A Stereoscopic Endoscopic Optical System for Measurement of the 3D Weld Geometry of Pipes: Concepts and Preliminary Results

Steel catenary risers (SCRs) have been considered a strong candidate in future sub sea developments at Campos Basin (Brazil). So there is an enormous effort to guarantee the structural integrity along its lifetime. In this context, inspection activities become significantly important in the manufacturing process. The welding is considered a critical process so it is the inspection focus. Studies and analysis have demonstrated that fatigue performance of one sided girth weld is considered a critical issue, mainly in the touch down point (TDP) area. In this context, control the girth weld misalignment would improve fatigue performance during fabrication. This paper presents an optical system developed to inspect the root region of the weld during fabrication. A pair of stereoscopic images is obtained when two images of the same scene are taken from slightly different viewing angles. Both images can be combined to give information about the 3D nature of the scene. This is naturally done by the brain of most animals and humans. It is also already a very well established computer vision technique to extract 3D information from objects. The authors developed an alternative endoscopic configuration to extract and analyze stereoscopic images from the inside of pipes is presented. Two cameras are coupled with conical optics to produce a pair of endoscopic stereoscopic images of a 360° measurement ring. A light projector is used to illuminate the inner surface and, alternatively, to generate a random pattern of dots that is projected into the inner surface of a pipe to make it easier to correlate the images. Both images are analyzed by cross-correlation technique. The corresponding position of the spots in each image are found and combined to compute the 3D coordinates of points laying on the inner surface of the pipe. As a result, a dense cloud of points of the inner surface is obtained and used to extract geometrical features of the pipe. The software is capable to represent the measured surface from different angles and through different cuts. A prototype using this principle was built and tested in laboratory. Some practical considerations and early measurement results are here presented. The aim is to measure and quantify the misalignment between the pipes and to detect common welding defects like lack and excess of penetration.