skip to main content

This content will become publicly available on December 4, 2024

Title: Generating synthetic as-built additive manufacturing surface topography using progressive growing generative adversarial networks
Numerically generating synthetic surface topography that closely resembles the features and characteristics of experimental surface topography measurements reduces the need to perform these intricate and costly measurements. However, existing algorithms to numerically generated surface topography are not well-suited to create the specific characteristics and geometric features of as-built surfaces that result from laser powder bed fusion (LPBF), such as partially melted metal particles, porosity, laser scan lines, and balling. Thus, we present a method to generate synthetic as-built LPBF surface topography maps using a progressively growing generative adversarial network. We qualitatively and quantitatively demonstrate good agreement between synthetic and experimental as-built LPBF surface topography maps using areal and deterministic surface topography parameters, radially averaged power spectral density, and material ratio curves. The ability to accurately generate synthetic as-built LPBF surface topography maps reduces the experimental burden of performing a large number of surface topography measurements. Furthermore, it facilitates combining experimental measurements with synthetic surface topography maps to create large data-sets that facilitate, e.g. relating as-built surface topography to LPBF process parameters, or implementing digital surface twins to monitor complex end-use LPBF parts, amongst other applications.  more » « less
Award ID(s):
2309483 1752069 2322322
Author(s) / Creator(s):
; ;
Publisher / Repository:
Date Published:
Journal Name:
Subject(s) / Keyword(s):
["additive manufacturing","surface topography","synthetic surface topography","generative adversarial \nnetworks"]
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    Despite its potential to overcome the design and processing barriers of traditional subtractive and formative manufacturing techniques, the use of laser powder bed fusion (LPBF) metal additive manufacturing is currently limited due to its tendency to create flaws. A multitude of LPBF-related flaws, such as part-level deformation, cracking, and porosity are linked to the spatiotemporal temperature distribution in the part during the process. The temperature distribution, also called the thermal history, is a function of several factors encompassing material properties, part geometry and orientation, processing parameters, placement of supports, among others. These broad range of factors are difficult and expensive to optimize through empirical testing alone. Consequently, fast and accurate models to predict the thermal history are valuable for mitigating flaw formation in LPBF-processed parts. In our prior works, we developed a graph theory-based approach for predicting the temperature distribution in LPBF parts. This mesh-free approach was compared with both non-proprietary and commercial finite element packages, and the thermal history predictions were experimentally validated with in- situ infrared thermal imaging data. It was found that the graph theory-derived thermal history predictions converged within 30–50% of the time of non-proprietary finite element analysis for a similar level of prediction error. However, these prior efforts were based on small prismatic and cylinder-shaped LPBF parts. In this paper, our objective was to scale the graph theory approach to predict the thermal history of large volume, complex geometry LPBF parts. To realize this objective, we developed and applied three computational strategies to predict the thermal history of a stainless steel (SAE 316L) impeller having outside diameter 155 mm and vertical height 35 mm (700 layers). The impeller was processed on a Renishaw AM250 LPBF system and required 16 h to complete. During the process, in-situ layer-by-layer steady state surface temperature measurements for the impeller were obtained using a calibrated longwave infrared thermal camera. As an example of the outcome, on implementing one of the three strategies reported in this work, which did not reduce or simplify the part geometry, the thermal history of the impeller was predicted with approximate mean absolute error of 6% (standard deviation 0.8%) and root mean square error 23 K (standard deviation 3.7 K). Moreover, the thermal history was simulated within 40 min using desktop computing, which is considerably less than the 16 h required to build the impeller part. Furthermore, the graph theory thermal history predictions were compared with a proprietary LPBF thermal modeling software and non-proprietary finite element simulation. For a similar level of root mean square error (28 K), the graph theory approach converged in 17 min, vs. 4.5 h for non-proprietary finite element analysis. 
    more » « less
  2. Abstract Laser powder bed fusion (LPBF) was utilized to create a series of aluminum alloy (i.e., AlSi10Mg) 5 mm-diameter support pillars with a fixed height of 5 mm containing varying filet angles and build orientations (i.e., 0 deg, 10 deg, 20 deg, 30 deg, 40 deg, 50 deg, and 60 deg from the normal surface) to determine surface roughness and water wettability effects. From experiments, anisotropic wetting was observed due in part to the surface heterogeneity created by the LPBF process. The powder-sourced AlSi10Mg alloy, typically hydrophobic, exhibited primarily hydrophilic behavior for build angles of 0 deg and 60 deg, a mix of hydrophobic and hydrophilic behavior at build angles of 10 deg and 20 deg, and hydrophobic behavior at 30 deg, 40 deg, and 50 deg build angles. Measured surface roughness, Ra, ranged from 5 to 36 µm and varied based on location. 3D-topography maps were generated, and arithmetic mean heights, Sa, of 15.52–21.71 µm were observed; the anisotropy of roughness altered the wetting behavior, thereby prompting some hydrophilic behavior. Build angles of 30 deg and 40 deg provided for the smoothest surfaces. A significantly rougher surface was found for the 50 deg build angle. This abnormally high roughness is attributed to the melt pool contact angle having maximal capillarity with the surrounding powder bed. In this study, the critical melt pool contact angle was near equal to the build angle, suggesting that a critical build angle exists, which gives rise to pronounced melt pool wetting behavior and increased surface roughness due to enhanced wicking followed by solidification. 
