close


Beginning LS-DYNA versions 971, a new option for SOFT=2 contact called DPRFAC in Optional Card ‘C’ in *CONTACT keyword allows transient removal of any initial penetrations. This option is useful to quickly zero out any initial penetrations rather than just to remember them. The following figure shows how the effective penetration that is used to compute the contact forces is gradually increasing and a subsequent drop in the initial penetration recorded during initialization.

(Click image to enlarge)

To demonstrate the practical use of the DPRFAC, we can consider a problem used in interference . The prolem has a rigid hollow sphere and a foam block and there is significant initial penetration at time zero. When DPRFAC is zero and and the option IGNORE is turned on, LS-DYNA always removed the initial recorded penetration from the current penetration which results in a very small force and subsequent smaller amount of foam compression. The transient plot for the case is shown below.

Without DPRFAC (Click image to animate)

When DRPFAC is set to 0.1 (reduce the initially recorded penetration by 10%/cycle), the penetrations are eventually forced out as in interference problem which causes foam block to deform and finally assume the geometry of the rigid sphere. This is shown below.

With DPRFAC (Click image to animate)

The following is the force time history comparing a case of DPRFAC=0, DPRFAC=0.01, and DPRFAC=0.1.

(Click image to enlarge)

Real world application would eventually be in places where we want to model a interference fit in full-scale models which are currently ignored. Consider a full vehicle which has several places where parts are press-fitted. In finite element models that exists today, if penetrations exists due to such interference, we simply use the IGNORE parameter to ignore them, which is a big advancement in itself when compared to the process of removing them manually, but in the case of DRPFAC, we will eventually be able to a capture such stress more accurately.

  • Sven says:

    Hi Suri,
    The LS-Dyna 971 Keyword manual (22-01-07) says that “DPRFAC does not apply to initial penetrations at the start of the calculation, only those that are measured at later times. This prevents nonphysical movement and energy growth at the start of the calculation.”

    Can you explain what this means since, at least in your example here, it seems that the initial penetrations (due to geometry overlap) *are* in fact affected by the DPRFAC option.

    The keyword manual seems to suggest that its usage is intended for keeping initial penetrations throughout a simulation and then removing them (over a few thousand cycles or so) at the *end* of a simulation, so that non-penetrating geometry can be output after a simulation. Your example however looks like its removing initial penetrations over the *first* few thousand cycles (ie, before 0.002 sec) so that physically correct geometry is used throughout the simulation.

    Cheers,
    Sven.

  • Suri Bala says:

    Hi Sven,

    It is true that DPRFAC does not apply to initial penetrations. What this really means is if the two parts which have initial penetrations between them (such as the rigid sphere and the foam block) and have NO further relative motion between them, then DRPFAC has no effect. This is because, as you see in Figure 1, the effective penetration is the difference between the current and the initial. If the current and the initial is the same thought the simulation, then the effective penetration is simply ZERO. DRPFAC comes into play ony when the two parts move closer to each other or when the contact segments penetrates further into the impacting segement which results in a non-zero effective penetration and which increases rapidly as step 2 continues to reduce the initial penetration by a fraction of DRPFAC. In the case above, I prescribe a displacement of 0.0001mm (very small) over the length of the simulation to allow DRPFAC to come into play.

    The original intention of DRPFAC was to provide a final geometry that has no penetrations. However, as you can see from the example, it serves well in other interference type situations as well and is necessarily not limited to be used towards the end of the simulation.

  • Sven says:

    Thanks Suri, that clears things up.

    Do you think that this method could be developed as an initial penetration removal tool? I’m of course thinking about complex geometry models with small penetrations of (mostly) deformable parts.

    My understanding is that current initial penetration removal algorithms occur before the simulation starts, and proceed by pushing and prodding nodes outside of initial contact regions. This can sometimes affect local geometries (element quality, surface smoothness etc), and also the initial element stresses are reset to zero despite this node movement.

    I’m now thinking of an application that instead uses DRPFAC to try to remove initial penetrations. Perhaps a few cycles at the start of a simulation with some kind of small vibration applied to parts likely to have initial penetrations (to avoid the ‘zero effective penetration’ you mentioned above). This would ideally let penetrating parts untangle themselves a little more naturally than before, and would also have the benefit of retaining stresses based on originally supplied geometries.

