introduction to numerical methods to hyperbolic pde’sx =0 conservation of mass conservation of...
TRANSCRIPT
![Page 1: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/1.jpg)
Introduction to numerical methods to hyperbolic PDE’s
Lecture 1: Basic concepts
Literature:
Randy LeVeque: Computational Methods for Astrophysical Fluid Flow, Sass Fee lecture notes, Springer Verlag, 1998
download this from: http://www.mat.univie.ac.at/~obertsch/literatur/conservation_laws.pdf
![Page 2: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/2.jpg)
Outline of the first lecture
![Page 3: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/3.jpg)
Euler equations of compressible gas dynamics:
(ρu)t + (ρu2 + p)x = 0
Et +!
u(E + p)"
x= 0
E =p
γ − 1+
1
2ρu2
ρt + (ρu)x = 0 conservation of mass
conservation of momentum
conservation of total energy
closure relationship - equation of state: polytropic gas
Here is a typical example of a system of PDEs we want to numerically discretise:
![Page 4: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/4.jpg)
a planar shock wave in a gas hitting three cylinders
t = 0
![Page 5: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/5.jpg)
Given smooth initial data for such equations, the solution will evolve into something not smooth.
ut +
!
u2
2
"
x
= 0u = u(x, t)consider ∈ R Burgers equation:
u
x
In order to study the numerical discretization of such equations we first study simpler equations.
![Page 6: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/6.jpg)
Linear advection equation
![Page 7: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/7.jpg)
Numerical methods use space- and time discretization:
t
x
the state of the gas is given initially
t = 0we then need to
determine it at later times.
∆x = h
∆t = k
xi
tn
xi+1xi−1
![Page 8: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/8.jpg)
Finite difference method
![Page 9: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/9.jpg)
Finite volume method
![Page 10: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/10.jpg)
Convergence
![Page 11: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/11.jpg)
Local truncation error τ
![Page 12: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/12.jpg)
Consistency
If
But:
![Page 13: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/13.jpg)
Consistency + Stability = Convergence
Fundamental therom:
![Page 14: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/14.jpg)
Lax-Richtmyer stabilty
![Page 15: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/15.jpg)
Stability of upwind
![Page 16: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/16.jpg)
Upwind as interpolation
![Page 17: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/17.jpg)
The CFL Condition
not sufficient
![Page 18: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/18.jpg)
Numerical domain of dependence
![Page 19: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/19.jpg)
The CFL Condition
![Page 20: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/20.jpg)
CFL ConditionStencil
Lax-Wendroff
![Page 21: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/21.jpg)
Hyperbolic systems
![Page 22: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/22.jpg)
3 equations with λ1 < 0 < λ2 < λ3
domain of dependence range of infuence
![Page 23: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/23.jpg)
CFL ConditionStencil
![Page 24: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/24.jpg)
Upwind for a linear system
![Page 25: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/25.jpg)
Symmetric methods
unstable
![Page 26: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/26.jpg)
Numerical dissipation
![Page 27: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/27.jpg)
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5Upwind at t=1
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5Upwind at t = 5
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5LaxïWendroff at t = 1
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5LaxïWendroff at t = 5
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5BeamïWarming at t = 1
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5BeamïWarming at t = 5
Lax-Friedrichs Lax-Friedrichs
qt + qx = 0 with periodic boundary conditions
initial conditions
x
x
![Page 28: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/28.jpg)
Lax-Wendroff
![Page 29: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/29.jpg)
Modified equations
![Page 30: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/30.jpg)
Modified Equation for Lax-Wendroff
![Page 31: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/31.jpg)
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5Upwind at t=1
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5Upwind at t = 5
