Simulating Long-Fiber Composites

Although the use of injection-molded long carbon fiber thermoplastic composites is limited in automotive, this is understandable for several reasons. For example, Leo Fifield, material scientist at the Department of Energy’s Pacific Northwest National Laboratory (PNNL; pnnl.gov), which has been undertaking research—with industry (e.g., Toyota, Magna), technical (e.g., PlastiComp [plasticomp.com], Autodesk  [Autodesk.com]) and academia (University of Illinois, Purdue University, Virginia Polytechnic Institute and State University)—says that they’ve been working on predictive engineering (PE) tools for the materials.

The reason? “Current development processes of composite components require carmakers to build molds, mold parts and test them. It’s a long, arduous process, slowing the advance of new, more cost-effective carbon fiber composites in automobiles,” says Fifield.

Why long fiber?
“Short fiber reinforced thermoplastic materials contain reinforcing fibers of various short lengths that are randomly orientated in the resin/pellet,” explains Eric Wollan, chief operating officer for PlastiComp, Inc. In “long” fiber materials, “the individual reinforcing fibers are aligned with respect to each other and are exactly as long. Long-fiber composites combine high levels of stiffness, strength and toughness in a single material. They can deliver superior, predictable mechanical performance compared to short fiber reinforced thermoplastic materials.”

In the PNNL research project, the carbon fiber pellets began ½ inch (12-13 mm) long. The fibers were manufactured through a pultrusion process (pulled extrusion), so the fibers were the full length of the pellet; however, after melt, extrude and injection in the mold, the fibers measured 4 mm or less. (“Short” fibers measure below 4 mm to start, and typically 1 mm 
after extrusion.)

Stiffness, strength and impact properties differ greatly between short and long fibers.

However, explains Fifield, “around 1 mm, you start to get close to the asymptotic improvement you might expect from length for stiffness of the parts. A 2-mm length fiber is not going to be a lot stiffer than a 1-mm length fiber because you’re getting in that regime of ‘long-ness.’”

Physically, short fibers in an injection molded part basically act within a plane. They are not long enough to flex and intertwine with surrounding fibers. Conversely, longer segments of fiber tend to intertwine three-dimensionally within the molded part. These longer segments help distribute pressure loading from one area of a part to another. That’s what gives long carbon fiber (LCF) parts stronger properties, the boost in toughness and impact resistance.

Simulating all this is “a challenge.” Fibers don’t intertwine exactly in the same pattern every time they’re injected into a mold. While they tend to align in the direction the material is moving, that’s not a 100-percent, perfect alignment. Plus, the thicker the part, the more random the fiber orientation. Current algorithms for molding simulations are not yet optimized for all of this.

Ironically, says Marc-Henry Wakim, senior design engineer at PlastiComp, another challenge is that long-fiber gives engineers “new design freedoms, which don’t exist with regular plastics.” For instance, fiber filling enables parts with variable thicknesses, but that alters fiber orientation. Many existing algorithms don’t work as well with parts having variable thicknesses. This is especially true with complex parts, such as those with ribs measuring 1-mm wide. Says Fifield, “We’re talking about 1- or 2-mm fibers. The behaviors change.” For instance, breaking strength becomes non-linear while stiffness becomes more linear. 

Generally, engineers design parts using long fibers by starting with the geometry they want to work with (in computer-aided design), replace the metals parts with plastic parts, add reinforcements or remove geometry as they saw fit, then run the design through finite element analysis (FEA). The engineers build in safety factors, as well as compare weld line areas to the FEA results to ensure weld lines or poor fiber orientation don’t exist in high-stress areas.

As necessary, explains Wakim, the engineers “manipulate the geometry to move those threats into other areas of the part that are less stressed, or they come up with innovative ways to remove those weld lines or poor fiber orientations altogether.”

The research team used Autodesk Moldflow (autodesk.com/products/moldflow) for the plastics molding simulation software. “The technology had been developed before for short fibers and glass fibers,” says Fifield. “We just ran it through its paces to try it out.” Moldflow predicted fiber-length distribution and fiber orientation based on inputs including material properties, fiber length and molding conditions (such as the fiber length distribution at the nozzle going into the mold). They also used a PNNL-developed module (EMTA-NLA) in Abaqus FEA from Dassault Systèmes that used the length and orientation properties to predict the stiffness of the molded part.

Concludes Fifield, “Using the engineering software validated by the PNNL-lead team, manufacturers will be able to ‘see’ what the structural characteristics of proposed carbon fiber composites designs would be like before it’s molded. The tools allow manufacturers and auto part designers to experiment and explore new ideas at a much faster rate.”