    Cheers,
    Sven.

  • Suri Bala says:

    Hi Sven,

    Yes. That is what is really exciting about DPRFAC. It can remove all interference in complex models using a SINGLE SURFACE type contacts and can record the stresses based on this.

    If you compare this to a INTEFERENCE type contact, it scales the stiffness from zero to maximum while DRPFAC scales the penetration from zero to maximum to acheive the same end result. The biggest advantage is the option to use in complex models where SINGLE SURFACE is used. The other disadvantage is ofcourse the problem of ensuring non-zero effective penetration by enforcing some sort of movement to the nodes which may be a vibration type as you are suggesting.

  • Victor says:

    Hi Suri,
    I try to apply parameters SOFT=2 and DPRFAC=0.1
    (*CONTACT_AUTOMATIC_SURFACE_TO_SURFACE ) for modeling
    interference fit with torque moment.
    I do not apply *CONTACT_NODES_TO_SURFACE_INTERFERENCE
    However I have problem – is not present interference pressure
    and stresses.

    When I applied *CONTACT_NODES_TO_SURFACE_INTERFERENCE
    stresses at action of the torque moment
    had strong fluctuations.

    Cheers,
    Victor

    PS. Sorry,I applied program – interpreter for
    Translation from Ukrainian on English.

  • Suri Bala says:

    Victor,

    I am confused. Are you referring to bolt prestress or interference fit ? It appears that you are referring to both.

    Suri Bala

  • Victor says:

    Hi Suri,

    I try to simulate two cylinders, which are interference fit.
    After one cylinder is loaded the torque, he will begin rotation
    I must be identification torque of destruction.

    Modeling interference fit the torque is function of
    time = revolv cylinder —> torque of destruction.

    Cheers,
    Victor

  • Suri Bala says:

    Victor,

    I am not clear but if I understand it correctly, you have two cylinders that have some initial interference and you want to model this in addition to applying a torque to one of them.

    If this is correct, I would recommend to first allow DRPFAC to take affect for some time and then you can apply the torque.

    Alternatively, you can use *CONTACT_SURFACE_TO_SURFACE_INTERFERENCE.

    If you are seeing strong fluctuations, I would recommend a slower ramp of the stiffness.

    Suri

  • Apoorva L says:

    Hi Suri,

    Iam really amazed at this option drpfac. I want to use this option. However i do not have the latest ls-prepost preprocessor to use this option. What would be the ls-dyna deck format for inputting this option manually. any help would be greatly appreciated.

  • Suri Bala says:

    Apoorva,

    DRPFAC is just a parameter in one of the optional cards in *CONTACT. Manually you can either input a positive value or a negative value where the latter would then cause LS-DYNA to look for a curve.

    Suri

  • Tim says:

    i have penetrations in my model, how do i remove them. what options should be switched on in contacts and how does it work. please let me know.

  • Suri Bala says:

    Tim,

    Please use the IGNORE option in *CONTROL_CONTACT.

    Suri Bala

  • Bjorn says:

    Hi Suri,

    I’m working with a folded airbag, i.e. there are several very thin (0.37mm) fabric layers on top of each other. Now I want to unfold the outer layer via a dynamic simulation which runs over 100 milliseconds. However, since I’m going to use the results in another simulation I want to avoid having any penetrations in the airbag self contact in the end of this simulation. I have defined a DPRFAC curve which is zero up to 98 milliseconds, then ramps up to 0.01 at 99 milliseconds and then remains at 0.01. However, I’m still getting about 100 crossed edges and 150 initial penetrations at the end in this simulation. Do you have any general recommendations for how to define the DPRFAC curve? Can you think of something else that can improve the situation in my case? For your information: I’m using *CONTACT_AIRBAG_SINGLE_SURFACE with SOFT = 2, SBOPT = 3 and DEPTH = 5.

    Cheers
    Bjorn

Leave a Reply

Your email address will not be published. Required fields are marked *


*