Hi all,
I am using DYNAMIC RELAXATION to apply preload (Gravity or Geo-stress) for a model, and checked the convergence of DR phase. Then do the transient analysis by invoking the drdisp.sif file in another Keywork file (idrflg set to 2). But the curve of stress vs time always oscilates before the stress wave arrive at the monitored point (near the boundary).
Could someone give me some suggestions?
Hi @csuznxg
Are the stress oscillations substantial? I’m not sure what units you are in or the nature of your simulation, but small stress oscillations can occur during settling due to any contacts.
Not very substantial. But it indeed has a negative effect on the numerical result. There is no contact defined. A simple model with preload, and then transient analysis.
I see, you are numerically applying a preload. You could try to lengthen the time duration of your load curve for the preload to see if that dampens the oscillations.
I checked that but there is no much difference. Also the convergence is checked, I am sure each model has reached convergence.
So you were running two separate simulations and condensed into one simulation?
I see, I did not realize that’s what your original question was saying. Interesting that it fixed the oscillations though...
Actually, if all the boundaries are constrained, there is no oscillation. But for the model with a free surface, the oscillation is very distinct.
For one keyword file, I tried two techniques, which are implicit-explicit converting and dynamic relaxation. Both of them meet my needs.
However, birth and death time for the boundary conditions is inactive in the dynamic relaxation phase. Thus, implicit-explicit switch is the best way when the boundary conditions need to be changed in transient analysis.