Implementation of Modelling Into the Mine Design Process

The only way numerical modelling will be fully implemented into the mine design process is if evidence of correlation between modelling results and rock mass response is demonstrated through case studies. To be of real value, the results from the numerical model must be directly related to rock mass response. This is best approached from three distinct points of view:

•parameters that are relevant to mine operators;

•physical aspects of rock mass behaviour; and,

•results from numerical modelling.

Rock response related characteristics, which have an important influence on mine design decisions must be identified and specified from an operating point of view. It is necessary to specify what parameters are of critical importance in decision making, and how these parameters actually come into play in the mine design process. Although it will be fairly easy to list these parameters, it will be much more difficult to define exactly how to use this information. Some examples could be:

•level of ground stability,

•stand-up time,

•risk of ore loss,

•level of ground support required.

The rock mass response must be characterized in terms of parameters, which can be directly measured underground, for example:

•level of joint alteration,

•displacement,

•strain,

•closure,

•stress change,

•but also in terms of parameters which are more subjective such as:

•ground support integrity,

•level of stability or amount of ground deterioration.

The rock mass at each mine site of interest must be characterized by use of scan line surveys and point load strength tests. Additional information in the form of displacements and strain measurements (extensometers and closure points) as well as stress change monitoring (CSIRO cells or IRAD gauges) should also be used where appropriate.

This work is very important because it provides a repeatable, objective measure of current ground conditions. By continuing to monitor the rock mass conditions in this way, a record of when and where conditions change is made. This not only permits one to compare different sites to one another, but also to develop a direct method to relate in situ ground conditions to measurable, physical parameters.

The specific numerical modelling results, which can be used to relate directly to the rock mass response, are strain, displacement, stress, safety factor and size of failed zone. In order for these predictions to be reliable, it is necessary that extensive calibration and back-fitting of input parameters (far field stress, mechanical rock properties, and rock strength properties) be conducted. This work can only proceed when the physical aspects of rock mass behaviour (discussed above), are characterized adequately.

The three items: response characteristics specified by the operators, the observed rock mass response, and results obtained from numerical modelling, must all be correlated to provide reliable predictive capability. Concrete relationships between predictions, measurements, and the parameters that are relevant to the mine operators must be developed. One must be able to reliably correlate for example, the amount of joint alteration with the level of ground stability.

Demonstrations will be necessary so that operators can get a feel for how reliable (or unreliable) the predictions are, and how seriously they should take these predictions. Several different methods should be pursued simultaneously so that over time, the most important and reliable indicators can be identified and verified. Note that the reliability will change with time as rock mass response characteristics are collected.

It is very important that the far field stress state be established with some confidence. The orientation can often be determined by observation of ground response (i.e. fault slip direction, micro-seismic event locations, ground stability dependence on orientation, dominant closure direction, spalling etc.). An active effort to compile this form of evidence should be undertaken. Direct measurements by over-coring or some other technique are often necessary (Wiles and Kaiser, 1994).

The most important deformation properties to be concerned with are the stiffness of the rock mass, and its inelastic response during yielding. These parameters are the key to making accurate displacement and strain predictions. A best guess estimate for stiffness can be made by use of rock mass rating (Hoek and Brown, 1988). However, it must be recognised that this is only an estimate, and it is necessary to further fine-tune the model.

It is important to distinguish between the actual elastic portion of the ground response, and the total response, which is a combination of the elastic deformation and non-linear yielding effects such as joint alteration. At a large distance from openings, one expects ground deformations to be primarily elastic, where-as near the excavation surfaces, or in yielding pillars, one expects a large part of the deformation to be inelastic, and highly dilational in nature.

Although the rock mass strength can be estimated by use of a rock mass rating (Hoek and Brown, 1988), laboratory compressive tests and point load test results are also very useful in this characterisation. Just as for stiffness however, it is necessary to fine-tune the strength parameters to the actual in situ response. This can only be achieved by use of parametric studies where one attempts model calibrations by trial and error for numerous different cases. With time, sufficient examples will be compiled to define a strength envelope, which is truly representative of the rock mass under study.

It should be mentioned here that both the far field stress and mechanical properties will influence model results, therefore, before any parametric studies are conducted, the best possible information must first be obtained for the former parameters. As better far field stress information and mechanical property information becomes available, strength parameters will have to be re-evaluated for accuracy, and adjusted to suit the new information.

This discussion is by no means complete, however it is presented here to give users a start on developing their own scheme for using modelling. In summary:

•Efforts should be made to clearly define the objectives of the modelling along the lines of providing operations with reliable ground control and ground stability predictions. All efforts should ultimately be directed towards achieving this objective. A plan should be put in place on how these objectives are to be achieved with the resources available.

•A ground stability reporting system should be established to systematically observe, document and report on ground control related problems. This should include a discussion on how the observed ground control problem could (or could not) be identified with instrumentation, numerical modelling etc., and controlled by use of modified sequence, layout or ground support. The underlying mechanism or cause should also be identified where possible. Implications with regards to important parameters defined by operations should be addressed, e.g. level of ground stability, stand-up time, risk of ore loss, level of ground support required etc. This reporting system can be used as a forum for communication and education on the use of instrumentation, numerical modelling and rock mechanics principles as applied to mine stability.

•When modelling, one should always test for very specific problems and conditions. Through a process of back-analysis, one can identify what stress states or stress paths lead to know problems underground. Some ideas on how to do this are presented in Figure 3. After many back-analyzes have been completed, one will have a good feel for the confidence that can be placed in the predictions. This confidence level may be very good or very poor, but it must be established. Once these have been identified, one can then attempt to make predictions with know reliability.

•Some form of routine monitoring should be implemented (e.g. visual, scan-line, closure, ground movement monitors, extensometers and stope outlines). Scan-lines can be repeated at the same location as ground conditions deteriorate as a quantitative log of ground response. These items can be directly related to mining problems and numerical modelling results.