ttncwt_JCP20_accept.pdf (10.61 MB)
Download fileApplication of the PODFS method to inlet turbulence generated using the digital filter technique
journal contribution
posted on 2020-05-29, 10:07 authored by NCW Treleaven, M Staufer, Adrian SpencerAdrian Spencer, Andrew GarmoryAndrew Garmory, Gary PageGary PageIn the past the digital filter technique has been used to successfully generate inflow turbulence for a number of academic and industrially relevant reacting and non-reacting flows. Weaknesses of the method include the requirement that the filter be computed over a structured mesh and can require extremely long computation times in cases where the turbulent length scales or filter widths are large as compared to the mesh spacing. Once computed, the inflow data may be saved and reused, however tens-of-thousands of time steps worth of data must be saved, copied and re-loaded into memory for each new computation and temporal interpolation must be used if the time step is adjusted. The newly developed PODFS (proper orthogonal decomposition Fourier series) method uses proper orthogonal decomposition (POD) to compress this large data set into a handful of modes that are optimally chosen to represent the high energy turbulent structures while a Fourier series representation of the temporal components of the POD modes means that the data becomes continuous, periodic and flexible in terms of time step. The PODFS is first applied to a set of inflow data generated using the digital filter method and used to simulate a turbulent planar jet with the results compared against a DNS simulation. A practical application of the method is then demonstrated where it is used to generate inlet turbulence from a time averaged URANS profile for a swirl stabilised combustor case. In this case, two PODFS models are linearly combined, one that represents acoustic forcing from downstream and one that represents turbulent fluctuations. This highlights a feature of the method which is its ability to represent different flow phenomena using the linear addition of two or more PODFS models. A subsequent LES calculation shows that the method results in the correct penetration of the airflow jets whilst neglecting the inlet turbulence results in the incorrect jet penetration depth.
Funding
CDT in Gas Turbine Aerodynamics : EP/L015943/1
CFD Modelling of the acoustic response of sprays : Karen Muncey
Rolls-Royce
EPSRC (Engineering and Physical Sciences Research Council) through the Centre for Doctoral Training in Gas Turbine Aerodynamics (grant ref. EP/L015943/1) and EPSRC grant ref EP/M023893/1.
History
School
- Aeronautical, Automotive, Chemical and Materials Engineering
Department
- Aeronautical and Automotive Engineering
Published in
Journal of Computational PhysicsVolume
415Issue
August 2020Publisher
Elsevier BVVersion
- AM (Accepted Manuscript)
Rights holder
© Elsevier Inc.Publisher statement
This paper was accepted for publication in the journal Journal of Computational Physics and the definitive published version is available at https://doi.org/10.1016/j.jcp.2020.109541.Acceptance date
2020-05-05Publication date
2020-05-14Copyright date
2020ISSN
0021-9991Publisher version
Language
- en