Breaking the barrier of drilling automation: measuring the distance drilled at the bottom of the well using an imaging while drilling tool

Abstract

Drilling technology has advanced rapidly over the past few decades. The petroleum industry is motivated to automate directional drilling operations, making them safer, more cost- effective, and efficient. The main input parameters fed to directional drilling tools (e.g., rotary steerable systems) are inclination, azimuth, and distance. Inclination and azimuth are measured downhole; however, the distance drilled is computed using a pipe tally system installed on the surface. Having continuous downhole measurements of distance, inclination, and azimuth allow the directional drilling tool to determine the drilling bit's location. Moreover, if most calculations were made downhole, then the bit could follow a preprogrammed trajectory and drill autonomously.

Many researchers have proposed different methods for measuring the distance drilled in the bottomhole as a solution for directional drilling automation. However, the suggested approaches have experienced challenges during field applications. For this reason, this dissertation focuses on solving one of the main limiting factors, i.e., real-time downhole measurements of the distance drilled.This work proposes a method that senses the distance drilled at the bottom of the well based on automated image analysis. The dissertation involves the development of a tool prototype with multiple identical imaging sensors spaced at known distances. The sensors acquire images of the formation at synchronized times, with the imaging sensors capturing an image of the same formation location at different times. Based on this concept, an algorithm was developed to identify the "fingerprints" of the images captured and register similar images according to those fingerprints. Under wellbore conditions, each fingerprint reveals unique marks left by the bit, natural geological features of the rocks drilled, and the topology of the borehole wall, all of which contribute to the image registration process. Therefore, when two images captured by two sensors match, the timestamps of each image and the distance between the sensors can be used to calculate the average rate of penetration in that time interval. Subsequently, integrating the rate of penetration results in an estimate of the interval distance drilled. Adding all intervals drilled is then equivalent to the total distance of the well, and combining this information with azimuth and inclination can provide an estimate of the depth drilled. This dissertation aims to demonstrate that an image-matching process is an attractive candidate for downhole measurements that compute the distance drilled.

In this work, an image-matching algorithm is developed with the sole intention of estimating the distance drilled. Tests were designed to analyze the effect of different variables on the accuracy of the distance calculation. Experiments were devised to investigate how sampling criteria, the distance between the sensors, velocity, matches missed, and noise affect the method's efficacy. The method was tested in homogeneous and heterogeneous scenarios (i.e., images with distinct and repetitive features) to ensure that the algorithm can match images when exposed to analog scenarios that model wellbore conditions. The results show that the image-matching algorithm can accurately match images even when wellbore features are considerably repetitive. Additionally, having an optimized velocity-dependent combination of sampling criterion and distance between the sensors can minimize errors, especially those associated with missed matches, and therefore, increase the accuracy of the distance calculation.

The main contribution of this Ph.D. dissertation is that it demonstrates that having identical sensors capture images showing topological features can be a robust method to measure the distance traveled. Additionally, this work provides initial research on how to optimize controllable variables to minimize errors. The results suggest that the proposed method has the potential to automate directional drilling and continuously determine the well path when combined with inclination and azimuth information.