vol 9, issue 7

Modelling & Simulation lifts aerospace control systems

15 July 2014
By Martin Courtney
Share |
Eurofighter Typhoon in flight

Eurofighter Typhoon engineers had to understand the closed loop simulation that brings stability in actual flight

Series 400 flight compartment

Honeywell used MATLAB software to update its Primus EFIS, used in a range of commercial aircraft

High Lift Prediction Workshop simulation

ANSYS participates in industry initiatives such as the High Lift Prediction Workshop so that its simulations can be validated

Fluent software

Fluent contains modelling capabilities needed to model flow, turbulence, heat transfer and reactions for industrial applications

Simulation of rotor wake from a rotor-body interaction: the isosurface of the Q-criterion is coloured by its vorticity

It’s critical to pre-determine the dynamics of propellers and blades that power advanced aeroplanes and helicopters

Lockheed C-130 Hercules antenna

SolidWorks software was used to build a satellite communications antenna on a Lockheed C-130 Hercules transport aircraft

M-346 fighter pilot training aircrafts

Modelling and simulation software was used to develop the autopilot system on the M-346 fighter pilot training aircraft

Modelling and simulation software plays a critical role in supporting blue sky thinking at the early stages of  aircraft flight control systems design and validation.

The aerospace sector continues to be on the rise as the aircraft that fill the skies above us proliferate and diversify. From personal helicopters to airliners and UAVs, the design requirements are being pushed into new areas of innovation and invention, and the ability to model and simulate how a new design will function once it gets into the air is becoming paramount.Modelling and simulation (M&S) software is used widely within the aerospace industry to craft physical components and optimise the aerodynamic performance of the airframe - and evaluate materials used to construct it. M&S also plays a lesser-known role in the design of flight control systems (FCS), which require the integration of both mechanical and control components within the simulation model to build much more complex prototypes.

The type of controllers found in an FCS computer commonly include control loops that direct the aircraft’s aileron and other control surfaces, for example, but also software used in the ‘black box’ flight recorder, and systems which handle failure management in the case of hydraulic power loss or damage to the airframe.

It is “all in the detail of how it is configured when you are designing a flight control system, which has no obvious function unless it is connected to something else”, explains Mark Walker, principal engineer at technical computing software specialist MathWorks UK. “A final aircraft includes a wide range of systems. If you think about the interaction between the FCS and the fuel distribution system, say, having a way to simulate that, or a way of running code in the final black box, you need to think about these integrated environments [along with] the communications bus on the aircraft, and simulate that in some way.”

ANSYS is another M&S software tool provider that supplies structural analysis tools used to understand the impact of aerodynamic loads on the control surfaces across the flight ‘envelope’, as well as electromagnetic-focused software, which helps design aircraft electromechanical systems such as actuators and motors that drive the movement of the control surfaces.

“Typically simulation of aerodynamic design or the structural response of materials is based on physics,” says Robert Harwood, ANSYS’s director of aerospace and defence industry. “For control systems, ANSYS is defining how it wants the system to behave, and setting-up mathematical models to represent this... [It is] solving the mathematical equations that define the fundamental laws of physics - conservation of mass, momentum and energy or Maxwell’s equations for electromagnetics.”

Sky’s the limit with variables

Aerospace M&S needs to be able to take account of an almost infinite range of operating environment variables. “Our tools are used to characterise the dynamic behaviour of complex engineering systems,” adds Paul Goossens, vice president, engineering solutions at mathematical and analytical software company Maplesoft. “Something as basic as a control surface on an aircraft wing to monitor airflow interaction and provide lift. If you combine that with activation by a hydraulic circuit, it means various different domains are interacting with each other with loading conditions which cannot always be predetermined.”

Avionics specialist Honeywell employed MathWorks’ MATLAB software to update the company’s Primus Electronic Flight Instrument System (EFIS) series that is used in a range of commercial aircraft, harnessing the software to enable one team to design, model and simulate the flight control laws, automatically generate flight-ready code, and reuse the same code on other designs.

Large portions of the code for the upgraded Primus series was generated from a Simulink model, with the revamped flight control system certified by the Federal Aviation Administration in 1998 and since integrated as the standard avionics system for Raytheon’s Hawker Horizon business jet.

Elsewhere, UK-based avionics company Smiths Aerospace - acquired by General Electric Aviation Systems for $4.8bn in 2007 - also uses modelling and simulation software to design flight control systems, as well as primary equipment such as thrust reversers, high-lift systems and systems actuators for civil and military aircraft.

BAE Systems-Crestview used SolidWorks 3D modelling and simulation software to build a prototype of a satellite communications antenna mounted on the escape hatch of a Lockheed C-130 Hercules military transport aircraft, as used by the Royal Air Force.

M&S software is particularly useful for testing an FCS against stringent safety and performance regulations laid down by the world’s aerospace regulation bodies - the testing and calibration of electronic control units (ECUs) and stability controllers is often of particular importance. ANSYS actively participates in industry initiatives such as the American Institute of Aeronautics and Astronautics (AIAA) High Lift Prediction Workshop and Drag Prediction Workshop to make sure that its aerodynamic simulations are validated as meeting industry requirements, for example.

