3

Process Hybridization in One Platform

Opening the second day of the workshop series, Workshop Co-Chair Tom Kurfess, Chief Manufacturing Officer, Manufacturing Demonstration Facility, Oak Ridge National Laboratory, welcomed panelists and participants and posed three key questions for discussion: (1) What is your vision of convergent manufacturing, according to your expertise and experience? (2) What are knowledge gaps for science, engineering, and implementation of convergent manufacturing? (3) What are one or two “moonshot” projects for convergent manufacturing?

Introducing the theme of the keynote presentation, he noted that democratization of innovation aims to move innovation from the smallest enterprise to production and operations, and ultimately to the population.

DEMOCRATIZATION OF INNOVATION

Tracy Frost, Director, Office of the Secretary of Defense Manufacturing Technology

Keynote speaker Frost explained that merging different materials, processes, and systems requires collaboration among several communities, many of which are accustomed to working in siloes. This convergence could lead to opportunities to democratize innovation. However, she described an ongoing challenge with scale-up and manufacturing, particularly among small businesses that lack access to capital-intensive facilities. The U.S. Department of Defense’s (DoD’s) initiative to address this barrier and support the democratization of manufacturing innovation is the Manufacturing USA Innovation Institutes (MIIs). Recommended by

the President’s Council of Advisors on Science and Technology, this initiative began in 2010–2011 after one-third of the manufacturing workforce had been lost, the manufacturing share of the gross domestic product had declined to 12 percent, and 85 percent of the textile industry workforce had been lost—creating a national economy and security issue. The framework for the MII program emphasizes public–private partnership (i.e., a whole-of-nation effort), a model that has been sustained for a decade. The three main pillars of the MII framework are (1) advancing research and technology (i.e., partnering with industry in applied research and industrially relevant manufacturing technologies); (2) securing human capital (i.e., developing manufacturing-specific education and workforce development resources to ensure that innovative technology is manufacturable); and (3) establishing and growing regional manufacturing hubs and ecosystems for long-term, national impact.

Frost noted that the first institute that was stood up—America Makes in Youngstown, Ohio—focused on additive manufacturing. Eight additional institutes have launched, the most recent of which focuses on bio-industrial manufacturing for non-medical products—BioMADE in Saint Paul, Minnesota. Although each institute has a headquarters, all have satellites and a broad presence across the United States: the institutes have more than 1,500 members across 49 states, the District of Columbia, and Puerto Rico. These industry-led, public–private partnerships have significant stakeholder commitments, with $1.5 billion invested from the federal government and more than $2 billion invested from private entities and states. She outlined the objective of the MII program to create enduring resources for these advanced manufacturing stakeholders across the nation.

Frost remarked that the MII model is intended to bridge the “valley of death,” where technology cannot be scaled up or adopted in the United States, or where funding ceases and technology stalls. She described key tenets of the MII model, the mission for which is to catalyze the establishment, effective operation, and integration of industry-led, public–private research partnerships that connect and develop people, ideas, and technology to accelerate the transition of new capabilities into defense products and systems. The MII model also focuses on industry-led, DoD-informed technical roadmapping of priorities. Joint roadmapping activities include all stakeholders across the institutes, which contributes to better, faster benefits. All members are invited to be involved in topic development and to lead or participate in a project, with particular emphasis on small- and medium-sized industry,¹ where much innovation occurs but does not become adopted broadly.

Frost emphasized that when manufacturing capabilities are too expensive and inaccessible to small- and medium-sized businesses, the pipeline of good ideas

___________________

¹ The United States considers small- and medium-sized industry to include firms with fewer than 500 employees. Small firms are generally those with fewer than 50 employees.

and the possibility to leverage manufacturing capabilities decreases. She stressed that the MII model provides opportunities for small- and medium-sized businesses to participate in projects with and leverage technologies from large manufacturers: more than 50 percent of the current membership is small- and medium-sized businesses. She shared a success story from AIM Photonics, which provides a state-of-the-art fabrication, packaging, and testing capability that is accessible for small- and medium-sized companies, academia, and government. AIM Photonics supports rapid, low-cost development with the provision of process design kits (PDKs) and multi-project wafers (MPWs). AIM Photonics PDKs include standard component libraries, are available in common electronics/photonics software platforms, support fabless design models, and facilitate entry for new designers and small businesses. PDKs have basic components, packaging, and modeling information to create and run a design through AIM’s foundry, reducing both design time and cost and increasing first-run success. MPWs, which allow companies to buy a piece of a wafer instead of a whole wafer, also lower cost and increase accessibility for small businesses.

Turning to a discussion of the MII pillar on securing human capital, Frost highlighted the value of education and workforce development. She explained that people often focus on pursuing science, technology, engineering, and mathematics education at the K–12 level, but reskilling and upskilling the existing workforce is equally important, for everyone from skilled technicians to PhDs. She advocated for developing more training opportunities, certificates, apprenticeships, and internships for technicians in particular. The 20 percent of MII members who are from academia help drive curriculum development as well as maintain existing pathways or offer alternative pathways to delivering education. She presented several examples of MII efforts in education and workforce development. First, Advanced Functional Fabrics of America embedded research fellows at defense facilities, providing cross-training opportunities. Second, AIM Photonics has developed online and in-person training courses. Third, BioFabUSA, which focuses on regenerative medicine (an area not often accessible to young students), developed games to engage middle and high school students in understanding biofabrication. Fourth, Advanced Robotics for Manufacturing (ARM) helped DoD to determine robotic workforce needs, training, and salary; as a result, ARM, in partnership with JROBOT, drafted three recommendations for robotic workforce positions.

