The Difficult Task of Using FEA to Simulate CFRP

Carbon fiber reinforced plastic isn’t easy to simulate, but Solvay does so using advanced FEA tools.

Finite element analysis (FEA) allows us to simulate various forces acting on virtual components and informs us of design decisions before we even cut a single piece of physical material. This can save a lot of time and money. However, FEA does some jobs better than others. For example, basic scenarios, such as a steel beam undergoing bending, are easier to simulate than a carbon fiber reinforced plastic (CFRP) beam undergoing the same bending force. The reason? A piece of steel is homogenous in terms of its composition, whereas a CFRP beam is composed of laminated layers and has fibers running at angles to each other. It exhibits anisotropic behavior, porosity and friction between the layers—factors that make behavior difficult to predict.

Solvay is a CFRP product manufacturer that has a range of products including Evolite, a thermoplastic CFRP, and SolvaLite, a thermoset CFRP. There are fundamental differences in how these materials behave, so there is no one-rule-fits-all simulation or manufacturing solution suitable for both types of product. Much research is needed in both the simulation and manufacture of these products.

CFRP car door. (Image courtesy of Solvay.)

CFRP car door. (Image courtesy of Solvay.)

Thankfully, material scientists, manufacturing engineers and chemists alike work toward increasing our knowledge of CFRP manufacturing simulation. They do so not just from a mechanical perspective, but for every stage of manufacture and beyond, including simulation of crash tests. They are achieving this with the help of various simulation software solutions developed by Troy Mich. based Altair Engineering, Inc. (Altair).

“There are three areas in the process that we are aiming to simulate,” said Fabio Bressan, Solvay virtual engineering manager. “We hope to simulate the material at nano and microscale levels, the tooling at macro levels, and also simulate the trim curing process. Curing simulation is a Multiphysics problem and is time, temperature and viscosity dependent.”

The dynamic relationship between factors affecting curing is one of the main issues that differentiates thermoplastic and thermoset CFRP behaviors. More on that later.

CFRP Manufacture 101

Before we delve into the actual details of software used, let’s take a brief look at CFRP products and how they are manufactured. In order, the sequence goes like this:

1)      Prepregmaterial cutting and assembly

2)      Preforming and draping

3)      Application of pressure and heated curing

4)      Demolding

The first step involves the cutting of prepreg material and assembly (layering) to produce thicker sheets. A prepreg material, as opposed to a dry material, consists of carbon fiber filaments or sheets with the plastic resin already impregnated into the filaments.

The next phase involves draping the sheets onto a mold base to provide the geometry of the near-net shaped preform component. This draping phase can cause issues if not performed correctly, and this is one step that can be simulated by using Altair’s HyperWorks simulation software suite. If the prepreg material is not draped correctly, the fiber direction can be altered, resulting in a non-optimum part. This step produces a 3D part with the matrix and fibers consolidated for the next stage.

Draping simulation. (Image courtesy of Altair and CEDREM.)

Draping simulation. (Image courtesy of Altair and CEDREM.)

The preform is then moved to a second mold where it is compressed and another resin is injected into the part. The timing of this step is critical. Pressure must be applied at the correct time because it is a thermoset plastic matrix, and as a result the viscosity and phase of the plastic are irreversible. Once that sweet-spot is missed, it can result in sub-standard parts.

Demolding is the final step. When the final part has finished curing it can be removed from the mold and is ready for inspection. Simulation can also assist here by determining the best location for the mold ejector pins.

Material Cards

The bulk of the simulation work takes place within the Altair HyperWorks environment.The various stages of manufacture can be simulated with use of plugins and fairly elaborate coding. The robustness of the results is fully dependent on the quality of the inputs.

“If you put garbage in, you get garbage out,” Bressan said.

In terms of inputs, the entire process of simulating CFRP begins with the creation of what are referred to as “material cards.” The material cards are basically virtual material for the types of prepreg material configurations and contain information related to directional stiffness and strength for each material, as well as other mechanical data such as modulus of elasticity, Poisson’s ratio, shear modulus and so on.

