资源描述
英文原文25th International Conference on Ground Control in MiningTHE EVOLUTION OF INTELLIGENT COALPILLAR DESIGN: 1981-2006Christopher Mark, Chief, Rock Mechanics SectionNIOSH-Pittsburgh Research LaboratoryPittsburgh, PA, USAABSTRACT： The first International Conference on Ground Control in Mining opened with the topic of pillar design. Two classic papers were presented, one by Bieniwski and the other by Wilson. Unfortunately, the two methods were so radically different from each other that it was nearly impossible to reconcile them. Adding to the confusion were the many other pillar strength formulas (such as the Salamon-Munro, the Holland-Gaddy, and the Obert-Duvall, just to name a few) that were also available. Little wonder that discussions of pillar design in those days often ended with anguished cries of “but which formula is the right one?” The past 25 years have seen substantial progress in the science of coal pillar design. Indeed, one testament to the improvement is the relative scarcity of papers on the topic at recent Conferences. Two factors have been largely responsible for the progress that has been made. The first has been the collection of large data bases of actual case histories of pillar performance in a variety of settings, from shallow room-and-pillar mines through deep cover longwalls. These have made possible the development of empirical design procedures that are closely linked to real world experience. The second important factor is the development of sophisticated computer models that can accurately simulate pillar behavior and roof/pillar/floor interactions. Together, these two lines of research have led to a new understanding of pillar mechanics that identifies three modes of pillar failure: Sudden, massive collapse, accompanied by airblast, for slender pillars (width/height4) Squeezing, or slow, non-violent failure, for most room and pillar applications (4w/h10)It is particularly satisfying that the insights gained from numerical models broadly support those obtained from the empirical studies. While far less controversial than in the past, pillar design problems continue to arise. One recent example is pillar design for highwall mining. NIOSH has just released a software package, called ARMPS-HWM, which employs a number of modern pillar design concepts. Since highwall mining web pillars are long and slender, the greatest danger is that of a sudden collapse. ARMPS-HWM suggests two possible prevention strategies, one which concentrates on the SF of the webs, and the other which creates a “pressure arch” using properly sized barrier pillars. The paper will close with a discussion of some current needs in coal pillar design, including: Updating older empirical methods, such as ALPS, where changes in technology (new types of roof support, more demanding ventilation requirements, faster retreat rates) may have made some of the original case histories obsolete. Methods for determining site-specific coal strengths, focusing on bedding plane strength and other factors that may effect confinement, as input for both empirical and numerical design. Improved methods for evaluating coal pillar performance for environmental issues, such as surface subsidence and hydrologic impacts, which consider such factors as depth, w/h ratio, water immersion/drainage, and time dependent seam strength. ACKNOWLEDGEMENT (AND APOLOGY) The topic of pillar design is one of the most important in the field of coal mine ground control, and the substantial progress that has been made has been due to the collective effort of many researchers and practitioners. In a brief overview like this one, it was only possible to mention a few of those who have made important contributions. As an apology to the many whose valuable work I was unable to include, I can only say that you are in very good company. INTRODUCTION Babcock et al. (1981), writing in their survey paper for the First Conference on Ground Control in Mining, traced the science of pillar design all the way back to Coulomb in 1773. During the ensuing century, a variety of researchers tested rock specimens of a variety of sizes and shapes. However, it was not until 1911 that Bunting (1991) proposed the first true pillar design method for coal mines. Bunting described the necessity for pillar design this way: To mine without adequate pillar support will result, sooner or later, in a squeeze; the inherent effects of which are crushing of the pillars, caving of the roof, and heaving of the bottom. In developing his formula, Bunting and his collaborators tested the strength of coal specimens in the laboratory and conducted back-analysis of full-scale pillar failures (“squeezes”) underground. Using essentially the same approach, a number of pillar design formulas were developed during the next 70 years around the world. These “classic” methods consisted of three steps: 