    more » « less
  3. The goal of this work is to quantify the link between the design features (geometry), in-situ process sensor signatures, and build quality of parts made using laser powder bed fusion (LPBF) additive manufacturing (AM) process. This knowledge is critical for establishing design rules for AM parts, and to detecting impending build failures using in-process sensor data. As a step towards this goal, the objectives of this work are two-fold: Quantify the effect of the geometry and orientation on the build quality of thin-wall features. To explain further, the geometry-related factor is the ratio of the length of a thin-wall (l) to its thickness (t) defined as the aspect ratio (length-to-thickness ratio, l/t), and the angular orientation (θ) of the part, which is defined as the angle of the part in the X-Y plane relative to the re-coater blade of the LPBF machine. Assess the thin-wall build quality by analyzing images of the part obtained at each layer from an in-situ optical camera using a convolutional neural network. To realize these objectives, we designed a test part with a set of thin-wall features (fins) with varying aspect ratio from Titanium alloy (Ti-6Al-4V) material — the aspect ratio l/t of the thin-walls ranges from 36 to 183 (11 mm long (constant), and 0.06 mm to 0.3 mm in thickness). These thin-wall test parts were built under three angular orientations of 0°, 60°, and 90°. Further, the parts were examined offline using X-ray computed tomography (XCT). Through the offline XCT data, the build quality of the thin-wall features in terms of their geometric integrity is quantified as a function of the aspect ratio and orientation angle, which suggests a set of design guidelines for building thin-wall structures with LPBF. To monitor the quality of the thin-wall, in-process images of the top surface of the powder bed were acquired at each layer during the build process. The optical images are correlated with the post build quantitative measurements of the thin-wall through a deep learning convolutional neural network (CNN). The statistical correlation (Pearson coefficient, ρ) between the offline XCT measured thin-wall quality, and CNN predicted measurement ranges from 80% to 98%. Consequently, the impending poor quality of a thin-wall is captured from in-situ process data. 
    more » « less
  4. Additive manufacturing (AM) as a disruptive technique has offered great potential to design and fabricate many metallic components for aerospace, medical, nuclear, and energy applications where parts have complex geometry. However, a limited number of materials suitable for the AM process is one of the shortcomings of this technique, in particular laser AM of copper (Cu) is challenging due to its high thermal conductivity and optical reflectivity, which requires higher heat input to melt powders. Fabrication of composites using AM is also very challenging and not easily achievable using the current powder bed technologies. Here, the feasibility to fabricate pure copper and copper-carbon nanotube (Cu-CNT) composites was investigated using laser powder bed fusion additive manufacturing (LPBF-AM), and 10 × 10 × 10 mm3 cubes of Cu and Cu-CNTs were made by applying a Design of Experiment (DoE) varying three parameters: laser power, laser speed, and hatch spacing at three levels. For both Cu and Cu-CNT samples, relative density above 90% and 80% were achieved, respectively. Density measurement was carried out three times for each sample, and the error was found to be less than 0.1%. Roughness measurement was performed on a 5 mm length of the sample to obtain statistically significant results. As-built Cu showed average surface roughness (Ra) below 20 µm; however, the surface of AM Cu-CNT samples showed roughness values as large as 1 mm. Due to its porous structure, the as-built Cu showed thermal conductivity of ~108 W/m·K and electrical conductivity of ~20% IACS (International Annealed Copper Standard) at room temperature, ~70% and ~80% lower than those of conventionally fabricated bulk Cu. Thermal conductivity and electrical conductivity were ~85 W/m·K and ~10% IACS for as-built Cu-CNT composites at room temperature. As-built Cu-CNTs showed higher thermal conductivity as compared to as-built Cu at a temperature range from 373 K to 873 K. Because of their large surface area, light weight, and large energy absorbing behavior, porous Cu and Cu-CNT materials can be used in electrodes, catalysts and their carriers, capacitors, heat exchangers, and heat and impact absorption. 
    more » « less
  5. Inundation mapping is a critical task for damage assessment, emergency management, and prioritizing relief efforts during a flooding event. Remote sensing has been an effective tool for interpreting and analyzing water bodies and detecting floods over the past decades. In recent years, deep learning algorithms such as convolutional neural networks (CNNs) have demonstrated promising performance for remote sensing image classification for many applications, including inundation mapping. Unlike conventional algorithms, deep learning can learn features automatically from large datasets. This research aims to compare and investigate the performance of two state-of-the-art methods for 3D inundation mapping: a deep learning-based image analysis and a Geomorphic Flood Index (GFI). The first method, deep learning image analysis involves three steps: 1) image classification to delineate flood boundaries, 2) integrate the flood boundaries and topography data to create a three-dimensional (3D) water surface, and 3) compare the 3D water surface with pre-flood topography to estimate floodwater depth. The second method, i.e., GFI, involves three phases: 1) calculate a river basin morphological information, such as river height (hr) and elevation difference (H), 2) calibrate and measure GFI to delineate flood boundaries, and 3) calculate the coefficient parameter ( α ), and correct the value of hr to estimate inundation depth. The methods were implemented to generate 3D inundation maps over Princeville, North Carolina, United States during hurricane Matthew in 2016. The deep learning method demonstrated better performance with a root mean square error (RMSE) of 0.26 m for water depth. It also achieved about 98% in delineating the flood boundaries using UAV imagery. This approach is efficient in extracting and creating a 3D flood extent map at a different scale to support emergency response and recovery activities during a flood event. 
    more » « less