National Academies Press: OpenBook

Foundational Research Gaps and Future Directions for Digital Twins (2024)

Chapter: 7 Toward Scalable and Sustainable Digital Twins

« Previous: 6 Feedback Flow from Virtual to Physical: Foundational Research Needs and Opportunities
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

7

Toward Scalable and Sustainable Digital Twins

Realizing the societal benefits of digital twins in fields such as biomedicine, climate sciences, and engineering will require both incremental and more dramatic research advances in cross-disciplinary approaches and accompanying infrastructure, both technical and human. The development and evolution of digital twins rely on bridging the fundamental research challenges in statistics, mathematics, and computing as described in Chapters 3 through 6. Bringing complex digital twins to fruition necessitates robust and reliable yet agile and adaptable integration of all these disparate pieces.

This chapter discusses crosscutting issues such as evolution and sustainability of the digital twin; translation of digital twin practices between different domains and communities; model and data sharing to advance digital twin methods; and workforce needs and education for digital twin production, maintenance, and use.

EVOLUTION AND SUSTAINABILITY OF A DIGITAL TWIN

As described in Chapter 2, digital twins build on decades of computational, mathematical, statistical, and data science research within and across disciplines as diverse as biology, engineering, physics, and geosciences. The results of this research are encapsulated in core components that form the foundation of a digital twin. Specifically, these components include the virtual representation of a given physical system and bidirectional workflows between the digital twin and the physical counterpart (see Figure 2-1). Response to varying and evolving changes in the physical system, availability of new observational data, updates to the digital model, or changes in the characteristics of the intended use may

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

dictate revision to the workflows of the digital twin. More fundamentally, robust and trustworthy innovation of the digital twin requires that component attributes (e.g., model, data, workflows) are formally described and changeable, and that the integrity and efficacy of the digital twin are preserved across its evolution. The National Science Foundation’s (NSF’s) Natural Hazards Engineering Research Infrastructure frameworks for data and simulation workflows illustrate an example of a starting point upon which developers could build as they bring trustworthy digital twins to fruition (Zsarnóczay et al. 2023).

Over time, the digital twin will likely need to meet new demands in its use, incorporate new or updated models, and obtain new data from the physical system to maintain its accuracy. Model management is a key consideration to support the digital twin evolution. A digital twin will have its own standards, application programming interfaces, and processes for maintaining and validating bidirectional workflows. It will require disciplined processes to accommodate and validate revisions. Self-monitoring, reporting, tuning, and potentially assisting in its own management are also aspects of digital twin evolution.

In order for a digital twin to reflect temporal and spatial changes in the physical counterpart faithfully, the resulting predictions must be reproducible, incorporate improvements in the virtual representation, and be reusable in scenarios not originally envisioned. This, in turn, requires a design approach to digital twin development and evolution that is holistic, robust, and enduring, yet flexible, composable, and adaptable. Digital twins operate at the convergence of data acquisition (sensors), data generation (models and simulations), large-scale computations (algorithms), visualization, networks, and validation in a secured framework. Digital twins demand the creation of a foundational backbone that, in whole or in part, is reusable across multiple domains (science, engineering, health care, etc.), supports multiple activities (gaining insight, monitoring, decision-making, training, etc.), and serves the needs of multiple roles (analyst, designer, trainee, decision-maker, etc.).

The digital twin benefits from having a well-defined set of services, within a modular, composable, service-oriented architecture, accompanied by robust life-cycle1 management tools (e.g., revision management). The characteristics of models that can be used within the digital twin workflow should be well specified. The attributes of the bidirectional workflows supported by the digital twin and the attendant resource provisioning required to support timely and reliable decisions are also key considerations for sustaining the digital twin. These workflows, too, may change over time and circumstance, so both formalism and flexibility are required in the design of the digital twin.

Existing literature and documented practices focus on the creation and deployment of digital twins; little attention has been given to sustainability and

___________________

1 For the purposes of this report, the committee defines life cycle as the “overall process of developing, implementing, and retiring … systems through a multistep process from initiation, analysis, design, implementation, and maintenance to disposal” as defined in NIST (2009).

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

maintenance or life-cycle management of digital twins. Communities lack a clear definition of digital twin sustainability and life-cycle management with corresponding needs for maintaining data, software, sensors, and virtual models. These needs may vary across domains.