1. Estimating the pillar load using tributary area theory; 2. Estimating the pillar strength using a pillar strength formula, and; 3. Calculating the pillar safety factor (SF). Step 1, estimating the load, was fairly straight forward for an industry that relied almost exclusively on room-and-pillar mining at relatively shallow depth. The tributary area estimate was considered sufficient, though it was recognized that in narrow panels the pillars near the edges might not experience the full load. More complex were the issues associated with pillar strength. The two big issues were the “size effect” and the “shape effect.” The size effect was most prominent in the laboratory, where coal strength testing showed that larger specimens were much weaker than small ones. The shape effect referred to the observation that slender (low width-to-height ratio) pillars were weaker than ones that were more squat. As the number of classic formulas proliferated, so did the arguments. Should the shape effect be represented as a straight line, or as an exponential equation? Was there such a thing as a “critical” specimen size? Could a “universal” formula even exist, or did each one have its own place? These issues were discussed at length in a number of survey papers that were a persistent theme in those days (Babcock et al., 1981; Logie and Matheson, 1983; Hustrulid, 1976). In some respects, Bieniawski represented the culmination of the classic approach to pillar design. In his paper at the First Conference in Morgantown, Bieniawski clearly described the issues involved in pillar design, and the advantages and shortcomings of the available methods. He then outlined a logical, step-by-step approach to sizing coal pillars. Indeed, Bieniawskis work has provided a firm foundation upon which many of the developments of the past 25 years have been built. However, the First Conference also contained a paper that described a radically new and different approach to pillar design. Arthur Wilson (1972, 1981) of the British National Coal Board had first proposed his “hypothesis concerning pillar stability” in 1972, but by 1981 he had expanded and refined it considerably. His frame of reference was deep longwall mining, where very large pillars were routinely employed. Here, the goal of pillar design was not to prevent a pillar collapse, but rather to ensure the serviceability of the gate entries. Wilsons first problem was the need to go beyond tributary area and consider the abutment loads brought about by full-extraction mining. His concept of the “load balance,”whereby the reduction of load in the gob equals the excess load carried by the chain pillars, allowed the first serious quantification of abutment loads. More fundamental were Wilsons innovations in defining pillar strength. In contrast to the empirical formulas, where “strength” was simply the failure load divided by the pillar area, Wilson treated the pillar as a complex structure, with non-uniform stresses throughout. His key insight was that the “shape effect” is caused by the build-up of confining stress within the pillar, which creates a high-strength “core” in the pillar center. While Wilsons mathematics contained some serious flaws (Mark, 1987; Salamon, 1992), his basic concepts are unchallenged today and underlie virtually all numerical models (Gale, 1996). For many First Conference participants, however, it was pretty difficult to see how Bieniawskis approach could ever be reconciled with Wilsons. While they both purported to address pillar design, the input parameters, mathematical formulas, and (most importantly) the predicted pillar sizes seemed to be radically different. In 1992, the situation seemed, if anything, to have become more confused. In that year the U.S. Bureau of Mines sponsored the first Workshop on Coal Pillar Mechanics and Design (Iannacchione et al., 1992), which featured 22 different papers from leading practitioners from around the world. Nearly every paper described a different approach, and these were approximately evenly split between empirical, analytical, and numerical methods. Their predictions for pillar strength varied widely, however, even in their trend. Some predicted that pillar strength would increase exponentially as the w/h ratio increased, others predicted it would tend towards a maximum limiting value, and still others predicted an intermediate, linear increase (figure 1). Stress measurements from 34 coal pillars were also analyzed, but were no help in narrowing the field (Mark and Iannacchione, 1992).Moreover, the Workshop participants could not even seem to reach agreement on something as fundamental as what constituted pillar failure. The classic approach contended that pillars will fail when the applied load reaches the compressive strength of the pillars and that the load-bearing capacity of the pillar reduces to zero the moment