Introduction

Additive manufacturing (AM) of metals is a potentially transformative set of technologies that is seeing growth in several important industrial sectors such as aerospace, health care, civilian nuclear, shipping, petroleum and natural gas, and defense. In all these industrial sectors, the introduction of AM-manufactured components generally started with components that are comparatively simple to build and that pose no safety risk if they fail. As experience has increased, so has the part complexity, application space, and business value. For example, in 2015, General Electric (GE) AviationFootnote 1 introduced the AM-produced T25 sensor housing for the GE90 turbofan, the company’s first AM component for a commercial jet engine. This introduction was quickly followed in 2016 by the launch by CFM International (a 50/50 collaboration between GE Aviation and Safran Aircraft Engines) of the Leading Edge Aviation Propulsion (LEAP) engine that incorporates 19 complex AM-produced fuel nozzles. On the structural side, in 2017, the FAA certified Norsk Titanium’s 33-cm-long, stress-bearing brackets that are used to anchor the Boeing 787 aft galley floor to the airframe. Looking to the near future, GE has announced that their upcoming GE9X turbofan includes numerous AM-manufactured components, including 28 fuel nozzles, 228 low-pressure turbine blades, a T25 sensor housing, a combustion mixer, eight inducers, and a heat exchanger. Expected delivery is in 2025. Despite this evidence for growth, qualification and certification (Q&C) of AM components remains a serious impediment for large-scale implementation. A key issue for AM-component Q&C is that the local processing conditions may vary with location throughout AM builds, as well as between builds, producing components where the local microstructures and properties can be highly non-homogeneous. Traditional coupon testing is of little value if the structure and properties of the coupons are not representative of the components that must be qualified and certified. Even if a given part is qualified and certified based upon extensive, and expensive, part-level testing, updating the part design or manufacturing process for a new application requires extensive retesting, greatly limiting the effectiveness of one of AM’s primary strengths—the innate ability to produce modified part designs quickly and easily.

Increased incorporation of computational materials (CM) into the Q&C process has been proposed as an effective approach for decreasing the time and cost of AM Q&C in the aviation industry.1 The gradual increase in incorporation of CM in the Q&C process would represent a significant paradigm shift and therefore faces considerable challenges, both for achieving acceptance by industry and certifying agencies, and for maturing and validating the various required modeling capabilities. A government/industry/academia steering group called Computational Materials for Qualification and Certification (CM4QC) is working on developing a roadmap for increasing the use of CM in the formal Q&C process. Another multi-institution organization, the Additive Manufacturing Benchmark Series (AM Bench), led by the National Institute of Standards and Technology (NIST), is working to develop and disseminate comprehensive measurement data designed specifically for validating location- and process-specific AM modeling capabilities across a broad range of material systems, AM build approaches, and time and length scales. The work of both organizations is described below.

Q&C of AM-Built Aviation Components

Q&C within the aviation industry is a highly effective formal process that has been developed over many decades to ensure that manned aircraft (e.g., passenger planes, military aircraft) operate reliably within well-defined conditions and flight envelopes. This process continues to evolve as lessons learned from in-service or production failures are incorporated back into the Q&C process requirements. The extraordinary safety record for civilian passenger aircraft is a testament to the success of this methodology. An inherent side-effect of the rigor of the Q&C process for aviation is that the introduction of new technologies may require long and careful scrutiny. Any attempt to broadly incorporate CM into the existing aviation Q&C process must start with a good understanding of how this process works and securing a buy-in from the relevant aviation companies and government certification authorities (i.e., government regulators).

It should be noted that there are no commonly accepted standards for the use of the terms Qualification and Certification across the different industries and certification authorities. Figure 1 is a notional high-level diagram illustrating some elements of the Q&C landscape for aviation,2 with qualification listed under industry and certification listed under certifying agencies (working in close collaboration with industry). For the purpose of this paper, the term qualification refers to the company-specific internal processes that industry uses to ensure that the products they produce meet their performance requirements (often specified by customers). Before these components can be used on aircraft, however, the design and company’s production system must also be certified by an agency that has certification authority. For civilian aircraft in the U.S., this agency is the Federal Aviation Administration (FAA). Other government agencies that possess certification authority over aircraft include the National Aeronautics and Space Administration (NASA) and the various U.S. armed forces. Certification requirements are clearly described in documentation available from the certifying agencies (e.g., Ref. 3 or relevant military standards).

