Abstract Extraction of soluble minerals, whether by natural or man-induced processes, can result in localized land-surface subsidence. The subsidence is caused by partial or total collapse of underground cavities resulting from dissolution of salt or other soluble evaporites. In many cases, subsidence is ultimately related to the strength limit of the overlying rocks that form the unsupported roof above the cavity. Downwarping results where strength of roof spans are exceeded. In other cases, collapse of the undermined roof causes stoping of the overburden rocks. If sufficient underground space is available for the loosely packed rock debris to collect, the void can migrate to the surface and produce surface subsidence, or in the extreme, catastrophic surface collapse. Another mechanism is subsurface erosion of susceptible layers (sandstone, silt, loess) overlying salt cavities. Ground water can erode and transport the loose material down subsidence-induced and natural cracks, or drill holes into the salt cavity. The voids formed in the higher eroded beds can then cause surface subsidence.

More than 2400 ft of exploratory core drilled for an underground installation was logged in terms of joint frequency, degree of weathering, relative hardness, core loss and broken core. A statistical regression analysis was computed relating weathering, hardness, core loss and broken core to joint frequency. A significant correlation existed between these parameters within practical limits. The correlation broke down when joint frequency exceeded 8 joints per foot, and when core loss and broken core was greater than 30 percent for a given core interval. The range of joint frequency between no joints and 8 joints per foot was arbitrarily divided into 10 equal divisions and designated as grades 1 (poor) to 10 (good). The corresponding values for weathering, hardness, core loss, and broken core were assigned to the 10 grades. A grade of φ was reserved for faults. A regression analysis was computed relating the rock grades to laboratory-determined physical and mechanical properties of core samples previously determined by logging. Significant correlations existed between the rock grades and dry bulk density, total porosity, Young’s modulus, shear modulus, and bulk modulus. Poisson’s ratio, however, did not show any significant correlation with rock grade. Underground mapping confirmed that rock of grades 8 to 10 presented no construction or stability problems. Rock of grades 5 to 7 was acceptable but tended to have moderate overbreak and required occasional steel sets for support. Rock of grades 3 and 4 was unstable and had to be supported. Rock of grades φ to 2 was incompetent.