Frost underscored the need to ensure that taxpayer dollars are used and leveraged for the right investments in these industry-led institutes. Although this is a national effort, she continued, there is an expectation to transition MII technologies to DoD to strengthen the military and better protect warfighters. For instance, Light Innovation for Tomorrow (LIFT) began working with Ricardo Defense Systems in 2017 to retrofit Humvees with antilock brake and electronic stability control systems, reducing rollovers by 74 percent. In March 2021, the

Army provided a contract for $89 million to Ricardo to develop 9,500 kits to retrofit additional Humvees over the next 3 years. Another example is NextFlex, which is working with Sentinel to develop a wearable atmospheric chemical sensor for personnel working in hazardous environments. Both MII partnerships demonstrate the relevance of innovations to many commercial and military applications.

Frost said that the MII program has become a long-term initiative owing to its success in expanding both the use of technologies and the ability of small- and medium-sized businesses and individuals to enter previously inaccessible spaces (i.e., democratizing innovation). Noting that institute assessments are conducted every 5 years, she expressed her hope that advanced manufacturing technologies would become fully adopted as traditional technologies.

Question and Answer Session

Workshop Co-Chair and Session Moderator Ajay Malshe, R. Eugene and Susie E. Goodson Distinguished Professor of Mechanical Engineering, Purdue University, echoed Frost’s assertions about the value of significant investments in and community collaboration for technology development. He reiterated that the future of combat is an asymmetric techno-socio-economic problem, and future solutions seek convergence of length scales, heterogeneous materials, and top-down and bottom-up processes in one platform (e.g., in a backpack, Humvee, or base station) to augment soldiers’ functionality and to reduce dependency on supply chains for critical materials and applications at the point of need. He invited Frost to share her initial commentary on the three key workshop questions. She replied that her team continues to advocate for institutes to work together on convergent manufacturing because siloed efforts are ineffective: public–private partnerships enable innovation. Although 16 institutes span the federal agencies, she continued, more could be needed in the future.

PANEL 3: HYBRID MANUFACTURING PROCESSES

Michael Sealy, Associate Professor of Mechanical Engineering, Purdue University

Sealy noted his interest in hybridization as a means to solve problems related to degradable implants, as well as those related to the food chain, supply chain, wearable structures, and lightweight structures. For example, when it costs $9,000/kg to launch something into space but only $5/kg for an airplane to fly from one city to another, the urgent need for lightweight structures as well as remote manufacturing (i.e., the ability to produce anything anywhere, whether in space or in a deployment zone) becomes evident.

Sealy described his research in hybrid additive manufacturing processes and defined convergent manufacturing by hybridization as having three components—hybrid additive manufacturing processes; hybrid additive manufacturing material, structure, and function; and hybrid additive manufacturing machines. He mentioned that in 2005, the research landscape for hybrid additive manufacturing was sparse, with only 5 key groups active in the area. By 2015, there were 11, and in 2021, there were more than 25—an exponential growth in the number of papers and the number of universities and investigators working in this space.

Sealy explained that traditional manufacturing platforms focus on producing surface integrity, where each manufacturing process can make unique changes to a part. With hybrid additive manufacturing, however, it is possible to combine manufacturing processes (e.g., deep rolling, milling, peening) to make changes layer by layer and to print the desired mechanical properties. He referred to this ability to make local changes with global implications as “glocal” integrity—that is, cumulative and evolving surface integrity and properties across multiple scales to achieve heterogeneous changes (see Sealy et al., 2018, 2019).

Sealy presented three knowledge gaps in convergent manufacturing: (1) Understanding thermal cancellation of residual stresses from applied heat flux on a previously peened layer as well as mechanical cancellation of compressive residual stress by new laser shock peening in a previously peened layer, for which more advanced computational tools are needed (see Madireddy et al., 2019; Sealy et al., 2019, 2020). (2) Moving toward more advanced solutions and identifying one solution for a given problem (e.g., using more advanced design tools to avoid simple fixed interval solutions). (3) Enabling anyone to measure changes from convergent manufacturing processes by bulk wave ultrasound (see Avegnon et al., 2021; Sotelo et al., 2020). Although an advanced degree or access to unique equipment is currently needed to take such measurements, the future could offer a convergent manufacturing measurement tool that could be plugged into an iPhone, revealing the microstructure of and the residual stress on a part. Sharing his moonshot for an intelligent manufacturing process that is accessible to all, he stressed that software should enable anyone to produce complex solutions for fatigue and corrosion problems; accessibility, understandability, and usability lead to true democratization.

Aaron Stebner, Associate Professor of Mechanical Engineering and Materials Science, Georgia Institute of Technology

Stebner explained that cooperative additive and subtractive tools improve precision, but the combination of tools increases challenges in build path planning. He discussed initial approaches that combined large-scale additive processes with computer numerical control machining processes to help with dimensional tolerance control as larger structures were built, and described the current movement

toward integration that is now being explored beyond simply additive and machining technologies.

Stebner noted that dividing tasks among multiple tools could increase cyberphysical security. He presented work from the Colorado School of Mines ADAPT (Alliance for the Development of Additive Processing Technologies) Research Center to integrate a femtosecond laser with a carbon dioxide (CO₂) laser in a laser powder bed system machine. It was possible to use ablative surface machining from the femtosecond laser to achieve three different surface textures, two different surface finishes, and structured surfaces (see Worts et al., 2019). This technology has implications for defense—for example, using ablative machining while building a part that, when held up to light, reveals an optical barcode. This provides an anticounterfeiting capability, where the plan for the placement of the barcodes is in the femtosecond laser, not part of the build file for the base part itself. An added benefit, he continued, is that 10 percent of the light from the femtosecond image can be used for real-time imaging and feedback control. The Colorado School of Mines is now working with Georgia Tech to add interferometry—enabling surface roughness measurement of the additive plus machine surfaces.

