A version of this article appears in the August 11/18 double issue of Aviation Week & Space Technology.

It is 2035, and a customer is taking delivery of not only a new aircraft but also a highly detailed digital model specific to that aircraft’s tail number—its airframe, engines and systems. 

Built up over the course of design, development, testing and production, and ultra-realistic down to the level of unique manufacturing flaws, the model will accompany the aircraft throughout its service life. Mirroring its flights exactly, the model’s simulations will be compared with data from the real aircraft to identify anomalies, predict maintenance needs and forecast remaining life.

The “digital twin” is one effort under way to push computational engineering tools to new levels of capability, from model-based design through virtual prototyping and flight testing, to simulation-based certification. To achieve the vision will require substantial government and industry investment in advancing and integrating design tools such as computational fluid dynamics (CFD) for aerodynamics and finite element modeling (FEM) for structures.

Computational engineering has made enormous strides over the past 50 years, but to fully numerically simulate aircraft and engines will require further development of the software tools. A March report by NASA outlining a vision for CFD in 2030 warns that significant advances in simulation-based analysis and design are required to achieve the goal of modeling a complete aircraft throughout its flight envelope.

“To enable use of CFD in all phases of flight, all corners of the design space, will require significant investment,” says Jeffrey Slotnick, Boeing technical fellow for computational sciences and aerodynamics. “In the late 1980s, early 1990s, there was a big increase in CFD investment, and we made huge gains. In the mid-1990s it basically stopped, and the next 20 years were spent validating the technology,” he says.

“When I joined Boeing 36 years ago...CFD was still rather young, and it was just ‘wing out of a wall,’ with no capability to do engines or winglets,” says Robert Gregg, chief aerodynamicist at Boeing Commercial Airplanes. “We have progressed to the point where, today, it is not clear that you need a wind tunnel to give you the drag prediction. Decisions are being made in CFD for the complete integrated aircraft.”

Since the first generation of jet airliners, there has been about a 40% improvement in aerodynamic efficiency and a 40% improvement in engine efficiency, Gregg says. “Probably about half of that has come from CFD,” he estimates, by enabling an understanding of how to apply new technologies. “We’ve also been able to use CFD to reduce the amount of wind-tunnel time—but not eliminate it,” he says, noting that tunnel occupancy was reduced by 25% from the 767 to the 777, and 30% from the 777 to the 787.

There was a steep decline in the number of wing designs tested in the wind tunnel, from 77 for the 767 in 1980 to 11 for the 777 in 1995, but since then the number of tests required has remained essentially constant, says Parvin Moin, professor in the department of mechanical engineering at Stanford University. “Computer power has increased by a factor of 100 every seven years, but since 1995 the number of required tests has plateaued,” he says. “My interpretation is the tools have plateaued.”

“The insight the engineer gains from understanding the detailed aerodynamics is invaluable in advancing the state of the art in component design,” says Stephen Morford, Pratt & Whitney chief engineer for systems analysis and aerodynamics. “Current engines have 40-50:1 overall pressure ratios [OPR] and do it with 10-15% fewer stages than 30 OPR engines 20 years ago,” he says, noting the contribution of CFD.

“That insight allows objectives to be achieved much more quickly and reliably,” Morford says, with Pratt performing only two high-pressure compressor rig tests for its latest PW1000G geared turbofan compared with 40 during development of the PW4000 in the 1980s and 1990s. Using CFD in the design of combustion chambers, which by intent accommodate separated flow, has resulted in a 75% reduction in the number of tests to develop a combustor, and a 60% reduction in emissions, he says.

But the perception of CFD in the industry is a “love/hate relationship,” says Morford. “CFD is critical to the success of our industry. We could not design an aircraft or engine without it —it would be very painful. The business would not withstand the investment required to do it by old methods,” he says. “That’s the love part. The hate part is it does not always you give the right answers. How do you decipher whether you are a getting a realistic answer out of CFD or not?

