Defensible Advanced Composite Damage Analysis Workflow

What Is It?

This article presents the complete workflow for conducting a defensible advanced composite damage analysis. It synthesises the concepts from the entire Advanced Damage & Failure Methods chapter into a structured, sequential process — from the initial structural requirement through to the final report. Each step in the workflow builds on the previous ones, and the chain must be internally consistent. The workflow is the closing cornerstone of the chapter: it is the practical framework that ties together calibration, validation, mesh sensitivity, energy balance, sensitivity analysis and failure mode verification into a coherent, defensible analysis process.

Why It Matters

A defensible analysis is one that can withstand technical scrutiny — by a certifying authority, by a customer, by an independent reviewer. It is not just a set of results; it is a documented chain of reasoning from requirements to conclusions, with each step justified by analysis, test evidence or established practice. An analysis that skips steps, uses unvalidated assumptions or does not address uncertainty is not defensible — its conclusions cannot be trusted. The workflow ensures that all necessary steps are addressed, in the correct order, with the appropriate evidence at each stage.

A defensible analysis can withstand technical scrutiny. It is a documented chain of reasoning from requirements to conclusions, with each step justified. The workflow ensures all necessary steps are addressed, in order, with appropriate evidence. Skip no step.

Step 1: Structural Requirement

The starting point is the structural requirement — what the structure must do, under what loads, with what damage present, and to what standard. The requirement defines the load cases (limit, ultimate, fatigue spectrum), the damage categories (BVID, VID, discrete source, manufacturing defects), the residual strength requirements and the applicable regulations and standards. Everything downstream flows from the requirement — the analysis is meaningless without a clear statement of what it must demonstrate.

• Define the load cases: static limit, static ultimate, fatigue spectrum, impact, discrete source
• Define the damage categories: BVID, VID, discrete source, manufacturing defects
• Define the residual strength requirements for each damage category
• Identify the applicable regulations and standards (e.g. CS/14 CFR 25.571, AMC 20-29)
• The requirement drives everything downstream — state it clearly and unambiguously

Step 2: Laminate and Material Definition

The laminate and material definition establishes what is being analysed. The material system (fibre, resin, prepreg, cure), the ply properties (E1, E2, G12, ν12, strengths, fracture energies) and the laminate lay-up (ply sequence, orientations, thicknesses) must be defined. The material properties must be from the qualified material allowables — not generic data. The fracture energies must be measured from fracture tests. The laminate definition must match the as-built configuration, including any manufacturing tolerances that are relevant to the analysis.

Material properties must be from qualified allowables — not generic data. Fracture energies must be measured from fracture tests — not assumed. The laminate definition must match the as-built configuration, including relevant manufacturing tolerances.

Step 3: Loads and Environment

The loads and environment define the conditions the structure experiences. The loads include the mechanical loads (tension, compression, shear, combined), the thermal loads (cure stress, operating temperature) and the environmental conditions (hot/wet, cold/dry). The loads must be the design loads — limit, ultimate and fatigue spectrum — applied in the correct sequence and with the correct boundary conditions. Environmental effects (temperature, moisture) on material properties must be included if they reduce the properties to critical values.

• Mechanical loads: tension, compression, shear, combined — at limit and ultimate levels
• Thermal loads: cure stress, operating temperature differentials
• Environmental conditions: hot/wet (matrix-dominated properties reduced), cold/dry
• Fatigue spectrum: load sequence, R-ratios, number of cycles
• Apply loads with correct boundary conditions and in the correct sequence

Step 4: Initial Defects and Damage

The initial defects and damage define the starting state for the damage analysis. This includes manufacturing defects (porosity, waviness, wrinkles, misalignment — at acceptance levels) and service damage (impact damage at BVID or VID levels, discrete source damage). The initial damage state must be representative of the damage category being assessed — established by test, NDE mapping or validated impact analysis. For manufacturing defects, the acceptance levels must be those established by the project-specific substantiation — not generic values.

Initial defects and damage must be representative of the damage category — established by test, NDE or validated analysis. Manufacturing defect acceptance levels must be project-specific — not generic. The initial damage state drives the entire downstream analysis.

