Characterization of Defects in Coiled Tubing From Magnetic-Flux-Leakage Data

Abstract

This article, written by Assistant Technology Editor Karen Bybee, contains highlights of paper SPE 100121, "Observations on Characterization of Defects in Coiled Tubing From Magnetic-Flux-Leakage Data," by T.R. McJunkin, K.S. Miller, and C.R. Tolle, Idaho Natl. Laboratory, prepared for the 2006 SPE/ICoTA Coiled Tubing and Well Intervention Conference and Exhibition, The Woodlands, Texas, 4-5 April. The full-length paper presents observations on use of magnetic-flux-leakage (MFL) data to detect flaws in coiled-tubing (CT) strings. Sixty-six flaws of various shapes and types, ranging from 0.3-mm-deep pits to 9.5-mm-long slots, artificially created in 44.45-mm-outside-diameter pipe, were analyzed. The detection algorithm and the information extracted from the data are described. Introduction CT defects reduce CT life and must be detected to prevent costly failures in the field. One method commonly used for CT nondestructive inspection is MFL, where a magnetic field is generated with a coil or permanent magnets and Hall-effect sensors positioned around the CT measure the variation in the field near the CT surface. Flaw characterization is important for failure-prediction models. The size and shape of a flaw influence MFL-signal amplitude and duration and are directly related to expected CT-string life. The full-length paper describes an effective method of detecting flaw signatures in an MFL signal. Equipment and Processes A laboratory-scale experimental CT MFL-sensor head was constructed. The coil used to create the field in and around the CT was an 89-mm-long solenoid with a 102-mm inside diameter. One-thousand four-hundred turns of 18-gauge copper wire were wound evenly along the length of the coil. Data presented in the full-length paper were acquired by use of a 5-A direct current in the coil. This current was sufficient to saturate the pipe and permit flaw detection on the inner surface of the CT. Plastic frames, or "shoes," were populated with three ratiometric linear Hall-effect sensors oriented orthogonally in cylindrical coordinates (i.e., radial, longitudinal, and circumferential). Five shoes were distributed circumferentially around approximately one-fourth of the circumference of the CT samples. Artificially created defects were moved past the sensors, with the defect centered approximately on the middle of the five shoes. Data from the Hall-effect sensors were acquired at a 2000-samples/s rate. A first-order low-pass Butterworth filter with a 100-Hz cutoff frequency was used as a digital filter. The time-series signal was converted into distance vs. amplitude by adjusting for scan velocity and sample rate.