peteroznewman
Bbp_participant
You don't understand aliasing and how to avoid it. You can't simply increase the output to 5000 points in the 0.05 s time. That is a sampling period of 1e-5 s. What is the Explicit Dynamics time step? It may be 1e-8 s, or 1000 times smaller. The point of the paper is that you have to have every single time step from the simulation. You can't take just one in every 1000 time steps and filter that decimated data. That is how you create an aliased signal. You have to apply the filter on every single time step from the simulation. Doing this requires limiting the output to a very small number of nodes to avoid filling your entire hard drive.
LS-Dyna has recognized this issue and provided an internal low pass filter that is applied to every single time increment in the solution. Did you use the *CONTROL keyword and set the IACCOP flag equal to 2 before you you requested the acceleration output? If you don't do this, then you are better off using the velocity data.