The Need

Powder Bed Fusion (PBF) is a subset of additive manufacturing processes that performs a layer-by-layer fabrication of metal components by selectively melting metal powder disbursed over the earlier layer. PBF processes encompass methods such as selective laser sintering (SLS), direct metal laser sintering (DMLS), and electron beam melting (EBM). The market for PBF printers is expected to reach USD 1.85 billion by 2023 with a estimated compound annual growth rate (CAGR) of 24.1%.

The market for PBF technologies is growing so quickly largely because PBF manufacturing can produce intricate parts in a variety of materials with superior mechanical properties to those manufactured by traditional methods. However, while their resultant strength may be higher, PBF parts tend to also display high levels of residual stresses, porosity, and anisotropy. These defects are the direct result of disparate cooling rates of the part as it is printed. In tandem with the high thermal gradient within the build chamber, the variable cooling rates create a temperature gradient within the part itself that influences its metallurgical phase formation, and in turn its mechanical properties. The high thermal stresses may also induce warping and heat-induced distortion of the part. Despite the importance of temperature control in PBF components, most PBF printers run without advanced process monitoring tools or feedback. While some infer cooling rates from temperature measurements of the melt pool, build chamber, or printer bed, all are limited to measuring surface temperature only, and methods estimating internal temperature accurately are needed to improve build success rates and component quality.

The Technology

Prof. David Hoelze and Graduate Research Associate Nathaniel Wood have developed a process-monitoring algorithm to estimate the internal temperatures of components manufactured through PBF in near-real time. The algorithm is a type of ensemble kamlan filter (EnKF), a subclass of state-observers that modifies a model of the internal temperature based on measurements of process inputs and outputs such as laser power and melt pool temperature. This approach forces the model's predictions to match the process realities thereby improving the model’s accuracy. The inventors have adopted the EnKF to the realities of PBF, modifying its internal structure to accommodate uncertainty inherent in the PBF process inputs.

Designed to work with the current generation of PBF printers, the algorithms' temperature field estimations offer comparable accuracy with a greatly reduced need for expansive training data sets relative to current methods. Accurate temperature predictions from the model allow a higher degree of process monitoring and control than previously available, such as near-real time fault detection and microstructure prediction during the manufacture of components.

Commercial Applications

• Signal processing tool for PBF system manufacturers
• Quality control device for near-real time detection of faults and out-of-specification prints

Benefits

• Near-real time modeling of internal temperature gradient
• Adaptable to existing PBF printers
• Comparably small amount of necessary calibration

1 BCC Research. (2018). Global Markets for 3D Printing. (Report Code IAS102C). Retrieved from https://www.bccresearch.com/market-research/instrumentation-and-sensors/global-markets-for-3d-printing.html