Adaptive Meshing using STAR-CCM+ Weekly meeting – 05.12.2010 Victor Pépin.

Adaptive Meshingusing STAR-CCM+

Weekly meeting – 05.12.2010

Victor Pépin

Plan

Curriculum Problem Software presentation Solution Results Next steps

Curriculum

Student at Ecole Centrale Paris French engineering school Non-specialized school 6 months internship in Manchester

• February to mid-July

Problem

How to improve the use of computation resources over the geometry of the simulation ?

Particularly, how to share resources relevantly, depending on the interest each area of the simulation represents ?

A solution : adaptive meshing

Software presentation

Star-CCM+, v4.04.011 (CD-Adapco) Allows the use of macros, written in Java A macro is a list of instructions, forming

an algorithm It interacts with the software via the user

interface The software is seen like a “black box”

Example of macro instructions

// Define the simulation as a variableSimulation simulation_0 = getActiveSimulation();// Define the Stopping Criterion as a variableStepStoppingCriterion stepStoppingCriterion_0 = ((StepStoppingCriterion) simulation_0. getSolverStoppingCriterionManager(). getSolverStoppingCriterion("Maximum Steps"));// Set the value of the Stopping Criterion to 1000stepStoppingCriterion_0.setMaximumNumberSteps(1000);// Run the simulationsimulation_0.getSimulationIterator().run();

Solution Scheme 1/2

Choose a quantity of interest, which will be the criterion of your refinement: e.g. velocity

Launch the simulation with a coarse mesh Export a table containing the velocity in each

cell of the mesh Consider each cell, and compare its velocity

with a reference velocity• Vref = Vmin + percentage x (Vmax – Vmin)

Solution 2/2 If V > Vref :

• Create a box (Volume Shape), centred on the cell, whose size is approximately the size of the cell

• Add the Volume Shape to a Source Volume Set a smaller base size for the mesh upon the

Source Volume Re-mesh the simulation Run the simulation Repeat the process

Results 1/3

2D lid-driven cavity test case Re = 50 Steady state of the flow

Results 1/3

2D turbulent lid-driven cavity Re = 15 000 Establishing flow The mesh follows the flow

Results 2/3

3D cylinder Re = 10 Flow entering the pipe on the left

Results 3/3

Other features of the macro Adapt easily to any case (shape,

boundaries, type of mesh…) Take user-defined field functions as a

criterion for mesh refinement• Gradients,

• Convergence between 2 times steps…

Refine the mesh considering several field functions

Next steps

Measure the gains and loss of accuracy Measure the benefits in terms of time,

and memory

The end

Thank you for listening !