AM Bench is a NIST-led organization that provides a continuing series of Additive Manufacturing (AM) benchmark measurements, challenge problems, and conferences with the primary goal of enabling modelers to test their simulations against rigorous, highly controlled additive manufacturing benchmark measurement data. All AM Bench data are permanently archived for public use using comprehensive, custom data management systems.
AM Bench partners broadly with the AM research community. If you have ideas for future AM benchmarks, or if you wish to partner with us on conducting such benchmarks, see the NIST AM Bench website for more information.
To enhance use with computational resources, the public AM Bench data is mirrored to SciServer. To gain access, users can join the Manufacturing Group. This will allow read-only access to the SciServer AM Bench data volume. Users can then make use of the data there or copy it to their user volume for additional utilization within a compute job. For more information, see the ‘SciServer’ section below. These data are freely available under the terms of the NIST License.
AM Bench History
The first round of benchmark measurements, challenge problems, and conference (AM Bench 2018) were completed in 2018 and the second round was completed in 2022, following a one-year delay caused by the COVID-19 pandemic. AM Bench follows a nominal 3-year cycle with the next set of measurements, challenge problems, and conference scheduled for completion in 2025. Additional asynchronous benchmarks are also being planned.
AM Bench 2022 Benchmarks
The AM Bench 2022 Measurements and Challenge Problems include both metals AM and polymers AM.
For metals, 3 sets of benchmarks focused on laser powder bed fusion (LPBF) of the nickel-based superalloy 718, and 2 sets of benchmarks included follow-on mechanical performance and microstructure measurements for the 2018 LPBF studies using the nickel-based superalloy 625. Taken together, these benchmarks cover the full processing-structure-properties range, including feedstock characterization, in situ measurements during the build process, heat treatments, 2D and 3D microstructure characterization, measurements of residual strains and part deflection, and mechanical behavior measurements.
For polymers, one set of benchmarks focused on material extrusion (MatEx) of polycarbonate test objects, and another set of benchmarks focused on vat photopolymerization measurements. Finally, an asynchronous set of benchmark measurements was completed in early 2022 and focused on melt pool and laser coupling dynamics during high-power laser interactions with Ti-alloy and Al-alloy bare metal surfaces. Links to descriptions of the 2022 benchmark measurements, challenge problems, and results are provided through the NIST AM Bench website.
AM Bench 2022 Data and Metadata
AM Bench 2022 data and metadata access is being supported through several different mechanisms that target different user needs. Each access method is summarized in the list below, with a link where available. More information on each method is available below.
- AM Bench website: Primary source for information and data links concerning the AM Bench measurements, data, challenge problems, and conference series
- NIST Science Data Portal and Public Data Repository: Primary data source with access to all public AM Bench measurement data
- Publications: Traditional publications in the journal Integrating Materials and Manufacturing Innovation, within the thematic section: AM-Bench 2022. These AM Bench 2022 measurement articles are expected to become available in late 2023.
- Measurement Catalog: searchable AM metadata curation system.
- SciServer: Free analysis and processing of large AM Bench datasets can be conducted directly on the data server using your own or provided online notebooks. This avoids the need to download Terabytes of measurement data and provides easy analysis using a Jupyter notebook environment. AM Bench data on SciServer is copied from the NIST Public Data Repository.
- AM Bench GitHub repository: AM Bench users will be able to share models and codes that can run on the AM Bench SciServer or at your home institution
AM Bench website
The NIST AM Bench website includes detailed background material on AM Bench, descriptions of all AM Bench measurements and challenge problems along with informational videos from the measurement teams, links to all public AM Bench data and metadata, schedules for upcoming AM Bench events, lists of the 2018 and 2022 AM Bench challenge problem award winners, and descriptions of the data management systems.
NIST Science Data Portal and Public Data Repository
All AM Bench public data are stored on the NIST Public Data Repository (PDR) and may be searched for through the NIST Science Data Portal which provides a user-friendly discovery and exploration tool for publicly available datasets at NIST. This portal is designed with FAIR principles and best practice for Federal Data Strategy.
For Fair Use of these data, please include the citation provided on the PDR homepage in your works, including the Digital Object Identifier (DOI). NIST DOIs are registered with the DataCite organization and provide globally unique persistent identifiers. The DOI also serves as a direct link to data homepages giving access to the full research publication description and underlying data files.
Direct links to the various AM Bench datasets in the PDR are provided on the data access pages of the AM Bench website.
Status as of March 16, 2023: All 2022 measurements are completed and data publications are being prepared, reviewed, and published as quickly as possible.
Measurement data are incomplete without the critical metadata describing the measurement instrument, instrument configuration, calibration, sample details, analysis methods, and many other factors. Metadata associated with AM Bench datasets are available through the AM Bench Measurement Catalog.
The AM Bench metadata are curated using the NIST Configurable Data Curation System (CDCS). The CDCS provides a method for capturing, sharing, and transforming unstructured data into a structured format based on the Extensible Markup Language (XML). Data and metadata are organized using AM Bench-developed templates encoded in XML Schema to create searchable data documents that are saved in a non-relational (NoSQL) document database.
Status as of March 16, 2023: Ambench2022.nist.gov is publicly accessible with a large subset of the AM Bench 2022 measurement data and metadata. The available data and metadata are being expanded as quickly as possible as data are published in the PDR.
AM Bench users can register for a SciServer account at https://sciserver.org. More information on how to use the SciServer platform may be found in the Help section of this website.
Once registered, you will be able to access AM Bench data through the Manufacturing Science Domain. To gain access first click on the ‘Science Domains’ tab in the SciServer dashboard. There select the “Manufacturing” entry and click the “Join” button on the right. Now you should see the AMBench Data Volume under the Files tab and it will also be available to be mounted when you create a Compute container.
Some of the AM Bench data sets are large (> 1 TB) and may require processing to extract desired quantities. Since it is impractical to require all AM Bench users to download such large datasets and to develop all their own codes for extracting meaningful results, the AM Bench project is providing server-side processing through SciServer Compute. AM Bench users can create a SciServer.org login to use virtual machines that include Jupyter notebooks and pre-installed software packages for AM Bench data analysis. A mirror of the AM Bench public measurement data on the PDR is maintained on the SciServer platform and search features are available.
Status as of March 16, 2023: The AM Bench SciServer is publicly accessible. The available data sets will be expanded as they are published in the PDR.
AM Bench GitHub repository
As mentioned above, many of the AM Bench datasets can be processed and analyzed to obtain valuable information for validating model predictions and exploring connections between disparate phenomena. Although some pre-written analysis codes are provided, it is expected that some AM Bench users will need to develop their own codes and algorithms for exploring these datasets. We are providing a public AM Bench GitHub where the AM Bench users can share codes, strategies, and results.
Status as of March 16, 2023: The AM Bench GitHub repository is currently under development.