Stebner said that near-term goals for hybrid manufacturing include moving beyond the paradigm of two tools/processes and one material. He described three new tools at Georgia Tech: a machining tool, a powder-blown laser deposition tool with different nozzle shapes, and a wire feed tool, which together make it possible to build faster cores of parts with wire; apply different surface coatings, different materials, or finer finishes with a powder tool; and have the machining capability to help with subtractive tasks and improve surface finishes in critical places, all iteratively in one environment while parts are built. He emphasized that hybrid processes and hybrid human–artificial intelligence (AI) cooperation would help realize the benefits and abilities for these controls, and added that it is critical to be able to characterize, measure, and qualify, as well as to integrate recycling.

Stebner mentioned previous work to replace and improve door hinges on armored vehicles when breaking in the field. After several design iterations, the hinge became 35 percent lighter, more instruments could be added to the vehicle, the hinge became much stronger, the part count was reduced from seven to one, and destructive testing was completed to generate qualification data (see Gallmeyer et al., 2019). His near-term moonshot for hinges that break in the field is to be able to feed a broken hinge into a machine and print a new one with an improved design that corrects the problem. This would require recycling, analyzing data, and qualifying implications of the new part on-demand at the point of need, an approach that integrates manufacturing processes, data informatics, human cooperation, resource utilization, and sustainability.

Brian Paul, Professor of Manufacturing Engineering, Oregon State University

Paul discussed some of the polymetal² additive manufacturing work he and his team have been engaged in since 2016, including a partnership within the RAPID (Rapid Advancement in Process Intensification Deployment) Institute, a Manufacturing USA Institute, involving Oregon State University, Pacific Northwest National Laboratory, STARS Technology Corporation, and Southern California Gas Company. He described an advanced thermochemical platform that turns natural/renewable natural gas into hydrogen for fuel cells. As a result, compact chemical plants are placed in a distributed manner, adjacent to the point of use, and the natural gas infrastructure is used to deliver methane to the plants at a low cost. Toyota and other automobile manufacturers have introduced fuel cell vehicles into the California market, but owing to the high cost of centralized hydrogen production plus distribution to filling stations, the price of hydrogen at the “pump” is currently too high to accelerate adoption. Through the RAPID partnership, additive manufacturing vendors were found capable of producing new compact microchannel reactor components economically, which is enabling an initial demonstration for fuel cell buses in California in 2022.

Paul noted that polymetal additive manufacturing could further extend the ability to miniaturize these reactor components through the use of thermally enhanced pins within the microchannels to direct heat transfer vertically between heat transfer channels while minimizing axial (i.e., lateral) heat loss. Polymetal techniques can further be used to lightweight high temperature reactor components by enabling the doping of metal alloys to produce metal matrix composites possessing higher creep resistance at equivalent density. His moonshot is to integrate a capacitive sensor into future flow components for measuring flow-induced vibrations to avoid high cycle fatigue. He stressed that all of these innovations require the ability to tailor existing alloys or grade between multiple materials within a single build.

Paul showed that his research on polymetal additive manufacturing demonstrates that it is possible to build high-quality metal matrix composites such as oxide dispersion–strengthened stainless steel. Furthermore, he and his colleagues have developed “programmable” alloys with the means to locally dope the microstructure at a voxel level, opening the means for product designers to specify different material properties within a single component (see Paul et al., 2020). He added that programmable alloys illuminate the challenges associated with differences in the way materials scientists and engineers think. For example, consider the grading between two metal alloys: materials scientists think in terms of gold standards such

___________________

² In chemistry or mining, polymetal or polymetallic is a substance composed of a combination of different metals.

as X-ray fluorescence as a means to validate the evolving microstructure. Engineers think in terms of indirect methods such as melt pool morphology as a means for process control and part certification. Much work is needed to develop process control methods that can reliably predict the composition and microstructure of components on a voxel-by-voxel basis during economical production. Advances are happening in the machine tool supply chain to make this a reality. Meltio, a small company formed between a U.S. technology startup and a Spanish three-dimensional printing equipment distributor, is selling laser-based directed energy deposition equipment with simultaneous powder-fed and wire-fed capabilities, which Paul and his team are showing capable of delivering programmable alloys. These capabilities are important for navigating the deleterious phases and residual stresses that can make it difficult to place two alloys (e.g., Inconel 625 and GRCop 42) side by side, producing intermediate transition layers to take advantage of the 13× difference in thermal conductivity within components, such as rocket nozzles and chemical reactor vessels.

In closing, Paul described four key knowledge gaps:

1. Exploiting voxel-level properties within design methodologies (including AI-assisted design tools), characterizing graded materials, and predicting microstructure based on process conditions;
2. Decoupling the mixing of alloys and phases within weld pools/beads from process parameters, creating composition tolerances for specifying local material properties and graded transitions, controlling voxel size while changing composition, and estimating high temperature material properties needed for process models;
3. Improving process control and data to support part certification; and
4. Enabling electromechanical integration.

He championed moonshots in the following two areas, which could be attainable with increased investment: (1) Chemical reactors—thermal circuits to direct the flow of exergy between exothermic and endothermic events as well as integrated catalyst scaffolds and catalyst loading. (2) Electromechanical systems—the programming of conductive and dielectric materials during a component build to enable the integration of sensing for equipment health monitoring in space, nuclear, aerospace, and defense applications.

