The facies and core data prediction techniques have not yet found their way in the CBM/CSG business, despite having been used successfully for more than a decade in the conventional O/G business; but we believe this new technique signals that the time has come for this to happen.
The method described here is based on the Facimage facies-prediction software run under Geolog. Reasonable qualitative estimates can be generated immediately from a full suite of logs but for each new basin or coal sequence, gas and ash content from cores are required to derive a predictive model and deliver calibrated, quantitative results - by tying core measurements to their log signature, as is done for conventional O/G analysis. Once blind testing has proven the model is satisfactory, gas and ash content together with the different facies can be predicted quantitatively from logs without cores.
Rigorous quality control measures ensure that the prediction models used are applicable or, where there is a difference, that there may/must be changes in the geology or a log acquisition problem. The processes used are designed to be devoid of operator bias and can be run as soon as logs are run to allow on-site decisions about testing, suspension or P&A. We demonstrate the principles and methods used to derive gas and ash content without core data, but assuming calibration has been achieved, and how we build the quality control model and how we address the problems of spatial resolution mismatch between core and logs and of shoulder-bed effects.
We present an example from an open file well in the Galilee Basin (Queensland) of Gas and Ash content prediction using a basic set of wireline logs as predictors.
The core of the method is algorithms designed to create facies models (MRGC-CFSOM) from any given set of logs and predict core measurements (k-NN) over the whole logged interval of any well, even un-cored wells.