Fig. 1
figure 1

United States landscape for aviation Q&C, from Ref. 2.

Full size image

CM4QC Steering Group

NASA has a long history of driving innovation for aviation, and it frequently holds workshops to gather information about current practices, trends, and visions for the future, with input from a broad range of industrial, government, and academic experts. Two such activities that had a major impact on the formation of the CM4QC steering group were NASA’s Vision 2040 Roadmap for Integrated, Multiscale Modeling and Simulation of Materials and Systems4 and the Aeronautics Research Mission Directorate (ARMD) Workshop on Materials and Methods for Rapid Manufacturing for Commercial and Urban Aviation.5 Side discussions between participants led to the conclusion that developing the necessary infrastructure and ecosystem for advancing Q&C for AM aviation applications would require input from industry, government, and academic stakeholders within a focused setting. Representatives from NASA, NIST, and the FAA therefore organized a 2-day NASA/NIST/FAA Technical Interchange Meeting (TIM) on Computational Materials Approaches for Qualification by Analysis for Aerospace Applications that was held at the NASA Langley Research Center on January 15–16, 2020.1

The TIM brought together approximately 60 subject matter experts representing eight aerospace manufacturers, seven government organizations, and two universities. The key objectives of the TIM were to understand existing gaps in CM processing and performance predictions for aerospace AM materials and components, and to explore how CM approaches could be matured to support material, process, and part-level Q&C. The key output of the TIM was a strong recommendation from several industrial participants that a small steering group should be assembled to provide input and guidance to industry and regulatory agencies on how CM methods should be matured so that they can be effectively used in the context of a Q&C framework. The resulting CM4QC steering group was formed in September of 2020.

CM4QC is composed of representatives from major U.S. aviation-focused companies, government agencies, and academia, as shown in Table I. Representatives from the organizations marked with ‘*’ serve as the CM4QC leadership team.

Table I CM4QC member organizations
Full size table

The primary goals of the CM4QC steering group are to:

  • Inform key stakeholder groups (primarily U.S. industry/government/academia) regarding the pre-competitive R&D investment opportunities to accelerate the development of CM-based approaches for Q&C.

  • Identify key considerations and enablers required to increase the airworthiness certifying authorities’ and industry’s acceptance of the use of computational methods for Q&C of structural process intensive materials (initially AM) parts.

  • Increase the dialogue among the stakeholder organizations by developing a common understanding of the state-of-the-art of CM use in the Q&C domain, including related gaps and challenges.

  • Seek opportunities and provide vehicles and venues for sharing capabilities, methods, tools, and best practices and for discussing regulatory considerations.

To fulfill these goals, CM4QC is developing a multi-year implementation plan, or roadmap, that can serve as a community vision spanning topics related to the use of CM capabilities as a component of industry’s Q&C framework. The primary topics include identification of the relevant regulatory gaps, enablers, and requirements, including acceptable levels of verification and validation (V&V); identification of key CM and enabling technologies; assessment of their current maturity levels; and required future development and opportunities for investment. As of late summer 2023, CM4QC has produced a preliminary draft of the roadmap and is on track to publish the roadmap by summer 2024, prior to a launch symposium being organized for the Materials Science and Technology (MS&T) 2024 conference.

AM Bench

As will be highlighted in the CM4QC roadmap, a key factor for introducing CM into the Q&C process is model validation. Validation is a process for determining the degree to which a model is an accurate representation of the real world from the perspective of the intended uses of the model. Excellent guides and references for model validation exist6,7,8,9,10,11,12,13,14 and the reader is referred to those for detailed information. Measurements conducted for model validation require close cooperation between the measurement and modeling teams, careful control over all initial conditions, boundary conditions, and auxiliary conditions needed for comparison with computer simulations, close attention to measurement uncertainties, and quantitative metrics for evaluating model predictions.

In response to the strong community need for validation data for AM applications, NIST proposed a framework for developing international benchmark measurements for the AM community at an AM workshop at the National Academies of Sciences, Engineering, and Medicine (NASEM) in Washington D.C., on October 7, 2015.15 The response was highly positive, and following a year of planning, AM Bench was launched in November of 2016.

