Point clouds from laser scans or photogrammetry are a valuable source of data for the generation of digital twins as the data captures the current actual state. This work focuses on the reconstruction of pipelines in point cloud data, as they are main object types in industrial infrastructures. Therefore, a synthetic point cloud dataset of 50 pipelines from straight parts, individual pipe bends or pipework of higher complexity is generated using Unreal Engine and Colosseum. For the individual pipe objects a skeletonization algorithm is used to determine the spline route. Furthermore, the pipe diameter along this route is determined using 2D projection and a RANSAC approach for circle fitting. After reconstruction the overall Intersection over Union of the pipe volumes is up to 76 %. Given the route and diameter as main properties of pipes, geometric digital twin models can be generated.