Mary Clare McCorry, Director of Technology and Process Development, Advanced Regenerative Manufacturing Institute (ARMI), BioFabUSA

McCorry outlined the biological considerations for hybrid manufacturing, emphasizing that biology makes a system more complicated, with concern for

how all of its components are interacting. ARMI’s goal is to protect and restore capabilities to warfighters by creating technologies that regenerate instead of only treat. ARMI is focusing on engineered tissue technologies that could restore form, function, and appearance to wounded warfighters—for example, restoring skeletal muscle function when there is muscle loss or engineering a custom-fit bone to replace a damaged bone. Another area of interest is restoring function of damaged nerves, with consideration for long-term effects of warfighting. For instance, there are efforts to address the osteoarthritis that often emerges among discharged warfighters after years of strain in the field. There is also work in small-size disease models with miniature tissues to develop personalized medicine approaches or to better screen drugs used for treatment.

McCorry stressed that the structure and mechanical performance of tissues are as important as the materials themselves. Cells are critical because they are the “tools” used in the manufacturing process to generate tissue; therefore, it is important to consider how cells will interact with and respond to materials, regarding both biological and mechanical characteristics. A tissue-engineered medical product has a triad of materials: cells, signaling factors, and scaffolds. She explained that many manufacturing approaches for generating cells, tissues, and organs are manual-intensive (e.g., sterile rooms, culture hoods, operators with pipettes). Now, there is a shift toward scalable, modular, automated, and closed (SMAC) manufacturing, but the steps of the manufacturing process remain siloed (i.e., tissue harvest and cell banking, expansion of culture, cell harvest and wash, scaffold fabrication, tissue assembly and maturation, preservation and packaging, and transport and logistics). A key challenge is shipping, especially in the transplant industry, which requires short time scales to deliver live tissue. In the case of personalized medicine, she continued, materials have to be tracked throughout the manufacturing process to ensure that the right product is delivered to the right patient.

McCorry’s moonshot is to simplify the process so that it is more accessible, to be able to deploy in remote environments, to develop consistent and quality products, to create autonomous and closed processes, to enable predictive understanding, to engage in informed decision making, and to create flexible processes and personalized medicine. These improvements require measurement system optimization, implementation of process analytic technologies, development of predictive models, technology integration, and implementation and application of AI and modeling.

Question and Answer Session

Cambre Kelly, Vice President of Research and Technology, restor3d, Inc., observed common themes across the panel discussion, including the need to program heterogeneous materials and the importance of developing inspection

and monitoring techniques. Serving as discussion moderator, Sudarsan Rachuri, Technology Manager, Advanced Manufacturing Office, Office of Energy Efficiency and Renewable Energy, U.S. Department of Energy, posed a question about moving from hybrid manufacturing to the convergence of different technologies at different scales. Sealy highlighted two areas that would benefit substantially from multiscale convergent manufacturing: tissue engineering and cellular-based food (i.e., integrating cellular behavior on scaffolds to achieve the desired taste, texture, and form of food). He added that convergent manufacturing is complex in that it requires multiple materials across multiple scales, processes, and systems. Stebner referenced the National Academies’ decadal survey for materials and manufacturing, which prioritized the movement from serial development cycles to continuously connected development (NASEM, 2019). As a result, convergent manufacturing relies on computing, data management, data connection, and mapping among different types of data and players in the supply chain. It is also critical to remove siloes in areas of expertise. He noted that because human minds think serially, parallel computing or computing with a graphics processing unit, which are not limited to that type of thinking, are important. He proposed that automation be used at the point of data curation, allowing humans to focus on abstract thought, extrapolation, and understanding complex relationships. A convergence of physics, science, engineering, computing, and social implications is key, he continued, because technologies are irrelevant if societies do not adopt them. Paul remarked that the design of hybrid manufacturing machine tools extends from an understanding of the physics that must be governed in the context of a manufacturing process. Thus, he suggested that “manufacturing process design” precede machine tool design and become part of a common engineering lexicon as a means for achieving manufacturing innovation. A key question remains about the best pathway to achieve the desired microstructure needed to produce tailored material properties within parts. The value of tailoring material properties can only be leveraged in regulated industries through the development of process control capable of supporting component certification. Although metrology is available for enabling dimensional control, he continued, further research is needed to reveal how best to enable microstructural control including ways to measure compositional tolerance. He advocated for more interaction across disciplines to think about problems from different perspectives, which could lead to the creation of new platforms to develop diverse products and capabilities.

Rachuri asked McCorry if SMAC manufacturing is a precursor to convergent manufacturing. She described SMAC as a good place to begin and outlined a roadmapping exercise that was used to better understand ARMI members’ thoughts about the current state of manufacturing. It became clear that no one had thought about manufacturing from the start; because the process is so complex, they relied on the way that they manufactured during the innovation stage, which requires

much flexibility. Operators were both performing manual steps and interacting with the U.S. Food and Drug Administration, which created anxiety about making changes in the manufacturing process that could affect the final product. McCorry’s team has been interacting with these companies early in their manufacturing processes to help develop methods that they can grow with as they begin to scale to meet clinical demand. She emphasized the importance of predictive models for scaling—much of the time, it is not possible to know how materials and cells will behave, so it is useful to predict performance at all scales.

