Structure-Preserving B-spline Methods for the Incompressible Navier-Stokes Equations John Andrew Evans Institute for Computational Engineering and Sciences,

Slides:



Advertisements
Similar presentations
Stable Fluids A paper by Jos Stam.
Advertisements

Formulation of linear hydrodynamic stability problems
Christopher Batty and Robert Bridson University of British Columbia
Compatible Spatial Discretizations for Partial Differential Equations May 14, 2004 Compatible Reconstruction of Vectors Blair Perot Dept. of Mechanical.
Navier-Stokes.
Continuity Equation. Continuity Equation Continuity Equation Net outflow in x direction.
Point-wise Discretization Errors in Boundary Element Method for Elasticity Problem Bart F. Zalewski Case Western Reserve University Robert L. Mullen Case.
Discrete Exterior Calculus. More Complete Introduction See Chapter 7 “Discrete Differential Forms for Computational Modeling” in the SIGGRAPH 2006 Discrete.
Introduction to numerical simulation of fluid flows
A Bezier Based Approach to Unstructured Moving Meshes ALADDIN and Sangria Gary Miller David Cardoze Todd Phillips Noel Walkington Mark Olah Miklos Bergou.
1 MECH 221 FLUID MECHANICS (Fall 06/07) Tutorial 7.
1 Internal Seminar, November 14 th Effects of non conformal mesh on LES S. Rolfo The University of Manchester, M60 1QD, UK School of Mechanical,
High-Order Adaptive and Parallel Discontinuous Galerkin Methods for Hyperbolic Conservation Laws J. E. Flaherty, L. Krivodonova, J. F. Remacle, and M.
LES of Turbulent Flows: Lecture 3 (ME EN )
Chapter 4 Fluid Flow, Heat Transfer, and Mass Transfer:
LAMINAR PLANE COUETTE AND OPEN CHANNEL FLOW
Modeling, Simulating and Rendering Fluids Thanks to Ron Fediw et al, Jos Stam, Henrik Jensen, Ryan.
FUNDAMENTAL EQUATIONS, CONCEPTS AND IMPLEMENTATION
ME 231 Thermofluid Mechanics I Navier-Stokes Equations.
Conservation Laws for Continua
Review (2 nd order tensors): Tensor – Linear mapping of a vector onto another vector Tensor components in a Cartesian basis (3x3 matrix): Basis change.
Department of Aerospace and Mechanical Engineering A one-field discontinuous Galerkin formulation of non-linear Kirchhoff-Love shells Ludovic Noels Computational.
CFD Modeling of Turbulent Flows
Ch. 10 Vector Integral Calculus.
Hybrid WENO-FD and RKDG Method for Hyperbolic Conservation Laws
A Look at High-Order Finite- Volume Schemes for Simulating Atmospheric Flows Paul Ullrich University of Michigan.
Animation of Fluids.
Chapter 9: Differential Analysis of Fluid Flow SCHOOL OF BIOPROCESS ENGINEERING, UNIVERSITI MALAYSIA PERLIS.
PTT 204/3 APPLIED FLUID MECHANICS SEM 2 (2012/2013)
Page 1 JASS 2004 Tobias Weinzierl Sophisticated construction ideas of ansatz- spaces How to construct Ritz-Galerkin ansatz-spaces for the Navier-Stokes.
A conservative FE-discretisation of the Navier-Stokes equation JASS 2005, St. Petersburg Thomas Satzger.
A particle-gridless hybrid methods for incompressible flows
Governing equations: Navier-Stokes equations, Two-dimensional shallow-water equations, Saint-Venant equations, compressible water hammer flow equations.
Prof. dr. A. Achterberg, Astronomical Dept., IMAPP, Radboud Universiteit.
Discontinuous Galerkin Methods for Solving Euler Equations Andrey Andreyev Advisor: James Baeder Mid.
Final Project Solution to a Hyperbolic problem using Mimetic Methods Caplan,Ron Jimenez, Rosa Martinez, Joan M. M 542 Intro. to Numerical Solutions to.
Reynolds Transport Theorem We need to relate time derivative of a property of a system to rate of change of that property within a certain region (C.V.)
Stable, Circulation- Preserving, Simplicial Fluids Sharif Elcott, Yiying Tong, Eva Kanso, Peter Schröder, and Mathieu Desbrun.
Dr. Wang Xingbo Fall , 2005 Mathematical & Mechanical Method in Mechanical Engineering.
CIS888.11V/EE894R/ME894V A Case Study in Computational Science & Engineering We will apply several numerical methods to find a steady state solution of.
CHAP 3 WEIGHTED RESIDUAL AND ENERGY METHOD FOR 1D PROBLEMS
Discretization Methods Chapter 2. Training Manual May 15, 2001 Inventory # Discretization Methods Topics Equations and The Goal Brief overview.
A Non-iterative Hyperbolic, First-order Conservation Law Approach to Divergence-free Solutions to Maxwell’s Equations Richard J. Thompson 1 and Trevor.
A New Discontinuous Galerkin Formulation for Kirchhoff-Love Shells
Generalized Finite Element Methods
The Stability of Laminar Flows
1 CONSTITUTIVE RELATION FOR NEWTONIAN FLUID The Cauchy equation for momentum balance of a continuous, deformable medium combined with the condition of.
Advanced Computer Graphics Spring 2014 K. H. Ko School of Mechatronics Gwangju Institute of Science and Technology.
Reverse Engineering of Point Clouds to Obtain Trimmed NURBS Lavanya Sita Tekumalla Advisor: Prof. Elaine Cohen School of Computing University of Utah Masters.
SIAM Conference on Geometric Desing & Computing Approximation of spatial data with shape constraints Maria Lucia Sampoli University of Siena, Italy.
1 Spring 2003 Prof. Tim Warburton MA557/MA578/CS557 Lecture 28.
Multipole-Based Preconditioners for Sparse Linear Systems. Ananth Grama Purdue University. Supported by the National Science Foundation.
1 LES of Turbulent Flows: Lecture 7 (ME EN ) Prof. Rob Stoll Department of Mechanical Engineering University of Utah Spring 2011.
Stability Investigation of a Difference Scheme for Incompressible Navier—Stokes Equations D. Chibisov, V. Ganzha, E.W. Mayr, E.V. Vorozhtsov.
Model Anything. Quantity Conserved c  advect  diffuse S ConservationConstitutiveGoverning Mass, M  q -- M Momentum fluid, Mv -- F Momentum fluid.
Computer Animation Rick Parent Computer Animation Algorithms and Techniques Computational Fluid Dynamics.
1 CHAP 3 WEIGHTED RESIDUAL AND ENERGY METHOD FOR 1D PROBLEMS FINITE ELEMENT ANALYSIS AND DESIGN Nam-Ho Kim.
Chapter 4 Fluid Mechanics Frank White
Convergence in Computational Science
Objective Review Reynolds Navier Stokes Equations (RANS)
Topic 6 FluidFlowEquations_Introduction
Advanced Computer Graphics Spring 2008
Objective Numerical methods Finite volume.
topic8_NS_vectorForm_F02
Introduction to Fluid Dynamics & Applications
Topic 5 NavierStokes Equations
Navier-Stokes The questions addressed in this lecture
topic8_NS_vectorForm_F02
Comparison of CFEM and DG methods
Presentation transcript:

