Probabilistic Casing Collar Locator for Untethered Downhole Tools

The residual magnetic field amplitude shows significant variations along the length of the casing joints and usually a sharper variation occurs around the collars. These can be detected with magnetometers instead of conventional casing collar locators (CCL). However, magnetic signals within the casing joints can mask the signals around the collars. Here we test a probabilistic method that detects casing collars using hydrostatic pressure and the residual magnetic fields of casings. The method uses hydrostatic pressure measurement along with a Kalman filter to measure depth. If a magnetic field peak is observed, the algorithm compares the measured depth of the detected peak with the expected depth of a collar from the casing tally. If the two depths are relatively close, higher confidence is assigned to the detected collar location. Low confidence peaks are discarded. The algorithm was tested using data collected from a slickline tool with reference depth. Depth calculation using hydrostatic pressure is sensitive to measurement accuracy and density variations. Therefore, there is an error margin in the calculation. However, the erroneous depth measurement provides a window to evaluate magnetic field signal around expected collar depths that are known from the casing tally. High pass filtering and thresholding the windowed magnetic field, resulted in several potential peaks that could belong to a collar. Depending on the threshold level applied, several false positives or false negatives were found. Comparing the detected peak depth to the expected collar depth provided a Bayesian probability for the detection. Once a collar was detected with high probability, it was used as a reference depth for the subsequent collars. This limits the depth window size to evaluate magnetic fields. The algorithm identified 21 out of 26 collars within 1150 ft of depth in a steel cased test well filled with water. The calculated depth from the pressure was corrected at each detected collar. Compared to the reference depth from the slickline tool, the calculated depth stayed within 1 ft, even with 10% error margin in the depth calculation using hydrostatic pressure. The introduced method enables the miniaturization of the CCLs and paves the way for autonomous untethered downhole tools.