Step 5: Structural Idealisation

The structural idealisation converts the physical structure into the analysis model. This includes the element type (shell, solid, layered), the mesh density, the boundary condition idealisation, the load introduction method and the level of detail (which features are included, which are simplified). The idealisation must capture the features that are critical to the damage behaviour — stress concentrations, ply drops, stiffener terminations, joints, free edges. Features that are not critical may be simplified to reduce model size, but the simplification must be justified.

• Element type: shell (efficient, less through-thickness detail), solid (full 3D, more detail, more costly)
• Mesh density: fine enough to resolve damage zone and stress concentrations
• Boundary conditions: idealised — assess sensitivity to the idealisation
• Load introduction: must represent the real load path without artificial stress concentrations
• Level of detail: include critical features (ply drops, stiffener ends, joints); simplify non-critical features

Step 6: Failure-Mode Expectation

Before running the analysis, state the expected failure mode. What failure mechanism is anticipated — fibre tension, fibre compression, matrix cracking, delamination, buckling-driven delamination, net-section fracture? The expectation is based on the structure, the loads, the laminate and experience. Stating the expectation upfront prevents confirmation bias — if the analysis predicts a different mode, the discrepancy is investigated, not ignored. The failure-mode expectation also guides the model setup: if delamination is expected, cohesive elements must be included; if fibre fracture is expected, the fibre damage model must be active.

State the expected failure mode before running the analysis. This prevents confirmation bias and guides the model setup. If the analysis predicts a different mode, investigate the discrepancy — do not ignore it.

Step 7: Initiation Criterion

Select and calibrate the damage initiation criterion. The criterion must be appropriate for the expected failure mode — a Hashin or Puck criterion for ply damage, a cohesive traction law for delamination. The criterion parameters (strengths) must be from qualified allowables. The criterion must be calibrated against coupon data that isolates the relevant failure mode. The selection, calibration and source of the criterion should be documented.

• Select criterion appropriate for expected failure mode
• Calibrate against coupon data that isolates the relevant mode
• Strengths from qualified allowables — not generic data
• Document the criterion selection, calibration and data source

Step 8: Damage Evolution

Select and calibrate the damage evolution law. The law must be appropriate for the material behaviour — linear, exponential or energy-based softening. For mesh objectivity, use fracture-energy-based evolution with a characteristic length. The fracture energies must be measured from fracture tests — not assumed. The evolution law parameters must be calibrated against coupon post-peak behaviour. The selection, calibration and source should be documented.

Use fracture-energy-based evolution with characteristic length for mesh objectivity. Fracture energies must be measured from fracture tests — do not invent them. Calibrate against coupon post-peak behaviour.

Step 9: Interlaminar Model

Select and calibrate the interlaminar damage model — cohesive elements or VCCT. The model must capture the delamination behaviour relevant to the structure: initiation (strength-controlled), propagation (fracture-energy-controlled) and mixed-mode behaviour. Cohesive elements are generally preferred for problems involving delamination initiation and growth; VCCT is suited to pre-existing crack growth. The cohesive mesh must be fine enough to resolve the process zone. The interlaminar properties (G_Ic, G_IIc, mixed-mode exponent) must be from fracture tests.

• Cohesive elements: preferred for delamination initiation + growth
• VCCT: suited to pre-existing crack growth
• Cohesive mesh: fine enough to resolve the process zone (l_e ≤ l_pz / 3 to l_pz / 5)
• Mixed-mode behaviour: use B-K or power-law criterion with measured exponent
• G_Ic, G_IIc from DCB, ENF tests — not assumed

Step 10: Mesh and Regularisation

Define the mesh and verify regularisation. The ply mesh must be fine enough to resolve the damage zone. The cohesive mesh must resolve the process zone. Regularisation (fracture-energy-based evolution + characteristic length) must be active for mesh objectivity. Run a mesh sensitivity assessment with at least two mesh densities and compare the failure load, damage pattern and energy dissipation. The results must be mesh-independent within an acceptable tolerance. Document the mesh, the regularisation and the mesh sensitivity assessment.

