numerical simulation of a viscous oldroyd b model
play

Numerical simulation of a viscous Oldroyd-B model Bangwei She, M - PowerPoint PPT Presentation

Numerical simulation of a viscous Oldroyd-B model Bangwei She, M aria Luk a cov a JGU-Mainz cooperation with Prof. M.Tabata, H. Notsu, A. Tezuka Waseda University IRTG 1529 Mathematical fluid dynamics Sino-German Symposium on


  1. Numerical simulation of a viscous Oldroyd-B model Bangwei She, M´ aria Luk´ aˇ cov´ a JGU-Mainz cooperation with Prof. M.Tabata, H. Notsu, A. Tezuka Waseda University IRTG 1529 Mathematical fluid dynamics Sino-German Symposium on advanced numerical methods for compressible fluid mechanics and related problems

  2. Oldroyd-B model  Re ( ∂ u β ∂ t + u · ∇ u ) = −∇ p + α ∆ u + We ∇ · σ  ∇ · u = 0 (1) ∂ t + ( u · ∇ ) σ − ∇ u · σ − σ · ( ∇ u ) T = 1 ∂ σ We ( I − σ )  u : velocity, p : pressure Re : Reynolds number σ : conformation tensor, symmetric positive definite, elastic We : Weissenberg number, relaxation time over characteristic time α : ratio of Newtonian viscocity in total viscocity. β = 1 − α . B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 2 / 20

  3. Existence results local in time existence: Guillope and Saut, 1990 Renardy 1991 Jourdain, Lelivre and Bris, 2004 Li and Zhang,2004 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 3 / 20

  4. Existence results local in time existence: Guillope and Saut, 1990 Renardy 1991 Jourdain, Lelivre and Bris, 2004 Li and Zhang,2004 global existence with small initial data: Guillope and Saut, 1990 Lin, Liu and Zhang, 2005 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 3 / 20

  5. Existence results local in time existence: Guillope and Saut, 1990 Renardy 1991 Jourdain, Lelivre and Bris, 2004 Li and Zhang,2004 global existence with small initial data: Guillope and Saut, 1990 Lin, Liu and Zhang, 2005 regularized model: regularity in 2D, Constantin,2012 global existence of regularized model: Barrett, 2014 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 3 / 20

  6. Existence results local in time existence: Guillope and Saut, 1990 Renardy 1991 Jourdain, Lelivre and Bris, 2004 Li and Zhang,2004 global existence with small initial data: Guillope and Saut, 1990 Lin, Liu and Zhang, 2005 regularized model: regularity in 2D, Constantin,2012 global existence of regularized model: Barrett, 2014 global existence and uniqueness in the discrete scheme: Lee, Xu, and Zhang, 2011 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 3 / 20

  7. Numerical results High Weissenberg number problem. 201 Re = 1, We = 1 x 10 3.5 3 2.5 2 1.5 1 0.5 0 0 5 10 15 20 25 30 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 4 / 20

  8. Positivity-preserving method: B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 5 / 20

  9. Positivity-preserving method: Log-transformation: ψ = log σ Fattal and Kupfermann, 2005 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 5 / 20

  10. Positivity-preserving method: Log-transformation: ψ = log σ Fattal and Kupfermann, 2005 Square-root: 1 ψ = ( σ ) 2 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 5 / 20

  11. Positivity-preserving method: Log-transformation: ψ = log σ Fattal and Kupfermann, 2005 Square-root: 1 ψ = ( σ ) 2 Euler-Lagrangian method: δ σ δ t = ∂ σ ∂ t + ( u · ∇ ) σ − ∇ u · σ − σ · ( ∇ u ) T is discretized as σ k +1 − F ( σ k ◦ X ) F T ∆ t d F dt = ∇ uF Trebotich, 2005 Lee and Xu, 2006 Lee, Xu, and Zhang, 2011 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 5 / 20

  12. A cartoon model: ∂ t + a ( x ) ∂φ ∂φ ∂ x − b ( x ) φ = − 1 We φ φ ( x , 0) = 0. a ( x ) , b ( x ) > 0 play the role as u , ∇ u . � x 0 exp( b ( x ′ ) − We − 1 ) dx ′ is exponential. Steady state φ ( x ) = a ( x ′ ) B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 6 / 20

  13. A cartoon model: ∂ t + a ( x ) ∂φ ∂φ ∂ x − b ( x ) φ = − 1 We φ φ ( x , 0) = 0. a ( x ) , b ( x ) > 0 play the role as u , ∇ u . � x 0 exp( b ( x ′ ) − We − 1 ) dx ′ is exponential. Steady state φ ( x ) = a ( x ′ ) What would happen if we do logarithm transformation? ψ = log φ . B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 6 / 20

  14. A cartoon model: ∂φ ∂ t + a ( x ) ∂φ ∂ x − b ( x ) φ = − 1 We φ φ ( x , 0) = 0. a ( x ) , b ( x ) > 0 play the role as u , ∇ u . � x 0 exp( b ( x ′ ) − We − 1 ) dx ′ is exponential. Steady state φ ( x ) = a ( x ′ ) What would happen if we do logarithm transformation? ψ = log φ . ∂ x − b ( x ) = − 1 ∂ψ ∂ t + a ( x ) ∂ψ We B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 6 / 20

