Cerfacs Enter the world of high performance ...

Mono-channel methods and modeling for LES of turbomachinery stages

 , , , , ,

Required Education : Formation ingénieur, Master ou equivalent
Start date : 1 October 2017
Mission duration : 3 ans
Deadline for applications : 30 November 2017

Context / Contexte

Large Eddy Simulation (LES) is clearly identified as the next generation of turbulent modeling tool [1] that will improve high Reynolds number flow predictions of complex systems.  Although, the demonstration has been evidenced in the context of fundamental research and transfer has been performed by scientists for specific applications such as combustion…, its application and generalization to any real industrial problems requires today access to dedicated and specific solvers which can be leveraged only by use of high performance computers.  This combined to the ever-increasing complexity and the increasing size of the computational domain needed to treat real problems, makes the use of advanced high fidelity CFD tools difficult in industry. The typical example applies to turbomachinery applications where even in academia transition to LES has only appeared locally due to specificity of these high-fidelity tools and their need for High Performance Computing (CPU) [2].  This specific transition has recently occurred within CERFACS for turbomachinery flows [3-10]. The need is now to continue this effort to ease this transfer to CERFACS' industrial partners.  To do so, SAFRAN ST and CERFACS will joint their respective efforts and knowledge to provide reliable solutions obtained within this PhD work in an attempt to improve state-of-the art of LES in Turbomachinery as detailed below.

Description / Description

Large Eddy Simulation (LES) of turbomachinery flows has recently appeared as a valuable CFD tool to improve design and understanding of such highly complex systems [1,2].   Despite the necessarily modeling difficulties present in real applications which are associated to turbulence, boundary layer dynamics with transition, re-laminarization, pressure gradients…, wall modeled LES has indeed produced interesting demonstrations [3].  These simulations however come today with important geometrical changes that are necessary to render accessible such CPU intensive simulations that are only possible on reduced size computational domains. In the end and because of these changes, the flow dynamics present in these LES prediction is not fully compliant with the reality: i.e. the deterministic features issued by the rotor/stator interactions are modified by the geometrical changes.

The above shortcoming has very early on been identified as a clear bottle-neck for any CFD method and alternatives have been proposed [11-21]. Among all possible solutions, the mono-channel context is of foremost interest since it proposes to produce a representative simulation of the real configuration based on a single passage simulation of all the different stages of the machine: i.e. the minimal computational domain possible [19,20]. The counter part is that because of the non-matching azimuthal extend of each individual channel when compared to its neighboring channels, reconstructions and models are required.  Several strategies are today available in the context of Unsteady or not Reynolds Average Navier Stokes (RANS or URANS) simulations [21]. These solutions however disregard totally the importance of turbulence and the link potentially present between deterministic and turbulent features in the context of rotor/stator interactions. A direct consequence is that these models are difficult if not impossible to use in the context of LES.

The objective of this PhD work is therefore to investigate and provide modeling solutions compatible for mono-canal LES.  To do so, three axes will be developed through the course of this work.

Year 1: A priori analyses of existing multi-passages and multi-stages LES predictions of a real CFD friendly configuration such as the one obtained for the CREATE configuration. The objective is here to obtained dedicated analyses on the coupling between deterministic and turbulent fields at the various rotor/stator interfaces present in this real application. Fundamental questions, such as the determination of the deterministic modes, their spatial coherence and intensity as well as the effective validation of a chorochronic context will be probed.  The specific use of a PoD basis as proposed by G. Mouret et al. for this problem will be addressed at this occasion [5-7,9,10].

Year 2: A posteriori validation of the PoD based mono-channel method of G. Mouret et al [11]. A series of tests with incremental complexity will be devised to better master the method and improve it if required.  The intent is here to bound and understand the limits of such a modeling techniques. Corrections and improvements will mainly rely on the understanding gathered through step 1/.

Year 3: End application of the modeling strategy on a real application such as the CREATE configuration.

All of the above described steps will take place within CERFACS, with the help of SAFRAN ST personnel.  Although most of the research will be performed at CERFACS and with CERFACS tools as well as using already available predictions, periods of time will be devoted to exchanges with SAFRAN ST for a maximum duration of 6 months.


Contacts / Contacts

Name: Gicquel Laurent

Phone: 05-61-19-30-46

Fax: 05-61-19-30-00

Email: lgicquel@cerfacs.fr

Name: Florent Duchaine

Phone: 05 61 19 30 73

Email: duchaine@cerfacs.fr


List of References  / Liste de Références:

[1] Sagaut P., Large Eddy Simulation for incompressible flows – An introduction, 2nd Edition, 2002, Theoretical, Mathematical & Computational Physics, Springer

[2] Laskowski G. et al, Future Directions of High-fidelity CFD for Aero-Thermal Turbomachinery Research, Analysis and Design,  AIAA Aviation, 13-17 June 2016, Washington D.C., 46th AIAA Fluid dynamics Conference.

