Composites Don't Fail at Random: Your Simulation Should Know Why

Overview

In most composite teams the designer builds the ply stack and the laminate sequence in CATIA, while the analyst rebuilds that structure in a separate FEA environment. The handover costs accuracy, since the analyst works from a drawing while the definition itself stays in CATIA.

The analyst re-creates the stack-up with assumed fiber orientations and simplified material cards, runs the model and reports the results. Those results describe something close to the design, and how close stays an open question.

The 3DEXPERIENCE platform removes that handover by giving CATIA and Abaqus one shared definition of the laminate.

The building-block problem in composite certification

Composite certification follows the building block approach, a pyramid of tests that starts at the material coupon and works up through subcomponents, components and the full structure. The method is rigorous, and it is expensive and slow.

Integrated simulation keeps the pyramid in place, since EASA and the FAA require physical evidence. It compresses the middle layers, where well validated virtual tests take the place of many physical sub-element and component tests. The physical tests that remain become more targeted and more informative.

How CATIA and Abaqus actually talk to each other

The 3DEXPERIENCE platform connects CATIA Composites Design to Abaqus through a shared data model, so the laminate exists once and both applications read the same definition.

The traditional workflow passes data through a neutral file such as IGES or STEP, or through a proprietary exchange format. That step strips most of the composites metadata, so the laminate definition is rebuilt by hand in the solver environment. Orientations drift, material cards transfer partially, and someone makes decisions along the way that quietly change the model.

Three approaches to composite definition in CATIA

CATIA Composites Design supports three ways to define a composite, and Abaqus reads all three directly on 3DEXPERIENCE:

• Ply-by-ply: Each ply carries its own contour, material, fiber orientation and stagger offset. This gives the finest control over the laminate and suits complex aerostructures where every ply boundary matters.
• Zone-based: Laminate zones come first, each carrying a thickness law or a stacking sequence, and the individual plies are generated from them. This suits large structures with gradual thickness transitions.
• Grid-based: Stacking sequences are managed through a grid or spreadsheet interface. This suits very large structures where many zone to ply permutations have to be managed systematically.

Approaches to modeling damage in Abaqus

Damage modeling in composite FEA is a layered set of approaches, and the right one depends on what you want to predict and at what fidelity. Abaqus provides tools across all of these levels.

Level 1: First-ply failure (FPF) using failure indices

The simplest approach applies a failure criterion as a post-processing check on a linear elastic stress state. You run a standard linear static analysis, extract the ply-level stresses, and evaluate whether any ply exceeds a defined failure surface. Material degradation stays out of the model, so the structure carries load as if undamaged throughout.

This is first-ply failure analysis. It gives the load at which the first damage event is predicted, along with the ply and the mode responsible. Criteria commonly used this way in Abaqus include:

• Maximum stress and maximum strain: Simple component-by-component checks, conservative and easy to implement, and they leave the interaction between stress components out.
• Tsai-Hill: An interaction-based polynomial criterion derived from the Hill yield criterion for anisotropic metals and adapted for composites. It accounts for stress interaction while treating tensile and compressive failure alike.
• Tsai-Wu: A more general polynomial criterion with separate tensile and compressive strengths and a biaxial interaction term. Industry uses it widely, though the interaction coefficient calls for biaxial test data that is often unavailable.

Level 2: Progressive failure analysis (PFA)

Progressive failure analysis carries on past the first damage event. The material stiffness is degraded at each integration point where a failure criterion is met, and the load redistributes to the surrounding structure. The analysis continues until global instability or a defined failure condition is reached.

This represents the physical behavior of composite laminates more closely, since they typically keep carrying load after initial matrix cracking. The result is a load-displacement curve showing progressive stiffness loss, in place of a single safe or unsafe threshold. PFA in Abaqus can be implemented in two main ways:

• Built-in Hashin damage model: Damage initiation is governed by Hashin's four in-plane failure modes, detailed in the next section. Once initiated, damage evolves according to an energy-based stiffness degradation law. This is the most accessible route for shell and continuum shell element models.
• User subroutines (UMAT, VUMAT, USDFLD): For solid 3D elements, or for failure theories outside Abaqus such as Puck's action-plane criterion or the LaRC05 family, PFA is implemented through Fortran-based subroutines. This calls for more expertise and offers full control over the failure physics, including 3D Hashin variants, crack-band regularization and custom softening laws.

Level 3: Delamination and interlaminar failure

In-plane criteria such as Hashin and Tsai-Wu leave delamination out, since it occurs at the interface between plies while they describe failure within a ply. Interlaminar stresses and fracture mechanics govern delamination, and ply-level strength values sit outside that calculation.

Abaqus provides two main approaches for delamination:

• Cohesive elements (CZM): Thin cohesive elements sit at ply interfaces where delamination is expected. The traction-separation law, typically bilinear, governs damage initiation and evolution at the interface. The mixed-mode formulation by Camanho and Davila (2002), which Abaqus implements, is the reference approach for mode I, mode II and mode III coupling.
• Virtual Crack Closure Technique (VCCT): An energy-based approach that computes the strain energy release rate at an existing crack front and compares it to mode-specific fracture toughness values. VCCT suits cases where a pre-existing crack or defect is known and its growth is to be tracked.

Level 4: Woven and non-UD composites

The failure criteria above are derived for unidirectional fiber-reinforced composites. Woven fabrics, braided textiles and chopped strand mats have bidirectional or random fiber architectures that call for different modeling approaches. Woven composites sit outside the built-in material models in Abaqus, so adapted criteria are implemented through user subroutines and material definition. Recent research has extended Hashin's framework to woven fabrics by adding bidirectional failure modes, with good correlation to experimental data in both quasi-static bending and high-velocity impact.

