I want to do a quasi static torsion of cylindrical sample in explicit dynamics

The explicit dynamics is asking for CFL for quasi static. What will be the CFL for avoiding inertial effects.

Best Answer

  • peteroznewmanpeteroznewman Member
    edited December 2020 Accepted Answer

    @rahulb

    It wants to know the CFL not to avoid inertial effects, but to reduce the number of hours you have to wait for a solution. It's up to you to configure the end time so that the inertial forces do not get too high. You plot the Kinetic Energy and compare that to the total energy in the result to check that KE is a small fraction. If you have to make the end time 10 times longer to do that, you have to wait 10 times longer for the solution to finish.

    Change the analysis settings back to the default, where it does not ask for a CFL and the solver will run.

    When it does run, you will see in the Solution Output what the Time Step is for you model, which is the CFL value it is looking for. The CFL time step is a combination of the minimum Characteristic Length (CL) in your mesh, the density and the Young's Modulus of your material. You will also see a prediction of how many hours the solution will take.

    If your mesh is perfectly uniform in element CL, then you won't benefit from using the Quasi-Static settings. But if your mesh has a few elements with much smaller CL than the rest, then the QS setting will artificially increase the density of those small CL elements to increase the Time Step.

    If your mesh is very uniform in element CL (plot that Mesh Metric to see the distribution), then you can reduce the solution time by artificially increasing the density of the materials in the model. Of course this is going to cause the KE to increase, so I'm not sure that you take any benefit from that over just reducing the end time.

    If your mesh has a few elements with small CL, try to use mesh controls or slicing the geometry to increase the minimum CL and run again to see the Time Step size. Suppose the Time Step was 1e-8, you can ask the solver to use a Time Step of 1e-7 by typing in that value where it asks for the CFL values in the Quasi-Static setting and it will change the density of those few elements to deliver the solution with a reduction in wait time by a factor of 10. Those elements will generate higher inertia forces, but if the elements are small, this is an acceptable price to pay for a large reduction in wait time.

Answers

  • peteroznewmanpeteroznewman Member
    edited December 2020 Accepted Answer

    @rahulb

    It wants to know the CFL not to avoid inertial effects, but to reduce the number of hours you have to wait for a solution. It's up to you to configure the end time so that the inertial forces do not get too high. You plot the Kinetic Energy and compare that to the total energy in the result to check that KE is a small fraction. If you have to make the end time 10 times longer to do that, you have to wait 10 times longer for the solution to finish.

    Change the analysis settings back to the default, where it does not ask for a CFL and the solver will run.

    When it does run, you will see in the Solution Output what the Time Step is for you model, which is the CFL value it is looking for. The CFL time step is a combination of the minimum Characteristic Length (CL) in your mesh, the density and the Young's Modulus of your material. You will also see a prediction of how many hours the solution will take.

    If your mesh is perfectly uniform in element CL, then you won't benefit from using the Quasi-Static settings. But if your mesh has a few elements with much smaller CL than the rest, then the QS setting will artificially increase the density of those small CL elements to increase the Time Step.

    If your mesh is very uniform in element CL (plot that Mesh Metric to see the distribution), then you can reduce the solution time by artificially increasing the density of the materials in the model. Of course this is going to cause the KE to increase, so I'm not sure that you take any benefit from that over just reducing the end time.

    If your mesh has a few elements with small CL, try to use mesh controls or slicing the geometry to increase the minimum CL and run again to see the Time Step size. Suppose the Time Step was 1e-8, you can ask the solver to use a Time Step of 1e-7 by typing in that value where it asks for the CFL values in the Quasi-Static setting and it will change the density of those few elements to deliver the solution with a reduction in wait time by a factor of 10. Those elements will generate higher inertia forces, but if the elements are small, this is an acceptable price to pay for a large reduction in wait time.

Sign In or Register to comment.