the ultimate strength is exceeded (Bieniawski, 1992). In this view, which was represented most strongly by the South African experience, the only true failures were those in which the panel width was very wide compared to the depth, and subsidence could actually be confirmed on the surface. Pillars with w/h ratios greater than about 10 were considered “indestructible” (Wagner, 1992). At the other extreme were those whose experience was framed by longwall mining. These experts had seen plenty of examples where pillars with w/h ratios well in excess of 10 had proved too small and resulted in poor ground conditions. Obviously such squat pillars had not “failed” in the classic sense that their load-bearing capacity had disappeared. Yet they had failed to perform their ground control function. In many of these cases, conditions improved when the pillars were made larger. Clearly pillar design was still essential to maintaining gate road stability. Observing the discussion, an outsider might have been forgiven for thinking that he had happened across a modern-day Tower of Babel. There were at least three groups, the empiricists, the modelers, and the theoreticians, each apparently speaking their own language. Even within each group there were bitter disputes. Yet just seven years later, by the time of the second Workshop on Coal Pillar Mechanics and Design (Mark et al, 1999), a rough consensus had been reached on a unified theory of coal pillar mechanics. What had happened? THE NEW PILLAR MECHANICS PARADIGM The explanation can be summarized by another ancient parable, the one about the three blind scholars and the elephant. Each explored a different part of the elephantone the trunk, another an ear, the third a leg. Based on his own observations, each one felt that he could describe the elephant, yet their descriptions were so different from one another that they could find no common ground. Only when they put all their observations together, however, could they get a true picture of the beast. The answer in this case was that while all coal pillars are made of the same basic material, not just their strength but their behavior can vary dramatically depending on their shape. In fact, three broad categories of pillar behavior and failure mode can be identified, eachdefined by an approximate range of width-to-height ratios (Mark, 1999): Slender pillars, whose w/h ratios are less than about 3 or 4. When these pillars are loaded to their maximum capacity, they fail completely, shedding nearly their entire load. When large numbers of slender pillars are used over a large area, the failure of a single pillar can set off a chain reaction, resulting in a sudden, massive collapse accompanied by a powerful airblast. Intermediate pillars are those whose w/h ratios fall between about 4 and 8. These pillars do not shed their entire load when they fail, but neither can they accept any more load.Instead, they deform until flexure of the overburden transfers some weight away from them. The result is typically a non-violent pillar “squeeze,” which may take place over hours, days, or even weeks. The large roof-to-floor closures that can accompany squeezes can cause hazardous ground conditions and entrap equipment. Squat pillars are those with w/h ratios that exceed 10. These pillars can carry very large loads, and may even be strain-hardening (meaning that they may never actually shed load, but just may become more deformable once they “fail.”). None the less, the pillar design may fail because excessive stress is applied to the roof, rib, or floor, or because the coal bumps. Moreover, the strength of squat pillars can vary considerably depending upon the presence of soft partings, weak roof or floor interfaces, and other geologic factors. Although derived from laboratory data, figure 2 illustrates how the post-failure behavior and the residual strength of coal pillars changes with their shape (Das, 1986).What was the evidence for this new model of pillar mechanics? In essence, two largely separate lines of research had converged upon very similar conclusions. One source was a new generation of empirical studies, the other sophisticated numerical modeling. EVIDENCE FROM MODERN EMPIRICAL METHODS Prior to 1990, most classic pillar design methods had been derived from curve-fitting to coal strength data obtained from laboratory or in situ testing. The most notable exception was Salamon and Munros formula, which was based entirely on statistical analysis of 98 unfailed and 27 collapsed pillar panels in South Africa. Salamon and Munro had developed their formula following the sudden, disastrous 1960 pillar collapse at the Coalbrook Colliery in which 437 lives were lost (Wagner, 1992)Salamons approach, that of using case histories involving full-scale pillars from actual mines, has a lot to recommend it. With such real-world data, it