Conclusion 7-1: The notion of a digital twin has inherent value because it gives an identity to the virtual representation. This makes the virtual representation—the mathematical, statistical, and computational models of the system and its data—an asset that should receive investment and sustainment in ways that parallel investment and sustainment in the physical counterpart.

Recommendation 4: Federal agencies should each conduct an assessment for their major use cases of digital twin needs to maintain and sustain data, software, sensors, and virtual models. These assessments should drive the definition and establishment of new programs similar to the National Science Foundation’s Natural Hazards Engineering Research Infrastructure and Cyberinfrastructure for Sustained Scientific Innovation programs. These programs should target specific communities and provide support to sustain, maintain, and manage the life cycle of digital twins beyond their initial creation, recognizing that sustainability is critical to realizing the value of upstream investments in the virtual representations that underlie digital twins.

With respect to data, a key sustainability consideration is the adoption of open, domain-specific and extensible, community data standards for use by the digital twin. These standards should also address both data exchange and curation. Data privacy and additional security considerations may be required within the digital twin depending on the nature of the data and the physical counterpart being represented. While this direction aligns with current trends with respect to research data, it requires additional emphasis across all digital twin domains. Specific findings, gaps, and recommendations for data are addressed in the upcoming section Translation Between Domains.

Scalability of Digital Twin Infrastructure

The successful adoption, deployment, and efficient utilization of digital twins at scale requires a holistic approach to an integrated, scalable, and sustainable hardware and software infrastructure. The holistic system-of-systems characteristic of a digital twin presents the challenge that digital twins must seamlessly operate in a heterogeneous and distributed infrastructure to support a broad spectrum of operational environments, ranging from hand-held mobile devices accessing digital twins “on-the-go” to large-scale centralized high-performance computing (HPC) installations and everything in between. The digital twin may support one

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

or multiple access models such as in situ, nearby, and remote access (Grübel et al. 2022). Digital twins necessitate a move away from fragmented components and toward a trusted and secured single hub of assets that captures, integrates, and delivers disparate bidirectional data flows to produce actionable information. The infrastructure challenge is how to create, access, manage, update, protect, and sustain this hub in today’s distributed digital infrastructure.

To support the rapid refresh cycle and persistent interactions required of digital twins, large hardware systems with massive numbers of processors (CPU [central processing unit] and GPU [graphics processing unit]), vast memory, and low-latency, high-bandwidth networks are needed. Newer methods like artificial intelligence (AI) and machine learning (ML) have been able to create new programming approaches that can take full advantage of the newer computational architectures to accelerate their computations. However, not all workloads are good matches for GPU architectures, such as discrete event simulations, due to the generation of non-uniform workloads and resulting potential for performance degradation. Cloud computing may be better suited to handle digital twin components that require dynamically changing computational power, but other components will require a consistent infrastructure.

Simulations for a digital twin will likely require a federation of individual, best-of-breed simulations rather than a single, monolithic simulation software system. A key challenge will be to couple them and orchestrate their integration. To allow a full digital twin ecosystem to develop and thrive, it will be necessary to develop interface definitions and application programming interfaces that enable individual and separate development of simulations that end up running coupled tightly together and influencing each other.

A barrier to realizing digital twins is the speed of the simulations. Classical physical simulations run far slower than real time in order to achieve the desired accuracy. Digital twins will require simulations that can run far faster than real time to simulate what-if scenarios, required fundamental changes in how simulations are designed, deployed, and run. Chapter 3 describes the design and use of surrogate models in order to speed up simulations. To be useful for a deployed digital twin in the field, these simulations may need to be accessible in real time and beyond real time, putting a heavy strain on the communications infrastructure of the end user.