“As we advance to much more application of CFD, the data derived is so overwhelming to the engineer—can they actually dig deep into the solution to determine the governing physics that have produced the result they see?” he asks. “Do they understand why the code predicted that answer? That is very challenging right now, and it’s a real gap going forward as we start talking about high-performance computing and multi-disciplinary optimization,” Morford says.

“Because of the success we’ve had with CFD we have taken a much more aggressive position toward developing engines,” he adds. “We do it in a shorter timescale than we ever have, about 25% shorter, but because of that when we do have a failure with CFD it becomes a really painful experience— and it has happened.”

Of the limitations to using CFD in designing jet engines, the biggest is that every component is analyzed at its design point. “It’s off-design that we run into problems,” Morford says. In the combustor, at its takeoff or cruise design points, the chemistry is so fast relative to the mixing of the fuel and air that it is an aerodynamically limited problem and thus computationally tractable, he says. But when looking at high-altitude starting, inlet pressures and temperatures are low: “Can you light the fuel? That’s a certification requirement—and not computationally capable today,” he says.

Another issue is speed. “We can throw all the computers in the world at these problems, but for industry to use them to actually affect the products we are designing today, the speed of the computations must be within the design iteration cycle,” says Morford. The aerodynamic, structural and manufacturing design of an airfoil is iterated “hundreds of times in a nine-month interval to develop a high compressor design,” he says. “How do I insert that high-fidelity, off-design, aero-structural interaction analysis and not extend the design cycle? That is a challenge going forward—capability and speed of capability within the design cycle.”

Other limitations of CFD include its inability to analyze aeroacoustics, thermoacoustic instability and to conjugate heat transfer. “All need significant advancement in computational capability and multi-disciplinary analysis if we are to advance to the next generation of jet engine,” Morford says.

The integration of multiple physics-based design and analysis tools are required to implement the Airframe Digital Twin vision of the U.S. Air Force Research Laboratory (AFRL). The Digital Twin is an integrated multi-physics, multi-scale, probabilistic simulation of an aircraft, in which ultra-high-fidelity physical models are updated with sensor data from the real aircraft’s on-board health management system, maintenance records and fleet history.

The simulation mirrors the life of its flying twin and continuously forecasts the aircraft’s health and remaining useful life. The system can also predict its twin’s response to safety-critical events, uncover previously unknown issues and mitigate damage to improve safety and extend life.

As envisioned, the Digital Twin incorporates precise models of the aircraft and its components as built, from FEM and CFD simulations of structural responses to aerodynamic forces down to molecular dynamics and crack growth, including material defects and fabrication anomalies measured during manufacture. The aircraft’s health management system continuously monitors aerodynamic and other loadings, and measures degradation of the vehicle; for example, monitoring strain near a fatigue crack or delamination in a composite material.

On-board sensor data are used to update the physics-based models to produce continuously refined predictions of aircraft health. The simulation can also be used to assess the impact of changes in mission profile on the future condition of the vehicle. As each Digital Twin is unique to a particular aircraft, it can be used to “fly” future missions during design. AFRL and NASA see the system as an enabler of virtual digital certification, replacing today’s empirical design rules based on experiment and experience with ultra-high-fidelity simulations and sensor systems.

The U.S. Air Force’s interest is being able to make real-time operational decisions for individual aircraft based on health awareness by tail number. Flying the digital model through the flight profiles recorded on the actual aircraft will update, calibrate and validate the simulation. As unanticipated damage is found, it will be added to the model so it continually reflects the current state of the aircraft. Flying the model through future missions will help determine when and where damage is likely to occur and when maintenance is needed.

AFRL launched Digital Twin Spiral 1 late last year, awarding $20 million contracts to GE Global Research and Northrop Grumman Aerospace Systems. Northrop aims to develop prognostic and probabilistic individual aircraft tracking that improves the prediction of fatigue cracking. This will integrate advanced flight-load simulation, in-situ damage detection and probabilistic structural reliability analysis into flight-by-flight aircraft tracking. The approach will be demonstrated by tracking fatigue damage in two full-scale structural experiments to be completed by early 2020.