The material cards are developed by first performing mechanical tests on physical samples while simultaneously modelling the mesoscale behaviour in a finite element environment.

The results of these parallel tests/simulations are combined to create the actual material cards, which are written for various commercial solvers including Altair’s RADIOSS solver for nonlinear dynamic analysis.

Next up, the cards are tested to see if they fit up to the physical experiments by running simulations on the card models while performing another set of physical tests. If the experimental/numerical results correlate, then it is a good card and can be used for future design and simulation work.

“Material cards are not simple ASTM data tables,” Bressan said. “They contain information regarding dynamic non-linear behaviors far beyond what a data table can provide. This is very important for analysis of complex failures of structures, such as vehicle crash analysis.”

Pressure / Curing

One of the issues with the current database of cards, however, is that many of them are based on thermoplastic polymers rather than thermoset polymers. Many applications prefer use of thermoset CFRP products because they behave the same over a range of temperatures and the phase changes are irreversible, unlike thermoplastics. This is particularly useful in aerospace, where a vehicle structure can be subjected to various temperature extremes depending on location and altitude.

Knowing the behavior of thermoset polymer based CFRPs is critical when it comes to simulating the curing process. Due to the irreversible nature of thermosets during application of pressure in the third step of manufacturing, timing is everything. There is a specific sweet-spot in terms of viscosity where the pressure must be applied. If that spot is missed, the final product can suffer imperfections resulting in loss of performance and critical part failure. This is not such an issue with thermoplastic resins because phase and viscosity are both reversible.

Relationship between temperature and viscosity.

Relationship between temperature and viscosity.

Simulation

We have seen a basic overview of the manufacturing process and challenges associated with providing good material models. How exactly are composite designers using software to simulate these processes?

The early composite design phases can be performed with OptiStruct software, which allows design variables to be altered to determine optimized designs. By using OptiStruct, designers can gain understanding of the factors that will influence both performance and safety of the product, before even cutting a single ply.

By using OptiStruct in the early stages of design, the engineers are able to determine optimal location of ply drops and ply build-ups, develop ply cut-out patterns and define the optimal stacking sequence for creating thick laminates.

Hot compression molding simulation (Image courtesy of Solvay.)

Hot compression molding simulation (Image courtesy of Solvay.)

For setting up Finite Element models, HyperMesh can be used. This allows users to generate laminate definitions efficiently and  visualize stacking sequence, ply orientation and draping angle deviations.

For analysis of static and dynamic scenarios, OptiStruct can be used for linear and nonlinear analysis involving static, dynamic and thermal loadings, and RADIOSS can be used for highly nonlinear problems under dynamic loading, such as crash analysis.

Because the Finite Element models have been well-defined at the laminate and ply level in HyperMesh, sophisticated post-processing at the laminate or ply level can be achieved and results assessed with HyperView.

Like any modern design workflow, the results of such analysis is combined with real-world experimental data and fed back into the iterative loop, which enables more robust and accurate simulations for future design work.

Future Developments

We can see how that even now simulation is being used at various stages of design and manufacture of CFRP products. But Solvay still has a lot of work to close the gap between idealized simulation and the realities of manufacture. As mentioned previously, a good simulation starts with a good material card. So, the more the material card database grows, so does the ability to generate accurate results.

But much of the error still comes from human sources. CFRP still requires a hands-on approach at certain stages, and at these points error can creep in.

“Automation is the future for composite materials,” Bressan said. “If you can simulate every step of manufacturing, then you can master the physics of the manufacturing process.”

When CFRP products can be manufactured with a fully autonomous process and consistent high-quality parts suited to high demanding industries—such as aerospace and automotive—produced, manufacturing costs will drop significantly.

Faster, better and cheaper. NASA tried that philosophy once, with mixed results. Will the CFRP industry succeed where many have failed before? Only time will tell.


Altair has sponsored this story. All opinions are mine, except where quoted or stated otherwise. —Phillip Keane