how to set time step for ls-dyna

Altair Forum User
Altair Forum User
Altair Employee
edited October 2020 in Community Q&A

Hi,

 

I recently mesh an aluminium tube with 3mm square shell element and applied a time step ('control_timestep') with parameter of TSSFAC 0.9 and DT2MS -0.05. So the calculated time step is 0.045ms. (unit system: Gpa, mm, ms, kg) As soon as I assigned crashing velocity as 150mm/s, the simulation went well and yet a reasonable result. 

 

Q1: according to the CFL condition, the time step can be calculated by the formula t=L/(sqrt(E/p*(1-v^2))). But if I use this formula, the time step is 0.55e-3ms which much much smaller than 0.045ms. Since the calculation gives the minimum time step, how is it possible to give me the correct simulation result by using a bigger one?

 

Q2: when I assigned another speed 0.33mm/s, I just got a straight line (see attachment) over the force-time curve. What should I put in the control_timestep card?

 

Please help.

 

 

Many thanks.

<?xml version="1.0" encoding="UTF-8"?>post-7386-0-43820800-1401251089_thumb.jp

Tagged:

Answers

  • Altair Forum User
    Altair Forum User
    Altair Employee
    edited May 2014

    I just realize if the formula t=L/(sqrt(E/p*(1-v^2))) is used to calculate the critical time step, since the density of metal is always with e-6, when it gets out of the square root, e-3 is maintained and the time step is always around 0.001ms which is too small. 

     

    Please help.

  • Altair Forum User
    Altair Forum User
    Altair Employee
    edited May 2014

    Hi,

     

    try to use the Hypermesh tool to calculate the minimum time step (based on elements, not contacts).

     

    Use on the Toolbar 'Tools->check elems->time' ... here you can see the minimum value, added masses and so on. Very helpful.

     

    Best regards,

    Mario

  • Altair Forum User
    Altair Forum User
    Altair Employee
    edited May 2014

    Hi Mario,

     

    Thanks for your tips and it is really helpful. 

     

    I still have three questions about CFL criterion:

    Q1: As far as I understand through the reading, the time step applied to the simulation must be less than that calculated from the CFL criterion. But from the HV, it gives the minimum time step. Does it mean I need to apply a time step bigger than it? If so, it conflicts with CFL criterion.

     

    Q2: I have also done a simulation with a reasonable time step 0.9*|-0.05| (TFFSAC*DT2MS in the control_timestep keyword) for a termination of 1000. (unit in ms) The simulation went well. This also conflict with the minimum time step given by HV.

     

    Q3: I have tried a very big time step 500 (termination time is 100000, unit in ms), as soon as I tried to run the simulation, the solver just simply crashed and told me to close the program. I have tried three times, but every time is the same. Is there a maximum accepted value by the solver? If so, how much is it? If not, why it didn't go through the simulation like in Q2 with a bigger time step?

     

    Thanks a lot.

  • tinh
    tinh Altair Community Member
    edited May 2014

    Hi

    yes, it conflicted but not, when you specify a timestep by DT2MS, if dyna calculate timestep smaller than your value, dyna will add mass to the nodes

    small elems contribute less mass to nodes so dyna add mass to these. a nodal mass model have critical timestep as 2/w and it's higher when more mass added

    please read chapter about time step control in manual...

    finally, if you specify bigger timestep, more mass is added to your model and this is error. simulation went well, but no correct result

     

    just look in d3hsp, it will show you smallest timestep, how many percent of mass added,  and how big timestep that you should not exceed

Categories