Optimization of X-Ray CT Measurement Accuracy for Metal AM Components

Manufacturing
Optimization of X-Ray CT Measurement Accuracy for Metal AM Components (LAR-TOPS-397)
Reducing cost and time of NDE certification processes
Overview
In aerospace and other industries, metal additive manufacturing (AM) technologies are poised to enable the rapid production of complex, custom components that provides a host of advantages. However, metal AM exhibits high complexity compared to conventionally manufactured parts (e.g., point-by-point deviations in grain structure, material properties, and defect conditions). These complexities make it difficult to ensure accuracy and detectability of critical defects via nondestructive evaluation (NDE). As a result, despite their promise, few metal AM parts have made their way into commercial use for safety-critical applications (aerospace, medical, etc.). X-Ray Computed Tomography (X-Ray CT) has become the gold standard for volumetric characterization of metal AM components. However, X-Ray CT can have defect detection measurement uncertainty, particularly for porosity (the most prominent safety-critical defect in metal AM). A recent study of different labs performing X-Ray CT-based inspection of metal AM parts found that, across 10 different facilities, maximum pore size measured varied by a factor of 4.5 and bulk porosity estimated varied by over a factor of 7! Clearly, a new highly accurate and reliable technique is required that can help develop and can improve the accuracy and reliability of part-specific X-ray CT NDE to speed up certification, enabling industry to realize the advantages of metal AM components for safety-critical applications.

The Technology
This NASA innovation is a method for quantifying and improving accuracy of X-Ray CT-based metal AM part inspection by comparing X-Ray CT data with a high-fidelity 2D surface imaging technique such as surface profilometry. Using surface profilometry that has a NIST-traceable calibration, the technique guarantees a specific level of high fidelity, allowing surface profilometry data to serve as a ground truth reference with which to judge X-Ray CT accuracy and detectability. To deploy the method, both 3D X-Ray CT data and 2D high-fidelity surface images are acquired on the same metal AM part. 3D X-Ray CT data is then segmented and reoriented to extract a 2D X-Ray CT surface image. Measurements of features (e.g., surface-breaking porosity) are then made in both datasets, followed by a comparison of various metrics. This comparison serves two purposes: (a) quantifying the accuracy of the X-Ray CT inspection performed, and (b) providing an objective function which can be minimized to optimize X-Ray CT inspection. The objective function allows engineers to tune X-Ray CT parameters to minimize the function. These optimized parameters can then be implemented to achieve higher accuracy and defect detection reliability in X-Ray CT imaging. Once an X-Ray CT process is optimized for a specific metal AM component, analysis and certification can be accelerated. This technology helps develop X-Ray CT-based metal AM part inspection processes with high accuracy and reliable detectability. Industries for which metal AM parts are desirable and safety, reliability, and fatigue life is of concern (e.g., aerospace, commercial space, automotive, medical) could benefit from the invention. Companies including X-Ray CT inspection system manufacturers using optical sensors and software, NDE data analysis software providers, and end-users in the industries may be interested in licensing this NASA invention.
Segmented 3D volumetric X-Ray CT data. Credit: NASA
Benefits
  • Empowers users to quantify and improve the accuracy and detection reliability of metal AM part-specific X-Ray CT analysis.
  • Enables the optimization of an X-Ray CT machine and post processing parameters to more closely mirror high-fidelity surface imaging and destructive testing techniques.
  • Democratizes metal AM component accessibility by providing a rapid, low-cost means to verify and optimize metal AM component X-ray CT NDE for low volume components.
  • Increases the inspectibility of complex geometry of metal AM components, facilitating AM component designs.

Applications
  • Quantifying the accuracy and defect detection reliability of X-Ray CT evaluation for distinct metal AM components: NASA’s method enables users to quantify the accuracy of an X-Ray CT test for a specific metal AM component by measuring deviations between X-Ray CT and high-fidelity surface imaging data.
  • Increase the inspectibility of difficult-to-certify AM components by improving X-Ray CT defect detection accuracy and reliability: This method produces an objective function which can be minimized by changing X-Ray CT variables (X-Ray CT system settings, reconstruction approach, post processing, image segmentation algorithm parameters) to achieve higher accuracy defect detection for metal AM components.
  • Intermittent quality control of certified X-Ray CT tests in metal AM production environments: This method can be used in metal AM component production environments as a quality control test to ensure X-Ray CT inspections maintain desired defect detection accuracy.
Technology Details

