts to quantitatively relate these reports to parameters such as type of under- ground mining, age of mines, topography, and mine overburden stratigraphy and thickness (_1_) . The overall objective of this study was to determine the impacts of two under- ground coal mines on streamflow and ground water supply levels. One study site (reported on in this paper as study site Y) is located in Greene County in southwestern Pennsylvania; the second site (study site Z), which is not re- ported on in this paper, is located in Monongalia County in northern West Vir- ginia (fig. 1). Both of these under- ground coal mines utilize the longwall raining method in the Pittsburgh Coal Seam. These two coal mines were selected for study because of the cooperation of the mining companies involved as well as the similarities in geologic setting, mined coal seam, and mining methods used. Both of these mines were studied for wa- ter quantity impacts by mining in an M.S. thesis study by Tieman (3_). The terrain in this portion of the unglaciated Appa- lachian Plateau consists of rounded hills separated by narrow stream valleys with a topographic relief of 300 to 450 ft. The mine overburden at both sites consists of shale, mudstone, claystone, fire clay, sandstone, limestone, and coal with a thickness of 650 to 1,100 ft above the Pittsburgh Coalbed. The effects of coal mining on water resources in Greene Coun- ty, PA, have been generally characterized by the U.S. Geological Survey (4_~_5)« METHODS OF INVESTIGATION A domestic water well and spring inven- tory at mine Y yielded data on 58 water LEGEND — State boundary City 100 J Scale, miles FIGURE 1.— Location map of mine study sites Y and Z. wells and springs in February, June, and July, 1986. The inventoried water sup- plies were located above mine Y and with- in 2,000 ft of the nearest longwall or room-and-pillar mine area (fig. 2). The data collected represented the (1) static water table depth, (2) spring yield, (3) total well depth, (4) depth of well casing, (5) observed subsidence damage, and (6) hydrologic changes to the water source. The data collected were from published reports, information from mine company personnel and owners of the do- mestic water supplies, and from measure- ments made by the senior author. The depth to static water level was deter- mined when the well top was accessible by utilizing a Soiltest electric water level meter. Spring discharge in gal- lons per minute was determined with a 2-gal graduated bucket and stopwatch. Stream discharge at 28 sampling points on 8 streams was measured by the authors on either or both June 6 and July 24, 1986. Streamflow was measured during baseflow conditions, with no measurable precipita- tion during the previous 24 h so as to have stable and easily measurable flows. Stream discharge in gallons per minute was measured using a 6-in PVC pipe and a 2-gal graduated bucket with a stopwatch. A Parshall flume with a 50-mm throat was 74 ■ ;33 ."' -■■ :.\\ *■:>. :■»■ ' . ■»' ■ i.VJ.~:U iVl^^f'-'-TA ' - - -^ (a ^ • 40 'fle 39 •„ c o| "48 £7 4 « 7 4>45 46 '55~54 V "52 36 37 "38 LEGEND — <^ Watershed drainage divide i2» Monitored ground water supply H ** Stream monitoring location Y/A Mined longwall panel 6-1-86 Date ot mining Stream 3,000 I i _J Scale, ft FIGURE 2.— Location of monitored ground water supplies and streams at mine site Y. 75 utilized for monitoring flow at stream H of mine Y owing to its relatively large size. The locations of the sampled water sup- plies and stream monitoring points were plotted on the appropriate 7.5-min topo- graphic map along with the mine outline and type of mining method used. Water- shed boundaries were then determined and plotted on this map for each stream moni- toring point. The watershed boundaries were constructed by first drawing two lines uphill in either direction from the stream monitoring point perpendicular to the elevation contours until the nearest watershed ridge top was intersected, and then tracing the topographic drainage divide along the ridge tops. The area of each watershed was finally determined by utilizing a dot matrix counting