Malshe wondered how to converge disciplinary siloes with the point of need so that linear thinking becomes spherical thinking. Paul said that convergent manufacturing involves “superprocesses” involving many separate traditional processes for which new design methodologies are critical. He suggested formulating the requirements before designing the process: What is the desired product? What is the annual production quantity? What material systems are needed? Stebner noted that during the pandemic, commercial supply chains were not robust enough to make personal protective equipment and respirators, but universities and individuals with printers could enter this supply chain and meet it at the point of need. This demonstrates the usefulness of a flexible supply chain to optimize at the point of need, although challenges arise in how to govern, certify, and ensure quality and data security. He added that military deployments and space missions also demonstrate the importance of the point of need, where the options are to wait 5 months for a capability to arrive or use what is available.

Rachuri inquired about the educational training needed to participate in convergent manufacturing. Sealy explained that although most universities have secured a metal additive manufacturing system over the past 5 years, it is too complex and unsafe for many at the bachelor’s level to use, which raises questions about how students will be trained to do metal additive manufacturing. To expand convergent manufacturing, he continued, more accessible and lower cost systems would lead to increased educational opportunities at the undergraduate level and make it less difficult to hire people with metal additive manufacturing skillsets. He asserted that a diversity in backgrounds and experiences among experts is essential to further convergent manufacturing. McCorry echoed the notion that there is significant demand for education and workforce development. Although advanced degrees are important for manufacturing, technicians would also benefit from more and better training. Her team is engaged in efforts to excite K–12 students about the opportunities in manufacturing careers, as well as efforts to reskill technicians. Because manufacturing tools are being continuously developed, she noted that it is difficult to determine what skillsets are needed, and hiring qualified people remains a challenge throughout the ecosystem.

In closing, Rachuri invited the panelists to share their key takeaways from the session. Paul commented that with so much open space for tailoring material

properties, engineers need to grow in their knowledge of materials science and material design. The required culture change emphasizes the need for cross-disciplinary education to address this gap. Focusing on the importance of process control for tailoring microstructures within components, he cautioned against underestimating the challenges of developing software to implement process control. Stebner said that AI and machine learning can tackle any problem with statistical relationships and enough samples taken to estimate those statistics, and he pointed out a lack of evaluation tools for hybrid additive manufacturing. He posited that better statistical models would lead to the scaling of information value versus information speed. In the near term, he suggested increased attention to information fusion, verification, validation, and uncertainty quantification. Sealy remarked on the opportunity to use AI and machine learning to help solve optimization problems in hybrid additive manufacturing, although this process has to become simple enough that an advanced degree is not needed. McCorry championed the value of the convergence of minds, as well as standards and control of data to enable distributed manufacturing.

PANEL 4: DESIGN AND MODELING OF HYBRID MANUFACTURING PROCESSES

Julie Chen, Vice Chancellor for Research and Innovation, University of Massachusetts Lowell

Chen emphasized that funding for fundamental research and development has encouraged the creation and evolution of models amidst the modification of materials and the emergence of more complex structures. As these models continue to evolve, AI and machine learning play an important role; for example, additive manufacturing processes and materials combinations are becoming so complex that machine learning coupled with physics-based models offers a means to optimize and move beyond the make-and-break process and product development stage. She described the complexity of a current Army Research Laboratory–funded project, which has a plastics engineer, a mechanical engineer, and a computer scientist working together to develop a model and process for one manufacturing capability.

Chen highlighted the Fabric Discovery Center at the University of Massachusetts Lowell, which is supported by the Commonwealth of Massachusetts as well as three Manufacturing USA Institutes (Advanced Functional Fabrics of America, Flexible Hybrid Electronics, and ARM), to demonstrate the value of collaboration in the creation of products or systems. Despite the Center’s strong modeling capability to make an organic photovoltaic fiber and weave that fiber into a fabric that could be put into a soldier’s uniform, this process is complicated when there is a photovoltaic fiber that needs to be connected to power the device for the soldier. Noting a gap in

the development of modeling capabilities at the systems level, she explained that less funding has been available for connectors and system-level work (especially how to model behavior after developing a product)—and few of the sensors created in the laboratory are ever used in the field. She added that it is important to consider how to create the smallest unit that could be accessible for warfighters in the field. While it is possible to additively manufacture a semiconductor chip, it would not be very efficient; the alternative would be to have access to many chips and sensors and use additive manufacturing for the connections and packaging in the field.

Chen suggested the following discussion topics on the challenges for design and modeling of hybrid systems: (1) interconnects, interfaces, packaging, and mixed materials; (2) standards, test methods, and material databases; and (3) workforce development. She said that manufacturing has a reputation of not being a desirable field because people only think of “dirty and dangerous” manufacturing from many decades ago, and technicians through PhDs are in demand, including a broader and more diverse population within those fields.

Mark Benedict, Computational Materials Scientist and Program Manager in the Propulsion, Structures, and Industrial Technologies Branch, Manufacturing Technology Division, Materials and Manufacturing Directorate, Air Force Research Laboratory (AFRL)

Benedict explained that most manufacturing for the Air Force has to be qualified for air worthiness, which is a rigorous process related to stability, producibility, characterization, predictability, and maintainability. Modeling and advanced design have the potential to accelerate the acceptance of advanced manufacturing concepts, he continued, particularly in convergent manufacturing.

