Flow control applied to transitional flows inputoutput analysis

  • Slides: 52
Download presentation
Flow control applied to transitional flows: input-output analysis, model reduction and control Linné Flow

Flow control applied to transitional flows: input-output analysis, model reduction and control Linné Flow Centre KTH Mechanics Dan Henningson collaborators Shervin Bagheri, Espen Åkervik Luca Brandt, Peter Schmid 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 1

Outline • Introduction with input-output configuration • Matrix-free methods for flow stability using Navier-Stokes

Outline • Introduction with input-output configuration • Matrix-free methods for flow stability using Navier-Stokes snapshots Edwards et al. (1994), … • Global modes and transient growth Cossu & Chomaz (1997), … Linné Flow Centre KTH Mechanics • Input-output characteristics of Blasius BL and reduced order models based on balanced truncation Rowley (2005), … • LQG feedback control based on reduced order model • Conclusions 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 2

Message • Need only snapshots from a Navier-Stokes solver (with adjoint) to perform stability

Message • Need only snapshots from a Navier-Stokes solver (with adjoint) to perform stability analysis and control design for complex flows • Main example Blasius BL, but many other more complex flows are now tractable … Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 3

Linearized Navier-Stokes for Blasius flow Continuous formulation Discrete formulation Linné Flow Centre KTH Mechanics

Linearized Navier-Stokes for Blasius flow Continuous formulation Discrete formulation Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 4

Input-output configuration for linearized N-S Linné Flow Centre KTH Mechanics 7 th European Fluid

Input-output configuration for linearized N-S Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 5

Solution to the complete input-output problem Linné Flow Centre KTH Mechanics • Initial value

Solution to the complete input-output problem Linné Flow Centre KTH Mechanics • Initial value problem: flow stability • Forced problem: input-output analysis 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 6

The Initial Value Problem Linné Flow Centre KTH Mechanics • Asymptotic stability analysis: Global

The Initial Value Problem Linné Flow Centre KTH Mechanics • Asymptotic stability analysis: Global modes of the Blasius boundary layer • Transient growth analysis: Optimal disturbances in Blasius flow 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 7

Dimension of discretized system Linné Flow Centre KTH Mechanics • Matrix A very large

Dimension of discretized system Linné Flow Centre KTH Mechanics • Matrix A very large for complex spatially developing flows • Consider eigenvalues of the matrix exponential, related to eigenvalues of A • Use Navier-Stokes solver (DNS) to approximate the action of matrix exponential or evolution operator 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 8

Krylov subspace with Arnoldi algorithm • Krylov subspace created using NS-timestepper • Orthogonal basis

Krylov subspace with Arnoldi algorithm • Krylov subspace created using NS-timestepper • Orthogonal basis created with Gram-Schmidt • Approximate eigenvalues from Hessenberg matrix H Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 9

Global spectrum for Blasius flow • Least stable eigenmodes equivalent using time-stepper and matrix

Global spectrum for Blasius flow • Least stable eigenmodes equivalent using time-stepper and matrix solver • Least stable branch is a global representation of Tollmien. Schlichting (TS) modes Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 10

Global TS-waves • Streamwise velocity of least damped TS-mode • Envelope of global TS-mode

Global TS-waves • Streamwise velocity of least damped TS-mode • Envelope of global TS-mode identical to local spatial growth • Shape functions of local and global modes identical Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 11

Optimal disturbance growth • Optimal growth from eigenvalues of • Krylov sequence built by

Optimal disturbance growth • Optimal growth from eigenvalues of • Krylov sequence built by forward-adjoint iterations Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 12

Evolution of optimal disturbance in Blasius flow • Full adjoint iterations (black) of TS-branch

Evolution of optimal disturbance in Blasius flow • Full adjoint iterations (black) of TS-branch modes only (magenta) • Transient since disturbance propagates out of domain sum Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 13

Snapshots of optimal disturbance evolution • Initial disturbance leans against the shear raised up

Snapshots of optimal disturbance evolution • Initial disturbance leans against the shear raised up by Orrmechanism into propagating TS-wavepacket Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 14

The forced problem: input-output • • • Ginzburg-Landau example Input-output for 2 D Blasius

The forced problem: input-output • • • Ginzburg-Landau example Input-output for 2 D Blasius configuration Model reduction Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 15

Ginzburg-Landau example • Entire dynamics vs. input-output time signals Linné Flow Centre KTH Mechanics

Ginzburg-Landau example • Entire dynamics vs. input-output time signals Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 16

Input-output operators • Past inputs to initial state: class of initial conditions possible to

Input-output operators • Past inputs to initial state: class of initial conditions possible to generate through chosen forcing • Initial state to future outputs: possible outputs from initial condition • Past inputs to future outputs: Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 17

Most dangerous inputs and the largest outputs Linné Flow Centre KTH Mechanics • Eigenmodes

Most dangerous inputs and the largest outputs Linné Flow Centre KTH Mechanics • Eigenmodes of Hankel operator – balanced modes • Controllability Gramian • Observability Gramian 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 18