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5LaxïWendroff at t = 1
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5LaxïWendroff at t = 5
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5BeamïWarming at t = 1
0 0.2 0.4 0.6 0.8 1ï0.5
0
0.5
1
1.5BeamïWarming at t = 5
![Page 32: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/32.jpg)
Dispersion relation
![Page 33: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/33.jpg)
Dispersive behavior
![Page 34: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/34.jpg)
0 0.2 0.4 0.6 0.8 1ï1
ï0.5
0
0.5
1
1.5Upwind at t = 1
0 0.2 0.4 0.6 0.8 1ï1
ï0.5
0
0.5
1
1.5Upwind at t = 10
0 0.2 0.4 0.6 0.8 1ï1
ï0.5
0
0.5
1
1.5LaxïWendroff at t = 1
0 0.2 0.4 0.6 0.8 1ï1
ï0.5
0
0.5
1
1.5LaxïWendroff at t = 10
0 0.2 0.4 0.6 0.8 1ï1
ï0.5
0
0.5
1
1.5Superbee at t = 1
0 0.2 0.4 0.6 0.8 1ï1
ï0.5
0
0.5
1
1.5Superbee at t = 10
![Page 35: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/35.jpg)
![Page 36: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/36.jpg)
![Page 37: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/37.jpg)
Introduction to numerical methods to hyperbolic PDE’s
Lecture 2: High resolution finite volume methods
download review from: http://www.mat.univie.ac.at/~obertsch/literatur/conservation_laws.pdf
![Page 38: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/38.jpg)
Outline
![Page 39: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/39.jpg)
Qn
i =1
h
! xi+ 1
2
xi−
12
q(x, tn)dx
qt + f(q)x = 0Finite volume method
![Page 40: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/40.jpg)
Finite volume method
![Page 41: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/41.jpg)
Godunov’s method for advection
![Page 42: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/42.jpg)
Godunov’s method
![Page 43: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/43.jpg)
First order REA Algorithm
![Page 44: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/44.jpg)
First order REA Algorithm
![Page 45: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/45.jpg)
Cell update
![Page 46: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/46.jpg)
Second-order REA Algorithm
linear
![Page 47: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/47.jpg)
Second-order REA Algorithm
![Page 48: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/48.jpg)
Choice of slopes
![Page 49: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/49.jpg)
Oscillations
![Page 50: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/50.jpg)
High-resolution methods
![Page 51: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/51.jpg)
Minmod slope
![Page 52: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/52.jpg)
Piecewise linear reconstruction
![Page 53: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/53.jpg)
TVD Methods
![Page 54: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/54.jpg)
TVD REA Algorithm
TV (qn) ≤ TV (Qn).with the property that
linear
Note:
![Page 55: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/55.jpg)
Some popular limitersθj =
Uj − Uj−1
Uj+1 − Uj
![Page 56: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/56.jpg)
6.9 Slope-Limiter Methods 111
6.9 Slope-Limiter Methods
We now return to the derivation of numerical methods based on piecewise linear reconstruc-tion, and consider how to limit the slopes so that (6.23) is satisfied. Note that setting σ n
i ≡ 0works, since the piecewise constant function has the same TV as the discrete data. Hencethe first-order upwind method is TVD for the advection equation. The upwind method maysmear solutions but cannot introduce oscillations.
One choice of slope that gives second-order accuracy for smooth solutions while stillsatisfying the TVD property is the minmod slope
σ ni = minmod
(Qn
i − Qni−1
"x,
Qni+1 − Qn
i
"x
), (6.26)
where the minmod function of two arguments is defined by
minmod(a, b) =
⎧⎪⎨
⎪⎩
a if |a| < |b| and ab > 0,
b if |b| < |a| and ab > 0,
0 if ab ≤ 0.
(6.27)
If a and b have the same sign, then this selects the one that is smaller in modulus, else itreturns zero.
Rather than defining the slope on the i th cell by always using the downwind difference(which would give the Lax–Wendroff method), or by always using the upwind difference(which would give the Beam–Warming method), the minmod method compares the twoslopes and chooses the one that is smaller in magnitude. If the two slopes have differentsign, then the value Qn
i must be a local maximum or minimum, and it is easy to check inthis case that we must set σ n
i = 0 in order to satisfy (6.23).Figure 6.2(a) shows results using the minmod method for the advection problem con-
sidered previously. We see that the minmod method does a fairly good job of maintaininggood accuracy in the smooth hump and also sharp discontinuities in the square wave, withno oscillations.