is not necessary to fully understand the mechanics, though a “reasonably clear understanding of the phenomenon in question” (Salamon, 1989) is needed to guide both the data collection and the statistical analysis. Moreover, the design equation that results from the analysis is generally simple, realistic, and thoroughly verifiable. In essence, it makes the past experience of a broad segment of the industry available to mine planners in a practical form. The Analysis of Longwall Pillar Stability (ALPS) was the first modern pillar design method to employ a large case history data base like Salamons (Mark, 1990; Mark, 1992). While the original ALPS research focused on defining longwall abutment loads using stress measurements, the real crux was identifying the proper SF to use for design. The case history data, obtained from a broad cross-section of mines across the U.S., showed that both successes and failures, defined in terms of tailgate serviceability, occurred over a wide range of pillar SFs. Clearly other factorslike the strength of the roofwere involved.This observation fit well with studies conducted as early as the 1960s that had concluded that whether or not the stress from an extracted longwall panel will influence a roadway depends more on the strength of the rocks which surround the roadway itself than on the width of the intervening pillar (Carr and Wilson, 1982). Yet the variety and complexity of geologic environments had defied effective measurement, making it difficult to incorporate rock strength into design. The Coal Mine Roof Rating (CMRR) overcame this obstacle by providing a quantitative measure of the structural competence of coal mine roof (Molinda and Mark, 1994). When the CMRR was included in the analysis, ALPS could successfully predict the outcome in 85% of the case histories (figure3). The analysis indicated that under very strong roof, the SF could be as low as 0.7, while under weak roof, an SF of 1.3 was required (Mark et al., 1994).Building on the success of the ALPS method, the research that culminated in the Analysis of Retreat Mining Pillar Stability (ARMPS) program employed an even larger case history data base. Here, most of the failures (unsatisfactory designs) involved pillar squeezes. For much of the data base, an SF of 1.5 seemed to separate the successful designs from the unsatisfactory ones (Mark and Chase, 1997). There were two interesting exceptions, however.At one extreme, the ARMPS data base included case histories of 12 massive pillar collapses, each of which had occurred so suddenly that they generated powerful airblasts (Mark et al., 1997). Like the more common squeezes, the collapses all involved cases where the ARMPS SF was less than 1.5. What really distinguished the sudden collapses from the slow squeezes, however, was the pillar=s w/h ratio. Every massive pillar collapse involved slender pillars whose w/h was 3 or less (figure 4). Subsequently, it was noted that all of Salamons South African collapse cases also involved pillars with w/h less than 4. Apparently, these types of failures form a separate class, distinct from the squeezes that are more common in the U.S.The other anomaly occurred with the cases where the depth of cover exceeded 750 ft. In this group, both successes and failures occurred with SF that were well under 1.5, and it was much more difficult to separate them. A later study (Chase et al., 2002) added nearly 100 more deep-cover cases to the data base. Most of the failures were still squeezes, but bumps became more common at greater depth and with stronger roof. The study concluded that the apparent pillar strength for these squat, deep cover pillars was more variable than it was for the typically more slender, shallower pillar cases (figure 5). Roof quality was found to be significant; as was the use of barrier pillars (which no doubt reduced the applied load).CONCLUSIONS Recent years have seen significant advances in the state-of-the-art in coal pillar design. From a practical standpoint, the development of reliable empirical methods like ALPS and ARMPS has been particularly valuable. They have been widely accepted throughout the mining community because they have been verified by extensive data bases of real-world case histories, and because they have been readily available in user-friendly computerized formats. The tremendous advances in numerical modeling have been another important success story. The research has led to some other important conclusions, including: Laboratory testing of small coal samples, particularly uniaxial compressive strength tests, are not useful for predicting pillar strength; The strength becomes more difficult to predict as the pillar becomes more squat; The w/h ratio is important for predicting not just the pillar strength, but the mode of failure, and; Many ground control problems must be considered fr
展开阅读全文