The end user will often be in a position to use classical infrastructure like wired networking and other amenities associated with an office environment to access and control the digital twin. However, a large number of high-value applications for digital twins in areas such as emergency response, natural disaster management and others will depend on mobile infrastructure with significantly lower bandwidth and higher latencies. Mobile networks such 4G and 5G (and even the future 6G) cannot compete with fixed infrastructure, which is where most of the current development happens. Effective access to a digital twin in these constrained conditions will require new methods to present, visualize, and interact with the simulations.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

One possible solution to this challenge lies in pushing surrogate models to their limits by deploying them on the end-user devices. This will require truly innovative methods of simplifying or restructuring simulations in order to allow their execution on mobile devices. These kinds of specialized simulations are a good match for ML approaches, which are less concerned with the underlying physical aspects and more with the phenomenological results of the simulations. Surrogate modeling approaches, including ML, are fundamentally asymmetric: developing and training the model is a computationally intensive process that is well suited to large, centralized infrastructure in the form of HPC centers. But the resulting model can, depending on the complexity of the simulation, be fairly compact. This would allow transferring the model into the field quickly, and modern mobile devices already have or, in the near future, are going to have sufficient computational capacity to execute learned models.

In addition to these extreme cases (model running on HPC versus model running on mobile device), intermediate solutions are possible, in which the model is run on an edge system close to the end user, that can provide higher computational capacity than the mobile device, but at a closer network distance and therefore at lower latency. Such capacity could make it feasible to run complex models in time-sensitive or resource-constrained environments.

Developers do not need to replicate infrastructure at all sites where the digital twin needs to be utilized. Instead, a distributed heterogeneous infrastructure capable of routing data and computational resources to all places where the digital twin is used may be preferable. Sustaining a robust, flexible, dynamic, accessible, and secure digital twin is a key consideration for creators, funders, and the diverse community of stakeholders.

CROSSCUTTING DIGITAL TWIN CHALLENGES AND TRANSLATION ACROSS DOMAINS

As can be seen throughout this report, there are domain-specific and even use-specific digital twin challenges, but there are also many elements that cut across domains and use cases. Many common challenges were noted across the three information-gathering workshops on atmospheric, climate, and sustainability sciences (NASEM 2023a); biomedical sciences (NASEM 2023b); and engineering (NASEM 2023c). In particular, the bidirectional interaction between virtual and physical together with the need for on-demand or continual access to the digital twin present a set of challenges that share common foundational research gaps—even if these challenges manifest in different ways in different settings. This blend of domain specificity and commonality can be seen in each of the elements of the digital twin ecosystem. When it comes to the digital twin virtual representation, advancing the models themselves is necessarily domain specific, but advancing the digital twin enablers of hybrid modeling and surrogate modeling embodies shared challenges that crosscut domains. Similarly,

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

for the physical counterpart, many of the challenges around sensor technologies and data are domain specific, but issues around handling and fusing multimodal data, sharing of data, and advancing data curation practices embody shared challenges that crosscut domains. When it comes to the physical-to-virtual and virtual-to-physical flows that are so integral to the notion of a digital twin, there is an opportunity to advance data assimilation, inverse methods, control, and sensor-steering methodologies that are applicable across domains, while at the same time recognizing domain-specific needs, especially as they relate to the domain-specific nature of decision-making. Finally, verification, validation, and uncertainty quantification (VVUQ) is another area that has some domain-specific needs but represents a significant opportunity to advance digital twin VVUQ methods and practices for digital twins in ways that translate across domains.

As stakeholders consider architecting programs that balance these domain-specific needs with cross-domain opportunities, it is important to recognize that different domains have different levels of maturity with respect to the different elements of the digital twin ecosystem. For example, the Earth system science community is a leader in data assimilation (NASEM 2023a); many fields of engineering are leaders in integrating VVUQ into simulation-based decision-making (NASEM 2023c); and the medical community has a strong culture of prioritizing the role of a human decision-maker when advancing new technologies (NASEM 2023b). Cross-domain interaction through the common lens of digital twins is an opportunity to share, learn, and cross-fertilize.

Throughout the committee’s information gathering, it noted several nascent digital twin efforts within communities but very few that spanned different domains. The committee also noted that in some areas, large-scale efforts in Europe are being initiated, which offer opportunity both to collaborate and to coordinate efforts (e.g., Destination Earth,2 European Virtual Human Twin [EDITH 2022]).

Finding 7-1: Cross-disciplinary advances in models, data assimilation workflows, model updates, use-specific workflows that integrate VVUQ, and decision frameworks have evolved within disciplinary communities. However, there has not been a concerted effort to examine formally which aspects of the associated software and workflows (e.g., hybrid modeling, surrogate modeling, VVUQ, data assimilation, inverse methods, control) might cross disciplines.

