The production of composite parts relies on an ever increasing number of processing technologies which have a large impact on the performance of the part. The manufacturing step, or processing, is usually optimized for specific materials by using process simulation tools. Commonly used technologies are injection molding, injection-compression molding, compression molding, fiber draping or placement software. All of these approaches are applied in combination with short, long, and continuous fiber reinforced plastics for additional strength. Composite materials are heavily influenced by the processes used to create the part, which can lead to non-uniform material behavior throughout the final manufactured part, adding another challenge to the design process.

Manufacturing, local microstructure, material properties, and final performance are interdependent.

To reach the best possible design for some specific performance constraint, such as the overall weight, it is important to go through optimization cycles for the processing step, the structural design, and even of the material itself. The final performance of the part will depend on all three at the same time, and therefore it is important to consider them simultaneously. Today it is possible to set up multi-scale simulations where all three influences are fully coupled and thus can be investigated in one unique approach, resulting in more accurate analysis and shorter design cycles.

Composite materials are typically composed of a material matrix with embedded inclusions. The matrix and inclusion phases are adjusted according to their properties and performance with respect to a specific application. These material properties are also tweaked by changing the microstructure of the composite, resulting in a range of different macroscopic responses. Possible variations could include, for example, the amount or the shape of the inclusions used, or the use of short, long, or continuous fibers. These parameters are varied in order to tailor the ultimate material performance.

Whatever composite is used, the tailored properties, and thus their real benefit, depend heavily on the microstructure of the material. This is where complexity of design sets in. Manufacturing has a direct impact on the material microstructure. This means the processing step will define local material properties in the part, which can vary from one location to another, leading directly to a range of material properties throughout the composite part. This distribution of properties will directly impact the overall performance of the composite design. Thus the manufacturing, local microstructure, material properties, and final performance of the part are interdependent.

As a consequence, to take full advantage of the tailored composite materials it is mandatory to take all of these layers of influence into account. In this context, the task is to set up coupled analyses that build on processing simulation and map that influence onto the structural analysis.

The key piece linking the processing to the performance is the material model. It must be capable of predicting the local composite properties based on input from the material microstructure as projected by the process simulation. This objective can be achieved by setting up multi-scale analyses based on micromechanical modeling. Such analyses start with an input of individual properties for the matrix and inclusion phases which then are re-combined based on additional information for the microstructure to calculate the local performance of the material. The resulting material model is a function of the microstructure and can describe the stiffness and failure of composites under various loading, and automatically includes the anisotropic nature of the material.

With the material model defined as a function of microstructure, another challenge in setting up coupled multi-scale analyses originates from predicting composite microstructures with a processing simulation. Depending on the material, different kinds of manufacturing methods can be thought of, which include but are not limited to the following:

  • Injection molding
  • MuCell®
  • (Injection-) Compression molding
  • Draping
  • Fiber placement
Coupled analysis for the MuCell process. Local fiber orientation and void information are mapped onto the structural mesh and used in a 3-phase micromechanical model.

For all the methods listed here, there are commercial software tools available to perform the processing simulation. It is important to understand that the processing simulations and resulting outputs vary between the different methods, and that the micromechanical model may also be specific to the material. Regardless, the fundamental approach and overall situation is always the same. The critical questions to be asked before a coupled analysis is performed are:

  1. Does a processing simulation exist for the desired manufacturing method?
  2. Does this processing simulation deliver information about the local microstructure that can be used in micromechanical modeling?
  3. Does a micromechanical material model exist to describe the desired composite performance?

When all of these questions can be answered positively, at least a basic approach towards coupled analyses can be implemented.

For short fiber reinforced plastics (SFRP) this type of coupled analysis is a common procedure. Based on the level of accuracy, coupled stiffness analyses including temperature and strain rate dependencies have developed into a standard over the past years. The process can also be applied to long fiber reinforced plastics (LFRP) as well, though here the situation differs from that for SFRP as the processing tools do not yet fully predict the local microstructure in a standard procedure. Still though, based on state-of-the-art technology as offered today, it is already feasible to perform stiffness and failure analysis in a coupled solution. The underlying assumptions with this approach are that local fiber orientations are predicted correctly by the processing software and that the long fibers are straight and exhibit no bundling. To which extent these assumptions are valid requires further validation.

Composite design for unidirectional and woven composites starts with the material characterization on the material level with test coupons. Whereas on the coupon level the fiber orientation in each ply of the stacking is easy to describe and to incorporate in a simulation model, things become increasingly complex when turning to real parts. Such parts exhibit curved surfaces which will influence the local orientation of the fibers when produced via some manufacturing method. Typical procedures used for this are draping or fiber placement, which result in warp and weft angles of the fiber that can be fed into micromechanical models to set up fully coupled analyses. The material model must be sensitive to the fiber orientation in addition to accounting for the complex failure mechanisms of the composite.

Comparison of draping and fiber placement in a coupled analysis. The focus of the investigation is the amplitude of displacement in the buckling of a composite panel.

In conclusion, the design of composite structures is a challenging task. Man ufacturing, local microstructure, resulting composite properties, and final performance of the part are inter-dependent. To take full advantage of tailored composite properties for design optimization, all of these influences must be taken into account. This objective can be tackled by setting up multi-scale analyses based on micromechanical modeling which starts with the availability of microstructure prediction from processing software. Then a validation of both the microstructure predictions from the processing software as well as the material models for different kinds of fiber reinforcements and performances is required.

As all industries continue to introduce more lightweight structures, composite designs will be a fundamental part of future success. Using simulation techniques such as coupled multi-scale analyses ensures high performance design while further shortening development cycles. The current technology of the multi-scale approach is a strong and highly promising foundation to respond to the ongoing demand for advanced simulation methods.

This article was written by Robert Schmitz, business development engineer, e-Xstream engineering (Mont-Saint-Guibert, Belgium); Dr. Roger Assaker, founder and CEO, e-Xstream engineering and chief material strategist, MSC Software (Newport Beach, CA); and Jan Seyfarth, Digimat product manager, e-Xstream engineering. For more information, click here .