Complex composites structures are becoming more common, especially in aeronautics. Composites preforms can now be manufactured by automated processes, such as automated fiber placement (AFP) and automated tape laying (ATL), to achieve the expected production volumes.

The AFP process is the most interesting since it can address complex double-curvature layup surfaces with good productivity rates and low material scrap, according to Coriolis Software. The main advantages of this technology are that material can be steered to address the problem of curvature, and fiber angular deviation from the engineering rosette can be easily managed. This opens interesting possibilities to optimize the design of composites structures and to change the current traditional “black metal” approach of design offices.

But these benefits also induce new problems to solve in terms of potential manufacturing defects, such as wrinkles, bridging, gaps, and overlaps. The offline programming system used to build the tape courses over the layup surface should be able to predict these defects but also help find an optimal draping strategy.

These systems should integrate all the programming features at the design stage, allowing the design engineers to make smart proposals to NC programmers, validated by stress analysis, and hence eliminating useless loops and trials.

Researchers from Coriolis used an aerospace use case to illustrate the required software features for optimal fiber-placement programming. Through the collaborative research project “Defi Composite,” Coriolis Composites and SAFRAN Aircelle worked together on a thrust reverser component demonstrator made with carbon-fiber-reinforced plastic (CFRP) material and an AFP process. This component is an inner fixed structure (IFS), made of two monolithic carbon skins and a core.

The goal of this project was to highlight the benefits of an AFP process compared to manual hand layup, in terms of layup quality and productivity rate.

Multiple constraints for optimal design

The most important thing to understand with AFP programming is that both engineering and manufacturing specifications need to be managed, and are often in contradiction. That’s why the software system should be designed to address not only NC programmers, but also composites designers, project leaders, and stress engineers.

Five constraints are identified with a key role for optimal design with an automated manufacturing process: fiber orientation, isotropy, fiber steering radius, fiber gaps and overlaps, and the manufacturing process. Coriolis presents an implemented solution for each constraint.

For example, the offline programming system offers the possibility to analyze isotropy, since the maximum fiber deviation constraint alone can be too restrictive. Isotropy is the measure of the relative fiber angle between two layers at a same measurement point, for example between a 0° and a 90° ply.

A specific development was done to export fiber angle deviation for each layer from the programming courses, in order for the stress engineer to validate fiber orientation and isotropy. This software add-on is integrated on top of the offline programming system to ease communication between designers and stress engineers.

The main steps for capturing the fiber deviation and isotropy for structural analysis are: 1.) Mapping of fiber centerlines on FE shell mesh, 2.) Core sampling at finite element centroid point, 3.) For each layer of the stacking sequence, get the closest fiber direction and evaluate angle deviation from the cartesian rosette, 4.) Writing of a composites shell property with “true” material angle, for each mesh element, and 5.) Factorization of stacking sequence using a custom angular tolerance, in order to minimize the number of composites properties.

Concerning fiber steering radius, the constraint is mainly caused by fiber in-plane curvature and material width, and may lead to wrinkles and tow buckling. This kind of defect is not perfectly well understood in terms of direct impact on structural strength.

Fiber steering is very complicated to interpret, since it depends on other parameters such as material tackiness, compacting pressure of roller, layup velocity, heating power, and intermediate vacuum bagging. For this reason, fiber steering radius analysis at the design stage should always be correlated to real sample tests in order to get the empirical knock-down factors.

The software system is able to detect this problem at least through a geometrical analysis of the in-plane curvature of tow centerline curve, in order to highlight critical areas. The designer can also easily do “what if” scenarios with different material widths (1/4-in or 1/8-in) or types (thermoset, dry fiber).

Steering issues can be solved at the design stage by either using a geodesic path approach—for example, through a complex rosette transfer—or by locally decreasing the layup velocity. This can be achieved by defining a local “process contour” in the software system (also called manufacturing strategy), after identifying the highly steered areas over the surface, and then applying a lower nominal machine velocity.

Experimental trials showed that it can significantly improve the final layup quality.

Automatic ply boundary splicing

To optimize the design for AFP, one solution is to split the ply boundary in several pieces, called sectors or regions. This task can be manually achieved, but it’s time consuming, and more important, it will not be easy to obtain an optimal result.

The optimal result of the problem is to minimize the number of sectors and tape courses within the ply, and to avoid too many material feeds and cuts (leading to slow layup velocity).

A specific software module was implemented in that way, to automate the sector and guide curves creation, starting from the following initial conditions: geometrical layup surface, engineering rosette axis, ply contour, initial seed-point, material width, maximum number of tows by course, maximal angular deviation, and minimum steering radius.

The system is based on an iterative algorithm, building a fixed angle path and parallels, and then cutting the ply boundary if the parallels present excessive angular deviation. The output result is a list of sectors, each made of a closed contour and one or several guide curves.

After many trials and experimental validations, the system was improved to integrate the following specifications:

• Avoid ply cuts and tow drop off located in a specified zone (example: stringer bonding area)

• Adjust the tape width (number of tows) if roller conformance is not compliant (excessive crush or fiber bridging defects)

• Minimize the gap between adjacent tape courses

• Management of sequence of several plies in the same layer (patch areas).

Compared to a traditional approach (ply with one single guide curve) or a manual ply sectors method, the automatic method proposal provides significant benefits. The major benefits include the programming time (limited required skills, possibility to manage iterative design loops) and the convergence to an optimal result, very difficult or impossible with other methods because of the large number of parameters and constraints.

A solution was successfully implemented on a complex aerospace case, with competitive results compared to a manual hand layup process in terms of structural strength, layup quality, and productivity rate. Improvements will be made for extending the automatic method to manage geodesic path as sector guide curve, and control the staggering distance of tow drop off areas.

This article is based on SAE International technical paper 2014-01-2261 by Yvan Blanchard of Coriolis Software. The paper is part of the “Manufacturing/Materials/Structures – Product Design and Manufacturing Integration” session taking place at the SAE 2014 Aerospace Manufacturing and Automated Fastening Conference & Exhibition.