Conclusion 7-2: As the foundations of digital twins are established, it is the ideal time to examine the architecture, interfaces, bidirectional workflows of the virtual twin with the physical counterpart, and community practices in order to make evolutionary advances that benefit all disciplinary communities.

___________________

2 The website for Destination Earth is https://destination-earth.eu, accessed June 13, 2023.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

Recommendation 5: Agencies should collaboratively and in a coordinated fashion provide cross-disciplinary workshops and venues to foster identification of those aspects of digital twin research and development that would benefit from a common approach and which specific research topics are shared. Such activities should encompass responsible use of digital twins and should necessarily include international collaborators.

Finding 7-2: Both creation and exploration of the applications of digital twins are occurring simultaneously in government, academia, and industry. While many of the envisioned use cases are dissimilar, there is crossover in both use cases and technical need within and among the three sectors. Moreover, it is both likely and desirable that shared learning and selective use of common approaches will accrue benefits to all.

Recommendation 6: Federal agencies should identify targeted areas relevant to their individual or collective missions where collaboration with industry would advance research and translation. Initial examples might include the following:

  • Department of Defense—asset management, incorporating the processes and practices employed in the commercial aviation industry for maintenance analysis.
  • Department of Energy—energy infrastructure security and improved (efficient and effective) emergency preparedness.
  • National Institutes of Health—in silico drug discovery, clinical trials, preventative health care and behavior modification programs, clinical team coordination, and pandemic emergency preparedness.
  • National Science Foundation—Directorate for Technology, Innovation and Partnerships programs.

MODEL AND DATA COLLABORATIONS TO ADVANCE DIGITAL TWIN METHODS

While several major models are used within the international climate research community, there is a history of both sharing and coordination of models. Moreover, there is a history of and consistent commitment to data exchange among this international research community that is beneficial to digital twins. This is not as pervasive, for example, among biomedical researchers, even accounting for data privacy. While data and model exchange might not be feasible in domains where national security or privacy might be compromised, model and data collaborations may lead to advancements in digital twin development.

There is a culture of global data collaboration in the weather and climate modeling community that results from the fact that observations from all over the world are needed to get a complete picture of the coupled Earth system—the

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

components of the Earth system, the atmosphere in particular, do not end at political boundaries. A fundamental underlying thread in weather and climate research is that, knowing the governing equations of atmospheric behavior (e.g., circulation, thermodynamics, transports of constituents), forecasts of the future state of the atmosphere can be made from a complete set of observations of its current state and the underlying surface and energy exchanges with Earth’s space environment. Because the atmospheric circulation develops and advects properties of the atmosphere on time scales of days, forecasts for a given locality depend on the state of the system upstream within the recent past; observations made over a neighboring jurisdiction are necessary to make a forecast.

The necessity for data exchange among nations was recognized in the mid-19th century, and a rich network of data observations and high-speed communications has developed over the decades since (Riishojgaard et al. 2021). Today, global observations of the atmosphere and ocean are routinely taken from a variety of instrument platforms—including surface land-based stations, rawinsondes, commercial and research ships, aircraft, and satellites—and instantaneously tele-metered to a set of national meteorological and hydrological services that act as hubs for the data via the Global Telecommunication System.3 A vast infrastructure supports this network. The real-time transmission of data is backed up by a network of archival facilities, such as the National Centers for Environmental Information,4 which archives and makes publicly available more than 700 TB of Earth observations and analyses each month. The global network of atmospheric observations used for weather forecasting is augmented for longer records in the Global Climate Observing System.5

The evolution of the global observing network has been advised by the use of the observations to initialize weather and climate forecasts via data assimilation in which a model of the Earth system (or one of its components) is used to generate an estimate of the current state of the system, which is then optimally combined with the observations to produce an initial condition for a forecast. The combination of the observing system, data assimilation system, and forward model conforms to the definition of a digital twin insofar as data from the real world are constantly being used to update the model, and the performance of the model is constantly being used to refine the observing system.

With respect to coordination of models, the United States Global Change Research Program (USGCRP)6 is an interagency body of the federal govern-

___________________

3 The website for the Global Telecommunication System is https://public.wmo.int/en/programmes/global-telecommunication-system, accessed September 20, 2023.