Structure-Preserving B-spline Methods for the Incompressible Navier-Stokes Equations John Andrew Evans Institute for Computational Engineering and Sciences, UT Austin Stabilized and Multiscale Methods in CFD Spring 2013

Motivation So why do we need another flow solver? Incompressibility endows the Navier-Stokes equations with important physical structure: Mass balance Momentum balance Energy balance Enstrophy balance Helicity balance However, most methods only satisfy the incompressibility constraint in an approximate sense. Such methods do not preserve structure and lack robustness.

Motivation Consider a two-dimensional Taylor-Green vortex. The vortex is a smooth steady state solution of the Euler equations.

Motivation Conservative methods which only weakly satisfy incompressibility do not preserve this steady state and blow up in the absence of artificial numerical dissipation. 2-D Conservative Taylor-Hood Element Q2/Q1 and h = 1/8

Motivation Methods which exactly satisfy incompressibility are stable in the Euler limit and robust with respect to Reynold’s number. Increasing Reynold’s number 2-D Conservative Structure-preserving B-splines k’ = 1 and h = 1/8

Motivation Due to the preceding discussion, we seek new discretizations that: Satisfy the divergence-free constraint exactly. Harbor local stability and approximation properties. Possess spectral-like stability and approximation properties. Extend to geometrically complex domains. Structure-Preserving B-splines seem to fit the bill.

The Stokes Complex The classical L 2 de Rham complex is as follows: From the above complex, we can derive the following smoothed complex with the same cohomology structure:

The Stokes Complex Scalar Potentials Vector Potentials Flow Pressures Flow Velocities The smoothed complex corresponds to viscous flow, so we henceforth refer to it as the Stokes complex.