AM Bench includes extensive sets of benchmark measurements, challenge problems for the AM modeling community, and an international conference series. Since AM materials strongly depend on the processing conditions which create them, these data include extensive process input (e.g., process parameters) and precursor materials information, in addition to in situ measurements. These are linked to the post-fabrication structure and performance measurement data, to compose a complete measurement set of the oft-cited process-structure-property (PSP) relationships.

The challenge problems include “blind” tests, as recommended in the NASA Vision 2040 roadmap,4 where modelers are given only the AM experiment conditions, materials information, and some preliminary measurement data, and must aim to replicate the full measurement results with their simulations. These challenges are posed as a modeling competition, where winners are announced at each AM Bench conference. However, modelers are strongly encouraged to continually make use of and compare against all the AM Bench measurement data, which extends beyond those used in the modeling competitions.

Organized around a nominal 3-year schedule, AM Bench completed its first round of measurements, challenge problems, and conference in 2018 and a second round in 2022 (delayed 1 year due to the COVID pandemic). Measurement data and metadata from this effort are freely available. Descriptions of the AM Bench rounds and directions for accessing the AM Bench data and metadata may be found in the AM Bench overview papers16,17 and on the AM Bench website at < https://www.nist.gov/ambench > .

For AM Bench 2022, more than 100 people directly contributed to the AM Bench measurements, data management, and conference organization, including experts from 10 NIST divisions and 21 external organizations. The contributing organizations are shown in Table II. All the organizations participating in the AM Bench measurements provided their own funding to participate.

Table II Organizations that contributed to the AM Bench 2022 measurements, data management, and conference organization
Full size table

AM Bench was established to provide location- and process-specific benchmark measurement data for all material classes and AM methods. However, the requirement for highly controlled experimental conditions and constrained resources limits the current scope. AM Bench 2022 included eight large sets of benchmarks, six on metal alloys and two on polymers. The metal benchmarks focused on laser powder bed fusion AM, with three-dimensional (3D) builds, single laser tracks on bare metal plates, and arrays of laser tracks on bare plates that match the scan patterns of the 3D builds. The hundreds of measurements that were conducted across the United States included simultaneous in situ measurements of laser absorptivity and synchrotron X-ray imaging of the melt pool dynamics, in situ thermographic measurements during the builds, 3D microstructure characterization with micrometer-scale resolution, residual stress measurements using synchrotron X-ray diffraction, neutron diffraction, and mechanical release, along with many more. The measurements cover the full range of PSP, with target alloys including nickel alloys 625 and 718, Ti-6Al-4V, and aluminum alloy 5182. The polymer benchmarks include thermoplastic material extrusion of polycarbonate and vat photopolymerization of methacrylate-based and acrylate-based resins.

Challenge problems were released to the modeling community for all sets of benchmarks. In all, AM Bench received 138 challenge problem submissions for the metal and polymer benchmarks, which is a significant increase in participation from 2018 when a total of 46 challenge problem submissions were received. Figure 2 shows a list of the benchmarks released in 2022, the corresponding challenge problem topics, and the numbers of submissions by the AM modeling community.

Fig. 2
figure 2

AM Bench 2022 challenge submissions. The first column is a graphic describing each set of benchmark measurements. Columns 2 and 3 give descriptive names and designations, respectively, for all the challenges. The blue titles indicate metal benchmarks, and the green titles indicate polymer benchmarks. Column 4 gives the number of submissions for each challenge (Color figure online).

Full size image

Another key factor for AM Bench is data management using FAIR data principles (Findable, Accessible, Interoperable, and Reusable).18 AM Bench provides multiple systems and pathways for users to access, download, search, and analyze AM Bench data and metadata. These capabilities are supported through integrated tools providing several different capabilities that target different user needs:

  • AM Bench Website—best source of information concerning AM Bench measurements, data, metadata, challenge problems, and conference series, including direct links to AM Bench datasets.

  • NIST Public Data Repository (PDR)—primary access to all public AM Bench measurement data.

  • Traditional Journal Articles—published in the journal Integrating Materials and Manufacturing Innovation (IMMI), within the thematic section: AM-Bench 2022.

  • Measurement Catalog—searchable sample and measurement characterization metadata with linked access to associated PDR datasets.

