➤


ESTECO, explained: ‘Tis means that, for a global decision, you may consider simplified formulas constructed on the basis of the previous simulation being done inside our environment. ‘But once you come to a decision – or


possible configuration – you may fire from the top down to the tiny details all the simulations, which are needed at the highest level of fidelity. So, splitting the problem of doing things at the right fidelity in different situations, and in different moments of the process,’ he added. For example, ESTECO is involved in the


Composelector project to develop a multi- scale modelling approach for the selection of composite materials. Poloni explained: ‘A material property at the global level is something that has to be used in the framework of the aircraſt simulation, but at the very low level you need the performance of molecules of the materials in relation to the mixture.’ If you look at a composite material, you


have to consider the fibres, resin and the interaction between the fibres and the resin at the atomic level. However, when you want to model this even in small components, you need to model these properties. Poloni said: ‘What we can provide is the way to include even the tiniest simulations at the very low scale, up to the higher scale and components, and use that information in the optimisation of the single part. ‘Te same analogy could be used at


a different scale considering smaller components compared to the complete aircraſt,’ he added.


Multi-fidelity Tere are many different scenarios where a multi-fidelity approach is required. For example, an engine manufacturer will need a different level of fidelity in their engine model compared to an OEM, who wants an engine model to integrate into a complete aircraſt simulation. Chris Hayhurst, consulting manager at MathWorks, said: ‘Tat [OEM] needs to run real-time in a pilot-in-loop simulator, and many times faster than real- time for desktop analysis. A supplier may provide a lower fidelity model, which can also serve to protect their IP, however this also adds the overhead of comparing the match or functionality between the two models.’ OEMS are starting to contractually require


suppliers to deliver executable models of components or subsystems, prior to delivery of the hardware itself, so that these models can be integrated into the OEMs system- level models, according to Hayhurst, who


28 SCIENTIFIC COMPUTING WORLD


SIMULATION IS THE


ONLY COST-EFFECTIVE WAY TO ENSURE THESE COMPONENTS CAN SURVIVE THE


ENVIRONMENT THEY WILL OPERATE IN BEFORE THEY ARE TESTED IN VERY EXPENSIVE EXPERIMENTS


added: ‘At a technical level, such shared models can foster close engineer-to-engineer co-operation, greater understanding of component behaviour and performance and, at a business level, reduced risk.’ Computational fluid dynamics experts also


need to utilise many tools to simulate various vehicles, such as airplanes and rockets. Tere are a range of flow conditions, such as speeds (low versus high speed), turbulence levels (laminar versus turbulent), time representations (steady versus unsteady). Kiris said: ‘Certain numerical tools are better at certain conditions, and it is the realm


of the computation fluid dynamics (CFD) practitioner to select the correct tool to represent the flow physics at the required fidelity. ‘Within NAS, we heavily utilise Navier-


Stokes equation based flow solvers, but are now branching out into other representations, such as Lattice Boltzmann-based. It is also common to combine the higher-fidelity methods with reduced order models.’ ANSYS is interested in the Integrated


Computational Materials Engineering workflow (ICME) to enable simulation ‘from the atom to the airplane’, according to Harwood, who said: ‘While the practical use of full ICME methods is still somewhat of a vision, progress is being made. Today, it is common to use both global models of the whole aircraſt (that use reduced levels of fidelity) to get an idea of load paths and magnitude, and then use these to seed more detailed local models of individual components and subsystems and analyse them in full detail, including multi-scale effects. ‘Of course, as computational power


increases and soſtware capabilities become better, the line between these global and local models is blurring, with local models becoming bigger and less local and global


@scwmagazine l www.scientific-computing.com


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36