Mining Publication: Deep Coal Longwall Panel Design for Strong Strata: The Influence of Software Choice on Results
Software is often used to construct design models used in longwall panel design. These models necessarily reduce the abundant variability found in nature to a simplified representation that, ideally, captures the relevant characteristics of ground response to mining. The choice of stress analysis software is an important step in the modeling process. The importance of this step can be easily overlooked, yet assumptions inherent in modeling software can have a decisive influence. An appreciation of this step is important for both practitioners and users of design model results, particularly when decision-makers are integrating results into design decisions or evaluating the adequacy of design specifications. This paper examines four different stress analysis programs or tools for analyzing stress around a single longwall panel at 610 m (2,000 ft) depth beneath overburden containing some strong strata. The four tools are: an empirical model that underlies the Analysis of Longwall Pillar Stability (ALPS) program; the three-dimensional displacement-discontinuity program MULSIM/NL; the three-dimensional flexible overburden program LaModel, which has largely superseded MULSIM/NL; and the two-dimensional volume-element program FLAC. This study examines key aspects of panel simulation, including stress transfer through the gob, stress concentration in abutment ribs, and stress transfer distance into abutments. MULSIM/NL and FLAC results were the most similar. LaModel results varied greatly in terms of peak stress and stress transfer, while ALPS produced the least peak stress. This study also examines the transferability of calibrated input properties between MULSIM and LaModel. For instance, displacement-discontinuity codes have been replaced, at least in part, by LaModel, yet considerable experience exists in calibrated models with older tools. Whether and how this experience can be incorporated into analyses using other tools is a subject of some controversy and must be approached carefully. In all cases, selecting models and properties appropriate to site conditions is extremely important.