4 The website for the National Centers for Environmental Information is https://www.ncei.noaa.gov, accessed June 26, 2023.

5 The website for the Global Climate Observing System is https://public.wmo.int/en/programmes/global-climate-observing-system, accessed September 20, 2023.

6 The website for the U.S. Global Change Research Program is https://www.globalchange.gov/about, accessed July 7, 2023.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

ment established by Congress in 1990 to coordinate activities relating to global change research, including Earth system modeling. Several federal agencies support independent Earth system modeling.7 Any of the individual models could be viewed as a digital twin of the Earth’s climate system, and at least three of them—Unified Forecast System, Earth System Prediction Capability, and Goddard Earth Observing System—are ingesting observations in real time for prediction purposes. Furthermore, the USGCRP is coordinating these modeling activities through an inter-agency working group (USGCRP n.d.).

While other disciplines have open-source or shared models (e.g., Nanoscale Molecular Dynamics, Gromacs, or Amber within molecular dynamics), few support the breadth in scale and the robust integration of uncertainty quantification that are found in Earth system models and workflows. This lack of coordination greatly inhibits decision support inherent in a digital twin. A greater level of coordination among the multidisciplinary teams of other complex systems, such as biomedical systems, would benefit maturation and adoption of digital twins. More international joint funding mechanisms—such as the Human Frontier Science Program,8 which aims to solve basic science to solve complex biological problems, and the Human Brain Project in the European Union9—offer the size and interdisciplinary makeup to accelerate digital twin development while ensuring cross-system compatibility. The creation of the human genome demonstrates a successful worldwide cooperative effort that advanced common and ambitious research goals. Another objective in establishing these international interdisciplinary collaborations might be to lay the groundwork for establishing norms and standards for evaluation, protection, and sharing of digital twins. While there are similar examples of community data sets (e.g., Modified National Institute of Standards and Technology, National Health and Nutrition Examination Survey, Human Microbiome Project, Laser Interferometer Gravitational-Wave Observatory,10 U.S. Geological Survey Earthquake Hazards Program11), few communities have the comprehensive and shared set of global observational data available to the Earth system modeling community, which obtains a synoptic view of the planet’s coupled components. Such a level of sharing requires the

___________________

7 Including the Department of Energy (https://e3sm.org), the National Science Foundation (https://www.cesm.ucar.edu), the National Aeronautics and Space Administration (https://www.giss.nasa.gov/tools/modelE), the National Oceanic and Atmospheric Administration (https://www.gfdl.noaa.gov/model-development and https://ufscommunity.org), and the Navy (Barton et al. 2020; https://doi.org/10.1029/2020EA001199).

8 The website for the International Human Frontier Science Program Organization is https://www.hfsp.org, accessed June 20, 2023.

9 The website for the Human Brain Project is https://www.humanbrainproject.eu/en, accessed June 26, 2023.

10 The website for the Laser Interferometer Gravitational-Wave Observatory data set is https://www.ligo.caltech.edu/page/ligo-data, accessed June 14, 2023.

11 The website for the U.S. Geological Survey Earthquake Hazards Program is https://www.usgs.gov/programs/earthquake-hazards, accessed June 26, 2023.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

establishment of agreements among a diverse group of stakeholders that must be reviewed and revised as circumstances evolve. Incentives and frameworks (including frameworks that go beyond mere aggregation of de-identified data) for comprehensive data collaborations, standardization of data and metadata, and model collaborations would likely aid in this effort.

Conclusion 7-3: Open global data and model exchange has led to more rapid advancement of predictive capability within the Earth system sciences. These collaborative efforts benefit both research and operational communities (e.g., global and regional weather forecasting, anticipation and response to extreme weather events).

Conclusion 7-4: Fostering a culture of collaborative exchange of data and models that incorporate context through metadata and provenance in digital twin–relevant disciplines could accelerate progress in the development and application of digital twins.

Recommendation 7: In defining new digital twin research efforts, federal agencies should, in the context of their current and future mission priorities, (1) seed the establishment of forums to facilitate good practices for effective collaborative exchange of data and models across disciplines and domains, while addressing the growing privacy and ethics demands of digital twins; (2) foster and/or require collaborative exchange of data and models; and (3) explicitly consider the role for collaboration and coordination with international bodies.

