157 / 2023-08-30 10:18:28

Three-dimensional Automatic Modeling Method of Steep Thin Orebody and Complex Stratum Based on Multi-point Geostatistics

Extremely thin coal seams,Multipoint geostatistics,Principal component analysis,Uncertainty analysis,Training image

The geological environment of coal mines often includes thin coal seams and other Quaternary loose surface strata, and its geological structure and geological phenomena have great randomness and uncertainty in microcosmic. Refined three-dimensional coal seam geological model and uncertainty quantification intuitively point out the size and distribution of uncertainty in the model, which is helpful to guide the mining work and ensure the safe and efficient mining of mines. Traditional multi-point geostatistics usually uses three-dimensional conceptual models or two-dimensional profiles as training images, which lacks the simulation of scattered and uneven low-dimensional borehole data, and does not fully utilize the occurrence information of geological structures. Based on the vertical and horizontal distribution characteristics of boreholes, this paper puts forward a multi-point geostatistical stochastic simulation method and stratum uncertainty analysis method for occurrence constraints of borehole data in extremely thin coal seams, and realizes automatic geological modeling and uncertainty evaluation of extremely thin coal seams. First, we re-sampled the borehole data and calculated the occurrence and principal direction of each formation (coal/rock) by principal component analysis. On this basis, the spatial correlation of strata between different underground units is characterized by the autocorrelation function constrained by strata occurrence, and the probability of specific strata in a given underground unit is determined according to the spatial correlation, and the prior probability and prior model are obtained. This method is different from other modeling methods that depend on borehole data, does not need to consider the connection relationship between strata, reduces the uncertainty of the model, and also provides the possibility for modeling extremely thin coal seam, which is a geological body with small volume and obvious occurrence. Secondly, we propose Simulated Path Optimization (SPO) strategy and Occurrence Local Scan (OLS) strategy to make simulation nodes scan within the range of attitude constraints and establish the final model. These strategies ensure the attitude constraint and partition simulation, and eliminate the discontinuity between partition boundaries. Finally, we validated the sensitivity of the modeling parameters and the effectiveness of the algorithm proposed in this paper using test data, Perth data from Western Australia, and Shuangyang Coal Mine in Heilongjiang, China, and compared it with other modeling methods. The results indicate that the proposed 3D geological modeling and uncertainty quantification method can effectively reveal the structural form and coverage relationship of coal and rock layers, conveniently obtain the complexity and model reliability of any given underground unit, and provide decision support for intelligent mining of thin coal seams.