technique. The sampled wells, springs, and stream points were then plotted on various fig- ures to determine the horizontal and ver- tical spacings between these features and the mine. The stratigraphic position of each water supply and stream point was also determined. Finally, horizontal and vertical spacings were measured from the sampled water supplies to both the near- est stream and the regional base level, and the topographic position of these supplies was also determined. This in- formation was then used to determine the extent of strata dewatering and to quan- tify the relationships of certain hydro- geologic parameters with dewatering. RESULTS FOR MINE Y PREMINING HYDROLOGIC BALANCE Hydrologic conditions existing prior to mining at the mine Y site were determined by examining nearby local streams and wa- ter supplies that were beyond the influ- ence of mining and up to 2,000 ft away from the nearest mine. The shallow water table for hilltops was typically encoun- tered at a depth of 10 ft, and the shal- low water table under the stream valleys was encountered at depths of 10 ft and less. The depth to static level in deep- er wells averaged 25 ft in the stream valleys and 100 ft on the hilltops. The most dependable water wells were devel- oped into the Jollytown Sandstone Unit in the western portion of the mine. In the eastern portion of the mine the Waynes- burg Sandstone was the most prolific aquifer. These units yielded the most dependable water supplies, especially when the water wells developed in them were located within 100 ft vertically of a stream valley. In two cases such wells were artesian, and even nonartesian valley wells often provide enough water for four or five families each. Devel- oped springs and water wells that are located on hill tops or upper hill slopes appear to have more variable, less de- pendable yields or water levels affected by seasonal changes in precipitation and evapo transpiration. Based on the analysis of control streams in watersheds that were not un- dermined, the bulk of the streams can be classified as small perennial streams. During dry periods stream discharges were composed of just ground water runoff which usually sustained some measurable streamflow. Such streamflow (or base- flow) was proportional to watershed area. The normal median streamflow calculated relative to watershed area from baseflow measurements was 0.0645 gal/(min*acre) on June 6, 1986, and 0.0136 gal/(min*acre) on July 24, 1986, during a much drier period than in June. 76 Stream G Streamy H6 Unmined coal 6-85 9-84 9-83 LEGEND Watershed drainage divide i2» Monitored ground water supply H ** Stream monitoring location V7A Mined longwall panel P2 Longwall panel 7- 24-86 Date ot mining Stream 2,000 Scale, ft FIGURE 3.— Location of monitored ground water supplies and streams overlying or in close proximity to longwall panels at mine site Y. 77 MINING IMPACTS ON HYDROLOGY Wells and Springs Water supplies overlying or in near proximity to room-and-pillar or mine en- try sections have had no reported nega- tive impacts associated with this type of mining. This situation appears to be in response to the relatively thick mine overburden of 490 ft or more between these water supplies and mine Y. Analysis of the water supplies devel- oped in the subsided strata overlying longwall panels (as located in figure 3) indicated that 8 of 11 domestic water supplies that were monitored both be- fore and after mining were partly to completely dewatered (fig. 4). The maxi- mum amount of dewatering appears to have been more extensive near longwall panel centers, such as zone 3 of figure 4, judging by the lowest position of com- pletely dewatered supplies. Dewatering appears to have been limited to the strata located at least 655 to 700 ft above the base of the Pittsburgh Coal. An examination of the ratio of panel width to mine overburden thickness (fig. that dewatering at mine Y a range of ratio values of This range of values cor- a minimum mine overburden thickness of 655 ft (as shown in figure 4), above which dewatering was reported to have occurred. Water supplies located adjacent