Run a mesh sensitivity assessment with at least two densities. Verify that the failure load, damage pattern and energy dissipation are mesh-independent. Document the mesh, the regularisation and the assessment. Use the converged mesh for the final analysis.

Step 11: Nonlinear Solution

Set up the nonlinear solution. Progressive damage analysis is inherently nonlinear — the stiffness changes as damage evolves. The solution method (implicit or explicit), the load step size, the convergence criteria (for implicit) and the time step (for explicit) must be appropriate for the problem. For implicit, the convergence tolerance must be tight enough to capture the damage progression accurately. For explicit, the mass scaling (if used) must not affect the physical response. The solution parameters should be verified — confirm that the analysis converges (implicit) or is stable (explicit) and that the results do not change with finer step sizes.

• Implicit: appropriate load step size, tight convergence tolerance for damage progression
• Explicit: appropriate time step; mass scaling must not affect physical response
• Verify solution parameters: convergence (implicit) or stability (explicit)
• Confirm results do not change with finer step sizes or tighter tolerances
• For explicit: check the energy balance (see Step 13 and energy balance article)

Step 12: Progressive Damage

Run the progressive damage analysis. Examine the damage progression: where does damage initiate, how does it propagate, how do the damage modes interact, what is the final failure mechanism? Compare the predicted failure mode to the expected failure mode (Step 6) — if they differ, investigate. Examine the damage contours, the damage variable history and the load redistribution. The progressive damage output is the core of the analysis — it provides the damage pattern, the failure mode and the failure load.

Compare the predicted failure mode to the expected failure mode (Step 6). If they differ, investigate — do not ignore the discrepancy. Examine damage contours, damage variable history and load redistribution. The progressive damage output is the core of the analysis.

Step 13: Damage Pattern

Examine the predicted damage pattern — the size, shape, location and through-thickness extent of the damage. The damage pattern should be physically plausible: the delamination shape should be consistent with the lay-up, the damage should be in the expected location, the through-thickness distribution should be consistent with the stress state. An implausible damage pattern indicates a model problem — incorrect parameters, mesh issues or missing physics. The damage pattern will be compared against test evidence (NDE, sectioning) in the test correlation step.

• Damage size: is the predicted damage area plausible for the load and structure?
• Damage shape: is the delamination shape consistent with the lay-up (staircase pattern)?
• Location: is the damage where expected from the stress state?
• Through-thickness: are the damaged interfaces and plies consistent with the stress distribution?
• Implausible patterns indicate model problems — investigate before proceeding

Step 14: Residual Strength

Determine the residual strength from the analysis — the maximum load the damaged structure can carry. The residual strength is the primary output for damage tolerance assessment: it must be compared against the required design load for the applicable damage category. If the residual strength exceeds the requirement with adequate margin, the structure is damage-tolerant for that damage category. If not, the structure must be redesigned, the damage category redefined or the inspection regime adjusted. The residual strength should be reported with the uncertainty bounds from the sensitivity analysis.

Residual strength assessment:

P_residual = maximum load from progressive damage analysis

Requirement:
  P_residual ≥ P_required × (1 + margin)

where P_required depends on damage category:
  BVID:            Limit Load
  Visible damage:  Limit Load (shorter inspection interval)
  Discrete source: Continue Safe Flight Load

Report:
  P_residual (mean)
  P_residual (lower bound from sensitivity/uncertainty)
  Margin = P_residual / P_required − 1

Step 15: Mesh Sensitivity

Revisit mesh sensitivity as a formal step in the workflow. The mesh sensitivity assessment (Step 10) was done during the model setup; this step confirms that the final analysis, with the final mesh and the final parameters, is mesh-independent. If the analysis was run with only one mesh, run it again with a refined mesh and compare. If the results change, the mesh is not adequate — return to Step 10. The mesh sensitivity confirmation is part of the defensible analysis evidence.

Confirm mesh independence of the final analysis. If only one mesh was used, run a second and compare. If results change, the mesh is not adequate. Mesh sensitivity confirmation is part of the defensible analysis evidence — do not skip it.

