Design of Robotically Manufactured Composites for Optimal Structural Performance, Incorporating Manufacturing Constraints

Abstract

Robotic manufacturing of composites has revolutionized the aerospace industry. Traditional manufacturing of carbon fiber composite laminates involved manual hand lay-up of resin pre-impregnated sheets (prepregs) of pristine material. Such lay-up used to be cumbersome and time consuming, as well as inefficient. There is no repeatability of parts, and the method produced significant scrap material. The challenge significantly rises as the parts become larger and complex. Robotic Automated Fiber Placement (RAFP) and Automated Tape Laying (ATL) are the two robotic manufacturing techniques for CFRP laminates. While RAFP lays down “tows” of prepreg material, ATL is used for dry layup which shall require a resin infusion before curing. RAFP has started to be widely used in large scale manufacture of aerospace structures. With aircraft like the Boeing 787, Airbus A350-XWB and Airbus A220 having significant percentage of load bearing composite members, it is imperative to resort to faster and repeatable manufacturing techniques.

RAFP technology also opens up a design space that was previously not explored in traditional manufacture of laminates. An aircraft structural designer now has the capability to derive optimal fiber paths that could be steered, to be spatially varying based on the applied loads and boundary conditions. While the idea of steered fiber paths have been explored since the early 1990’s, there has been a recent interest in designing parts for optimal structural performance. It is noted that while RAFP has many benefits, the drawback of the technology is manufacturing induced defects like gaps, overlaps and wrinkles of the fiber “courses”, generally called the manufacturing signature (MS). In this work, explicit care has been given to incorporate parameters that drive the manufacturing signatures within the optimization framework, so as to produce realistic, manufacturable structural parts for improved structural performance. Some of the distinctive contributions of this work include- use of parametric curves to model center-lines of individual fiber paths, use of a global manufacturing mesh to reduce the number of optimization variables, explicitly incorporating MS into the finite element framework and including the geometrical changes arising due to compaction during manufacture, and an optimization framework in conjunction with a surrogate model built using machine learning algorithms.

Two design problems are studied - a flat plate under uni-axial and bi-axial, in-plane compressive loading, and a flat plate with an elliptical cut-out under in-plane tensile loading. The optimal designs for the uni-axial buckling are manufactured and studied for the manufacturing signatures using non destructive testing, and then subjected to in-plane compression to evaluate the laboratory performance to compare against analytical models. Further,a study on the optimal steered fiber paths is conducted for a rectangular plate with an elliptical cutout. Here the objective is to generate designs that incorporate the manufacturing signature and produce minimum stress concentration.