PREPARING AN INTERDISCIPLINARY WORKFORCE FOR DIGITAL TWINS

While digital twins present opportunity for dramatic improvement in accurate predictions, decision support, and control of highly complex natural and engineered systems, successful adoption of digital twins and their future progress hinge on the appropriate education and training of the workforce. This includes formalizing, nurturing, and growing critical computational, mathematical, and engineering skill sets at the intersection of disciplines such as biology, chemistry, and physics. These critical skill sets include but are not limited to “systems engineering, systems thinking and architecting, data analytics, ML/AI, statistics/probabilistic, modeling and simulation, uncertainty quantification, computational mathematics, and decision science” (AIAA Digital Engineering Integration Committee 2020). Unfortunately, few academic curricula foster such a broad range of skills.

New interdisciplinary programs encompassing the aforementioned disciplines, ideally informed by the perspectives of industry and federal partners,

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

could help prepare tomorrow’s workforce (AIAA Digital Engineering Integration Committee 2020). Successful workforce training programs for digital twins require a multipronged approach, with new interdisciplinary programs within the academic system (e.g., interdisciplinary degree programs or online certificate programs) led by governmental agencies (e.g., fellowships at national laboratories) or offered in collaboration with industrial partners (e.g., internships).

In the context of workforce development for digital twins, the committee identifies three core areas where foundational improvements can have significant impact: interdisciplinary degrees, research training programs, and faculty engagement.

Interdisciplinary Degrees

Progress in both advancing and adopting digital twins requires interdisciplinary education. Workforce development may require new curricula, but it can be difficult to create interdisciplinary curricula within the existing structure of most universities. Crosscutting and interdisciplinary research that is foundational, rather than only applied, requires incentives and specific support for a culture change that can foster horizontal research across institutions.

Workforce needs for digital twins require students who are educated across the boundaries of computing, mathematics and statistics, and domain sciences. This is the domain of the interdisciplinary fields of computational science and engineering (CSE) and, more recently, data science and engineering (DSE). Interdisciplinary educational degree programs in CSE and DSE have been growing in number across the nation but remain less common than traditional disciplinary programs, especially at the undergraduate level. Traditional academic structures, which tend to reward vertical excellence over interdisciplinary achievement, are often not well suited to cultivate interdisciplinary training.

Some good models for overcoming disciplinary silos and barriers at universities include new interdisciplinary majors (e.g., the Computational Modeling and Data Analytics program at Virginia Tech) and new classes. There could also be interdisciplinary centers that could serve as a fulcrum for engagement with other universities, government agencies, and industry partners (e.g., the Interdisciplinary Research Institutes at Georgia Tech). The national laboratories provide a good model for interdisciplinary research. However, a driving force (e.g., an overarching problem) to serve as a catalyst for such initiatives is needed. Programs such as the NSF Artificial Intelligence Institutes support a culture of interdisciplinary research, but additional incentives are needed to foster broader recognition and adoption of interdisciplinary research within academic institutions.

Finding 7-3: Interdisciplinary degrees and curricula that span computational, data, mathematical, and domain sciences are foundational to creating a workforce to advance both development and use of digital twins. This need

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

crosses fundamental and applied research in all sectors: academia, government, and industry.

Recommendation 8: Within the next year, federal agencies should organize workshops with participants from industry and academia to identify barriers, explore potential implementation pathways, and incentivize the creation of interdisciplinary degrees at the bachelor’s, master’s, and doctoral levels.

Research Training Programs

In order for the necessary growth in both research and standardization of digital twin approaches across domains as diverse as climate science, engineering, and biomedicine to occur, a targeted interdisciplinary effort must be undertaken that engages academia, industry, and government as part of the digital twin community.

Two types of training will be required to advance the benefits of and opportunities for digital twins. One type of training focuses on the exploration and development of new capabilities within digital twins, while the other type of training focuses on effective use of digital twins. This training will be dynamic as digital twins mature and will need to occur at various levels, including at community colleges and trade schools (e.g., certificate programs).

There are few examples of successful research training programs for interdisciplinary work. The Department of Energy’s (DOE’s) Computational Science Graduate Fellowship program12 is an exemplary model of an effective fellowship program that can be emulated for graduate training of digital twin developers. This program requires that graduate students follow an approved interdisciplinary course of study that includes graduate courses in scientific or engineering disciplines, computer science, and applied mathematics. The fellows are required to spend a summer at a DOE laboratory conducting research in an area different from their thesis, consistent with the interdisciplinary emphasis of the program. More recently, NSF offers student programs for supporting internships in industry and at national laboratories.

