2023 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2023): Process Development: Powder Bed Fusion Process
Program Organizers: Joseph Beaman, University of Texas at Austin
Wednesday 1:10 PM
August 16, 2023
Room: 400/402
Location: Hilton Austin
Session Chair: Lianyi Chen, University of Wisconsin-Madison
1:10 PM
Revealing and Controlling Process Instabilities for Defect Lean Metal Additive Manufacturing: Lianyi Chen1; 1University of Wisconsin-Madison
The process instabilities intrinsic to the localized laser-powder bed interaction cause the formation of various defects in laser powder bed fusion (LPBF) additive manufacturing process. In this talk, I will first present our research on revealing the process instabilities by in-situ high-speed high-energy synchrotron x-ray imaging and diffraction. Then, I will introduce the defects mitigation approaches we developed based on the new mechanisms we discovered and the new insights we obtained from the in-situ characterization study for achieving defect lean metal additive manufacturing.
1:30 PM
Acoustic Mechanisms of Laser Powder Bed Fusion through an Analogous Whistle Model: Yuchen Sun1; Aiden Martin1; Sanam Gorgannejad1; Jenny Wang1; Maria Strantza1; Nicholas Calta1; 1Lawrence Livermore National Laboratory
In laser powder bed fusion (LPBF), various non-destructive in situ diagnostic techniques are being developed including pyrometry, plasma spectroscopy, and acoustic monitoring, where acoustic monitoring represents an inexpensive, low spatial footprint, and computationally lightweight approach. Recent works have correlated acoustic signals from the printing process with print parameters and overall print quality. This work furthers the correlative understanding of acoustic monitoring and introduces a mechanistic understanding of acoustic generation in LPBF. We first present time-resolved frequency-domain acoustic data of Ti64 to demonstrate its correlation with laser energy density. We then perform non-dimensionalized analyses using Strouhal’s number and analogize the LPBF vapor depression to hole tone systems—exemplified by steam kettle whistles. This comparison reveals two mechanisms of acoustic generation: Helmholtz resonance at lower laser powers, duct resonance and vortex shedding at higher laser powers. This mechanistic understanding of vapor depressions further demonstrates the diagnostic capacity of acoustic monitoring in LPBF.This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
1:50 PM
Surface Roughness Formation and Measurement for Metals Built with Laser Powder Bed Fusion: Edwin Glaubitz1; Joy Gockel1; Jason Fox2; Orion Kafka2; Claire Casey1; 1Colorado School of Mines; 2National Institute of Standards and Technology
As-Built surfaces in laser powder bed fusion (PBF-LB) have rougher and more complicated surfaces than those from conventional techniques. Several surface features are unique to additive processes: layering artifacts, adhered particles, and large valleys. Surface valleys have been shown to negatively influence the fatigue life by serving as locations for crack initiation, while adhered particles can obstruct measurement of mechanically relevant surface features from contact and optical measurement techniques. Standard PBF-LB metals such as nickel superalloy 718, AlSi10Mg, and stainless steel 316L are built with a wide range of contour parameters and across multiple types of machines. Contour melt pool size and shape are measured and characterized. The surfaces are measured using laser confocal scanning microscopy and X-ray micro-computed tomography. Understanding of relationships between contour melting behavior and resulting surface roughness will inform process parameter development and improve correlations between measurement and part performance.
2:10 PM
Pushing Boundaries: Machine Learning Applied to Selective Laser Melting: Mary Daffron1; Steven Storck1; Brendan Croom1; Timothy Montalbano1; Salahudin Nimer1; 1Johns Hopkins University Applied Physics Lab
A primary concern in advancing selective laser melting (SLM) is developing a method to rapidly establish laser parameters for the end application. Establishing a set of processing parameters is complicated by the infinite number of possible combinations of machine variables. Additionally, certain processing domains can be unstable resulting in failure to extract valuable statistical information. Leveraging total machine capacity while balancing laser processing parameters is vital to the scalability of AM at quality. Intelligent parameter development via machine learning techniques unlocks the full processing space to enable unrealized potential including application specific properties. Rapid characterization techniques combined with strategic evaluation of microstructure inform the machine learning model and parameter development process. This results in a thorough exploration of the processing space in fewer build cycles. Examples will be presented showing the identification of new processing domains with increased density and targeted mechanical behavior optimization compared manufacturer recommended parameter sets.