  15. A cartoon model: ∂φ ∂ t + a ( x ) ∂φ ∂ x − b ( x ) φ = − 1 We φ φ ( x , 0) = 0. a ( x ) , b ( x ) > 0 play the role as u , ∇ u . � x 0 exp( b ( x ′ ) − We − 1 ) dx ′ is exponential. Steady state φ ( x ) = a ( x ′ ) What would happen if we do logarithm transformation? ψ = log φ . ∂ x − b ( x ) = − 1 ∂ψ ∂ t + a ( x ) ∂ψ We Numerical methods based on polynomials is difficult to catch the exponential growth!! Re = 1, We = 2, t = 2 1 250 300 0.9 250 0.8 200 0.7 200 0.6 150 0.5 τ 11 150 0.4 100 100 0.3 0.2 50 50 0.1 0 0 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 y (x=0.5) B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 6 / 20

  16. some stability technique For any matrix ∇ u and symmetric positive definite matrix σ : ∇ u = B + Ω + N σ − 1 . N , Ω are anti-symmetric, B is symmetric and commutes with τ . B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 7 / 20

  17. some stability technique For any matrix ∇ u and symmetric positive definite matrix σ : ∇ u = B + Ω + N σ − 1 . N , Ω are anti-symmetric, B is symmetric and commutes with τ . ∂ t + ( u · ∇ ) σ − ∇ u · σ − σ · ( ∇ u ) T = Eq.(1) 3 ∂ σ 1 We ( I − σ ) can be written as 1 ∂ σ ∂ t + ( u · ∇ ) σ = Ω σ − σ Ω + 2 B σ + We ( I − σ ) B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 7 / 20

  18. some stability technique For any matrix ∇ u and symmetric positive definite matrix σ : ∇ u = B + Ω + N σ − 1 . N , Ω are anti-symmetric, B is symmetric and commutes with τ . ∂ t + ( u · ∇ ) σ − ∇ u · σ − σ · ( ∇ u ) T = Eq.(1) 3 ∂ σ 1 We ( I − σ ) can be written as 1 ∂ σ ∂ t + ( u · ∇ ) σ = Ω σ − σ Ω + 2 B σ + We ( I − σ ) � Log-transformation ψ = log( σ ): diag ( λ i ) = R T σ R , ψ = R diag ( log λ i ) R T . B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 7 / 20

  19. some stability technique For any matrix ∇ u and symmetric positive definite matrix σ : ∇ u = B + Ω + N σ − 1 . N , Ω are anti-symmetric, B is symmetric and commutes with τ . ∂ t + ( u · ∇ ) σ − ∇ u · σ − σ · ( ∇ u ) T = Eq.(1) 3 ∂ σ 1 We ( I − σ ) can be written as 1 ∂ σ ∂ t + ( u · ∇ ) σ = Ω σ − σ Ω + 2 B σ + We ( I − σ ) � Log-transformation ψ = log( σ ): diag ( λ i ) = R T σ R , ψ = R diag ( log λ i ) R T . ∂ ψ 1 We ( e − ψ − I ) ∂ t + u · ∇ ψ = Ω ψ − ψ Ω + 2 B + B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 7 / 20

  20. kinetic energy 0.018 We=0.5 We=1 0.016 We=3 0.014 0.012 0.01 0.008 0.006 0.004 0.002 0 0 5 10 15 20 25 30 time No blow up!! B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 8 / 20

  21. kinetic energy 0.018 We=0.5 We=1 0.016 We=3 0.014 0.012 0.01 0.008 0.006 0.004 0.002 0 0 5 10 15 20 25 30 time No blow up!! Table: L 2 error || σ h − σ h / 2 || h We = 0 . 5 We = 1 We = 3 1/32 0.3502 1.5846 4.8967 1/64 0.5006 3.3141 10.8242 1/128 0.7181 5.3517 18.5389 B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 8 / 20 Results do not converge.

  22. A viscous Oldroyd-B model in log-transformation: ∂ u β  We ∇ · e ψ ∂ t + u · ∇ u = −∇ p + α ∆ u +  ∇ · u = 0 We ( e − ψ − I ) + ε ∆ ψ ∂ ψ 1  ∂ t + u · ∇ ψ = Ω ψ − ψ Ω + 2 B + (2) B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 9 / 20

  23. A viscous Oldroyd-B model in log-transformation: ∂ u β  We ∇ · e ψ ∂ t + u · ∇ u = −∇ p + α ∆ u +  ∇ · u = 0 We ( e − ψ − I ) + ε ∆ ψ ∂ ψ 1  ∂ t + u · ∇ ψ = Ω ψ − ψ Ω + 2 B + (2) The viscous model satisfies the inequality: d � β � |∇ u | 2 + tr ( σ + σ − 1 − 2 I ) ≤ 0 dt F ( u , σ ) + α (3) 2 We 2 D D and the free-energy F ( u , σ ) decreases exponentially fast to zero in time. B. She, M. Luk´ aˇ cov´ a (JGU-Mainz) Numerical tests of a viscous Oldroyd-B model 2014-05-24 9 / 20

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend