What if the U.S. Department of Defense had a computer model so supremely detailed, it could reproduce exactly the operations of a complex military system? If the Pentagon ran the alongside real-world applications, leaders would know when systems were working properly, and when they were going awry.
Navy leaders say this notion of a “digital twin” is not just an interesting hypothesis, it’s an urgent imperative.
“The digital twin concept is critical,” Donald McCormack, executive director for the Naval Surface Warfare Center, said in recent Navy documents. “To pace the threat, we must have an agile testing methodology, which allows for the complexities presented by new automation and technologies. We need to understand how we test in the future with artificial intelligence.”
Military planners and technology leaders on the industry side agree that with the rise of big data, and the ready availability of massive compute power, digital twins could dramatically improve maintenance regimens and lead to more efficient introduction of new and emerging systems.
As the name suggests, a digital twin is a computer-based model of a mechanical or electrical system.
“You take a machine and make a computerized replica of that machine,” said Kelly Jones, a systems engineer at Cisco, which has been collaborating with the Navy on the development of such systems. “For the Navy it solves a lot of critical issues.”
Those issues primarily have to do with maintenance and, by extension, readiness.
Military leaders struggle to keep their legacy systems in working order. A recent report from the Navy for found that fewer than half the Navy’s 546 Super Hornets are mission capable. Digital twins could help, advocates say, by paring back routine maintenance in favor of as-needed repairs.
“Rather than saying a ship must come in at six months, you can maybe keep that ship out there longer, because you are working with known data as opposed to just guessing,” Jones said. “That gives us greater mission readiness and it gives us cost reduction. The less often you pull in a ship for maintenance, the less it costs.”
Digital twins also could help the Navy to rapidly prototype new systems, and to ensure those systems align with real-world needs.
“The idea is to take the lessons learned from working with digital models and use that information to improve designs of the future fleet,” said Dave Drazen, staff specialist to the Office of the Undersecretary of Defense for Research and Engineering. He spoke earlier this year at the Carderock Division of the Naval Surface Warfare Center.
For this vision to come to fruition, however, industry and military leadership will need to lean heavily on big data processing techniques and emerging artificial intelligence capabilities.
“It’s the organization, structuring, contextualization and analysis of data to produce actionable information and to help us make decisions,” said Trisha Shields, lead of the aviation data analytics projects for Carderock’s Sea-Based Aviation and Aeromechanics Branch. She spoke at an earlier Naval Surface Warfare Center event. “Right now, we are at a point where the generation of data is so easy and so cheap that it would be foolish of us not to take advantage of it.”
While engineers can deliver detailed digital models of complex systems, it’s the artificial intelligence that brings those models to life, allowing them to mimic the operation of real-world counterparts. “It needs the ability to think like a human, to know when something is not working right,” Jones said.
Digital twinning already happens in the cyber world, where virtual computing allows engineers to easily replicate entire computer networks. If a system is compromised, they need only access the master copy to replace the corrupt version with a pristine iteration.
For the military to implement a mechanical version of this – to model in exact detail a jet engine, for instance, and then make maintenance decisions based on that model – a new mindset may be required.
“The technical framework is there for it. Now we need to expand people’s imagination and expand their comfort zone,” Jones said. “If we’re talking about key pieces of aircraft structure, people are going to have some reservations about that. Right now, it’s about gaining trust in the technology.”
The Digital Age is Transforming Military Logistics
Many people don’t recognize how vital technology is to logistics and mission readiness.
Imagine being an astronaut almost 200,000 miles from Earth on the way to the moon, and there is a noise — and not a good noise. Something has gone wrong and both the mission and lives are at risk. This happened on Apollo 13 in 1970 and was the subject of a movie. It has been 40 years since the event and 23 years since the film was released, so there is little risk of a spoiler — the astronauts return safely to Earth.
Successful resolution of the mishap involved logistics. It is a gripping case study of operational sustainment under difficult and evolving mission conditions and one of the early examples of digital twin technology — i.e., the ground simulator.
In the movie, physical and digital models were used by Gary Sinise — as astronaut Ken Mattingly — to determine the right boot sequence for the command module, drawing power from the Grumman-built lunar module that had been serving as the crew’s lifeboat. In this case, the digital twin was a physical near-clone of the actual flight system and its use allowed the astronauts to return safely. More on digital twins later, but the story serves as some context for the role that technology plays in logistics and readiness.
Looking to the future — on both the platforms that Northrop Grumman develops and on those developed by other aerospace contractors — there is a continued focus on technology driven breakthroughs that we can bring to customers.
There are four big technology-driven trends that are impacting the logistics mission.
The first is sensors and analytics. At the heart of future trends in logistics, especially with sophisticated platforms like Global Hawk, is data. The military has been exploiting ubiquitous data to move up the value chain pyramid from data to information and ultimately knowledge and wisdom, that can be used to make decisions that drive outcomes.
Sensors are undoubtedly an enabler for this purpose because they are fundamentally changing how sustainment is viewed. Maintenance, for example, has been largely driven by schedule, like changing the oil in a car every 3,000 miles. Taking advantage of the internet-of-things era, there is greater use of instrumentation sensors for condition-based maintenance (CBM). With CBM, rather than changing engine oil on a set schedule, the maintenance interval is based on driving style, the oil’s viscosity and particulate level.
Data analytics research and development is focused on the forecasting of failure modes to enable predictive maintenance. By applying batch and real-time data analytics to “traditional” maintenance methods, the result is higher mission availability and mission effectiveness — at a reduced cost.