[3] Papadogiannis, D., Duchaine, F., Gicquel, L. Y. M., Wang, G. and Moreau, S. (2016) Effects of Subgrid Scale Modeling on the Deterministic and Stochastic Turbulent Energetic Distribution in Large-Eddy Simulations of a High-Pressure Turbine Stage, Journal of Turbomachinery, 138 (9), pp. 091005-091005-10

[4] Papadogiannis, D., Duchaine, F., Gicquel, L., Wang, G., Moreau, S. and Nicoud, F. (2015) Assessment of the indirect combustion noise generated in a transonic high-pressure turbine stage, Journal of Engineering for Gas Turbines and Power-Transactions of the ASME, 138 (4), pp. 0415030 (8 pp.), doi:10.1115/1.4031404

[5] Wang, G., Duchaine, F., Papadogiannis, D., Duran, I., Moreau, S. and Gicquel, L. Y. M. (2014) An overset grid method for large eddy simulation of turbomachinery stages, Journal of Computational Physics, 274 (October), pp. 333-355

[6] J. de Laborderie et al, Analysis of a high-pressure multistage axial compressor at off-design conditions with coarse large eddy simulations, Proceedings of the 12th European Conference on Turbomachinery Fluid dynamics & Thermodynamics, ETC12, April 3-7, Stockholm, Sweden, 2017.

[7] J. de Laborderie et al, Application of an overset grid method to the large eddy simulation of high-speed multistage axial compressor, Proceedings of the ASME Turbo Expo 2016, GT2016, June 13-17, Seoul, South Korea, 2016.

[8] L.M. Segui et al, LES of the LS89 cascade : influence of inflow turbulence on the flow predictions, Proceedings of the 12th European Conference on Turbomachinery Fluid dynamics & Thermodynamics, ETC12, April 3-7, Stockholm, Sweden, 2017.

[9] de Laborderie, J., Duchaine, F., Vermorel, O., Gicquel, L. Y. M. and Moreau, S. (2016) Application of an overset grid method to the Large-Eddy Simulation of a high speed multistage axial compressor, ASME Turbo Expo 2016. Seoul, South Korea 2016, p. GT2016-56344

[10] de Laborderie J. et al, Numerical Analysis of a high-order unstructured overset grid method of compressible LES of Turbomachinery, submitted to Journal of Computational Physics, March 2017.

[11] Mouret, G. et al, Adaptation of Phase-Lagged Boundary Condition to Large Eddy Simulation in Turbomachinery configurations, Journal of Turbomachinery, April 2016, Vol. 138, pp 041003-1 to -11.

[12] M. M. Rai. Navier-Stokes simulations of rotor/stator interaction using patched and overlaid grids. J. Propulsion, 3(5):387-396, 1987.

[13] F. Bardoux. Modélisation des interactions instationnaires rotor-stator en turbomachine multi-etages. PhD thesis, Ecole Centrale de Lyon, 2000.

[14] B. Greschner and F. Thiele. Wall Modeled LES simulation of Rotor-Stator-Cascade Broadband Noise. In 32nd AIAA Aeroacoustics Conference, Portland, Oregon, USA, 2011.

[15] W. A. McMullan and G. J. Page. Towards Large Eddy Simulation of gas turbine compressors. Progress in Aerospace Sciences, 52(Jul.):30-47, 2012.

[16] A. Arnone and R. Pacciani. Rotor-stator interaction analysis using the Navier- Stokes equations and a multigrid method. J. of Turbomachinery, 118(10):679-689, 1996.

[17] A. Fourmaux. Assessment of a low storage technique for multistage turbomachinery Navier Stokes computations. In ASME winter annual meeting, 1994.

[18] C. Cornelius, T. Biesinger, L. Zori, R. Campregher, P. Galpin, and A. Braune. Efficient time resolved multistage CFD analysis applied to axial compressors. In Proceedings of ASME Turbo Expo 2014: Turbine Technical Conference and Exposition, pages 1-14, Dusseldorf, Germany, 2014.

[19] J. I. Erdos and E. Alzner. Computation of Unsteady Transonic Flows Through Rotating and Stationary Cascades. I – Method of Analysis. Technical report, NASA, 1977.

[20] G. A. Gerolymos and V. Chapin. Generalized Expression of Chorochronic Periodicity in Turbomachinery Blade-Row Interaction. La Recherche Aerospatiale, 5:69-73, 1991.

[21] L. He. An euler solution for unsteady flows around oscillating blades. J. of Turbomachinery, 112(4):714-722, 1989.

[22] L. Zori, P. Galpin, R. Campregher, and J.C. Morales. Time transformation simulation of 1.5 stage transonic compressor. In Proceedings of ASME Turbo Expo 2015: Turbine Technical Conference and Exposition, pages 1{13, Montreal, Canada, 2015.