Concrete Crack Modeling for Nuclear Containment Structure Safety Analysis

Overview of Technical Issues:

Current modeling approaches insufficiently represent crack propagation behavior in concrete containment structures, leading to uncertainty in predicting structural failure modes and inadequate safety margin assessment under nuclear accident conditions; the goal is to develop accurate crack modeling methods that reliably predict containment integrity and support safety analysis for nuclear facilities.

Problem Direction 1 :
ImproveModel prediction accuracy for crack behavior
VS
ConstraintComputational resource consumption
Inspiration 1 : Cross-domain reference
Application Principle: #1 Segmentation
Cross-domain Case Inspiration
This patent improves measurement precision (3D reconstruction accuracy) through iterative refinement of segmented scan zones while preventing excessive energy use (computational resources) by starting with lower-resolution models and selectively refining only critical areas. It demonstrates how segmentation combined with adaptive resolution allocation resolves the contradiction between accuracy and resource consumption, directly paralleling the current need to achieve <10% crack prediction deviation within constrained computational budgets.
Determining a three-dimensional model of a scan target
Innovative Solution View detail
Hierarchical domain decomposition with progressive crack zone refinement for containment integrity analysis
Divide containment into independent zones with adaptive mesh density
How to solve :
  • Partition containment structure into hierarchical analysis domains: Level-1 bulk concrete (50mm mesh), Level-2 high-stress regions at penetrations/joints (10mm mesh), Level-3 active crack zones (2mm mesh)
  • Implement progressive refinement protocol — run Level-1 global analysis first (2-4 hours), identify stress concentration zones exceeding 0.7×tensile strength, activate Level-2 refinement only in flagged domains (8-12 hours), trigger Level-3 micro-mesh when crack initiation detected (strain discontinuity >150 με)
  • Apply interface coupling via boundary force transfer — coarse domain provides displacement boundary conditions to fine domain, fine domain returns reaction forces
  • iterate until force equilibrium error <2%, typically 3-5 cycles
  • Quality control: verify mesh transition ratio ≤1:3 between adjacent zones, validate stress continuity error <5% at domain interfaces using patch test, benchmark against uniform 2mm mesh on 1/8 scale specimen (acceptance: crack width deviation <8%)
Expected Effect : Accuracy <10% deviation; computation 12-18× baseline; cycle time 24-36 hours
Risk Control :
  • interface coupling convergence failure
  • domain boundary placement subjectivity
  • crack path crossing domain boundaries
Inspiration 2 : Technology in this field
Search: Adaptive mesh refinement, Element size optimization, Stress intensity factor prediction, Machine learning crack prediction, Cohesive zone modeling
Existing SolutionView detail
Hierarchical Adaptive Mesh Refinement with Dimensionality Reduction for Crack Propagation Modeling
Strategic mesh refinement combines dimensionality reduction and adaptive element sizing to balance accuracy and efficiency
How to solve :
  • Apply 3D-to-2D plane strain dimensionality reduction for crack propagation zones (reduces computation to 6.77% per ref 1) with element size 0.3-0.6mm near crack tips (7-16× concrete grain size per refs 4,6,7)
  • Implement hierarchical mesh strategy with three zones: ultra-fine elements (0.3mm) at crack tip cylindrical region, medium elements (0.6-1mm) in propagation path, coarse elements (3-5mm) in far-field (refs 5,9,13)
  • Use modified superconvergent patch recovery (SPR) error estimator with analytical crack-tip field integration for adaptive remeshing triggers, updating mesh every 50 load increments when local error exceeds 5% threshold (refs 2,3)
Expected Effect : Prediction error <8% with 12-18× computational cost versus current uniform mesh approaches
Risk Control :
  • Element aspect ratio control to avoid numerical instability at crack tips
  • Transition zone mesh quality between refinement levels
  • Calibration of remeshing frequency versus crack propagation rate
Problem Direction 2 :
ImproveModel prediction accuracy for crack behavior
VS
ConstraintAnalysis cycle duration
Inspiration 1 : Cross-domain reference
Application Principle: #10 Preliminary action
Cross-domain Case Inspiration
This patent improves measurement precision (maintaining accurate building models through continuous data collection) while preventing loss of time (eliminating manual data entry and outdated information delays). It applies preliminary action by establishing an automated data infrastructure that pre-populates and maintains system models, directly paralleling the need to pre-compute and store crack propagation scenarios for instant retrieval during safety reviews.