Sharper resolution of discontinuities can be achieved with other limiters that do notreduce the slope as severely as minmod near a discontinuity. Figure 6.5(a) shows somesample data representing a discontinuity smeared over two cells, along with the minmodslopes. Figure 6.5(b) shows that we can increase the slopes in these two cells to twice thevalue of the minmod slopes and still have (6.23) satisfied. This sharper reconstruction willlead to sharper resolution of the discontinuity in the next time step than we would obtainwith the minmod slopes.
(a) (b)
Fig. 6.5. Grid values Qn and reconstructed qn(·, tn) using (a) minmod slopes, (b) superbee or MCslopes. Note that these steeper slopes can be used and still have the TVD property.
6.9 Slope-Limiter Methods 111
6.9 Slope-Limiter Methods
We now return to the derivation of numerical methods based on piecewise linear reconstruc-tion, and consider how to limit the slopes so that (6.23) is satisfied. Note that setting σ n
i ≡ 0works, since the piecewise constant function has the same TV as the discrete data. Hencethe first-order upwind method is TVD for the advection equation. The upwind method maysmear solutions but cannot introduce oscillations.
One choice of slope that gives second-order accuracy for smooth solutions while stillsatisfying the TVD property is the minmod slope
σ ni = minmod
(Qn
i − Qni−1
"x,
Qni+1 − Qn
i
"x
), (6.26)
where the minmod function of two arguments is defined by
minmod(a, b) =
⎧⎪⎨
⎪⎩
a if |a| < |b| and ab > 0,
b if |b| < |a| and ab > 0,
0 if ab ≤ 0.
(6.27)
If a and b have the same sign, then this selects the one that is smaller in modulus, else itreturns zero.
Rather than defining the slope on the i th cell by always using the downwind difference(which would give the Lax–Wendroff method), or by always using the upwind difference(which would give the Beam–Warming method), the minmod method compares the twoslopes and chooses the one that is smaller in magnitude. If the two slopes have differentsign, then the value Qn
i must be a local maximum or minimum, and it is easy to check inthis case that we must set σ n
i = 0 in order to satisfy (6.23).Figure 6.2(a) shows results using the minmod method for the advection problem con-
sidered previously. We see that the minmod method does a fairly good job of maintaininggood accuracy in the smooth hump and also sharp discontinuities in the square wave, withno oscillations.
Sharper resolution of discontinuities can be achieved with other limiters that do notreduce the slope as severely as minmod near a discontinuity. Figure 6.5(a) shows somesample data representing a discontinuity smeared over two cells, along with the minmodslopes. Figure 6.5(b) shows that we can increase the slopes in these two cells to twice thevalue of the minmod slopes and still have (6.23) satisfied. This sharper reconstruction willlead to sharper resolution of the discontinuity in the next time step than we would obtainwith the minmod slopes.
(a) (b)
Fig. 6.5. Grid values Qn and reconstructed qn(·, tn) using (a) minmod slopes, (b) superbee or MCslopes. Note that these steeper slopes can be used and still have the TVD property.
Piecewise linear reconstruction
6.9 Slope-Limiter Methods 111
6.9 Slope-Limiter Methods
We now return to the derivation of numerical methods based on piecewise linear reconstruc-tion, and consider how to limit the slopes so that (6.23) is satisfied. Note that setting σ n
i ≡ 0works, since the piecewise constant function has the same TV as the discrete data. Hencethe first-order upwind method is TVD for the advection equation. The upwind method maysmear solutions but cannot introduce oscillations.
One choice of slope that gives second-order accuracy for smooth solutions while stillsatisfying the TVD property is the minmod slope
σ ni = minmod
(Qn
i − Qni−1
"x,
Qni+1 − Qn
i
"x
), (6.26)
where the minmod function of two arguments is defined by
minmod(a, b) =
⎧⎪⎨
⎪⎩
a if |a| < |b| and ab > 0,
b if |b| < |a| and ab > 0,
0 if ab ≤ 0.