Italian aircraft manufacturer Alenia Aermacchi, meanwhile, used modelling and simulation software to develop and certify the autopilot system on its M-346 Master fighter pilot training aircraft. The final system needed to be complaint with the Federal Aviation Administration DO-178B Level A modelling standard, which defines safety requirements for software used in airborne environments, published as the ED-12B standard by the European Organisation for Civil Aviation Equipment.

The Alenia Aermacchi engineering team used Mathworks’ Simulink and Stateflow for ARP-4754 software (Aerospace Recommended Practice ARP-4754 is a guideline that deals with the development processes which support certification of Aircraft systems), alongside IBM’s Rational Dynamic Object Oriented Requirements System, to define six pre-primary autopilot states, the transitions between them and other control logic. They then ran simulations to validate the system behaviour before optimising performance and incorporating safety standards to meet the certification constraints.

Any failure to thoroughly validate an FCS in as wide a variety of operational conditions and environments as possible can have potentially catastrophic and expensive consequences. Boeing’s fleet of 787 Dreamliners was grounded by global aviation industry regulators in January 2014, after a small electrical fire broke out on a Japan Airlines aircraft during take-off from Tokyo airport. Engineers diagnosed a problem with a leaking lithium-ion battery.

The Dreamliner’s two on-board batteries are used to power key systems including flight controls, emergency lighting and cockpit recorders - all of which have been built around the specific charge the batteries are able to deliver.

“You can try to predict how loading is going to be on those batteries, but to really cover everything you have to look at loading conditions to get an idea of duty cycle,” says Maplesoft’s Goossens. “In normal runways, a battery may work fine - but there can be an unfortunate set of circumstances which takes it over the edge, especially when more heat is generated during thermal runaway events.” An FCS is much more involved than the traditional autopilot in terms of the control it has over the aircraft and the sheer volume of the electronic systems it manages; this consideration means that simulation has to be that much more accurate to avert potential mishaps.

“If you take the [British Aerospace] Eurofighter Typhoon, for example, the airframe has been deliberately designed to be unstable in order to get maximum performance and the way it is then stabilised is to have a control loop running it,” Walker at MathWorks explains. “If the pilot had maximum control, the aircraft would fall out of the sky. So when exploring the design concept, the engineers had to understand the closed-loop simulation.”

The foremost benefit for aerospace engineers of using M&S applications is a much faster development process, which sees designs tested and validated in software before the model makes it to a prototype component, system or final product. Engineers can get feedback on the viability of the model very early, compared to discovering problems at a later stage.

“Basic control theory lets you do something with one loop at a time,” Walker points out. “Simulation software lets multiple loops - and more complex scenarios - be considered.”

By automating low-level certification activities, including requirements coverage analysis, running-time error and standards compliance checking, Alenia Aermacchi’s M-346 engineers estimated they had shaved up to 30 per cent of the time previously spent reviewing certification requirements. They managed to cut the overall time to flight period between developing the auto pilot system and installing it within the aircraft itself by about 20 per cent.

However, while M&S software provides much greater scope to do ‘Monte Carlo’ testing, and other types of simulation that repeatedly process large and diverse volumes of data to get more probabilistic results, increasing the accuracy of the findings cannot be guaranteed simply by throwing more hardware capacity at the problem.

“If you have a bad model, which is too computationally intensive or detailed, that can be solved by chucking more hardware capacity at it,” observes Walker, “but the other way is to have a more efficient model.”

Aerospace design engineers tend to follow a set path of testing, which involves various types of simulations, from model-in-the-loop (MIL) or software-in-the-loop (SIL) in the first instance, followed by hardware-in-the-loop (HIL). Maplesoft’s aerospace customers stem primarily from ‘Skunk Works’ type groups, small units of designers/developers given free range to explore new concepts with few management constraints.

The ‘Skunk Works’ term was initially coined by Lockheed Martin and is still applied to the company’s Advanced Development Programs (ADP) initiative responsible for innovative military aircraft designs, including the U-2, SR-71 Blackbird, F-117 Nighthawk and F-22 Raptor.

“In the past individual groups have worked on one subsystem in the aircraft, using the concept of the ‘iron bird’, where you lay out all those systems before you build it then take it through a notional flight,” reports Maplesoft’s Paul Goossens. “With Maplesoft tools and the high computational capabilities from HIL and SIL testing, there are fewer reasons to do that.”

Early in the process of conceiving a new flight control computer, “you would be interested in how the aircraft operates and would need to design it within a specific performance envelope”, explains Walker. “This can be simulated using a range of different calibrations against usually fairly low-fidelity models of the airframe - so you can make design trade-offs between better performance in one area versus another while staying within that envelope.”

Maplesoft applications are now being used primarily in robotic space systems and satellite control rather than aircraft, though there are many crossover requirements in either case and both use similar design and development tools.

The Canadian Space Agency (CSA) used MathWorks’ MATLAB and Simulink software to build one 17m-long and two 3.5m robotic arms on the Mobile Service System, a manipulator used to help build and maintain the International Space Station, unload equipment from shuttles, and move heavy objects. The CSA used the M&S tools to verify that every operation would work in space, not least because the hardware was so light that it could not support its own weight, and could not therefore be physically tested on earth where gravitational conditions were very different.

