Publications Details

Machine learning methods for estimating down-hole depth of cut

Sacks, Jacob; Choi, Kevin; Bruss, Kathryn; Su, Jiann-Cherng S.; Buerger, Stephen B.; Mazumdar, Anirban; Boots, Byron

Depth of cut (DOC) refers to the depth a bit penetrates into the rock during drilling. This is an important quantity for estimating drilling performance. In general, DOC is determined by dividing the rate of penetration (ROP) by the rotational speed. Surface based sensors at the top of the drill string are used to determine both ROP and rotational speed. However, ROP measurements using top-hole sensors are noisy and often require taking a derivative. Filtering reduces the update rate, and both top-hole linear and angular velocity can be delayed relative to downhole behavior. In this work, we describe recent progress towards estimating ROP and DOC using down-hole sensing. We assume downhole measurements of torque, weight-on-bit (WOB), and rotational speed and anticipate that these measurements are physically realizable. Our hypothesis is that these measurements can provide more rapid and accurate measures of drilling performance. We examine a range of machine learning techniques for estimating ROP and DOC based on this local sensing paradigm. We show how machine learning can provide rapid and accurate performance when evaluated on experimental data taken from Sandia's Hard Rock Drilling Facility. These results have the potential to enable better drilling assessment, improved control, and extended component life-times.