Federal agencies can also stimulate the digital twin interdisciplinary aspects through federally funded research and development centers, institutes, and fundamental research at the intersection of disciplines. These efforts provide stimulation both for small-team efforts as well as at-scale research. One might also nurture cross-sector collaborations among industry, academia, and federal agencies to benefit from the strengths of each. In such collaborations, parties must

___________________

12 The website for the Computational Science Graduate Fellowship program is https://science.osti.gov/ascr/CSGF, accessed July 3, 2023.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

be cognizant of different expectations and strengths (e.g., reciprocal benefits), different time scales and timelines, and different organizational cultures.

Faculty Engagement

A successful interdisciplinary research, teaching, and training program requires that the faculty perform interdisciplinary research. The resulting expertise is reflected in course content as well as in digital twin research contributions. It is impractical to have CSE faculty cover the breadth and depth of digital twins from foundational methodology to implementation at scale, so instead the committee suggests tutorials, summer schools, and hack-a-thons. Effective digital twin leadership requires disciplinary depth. Our academic institutions nurture this depth well. However, leadership in digital twins also requires the transformative power of interdisciplinary research. It is essential to be able to provide appropriate recognition for interdisciplinary contributions.

Specific to faculty career advancement (i.e., promotion), a challenge is that interdisciplinary research often involves larger teams, with computer science, mathematical, and statistical contributors being in the middle of a long list of authors. As a result, assessing progress and contribution is difficult for both employers (e.g., universities) and funding agencies. Various solutions have been proposed and provide a good starting point to increase the attractiveness of interdisciplinary research (Pohl et al. 2015). The first steps in this direction are done by, for example, the Declaration on Research Assessment,13 which redefines how scientists should be evaluated for funding and career progress. The European Commission, for example, has recently signed the declaration (Directorate-General for Research and Innovation 2022) and will implement its assessment values for funding awards. Other examples include an increasing number of journals allowing for publication of software and data (e.g., application note by Bioinformatics [Oxford Academic n.d.] or Data14). Github has facilitated the distribution of software tools in a version-controlled manner, allowing the referencing of computer science contributions. Another example is Code Ocean.15

Faculty play a critical role in identifying, developing, and implementing interdisciplinary programs; their support and engagement is essential.

KEY GAPS, NEEDS, AND OPPORTUNITIES

In Table 7-1, the committee highlights key gaps, needs, and opportunities for building digital twins that are scalable and sustainable. This is not meant to be an exhaustive list of all opportunities presented in the chapter. For the purposes

___________________

13 The website for the Declaration on Research Assessment is https://sfdora.org, accessed September 12, 2023.

14 The website for Data is https://www.mdpi.com/journal/data, accessed July 3, 2023.

15 The website for Code Ocean is https://codeocean.com, accessed July 2, 2023.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

TABLE 7-1 Key Gaps, Needs, and Opportunities for Scalable and Sustainable Digital Twins

Maturity Priority
Early and Preliminary Stages
Incentives and frameworks for comprehensive data collaborations, standardization of data and metadata (including across public data sets), and model collaborations are needed. Frameworks are needed that go beyond existing open science frameworks that largely rely on aggregating de-identified data into publicly accessible repositories. 1
Research Ongoing But Limited Results
Existing literature and documented practices focus on the creation and deployment of digital twins; little attention has been given to sustainability and maintenance or life-cycle management of digital twins. Communities lack a clear definition of digital twin sustainability and life-cycle management with corresponding needs for maintaining data, software, sensors, and virtual models. These needs may vary across domains. 1

of this report, prioritization of a gap is indicated by 1 or 2. While the committee believes all of the below gaps are of high priority, gaps marked 1 may benefit from initial investment before moving on to gaps marked with a priority of 2.

REFERENCES

AIAA (American Institute of Aeronautics and Astronautics) Digital Engineering Integration Committee. 2020. “Digital Twin: Definition & Value.” AIAA and AIA Position Paper.