  • SciServer—public platform that allows users to provision a workspace with compute and storage resources for running supplied data analysis and user-developed codes with direct access to a full mirror volume of AM Bench measurement datasets.

  • AM Bench GitHub—AM Bench users will be able to share codes and models that can run on the AM Bench SciServer or at their home institution.

Detailed descriptions and links for these data management systems are available on the AM Bench website at < https://www.nist.gov/ambench/am-bench-data-management-systems > .

Connections and Conclusion

CM4QC and AM Bench are both broad-based initiatives focused on supporting different, but interconnected, aspects of the AM ecosystem. Both groups understand the crucial role of AM model validation and there is substantial membership overlap between the CM4QC steering group and the AM Bench organizing committee and measurement teams. These overlapping interests and personnel have led to collaborative activities such as including an embedded workshop on AM Q&C for aviation applications at the AM Bench 2022 conference. This workshop included a series of opening remarks by NASA Marshall Space Flight Center, NASA Langley Research Center, FAA, Lockheed Martin, Pratt & Whitney, and NIST to set the stage for a NIST-chaired panel discussion including representatives of the listed organizations. A summary of the workshop panel discussion can be found in the AM Bench 2022 overview paper.17 The leadership teams for CM4QC and AM Bench recognize the strong synergy between both organizations and collaboration and coordination will continue.

Another critical component of the AM ecosystem is technical standards. Such standards have many uses, including specifying requirements, communicating guidance and best practices, defining test methods and protocols, documenting technical data, accelerating adoption of new technologies, enabling trade in global markets, and ensuring human health and safety. Standards development in the U.S. is conducted through voluntary participation and consensus. Several standards bodies and industry-government consortia have activities relevant to AM, including ASTM International Committee F42 on Additive Manufacturing Technologies, International Organization for Standardization (ISO) Technical Committee 261 on Additive Manufacturing, European Committee for Standardization (CEN) Technical Committee 438 on Additive Manufacturing, American Society of Mechanical Engineers (ASME) committee on Verification, Validation and Uncertainty Quantification (VVUQ), SAE International Aerospace Material Specifications for Additive Manufacturing (AMS-AM), Metallic Materials Properties Development and Standardization (MMPDS), and many more.

Just as there is significant synergy between CM4QC and AM Bench, similar connections exist between these organizations and standards activities. Standards often play a critical role for Q&C because they provide confidence that standardized measurement procedures and documentary standards are rigorous and broadly accepted. Also, government regulatory agencies and certifying bodies may reference some of the publicly available standards in their requirements, or as acceptable methods of compliance. The interface between AM Bench and standards activities is equally significant. The requirements for developing rigorous benchmark measurements are similar to those required for developing standards, and many of the people who conduct the AM Bench measurements serve on AM standards committees. AM Bench measurements can also serve as the inspiration and foundation for new measurement standards; development of AM Bench-inspired draft standards are currently in progress through the ASTM F42 committee.

While some standards already exist for model V&V (e.g., Refs. 6 and 7), these are still generalized for common computational modeling approaches such as finite element (FE) and computational fluid dynamics (CFD). Much work still needs to be done to establish V&V standards and best practices that address the multiscale and multiphysics simulations that AM Bench typically supports. Looking even further ahead, AM Bench, CM4QC, and standards organizations are all working to understand how V&V for data-driven approaches should best be handled and supported. Code verification, model validation, and model uncertainty quantification for data driven approaches are fundamentally different from traditional physics-based modeling.

Lastly, CM4QC, AM Bench, standards development, and other relevant activities and organizations are critical for accelerating the adoption of AM for industrial applications. It is therefore gratifying to see The Minerals, Metals & Materials Society (TMS) play an active role in promoting these activities and bringing these communities together for the benefit of their members. TMS has partnered with AM Bench from the beginning, and the 2018 and 2022 AM Bench conferences were TMS-affiliated conferences. Similarly, the primary launch event for the CM4QC roadmap is planned as part of a Q&C symposium that CM4QC is organizing at the MS&T 2024 conference. Finally, this journal of The Minerals, Metals & Materials Society (JOM) collection is part of a new TMS effort to bring people involved in technical standards and materials research closer together, with an initial thrust in AM. As one of the largest professional societies for materials research, TMS is fulfilling its role of building broader communities.