Benedict described DoD’s opportunities in convergent manufacturing related to (1) persistent design, (2) qualification and certification for a unique part, and (3) iterative codesign. First, he reflected on his experience supporting many legacy applications for which the original design data were from the distant past or were lost. When so few data are available to replace a part, inferences have to be made and different advanced techniques used to create new parts. As almost every aspect of the design process becomes digital, persistent design is possible; however, each expert has a modeling stack that does not integrate well and does not outlive the production of the part. To achieve persistent design, he said that data should be co-located so that designs can live in the future and become live entities that can be updated and modified, instead of being frozen or forgotten. Persistent design thus requires coding existing knowledge in the design space into models that can live for significant periods of time.

Second, questions remain about how to best qualify and certify a unique part (i.e., a lot size of 1). Predictability is key, Benedict continued, and modeling is

integral to accepting the risk of the use of that part. A stack of models (e.g., microstructure models, process models, performance models) could be brought to bear to bring knowledge and insight into the potential quality of the part. His moonshot is to have flexible convergent processes that respond to a perceived operational need with a new and novel solution, accept the risk of use on the first part produced, and allow that to change as the mission evolves.

Third, Benedict discussed the concept of iterative codesign, which does not yet exist for multiple processes. He described an opportunity to make near-term investments that would significantly impact convergent manufacturing: the processes would be aware of the needs of what precedes and will follow them, and an iterative process would occur to determine the best article to produce at any one stage in the process to reduce total system delivery time, to increase quality, or to reduce the cost of the item being produced.

Paul Witherell, Mechanical Engineer, Measurement Science for Additive Manufacturing Program, Systems Integration Division, Engineering Laboratory, National Institute of Standards and Technology

Witherell offered a systems perspective of design and modeling for hybrid manufacturing processes. He explained that systems integration activities have long benefitted from and contributed to maturing design, modeling, and simulation capabilities (e.g., virtual to virtual and virtual to physical). The evolution of systems integration can be characterized by the technology, the application, and the problem: technology acts as a driver for requirements, applications act as a driver for scoping domain needs, and problems evolve with new technologies and applications. Key characteristics of hybrid manufacturing through a systems perspective could include increased use of autonomy as well as multiple scales, materials, lasers, and machines.

Witherell turned to a discussion of systems technologies and applications in hybrid manufacturing. As technologies continue to advance, hybrid enablers include new laser systems and other processing technologies, advanced sensors and sensor networks, new automation capabilities, new materials, faster communication with improved wireless access, improved computational capabilities, new information paradigms, and new data analytics. These technology advancements lead to new systems challenges such as increased on-demand data access and storage; real-time system-to-system communication; real-time network (re)configurations; real-time data analysis with explainable results; varying data structures with increased data heterogeneity; and increased redundancies in instructions, observations, and behaviors. The desired platform characteristics to overcome these challenges include local, edge, and cloud support; on-demand access to relevant data; transfer learning capabilities; data compression;

information security; established baseline truths; a unifying data structure for disparate data sources to capture convergence; and a semantically unifiable structure.

As technology matures, digitization progresses, and new applications emerge, Witherell continued, hybrid enablers include new tooling requirements; new machine-to-machine interactions; integrated inspection; new (multi)material delivery and removal systems; scaling of processes/deposition rates in real time; process monitoring, diagnostics, and feedback systems; and accounting for the human-in-the-loop. These emerging application areas present new systems challenges, such as real-time, system-to-system communication; process interruption and control; machine health and prognostics monitoring and communication for machine, build, and facility; awareness between systems’ behaviors; and local and global response control to changing behaviors. Desired platform characteristics include data and decision convergence support, real-time automatic reconfigurability (virtual and physical), well-characterized and integrated behavior models, integrated safety features, integrated security features, and standards-based communication between systems.

Witherell emphasized that advanced hybrid manufacturing systems extend beyond the additive and subtractive. Advanced hybrid manufacturing systems create unique systems challenges, which can be overcome by solving evolving systems problems: these problems become increasingly complicated as the components of the system continue to increase in size, complexity, and scope while increased demands are placed on control. He stressed that modeling and simulation are key enablers and benefactors, and are necessary for problem formulation and resolutions. Modeling and simulation solve systems problems through activities such as supporting systems integration, developing systems interfaces, enabling communication between systems, understanding and facilitating system scaling, performing system optimization, defining system structure, understanding and predicting systems behavior, and assessing system performance. He added that verification, validation, and uncertainty quantification are critical.

John Keogh, Director of Engineering, LIFT

Keogh explained that technology used for instruments such as spectrometers and radio telescopes demands a multidisciplinary approach toward hybrid systems. A synthesis among various disciplines’ capabilities forms an overall functioning system that could do something more complex than what the individual components could do. His vision of convergent manufacturing includes materials processes (i.e., additive, subtractive, metamorphic, or transformative), the digital systems that integrate them, and the workforce—convergent manufacturing requires accurate and careful problem definition before identifying the materials processes, digital systems, and talent, which is a task on which industry could improve.

Keogh shared his overall approach to design and modeling for hybrid manufacturing processes. Problem definition is typically a synthesis between physical systems (the “napkin sketch” to begin to design the machinery or capability) and virtual systems (the modeling and simulation, which is a significant knowledge gap). Simulation requirements are then defined for the process and the material, and a preliminary digital thread is developed. The next steps are to monitor key metrics, manage and interrogate data, perform virtual commissioning to optimize and simulate the hybrid process, iterate (i.e., alternate between the physical and the virtual until a solution to address the original problem statement emerges), and either build the physical system or continue to iterate on the digital twin to improve function. He reiterated that a holistic approach toward hybrid manufacturing integrates many disciplines to solve a specific problem.

