Testing models for bottom-of-hole temperature recovery, Cooper Basin, South Australia

Abstract

A set of temperature data from 61 deep petroleum wells drilled in the Cooper Basin of South Australia has been used to test the accuracy of four commonly used models of borehole thermal re-equilibration. Models tested include the Horner plot as derived from Bullard (1947), the theoretical dual-media zero circulation cylindrical model of Cooper and Jones (1959), the empirical semi-log plot of Pitt (1986) and the exponential model of Nakaya (1953). The selected models have been fitted to bottom-of-hole temperature data recorded during the recovery period of each well, and are used to predict a value for True Formation Temperature (TFT). These values are then compared with actual temperature measurements derived from Cement Bond Logs (CBL). Recorded on average 482 days after the end of drilling, these temperature measurements are likely to represent TFT. Analysis of the accuracy of model prediction relative to CBL temperature suggests that on average most models tend to under­estimate TFT to some degree. The magnitude of this bias is found to be dependent upon assumptions implicit in each model. In most cases, model prediction is improved where at least one perturbed bottom hole temperature (BHT) has been recorded = 20 hours after the end of drilling. To date the best results are derived from the semi-log plot extrapolated to 50 hours.