A measurement chain, not another printer Fraunhofer IMWS published AddiQ results on 15 October 2026 aimed at in-process quality assurance for large-format Fused Granulate Fabrication. The project does not claim that sensors qualify a part by themselves. Its objective is to connect measured machine, material and environmental data with appropriate coupons and finished-part tests.

Fraunhofer IMWS worked with SLV Halle, IKTR and studio.201. A Yizumi pellet printer was instrumented to measure acceleration, ambient temperature, granulate temperature and humidity, melt pressure and mass on the build platform. Each signal is sensitive to different mechanisms, so they cannot be collapsed into one generic machine-health number.

What is measured and why? Granulate moisture and temperature can affect melt behaviour, gas formation and interlayer bonding. Melt pressure may respond to feeding, viscosity, blockage or tool wear. Acceleration can reveal vibration and mechanical anomalies, while measured platform mass can be compared with theoretical material consumption. Ambient temperature is one background variable in cooling and warpage of a large part.

Fraunhofer's central point is that data must be linked to the resulting geometry and mechanical properties. A pressure spike is not automatically a reject. The team first needs evidence that a signal pattern repeatedly accompanies a defined defect, then a decision threshold that manages both false alarms and missed defects.

Nozzle wear and maintenance The team also developed a retrofittable retract system and automated image evaluation for the nozzle. Worn or damaged outlet geometry can change bead shape, deposition width and flow stability. Periodic imaging becomes maintenance evidence only when lighting, magnification, position and the evaluation rule are controlled.

Predictive maintenance here should mean a relationship between condition trend and print quality, not replacement by calendar. The public summary reports no detection accuracy, alarm threshold or false-positive and false-negative rate. The method is therefore a demonstrated research result, not a universal off-the-shelf qualification product.

Scope of the rail demonstrator The project used a 500 mm-long version of an ICE train front apron, scaled to 38 percent of the original 1,300 mm component because the full part exceeded the available build volume. This is a relevant geometric demonstrator, but it does not automatically prove compliance for a full-size part exposed to service loads, weather, fire requirements and railway approval rules.

The stated next step is application-specific qualification of customer components with industrial partners. The value of the result is a possible evidence architecture, not a claim that every FGF spare part is now releasable.

Practical consequence A factory should not begin by installing the maximum number of sensors. It should first name the critical quality characteristic — dimension, interlayer strength, mass or a surface defect — and then choose a process signal with a physical link to it. Signal, coupon and finished-part inspection together form the evidence.

Platform mass can provide a rapid material-balance check but cannot locate an excess or deficit. Melt pressure may be fault-sensitive, but material viscosity and temperature also change it. Variables therefore need synchronised timestamps, calibrated measurement chains and a version-controlled recipe.

LAYERTIMES assessment AddiQ's strongest result is the deliberate link between process data and property testing. That is more meaningful than the mere presence of a camera or sensor. The public material does not provide sample count, uncertainty, predictive performance or qualification of a full-size rail component. The technology is a promising QA foundation that still requires validation for each material, machine and geometry combination.