to but not above longwall panels were also exam- ined for dewatering trends. Analysis of these supplies determined that dewater- ing zones were present, as defined by an angle of influence (fig. 6). The angle of influence is defined as the angle be- tween a vertical line projected upward from the edge (rib or end) of a longwall panel and a line projected to the fur- thest point of dewatering effects from the longwall panel (6). Eleven of 13 5) indicated occurred over 0.75 to 1.00. responded to 900 Q UJUI via: DOtl yo-8C KEY Monitored ground water supplies undermined by longwall panel (a) a No dewatering a Partial dewatering ■ Complete dewatering a, i Partial recovery Dewatering and recovery unknown Drilled or dug wells (with varying degrees of development, recovery, and / or dewatering) b Ground water supply developed post mining Monitored water supply i etc 700- C53 ZCo en & ^ | - 03 |_ 600- 500 400 930 H '29 |H 1 I i3 31 c ■ h ' Dewatering ~ zone I8-QI9 I __*»_ 15 ■ "i — a oie 58 07 oo 10 1 No > dewatering zone D5 A . Zone 1 Zone 2 I Zone 3 i i 100 200 300 400 500 HORIZONTAL DISTANCE OF MONITORED GROUND WATER SUPPLY INSIDE FROM NEAREST EDGE OF LONGWALL PANEL, ft FIGURE 4.— Vertical extent of dewatering for wells and springs over longwall panels at mine site Y. i r KEY Monitored ground water supplies at mine Y W/ 1^3 Complete dewatering y///, mm Partial dewatering //// m m I iNo dewaterinc i m W/< :::::i i 0.75 0.85 0.95 1.05 1.15 1.25 RATIO OF PANEL WIDTH TO OVERBURDEN THICKNESS FIGURE 5.— Dewatering as a function of the ratio of longwall panel width to mine overburden thickness for monitored ground water suplies at mine site Y. 78 KEY Monitored ground water supplies undermined by room-and-pillar section (a), or not undermined (A) Drilled or dug wells (with varying degrees of development, recovery, ,etc and /or dewatering) o Ground water supply developed post mining 28 Monitored water supply A No dewatering A, A Partial dewatering A, A Complete dewatering ik, A Partial recovery M, Jik Complete recovery A^,-A^ Dewatering and recovery unknown *s^l2 I Dewatering zone dewatering nr zone A5 200 400 600 800 1,000 1,200 1,400 HORIZONTAL DISTANCE FROM NEAREST LONGWALL PANEL, ft FIGURE 6.— Extent of dewatering for water supplies adjacent to longwall panels at mine site Y. domestic water supplies that were moni- tored both before and after mining and located within the maximum 42° angle of influence zone were reported to have been partly to completely dewatered by the longwall mining (fig. 6). This dewatered zone adjacent to longwall panels appears to be limited to strata located at least 700 to 720 ft above the base of the Pittsburgh Seam, judging by affected wa- ter supplies that are located in or most- ly in such strata. Partial to complete dewatering of water supplies was found to usually extend no lower stratigraphically than the base of the Jollytown Sandstone (fig. 7). Strat- igraphic cross section A-A'-A" extends the length of the center of panel 2 as indicated in figure 2. The Jollytown Sandstone is situated at or just above the hypothesized regional base level that exists at the position of the major stream (stream H) for this portion of Mine Y. The extent of strata dewatering 79 KEY Location of ground water supplies with respect to stratigraphic cross section A-A-A" monitored ground water supplies undermined by longwall panel (p), or by room-and- pillar section (a) a, a No dewatering A i a Partial dewatering A - ■ Complete dewatering A, A, a, k Partial recovery -A,- Dewatering and recovery unknown o 25 Drilled or dug wells (with varying degrees of development, recovery, and/or dewatering ) Ground water supply developed post mining Monitored water supply UJ > UJ < L±J if) < UJ LU > o < < > UJ _1 UJ c a> o c a> D a> h t- (J3 o u. c c o o c £ 1,000 1,500 2,000 2,500 \ 4,500 5,000 5,500| FIGURE 7.