Prediction of drill-bit performance using multi-variable linear regression analysis

The principal factors which require consideration in predicting drilling rates are the bit-type, the bit operating parameters and the characteristics of the penetrated rock. A single index is not a sufficient basis for predicting performance. A variety of rock types were drilled in the laboratory with polycrystalline diamond compact (PDC) and impregnated diamond core bits using a fully instrumented drilling rig at different rotational speeds and with a range of weights on the bit. Quantitative determinations of a wide range of textural, mechanical and intact rock properties were also made. Multi-variable linear regression analysis was then used to develop predictive models of rates of penetration from the operating and rock property data. The models have been verified by statistical tests. The analysis indicated that the models for the pin and hybrid PDC bits have potential for practical application. However, although the model for the impregnated diamond bit appears to be reasonable from a purely statistical point of view, it may not be satisfactory for drilling uses as it is based on two measures of rock hardness alone. The application of multiple regression analysis to PDC drilling performance is novel and the technique shows promise for the prediction of rock drillability with PDC bits. The ability of the technique to provide a mathematical characterisation of the performance of new drill bits in specific rocks may prove to be of value to drill manufacturers and users