The Stokes Complex For simply-connected domains with connected boundary, the Stokes complex is exact. grad operator maps onto space of curl-free functions curl operator maps onto space of div-free functions div operator maps onto entire space of flow pressures

The Stokes Complex Gradient Theorem: Curl Theorem: Divergence Theorem:

The Discrete Stokes Subcomplex Now, suppose we have a discrete Stokes subcomplex. Then a Galerkin discretization utilizing the subcomplex: Does not have spurious pressure modes, and Returns a divergence-free velocity field. Discrete Scalar Potentials Discrete Vector Potentials Discrete Flow Pressures Discrete Flow Velocities

The Discrete Stokes Subcomplex

Structure-Preserving B-splines Review of Univariate (1-D) B-splines: Knot vector on (0,1) and k-degree B-spline basis on (0,1) by recursion: Ξ = {0, 0, 0, 0.2, 0.4, 0.6, 0.8, 0.8, 1, 1, 1}, k = 2 Knots w/ multiplicity Start w/ piecewise constants Bootstrap recursively to k

k=2 Open knot vectors: Multiplicity of first and last knots is k+1 Basis is interpolatory at these locations Non-uniform knot spacing allowed Continuity at interior knot a function of knot repetition Review of Univariate (1-D) B-splines: Structure-Preserving B-splines

Review of Univariate (1-D) B-splines: Derivatives of B-splines are B-splines k=4k=3 onto Structure-Preserving B-splines

Review of Univariate (1-D) B-splines: We form curves in physical space by taking weighted sums. - control points- knots Quadratic basis: “control mesh” Structure-Preserving B-splines

Multivariate B-splines are built through tensor-products Multivariate B-splines inherit all of the aforementioned properties of univariate B-splines In what follows, we denote the space of n-dimensional tensor- product B-splines as polynomial degree in direction i i th continuity vector Review of Multivariate B-splines: Structure-Preserving B-splines

We form surfaces and volumes using weighted sums of multivariate B-splines (or rational B-splines) as before. Review of Multivariate B-splines: Control mesh Mesh Structure-Preserving B-splines

Define for the unit square: In the context of fluid flow: and it is easily shown that: Two-dimensional Structure-Preserving B-splines Structure-Preserving B-splines

Two-dimensional Structure-Preserving B-splines: Mapped Domains On mapped domains, the Piola transform is utilized to map flow velocities. Pressures are mapped using an integral preserving transform. Structure-Preserving B-splines

We associate the degrees of freedom of structure-preserving B-splines with the control mesh. Notably, we associate: Structure-Preserving B-splines Two-dimensional Structure-Preserving B-splines

Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2: Structure-Preserving B-splines

Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2:

Structure-Preserving B-splines Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2:

Structure-Preserving B-splines Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2:

Structure-Preserving B-splines Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2:

Structure-Preserving B-splines Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2:

Structure-Preserving B-splines Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2:

Structure-Preserving B-splines Two-dimensional Structure-Preserving B-splines, k 1 = k 2 = 2:

Define for the unit cube: It is easily shown that: Structure-Preserving B-splines Three-dimensional Structure-Preserving B-splines

Flow velocities: map w/ divergence-preserving transformation Flow pressures: map w/ integral-preserving transformation Vector potentials: map w/ curl-conserving transformation Structure-Preserving B-splines Three-dimensional Structure-Preserving B-splines

We associate as before the degrees of freedom with the control mesh. Structure-Preserving B-splines Three-dimensional Structure-Preserving B-splines

Control points: Scalar potential DOF Control edges: Vector potential DOF Structure-Preserving B-splines Three-dimensional Structure-Preserving B-splines

Structure-Preserving B-splines Weak Enforcement of No-Slip BCs Nitsche’s method is utilized to weakly enforce the no-slip condition in our discretizations. Our motivation is three- fold: Nitsche’s method is consistent and higher-order. Nitsche’s method preserves symmetry and ellipticity. Nitsche’s method is a consistent stabilization procedure. Furthermore, with weak no-slip boundary conditions, a conforming discretization of the Euler equations is obtained in the limit of vanishing viscosity.

Structure-Preserving B-splines Weak Enforcement of Tangential Continuity Between Patches On multi-patch geometries, tangential continuity is enforced weakly between patches using a combination of the symmetric interior penalty method and upwinding.

Summary of Theoretical Results Well-posedness for small data Optimal velocity error estimates and suboptimal, by one order, pressure error estimates Conforming discretization of Euler flow obtained in limit of vanishing viscosity (via weak BCs) Robustness with respect to viscosity for small data Steady Navier-Stokes Flow

Summary of Theoretical Results Unsteady Navier-Stokes Flow Existence and uniqueness (well-posedness) Optimal velocity error estimates in terms of the L 2 norm for domains satisfying an elliptic regularity condition (local-in-time) Convergence to suitable weak solutions for periodic domains Balance laws for momentum, energy, enstrophy, and helicity Balance law for angular momentum on cylindrical domains

Spectrum Analysis Consider the two-dimensional periodic Stokes eigenproblem: We compare the discrete spectrum for a specified discretization with the exact spectrum. This analysis sheds light on a given discretization’s resolution properties.