Keogh asserted that the successful execution of hybrid manufacturing revolves around people (see Figure 3.1). He suggested continuous and repeated movement through the following cycle to optimize parameters toward certification: a model or a digital twin, process parameters that interface with the physical process to drive the system, in-process monitoring, big data, machine learning and AI, simulation

and analysis, and optimization. Education and workforce development are critical, as each node demands competent personnel.

Keogh described a current LIFT hybrid manufacturing cell with an additive/subtractive approach that is working toward monitoring thermal and melt pool data, torch parameters, and machining chatter, and feeding those data back through a centralized data hub that is interrogated with machine learning algorithms to improve function. Moonshot projects include any synthesis of additive, subtractive, or metamorphic capabilities to control the thermal history and microstructure of components, which could provide information on the properties that could be yielded and subsequent performance. He mentioned other innovative work under way to achieve a software- or process-agnostic approach toward a fully integrated tool chain for computational materials engineering—a middleware wrapper that could accurately and thoughtfully pass data between various software packages to move across the chemistry-process-microstructure-properties-performance continuum would be very impactful.

Question and Answer Session

Serving as session moderator, Kelly reiterated that problem definition continues to be challenging, especially given the number of disparate fields with different vocabulary, tool kits, and understanding. She asked about best practices to define explicit requirements and problems, as well as about tactical approaches to improve education and workforce development so that the next generation is ready to define problems clearly from multiple angles. Keogh observed that problem definition is situationally dependent, but, in general, a clear approach to the problem requires understanding system nuances and asking a wide range of questions. Chen remarked that researchers are often unaware of DoD’s problems; increased communication between those researchers and people in the field would help better define problems. In terms of education, students become more excited and engaged when they are presented with real-world problems. Kelly expressed her support for problem-based learning, which is also a step toward creating better requirements from the early stages of a project. Benedict championed persistent design and noted that model-based definition for requirements (instead of fixed requirements) allows some flexibility in creating a system or component. Witherell emphasized the need to understand the problem for which a digital twin is being developed. Clearly articulating the problem makes it possible to capture and communicate the right information and to focus on solving one problem instead of trying to address everything at once or nothing at all. Modeling and simulation are key to solving specific problems, he continued, but a problem has to first be formulated in a way that a machine can understand.

Malshe wondered whether mechanical modeling could be an effective on-site tool to augment a soldier’s ability to quickly respond to an emerging situation.

He also asked if a digital twin could integrate in real time with little latency so that modeling databases could drive processes to respond at the point of need. Keogh remarked that although it depends on the intention and the situation, accurate programming of decision-making algorithms is essential. The automation of modeling and simulation is limited computationally and in terms of the ability to capture human decision-making processes. He said that a well-refined decision tree is still a distant goal. Benedict noted that some real-time capabilities supported by cloud computing resources are available, although not yet on the battlefield. He envisioned a future state with large computational resources and edge devices or wearables that can interrogate the environment and receive decision support using modeling and manufacturing knowledge. Chen endorsed the notion of decision support, which provides guidance so that the human does not have to sort through a large volume of information and can focus on what is most important. Malshe highlighted the opportunities to unite modeling talent and knowledge in real time for systems-level decision making at the point of need via convergent manufacturing. Witherell reiterated that the digital twin is also situational, because changing environments affect decision making. Thus, one could sensor surroundings, model them, and embed them in the simulation to help inform appropriate behavior and determine corrective action.

Kelly asked how to reframe the design and modeling of hybrid systems when targeting specialized applications (e.g., a lot size of 1). Benedict replied that AFRL is working on advanced demonstration concepts that embrace risk and allow for novel approaches (e.g., modularity, a Lego-like approach to manufacturing). The goal is to open the design space to embrace variability in the available manufacturing process and materials as well as the perceived need; however, tools do not communicate well with each other, which creates a challenge. Witherell noted that if the goal is reconfiguration, the first step is to consider how the available material could be repurposed to provide a different function. The introduction of hybrid processes offers new options, so different levels of composability and modularity have to be considered. Keogh supported having a well-developed modeling and simulation cycle and history coupled to the manufacturing process, as well as certifying the process itself, to advance toward non-destructive certification or qualification of components in low batch numbers. Chen suggested thinking about how to create an opportunity for many types of companies and researchers to offer new ideas to solve the same problem; making it easier to exchange different variations on a particular theme will encourage more creative and innovative problem solving.

Kelly invited the panelists to share their key takeaways from the session. Chen asserted that workforce development is a top priority. It is also important to understand multiprocessing, multifunctional, and multimaterial systems (i.e., how the part is made and how it connects to the rest of the system). She pointed out that university funding often does not support an across-the-system perspective

even though manufacturing is a systems problem, and suggested that universities exert more effort toward the systems approach and eliminating siloes between disciplines. Keogh noted the lack of connection between industry requirements and what is being taught in the university classroom. He also supported more multidisciplinary education to better understand hybrid systems and the future of advanced manufacturing. Benedict explained that design fulfills a requirement, and modeling informs the risk of achieving that requirement; risk acceptance is key to moving faster. He advocated for a new phase of manufacturing, in which the voice at the point of need is amplified: the flight line knows what it needs and what risks it is willing to take, which should be communicated to the manufacturing floor. Witherell highlighted the important role of standards for systems with many moving parts, although specific hybrid manufacturing standards have not yet been developed. Kelly added that standards around a data management strategy would also be valuable.

GROUP QUESTION AND ANSWER SESSION: SCIENCE, ENGINEERING, AND APPLICATION GAPS

Moderator Amy Peterson, Associate Professor of Plastics Engineering, University of Massachusetts Lowell, asked the panelists to discuss successes in and areas for improvement with democratization of innovation. Sealy responded that lowering the cost of equipment would improve democratization of convergent manufacturing and serve as an important step in building the workforce. Benedict commented that making design tools more accessible and affordable allows people to become more comfortable with them. He reiterated that standards are the next step and advocated for government investment in those standards. Witherell added that standards development and participation enable a shift in mindset toward democratization. Chen acknowledged the benefit of more people at different levels of the workforce learning how to use tools, especially via retraining in small businesses. Acquiring lower-cost tools is important, she continued, but meanwhile people should have the opportunity to engage with more expensive tools on a trial basis (e.g., a company that is experimenting and is not ready to convert to new equipment). McCorry mentioned regional hubs that create opportunities for people to test out equipment. She also noted that when the Manufacturing USA Institutes were launched, a standards coordinating body was formed, which accelerated timelines for the development of new standards in regenerative medicine. She asserted that more standards for integration would be useful. Keogh posited that democratization of equipment and software is key for hybrid or advanced manufacturing—for example, plug-and-play capabilities and better education on integration. Because the high cost of and limited access to software continue to present challenges for those in the manufacturing space, particularly the small- and

medium-sized companies, he proposed developing better partnerships with software providers and perhaps offering trials of software packages.

Peterson posed a question about a moonshot related to iterative codesign. Benedict described a 20-year vision for an all-electric air vehicle with an aggressive price target. A smaller moonshot would be a design that could host a modest payload for a certain distance with a fixed cost target, realized with trade-offs between exquisite design and affordability. Most of the cost savings are in the integration of manufacturing processes. Peterson also asked whether intelligent manufacturing processes could create material substitutions to avoid the use of harmful chemicals. Keogh gave two examples of emerging approaches toward mitigating hexavalent chromium pollution: (1) the application of cold spray for thin layer deposition and cladding with metallic chromium, and (2) the use of ionic liquids that allow the use of various chemical forms of non-hexavalent chromium for electromechanical deposition. He championed applying historical knowledge to contemporary problems to achieve cutting-edge manufacturing. Benedict discussed a mirror system being designed by Raytheon with advanced manufacturing to topologically optimize additive designs with a conventional aluminum. Although the result creates a modest penalty for performance, other trades can be made to offset it. While this project demonstrates that it is possible to displace an unwanted material, he cautioned that sometimes displacing the material is not the best approach. Witherell added that the digitalization of manufacturing makes it possible to choose functionally graded materials and design at the microscale to achieve results at the mesoscale and macroscale.

Peterson inquired as to whether convergent manufacturing takes the full life cycle into consideration or if there is a risk that one-off designs are going in the opposite direction. Stebner explained that this question is being explored in the Advanced Manufacturing Pilot Facility at Georgia Tech, where they consider recycling integral to the characterization, the feedstock, and the widget. The difficulty with one-offs is qualification, and a challenge with convergence is doing certification and qualification simultaneously across disciplines, materials, parts, and build paths. He said that thought leaders and innovators at the top level as well as regional depots and small businesses that excel at the point of need could push the field in new, convergent directions. McCorry noted that three-dimensional printing and additive manufacturing are being used to create personal therapeutics that are being filed as one-offs for particular patients, even though the same manufacturing approach is used every time. At some point, she continued, the therapeutic should not be considered a one-off, and the material should be qualified to accelerate approvals and avoid the need to run a full clinical study for each use of the material.

Peterson wondered about international efforts in convergent manufacturing. Witherell replied that the international community has excelled in collaboration and cooperation on standards development. He referenced an agreement between

ASTM and the International Organization for Standardization on additive manufacturing, but international standards for convergent and hybrid manufacturing are not as mature. He emphasized the need for investment in new applications for existing technologies. Sealy added that the earliest patents for hybrid additive manufacturing processes emerged from China and the United Kingdom, many of which were driven by aerospace applications. He mentioned that some of the oldest work was published internationally and funded by the military overseas—the United States is lagging.

In closing, Peterson posed a question about strategies to increase the appeal of manufacturing careers among a broader community. Chen emphasized that manufacturing is not only the act of making something but also the pathway to solving challenging technical problems. Keogh stressed that manufacturing offers much opportunity for professional growth, intellectual engagement, and reward.

DAY 2 SUMMARY

Malshe challenged future innovators to be inspired by the opportunity to improve people’s lives. At the conclusion of the day’s panel discussions, he identified the following as gaps and opportunities for convergent manufacturing:

1. Facilities could connect to achieve and surpass Industry 5.0.³
2. The workforce could be trained with experiential learning that is driven by problems, not disciplines.
3. Because the point of need demands functions, siloes could be removed and features and functions could be converged.
4. Modeling could augment soldiers in the field in real time who are making and rationalizing decisions.
5. Low-cost material could be used for high-value functions, increasing accessibility and affordability.
6. Removing a manufacturing factory is a significant step toward democratization (e.g., Mother Nature manufactures without a traditional factory).
7. Existing facilities could reduce barriers to access and create opportunities for equitable manufacturing.
8. Social scientists, economists, and anthropologists could be part of the conversation about democratizing manufacturing, which is not only a technology problem but also a policy problem.

___________________

³ Industry 5.0 is a new production model where the focus lies on the interaction between humans and machines. Industry 5.0 takes the next step, which involves leveraging the collaboration between increasingly powerful and accurate machinery and the unique creative potential of the human being.

Kurfess suggested that these long-term goals could be achieved by leveraging human resources in conjunction with generative design, as well as by leveraging computing capabilities. He emphasized that there is a plethora of paths to the future.

REFERENCES