Another information-driven area that is trending within the industry is the digital twin. As the name suggests, it is a representation of an as-built physical system. Digital models have been used for a long time, but the level of detail in today’s twins is far greater than legacy digital models — the digital twin is a model plus sensor data. It’s a twin of a specific serial number. It recognizes that it’s not just a plane; it’s a specific tail number. From the time of manufacture, the as-built configuration is captured. The delivery of a new platform consists of both the physical platform as well as a digital copy of the platform. And the twin can be updated based on any modifications made to the physical system to account for repairs and upgrades.
The internet of things and greater computational power have made digital twin technology cost effective to implement. Applying digital twin lessons learned in the virtual world to the physical one results in a better prognosis — especially as it relates to logistics. Northrop Grumman has a lot of experience here — digital twins have been used in the space industry for a long time, for example Apollo 13, and continue through today.
An example is spacecraft command and control. Since one can’t typically touch or inspect the spacecraft after launch, a digital twin is maintained on the ground, keeping track of all aspects of the spacecraft like the state of charge of the batteries, remaining fuel and how well the mission is being performed.
The company is leveraging this experience and using model-based design on programs, collecting design parameters so that digital twins can be used during the long lifecycles. The technology is moving from one-of-a-kind spacecraft to aircraft and even the payloads that go onboard.
The next trend involves software. There is a continued growth in software-defined, hardware-enabled systems. Consider the standalone GPS unit in cars 10 years ago. It was purpose-built and did one thing. Map updates weren’t easy to get and traffic data was only on major roadways. What was bulky and cumbersome in the previous decade has moved rather simply and efficiently to smartphones in today’s environment.
Similarly, across the platforms and mission equipment, there are opportunities to continually innovate through the creation and introduction of new software on the systems developed, maintained and modernized. One aircraft Northrop Grumman is supporting today has 5 million lines of code in 15 distinct computer languages.
The software suite must also accommodate legacy hardware upgrades that have occurred. The original equipment manufacturers may not be around to consult. The code almost certainly was written by people other than those doing the support.
In spite of such hurdles, proprietary tools enable the company to do this effectively and efficiently, accommodating new mission requirements along the way. As mission needs evolve, software updates are used to expand functionality and extend the useful lifetime of systems.
Advanced manufacturing is another important trend in logistics. It is having and will continue to have an impact on the aerospace industry. A key dimension of advanced manufacturing is 3D printing.
Three-D printed plastic objects are common. One use is for stand-ins of conventional parts. Northrop Grumman uses printed objects as full-scale models to do fit checks and is now working to integrate these into operational assets. It is also placing printers closer to the point of need rather than keeping them only in the factory, which reduces production and storage requirements and shortens supply lines. This has a substantive impact on mission performance. This logistical approach is being piloted today with a military customer via a cooperative R&D agreement.
Computational power is being harnessed in a way not achievable in the past to create optimized designs. For decades, engineers have developed symmetric designs that could be created using conventional, subtractive manufacturing methods. Today, when optimizing designs for the freedom of advanced manufacturing, the outputs are often surprising, taking on unexpected shapes. They are functionally equivalent to human-designed parts but sometimes weigh 50 percent less and occupy less volume. Work continues on methods for multi-functional structural parts with features like embedded electronics such as sensors for condition-based maintenance, or embedded energy harvesting.
Advanced manufacturing also has a role in the global supply chain. For example, two parts can be manufactured that can be visually identical. They have the same volume and the same mass. Only one difference — the file used to print one has been manipulated to print the part with modified internal scaffolding. This flawed part won’t have the same structural performance as the genuine part.
In a supply chain, it would be difficult to distinguish between these two parts using conventional methods because of part complexity and limitations in non-destructive inspection techniques available to identify defects. Recognizing the importance of the complete supply chain, including digital elements vulnerable to cyber exploitation, is vital and Northrop Grumman is working with a university partner on how to cost-effectively distinguish between these two kinds of parts.
The final trend is robotics. In today’s environment, industrial robots are commonly used in high-volume, low-mix industries like automobile manufacturing, which involves millions of production units that don’t vary all that much from one another. Robots largely operate in a segregated, structured space in a fixed manner. Experts are required for programming and the actual tasks performed are relatively simple and repetitive.
Going forward, however, robots are expected to be economically useful in low-volume, high-mix arenas like the aerospace industry. Robots will become increasingly mobile and “human safe,” sharing space with workers doing the manufacturing or repair. Consider a robot used for aircraft maintenance that can fit in a fuel tank for inspection and repair.
Instead of constant programming, robots will rely on deep reinforcement learning, whereby they learn from watching a task being performed or even a simulation of a task being performed. The tasks themselves will also get more complex and variable in the future, including repairing or “hand finishing,” which will take advantage of new sensors, advanced control algorithms and multi-arm manipulation. Robots will be able to collect data on how a part was built or repaired including how much material was used and how much force was applied. The data all become part of a digital twin and feed into predictive maintenance routines.
The value of logistics obviously goes beyond managing manpower and spare parts. Logistics in the digital age allows for more condition-based, predictive maintenance that will ensure mission readiness. This readiness posture is especially pertinent since this digital age also brings new threats to the table, including those in the cyber arena. Maintaining a technological edge to support customers is paramount. By focusing on innovation, one can anticipate tomorrow’s challenges in today’s environment.
Gulu Gambhir is vice president and chief technology officer at Northrop Grumman’s Technology Services Sector.