— Location of ground water supplies with respect to stratigraphic cross section AAA" at mine site Y. with respect to regional base level is shown in figure 8 for supplies located reasonably close to stream H. Strata de- watering was found to extend down to a level located about 50 ft above the re- gional base level defined by stream H. Water supplies developed mostly below this 50-ft level appeared to be unaf- fected by the mining. Figure 9 shows that water supplies located mostly above the level of nearby perennial streams (not counting stream H) were dewatered within the 42° angle of influence zone adjacent to longwall panels. Those water supplies developed mostly below nearby streams in elevation appear to have been unaffected by the mining. Perennial streams undermined by longwall panels, as shown in figure 10, seem to be associated with more strata dewatering than such streams not over panels (fig. 9). Dewa- tered supplies extend to about 20 ft or more below the elevation of nearby panel streams, with the maximum dewatering ex- tending to 60 ft below stream elevation at the panel center position, as indi- cated by the water level positions mea- sured by Moebs and Barton (_1_) and Walker, Green, and Trevits (7). 80 KEY Monitored ground water supplies undermined by longwall panel (□), or by room-and-pillar section (a) a, a No dewatering A, ■ Complete dewatering 4 Partial recovery -A Dewatering and recovery unknown o-a, A— A Drilled or dug wells o Ground water supply developed post mining 58 Monitored water supply 900 800 .r 700 < Ll) oc 600 \- - 00 ^_ O u UJ Q. 2 ^ 1 UJ (Z 7* Id o 2 _J _J Q < UJ $ X (.9 tn F" Lt o UJ _l I- < £ 100 80 60 KEY B2m Stream sampling point \ 40 20 BI-B2 >B2-B3 _Decreased streamflow_ zone with dewatering fT i i Normal streamf low zone with no dewatering Increased streamf low zone • A2-A3 EI-E2,F2, Gil, FI-F2 Dl, D2 I •0.80 -0.60 -0.40 -0.20 0.20 0.40 0.60 RATIO OF ACTUAL MINUS MEDIAN NORMAL STREAM DISCHARGE TO WATERSHED AREA, gal/(min- acre) FIGURE 12.— Ratio of actual minus median normal stream discharge to watershed area versus percentage of watershed under- mined by longwall panels at mine Y. Stream G (fig. 3) was monitored on June 6 at 4 sites and on July 24 at 10 sites to test for a possible angle of influence effect on this watershed from panel 3; see figures 12 and 13 for plotted stream data for June 6, and July 24, respective- ly. Figure 13 shows stream segment G1-G2 near stream H to have been completely dry on July 24. Upstream, segment G2-G3 had pooled water with no measurable flow. Between stations G3 and G9 the stream ap- peared to have been steadily losing water as it flowed downstream toward stream H. As shown in figure 13, there was an apparent angle of influence of 31°, as measured from the panel end to point G9 where streamflow had peaked for stream G. Stream G lost water in this 31° dewater- ing zone as well as in the 21° zone lo- cated inward from the panel end (fig. 13). Upstream from station G9 and beyond the influence of panel 3 there appeared to be normal streamflow. Stream H, the major stream in the mined study area, was monitored on July 24 at two locations. Monitoring station HI was located downstream from the area of long- wall mining, and monitoring station H6 84 X o en ZD 00 CO \- \- »♦— n CO UJ r- X Z H d U. 0_ o o UJ z CO _l < CL 00 s -z. < CO UJ UJ ^ ^ < h- 111 UJ a: 00 \- co (D CO o III z z < o Q UJ X CO r- _l < 2 UJ o o Q LT 3 00 C£ UJ > o 800 700 ? 600 500- 400- 300 200- 100- Pooled water 1 KEY ' 65 a Stream sampling point 610 " Angle for complete stream loss Vr Angle for partial stream loss- Longwall panel -300 -200 01 uj2 C3|_ UJ 1,200 _J < UJ (0 z 1,100 < UJ 2 UJ 1,000 o CD < z o 900 r- < > UJ Ul 800 500 1,000 1,500 2,000 2,500 3,000 3,500 DISTANCE, ft FIGURE 14.— Location of selected stream monitoring stations with respect to stratigraphic cross section A-A' at mine Y. 86 Premining potentiometric surface Maximum mining- impacted potentiometric FIGURE 15.