2:30 PM Cancelled
Quality Prediction of AM Processes Using Volumetric CNNs with Spatialized Representations of Structure-borne Sound Sensor Data: Jork Groenewold1; Lukas Weiser1; Florian Stamer1; Gisela Lanza1; 1wbk Institute of Production Science, Karlsruhe Institute of Technology (KIT)
The low reproducibility of process quality is a challenge in additive manufacturing processes like powder bed fusion by laser beam melting (PBF-LB/M). Therefore, in-process monitoring is often used to detect process defects such as pores. One possible measurement technique are structure-borne sound sensors, which allow the laser-induced sound emissions to be monitored. Currently, the measured signals are evaluated individually for each manufactured layer. A key drawback of this approach is that the spatial arrangement of the measured data over multiple layers is not considered. This paper presents a new approach that aims to improve model performance by better exploiting the spatial context within the data. The approach consists of three steps: (1) data-preprocessing for three-dimensional assembly (2) optimization of a volumetric convolutional neural network and (3) training of models. The overall goal is to enable the models to close a quality control loop and thus enable reproducible process quality.
2:50 PM Break
3:20 PM
Localized Porosity Prediction in Laser Powder Bed Fusion via Deep Learning of Multi-modal Melt Pool Signatures: Haolin Zhang1; Chaitanya Krishna Vallabh2; Alexander Caputo3; Richard Neu3; Xiayun Zhao1; 1University of Pittsburgh; 2Stevens Institute of Technology; 3Georgia Institute of Technology
Laser powder bed fusion (LPBF) additive manufacturing utilizes laser to sinter or melt powders for fast production of complex parts. However, due to the complex interplays among laser, powder, printed part, and gas flow, the LPBF process tends to generate severe defects such as pores, which are detrimental to the final part performance. In this work, we develop a deep learning aided porosity prediction framework utilizing in-situ monitored melt pool signatures including multiple thermal, geometrical, and spatter metrics that are derived from our high-speed on-axis single-camera two-wavelength imaging pyrometer and an off-axis camera jointly. Scalograms, transformed from the obtained time-domain MP signatures are used as input to train deep convolutional neural network models for correlating to ex-situ porosity characterization data from X-ray computed tomography. The developed method is shown to be capable of quantifying localized porosity and holds promise to qualify LPBF processes and parts.
3:40 PM
Investigation of the Influence of Process Parameters to Increase Productivity in the LPBF Process for the Material Inconel 718: Christian Bodger1; Stefan Gnaase1; Dennis Lehnert1; Thomas Tröster1; 1Paderborn University
The nickel-based alloy Inconel 718, which is used in aerospace technology, poses a great challenge to machining due to its high work hardening and toughness. Here, the LPBF process offers an alternative with potantial savings if sufficiently high productivity can be achieved. Based on the parameter study carried out, starting from the standard parameters for the production of components exposure parameters could be developed to realize manufacturing with 120 μm and 150 μm layer thickness. For this purpose, the process parameters of laser power, focus diameter, hatch distance and scan speed were varied. The negative defocusing of the laser showed a positive effect on the density of the parts, realizing densities ≥ 99.94 %, with high dimensional stability and good mechanical properties. Considering the reduced manufacturing time of up to 61 %, a significant increase in productivity was achieved.
4:00 PM
High Frequency Ultrasonic Detection of Security Markers in Additive Manufacturing Components: Farhang Honarvar1; Sagar Patel2; Katayoon Taherkhani2; Peyman Alimehr2; Mihaela Vlasea2; 1K. N. Toosi University of Technology; 2University of Waterloo
Embedding security markers in additive manufacturing (AM) components have been investigated in literature as means to ensure product authentication and mitigate risks of counterfeiting and reverse engineering. These security markers have been previously detected using X-ray computed tomography (XCT), which is a method also commonly used for defect detection in AM. Ultrasonic testing (UT) has been used for offline detection of defects in AM, but UT has not been implemented for complex security markers such as QR codes in AM. In this work, UT and image processing are used for the first time in successfully detecting QR codes in laser powder bed fusion (LPBF) parts, with feature sizes as low as 0.5 mm. Additionally, UT is also implemented as means to detect and differentiate LPBF parts with lack of fusion and keyhole mode defects, which presents a low-cost alternative to XCT for defect detection in AM.