Acoustic Emission Data-Driven Modeling Of Rock Failure Process And Its Application

The depletion of shallow and easy-exploited ore reserves is forcing human to exploit mineral resources at ever-increasing depths.Deep open pit mining,mining transition from open pit to underground,deep underground mining have become the development trend of mining,which brings more frequently geological hazards,e.g.,landslide,collapse,rockburst,etc.It makes mineral exploitation a challenging task.In essence,the occurrence of any geological hazard is a result of rock damage and failure.Therefore,it is of great significance for safety production to study the rock failure and associated hazard mechanism,seek the precursors before rock failure and make a corresponding forecast.Both acoustic emission(AE)monitoring and numerical modelling are widely used in the research of rock damage and failure.However,only by use of AE monitoring,it cannot provide a reasonable explanation for the rock damage from the view of mechanics.And only by use of numerical modelling,it lacks of reliability and verification on the model parameters.Hence,we propose a rock failure process analysis method by integrating the AE monitoring and numerical modelling.In this method,taking AE physical data as the basis,characterization relationship of rupture source energy and rock damage as the bridge,and AE inversed damage as the input,an AE data-driven model is developed and applied to simulate and predict the rock damage and failure in laboratory and field scales.The main research contents and conclusions can be summarized as follows:(1)The tensile strength and fracture toughness are studied using notched three point bending tests.Compared with the Brazilian disc test,the tensile strengths of sandstone from notched three point bending are more reliable.The differences of three popular formulas for fracture toughness are analyzed via theoretical analysis,laboratory test and numerical modelling.The rock fracture toughness is determined.It provides reliable input parameters for subsequent numerical modelling.(2)AE monitoring is conducted on the yellow sandstone.The auto regression model and AIC method are applied to determine the P-wave arrival time.The AE event location is calculated with the Simplex algorithm.The results of onset time picking and AE location are much well.The AE event locations are in good agreement with the ultimate fractures of specimens.The influences of loading condition and failure pattern on the AE event rate,b value,and fractal dimension value are studied.Multi-parameter conjoint analysis would improve the reliability of rock failure prediction.(3)The rock failure mechanism is discussed and the rationality of tensile failure model is argued.An AE data-driven damage model is developed based on tensile assumption.The quantitative relationship between AE energy and rock damage is built via this tensile model.The location,size and degree of rock damage is characterized,respectively,using the AE data.The quantification of rock damage during the whole loading process is achieved.All the AE inversed damage radii are less than 1 mm and the inversed damage degrees are close to 1.(4)The validity of rock dynamic analysis under the quasi-static loading is testified,and the selection principle of time step is studied.The relationship between AE monitoring and numerical modelling is established in time dimension through the dynamic analysis,which lays the foundation of interactive modelling integrating AE and numerical computation.The inversed damage is input as the initial and current conditions for the numerical model,and the AE data-driven simulation of rock failure process is implemented.The Brazilian disc and uniaxial compression tests were used to validate the model.The rock failure pattern can be predicted with this AE data-driven model when the stress level of AE data-driven damage inversion is 0?0.6?c.(5)A microseismic data-driven model for rock mass damage is proposed on the basis of consideration of joint and groundwater seepage effects.Using data from the Shirengou iron mine,joints are used to reduce the global properties of rock mass,water is used to reduce the local properties of rock mass,and microseismicity are used to reduce the point properties of the rock mass.The influences of multiple factors on the properties of rock mass are considered collectively.The coupled fluid-solid model considering the effects of joints and water successfully simulates the rock mass damage evolution which is in good agreement with the microseismic monitoring result.Further modification of rock mass properties using microseismic moment tensor makes the model better resemble the actual situation,which is conducive to the modification of numerical result and prediction of rock mass damage development.