AUTOMATED COMPOSITES

The most straightforward method to inspect the fidelity of a composite part is manually, via one or more knowledgeable inspectors. Inspectors must be intimately familiar with the manufacturing specifications, allowable and tolerances.

With this knowledge, they can use calipers, scales, magnifying glasses and other tools to verify the part produced matches the necessary parameters.

While straightforward, this process becomes more complicated when applied to large parts. For example, manual inspection of an aircraft fuselage or wing skin could take multiple inspectors considerable time. Inspection and rework are the most significant contributors to the time a part requires to be produced.

For a given aircraft fuselage production, over half of the time spent in the fabrication work cell was dedicated to inspection and rework of process failures, as seen above.

Another concern for manual inspection regards accuracy. For many industrial automated composites manufacturing hardware vendors, process failure rates are approximately one in every three thousand events. Over the entire process of a part’s production, for commercial aircraft fuselage sections, there could be up to 180,000 events.

From this perspective, ensuring that quality inspectors catch all three thousand process failures, and their resulting defects during part production every time is a daunting requirement.

Because manual inspection relies heavily on human elements, inspection results can be subjective and non-repeatable. Mistakes can and will be made, given enough opportunity.

Unfortunately, many industries that use composites, including the aerospace, automotive and naval industries, cannot tolerate part failure. Due to the inherent high factor of safety, validating quality is a requirement and something that human inspection is ill-equipped to do, repeatedly.

Each time a mandrel or form is mounted on a machine, it may be necessary to probe a few locations on the form to determine its exact location. The information collected in this process is typically used to adjust the datum of the machine’s coordinate system, before laying material on the form. A probe, usually a ball of precisely known diameter, is mounted on the machine’s head, and a program drives it around a sequence of positions. At each probed location, there is a hole of known size or other locating features. Once the probe is close to the feature, a probing subroutine or macro is invoked to determine the feature’s precise machine coordinates.

There are alternatives to an automated probing location process, but by comparison, they take much longer and rely heavily both on human expertise and iterative methods.

Other assisted quality processes exist, as well. Long-throw laser projectors as used to aid in the faster location of critical part features. Overhead projectors help quality inspectors quickly visualize the designed edge of part boundaries to make measuring against this datum more straightforward. Planned and programmed overlaps of material are identified so that inspectors can confirm the material was placed as desired. In this way, material overlaps not identified by the laser projector are determined to be process defects, and necessitate rework efforts.

In-process inspection lies on the other end of the spectrum from fully manual efforts. Scanning instruments, usually profilometers, are attached to the same motion platform as deposits material on the form.

As the material is deposited, the profilometers reside directly behind the point of material placement and can, therefore, scan the just-placed material. No other machine motion is required to scan the composite part, which significantly reduces inspection time.

Point cloud inspection data generated by the scanners are analyzed in real-time for manufacturing artifacts, including overlaps, gaps, FOD, briding, wrinkling, among others. Detected conditions are compared with the design intent to determine whether issues exist.

As in-process inspection continues to grow in capability and maturity, some composite manufacturers are taking a statistical approach to quality. Instead of manufacturing specifications that require quality inspectors to rework every detected manufacturing artifact, some firms are choosing to rework process artifacts until some statistical threshold is met.

This process differs significantly from traditional rework practices, many of which required every detected artifact and defect to be rectified before proceeding.

The additional insight provided by real-time data gathering and scanning produces a higher level of visibility into the part’s production as a whole. These statistical decision-making approaches are validated with FEA analysis, which can determine the knock-down factors required for the detected defects.