Internship/Evaluation and Implementation of a Jacobian-Free Solver in the Flusepa3D Code (m/f)

Description de l'offre

Internship/Evaluation and Implementation of a Jacobian-Free Solver in the Flusepa3D Code (m/f)

Airbus Safran Launchers Les Mureaux

A 50–50 joint venture established by Airbus Group and Safran, Airbus Safran Launchers brings together all the expertise and assets of these two industry leaders in the fields of civil and military launch vehicles.
Its combination of a unique technological and industrial heritage and the dynamism of a truly passionate workforce is what makes Airbus Safran Launchers what it is today – the world leader in access to space with big ambitions.
With agility embedded right across its business, Airbus Safran Launchers can offer its customers particularly innovative, competitive solutions, striving to boost European space industry to the highest level and guarantee independent access to space.

Airbus is a global leader in aeronautics, space and related services. In 2016, it generated revenues of € 67 billion and employed a workforce of around 134,000. Airbus offers the most comprehensive range of passenger airliners from 100 to more than 600 seats. Airbus is also a European leader providing tanker, combat, transport and mission aircraft, as well as Europe's number one space enterprise and the world's second largest space business. In helicopters, Airbus provides the most efficient civil and military rotorcraft solutions worldwide.

Our people work with passion and determination to make the world a more connected, safer and smarter place. Taking pride in our work, we draw on each other's expertise and experience to achieve excellence. Our diversity and teamwork culture propel us to accomplish the extraordinary - on the ground, in the sky and in space.

Description of the job

Airbus Safran Launchers (Les Mureaux) is looking for an intern (m/f) for a 6-month internship.

The FLUSEPA (1) code was initially developed by Airbus Safran Lauchers to simulate compressible unsteady, viscous and reactive flows around bodies in relative motion (stage separations and space launchers lift-off). When movements are slow or non-existent, the unsteady character of phenomena becomes secondary in the hypothesis of averaged turbulent flows (RANS) and the use of an implicit temporal integration scheme becomes practically inevitable (in order to reduce the costs of simulations). The explicit solver of FLUSEPA now includes an Eulerian modelling of particle flows. The objective of this internship is to implicate this model by considering its coupling with a gas model in a MPI/OPENMP context within FLUSEPA.
FLUSEPA uses a Newton/GMRES algorithm for implicit problem-solving involving aerodynamics, chemistry, as well as turbulence. Within a Newton iteration, the various contributions of chemistry, turbulence and the purely aerodynamic part are decoupled to avoid the inversion of a very large matrix during resolution (and thus memory consumption and important computing time). If the chemistry/aerodynamic decoupling can be considered as acceptable at first glance, the strong particle/aerodynamic coupling however must absolutely be taken into account. For this reason, a completely decoupled resolution for particles seems irrelevant.
The Jacobian-free approach is a promising approach which could win in two areas, i.e. in memory storage and in a complete physical coupling (turbulence, particles, chemistry, and aerodynamics). This method consists in replacing matrix/vector products in the GMRES algorithm by Fréchet derivatives (2) of residues while directly using the explicit solver. This will not only allow the memory of elements to not be stored in the exact Jacobian matrix, but also to perform a total coupling of all the physical phenomena modelled.

This internship will start on 2nd May 2017 (subject to some flexibility).

This position will require a security clearance or will require being eligible for clearance by the recognised authorities.

(1) Finite volume centred on elements, non-structured hybrid, conservative CHIMERE model (overlapping geometric meshing), and upstream off-centre convective flow (GOUDOUNOV), etc.
(2) It will then be necessary to determine an "acceptable" value (which will allow the correct convergence of the method) for small disturbances used in the approximation of derivatives by their numerical differential for each unknown value

Internships at Airbus

Profil recherché

Tasks & accountabilities

Your tasks will include:

·  Carrying out a literature review on Jacobian-free methods, by looking at e.g. the "Jacobian-free Newton-Krylov Methods: A survey of Approaches and Application" paper by Knoll and Keyes, which proposed a good state-of-the-art of the techniques considered,
·  Building a solver using the Jacobian-free method for the aero-particle part and analysing its performances,
·  Completing the developed solver to take into account turbulence and chemical reactions.

Required skills

You are in the final year of engineering school, completing the 2nd year of a Master's degree or in the 5th year of university or in a gap year, in the Engineering field, specialising in Aerodynamics.

You ideally have initial work experience and you have an interest in particular in applied mathematics.

You have good knowledge of methods and numerical calculations and an interest in IT aspects: programming, script writing.

Knowledge of Fortran and/or Python languages would be a plus.
Practical experience with meshing tools (e.g. ANSA, ICEM or GAMBIT) would be a plus.

You are recognised for your ability to work independently.
You are a good team player and have excellent interpersonal and communication skills.

English: advanced level,
French: negotiation level.