Model coal pillars from NX diamond drill cores from four coals were broken in compression; the model pillars were 54 mm in diameter with six diameter-to-length ratios and eight end conditions. The purpose was to define relationships between strength, constraint, geometry, and failure mode. The test results indicate that the compressive strength and the burst proneness are statistically significant functions of the end constraint, the geometry, the particular coal, and the interaction between the end constraint and the geometry. The strengths ranged from 15.9 MPa (2,310 psi) to over 148.8 MPa (unbroken at 21,578 psi). The burst proneness ranged from 0 to 90 percent, defined to be the percentage of the weight thrown off a press table 30 inches square. Both long and short pillars could be made to crush or burst, given suitable end conditions. This reveals that the geometry alone is an adequate basis for pillar design.
The prevailing pillar design equation, based on the early work of Wilson and Ashwin (1972), assumes a confined core and does not explain burst behavior. Because extensive bursting occurred in the present tests, this work will be related to the work of Johnson (1897). Babcock, et al. (1981) summarized many questions in the literature with the equation (mathematical equation)(available in full paper)
In Johnson's work (1897), A =0.778, B=0.222, and a = ß =1. The pillar strength for any W/H ratio, s, is related to the strength of a pillar with W/H = 1, denoted by s1:1 . The volume, one factor in pillar strength, is automatically covered by the o1:1 factor, which can be for a test volume of any size. W and H are the pillar width and height, respectively. A common assumption in pillar design is that the laboratory test strength is many times larger (say 5) than the in situ pillar strength. This is often assigned to the difference in the size between the laboratory sample and the mine pillar. This is perhaps a factor of 106 times. Some of the difference in strength could be the result of the difference in end conditions, steel in the laboratory, and rock in the mine.
3. EXPERIMENTAL PROCEDURE
Over 500 photographs were taken of 206 tests to record the failure modes produced. Four sets of samples tested for analyses of variance had the geometries shown (Figure 1). Three of the coals tested were from Colorado, and the fourth was from Utah. The eight different end conditions used to provide differences in end constraint are denoted El through E8 (Figure 2). The designations S, C, C1, Cd, Wc, Ss, T, and SI stand for the layering sequence from top to bottom and denote steel platen, coal sample, m-thick powdered clay layer, 1.3 m-thick powdered coal layer, wet coal, sandstone, TeflonR1 , and slate, respectively. Both clay and coal dust were -35 mesh in size. The sandstone and slate layers were 13 m thick and 100 m in diameter.
FIGURE 1. - One of four sets of samples tested for analyses of variance.(available in full paper)