EFFICIENT NUMERICAL MODELING OF FAULTED ROCK USING THE BOUNDARY-ELEMENT METHOD

This paper presents an efficient method of modelling faulted rock usin g the multiregion boundary element method. The boundary element method is ideally suited for modelling problems in rock mechanics because of the requirement to descretize surfaces only. This contrasts with the volume discretization needed for the finite element method, finite dif ference method and the distinct element method. The disadvantage of th e boundary element method is that the rock mass is assumed to be isotr opic and elastic. However, it is possible to model the non-linear beha viour on fault planes by using the concept of multiple regions and by connecting these regions via non-linear springs or Goodman type joint elements. The disadvantage of this method has been that joint stiffnes ses have to be specified and that a great number of iterations are nee ded because the stiffness coefficient of a spring may change very rapi dly from a very high value (for example when a joint is in compression ) to a very low value (when a joint is in tension). This paper describ es a novel method for the treatment of faults with the boundary elemen t method which eliminates the need to specify joint stiffness paramete rs and which is able to model significant fault movement with very few iterations. The numerical scheme makes very efficient use of computer resources because the non-linear iterations only involve degrees of f reedom at the interfaces. Examples of applications of the new model to practical problems in mining and geological engineering are given.