Barton, N., E.J. Metzger, C.A. Reynolds, B. Ruston, C. Rowley, O.M. Smedstad, J.A. Ridout, et al. 2020. “The Navy’s Earth System Prediction Capability: A New Global Coupled Atmosphere-Ocean-Sea Ice Prediction System Designed for Daily to Subseasonal Forecasting.” Advancing Earth and Space Sciences 8(4):1–28.

Directorate-General for Research and Innovation. 2022. “The Commission Signs the Agreement on Reforming Research Assessment and Endorses the San Francisco Declaration on Research Assessment.” https://research-and-innovation.ec.europa.eu/news/all-research-and-innovation-news/commission-signs-agreement-reforming-research-assessment-and-endorses-san-francisco-declaration-2022-11-08_en.

EDITH. 2022. “Implementation.” https://www.edith-csa.eu/implementation.

Grübel, J., T. Thrash, L. Aguilar, M. Gath-Morad, J. Chatain, R. Sumner, C. Hölscher, and V. Schinazi. 2022. “The Hitchhiker’s Guide to Fused Twins: A Review of Access to Digital Twins In Situ in Smart Cities.” Remote Sensing 14(13):3095.

NASEM (National Academies of Sciences, Engineering, and Medicine). 2023a. Opportunities and Challenges for Digital Twins in Atmospheric and Climate Sciences: Proceedings of a Workshop—in Brief. Washington, DC: The National Academies Press.

NASEM. 2023b. Opportunities and Challenges for Digital Twins in Biomedical Research: Proceedings of a Workshop—in Brief. Washington, DC: The National Academies Press.

NASEM. 2023c. Opportunities and Challenges for Digital Twins in Engineering: Proceedings of a Workshop—in Brief. Washington, DC: The National Academies Press.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×

NIST (National Institute of Standards and Technology). 2009. “The System Development Life Cycle (SDLC).” ITL Bulletin. https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=902622.

Oxford Academic. n.d. “Instructions to Authors.” https://academic.oup.com/bioinformatics/pages/instructions_for_authors#Scope. Accessed June 29, 2023.

Pohl, C., G. Wuelser, P. Bebi, H. Bugmann, A. Buttler, C. Elkin, A. Grêt-Regamey, et al. 2015. “How to Successfully Publish Interdisciplinary Research: Learning from an Ecology and Society Special Feature.” Ecology and Society 20(2).

Riishojgaard, L.P., J. Zillman, A. Simmons, and J. Eyre. 2021. “WMO Data Exchange—Background, History and Impact.” World Meteorological Organization 70(2).

USGCRP (U.S. Global Change Research Program). n.d. “Interagency Group on Integrative Modeling.” https://www.globalchange.gov/about/iwgs/igim. Accessed June 7, 2023.

Zsarnóczay, A., G.G. Deierlein, C.J. Williams, T.L. Kijewski-Correa, A-M. Esnard, L. Lowes, and L. Johnson. 2023. “Community Perspectives on Simulation and Data Needs for the Study of Natural Hazard Impacts and Recovery.” Natural Hazards Review 24(1):04022042.

Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 99
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 100
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 101
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 102
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 103
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 104
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 105
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 106
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 107
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 108
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 109
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 110
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 111
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 112
Suggested Citation:"7 Toward Scalable and Sustainable Digital Twins." National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. doi: 10.17226/26894.
×
Page 113
Next: 8 Summary of Findings, Conclusions, and Recommendations »
Foundational Research Gaps and Future Directions for Digital Twins Get This Book
×
 Foundational Research Gaps and Future Directions for Digital Twins
Buy Paperback | $35.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Across multiple domains of science, engineering, and medicine, excitement is growing about the potential of digital twins to transform scientific research, industrial practices, and many aspects of daily life. A digital twin couples computational models with a physical counterpart to create a system that is dynamically updated through bidirectional data flows as conditions change. Going beyond traditional simulation and modeling, digital twins could enable improved medical decision-making at the individual patient level, predictions of future weather and climate conditions over longer timescales, and safer, more efficient engineering processes. However, many challenges remain before these applications can be realized.

This report identifies the foundational research and resources needed to support the development of digital twin technologies. The report presents critical future research priorities and an interdisciplinary research agenda for the field, including how federal agencies and researchers across domains can best collaborate.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!