As underground coal mining continues to evolve in the U.S. more reserves are being mined under deeper cover, with worse roof, or with interactions from previous workings. At the same time, the mining community is responding to higher safety standards, new subsidence regulations, and intense competitive pressures. The need for accurate pillar design has never been greater. During the 1970's and 80's, a number of field studies were conducted in coal pillars underground. As research funding from both government and private sector has diminished, the emphasis has shifted to empirical studies and numerical modeling. Empirical methods emphasize the collection and interpretation of case histories of pillar performance. Statistical methods are employed to determine those variables which are most important to the success of a pillar design. Large data bases of real-world pillar successes and failures have been compiled for the Analysis of Retreat Mining Pillar Stability (ARMPS) and Analysis of Longwall Pillar Stability (ALPS) formulas. Using these data, valuable insights into the importance of the width-to-height ratio (w/h), the role of coal strength testing, and the interaction between pillar performance and roof quality have been obtained. Numerical models used for pillar design may be divided into two categories. Finite element (FEM)and finite difference (FDM) models are best suited for investigating the behavior of detailed cross-sections of pillars and surrounding strata. Boundary Element (BEM) can analyze stress distributions in large three-dimensional areas of tabular deposits. FEM were used to investigate the effects of roof and floor frictional interfaces, the pillar width-to-height ratio, and clay partings on pillar strength. The results were compared to underground stress measurements and to empirical formulas. A new BEM program, called LAMODEL, incorporates laminated overburden which more accurately simulates overburden behavior than earlier versions. It is used to investigate the effects of multiple seams, mining steps, complex pillar layouts and variable topography on the pillar loading. Despite the differences in technique, the empirical and numerical lines of inquiry have converged on several important conclusions. For example, three broad categories of pillar behavior have been identified: Slender Pillars (w/h), which are subject to sudden collapse, Intermediate (3<w/h<8), in which pillar squeezes seem to be the most common failure mode, and, Squat pillars (w/h>8), which are dominated by entry failure (rib, roof, or floor) and coal bumps. There is also agreement that coal seam discontinues and roof-pillar-floor interactions are critical to pillar strength. A key outstanding issue is the need for simple field techniques to evaluate the in situ strength of the floor and the coal seam.
National Institute for Occupational Safety and Health, Pittsburgh Research Laboratory, Pittsburgh, PA 15236