(6.27)
If a and b have the same sign, then this selects the one that is smaller in modulus, else itreturns zero.
Rather than defining the slope on the i th cell by always using the downwind difference(which would give the Lax–Wendroff method), or by always using the upwind difference(which would give the Beam–Warming method), the minmod method compares the twoslopes and chooses the one that is smaller in magnitude. If the two slopes have differentsign, then the value Qn
i must be a local maximum or minimum, and it is easy to check inthis case that we must set σ n
i = 0 in order to satisfy (6.23).Figure 6.2(a) shows results using the minmod method for the advection problem con-
sidered previously. We see that the minmod method does a fairly good job of maintaininggood accuracy in the smooth hump and also sharp discontinuities in the square wave, withno oscillations.
Sharper resolution of discontinuities can be achieved with other limiters that do notreduce the slope as severely as minmod near a discontinuity. Figure 6.5(a) shows somesample data representing a discontinuity smeared over two cells, along with the minmodslopes. Figure 6.5(b) shows that we can increase the slopes in these two cells to twice thevalue of the minmod slopes and still have (6.23) satisfied. This sharper reconstruction willlead to sharper resolution of the discontinuity in the next time step than we would obtainwith the minmod slopes.
(a) (b)
Fig. 6.5. Grid values Qn and reconstructed qn(·, tn) using (a) minmod slopes, (b) superbee or MCslopes. Note that these steeper slopes can be used and still have the TVD property.
6.9 Slope-Limiter Methods 111
6.9 Slope-Limiter Methods
We now return to the derivation of numerical methods based on piecewise linear reconstruc-tion, and consider how to limit the slopes so that (6.23) is satisfied. Note that setting σ n
i ≡ 0works, since the piecewise constant function has the same TV as the discrete data. Hencethe first-order upwind method is TVD for the advection equation. The upwind method maysmear solutions but cannot introduce oscillations.
One choice of slope that gives second-order accuracy for smooth solutions while stillsatisfying the TVD property is the minmod slope
σ ni = minmod
(Qn
i − Qni−1
"x,
Qni+1 − Qn
i
"x
), (6.26)
where the minmod function of two arguments is defined by
minmod(a, b) =
⎧⎪⎨
⎪⎩
a if |a| < |b| and ab > 0,
b if |b| < |a| and ab > 0,
0 if ab ≤ 0.
(6.27)
If a and b have the same sign, then this selects the one that is smaller in modulus, else itreturns zero.
Rather than defining the slope on the i th cell by always using the downwind difference(which would give the Lax–Wendroff method), or by always using the upwind difference(which would give the Beam–Warming method), the minmod method compares the twoslopes and chooses the one that is smaller in magnitude. If the two slopes have differentsign, then the value Qn
i must be a local maximum or minimum, and it is easy to check inthis case that we must set σ n
i = 0 in order to satisfy (6.23).Figure 6.2(a) shows results using the minmod method for the advection problem con-
sidered previously. We see that the minmod method does a fairly good job of maintaininggood accuracy in the smooth hump and also sharp discontinuities in the square wave, withno oscillations.
Sharper resolution of discontinuities can be achieved with other limiters that do notreduce the slope as severely as minmod near a discontinuity. Figure 6.5(a) shows somesample data representing a discontinuity smeared over two cells, along with the minmodslopes. Figure 6.5(b) shows that we can increase the slopes in these two cells to twice thevalue of the minmod slopes and still have (6.23) satisfied. This sharper reconstruction willlead to sharper resolution of the discontinuity in the next time step than we would obtainwith the minmod slopes.
(a) (b)
Fig. 6.5. Grid values Qn and reconstructed qn(·, tn) using (a) minmod slopes, (b) superbee or MCslopes. Note that these steeper slopes can be used and still have the TVD property.