Spectrum Analysis: Structure-Preserving B-splines Spectrum Analysis

Spectrum Analysis: Taylor-Hood Elements Spectrum Analysis

Spectrum Analysis: MAC Scheme Spectrum Analysis

Selected Numerical Results Steady Navier-Stokes Flow: Numerical Confirmation of Convergence Rates 2-D Manufactured Vortex Solution

Selected Numerical Results Steady Navier-Stokes Flow: Numerical Confirmation of Convergence Rates 2-D Manufactured Vortex Solution

Selected Numerical Results Steady Navier-Stokes Flow: Numerical Confirmation of Convergence Rates 2-D Manufactured Vortex Solution

Selected Numerical Results Steady Navier-Stokes Flow: Numerical Confirmation of Convergence Rates Re Energy 1.40e-2 H 1 error - u 1.40e-2 L 2 error - u 2.28e-4 L 2 error - p 3.49e e-41.96e-4 Robustness with respect to Reynolds number k ’ = 1 and h = 1/16 2-D Manufactured Vortex Solution

Selected Numerical Results Steady Navier-Stokes Flow: Numerical Confirmation of Convergence Rates Re H 1 error - u 6.77e e-42.26e-32.16e-2X L 2 error - u 6.54e-46.54e-66.79e-61.97e-51.86e-4X L 2 error - p 1.96e-4 X Instability of 2-D Taylor-Hood with respect to Reynolds number Q2/Q1 and h = 1/16 2-D Manufactured Vortex Solution

Steady Navier-Stokes Flow: Lid-Driven Cavity Flow H H U Selected Numerical Results

Steady Navier-Stokes Flow: Lid-Driven Cavity Flow at Re = 1000 Selected Numerical Results

Steady Navier-Stokes Flow: Lid-Driven Cavity Flow at Re = 1000 Methodu min v min v max k’ = 1, h = 1/ k’ = 1, h = 1/ k’ = 1, h = 1/ k’ = 2, h = 1/ k’ = 3, h = 1/ Converged Ghia, h = 1/ Selected Numerical Results

Unsteady Navier-Stokes Flow: Flow Over a Cylinder at Re = 100 L H U D L out Selected Numerical Results

Unsteady Navier-Stokes Flow: Flow Over a Cylinder at Re = 100 Patch 1 Patch 2 Patch 3 Patch 4Patch 5 Selected Numerical Results

Unsteady Navier-Stokes Flow: Flow Over a Cylinder at Re = 100 Simulation Details: - Full Space-Time Discretization - Method of Subgrid Vortices - Linears in Space and Time - D = 2, H = 32D, L = 64D - Time Step Size: Trilinos Implementation - GMRES w/ ILU Preconditioning time-steps (shedding initiated after 1000 time-steps) - Approximately 15,000 DOF/time step Selected Numerical Results

Unsteady Navier-Stokes Flow: Flow Over a Cylinder at Re = 100 Selected Numerical Results

Unsteady Navier-Stokes Flow: Flow Over a Cylinder at Re = 100 Quantity of Interest: Strouhal Number: St = fD/U Computed Strouhal Number: Accepted Strouhal Number: Selected Numerical Results

Unsteady Navier-Stokes Flow: Three-Dimensional Taylor-Green Vortex Flow Selected Numerical Results 3-D Periodic Flow Simplest Model of Vortex Stretching No External Forcing

Unsteady Navier-Stokes Flow: Three-Dimensional Taylor-Green Vortex Flow Selected Numerical Results Time Evolution of Dissipation Rate Reproduced with permission from [Brachet et al. 1983]

Unsteady Navier-Stokes Flow: Three-Dimensional Taylor-Green Vortex Flow at Re = 200 Selected Numerical Results Enstrophy isosurface at time corresponding to maximum dissipation

Unsteady Navier-Stokes Flow: Three-Dimensional Taylor-Green Vortex Flow at Re = 200 Selected Numerical Results Convergence of dissipation rate time history with mesh refinement (k ’ = 1)

Unsteady Navier-Stokes Flow: Three-Dimensional Taylor-Green Vortex Flow at Re = 200 Selected Numerical Results Convergence of dissipation rate time history with degree elevation (h = 1/32)