Hashin criteria and the built-in PFA model

Among the progressive failure approaches, the built-in Hashin damage model in Abaqus is the most practical starting point for most composite structural analyses. What it covers, and where it stops, decides whether it fits a given analysis.

The four failure modes

Hashin's insight is that a fiber-reinforced composite fails through four physically distinct mechanisms, each driven by a different stress state, while a homogeneous material fails through one. The 1973 and 1980 formulations distinguish these modes explicitly:

From initiation to fracture, damage evolution

Damage initiation gives the point where a composite starts to break. Damage evolution gives what follows, as material stiffness degrades while cracks propagate. Abaqus implements this through an energy-based softening model, where the rate of stiffness reduction after initiation is governed by the fracture energy associated with each damage mode, fiber or matrix, tension or compression.

This progressive failure capability separates a serious composite analysis from a simple first-ply failure check. You get a map of how failure initiates and spreads through the laminate under increasing load, in place of a single safe or unsafe answer. That map informs both structural optimization and test program design.

Analysis types in the Composite Structures Analysis Engineer role

The Composite Structures Analysis Engineer role on 3DEXPERIENCE gives access to the full range of Abaqus analysis types. The table sets out each one and the composite work it suits:

The workflow from concept to validated design

The Composite Structures Analysis Engineer role on 3DEXPERIENCE organizes the analysis into five stages. The sequence starts with geometry preparation and ends with signed-off results that trace back to the design configuration they came from.

Iterating on ply design inside the same model

The integrated workflow lets you test design alternatives quickly, since the ply definition in CATIA connects directly to the simulation mesh. A change in fiber orientation, ply count or stacking sequence carries into the next Abaqus run on the same model.

A common optimization loop starts with a wing rib built on a standard [0/+45/-45/90]s quasi-isotropic layup. A linear static analysis shows the stiffness is adequate, while the first-ply failure index in bending sits close to the limit. The designer moves the off-axis angle from 45 degrees to 35 degrees and recomputes the draping check in CATIA, then the analyst reruns the Abaqus job. The model is shared across both applications, so the loop closes in hours.

Optimization studies the integrated workflow supports

• Parametric ply orientation studies change the angle plies across a range and evaluate the stiffness and failure index response. The connection to CATIA checks drapeability in the same iteration, so an orientation that looks structurally attractive but fails the drape check gets caught early.
• Thickness tapering optimization varies the ply drop-off regions to find the minimum weight configuration that still satisfies the buckling load factors and the damage tolerance requirements.
• Load case coverage mapping runs the maneuver, gust, ground and fatigue cases together and builds a critical load envelope, and the most demanding combinations then drive the laminate design.

Industry contexts

Aerospace

Weight, certification and damage tolerance drive the design work in aerospace, where composites make up more than half of a modern airliner structure by weight. The certification evidence for those structures must be traceable, reproducible and tied to one specific configuration. The digital thread in 3DEXPERIENCE handles that configuration management, which is the part that has made composite certification so document-heavy.

Automotive and motorsport

Motorsport teams redesign structures frequently and work on a short cycle between simulation and the physical build. Running a progressive failure analysis on an impact scenario against the existing model is worth real money on that schedule. Carbon fiber tubs, suspension components and body panels all sit in complex 3D stress states under crash loading, which Abaqus/Explicit handles well.

Pressure vessels and energy

Type IV composite pressure vessels for compressed hydrogen storage are among the hardest composite analysis problems to get right. They combine winding-induced residual stresses, pressure cycling fatigue and debonding at the liner-to-composite interface. The Abaqus wound composite modeler (WCM) plugin models the dome geometry and the helical winding parameters accurately. That accuracy decides the answer for vessels where dome zone failure governs the burst pressure.

The practical case

Many teams still export from CATIA, re-create the laminate definition in a separate preprocessor and manage material data by hand between design revisions. The shared data model replaces those three steps and keeps the Abaqus knowledge the team already has.

The physics of composite failure stay complex, and Hashin criteria, progressive damage evolution and cohesive zone delamination each demand engineering judgment to apply correctly. The 3DEXPERIENCE workflow removes the data handling around those decisions, so analysts spend their time on the mechanics.

Brice Thivolle, Senior Solutions Consultant at Dassault Systèmes, demonstrates the full workflow in this webinar.

References and further reading

Hashin, Z. and Rotem, A. (1973). A fatigue criterion for fiber-reinforced materials. Journal of Composite Materials, 7, 448-464.
Hashin, Z. (1980). Failure criteria for unidirectional fiber composites. Journal of Applied Mechanics, 47, 329-334.
Camanho, P.P. and Davila, C.G. (2002). Mixed-mode decohesion finite elements for composite delamination. NASA/TM-2002-211737.
Falzon, B.G. and Tan, W. (2017). Virtual testing of composite structures. In: The Structural Integrity of Carbon Fiber Composites. Springer, Cham.
Aerospace America / AIAA. The case for more virtual testing. aerospaceamerica.aiaa.org
Dassault Systèmes. CATIA Composites Design 3 (CPD) product documentation. 3ds.com
Abaqus Documentation: Damage and failure for fiber-reinforced composites, overview. Simulia / MIT Engineering.

Written by Melad Fahed for LaunchTech, an authorized Dassault Systèmes partner serving the Middle East. Technical claims come from the referenced sources, and the simulation capability details reflect the 3DEXPERIENCE platform.