— Schematic vertical cross section showing generalized dewatering effects at mine Y. presence of an aquiclude zone located below the regional base level (fig. 15). That zone is predicted by mine subsidence theory, since the minimum mine overburden thickness (500 ft) far exceeds the proba- ble vertical extent of the lower frac- tured and caved zone [30 to 60 times the mined coal seam thickness (8)]. Also, the greater lateral compressive stress and elastic rebound of subsided strata within the aquiclude zone should keep fracture permeability relatively low there (8^. Lack of significant vertical movement of water through this zone is indicated by several nondewatered wells within it and the formation of five new springs just above it. Additional evi- dence against deep mine recharge is that company personnel at mine Y reported that this mine received only about 0.01 gal/ (min*acre) of recharge. Stream H was investigated as a likely major stream to be receiving lost water runoff. The apparent excessive increase in streamflow for stream H must be ac- counted for by the inclusion of an addi- tional water source that is not normally a part of the stream's hydrologic budget. Otherwise, stream H would have had a nor- mal stream discharge to watershed area ratio, which would have been typical for an unmined stream in the area. The most probable additional water source was ex- tra water saved from evapotranspiration due to the effect of mining. To determine the possibility of an evapotranspiration saving accounting for the increased streamflow at stream H, it was first necessary to estimate the evap- otranspiration rate (ET) for the study area. The U.S. Geological Survey (Pitts- burgh office) has reported average precipitation to be 39.4 in/yr for the period 1941-80 based on a rain gauge at Waynesburg near mine Y (_5)« They also estimated that ET averaged 61 pet of to- tal precipitation for this same period for a portion of the watershed of South Fork Tenmile Creek that includes mine Y (_5_). Based on these data, the ET should have averaged 24.0 in/yr for the 1941-80 period at mine Y. This is equivalent to 6.52 x 10 +5 gal/(acre*yr). That ET value was then multiplied by the 535 acres of longwall mine area (including the zone of the 42° angle of dewatering influence) that could have been contributing to ground water runoff entering stream H. This produced a yield of 3.49 x 10 +8 gal/ yr, or 663 gal/min, of average ET from the longwall mine area. The ratio of ex- cess streamflow in the measured stream H segment (17.7 gal/min) to the average ET over the mine is 2.7 pet. Therefore that excess streamflow could be accounted for by a 2.7-pct saving in ET over the mine. This water saving could have been real- ized following subsidence by a diver- sion of shallow ground and stream waters along subsidence fractures to deeper lev- els closer to the regional base level, where the water would be less suscepti- ble to evaporation or transpiration by plants. The water that was thereby spared from ET would have been added to the ground water runoff portion of streamflow for stream H. WATER RECOVERY FOLLOWING MINING Some recovery of water levels occurred for domestic water supplies and streams following mining at mine Y. Of the two accessible water supplies that had been partly dewatered over longwall panels, both showed partial but not complete re- coveries. Of the four accessible sup- plies that were completely dewatered over longwall panels, only two had a partial recovery and none had a complete recov- ery. Of nine affected water supplies for which recovery data were available and which were not undermined by a longwall panel but were located within the zone of the 42° angle of influence, eight had a partial recovery and one had a complete recovery. Water supplies that showed partial to complete recoveries did so within 1 to 3 yr after dewatering oc- curred and were typically located close to or below nearby perennial streams or hydrologic base level. Supplies located over panel centers showed less overall recovery compared to supplies located near panel edges or adjacent to panels. Only one supply (well 2) has exhibited complete recovery so far. While some old affected springs showed partial recovery, several new springs ap- peared or were developed at lower eleva- tions following mining. Two new seasonal springs were developed over panel 2; one new perennial spring was developed over panel 3 and is being used to supply five families; one new seasonal spring was de- veloped over panel 4; and one new spring was developed over