Source-linked AI summary
Integrated Laser Scanning and Image-Based Topology Optimization Techniques for Detection and Quantification of Visible and Subsurface Structural Defects
Mehrdad Shafiei Dizaji, Devin Harris
TL;DR
The paper addresses characterization of both visible surface defects and non-visible structural abnormalities. It combines direct laser-scanning geometry with 3D-DIC, finite element updating, and topology optimization, finding complementary quantitative performance across controlled and randomly distributed damage.
Problem
Irregular structural defects are difficult to represent accurately, while some abnormalities are not directly visible from the inspected surface.
Method
The framework combines high-resolution laser scanning and point-cloud comparison for visible damage with 3D-DIC, finite element model updating, and topology optimization for inferring subsurface abnormalities.
Results
Both approaches identified and quantified defect geometry, with laser scanning achieving 3.6% mean absolute percentage deviation and topology optimization 5.0% for randomly distributed defects.
Takeaways & Limitations
Laser scanning and topology optimization provide complementary information for non-contact assessment of visible and partially hidden structural damage.
Takeaways & Limitations
Quantitative reconstruction remains sensitive to defect scale, measurement resolution, mesh density, post-processing, material assumptions, boundary conditions, noise, and regularization.
Abstract
from arXiv · showhide
Reliable characterization of structural defects requires methods capable of resolving both directly observable surface damage and damage that is not visible from the inspected surface. This study presents two complementary non-contact, vision-based approaches for the detection and quantitative characterization of defects in structural components. The first approach employs high-resolution laser scanning to generate three-dimensional (3D) point clouds of damaged steel specimens. Comparative processing of measured and reference point clouds is used to localize damaged regions, quantify geometric loss, and transfer the measured defect geometry to a finite element representation. The second approach combines full-field surface deformation measurements obtained using three-dimensional digital image correlation (3D-DIC) with finite element model updating and topology optimization. In this inverse framework, measured surface response is used to infer subsurface abnormalities through their influence on the spatial distribution of structural response. Experimental steel-beam specimens containing controlled smooth defects and randomly distributed defects are used to evaluate the approaches. Comparisons with milling-based ground-truth measurements demonstrate that both methods can identify and quantify defect geometry, while providing complementary information for visible and subsurface damage assessment. The combined framework establishes a pathway toward high-fidelity, non-contact structural condition assessment and model updating for components with complex and irregular damage.
1. INTRODUCTION AND METHODOLOGICAL FRAMEWORK
The study addresses irregular structural defects with two complementary, measurement-driven vision-based techniques: laser scanning for visible surface damage and 3D-DIC coupled with FEMU and topology optimization for subsurface abnormalities.
- Irregular defect geometries are difficult to represent with conventional idealized numerical models.Local geometry and material-property variations can significantly influence structural response.
- High-resolution laser scanning and 3D point-cloud processing target visible surface damage.The approach generates high-fidelity representations of measured surface defects.
- 3D-DIC, finite element model updating, and topology optimization infer non-visible abnormalities from their effects on structural response.The inverse framework uses measured surface deformation fields to identify subsurface damage.
2. LASER-SCANNING-BASED SURFACE DAMAGE CHARACTERIZATION
The laser-scanning approach directly measures steel-beam geometry, compares measured and reference point clouds, and transfers localized defect geometry into finite element models.
- High-resolution laser scanning directly measures 3D geometry to characterize surface defects and update numerical-model geometry.The measured geometry can support localized element removal or reduction, geometric imperfections, or connectivity loss.
- Measured and as-built or reference point clouds are compared to locate localized geometric deviations associated with damage.Computer-vision processing isolates and quantifies damaged point-cloud subsets rather than reconstructing the complete component surface.
- Surface-scan data collection is performed on a 4-ft steel beam containing randomly distributed defects on the rear surface.
- The experimental specimens include intact beams, beams with controlled smooth rear-surface defects, and beams with randomly distributed rear-surface defects.
3. 3D-DIC AND TOPOLOGY-OPTIMIZATION-BASED SUBSURFACE DAMAGE IDENTIFICATION
The subsurface-damage approach infers internal abnormalities from full-field surface deformation, combining 3D-DIC measurements with finite element response fields in an inverse topology-optimization framework.
- Internal abnormalities are inferred from their influence on measured surface deformation fields.Spatial perturbations in displacement and strain can provide indirect information about damage location, shape, and severity.
- The framework combines full-field 3D-DIC deformation measurements with analogous response fields predicted by a finite element model.These data are incorporated into an inverse structural-identification procedure formulated within topology optimization.
- Topology-optimization results are compared for steel beams with controlled smooth defects and randomly distributed defects.
4. RESULTS AND DISCUSSION
The experiments show that defect-volume accuracy depends on defect size and morphology: visual inspection was closest for controlled smooth defects, while laser scanning and topology optimization performed better for randomly distributed defects. The complementary approaches support direct characterization of accessible surface damage and response-based inference of less accessible abnormalities, while small-defect reconstruction remains sensitive to resolution and modeling choices.
- Controlled Smooth Defects: All three methods captured the overall magnitude of the three controlled smooth defects, but their agreement with milling CAD references varied by defect and method.The reference volumes were 2.88, 1.44, and 2.88 for Defects 1, 2, and 3, respectively.
- Controlled Smooth Defects: For Defects 1 and 3, visual inspection produced the closest numerical agreement, with absolute differences of approximately 4.2% and 5.9%, respectively.Laser scanning and topology optimization differed by approximately 8.3% and 9.4% for Defect 1, and 7.6% and 10.8% for Defect 3.
- Controlled Smooth Defects: For the smallest controlled defect, Defect 2, visual inspection differed by approximately 5.6%, compared with 22.9% for laser scanning and 37.5% for topology optimization.The reference volume was 1.44; modest absolute differences therefore produced comparatively large percentage errors.
- Controlled Smooth Defects: Across controlled smooth defects, mean absolute percentage deviations were approximately 5.2% for visual inspection, 13.0% for laser scanning, and 19.2% for topology optimization.Topology optimization nevertheless infers defects from their influence on measured structural response rather than solely from direct geometric observation.
- Randomly Distributed Defects: For randomly distributed defects, mean absolute percentage deviations were approximately 11.3% for visual inspection, 3.6% for laser scanning, and 5.0% for topology optimization.Laser scanning provided the highest overall quantitative accuracy for this specimen, with topology optimization following closely.
- Comparative Discussion: Performance depended on defect morphology and size: laser scanning’s dense point clouds better represented irregular three-dimensional boundaries than limited manual dimensions.This capability is relevant to naturally occurring deterioration with nonuniform boundaries and spatially varying depth.
- Comparative Discussion: Laser scanning and topology optimization are complementary approaches, respectively providing accessible surface geometry and response-based inverse identification.Topology optimization achieved an approximately 5.0% mean deviation for randomly distributed defects despite using indirect structural-response information.
- Comparative Discussion: Small-defect reconstruction and inverse identification remain sensitive to resolution, mesh density, post-processing, material assumptions, boundary conditions, noise, and regularization.Future studies should establish detection limits and uncertainty bounds for quantitative defect reconstruction.
5. CONCLUSIONS
The study demonstrates complementary non-contact approaches for characterizing visible surface defects and non-visible structural abnormalities. Their accuracy varies with defect morphology and method, supporting integrated measurement-driven finite element model updating.
- High-resolution laser scanning directly localizes surface damage, quantifies material loss, and transfers measured defect geometry into finite element models.
- The 3D-DIC/topology-optimization framework infers abnormalities that may not be directly observable from the inspected surface using full-field deformation measurements.
- For controlled smooth defects, mean absolute deviations were approximately 5.2% for visual inspection, 13.0% for laser scanning, and 19.2% for topology optimization.
- For randomly distributed defects, mean absolute deviations were approximately 11.3% for visual inspection, 3.6% for laser scanning, and 5.0% for topology optimization.
- The two non-contact approaches provide complementary rather than redundant capabilities, offering a basis for measurement-driven condition assessment and model updating.