Maplesoft tools have also been adopted by Nasa’s Jet Propulsion Laboratory (JPL), whose ongoing projects include designing spacecraft to explore comets, asteroids and other areas of the solar system. The agency is using Maple and MapleSIM to develop engineering models including multi-domain and multi-body systems, plant modelling and control design.

Competitor Autodesk, meanwhile, says that its tools have been widely used in design projects for defence-related UAVs which have a notoriously short development lifecycle, though a desire for secrecy means it is not allowed to discuss them. Autodesk is also used to design aircraft which use solar cells on the top side of the wings to enable very light aircraft to fly for days or weeks on end.

“Defence contractors are developing aerodynamic systems, traditional control systems and hydraulics, and now moving onto electronic controls especially on small UAVs to ensure durability performance,” reports Jonah Normand, product specialist in design, lifecycle and simulation.

Fuel management

A key example of this necessary integration between an FCS and flight performance characteristics is a fuel control system, something which experts say is very hard to simulate using other methods.

Even ascertaining how much fuel an aircraft’s tanks contain at any point during flight is challenging: aircraft are usually varied through angular orientations, and as such accurate measurement of fuel quantity on board is difficult to obtain.

In 2011, for instance, researchers from De Montfort University and Glasgow Caledonian University used COMSOL’s Multiphysics softare to successfully model a capacitive fuel gauging unit for an aircraft. At ANSYS, meanwhile, Harwood says: “As the industry is facing pressure to deliver more fuel-efficient aircraft, high fidelity simulation of the entire system - that is, control software and virtual hardware in an integrated simulation - delivers insight that cannot be gained through disconnected and siloed design processes.”

MathWorks software was employed to develop the fuel management system for one of the world’s most popular and successful commercial aircraft, the Airbus A380. The system was designed to handle fueling and defueling operations on the ground, and fuel flow to the four engines and 11 fuel tanks to optimise its centre of gravity, reduce wing bending and keep fuel within a safe, acceptable temperature range and control in flight fuel jettison.

“An aircraft weighs 300 [metric] tonnes at take-off and 250 tonnes of that is fuel - during the flight you have to manage where that fuel is spread, which means controlling 21 pumps, 43 valves, and other mechanical components,” explains MathWorks’ Walker. “This is actually something which is very difficult to write documentation for just using experience.”

The engineering team working on this fuel management system developed a paramaterised model of the tanks, pumps, valves and electrical components using MathWorks’ Simulink tools, and after running closed-loop simulations of each individual component integrated them into a complete model for system-level simulations. It also developed HIL tests well before the actual physical hardware was available.

It then used MathWorks’ Parallel Computing Toolbox and MATLAB Distributed Computing Server to carry out ‘Monte Carlo’ testing simulations on a 50-worker computing cluster (made up of multiple interconnected workstations), running 100,000 simulated flights under varied environmental conditions and operational scenarios over a single weekend.
This helped it validate the requirements months earlier than was previously possible, engineers confirmed, being able to predict problems which could result from combinations of relatively minor failures that much earlier.

Engineers estimated that earlier projects had taken up to nine months to integrate the fuel system design with the simulated cockpit, or iron bird, compared to less than a month using the model-based design. They were also able to re-use the HIL test rig to shorten the development time spent bringing the initial concept to flight by around three months. *

Share |

Jargon Buster

Closed loop  A plant simulation used to test the physical behaviour of a particular component or controller in isolation from the rest of the systems in a particular model.

Customer-in-the-loop  Systems under test in uncontrolled, customer environment, similar to a beta test.

Environment-in-the-loop  Real plant running in an uncontrolled environment, much like a field test.

Hardware-in-the-loop  Simulation-based hardware evaluation which uses simulated input conditions to assess performance, using mathematical representations of all related dynamic systems or sub-systems and usually including electrical emulation of sensors and actuators. Typically done as a final test before system integration and field testing.

Human-in-the-loop  The use of simulated system models for staff training and evaluation.

Iron-in-the-loop  Controlling the real plant in a test cell or other controlled environment.

Model-in-the-loop  Defines a scenario where the controller and the environment are simulated and connected in a closed loop simulation.

Monte Carlo Methods  A class of testing simulations that draw on repeated random samplings of the same job to provide more accurate probalistic, numerical results.

Processor-in-the-loop  Verifies compiled object code which is intended to be deployed in the production system, either on physical hardware or within an simulated environment which emulates that hardware.

Software-in-the-loop  A simulation-based software evaluation which verifies the behaviour of software code under simulated input conditions.

Related forum discussions
forum comment To start a discussion topic about this article, please log in or register.    

Latest Issue

E&T cover image 1411

"This issue we honour a national hero, and the subject of Benedict Cumberbatch's latest film, codebreaker Alan Turing"

E&T jobs

E&T Marketplace

The essential source of engineering products and suppliers.

Subscribe

Choose the way you would like to access the latest news and developments in your field.

Subscribe to E&T