| Abstract |
There have been significant improvements in drilling performance in petroleum operations. For example, wells to 14,000 ft in Wyoming from spud to rig release have gone from 60 days to 6 days. While there were many ways in which this was accomplished, systematic analysis of the drilling data both from surface and downhole sensors is a key process. Among the many things to analyze, large gains in drilling efficiency result from ensuring that the energy applied in the drilling system is almost entirely utilized in destroying the rock, rather being lost through various forms of drilling dysfunction such as bit bounce, BHA whirl, etc. One of the greatest challenges in reducing these forms of drilling dysfunction arises from the fact that different lithologies can drastically change how the drillstring responds with a given set of drilling parameters, making it difficult to predict what drilling parameters are optimal in reducing these dysfunctions. This paper describes the use of high frequency data acquisition and analysis to relate the lithology quantitatively to surface drilling operational measurements that characterize drilling performance. A standard mining coring rig outfitted with a diamond impregnated core bit was used to collect core samples at the Colorado School of Mine’s Edgar Experimental Mine. The formations cored were composed of hard igneous and metamorphic rocks as are often encountered during geothermal operations. The rig was outfitted with high frequency (20 kHz) vibration sensors during the operation along with force, torque, rotation, and position sensors. Data analysis in the frequency and time domains, showed a clear connection to the drilling performance and the lithology as shown by the recovered cores. The accelerometer high frequency data was also synchronized with core location. An image analysis of cores was combined with the accelerometer data to correlate lithology with drilling performance. A machine learning clustering function was used to estimate the percent composition of a given rock matrix component. This showed confirmation of drilling observational data that indicated lighter cuttings return was a marker of efficient performance and dark cuttings return was a marker of inefficient performance. While more work will be accomplished, the combination of different types of high frequency data (high definition images, accelerometers, and other drilling performance data) can be used to optimize operational parameters with a subsequent improvement in drilling performance. |