Data-driven surface reconstruction for assessing welding-induced distortions in ship-deck panels
Year: 2025
Type: article
Abstract: Increasingly stringent regulations on greenhouse gas emissions have sparked interest in lightweight, thin-walled deck panels for cruise ship superstructures. Throughout the ship assembly, irregular welding-induced distortions accumulate on thin plates, affecting their load-carrying capacity significantly. Due to complex geometries, the assessment of distorted thin-walled panel units relies on accurate finite element modeling based on high-density optical scanning measurements, provided as unorganized point clouds (PC). This poses challenges in terms of processing of the data for surface reconstruction and data storage. This paper presents a computationally efficient, iterative B-spline surface fitting procedure for surface reconstruction from an unorganized 3D PC for the finite element analysis of distorted thin-deck panels. The method achieves a minimum geometric reconstruction accuracy of 0.08 mm and structural strain predictions mostly within 89% accuracy, validated against uni-axial monotonic tensile experiments on full-scale panels. Additionally, the proposed procedure reduces the file size to less than 1% of the original, thus representing a valuable solution for the handling of large sets of data from the scanning of multiple decks in ship superstructures. These results highlight the method’s potential for improving the assessment and quality control of thin-walled ship superstructures. • Reconstruction algorithms are evaluated for geometry preparation of distorted panels. • B-spline surface fitting is validated for real distortions under tensile loading. • FEA with reduced data shows excellent agreement with full-scale panel experiments. • The proposed reconstruction yields excellent accuracy with reduction in data size.
Source: Marine Structures
Language: English
Subfield: Computational Mechanics
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $4,320
Funder: Business Finland
OpenAlex ID: W4416415241