Resolvent modelling of near-wall coherent structures in turbulent channel flow

https://doi.org/10.1016/j.ijheatfluidflow.2020.108662Get rights and content

Highlights

  • A good agreement between SPOD and resolvent modes were found for parameter combinations where the lift-up mechanism is present.

  • The lift-up is naturally selected as the preferred mechanism giving rise to near-wall coherent structures in turbulent channel flow.

  • Resolvent analysis provides a simplified model leading to an accurate representation of the streaky structures associated to the lift-up mechanism.

  • Resolvent analysis allows simple predictions of the dominant structures in wall-bounded turbulent flows.

Abstract

Turbulent channel flow was analysed using direct numerical simulations at friction Reynolds numbers Reτ=180 and 550. The databases were studied using spectral proper orthogonal decomposition (SPOD) to identify dominant near-wall coherent structures, most of which turn out to be streaks and streamwise vortices. Resolvent analysis was used as a theoretical approach to model such structures, as it allows the identification of the optimal forcing and most amplified flow response; the latter may be related to the observed relevant structures obtained by SPOD, especially if the gain between forcing and response is much larger than what is found for suboptimal forcings or if the non-linear forcing is white noise. Results from SPOD and resolvent analysis were compared for several combinations of frequencies and wavenumbers. For both Reynolds numbers, the best agreement between SPOD and resolvent modes was observed for the cases where the lift-up mechanism from resolvent analysis is present, which are also the cases where the optimal resolvent gain is dominant. These results confirm the outcomes in our previous studies (Abreu et al., 2019, Abreu et al., 2020), where we used a DNS database of a pipe flow for the same Reynolds numbers.

Introduction

The subject of wall-bounded turbulence has been the topic of many studies in the past decades. Coherent turbulent structures have been studied in several frameworks in wall turbulence, and near-wall streaks are the most typically observed structures (Kline et al., 1967, Gupta et al., 1971). Streaks are elongated structures in the streamwise direction for low Reynolds numbers and close to the wall, composed by regions of alternating low and high momentum located in the viscous and buffer layers with a characteristic spanwise spacing of 100ν/uτ, where uτ is the friction velocity and ν is the kinematic viscosity of the fluid (Marusic et al., 2017). Farther from the wall, larger structures with similar streaky shape are observed (Marusic et al., 2017). The study of a turbulent flow along the surface of a solid body and its interaction with the wall is one of the most fundamental problems in fluid mechanics. Thus, the pursuit for more effective methods to characterize and model near-wall coherent structures is a very relevant problem.

Statistical methods can be used to identify coherent structures present in turbulent flows. Certain turbulence statistics, taken in frequency and wavenumber domain whenever possible, can be optimally represented using a spectral proper orthogonal decomposition (SPOD), described in previous works (Abreu et al., 2017, Abreu et al., 2018, Abreu et al., 2019, Schmidt and Colonius, 2020). Essentially, SPOD extracts the most likely spatial structures that arise in turbulent-flow realisations, called SPOD modes, sorted in terms of their contribution to the overall energy; SPOD is thus a useful method to extract energetic coherent structures in turbulence.

A theoretical approach, involving the Navier–Stokes equations, called resolvent analysis, can be used to analyse the dynamics and to model coherent structures in turbulent flows. Resolvent analysis was introduced and explored in Jovanovic and Bamieh, 2005, Bagheri et al., 2009, McKeon and Sharma, 2010. The method consists in taking the singular value decomposition of the linearised Navier–Stokes operator, considering the mean field as base flow for linearisation, and treating non-linear terms, neglected in the linearisation, as an unknown external forcing. This provides two orthonormal bases, one for forcings and the other for associated flow responses, related by a gain. Response modes with high gains are expected to be dominant in the flow, and associated forcing shows non-linear terms that excite such amplified responses, leading to relevant information about how these are generated. Such linearised responses can often be related to results of hydrodynamic stability theory, with modes corresponding to linear stability eigenfunctions or to non-modal mechanisms such as lift-up (Jovanović and Bamieh, 2005), but in recent years resolvent analysis has been used to study coherent structures in turbulent flows (see the review of (McKeon, 2017) and references therein).

