SBIR/STTR Award attributes
It is critical to characterize subsurface formations for energy resource exploration and production. Radioisotope-based well logging devices have been widely used to measure the physical properties of subsurface rocks such as density, porosity, mineralogy, hydrogen content, etc. Those devices, however, pose radiological security risks, thus, there is an essential need to explore replacement technologies to determine these subsurface parameters without using high-activity radioisotope sources. To this end, we propose to develop an alternative methodology to produce well logs equivalent to gamma ray (GR) and neutron logs from the high-resolution drill-bit vibration data that can be acquired at no risk nor significant cost along the entire well length. Our proposed technology benefits from advanced signal processing and a powerful physics-informed machine learning model to extract the targeted physical properties from bit-vibration signals. The proposed method is based on the concept that a large volume of data is continuously collected by the drill-bit as the first logging tool touching the rock formations while drilling. These data include information about the physical and mechanical rock properties, fluid properties, in-situ stresses, and mineralogy. This information is embedded in a series of high-resolution high-scanning rate vibration signals generated by the rock-bit interaction and of course drilling system configuration. With a good understanding of bit-rock interaction physics and by means of an efficient signal processing approach, it is possible to segregate the signal and extract different information out. In this project, we will develop a signal processing workflow and a physics-informed machine learning (PhiML) model based on bit-rock interaction physics. We will also conduct experimental research at the Drilling Vibration Laboratory of the Oklahoma University to better understand the complex bit-rock interaction and perform sensitivity analysis on the effective parameters. The Phase I outcome will be a combined signal processing and machine learning workflow that determines density, porosity, and clay mineral/organic matter content along the entire well length. In Phase II, we propose to further develop and package the methodology as a product, conduct field testing in different geological formations settings, and commercialize the technology.