Controllability Gramian for GL-equation Linné Flow Centre KTH Mechanics • Correlation of actuator impulse

Controllability Gramian for GL-equation Linné Flow Centre KTH Mechanics • Correlation of actuator impulse response in forward solution • POD modes: • Ranks states most easily influenced by input • Provides a means to measure controllability Input 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 19

Observability Gramian for GL-equation Output Linné Flow Centre KTH Mechanics • Correlation of sensor

Observability Gramian for GL-equation Output Linné Flow Centre KTH Mechanics • Correlation of sensor impulse response in adjoint solution • Adjoint POD modes: • Ranks states most easily sensed by output • Provides a means to measure observability 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 20

Balanced modes: eigenvalues of the Hankel operator • Combine snapshots of direct and adjoint

Balanced modes: eigenvalues of the Hankel operator • Combine snapshots of direct and adjoint simulation • Expand modes in snapshots to obtain smaller eigenvalue problem Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 21

Snapshots of direct and adjoint solution in Blasius flow Direct simulation: Adjoint simulation: Linné

Snapshots of direct and adjoint solution in Blasius flow Direct simulation: Adjoint simulation: Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 22

Balanced modes for Blasius flow adjoint forward Linné Flow Centre KTH Mechanics 7 th

Balanced modes for Blasius flow adjoint forward Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 23

Properties of balanced modes • Largest outputs possible to excite with chosen forcing •

Properties of balanced modes • Largest outputs possible to excite with chosen forcing • Balanced modes diagonalize observability Gramian • Adjoint balanced modes diagonalize controllability Gramian • Ginzburg-Landau example revisited Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 24

Model reduction • Project dynamics on balanced modes using their biorthogonal adjoints • Reduced

Model reduction • Project dynamics on balanced modes using their biorthogonal adjoints • Reduced representation of input-output relation, useful in control design Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 25

Impulse response Disturbance Actuator Sensor Objective Linné Flow Centre KTH Mechanics Disturbance Objective DNS:

Impulse response Disturbance Actuator Sensor Objective Linné Flow Centre KTH Mechanics Disturbance Objective DNS: n=105 ROM: m=50 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 26

Frequency response From all inputs to all outputs Linné Flow Centre KTH Mechanics DNS:

Frequency response From all inputs to all outputs Linné Flow Centre KTH Mechanics DNS: n=105 ROM: m=80 m=50 m=2 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 27

Feedback control • • LQG control design using reduced order model Blasius flow example

Feedback control • • LQG control design using reduced order model Blasius flow example Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 28

Optimal Feedback Control – LQG cost function w Linné Flow Centre KTH Mechanics g

Optimal Feedback Control – LQG cost function w Linné Flow Centre KTH Mechanics g (noise) Ly f=Kk z controller Find an optimal control signal f (t) based on the measurements y(t) such that in the presence of external disturbances w(t) and measurement noise g(t) the output z(t) is minimized. Solution: LQG/H 2 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 29

LQG controller formulation with DNS • Apply in Navier-Stokes simulation Linné Flow Centre KTH

LQG controller formulation with DNS • Apply in Navier-Stokes simulation Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 30

Performance of controlled system controller Noise Linné Flow Centre KTH Mechanics Sensor Actuator Objective

Performance of controlled system controller Noise Linné Flow Centre KTH Mechanics Sensor Actuator Objective 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 31

Performance of controlled system Linné Flow Centre KTH Mechanics Noise Sensor Actuator Objective Control

Performance of controlled system Linné Flow Centre KTH Mechanics Noise Sensor Actuator Objective Control off Cheap Control Intermediate control Expensive Control 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 32

Conclusions • Complex stability/control problems solved using Krylov/Arnoldi methods based on snapshots of forward

Conclusions • Complex stability/control problems solved using Krylov/Arnoldi methods based on snapshots of forward and adjoint Navier-Stokes solutions • Optimal disturbance evolution brought out Orr-mechanism and propagating TS-wave packet automatically Linné Flow Centre KTH Mechanics • Balanced modes give low order models preserving inputoutput relationship between sensors and actuators • Feedback control of Blasius flow Reduced order models with balanced modes used in LQG control Controller based on small number of modes works well in DNS • Outlook: incorporate realistic sensors and actuators in 3 D problem and test controllers experimentally 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 33

Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18

Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 34

Background • Global modes and transient growth Ginzburg-Landau: Cossu & Chomaz (1997); Chomaz (2005)

Background • Global modes and transient growth Ginzburg-Landau: Cossu & Chomaz (1997); Chomaz (2005) Waterfall problem: Schmid & Henningson (2002) Blasius boundary layer, Ehrenstein & Gallaire (2005); Åkervik et al. (2008) Recirculation bubble: Åkervik et al. (2007); Marquet et al. (2008) • Matrix-free methods for stability properties Krylov-Arnoldi method: Edwards et al. (1994) Stability backward facing step: Barkley et al. (2002) Optimal growth for backward step and pulsatile flow: Barkley et al. (2008) Linné Flow Centre KTH Mechanics • Model reduction and feedback control of fluid systems Balanced truncation: Rowley (2005) Global modes for shallow cavity: Åkervik et al. (2007) Ginzburg-Landau: Bagheri et al. (2008) 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 35

