Study on the Developmental Characteristics of Water-Conducting Fracture Zones in Multi-Slice Mining of Thick Coal Seams within the Gondwana Stratigraphy
Authors:
Xuwei Mao, Liang Li, Jian Cao, Xueyi Yu, Lidong Ma, Zhiqiang Zhao
Keywords:
Gondwana Strata; Height of the Water-conducting Fracture Zone; Leakage Observation; BP Neural Networkt.
Doi:
10.70114/acmsr.2026.6.1.P161
Abstract
The geological strata of the Barapukuria Coal Mine in Bangladesh are affiliated with the Gondwana Formation. To ensure safe future mining of thick coal seams under aquifers, the development law of the Water-Flowing Fractured Zone (WFZ) during slice extraction was investigated.The development height of the WFZ was predicted via numerical simulation.. Field measurements of the WFZ height were subsequently conducted for each mining slice, and regression analysis was performed to establish a mine-specific predictive formula.Analysis of extensive case studies identified mining depth, overburden lithology, dip angle, mining method, and extraction thickness as the primary factors influencing WFZ development. A predictive model for the WFZ based on a Back-Propagation (BP) neural network was developed using MATLAB R2014a. The model was subsequently trained and tested, and then applied to predict the WFZ height in the mine's panels. These predictions were compared against the empirical formula from the “Under-Three” Regulations, the site-specific regression formula, and actual field measurements.The results demonstrated that the WFZ height increased approximately linearly with the cumulative extraction thickness after the first and second slices. During the third mining slice, the rate of WFZ height increase was significantly reduced. Among all predictive models evaluated, the BP neural network-based model achieved the highest accuracy. Finally, the predicted WFZ height for the third mining slice was presented.