panel 5. All of these springs appeared immediately to 1 yr after undermining occurred. The new springs appearing above panels 2 and 3 are located stratigraphically at the top of the Jollytown Sandstone, whereas the new springs above panels 4 and 5 are situated approximately 20 ft above the stratigraphic level of the Jollytown Sandstone (fig. 14). In addition, a do- mestic well that was not dewatered over the mine had a reported increase in yield following mining; this well penetrates to below the Jollytown Sandstone. These enhanced water supplies again indicate that overlying aquifer waters were di- verted downward and that partial recovery of ground water often occurred close to the local or regional base (stream) level within 1 yr of mining. A comparative analysis of streamflow to watershed area ratios for June showed in- directly that panel 1 had a complete re- covery and panels 2 and 3 had partial re- coveries since longwall mining occurred. The portion of panel 1 that undermined subwatershed B3 was between 3.0 and 3.5 yr old, whereas the portion of panel 2 that undermined subwatershed B2 to B3 was between 2.0 and 2.5 yr old. Subwatershed B2 to B3 was calculated to be losing 0.811 gal/(min*acre of longwall panel) to 88 the subsurface. In contrast, the portion of panel 3 that undermined the stream G watershed was between 1 and 2 yr old, and subwatershed G4 to Gil was calculated to be losing 2.08 gal/(min*acre of long- wall panel) to the subsurface. No cal- culations of maximum streamflow loss, such as those above, were possible for panels 4 and 5 owing to the dry streams encountered over these panels. The obvi- ous trend from these data is that streams show progressive recovery with time between about 1 and 3 yr of longwall min- ing and recover their normal premining flow by about 3 yr after mining. CONCLUSIONS FOR MINE Y 1. Roora-and-pillar mining (with no subsidence) has not affected water sup- plies or streams. 2. Above longwall panels, 8 of 11 wa- ter supplies appeared to have been partly to completely dewatered. Dewatering oc- curred in a zone about 655 ft and higher above the mined Pittsburgh Coal Seam, but not closer to this seam. Dewatering was greatest above the center of the longwall panels. All supply dewatering over long- wall panels occurred within a range of 0.75 to 1.00 for the ratio of longwall panel width to mine overburden thickness. 3. Off of and away from the longwall panels, 11 of 13 water supplies were partly to completely dewatered within a maximum angle of dewatering influence of 42° from the nearby panel ribs or ends. Dewatering occurred in a zone of about 720 ft and higher above the mined Pitts- burgh Coal Seam, but not within 720 ft of this seam. 4. Dewatering did not occur below the regional base level, as represented by the major stream over the mine. Water supplies developed within 800 ft lateral- ly of the major stream and no higher than 50 ft above it were not dewatered, but such supplies over 50 ft above it were partly to totally dewatered. 5. Water supplies developed at or be- low the elevation of the nearest peren- nial stream and outside of panels but in- side the 42° angle of influence zone did not appear to be dewatered. However, such supplies developed mostly above the level of the nearest perennial stream were at least partly dewatered. Most wa- ter wells developed more than 20 ft below the elevation of the nearest perennial stream and undermined by longwall panels were not dewatered, but such wells lo- cated above this level were partly to totally dewatered. Above the center of one longwall panel, a well was dewatered to a depth of 60 ft below the elevation of the nearest perennial stream but not below regional base level. 6. Streams located above regional base level and undermined by longwall panels less than 2.5 yr old were partly to com- pletely dewatered during baseflow condi- tions. An angle of dewatering influence of 31° existed outside the end of such a panel for one measured stream, with par- tial stream dewatering having occurred within this area adjacent to the panel. Streams located above regional base level and also above panels at least 3 yr old had normal flows. 