Aerospace Moving Forward with Modeling and Simulation

Print Friendly, PDF & Email

In this special guest feature from Scientific Computing World, Gemma Church writes that the aerospace industry is stuck in the past – but it isn’t due to a lack of new simulation and modeling techniques.

Increasing regulation under tightening budgetary constraints is crippling innovation in the commercial sector. Every new material, process or manufacturing tool must be certified before it can be used, creating tremendous expenditure and a risk-averse environment.

Roel Van De Velde, director of aerospace and defence at Esteco, said: “While there have been incremental improvements, in terms of creating quieter and more fuel-efficient aeroplanes, there is not a lot of innovation coming in, compared to the last century when we created aeroplanes moving at the speed of sound and made air travel available to the masses.”

Van De Velde added: “Some large aircraft companies have developed conceptual designs that look radically different compared to the current ‘wing-and-tube” model; for example, the blended wing or the closed wing concepts. They know that [these conceptual designs] have superior performance thanks to simulation and modeling but, even having that knowledge, they do not want to bite the bullet and make a major change. It takes a massive up-front investment, but if we do not commit to a change, nothing will change.”

Aerospace companies are instead focusing on using simulation and modeling tools to optimize manufacturing processes and streamline existing designs. Slaheddine Frikha, aerospace director at the ESI Group, explained: “Since 2010, there”s a strong pressure to drive down costs, leading engineering teams towards the optimization of manufacturing processes (and manufacturing activity as a whole). The focus [then] shifted towards reducing scrap rates and introducing new materials and processes, while increasing production rates.”

Frikha added: “As a result of this evolving focus, the aerospace industry started to implement simulation in its standard processes, to better design and engineer manufacturing capabilities and planning.”

Data analytics is increasingly being used by aerospace manufacturers to achieve sustainable industrial operational excellence, according to Frikha, who added:”The factory of the future is generating lots of precious information via connected objects in the IoT. This data can be analyzed to understand the root cause of product deviations, to fight scrap rates and tackle quality issues, while making sure the machines are utilized at full potential, and to increase productivity.”

Virtual reality is also a fast emerging tool in the industry. “By virtually simulating processes step by step, in an immersive real-scale environment, manufacturers eliminate costly design errors early in design cycle. Predicting human interactions with machines and factory environment enables manufacturers to fine-tune assembly and disassembly sequences to allow production ramp-ups, and to plan more efficient maintenance operations,” Frikha added.

For example, ESI’s virtual reality tool, IC.IDO, was recently used by aircraft engine nacelle producer Safran Nacelles to visualize and validate their process designs – without building full-sized prototypes. This tool helped them reduce the number of product and process development cycles required.

Away from the factory floor, manufacturers not only want to maintain profitability but make sure they meet the industry regulations and high expectations of customers – who want to travel safely with maximum comfort that is possible.

While modeling and simulation can speed up the certification process, there”s an increasing need to validate across multiple domains. Kyle
Indermuehle, vice president of engineering operations, Americas, at MSC Software, said: “In the civil aerospace world, certification is costly, especially as they try to push the boundaries of design, materials and mechatronics. We need to recertify in all three domains and reduce physical testing to reduce costs.”

When it comes to customer expectations, Jousef Murad, community and academic program manager at SimScale, said: “The aerospace industry is under significant pressure and is investing a large budget to achieve low-noise air travel and integrate lightweight components for fuel saving, to produce more efficient and ‘greener” aircraft.”

As a result, an ‘exorbitant amount of money goes into the development of prototypes,” according to Murad, and aircraft companies are trying to reduce this expenditure.

We don”t think completely eliminating physical prototypes in aerospace will be possible, as physical experiments and simulation go hand in hand. One cannot exclude the other. Simulation is used early in the design process to test various design versions and scenarios, while physical prototypes are built later in the process when an optimized design has been already created,” Murad added.

“Aerospace companies are seeking to reduce the number of physical tests and to replace them by virtual tests in order to cut down development costs and lead times. As an example, the French procurement agency DGA-EP is currently developing a virtual test bench for testing aircraft engines in flight conditions for both military and civil applications, using ESI’s system simulation solutions,” Frikha explained.

Manufacturers are also demanding a “continuous digital thread” to seamlessly shift between initial 1D models and 3D drawings when designing an aircraft. Dhiren Marjadi, vice president of global aerospace business at Altair, explained: “Manufacturers start with the 1D model, where all the equations of the aircraft are optimized. There are no 3D drawings at this initial stage and thousands of models are run through to come up with the system specification.”

Marjadi added: “When they move to 3D modeling, there is a connectivity problem. They want to go from 1D to 3D models seamlessly to add value, but there is a technical barrier to overcome to integrate the tools between the 3D and 1D worlds. Also, 1D and 3D modelers have different capabilities and these also need to be overcome going forward.”

There”s also demand to streamline not just simulation and modeling techniques, but also streamline collaboration between manufacturers and their partners. For example, Lockheed Martin recently selected Esteco’s VOLTA software platform to advance multidisciplinary design optimization (MDO) technologies as part of the Expedite program.

The US government-funded Expanded MDO for Effectiveness Based Design Technologies (Expedite) program seeks to advance and expand the use of MDO technologies to disciplines that go beyond aerodynamics and structures, including power and thermal management systems, aircraft performance, manufacturability, robustness, reliability and cost.

Esteco is one year into the four-year program, where the baseline models have already been developed, according to Van De Velde.

