Study On The Data-intensive Numerical Modeling Method Of Strata Movement

In the context of big data and precision coal mining,fully utilizing geological borehole logs and mechanical test data of rock specimens to achieve precise analysis and control of strata movement is an important trend.It has important practicical siginificance for conducting the concept of precision coal mining and also has great theoretical and engineering significance for improving the reliability and accuracy of mining subsidence prediction.At present,numerical modeling of large-scale rock masses often ignores the undulations of beddings and the changes of rock lithology and its physical-mechanical properties along the bedding.For this reason,the intensive use of geological borehole logs and mechanical test data of rock specimens was taken as ideological guidance,and literature research,orthogonal expriments,theoretical analysis,and numerical simulations were used to discuss the claasification of dataintensive degree(L0~L4)of strata movement,the computing framework of dataintensive numerical simulation,and other basic issues.The data-intensive 3DEC modeler was developed in this paper,and the L2 data-intenvive numerical modeling theory was also researched and verified from multiple angles.The main research findings,research results and research conclusions obtained in this paper are as follows:(1)The large-scale strata movement analysis model has an oversimplification of the geological prototype,and it is necessary to perform data-intensive numerical modeling.The type and volume of strata-related data are time-dependent,and different periods will show different characteristics of data utilization.According to the current data scale of geophysical exploration and physical-mechanical test of rock specimens,strata-data can be divided into five levels of L0 to L4,which can represent the characteristics of each level,and the current data density of strata movement numerical simulation is between L0 and L2.(2)The computing framework of data-intensive modeling can be composed of four parts: data layer,logical decision-making layer,software implementation layer,and computational mechanics operation layer;the L2 data-intensive numerical computing framework can be composed of four parts: L2 data-intensive overburden structure model,L2 data-intensive rock mass physical-mechanical parameter estimation,dataintensive numerical calculation pre-processing software,and ITASCA strata movement analysis solution.(3)The L2 data-intensive overburden structure model and the L2 data-intensive rock mass physical-mechanical parameter estimation method are constructed.The L2 data-intensive overburden structure is a "structure-structural plane" combination model formed by cutting the geological bodies from two directions,i.e.the horizontal NURBS surfaces and vertical planes.Among them,the horizontal surfaces are mainly sedimentary structural planes,and the vertical planes are the potential rock layer fractures;the L2 data-intensive rock mass physical-mechanical parameter estimation method is comprehensively obtained by using analogy analysis,statistical analysis,GSI index analysis,and orthogonal testing.(4)The data-intensive 3DEC modeler that is suitable for the L2 data-intensive overburden structure model,is developed,which can realize the automatic modeling of horizontal curved surfaces and vertical fractures,and can greatly improve L2 dataintensive 3DEC modeling efficiency.Bases on VB.NET computer language,the detailed algorithms of key rock layer module,NURBS curve module and 3DEC grid output module are given.(5)Under the FLAC3 D computing framework,the L2 data-intensive overburden structure model and the L2 data-intensive rock mass physical-mechanical parameter estimation method were proved to be reasonable and effective.The FLAC3 D simulation analysis,based on the 2201 panel of the Ying-Pan-Hao coal mine,shows that when compared with the CMM-based prediction without parameter calibration,the DSMM-based prediction has obvious advantages: 1)the maximum subsidence relative error is reduce by 93.7% on average;2)the root mean square error of 70 points is decreased by 39.0%.After parameter calibration,the DSMM-based prediction still has obviuous advantages: 1)1)the maximum subsidence relative error is reduce by 63.3% on average;2)the root mean square error of 70 points is decreased by 13.8%.(6)Under the 3DEC computing framework,the L2 data-intensive overburden structure model and the L2 data-intensive rock mass physical-mechanical parameter estimation method were once again proved to be reasonable and effective.The 3DEC simulation analysis,based on the 2201 panel of the Ying-Pan-Hao coal mine,shows that: 1)increasing the data-intensive level can significantly improve the prediction accuracy of the maximum surface subsidence value and the overburden fracture development height;2)the spacing of the vertical structural surface has a significant impact on the 3DEC prediction results,and the vertical structural surface spacing cannot be simply equated to the thickness of rock layers,and 3)the determination method for the vertical structural surface spacing in the L2 data-intensive overburden structure model can significantly improve the accuracy of 3DEC prediction.This thesis includes 76 Figures,24 Tables,and 224 References.