You are on page 1of 2

REPORT ON TEMPORAL SCHEMES ON TOPOCHANGING MESH

With layer removal, the minThickness and maxThickness which will specify the size of the cell
before and after removal is quite important. Inconsistent layer removal results in pressure
fluctuations and wiggling interfaces in multiphase simulations. With a properly constructed case,
the effect of min. and max. thickness will be studied.

We have a 457.2 mm long 2D shot sleeve with pure hex elements. The blockMesh grading will be
compared along with the working combination in a table. By doing that, we might have an idea how
the parameters can be set in order to get a proper flow profile.

Case Name Grade Cell length minThickness maxThickness ddt scheme Remark
(shotSleeve) (m) (m) (m)

dambreak 270 0.00169 0.0008 0.0015 Euler No wiggles


dambreak1 270 0.00169 0.0008 0.0015 CrankNicolson 0.5 No wiggles
dambreak2 270 0.00169 0.0008 0.0015 CrankNicolson 1.0 Wrong
dambreak3 270 0.00169 0.0008 0.0015 backward No wiggles
dambreak4 270 0.00169 0.0008 0.0015 SLTS phi rho 0.5 No wave
profile
dambreak5 270 0.00169 0.0008 0.0015 CoEuler phi maxCo No wave
0.5 profile

Right from the first two cases, we can see that the numerical schemes also play a role in simulations
with topochanging mesh. A brief summary of the ddt schemes is provided to gain an understanding
as to why the simulation gives erroneous results. Keeping the other schemes constant, using
CrankNicolson 1 gives unrealistic results.

Alpha values (value to be bounded between 0 and 1):

Euler gives bounded values.


CN with blending scheme gives bounded values.
Backward also gives bounded values.
CN 1.0 gives unbounded values.
SLTS is not applicable as it is meant for pseudo steady state problems.
CoEuler is also not applicable.

Transient problems require ddt schemes to be consistent with the mesh motion and correctly map
the field in case of cell layer removal or addition. From cfd-online forum, it seems that the field
mapping after a layer removal introduces some problems as compared to layer addition as the
previous point field gets changed and is not available for flux correction.

Pressure comparison for the different schemes used in shotSleeve_dambreak case.

∂ϕ
ddtSchemes ( Schemes)
∂t

Time schemes define how a property is integrated as a function of time ∂ ϕ /∂ t . Depending on the
choice of scheme, field values at previous time steps are required, represented in the following as φo
and φoo for the old and old-old time levels. OpenFOAM provide the following schemes for temporal
discretization.

1. Backward time scheme


2. Crank-Nicolson time scheme
3. Euler implicit time scheme
4. Local Euler implicit/explicit time scheme
5. Steady state time scheme.

Backward time scheme:


∂ (ϕ )= 1 ( 3 ϕ −2 ϕ o + 1 ϕ oo )
∂t Δt 2 2
ddtSchemes
{
default backward;
ddt(phi) backward;
}

Properties of backward scheme:


1. Implicit
2. Second order
3. Transient
4. Boundedness not guaranteed
5. Conditionally stable

You might also like