Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

Lecture 9:

Diffusion on Unstructured Meshes


Last Time…

We looked at
z Von Neumann stability analysis
» Apply to explicit scheme and establish stability
limits
z Diffusion on orthogonal unstructured meshes
» Showed that if mesh is orthogonal, structure of
discrete set is like that for Cartesian meshes
This Time…

We will look at
z Consider 2D steady diffusion on non-orthogonal
unstructured meshes
z Realize that the normal diffusion flux is more complex
because of non-orthogonality
z Figure out a way to deal with the extra terms that
result from non-orthogonality
Non-Orthogonal Mesh

Orthogonal

Non-Orthogonal
Cell Balance

z Governing equation:
z Manipulating as before:

z Here, face area vector

z Therefore:
Area Vector

A f = Af n f
n f is outward - pointing face normal
Af is face area
Coordinate Transformation

z Writing the face flux as

is not useful for discretization because the derivative


cannot naturally be written in terms of cell-centroid
values
z Transform to (ξ,η) coordinates
z Note: not orthogonal!
Coordinate Transformation (cont’d)

z Using chain rule:

z Solve for φx, φy:

z Here, Jacobian J
Face Flux

z Using coordinate transformation:

z How to find geometric factors?


Geometric Factors
Geometric Factors (cont’d)

z Furthermore:
Face Flux Terms

z Recall face flux expression

z Let’s look at φξ
Face Flux Terms (cont’d)

z Similarly φη
Overall Face Flux Term

z Overall term:

Primary Gradient Secondary Gradient:

Secondary Gradient:
Discretization

z Make linear profile assumption between cell centroids:

Primary Gradient
Secondary Gradient:
Finding Secondary Gradient

z On structured meshes, there is no problem

⎛ ∂φ ⎞ ( φb − φa )
⎜ ⎟ =
⎝ ∂η ⎠ f ∆η
Discussion

z For orthogonal mesh, what is the value of:

z What is the value of

z How to evaluate secondary gradient ?


Secondary Gradient (cont’d)

z On 2D unstructured mesh, there is no problem:

⎛ ∂φ ⎞ ( φb − φa )
⎜ ⎟ =
⎝ ∂η ⎠ f ∆η

z What about 3D
» No unique η direction
» There is a unique ξ direction, though
Secondary Gradient (cont’d)

z Rewrite secondary gradient as (total-primary):

Primary

Secondary = Total - Primary

Need to find ( ∇φ ) f
Notice: no mention of η direction
Secondary Gradient (Cont’d)

z For a uniform mesh, if we have cell-gradients, we


could find face value as

z Will find better ways to interpolate later


Discrete Equation

z Arranged so as to
have diagonal
dominance
z But extra terms
contained in b
term
z Will need outer
iteration even for
linear problems
z Boundedness?
Closure

In this lecture we:


z Considered the steady diffusion equation
z Basic procedures are the same as for structured
meshes
z Non-orthogonality produces a secondary gradient term
z Easy to find secondary gradient in 2D unstructured
meshes, but problems in 3D
z Addressed 3D using (total-primary) decomposition

You might also like