sliding interfaces for eddy current simulations
play

Sliding Interfaces for Eddy Current Simulations Raffael Casagrande - PowerPoint PPT Presentation

Sliding Interfaces for Eddy Current Simulations Raffael Casagrande Supervisor: Prof. Dr. Ralf Hiptmair Seminar of Applied Mathematics ETH Zrich April 17th, 2013 Raffael Casagrande (ETH Zrich) Sliding Interfaces for Eddy Current April


  1. Sliding Interfaces for Eddy Current Simulations Raffael Casagrande Supervisor: Prof. Dr. Ralf Hiptmair Seminar of Applied Mathematics ETH Zürich April 17th, 2013 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 1 / 25

  2. Outline Introduction 1 Motivation Deriving the eddy current model 2 Maxwell’s Equations in a moving frame The eddy current model in a moving frame Discontinuous Galerkin Formulation 3 DG Theory Aspects of the implementation Results and Conclusion 4 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 2 / 25

  3. Motivation Generator circuit breakers ◮ translational motion Electric engines ◮ rotation Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 3 / 25

  4. Motivation Generator circuit breakers ◮ translational motion Electric engines ◮ rotation Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 3 / 25

  5. Outline Introduction 1 Motivation Deriving the eddy current model 2 Maxwell’s Equations in a moving frame The eddy current model in a moving frame Discontinuous Galerkin Formulation 3 DG Theory Aspects of the implementation Results and Conclusion 4 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 4 / 25

  6. Outline Introduction 1 Motivation Deriving the eddy current model 2 Maxwell’s Equations in a moving frame The eddy current model in a moving frame Discontinuous Galerkin Formulation 3 DG Theory Aspects of the implementation Results and Conclusion 4 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 5 / 25

  7. Maxwell’s Equations curl E + ∂ B div B = 0 ∂ t = 0 div E = ρ ∂ E curl B − 1 ∂ t = µ 0 ( j f + j i ) . c 2 ε 0  Quasistatic model for ⇓ j f = σ E , c → ∞  slowly varying Electric fields (High conductivities)  Eddy Current Model curl E + ∂ B div B = 0 ∂ t = 0 div E = ρ � σ E + j i � curl B = µ 0 ε 0 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 6 / 25

  8. Maxwell’s Equations curl E + ∂ B div B = 0 ∂ t = 0 div E = ρ ∂ E curl B − 1 ∂ t = µ 0 ( j f + j i ) . c 2 ε 0  Quasistatic model for ⇓ j f = σ E , c → ∞  slowly varying Electric fields (High conductivities)  Eddy Current Model curl E + ∂ B div B = 0 ∂ t = 0 div E = ρ � σ E + j i � curl B = µ 0 ε 0 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 6 / 25

  9. Maxwells equations are invariant under Lorentz transformation if E and B transform as E = γ ( E + V × B ) − ( γ − 1 )( E · ˆ ˜ V )ˆ V � B − V × E � ˜ − ( γ − 1 )( B · ˆ V )ˆ B = γ V c 2 1 ˆ V = ˆ γ := V / | ˆ V | � 1 − v 2 / c 2 ⇓ c → ∞ ˜ E = E + V × B ˜ B = B It can be shown that the eddy current model is also invariant under Rotation !!! Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 7 / 25

  10. Maxwells equations are invariant under Lorentz transformation if E and B transform as E = γ ( E + V × B ) − ( γ − 1 )( E · ˆ ˜ V )ˆ V � B − V × E � ˜ − ( γ − 1 )( B · ˆ V )ˆ B = γ V c 2 1 ˆ V = ˆ γ := V / | ˆ V | � 1 − v 2 / c 2 ⇓ c → ∞ ˜ E = E + V × B ˜ B = B It can be shown that the eddy current model is also invariant under Rotation !!! Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 7 / 25

  11. Two eddy current formulations Temporal gauged Potential formulation : µ curl A + σ∂ A curl 1 ∂ t = j i A ( t = 0 ) = 0 curl A × n = 0 on ∂ Ω H -formulation : σ curl H + µ∂ H curl 1 ∂ t = curl 1 σ j i H ( t = 0 ) = 0 H = 0 on ∂ Ω Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 8 / 25

  12. Two eddy current formulations Temporal gauged Potential formulation (Rest frame): µ curl A + σ∂ A curl 1 ∂ t = j i + σ V × curl A A ( t = 0 ) = 0 curl A × n = 0 on ∂ Ω H -formulation (Rest frame): σ curl H + µ∂ H curl 1 ∂ t = curl 1 σ j i + curl ( µ V × H ) H ( t = 0 ) = 0 H = 0 on ∂ Ω Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 8 / 25