The connection between SPOD modes and the flow responses to stochastic forcing have been explored in recent works (Towne et al., 2018, Abreu et al., 2017, Abreu et al., 2018, Abreu et al., 2020). Indeed, resolvent modes have a direct relationship with energetic structures observed in experimental or numerical flow databases. An important result is that if the forcing can be modeled as spatial white noise, the response modes provided by resolvent analysis should be exactly equal to those obtained by SPOD (Towne et al., 2018). Moreover, for a flow with a dominant optimal forcing, leading to a gain much larger than that of suboptimal ones, the cross-spectral density matrix (CSD) will often be dominated by the leading response obtained in resolvent analysis (Cavalieri et al., 2019). However, despite such general trends, the lack of knowledge on non-linear forcing leads to no a priori guarantee of the accuracy of resolvent analysis in representing turbulent structures, and a quantitative comparison between modelled resolvent modes and educed SPOD modes is carried out on a case-by-case basis.

This paper focuses on the combined analysis of a turbulent channel flow, with SPOD on the one hand, serving to decompose turbulent flow fluctuations, and resolvent analysis on the other hand, used as a theoretical framework. This enables a reduced-order model of the dynamically-relevant flow features. The investigated flows have friction Reynolds numbers Reτ=180 and 550 (note that Reτ is defined in terms of uτ,ν and the channel half-height h). Our previous studies (Abreu et al., 2020, Abreu et al., 2019) dealt with SPOD and resolvent analysis for turbulent pipe flow also at Reτ=180 and 550, and in the present work we apply a similar approach to study turbulent channel flow. This allows a comparison between these two canonical flows for wall turbulence, with a quantitative examination of how resolvent analysis is capable to model coherent structures for both flows. We expect to find similar results for channel and pipe flows, since the same kind of turbulent coherent structures near the wall are found for channel, pipe and also for a zero pressure gradient turbulent boundary layer (ZPGTBL) (Monty et al., 2009), with differences only related to the amplitude of the forcing/pressure gradient (zero for the ZPGTBL, lower for pipes, higher for channels), the surface curvature and the potential interaction with the wake structures.

The remainder of the paper is organised as follows. Section §2 presents a brief description of the simulation databases, and some results of turbulence statistics and spectral analysis. Section §3 proceeds with a description of the mathematical models used to analyse the databases, with a description of the SPOD, used as a data-driven approach, and the mathematical formulation of the resolvent analysis, used as a theoretical framework. Section §4 shows the results of a detailed comparison between SPOD and resolvent modes. The paper is completed with the conclusions in §5.

Section snippets

Numerical simulation description

The direct-numerical-simulation (DNS) databases employed in this work were produced in Martini et al. (2020) (Reτ180) and Morra et al. (2020) (Reτ550) for a turbulend channel flow. The corresponding Reynolds numbers based on the bulk velocity Ub are Reb=2800 and 10000, respectively. The simulations were carried out using the code Channelflow (see Channelflow 2.0 (2018) for details), which is based on a pseudo-spectral spatial discretisation using Fourier modes in streamwise and spanwise

Data-driven approach

Proper orthogonal decomposition (POD) was first introduced by Lumley (Lumley, 1967) in the context of turbulence, and consists in obtaining, from an ensemble of realizations of the flow field, a basis functions that maximises the mean square energy. An approach, also referred to as space-only POD (Towne et al., 2018), is the most commonly employed form of POD to generates spatial modes. Alternatively, POD in frequency domain, or spectral proper orthogonal decomposition (SPOD) as introduced in

Projection coefficient

In order to perform detailed quantitative comparisons between the first SPOD mode and the optimal response from resolvent analysis, we adopted the metric used in our previous work (Abreu et al., 2020), which evaluates the agreement for several values of wavelengths λx and λz at a fixed frequency ω, defined as:β=q̂1SPOD,q̂1RES||q̂1SPOD||·||q̂1RES||,where q̂1SPOD=[û1SPOD,v̂1SPOD,ŵ1SPOD] is the first SPOD mode; q̂1RES=[û1RES,v̂1RES,ŵ1RES] is the optimal response from resolvent