7. Water from lost ground water sup- plies and streams did not penetrate to the mine, but instead migrated downward through probable subsidence fractures to near regional base level, where it mi- grated laterally through the Jollytown Sandstone to finally discharge to the largest area stream over the mine. Re- ported ground water recharge to the deep mine was only 0.0.1 gal/(min*acre). The major area stream flowing over the mine received about 17.7 gal/min more than the expected normal runoff contribution on July 24, 1986. This extra slug of runoff during dry baseflow conditions was most likely contributed by a potential evapo- transpiration reduction of about 2.7 pet over the mine that resulted from reduced stream and ground water levels at higher elevations. 8. Several streams and water supplies showed partial to total recovery follow- ing mining effects. Streams appeared to 89 have had complete recovery within 3 yr after longwall mining occurred. Of the accessible ground water supplies over longwall panels, all partly dewatered supplies had partial recovery, but only one-half of the completely dewatered sup- plies had a partial recovery, with no complete recovery observed. Of the af- fected accessible supplies adjacent to longwall panels, all showed at least par- tial recovery. Several new springs also formed close to stream level within 1 yr of nearby mining. Any noted recovery for affected wells and springs occurred with- in 1 to 3 yr of the initial mining ef- fects, with more extensive and rapid re- covery for supplies near stream level or not over panel centers. Only one af- fected water supply had shown complete recovery as of February 1986. Affected streams exhibited more complete recovery than did wells and springs following mining. REFERENCES 1. Moebs, N. M. , and T. M. Barton. Short-Term Effects of Longwall Mining on Shallow Water Sources. Paper in Mine Subsidence Control. Proceedings of Tech- nology Transfer Seminar, Pittsburgh, PA, September 19, 1985, comp. by Staff, Bu- reau of Mines. BuMines IC 9042, 1985, pp. 13-24. 2. Merritt, P. C. U.S. Sees Marked Growth in New Longwall in 1985 (1985 Longwall Census). Coal Age, v. 90, No. 8, 1985, pp. 47-65. 3. Tieman, G. E. Study of Dewatering Effect at Two Underground Longwall Mine Sites in the Pittsburgh Coal Seam of the Northern Appalachian Coal Field. Unpub- lished M.S. thesis in Geology. WV Univ., Morgantown, WV, 1986, 147 pp. 4. Stoner, J. D. Probable Hydrologic Effect of Subsurface Mining. Ground Wa- ter Monitoring Review, v. 3, No. 1, 1983, pp. 128-137. 5. Stoner, J. D. , D. R. Williams, T. F. Buckwalter, J. K. Felbinger, and K. L. Pattison. Hydrogeology , Water Resources, and the Hydrologic Effects of Coal Mining, Greene County, Pennsylvania. U.S. Geol. Surv. , 4th Ser. , Water Resour. Rep., PA Geol. Surv., (in press, 1987). 6. Cifelli, R. C. , and H. W. Rauch. Dewatering Effects From Selected Under- ground Coal Mines in North-central West Virginia. Paper in Proceedings of Second Workshop on Surface Subsidence Due to Underground Mining. WV Univ. , Morgan- town, WV, 1986, pp. 249-263. 7. Walker, J. S., J. B. Greene, and M. A. Trevits. A Case Study of Water Level Fluctuations Over a Series of Long- wall Panels in the Northern Appalachian Coal Region. Paper in Proceedings of Second Workshop on Surface Subsidence Due to Underground Mining. WV Univ. , Morgan- town, WV, 1986, pp. 264-269. 8. Coe, C. J., and S. M. Stowe. Eval- uating the Impact of Longwall Coal Mining on the Hydrologic Balance. Paper in Pro- ceedings of Conference on the Impact of Mining on Ground Water. National Water Well Association, 1984, pp. 348-359. 1187 234 US GOVERNMENT PRINTING OFFICE 1987 605 01 7 6*. . §?|1k2 * «? ^ oMW * aV "Jv. . W \c? SMS;* % s*.-$&\ J#fe'.\ ^'&&X << 'K +±J a,0° ^ « -• ^ 4 CJ. -oV ^o. ♦•To a0^ *^j K :«: V ho, °ww: ***** * *o, ** °^%^> ,0* ».^ V 4* V 4.9 •!••»*>' V »•!!£/♦ «>& a9 • !*£» *> aV^ <* & * • ©., V* U A* \g ' * jV /..i^.*°o > V ..55K%V d»*..ii&.% 1 «* **" •• f**^ c5^ Cr o, *••»• A <» *«vv»* *%, *•••»* A <. ♦/TV*' »0 V >», "« £ ^. "^/ 'ISEPT.-OCT.1987 ■^Z %^ ^V •bV" ■v-o"