6.9 Slope-Limiter Methods 111
6.9 Slope-Limiter Methods
We now return to the derivation of numerical methods based on piecewise linear reconstruc-tion, and consider how to limit the slopes so that (6.23) is satisfied. Note that setting σ n
i ≡ 0works, since the piecewise constant function has the same TV as the discrete data. Hencethe first-order upwind method is TVD for the advection equation. The upwind method maysmear solutions but cannot introduce oscillations.
One choice of slope that gives second-order accuracy for smooth solutions while stillsatisfying the TVD property is the minmod slope
σ ni = minmod
(Qn
i − Qni−1
"x,
Qni+1 − Qn
i
"x
), (6.26)
where the minmod function of two arguments is defined by
minmod(a, b) =
⎧⎪⎨
⎪⎩
a if |a| < |b| and ab > 0,
b if |b| < |a| and ab > 0,
0 if ab ≤ 0.
(6.27)
If a and b have the same sign, then this selects the one that is smaller in modulus, else itreturns zero.
Rather than defining the slope on the i th cell by always using the downwind difference(which would give the Lax–Wendroff method), or by always using the upwind difference(which would give the Beam–Warming method), the minmod method compares the twoslopes and chooses the one that is smaller in magnitude. If the two slopes have differentsign, then the value Qn
i must be a local maximum or minimum, and it is easy to check inthis case that we must set σ n
i = 0 in order to satisfy (6.23).Figure 6.2(a) shows results using the minmod method for the advection problem con-
sidered previously. We see that the minmod method does a fairly good job of maintaininggood accuracy in the smooth hump and also sharp discontinuities in the square wave, withno oscillations.
Sharper resolution of discontinuities can be achieved with other limiters that do notreduce the slope as severely as minmod near a discontinuity. Figure 6.5(a) shows somesample data representing a discontinuity smeared over two cells, along with the minmodslopes. Figure 6.5(b) shows that we can increase the slopes in these two cells to twice thevalue of the minmod slopes and still have (6.23) satisfied. This sharper reconstruction willlead to sharper resolution of the discontinuity in the next time step than we would obtainwith the minmod slopes.
(a) (b)
Fig. 6.5. Grid values Qn and reconstructed qn(·, tn) using (a) minmod slopes, (b) superbee or MCslopes. Note that these steeper slopes can be used and still have the TVD property.
6.9 Slope-Limiter Methods 111
6.9 Slope-Limiter Methods
We now return to the derivation of numerical methods based on piecewise linear reconstruc-tion, and consider how to limit the slopes so that (6.23) is satisfied. Note that setting σ n
i ≡ 0works, since the piecewise constant function has the same TV as the discrete data. Hencethe first-order upwind method is TVD for the advection equation. The upwind method maysmear solutions but cannot introduce oscillations.
One choice of slope that gives second-order accuracy for smooth solutions while stillsatisfying the TVD property is the minmod slope
σ ni = minmod
(Qn
i − Qni−1
"x,
Qni+1 − Qn
i
"x
), (6.26)
where the minmod function of two arguments is defined by
minmod(a, b) =
⎧⎪⎨
⎪⎩
a if |a| < |b| and ab > 0,
b if |b| < |a| and ab > 0,
0 if ab ≤ 0.
(6.27)
If a and b have the same sign, then this selects the one that is smaller in modulus, else itreturns zero.
Rather than defining the slope on the i th cell by always using the downwind difference(which would give the Lax–Wendroff method), or by always using the upwind difference(which would give the Beam–Warming method), the minmod method compares the twoslopes and chooses the one that is smaller in magnitude. If the two slopes have differentsign, then the value Qn
i must be a local maximum or minimum, and it is easy to check inthis case that we must set σ n
i = 0 in order to satisfy (6.23).Figure 6.2(a) shows results using the minmod method for the advection problem con-
sidered previously. We see that the minmod method does a fairly good job of maintaininggood accuracy in the smooth hump and also sharp discontinuities in the square wave, withno oscillations.
Sharper resolution of discontinuities can be achieved with other limiters that do notreduce the slope as severely as minmod near a discontinuity. Figure 6.5(a) shows somesample data representing a discontinuity smeared over two cells, along with the minmodslopes. Figure 6.5(b) shows that we can increase the slopes in these two cells to twice thevalue of the minmod slopes and still have (6.23) satisfied. This sharper reconstruction willlead to sharper resolution of the discontinuity in the next time step than we would obtainwith the minmod slopes.