Manufacturing
LAR-TOPS-397
LAR-20512-1
“Assessment of Segmentation-Induced Deviations of Porosity Metrics in Powder Bed Fusion Additively Manufactured Components” Peter W. Spaeth, Erik L. Frankforter, Samuel J. Hocker, and Joseph N. Zalameda, SPIE Smart Structures + Nondestructive Evaluation (2024). Conference Presentation "X-Ray computed tomography integration with surface imaging to improve additive manufacturing porosity quantification." Erik L. Frankforter and Peter W. Spaeth. SPIE Smart Structures + Nondestructive Evaluation (2025). https://ntrs.nasa.gov/citations/20240002879 Data Release "Additive Manufacturing Disk Specimen Nondestructive Evaluation Dataset" Peter W. Spaeth, Erik L Frankforter, Samuel J. Hocker, Joseph N. Zalameda (2024). https://ntrs.nasa.gov/citations/20240002767
Similar Results
front image
Interim, In Situ Additive Manufacturing Inspection
The in situ inspection technology for additive manufacturing combines different types of cameras strategically placed around the part to monitor its properties during construction. The IR cameras collect accurate temperature data to validate thermal math models, while the visual cameras obtain highly detailed data at the exact location of the laser to build accurate, as-built geometric models. Furthermore, certain adopted techniques (e.g., single to grouped pixels comparison to avoid bad/biased pixels) reduce false positive readings. NASA has developed and tested prototypes in both laser-sintered plastic and metal processes. The technology detected errors due to stray powder sparking and material layer lifts. Furthermore, the technology has the potential to detect anomalies in the property profile that are caused by errors due to stress, power density issues, incomplete melting, voids, incomplete fill, and layer lift-up. Three-dimensional models of the printed parts were reconstructed using only the collected data, which demonstrates the success and potential of the technology to provide a deeper understanding of the laser-metal interactions. By monitoring the print, layer by layer, in real-time, users can pause the process and make corrections to the build as needed, reducing material, energy, and time wasted in nonconforming parts.
Purchased from Shutterstock 
https://www.shutterstock.com/image-photo/detail-3d-printer-printing-metal-piece-450749968
Additive Manufacturing Model-based Process Metrics (AM-PM)
Modeling additive manufacturing processes can be difficult due to the scale difference between the active processing point (e.g., a sub-millimeter melt pool) and the part itself. Typically, the tools used to model these processes are either too computationally intensive (due to high physical fidelity or inefficient computations) or are focused solely on either the microscale (e.g., microstructure) or macroscale (e.g., cracks). These pitfalls make the tools unsuitable for fast and efficient evaluations of additive manufacturing build files and parts. Failures in parts made by laser powder bed fusion (L-PBF) often come when there is a lack of fusion or overheating of the metal powder that causes areas of high porosity. AM-PM uses a point field-based method to model L-PBF process conditions from either the build instructions (pre-build) or in situ measurements (during the build). The AM-PM modeling technique has been tested in several builds including a Ti-6Al-4V test article that was divided into 16 parts, each with different build conditions. With AM-PM, calculations are performed faster than similar methods and the technique can be generalized to other additive manufacturing processes. The AM-PM method is at technology readiness level (TRL) 6 (system/subsystem model or prototype demonstration in a relevant environment) and is available for patent licensing.
First 3-D Printed Jet Engine
In-situ Characterization and Inspection of Additive Manufacturing Deposits using Transient Infrared Thermography
Additive manufacturing or 3-D printing is a rapidly growing field where solid, objects can be produced layer by layer. This technology will have a significant impact in many areas including industrial manufacturing, medical, architecture, aerospace, and automotive. The advantages of additive manufacturing are reduction in material costs due to near net shape part builds, minimal machining required, computer assisted builds for rapid prototyping, and mass production capability. Traditional thermal nondestructive evaluation (NDE) techniques typically use a stationary heat source such as flash or quartz lamp heating to induce a temperature rise. The defects such as cracks, delamination damage, or voids block the heat flow and therefore cause a change in the transient heat flow response. There are drawbacks to these methods.
Credit: NASA
Advanced Thermal Inspection with Pulsed Light Emitting Diodes (PLED) Technology
NASA’s PLED thermal inspection system consists of an array of high- powered LED chips configured to deliver controlled pulses of visible light. The system includes 8 LED chip arrays, mounted on an aluminum heat sink and housed in a hood configuration. The inspection hood is specially designed with filters to prevent internal reflections. The LEDs are powered by regulated power supplies and controlled via a computer interface that synchronizes heat pulses with an infrared camera. An acrylic filter is placed over the LEDs to block residual infrared radiation, ensuring that only visible light reaches the target surface. The system’s infrared camera, operating in the mid-wave infrared (MWIR) range does not detect the visible light and captures the transient thermal response of the material, allowing for precise defect detection. By eliminating the need for high-intensity broadband infrared sources, the PLED system provides a cleaner and more accurate thermal response, particularly for unpainted metals and additively manufactured (AM) components. Performance validation of the PLED system has demonstrated significant advantages over traditional flash thermography. In tests on aluminum samples with material loss and AM Ti-6Al-4V metal specimens, the PLED system successfully detected defects with superior contrast and no heat source reflections. Principal Component Analysis (PCA) applied to PLED inspection data revealed clearer defect indications compared to flash-based methods, which introduced unwanted artifacts due to transient reflections. Additionally, the PLED system enabled quantitative thermal diffusivity measurements, offering a new approach to single-sided material characterization. NASA's PLED thermal inspection technology is available for patent licensing. Potential applications include corrosion detection in aerospace components, quality control of AM metal parts, structural health monitoring of industrial materials, and more.
X-Ray Crack Detectability
NASAs software technology uses an Image Quality Indicator (IQI)-based model that can predict whether cracks of a certain size can be detected, as well as a model that can provide appropriate conditions to optimize x-ray crack detection setup. Because this modeling software can predict minimum crack sizes that can be detected by a particular X-ray radiography testing setup, users can test various setups until the desired crack detection capabilities are achieved (predicted) by the modeling system. These flaw size parameter models use a set of measured inputs, including thickness sensitivity, detector modulation transfer function, detector signal response function, and other setup geometry parameters, to predict the minimum crack sizes detectable by the testing setup and X-ray angle limits for detecting such flaws. Current X-ray methods provide adequate control for detection of volumetric flaws but do not provide a high probability of detection (POD), and crack detection sensitivity cannot be verified for reliable detection. This results in reduced confidence in terms of crack detection. Given that these cracks, if undetected, can cause catastrophic failure in various systems (e.g., pressure vessels, etc.), verifying that X-ray radiography systems used for NDE can detect such cracks is of the utmost importance in many applications.
Stay up to date, follow NASA's Technology Transfer Program on:
facebook twitter linkedin youtube
Facebook Logo X Logo Linkedin Logo Youtube Logo