Trying to model a bird impact on a flat plate, it seems that I have an issue with the contact. The bird goes pass the plate without any collision. Can anyone help me in understanding what’s going wrong with the model? While running I saw a message regarding contact which says “The LS-DYNA time step size should not exceed 6.538E-06 to avoid contact instabilities. If the step size is bigger then scale the penalty of the offending surface.” Does the contact issue related to this any way? First video shows the issue and the keyword file also attached.
Plate material properties assigned:
Mass density = 2.59000E-4 lbf.s2/in4
E = 10.5e6 psi
Poissons Ratio = 0.33
Bird material properties:
RO = 8.889e-05
PC = -14.50
MU = 3.916e-07
From the looks of things you have put everything into the contact cards correctly, so passing right through is definitely strange.
You are using the default contact formulation with a really thin plate though, have you tried increasing the thickness a little like .5, 1.0? The nodes might be passing right through between timesteps, and the funny part is that if they keep passing through the timestep will stay too high.
Invoking a SOFT contact miiiight help. They are usually the go-to for node issues since they are segment based, but to be honest I am not sure if they are as effective for SPH-Lag. I'd say try that too.
Thank you so much for your response @sbrrr. I tried with an increased thickness of 1.00” and I see things improved a little bit. Now few particles ahead of the bird impact with plate but most of these particles then leak through the plate itself. Still the issue exists as shown in the first part of the attached video. In next iteration, I also changed the TSSFAC value from 0.5 to 0.15. TSSFAC value change make the contact works correct, as shown in attached video. Is it possible to use a low value of TSSFAC = 0.15 for analysis? Can anyone make me understand why this change in TSSFAC value make the contact works correct?
@paulvazz Hmmmm, there is a lot to unpack there. What you're asking is kinda the core principal of explicit analysis.
To answer your question directly, TSSFAC is a coefficient that gets applied to the next timestep LSDyna calculates to be the minimum of all possibilities in your simulation. You can find it in vol 1 of the manual, but it really is just a coefficient next to a min() operation.
The reason TSSFAC is helping you is because timesteps are functions of the how long it takes to propagate information through your medium. For a Lagrangian mesh the speed of sound through a given element is a determined by the lengths of each side, because those are the nodal connectors, and the material properties. So as an element begins getting crushed, Dyna tries to determine acceptable timesteps for each iteration of solving. Smaller elements, smaller timesteps.
Sph is a has a similar effect, but obviously they aren't "elements" any more so it uses a velocity field as a spacial component if I remember correctly; it's definitely in the LSDyna theory book. That's probably why the first half of you video was sliced in half. After enough particles were disturbed the timestep remained fine enough to determine every collision.
As for contact, you would definitely be better off looking through the Dyna support pages than having me explain them. They have some nice pictures and such. I think the key thing to your problem, is that having such a thin plate BETWEEN two layers of Sph exerts no disruptive force on any of the particles. So until a few of them happen to land on it, in the exact moment of a given timestep, they have no idea it's there. That's why a thicker plate also helps. If you had a plate whose thickness is the same as your Sph spacing, then I don't think you'd have any issue with leaving TSSFAC at the default. Hopefully that will make sense once you skim through the contact section.
Hmmm. . I tried my best to be concise with those but I probably ended up leaving out too much important stuff. Anyway, you asked about two vital pieces of using explicit fea, I'd definitely recommend hitting the manuals for those topics because there is quite a lot more than I can type 😅.
include * CONTROL_SPH card . . . ( here NMNEIGH should be varied accordingly )
Also your Lsdyna Version matters.