(a) (b)
Fig. 6.5. Grid values Qn and reconstructed qn(·, tn) using (a) minmod slopes, (b) superbee or MCslopes. Note that these steeper slopes can be used and still have the TVD property.
![Page 57: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/57.jpg)
Sweby diagram Regions in which function values must
lie in order to give TVD and second order TVD
methods.
φ(θ)
TVD TVD & 2nd order
TVD & 2nd orderTVD & 2nd order
all methods in this diagram leave the
TVD region
![Page 58: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/58.jpg)
Order of accuracy isn’t everything
![Page 59: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/59.jpg)
Upwind algorithmmonotone convection
![Page 60: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/60.jpg)
Lax-Wendroff
![Page 61: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/61.jpg)
slope limited, MC limiter
![Page 62: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/62.jpg)
Upwind algorithmWave packet:
![Page 63: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/63.jpg)
Lax-Wendroff
![Page 64: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/64.jpg)
slope limited, Superbee limiter
![Page 65: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/65.jpg)
Slope limiters and flux limiters
![Page 66: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/66.jpg)
Wave limiters
![Page 67: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/67.jpg)
Extension to linear systems
![Page 68: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/68.jpg)
Nonlinear scalar conservation laws
Upwind methods for u > 0.
![Page 69: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/69.jpg)
Weak solutions depend on the conservation law
![Page 70: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/70.jpg)
Conservation form
![Page 71: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/71.jpg)
Importance of conservation form
![Page 72: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/72.jpg)
nonconservative method
conservative method
Burgers' equation solved with an upwind method, to demonstrate that this does not approximate the weak solution properly.
![Page 73: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/73.jpg)
Lax-Wendroff Theorem
Note:
![Page 74: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/74.jpg)
Boundary conditions and ghost cells
![Page 75: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/75.jpg)
Sample boundary conditions
![Page 76: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/76.jpg)
Computed with 400 grid points and the MC limiter.
inflow
inflow
reflecting boundary
reflecting boundary
sinusoidal disturbance enters here
Acoustics equationsZero initial condition.
28 2 Conservation Laws and Differential Equations
and since p0 = P(ρ0), we obtain
p ≈ P ′(ρ0)ρ.
Also we have
ρu = (ρ0 + ρ)(u0 + u) = ρ0u0 + ρu0 + ρ0u + ρu,
and so
!ρu ≈ u0ρ + ρ0u.
Using these expressions in the equations (2.47) and performing some manipulations(Exercise 2.1) leads to the alternative form of the linear acoustics equations
pt + u0 px + K0ux = 0,
ρ0ut + px + ρ0u0ux = 0,(2.48)
where
K0 = ρ0 P ′(ρ0). (2.49)
The equations (2.48) can be written as a linear system"
pu
#
t
+"
u0 K0
1/ρ0 u0
# "pu
#
x
= 0. (2.50)
Here and from now on we will generally drop the tilde on p and u and use
q(x, t) ="
p(x, t)u(x, t)
#
to denote the pressure and velocity perturbations in acoustics.The system (2.50) can also be derived by first rewriting the conservation laws (2.38) as
a nonconservative set of equations for u and p, which is valid only for smooth solutions,and then linearizing this system; see Exercise 2.2.
An important special case of these equations is obtained by setting u0 = 0, so that weare linearizing about the motionless state. In this case the coefficient matrix A appearing inthe system (2.50) is
A ="
0 K0
1/ρ0 0
#(2.51)
and the equations reduce to
pt + K0ux = 0,
ρ0ut + px = 0.(2.52)
In Section 2.12 we will see that essentially the same set of equations can be derived forone-dimensional acoustics in an elastic solid. The parameter K0 is called the bulk modulusof compressibility of the material; see Section 22.1.2 for more about this parameter.
