Why AFP Part Quality Changes at Corners and Course Ends

Constant power plus variable speed equals variable heat input. The machine log says the course ran fine; the laminate disagrees.

In automated fiber placement, the heat source delivers constant power — but the placement head does not move at constant speed. That mismatch is one of the most common and least discussed causes of quality variation within a single part.

What actually changes with feed rate?

The governing relationship is simple: faster movement means less exposure time to the heat source. This holds whether the source is a laser, a hot gas torch, an infrared heater or pulsed light.

Head velocity is not constant in practice. It varies during:

Acceleration at the start of every course
Deceleration at the end of every course
Curved paths, where the head slows to hold the programmed tolerance
Short courses, which may never reach commanded speed at all

With power held constant, each of those regions receives a different energy density at the nip point than the straight, at-speed middle section of the same course.

Why does this affect part quality?

Changing the exposure time changes both conduction into the substrate and advection along the placement direction. The result is a temperature gradient along the plied layers rather than a uniform thermal history.

Because degree of consolidation and interlaminar bond strength both depend on the temperature reached and the time held there, that thermal gradient becomes a mechanical property gradient. The part is inhomogeneous even though every course was placed exactly as programmed.

How is it compensated?

Three approaches are common, and they are not equally good:

Slow the whole course to worst-case speed. Simple, effective, and expensive — it discards throughput everywhere to fix a problem that exists in a few zones.
Fixed power derate in known accel and decel zones. Cheap to implement, but open loop: it assumes the zones and the correction are the same on every part.
Power modulated as a function of commanded feed rate. The correct approach in principle, and the one that keeps cycle time. It requires knowing the right power-versus-velocity relationship for your material and head geometry.

Where simulation helps

The power-versus-velocity relationship can be derived rather than tuned by trial. Simulating the optical and thermal behaviour across the velocity range gives the compensation curve directly, before any material is placed — which is considerably cheaper than discovering it through scrapped parts. See real-world simulation and engineering services.