Reflection seismic sources for defining coal seam structure

Abstract

The prime objective of the investigations described in this paper is to improve the reliability of structural information in advance of underground mining. Kembla Coal and Coke extracts high-grade coking coal by longwall mining methods. In 1984 major disruption to mining was caused by an unexpected fault which crossed the longwall panels. The costly consequences of this faulting provided the Company with a strong incentive to obtain accurate structural data in advance of mining. Seismic reflection work carried out in the 1970s using large explosive charges, produced poor data of very low resolution. In 1987, the Mini Sosie technique was used to obtain seismic data. Fair data were obtained with moderate resolution. Work carried out in 1988 established that a small explosive charge at the base of weathering produced still better data. The weathering depth was determined by a continuous seismic refraction survey. In order to reduce the costs of obtaining seismic data, as well as seeking further improvements in data quality and resolution, experiments were carried out in 1990 and 1991 using alternative sources. Data from a Bolt LSS-3B air gun, a Mertz M26 vibrator and 1.5 m deep explosive charges were compared with data from explosives at the base of weathering. Experiments were also carried out using shaped charges and high velocity of detonation explosives (up to 8600 m/s). The results from the various experiments and production work have demonstrated that all the surface and near-surface sources used, generated substantial ground roll, which reduced the dynamic range of the recording instruments and degraded the signal-to-noise ratio. Weathered near-surface layers also significantly attenuated the high frequencies, resulting in lower temporal resolution. A 240-9 conventional explosive with 6800 m/s velocity of detonation, placed below the base of weathering, proved to be the best available source for obtaining good quality high-resolution data. Substantial improvements in data quality and resolution were achieved. Further work is still required in other areas of data acquisition, processing and modelling to improve confidence in the interpretation of very small faults.