Most metal 3D printers currently available in the market is based on the powder bed fusion (PBF) process. In PBF, the part is built layer-by-layer using a direct energy source to melt and fuse a cross-section of the part into the layer of powder on the build platform. While PBF has gained prominence in applications ranging from automotive, marine, aerospace to medical fields, there are still issues that hinder its wide adoption. The inconsistency in the quality and reliability of PBF products is one of the critical limitations. The PBF products are susceptible to defects due to a) Variability in the size of powder particles; b) Complex heat transfer; c) Non-uniformity in metal powder deposition for a printing layer; d) Non-uniformity in the welded bead due to variability in laser power, chamber temperature, chamber gas concentrations; e) Local geometry of the part being printed; f) Printing parameters such as scanning speed, laser power.
The in-situ monitoring method for Powder Bed Fusion (PBF) process aims to provide real-time insights to the print quality of the part as the job progresses. The method entails the intelligent use of an optical sensor and an infrared (IR) sensor to acquire thermal and visual information of a localized region on the part being fabricated. Through machine learning and data analysis of the thermal map and optical image, surface anomalies and defect signatures (e.g. elevated areas, low energy input, or overheated regions) of the part being fabricated can be identified early for remedial actions.
The technology comprises of commercially available optical and IR sensors. The field of view (FOV), image resolution, IR wavelength and focal length are customised and calibrated according to the building chamber geometry of the particular PBF printer. Additionally, the integrated system is connected to a backend processor that performs sensors data fusion. Visual and thermal images are merged to produce a perception data that can be further processed and interrogated to validate against a referenced data set. Through machine learning and data analysis of the thermal map and optical image, surface anomalies and defect signatures of the part being fabricated can be identified.
The technology owner provides to collaborators the installation and calibration of the in-situ monitoring system, as well as a licensable software for quality management of the PBF products.
The Global Additive Manufacturing market, including hardware software, materials and services, stands at $9.3 billion in 2018, and is projected to reach $41.6 billion in 2027. Printing large components like propellers can take days or even weeks to complete. If anomalies occurs during the printing process affecting part quality, the whole part may have to be discarded. According to published cost models, the risk related cost due to build failure is the second largest cost, occupying 26% of the total unit cost. This highlights that process instability can severely affect the overall value proposition of AM. This technology thus has the potential to minimize risk-related costs and significantly reduce the production costs of the PBF process.