  13. Two eddy current formulations Temporal gauged Potential formulation (Moving frame) : A + σ∂ ˜ curl 1 A ˜ curl ˜ ˜ ∂ t = ˜ j i µ ˜ A ( t = 0 ) = 0 curl ˜ ˜ A × n = 0 on ∂ Ω H -formulation (Moving frame): H + µ∂ ˜ H curl 1 curl 1 ˜ curl ˜ ˜ ˜ ˜ j i ∂ t = σ σ ˜ H ( t = 0 ) = 0 ˜ H = 0 on ∂ Ω Note: If j i is smooth enough, 1 µ curl A = H ⇒ Do the same simulation and compare the two models (Primal & Dual formulation). Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 8 / 25

  14. Transformation laws The coordinates of the moving frame ( ˜ x ) are related to the rest frame ( x ) by x = T ( t )˜ x + r ( t ) . T : Rotation matrix. Transformation laws T ˜ T ˜ E = E + V × B B = B T ˜ j i = j i T ˜ H = H T ˜ j f = j f T ˜ V = − V � t T T grad ( V · A ) T ˜ A = A − T 0 ⇒ Use transformation laws to derive transmission conditions at sliding interface. Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 9 / 25

  15. Transformation laws The coordinates of the moving frame ( ˜ x ) are related to the rest frame ( x ) by x = T ( t )˜ x + r ( t ) . T : Rotation matrix. Transformation laws T ˜ T ˜ E = E + V × B B = B T ˜ j i = j i T ˜ H = H T ˜ j f = j f T ˜ V = − V � t T T grad ( V · A ) T ˜ A = A − T 0 ⇒ Use transformation laws to derive transmission conditions at sliding interface. Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 9 / 25

  16. Transformation laws The coordinates of the moving frame ( ˜ x ) are related to the rest frame ( x ) by x = T ( t )˜ x + r ( t ) . T : Rotation matrix. Transformation laws T ˜ T ˜ E = E + V × B B = B T ˜ j i = j i T ˜ H = H T ˜ j f = j f T ˜ V = − V � t T T grad ( V · A ) T ˜ A = A − T 0 ⇒ Use transformation laws to derive transmission conditions at sliding interface. Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 9 / 25

  17. Outline Introduction 1 Motivation Deriving the eddy current model 2 Maxwell’s Equations in a moving frame The eddy current model in a moving frame Discontinuous Galerkin Formulation 3 DG Theory Aspects of the implementation Results and Conclusion 4 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 10 / 25

  18. DG Formulation of the Eddy Current Model σ∂ A ∂ t + curl 1 µ curl A = j i curl A × n = 0 on ∂ Ω DG Variational formulation Find A ( i ) h ∈ V h , i = 1 , . . . , N such that for all A ′ h ∈ V h , we have � σ A ( i + 1 ) − A ( i ) � � � ( A ( i + 1 ) h h , A ′ + a SWIP , A ′ j i , ( i + 1 ) , A ′ h ) = h h h h δ t � 3 , P k � ∀ T ∈ T h , v | t ∈ P k � � P k � v ∈ L 2 (Ω) � Where V h := 3 ( T h ) d ( T h ) := d ( T ) . Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 11 / 25

  19. Symmetric-Weighted-Interior-Penalty Bilinear form a SWIP h � 1 a SWIP ( A h , A ′ µ curl h A h · curl h A ′ h ) = h h Ω � 1 � � � A ′ � � − · µ curl h A h h T F ω F ∈F i h � 1 � � � µ curl h A ′ − · [ A h ] T h F ω F ∈F i h ηγ µ, F � � � A ′ � + [ A h ] T · h T h F F F ∈F i h { A h } ω = ω 1 A h , 1 + ω 2 A h , 2 , [ A h ] T = n F × ( A h , 1 − A h , 2 ) (1) µ 1 µ 2 2 ω 1 = , ω 2 = , γ µ, F = (2) µ 1 + µ 2 µ 1 + µ 2 µ 1 + µ 2 Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 12 / 25

  20. Convergence Under regularity conditions on the mesh sequence (matching) and assuming the exact solution A is smooth enough we can prove � 1 / 2 � N √ σ ( A ( N ) − A ( N ) 2 � � � � A ( i ) − A ( i ) � � ) L 2 (Ω) + C stab δ t ≤ � � � � h h � � � SWIP i = 1 C 1 h k + C 2 δ t 1 / 2 � � Ct F � � � ∂ 2 A ( t ) where C 1 = max t ∈ [ 0 , t F ] | A ( t ) | H k + 1 (Ω) and C 2 = max t ∈ [ 0 , t F ] L 2 (Ω) The � � ∂ t 2 � constants C 1 , C 2 and C are independent of h and δ t . 1 / 2   2 � � γ mu , F 1 � � [ A ] T � 2 � � | A | SWIP := + √ µ curl h A � �  L 2 ( F )  h F � � L 2 (Ω) F ∈F h Raffael Casagrande (ETH Zürich) Sliding Interfaces for Eddy Current April 17th, 2013 13 / 25

Recommend


More recommend