Passive and active wireless building management system and method
Innovative Solution View detail
Pre-computed crack scenario library with real-time interpolation engine for rapid safety assessment
Build validated crack library before design iteration
How to solve :
  • Execute 200-300 high-fidelity crack simulations offline covering design basis accidents (LOCA at 0.4-0.7 MPa, seismic 0.2-0.5g PGA, thermal gradients 50-150°C) with parametric variations in concrete strength (30-50 MPa), rebar ratio (0.3-0.8%), and prestress (0.8-1.2 MPa)—store crack width, propagation path, and failure load as indexed scenario database
  • Deploy multi-dimensional interpolation engine using radial basis function with Gaussian kernel (shape parameter α=0.5-1.5) to estimate crack behavior for new design parameters within 10-30 minutes by weighted averaging of 8-12 nearest pre-computed scenarios—achieves <10% deviation through dense scenario coverage
  • Implement real-time validity checker that flags when design parameters fall outside pre-computed envelope (>15% extrapolation distance) and triggers targeted high-fidelity simulation only for out-of-bounds cases—maintains 3-5 day cycle by limiting full simulations to <5% of design iterations
Expected Effect : Analysis time reduced from 2-4 weeks to 10-30 min per iteration; prediction accuracy <10% deviation; 95% design variants resolved via interpolation
Risk Control :
  • scenario library coverage gaps causing extrapolation errors
  • interpolation accuracy degradation at parameter boundaries
  • initial library generation requiring 3-6 months upfront investment
Inspiration 2 : Technology in this field
Search: Automated crack propagation simulation, Machine learning prediction methods, Enhanced FEM accuracy techniques, Multi-inspector error correction, Rapid cycle analysis workflow
Existing SolutionView detail
Hybrid Dimensionality Reduction with Probabilistic Crack-Propagation Prediction Framework
Integrate dimensionality reduction and probabilistic prediction to balance accuracy and speed
How to solve :
  • Apply shell element dimensionality reduction from 3D solid models (error ≤1.25%) combined with local mesh coarsening in non-critical regions (fine mesh at crack tip radius ≤3× crack length, coarse mesh elsewhere) to reduce simulation time to 3.33% of full-order model while maintaining error ≤1.28%
  • Implement Monte Carlo-based crack-propagation curve estimation with N≥1000 combinations of material parameters (concrete tensile strength variation ±15%, fracture energy ±20%, environmental factors) originating from corrected initial crack measurements to generate probabilistic failure envelopes
  • Establish inspector-method correlation databases through simulated inspections on test specimens with known crack geometries, applying correlation diagrams to correct field measurements and estimate actual crack length probability distributions, feeding corrected initial conditions into propagation models
Expected Effect : Prediction deviation <10%, analysis cycle 3-5 days, computational cost reduction to 3-5% of baseline
Risk Control :
  • Correlation database representativeness for concrete-specific crack morphologies
  • Monte Carlo convergence verification for containment-scale models
  • Mesh transition quality between fine-coarse regions affecting stress singularity accuracy
Problem Direction 3 :
ImproveCrack characteristic measurement resolution
VS
ConstraintComputational resource consumption
Inspiration 1 : Cross-domain reference
Application Principle: #2 Taking out (Extraction)
Cross-domain Case Inspiration
This patent improves detection capability (real-time depth monitoring) while avoiding excessive computational burden by using distributed sensor architecture that [extracts] only critical parameters (depth, attitude) at source before centralized processing, preventing the worsening of energy/resource consumption—directly paralleling the current contradiction of enhancing crack detection resolution without proportionally increasing computational load.
Dark pine soil preparation work quality detecting system
Innovative Solution View detail
Hierarchical crack feature extraction with sensor-level preprocessing for sub-0.1mm detection
Deploy sensor-level preprocessing to extract crack features before central processing
How to solve :
  • Embed FPGA-based edge processors at distributed fiber optic interrogator units to execute real-time strain gradient thresholding (∂ε/∂x > 500 με/m triggers crack flag) and spatial discontinuity detection, transmitting only crack coordinates, width, and orientation vectors (≤2 kB per event) instead of raw strain field data (≥50 MB per scan)
  • Implement three-tier data hierarchy: Tier-1 sensors capture full-field at 0.65mm spatial resolution via Rayleigh backscatter
  • Tier-2 edge units apply Canny edge detection with adaptive thresholding (sensitivity 0.08mm crack opening) and compress to crack skeleton graphs
  • Tier-3 central processor reconstructs 3D coalescence from skeleton inputs using graph-based topology mapping (Delaunay triangulation of crack nodes)
  • Quality control via dual-threshold validation—edge units flag potential cracks at 0.08mm (high sensitivity), central processor confirms at 0.10mm using multi-sensor fusion (≥3 adjacent sensors agree within 0.02mm tolerance)
  • automated calibration every 24 hours using reference strain standards (±10 με accuracy)
Expected Effect : Data volume reduced 96%; processing load 4.2× current; <0.1mm crack detection confirmed; 3D coalescence mapping in 15 min
Risk Control :
  • FPGA algorithm stability under temperature drift ±20°C
  • false positive rate from vibration noise
  • sensor-processor synchronization latency >50ms