Jet in cross-flow Countair rotating vortex pair Shear layer vortices Horseshoe vortices Linné Flow

Jet in cross-flow Countair rotating vortex pair Shear layer vortices Horseshoe vortices Linné Flow Centre KTH Mechanics Wake region 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 36

Direct numerical simulations • DNS: Fully spectral and parallelized • Self-sustained global oscillations •

Direct numerical simulations • DNS: Fully spectral and parallelized • Self-sustained global oscillations • Probe 1– shear layer • Probe 2 – separation region Linné Flow Centre KTH Mechanics 1 2 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 2 Vortex identification criterion 37

Basic state and impulse response • Steady state computed using the SFD method (Åkervik

Basic state and impulse response • Steady state computed using the SFD method (Åkervik et. al. ) Steady state Unsteady Time-averaged Steady-state Linné Flow Centre KTH Mechanics • Energy growth of perturbation Perturbation 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 38

Global eigenmodes • Global eigenmodes computed using ARPACK • Growth rate: 0. 08 •

Global eigenmodes • Global eigenmodes computed using ARPACK • Growth rate: 0. 08 • Strouhal number: 0. 16 1 st global mode Linné Flow Centre KTH Mechanics time Perturbation energy Global mode energy 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 39

Optimal sum of eigenmodes Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics

Optimal sum of eigenmodes Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 40

Global view of Tollmien-Schlichting waves • Global temporal growth rate damped and depends on

Global view of Tollmien-Schlichting waves • Global temporal growth rate damped and depends on length of domain and boundary conditions • Single global mode captures local spatial instability • Sum of damped global modes represents convectively unstable disturbances • TS-wave packet grows due to local exponential growth, but globally represents a transient disturbance since it propagates out of the domain Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 41

3 D Blasius optimals • • Streamwise vorticies create streaks for long times Optimals

3 D Blasius optimals • • Streamwise vorticies create streaks for long times Optimals for short times utilizes Orr-mechanism Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 42

Input-output analysis Linné Flow Centre KTH Mechanics • Inputs: Disturbances: roughness, free-stream turbulence, acoustic

Input-output analysis Linné Flow Centre KTH Mechanics • Inputs: Disturbances: roughness, free-stream turbulence, acoustic waves Actuation: blowing/suction, wall motion, forcing • Outputs: Measurements of pressure, skin friction etc. • Aim: preserve dynamics of input-output relationship in reduced order model used for control design 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 43

A long shallow cavity • Basic flow from DNS with SFD: • Åkervik et

A long shallow cavity • Basic flow from DNS with SFD: • Åkervik et al. , Phys. Fluids 18, 2006 • Strong shear layer at cavity top and recirculation at the downstream end of the cavity Linné Flow Centre KTH Mechanics Åkervik E. , Hoepffner J. , Ehrenstein U. & Henningson, D. S. 2007. Optimal growth, model reduction and control in a separated boundary-layer flow using global eigenmodes. J. Fluid Mech. 579: 305 -314. 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 44

Global spectra • Global eigenmodes found using Arnoldi method • About 150 eigenvalues converged

Global spectra • Global eigenmodes found using Arnoldi method • About 150 eigenvalues converged and 2 unstable Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 45

Most unstable mode Linné Flow Centre KTH Mechanics • Forward and adjoint mode located

Most unstable mode Linné Flow Centre KTH Mechanics • Forward and adjoint mode located in different regions implies non-orthogonal eigenfunctions/non-normal operator • Flow is sensitive where adjoint is large 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 46

Maximum energy growth • Eigenfunction expansion in selected modes • Optimization of energy output

Maximum energy growth • Eigenfunction expansion in selected modes • Optimization of energy output Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 47

Development of wavepacket • x-t diagrams of pressure at y=10 using eigenmode expansion •

Development of wavepacket • x-t diagrams of pressure at y=10 using eigenmode expansion • Wavepacket generates pressure pulse when reaching downstream lip • Pressure pulse triggers another wavepacket at upstream lip Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 48

LQG feedback control cost function Linné Flow Centre KTH Mechanics Reduced model of real

LQG feedback control cost function Linné Flow Centre KTH Mechanics Reduced model of real system/flow Estimator/ Controller 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 49

Riccati equations for control and estimation gains Linné Flow Centre KTH Mechanics 7 th

Riccati equations for control and estimation gains Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 50

Feedback control of cavity disturbances • Project dynamics on least stable global modes •

Feedback control of cavity disturbances • Project dynamics on least stable global modes • Choose spatial location of control and measurements • LQG control design Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 51

Controller performance • Least stable eigenvalues are rearranged • Exponential growth turned into exponential

Controller performance • Least stable eigenvalues are rearranged • Exponential growth turned into exponential decay • Good performance in DNS using only 4 global modes Linné Flow Centre KTH Mechanics 7 th European Fluid Mechanics Conference, Manchester, 15 -18 September, 2008. 52