Conclusions

In the present study we used a data-driven approach based on SPOD, to identify space- and time-dependent coherent structures in a DNS database of turbulent channel flow for friction Reynolds numbers Reτ=180 and 550. Resolvent analysis was used as a theoretical approach to model such structures, and the mean velocity u¯ was used to linearise the Navier–Stokes equations. The statistically stationary time, and the homogeneous directions in x and z, of this simulation allow the evaluation of SPOD

CRediT authorship contribution statement

Leandra I Abreu: Conceptualization, Methodology, Software, Validation, Investigation, Writing - original draft, Formal analysis. André V.G. Cavalieri: Conceptualization, Supervision, Investigation, Validation, Writing - review & editing, Project administration, Funding acquisition, Resources. Philipp Schlatter: Data curation, Investigation, Validation, Writing - review & editing. Ricardo Vinuesa: Data curation, Investigation, Validation, Writing - review & editing. Dan S. Henningson:

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The authors acknowledge the financial support received from Conselho Nacional de Desenvolvimento Científico e Tecnológico, CNPq, under Grant No. 310523/2017-6, and by CAPES through the PROEX program. We also acknowledge funding from Centro de Pesquisa e Inovação Sueco-Brasileiro (CISB).

References (37)

  • G. Eitel-Amor et al.

    Simulation and validation of a spatially evolving turbulent boundary layer up to Reθ=8300

    Int. J. Heat Fluid Flow

    (2014)
  • C. Picard et al.

    Pressure velocity coupling in a subsonic round jet

    Int. J. Heat Fluid Flow

    (2000)
  • L.I. Abreu et al.

    Reduced-order models to analyse coherent structures in turbulent pipe flow

    Turbulence and Shear Flow Phenomena

    (2019)
  • Abreu, L.I., Cavalieri, A.V., Schlatter, P., Vinuesa, R., Henningson, D.S., 2020. SPOD and resolvent analysis of...
  • S.J. Kline et al.

    The structure of turbulent boundary layers

    J. Fluid Mech.

    (1967)
  • A. Gupta et al.

    Spatial structure in the viscous sublayer

    J. Fluid Mech.

    (1971)
  • I. Marusic et al.

    Scaling of the streamwise turbulence intensity in the context of inner-outer interactions in wall turbulence

    Phys. Rev. Fluids

    (2017)
  • L.I. Abreu et al.

    Coherent hydrodynamic waves and trailing-edge noise

    23rd AIAA/CEAS Aeroacoustics Conference

    (2017)
  • Abreu, L.I., Cavalieri, A.V.G., Schlatter, P., Vinuesa, R., Henningson, D., 2018. Wavepackets in a turbulent flow over...
  • L.I. Abreu et al.

    Resolvent analysis applied to acoustic analogies

    25th AIAA/CEAS Aeroacoustics Conference

    (2019)
  • O.T. Schmidt et al.

    Guide to spectral proper orthogonal decomposition

    AIAA J.

    (2020)
  • M.R. Jovanovic et al.

    Componentwise energy amplification in channel flows

    J. Fluid Mech.

    (2005)
  • S. Bagheri et al.

    Input-output analysis and control design applied to a linear model of spatially developing flows

    Appl. Mech. Rev.

    (2009)
  • B.J. McKeon et al.

    A critical-layer framework for turbulent pipe flow

    J. Fluid Mech.

    (2010)
  • M.R. Jovanović et al.

    Componentwise energy amplification in channel flows

    J. Fluid Mech.

    (2005)
  • B. McKeon

    The engine behind (wall) turbulence: perspectives on scale interactions

    J. Fluid Mech.

    (2017)
  • A. Towne et al.

    Spectral proper orthogonal decomposition and its relationship to dynamic mode decomposition and resolvent analysis

    J. Fluid Mech.

    (2018)
  • A. Cavalieri et al.

    Wave-packet models for jet dynamics and sound radiation

    Appl. Mech. Rev.

    (2019)
  • Cited by (0)

    View full text