Digital-X is a German aerospace center DLR program guided by the vision of an aircraft performing its first flight in a virtual environment. The project’s primary objective is to develop a flexible software platform for multi-disciplinary analysis and optimization (MDAO) of aircraft and helicopters based on high-fidelity simulation methods in each discipline, rather than the low-fidelity models now used. “Digital-X is a vision of not only aircraft design but also production realization based on numerical simulation,” says Chord Rossow, director of the DLR Institute of Aerodynamics and Flow Technology.

DLR sees numerical simulation as a key enabler for the reliable and efficient design of future aircraft, which may have unconventional configurations with non-linear flight characteristics. High-fidelity methods will be indispensable when designing and assessing such “step-change” aircraft, DLR says, for managing the risks and uncertainties in the design and achieving the best overall performance by integrating aerodynamics, structures and systems. This will require an improvement in simulation capability, particularly the efficiency, reliability and robustness of the CFD flow solver, Rossow says.

Running for almost four years, to the end of 2015, the first phase of Digital-X aims to develop a prototype MDAO simulation platform that integrates aerodynamic and structural design, versus the predominantly sequential process now used. The software platform “will also make it possible to efficiently and reliably perform maneuver simulations throughout the entire flight envelope, and thus permit the determination of aerodynamic and aeroelastic data for evaluating the handling qualities [of new aircraft configurations],” DLR says.

“CFD has evolved significantly over the last 35 years. It is complementary to wind-tunnel and flight testing and can be considered a mature tool for analyzing configurations at the cruise point in the flight envelope,” says Rossow. “What we call the ‘virtual X-craft’ requires us to go toward the full flight envelope, which means we have to tackle all the physical things we encounter there, including transition and separation. We have to include first the structural response and then the flight controls, which today are an inherent building block in the design of aircraft,” he says.

To leverage the potential of this approach to design requires “reliable and efficient simulations with respect to the complete aircraft, we have to deal with complex flows and a huge number of cases,” Rossow says. “These will be multi-disciplinary simulations and we have to make proper use of high--performance computing resources

“What is the potential? First it is time, cost and risk-reduction in design,” says Rossow. While real costs increase gradually as a program moves from design through development and test to production and operation, the “committed costs” that result from decisions made early in design rise steeply. It is here that numerical simulation is showing its potential to reduce overall costs, he argues.

Rossow also sees a possibility to “push the technical perimeter” by first experimenting with new technologies in the computer before wind-tunnel and flight testing. And he views numerical simulation as important for “knowledge preservation.” With design cycles today of five or 10 years, “as an engineer you will be lucky to have two or three full aircraft programs in your lifetime. How do you develop experience and preserve knowledge? This is where the value of numerical simulation comes into play.”

In the realization phase, high-fidelity numerical simulation will allow early consideration of specific production requirements, Rossow says. “You can also try to simulate the first flight in the computer to get information to predict the performance characteristics and handling qualities. This will save a significant amount of time and cost,” he says. “And numerical simulation may serve as a support for certification, and even as a substitute in certain parts of the process.”

Although simulation-based certification is controversial, Rossow cites as “proof that it is possible” the example of a flight-test program at DLR that encountered several problems at the end of last year. “By using CFD in a very quick way, we were able to redesign parts of the instrumentation in the test aircraft. This enabled us to change certain parts and then do the certification. If you use these tools as complementary, this is where you can get real value from them,” he says.

Modeling and simulation are expected to play an increasingly important role in generating the data needed to certify a new aircraft or engine, to reduce development costs and timescales, but regulators will need persuading, and testing in the tunnel and in flight will continue to be important. “We certify structures, not aerodynamics. CFD is used to define the conditions at which we must test to certify, but is not the primary means of demonstrating airworthiness. We have got to work with regulators so they understand the engineering processes being used,” says Morford.

“We need to dispel the idea of CFD replacing wind tunnels. CFD can do so much more than a wind tunnel,” says David Schuster, NASA technical fellow for aerosciences. “Do not think of CFD replacing wind tunnels, think of it as a tool, and of using the right tool,” says Rossow.