Step 16: Parameter Sensitivity

Conduct the parameter sensitivity analysis. Vary the key input parameters — strengths, fracture energies, stiffness, ply angle, impact energy, boundary conditions — within their uncertainty ranges and assess the effect on the failure load, damage pattern and failure mode. Identify the dominant parameters. Report the results with confidence bounds reflecting the dominant uncertainties. For critical structure, use a systematic approach (design of experiments) to capture interaction effects. The sensitivity analysis provides the confidence bounds on the prediction and identifies where additional characterisation effort is needed.

• Vary strengths, fracture energies, stiffness, ply angle, impact energy, boundary conditions
• Assess effect on failure load, damage pattern and failure mode
• Identify dominant parameters — focus additional characterisation effort there
• Report results with confidence bounds from the dominant uncertainties
• For critical structure: use systematic (DoE) approach to capture interactions

Step 17: Test Correlation

Correlate the analysis against test evidence across the building block. The correlation must address not only the failure load but also the failure mode, the damage pattern, the load-displacement response and the strain fields. The test evidence comes from coupon tests (material properties), element tests (structural details), subcomponent tests (structural features) and component/full-scale tests (overall validation). The correlation identifies where the model matches and where it does not — mismatches must be investigated and either resolved or acknowledged as limitations. The test correlation is the core of the validation evidence.

Correlate against test across the building block: coupon, element, subcomponent, component. Address failure load, failure mode, damage pattern, load-displacement and strain. Investigate mismatches — resolve or acknowledge as limitations. This is the core validation evidence.

Step 18: Uncertainty Quantification

Quantify the uncertainty in the analysis results. The uncertainty comes from material parameter scatter, model form uncertainty (the model may not capture all physics), mesh discretisation uncertainty, boundary condition idealisation and test data scatter. The uncertainty should be quantified and reported — not hidden. The residual strength should be reported as a range, not a single number, with the lower bound used for the damage tolerance assessment. For critical structure, the lower bound should be a B-basis or A-basis value, not the mean.

• Material parameter scatter: from coupon test data — B-basis or A-basis for critical structure
• Model form uncertainty: the model may not capture all physics — acknowledge limitations
• Mesh discretisation: from mesh sensitivity assessment
• Boundary condition idealisation: from sensitivity to boundary variations
• Report residual strength as a range — use the lower bound for damage tolerance assessment

Step 19: Reporting

The final step is reporting. The report documents the entire workflow — from the structural requirement through to the final residual strength with uncertainty bounds. The report must include: the requirement, the material and laminate definition, the loads and environment, the initial damage state, the structural idealisation, the failure-mode expectation, the model setup (initiation, evolution, interlaminar, mesh, solution), the progressive damage results (damage pattern, failure mode, failure load), the mesh sensitivity confirmation, the parameter sensitivity results, the test correlation evidence, the uncertainty quantification and the conclusions. The report is the defensible analysis — it must be sufficient for an independent reviewer to assess the analysis and its conclusions.

The report is the defensible analysis. Document every step — from requirement to conclusion — with the evidence at each stage. An independent reviewer must be able to assess the analysis and its conclusions from the report alone. No step should be undocumented.

Complete Workflow Summary

The complete workflow is a chain of 19 steps, each building on the previous. The chain is only as strong as its weakest link — a deficiency at any step undermines the entire analysis. The workflow ensures that all aspects of a defensible advanced composite damage analysis are addressed, in the correct order, with the appropriate evidence. It is the practical framework that ties together the entire Advanced Damage & Failure Methods chapter.

Key Takeaways

• The workflow is a 19-step chain from structural requirement to reporting — each step builds on the previous
• The chain is only as strong as its weakest link — a deficiency at any step undermines the entire analysis
• Start with the requirement — everything flows from it
• Material properties and fracture energies must be measured, not assumed or invented
• State the expected failure mode before running the analysis — investigate discrepancies
• Verify mesh objectivity, solution stability and energy balance
• Conduct parameter sensitivity analysis — identify dominant parameters, report with confidence bounds
• Correlate against test across the building block — verify failure mode, not just load
• Quantify and report uncertainty — use lower-bound values for critical structure
• Document every step — the report is the defensible analysis