With the arrival of manufacturing simulation in the aerospace domain, manufacturing data and associated models are becoming a cornerstone of the digital thread. Component manufacturers are starting to share data with OEMs that is linked to the manufacturing history and its parameters, in order to better understand global product performance across domains.

For example, ESI has developed a solution dedicated to the virtual prototyping of seats, encompassing manufacturing and performance testing. This end-to-end approach maximizes the benefits for both OEMs and their suppliers.

Flying AI

If we move to a smaller scale, new aircraft ideas are under development, including manned and unmanned aerial vehicles (UAVs). For example, MSC Software”s Adams multibody dynamics simulation solution was recently used by Saab to accurately capture interactions between the lifting forces on the rotor blades and downwash on one of its prototype UAVs.

The simulation model provided detailed information, such as the aerodynamic forces acting on each section of the blades. The model also made it possible to evaluate the performance of the UAV under a much wider range of conditions, compared to implementing a physical prototype due to the time, cost and risks involved in actual test flights.

Adams simulation subsequently saved the Saab team at least six months of development time, by reducing the amount of physical testing required.

Manned aerial vehicles (AKA flying cars) may seem more like the stuff of science fiction but they are also entering the conceptual development phase – with some firms promising a commercial rollout in as little as five years.

Furthermore, large standalone satellites designed to carry out a myriad of tasks are now being replaced with fleets of smaller satellites, where each satellite carries out a specific task.

These fleets of flying cars and small-scale satellites don”t only need to rely on simulation and modelling techniques to get them off the ground, but also to perfect their communication methods.

Marjadi said: “When a system of connected entities is created, be those flying cars or satellites, communications become increasingly important. We can”t rely on conventional air traffic control. These vehicles need to communicate with one another to avoid collisions.”

Modeling and simulation software is already being harnessed to design and test machine learning communication systems for this very purpose. Marjadi said: “There is a lot of data coming in from the satellites or vehicles, and that needs to be processed by the other vehicles. Each vehicle has to make a decision where to fly. But we cannot program that without help from AI and machine learning, due to the complexity and scale of the problem. It”s a very emerging technology and growing field.”

Machine learning is helping other areas of the aerospace industry, as Shan Nageswaran, vice president of aerospace, marine and heavy industry verticals at Altair, explained: “Machine learning is also increasingly used to help predictive maintenance. So, when a vehicle is in operation it sends signals on its status. Machine learning systems take that data and analyze it, producing a digital model of the plane to predict its life or the lifespan of a specific part.”

This is where work on digital twins in the aerospace industry is starting to have a real impact. Marjadi explained: “Today”s simulation tools are good enough to predict performance to a very good level of validation. Once we have these models, we can use the same digital model and change it to reflect each plane, so they have a digital twin. That digital twin keeps pace with the changes on the plane, allowing us to make accurate predictions, in terms of its life and maintenance.”

What”s more, ESI is using real-life data provided by connected objects to enrich simulation models. Once created, such hybrid twins can use machine learning techniques to keep the initial simulation model updated and prognose any unexpected product behavior.
Landing success

Computational fluid dynamics simulations are used to study the behavior of aircraft landing gear in reaction to the stresses as an effect of the airflow against the motion of aeroplanes.

Murad explained: “There is a great risk for damage to an aircraft due to a hard landing, or an improper landing technique. This may lead to high stress (due to the air or land impact) on the tires, as well as the shock strut, causing them to wear out sooner or even fracture.”

In a recent study, SimScale”s software was used to compare different designs, as well as various materials to be used for the multiple components of the landing gear. SimScale”s high-precision large eddy simulation (LES) method was used to analyse instantaneous velocity profiles and provide valuable insights into the flow field in wake regions around the landing gear.

However, to bring methods, such as LES, to large-scale applications, hardware issues need to be addressed. Murad explained: “Although hardware greatly evolved in recent years, there is still a long way to go in order to make LES a tool that can be used at any given time for any geometry. In addition, powerful hardware comes with enormous power consumption.”

A prerequisite for the models and simulations to be fully leveraged is that the hardware (as well as the algorithms) are on point and work as efficiently as possible for large computational domains on powerful HPC systems, according to Murad, who said: “At the moment, the aerospace industry is at the petascale and it is assumed that, by 2030, the industry will be reaching the exascale – this represents computing systems that are able to perform a quintillion calculations per second.”

With growing computing power, the world of simulation hopes to get one step closer to make LES a more common tool in the aerospace industry and be able to use direct numerical simulation for bigger applications,” he added.

An emerging technique known as co-simulation could also help the aerospace industry realize large-scale system assessments. Here, different subsystems that form a coupled problem are modeled and simulated in a distributed manner. Hence, the modeling is done on the subsystem level without having the coupled problem in mind.

Co-simulation could help manufacturers attempt to create a robust and sustainable simulation process that works across multiple domains. Indermuehle explained: “During a traditional analysis, we have separate teams for the structure, fluid dynamics and mechanics of the aircraft. Now, all these three areas need to be simulated at the same time.”

As aerospace firms invest increasingly in their HPC infrastructure, there are fewer hardware constraints to realize co-simulation. This is where the modeling and simulation space needs to pick up the baton of innovation to help the aerospace industry.

Indermuehle concluded: “Going forward, it”s not about the sophistication of the hardware in the aerospace industry, but what we can get out of the software, thanks to advancing simulation and modeling tools.”

This story appears here as part of a cross-publishing agreement with Scientific Computing World.

Sign up for our insideHPC Newsletter