28 2 Conservation Laws and Differential Equations
and since p0 = P(ρ0), we obtain
p ≈ P ′(ρ0)ρ.
Also we have
ρu = (ρ0 + ρ)(u0 + u) = ρ0u0 + ρu0 + ρ0u + ρu,
and so
!ρu ≈ u0ρ + ρ0u.
Using these expressions in the equations (2.47) and performing some manipulations(Exercise 2.1) leads to the alternative form of the linear acoustics equations
pt + u0 px + K0ux = 0,
ρ0ut + px + ρ0u0ux = 0,(2.48)
where
K0 = ρ0 P ′(ρ0). (2.49)
The equations (2.48) can be written as a linear system"
pu
#
t
+"
u0 K0
1/ρ0 u0
# "pu
#
x
= 0. (2.50)
Here and from now on we will generally drop the tilde on p and u and use
q(x, t) ="
p(x, t)u(x, t)
#
to denote the pressure and velocity perturbations in acoustics.The system (2.50) can also be derived by first rewriting the conservation laws (2.38) as
a nonconservative set of equations for u and p, which is valid only for smooth solutions,and then linearizing this system; see Exercise 2.2.
An important special case of these equations is obtained by setting u0 = 0, so that weare linearizing about the motionless state. In this case the coefficient matrix A appearing inthe system (2.50) is
A ="
0 K0
1/ρ0 0
#(2.51)
and the equations reduce to
pt + K0ux = 0,
ρ0ut + px = 0.(2.52)
In Section 2.12 we will see that essentially the same set of equations can be derived forone-dimensional acoustics in an elastic solid. The parameter K0 is called the bulk modulusof compressibility of the material; see Section 22.1.2 for more about this parameter.
28 2 Conservation Laws and Differential Equations
and since p0 = P(ρ0), we obtain
p ≈ P ′(ρ0)ρ.
Also we have
ρu = (ρ0 + ρ)(u0 + u) = ρ0u0 + ρu0 + ρ0u + ρu,
and so
!ρu ≈ u0ρ + ρ0u.
Using these expressions in the equations (2.47) and performing some manipulations(Exercise 2.1) leads to the alternative form of the linear acoustics equations
pt + u0 px + K0ux = 0,
ρ0ut + px + ρ0u0ux = 0,(2.48)
where
K0 = ρ0 P ′(ρ0). (2.49)
The equations (2.48) can be written as a linear system"
pu
#
t
+"
u0 K0
1/ρ0 u0
# "pu
#
x
= 0. (2.50)
Here and from now on we will generally drop the tilde on p and u and use
q(x, t) ="
p(x, t)u(x, t)
#
to denote the pressure and velocity perturbations in acoustics.The system (2.50) can also be derived by first rewriting the conservation laws (2.38) as
a nonconservative set of equations for u and p, which is valid only for smooth solutions,and then linearizing this system; see Exercise 2.2.
An important special case of these equations is obtained by setting u0 = 0, so that weare linearizing about the motionless state. In this case the coefficient matrix A appearing inthe system (2.50) is
A ="
0 K0
1/ρ0 0
#(2.51)
and the equations reduce to
pt + K0ux = 0,
ρ0ut + px = 0.(2.52)
In Section 2.12 we will see that essentially the same set of equations can be derived forone-dimensional acoustics in an elastic solid. The parameter K0 is called the bulk modulusof compressibility of the material; see Section 22.1.2 for more about this parameter.
![Page 77: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/77.jpg)
Advection equationZero initial condition.
inflow outflow
sinusoidal disturbance enters here
![Page 78: Introduction to numerical methods to hyperbolic PDE’sx =0 conservation of mass conservation of momentum conservation of total energy closure relationship - equation of state: polytropic](https://reader033.vdocuments.mx/reader033/viewer/2022042219/5ec50f33bb95b959ec1a4650/html5/thumbnails/78.jpg)
A standing wave in a closed tube with solid wall boundary conditions.Acoustics equations
reflecting boundary
reflecting boundary
reflecting boundary
reflecting boundary
reflecting boundary