Computational schemes for analysis of rock mass response to excavation, loading and other imposed changes, are employed pervasively in rock mechanics practice. Applications range in complexity from determination of stress and displacement distributions around openings, to prediction of the thermo-hydromechanical behaviour of a saturated fissured mass. While this may be taken as apparent sophistication in rock mechanics design activity, the indiscriminate use of computer methods may sometimes conceal inferior and inadequate engineering procedures. These inadequacies may not always be recognized, due to the general uncertainties which are inherent in rock mechanics investigations. This paper seeks to identify some common deficiencies in rock mechanics applications of computational schemes, and to propose ways of eliminating them. It also indicates some specific developments and applications of computational procedures, which are required to address persistent design problems in geomechanics practice.
Rock mechanics has evolved rapidly as an engineering discipline in the last twenty-five years. To some extent, this has been a response to the needs posed by the increasing physical scale and complexity of surface and underground rock structures, and the financial risks associated with these engineering ventures. Notable examples of such motivating ventures are large gravity and arch dam projects, deep level mining, hazardous waste isolation and comprehensive hydroelectric and stream diversion projects. In these and other cases, improved engineering practice has been based on sound scientific investigation and greater understanding of the fundamental processes controlling the load-deformation behaviour of rock masses. The emergence of rock mechanics as a coherent body of engineering science has been virtually contemporaneous with the widespread application of computers in scientific investigation and routine engineering activity. It is not surprising, therefore, that the techniques of numerical analysis are firmly embedded in design aspects of engineering rock mechanics. Certainly, the stage has been reached where a rock mechanics engineer, intimately involved in excavation design practice, can play a very limited role if not computationally literate. This implies, as a minimum, an understanding of the engineering principles exploited in various computational schemes, the valid range of conditions under which a particular scheme might be applied, and possible numerical difficulties that might be associated with particular algorithms, whether related to general numerical instability or machine dependence of the source code. Even though the engineer must understand the principles and limitations of the computational tools applied, this need not be to a level expected of a specialist in computational mechanics. It must, however, be sufficient to provide an informed and rational basis for defining computational activities required to support any particular investigation, and for interpretation of the meaning and significance of any analytical results. Definition of appropriate computational activities requires assessments of the engineering significance of the work to be done and the quality of the engineering data available to support any analytical investigations. These assessments can only be made by a rock mechanics engineer since a specialist in computational mechanics will seldom possess sufficient understanding of the problem or physical insight to make the necessary judgements.