Inspiration 2 : Technology in this field
Search: Digital Image Correlation (DIC), Subpixel Resolution Enhancement, 3D Crack Reconstruction, Deep Learning Crack Detection, High-Resolution Imaging
Existing SolutionView detail
Sub-Pixel Luminance-DIC Hybrid Crack Measurement System
Integrate sub-pixel crack detection with 3D tracking at controlled computational cost
How to solve :
  • Apply luminance information sum method with blur function modeling to estimate crack widths in sub-pixel units (0.35-1.0 μm/pixel resolution), using relational expressions incorporating background luminance, imaging posture, and resolution parameters to achieve <0.1mm detection without boundary pixel identification
  • Implement Digital Image Correlation (DIC) with optimized subset size 23×23 pixels and 11-pixel spacing to track displacement fields and strain concentrations, enabling 3D crack coalescence detection through correlation point analysis with spatial resolution ~25 μm/pixel and measurement accuracy <1/100 pixel
  • Deploy selective processing strategy using feature extraction (edge detection, template matching) to identify crack regions, then apply super-resolution and detailed DIC analysis only to extracted partial images, reducing computational load to 5-8× baseline while maintaining full-field measurement capability for crack initiation, propagation, and coalescence patterns
Expected Effect : Crack detection <0.1mm width with ±0.05mm accuracy; 3D coalescence tracking; computational cost 5-8× current levels
Risk Control :
  • Illumination condition stability affecting luminance-based measurements
  • DIC correlation accuracy in heterogeneous concrete microstructure
  • calibration consistency across multi-scale imaging systems
Problem Direction 4 :
ImproveCrack characteristic measurement resolution
VS
ConstraintAnalysis cycle duration
Inspiration 1 : Cross-domain reference
Application Principle: #10 Preliminary action
Cross-domain Case Inspiration
This patent improves detection efficiency (reducing loss of time) by [pre-configuring] multiple storage locations with prioritized access paths, while maintaining data quality and security. It demonstrates how [preliminary action] through pre-established infrastructure resolves the contradiction between measurement thoroughness (difficulty of detecting) and operational speed (loss of time), directly paralleling the current need to enhance crack detection resolution without extending analysis cycles.
Image forming apparatus
Innovative Solution View detail
Pre-embedded fiber optic sensor network with automated crack detection pipeline
Pre-embed distributed fiber optic sensors during specimen fabrication to eliminate setup time
How to solve :
  • Install distributed fiber optic sensors (spatial resolution 0.65mm, strain sensitivity ±1με) into concrete specimens during casting—embed Rayleigh backscatter interrogation fibers in 3D grid pattern with 5mm vertical spacing and 10mm horizontal spacing, eliminating 2-3 weeks of pre-test sensor installation and calibration
  • Deploy real-time edge computing units (FPGA-based interrogators operating at 100Hz sampling rate) co-located with fiber optic sensors to execute crack detection algorithms (strain gradient threshold ≥500με/mm for crack initiation) during loading—transmit only detected crack coordinates, width (±0.02mm accuracy), and 3D propagation vectors instead of raw terabyte-scale strain fields, reducing central processing load by 95%
  • Establish pre-configured analysis pipeline with automated crack reconstruction software that ingests sensor alerts and generates 3D crack maps within 2 hours of test completion—uses pre-trained machine learning models (trained on 150+ validation tests) to classify crack types and predict coalescence patterns, delivering model validation datasets within 1-2 weeks total cycle including 5-7 day loading test
Expected Effect : Resolution <0.1mm; cycle time 1-2 weeks; setup time zero; data load -95%
Risk Control :
  • fiber sensor survival during concrete curing and loading
  • edge algorithm false positive rate in noisy environments
  • pre-trained model generalization to new crack patterns
Inspiration 2 : Technology in this field
Search: X-ray micro-computed tomography (μCT), Digital volume correlation (DVC), Digital image correlation (DIC), Sub-pixel crack detection, Scanning electron microscopy (SEM-DIC)
Existing SolutionView detail
X-ray Micro-CT with Digital Volume Correlation for Sub-0.1mm 3D Crack Characterization in Concrete Containment
Integrate X-ray micro-computed tomography with digital volume correlation to capture 3D crack patterns in concrete specimens
How to solve :
  • Deploy industrial μCT scanner with 80-100μm voxel resolution, beam energy 200-225 kV, exposure 500ms per projection, achieving <0.1mm crack detection
  • Apply multi-pass DVC algorithm with FFT window deformation, interrogation areas 512→256 pixels, strain resolution <0.0002, extracting crack propagation direction and opening displacement
  • Implement strain-assisted crack segmentation combining grayscale thresholding with DVC strain fields (threshold >0.005 principal strain) to distinguish true cracks from imaging artifacts, enhancing detection integrity by 40%
Expected Effect : Sub-0.1mm crack resolution; 3D crack mapping; 10-12 day analysis cycle
Risk Control :
  • Sample size versus resolution trade-off requiring multiple scans
  • DVC correlation quality in low-contrast concrete requiring contrast enhancement protocols
  • Computational resource scaling for large datasets requiring parallel processing infrastructure
PatSnap Eureka Solution

If you want to get more technical solutions, you can try Eureka.