Hi,
Is there any way to calculate how much time the explicit finite element analysis need to be run.
Considering load duration, the load duration time (Delta_T) can be estimated by the following empirical rule:
Delta_T > 50 * delta_t_Courant (1)
where delta_t_Courant = delta_x / Cp, and Cp is the fastest bulk wave of the media, Cp=sqrt((lambda + 2*mu)/ro) with lambda and mu are Lame's constants and ro is the material density. If inequality (1) is violated you'll generally get visible oscillations behind the wave front.
Similar estimate can be applied to periodic loadings. In this case the empirical estimate to avoid high frequency noise is the following:
omega < 2*pi /(50 * delta_t_Courant) (2)
It also depends on the number of degrees of freedom of the model (number of nodes, number of DOFs per node, etc.).
Hi,
I am not taking about computational time, I want know how much time the analysis need to performed
solving time may be 10hrs or 50 hrs, I am not bothered at it, i want to know how much time i need to give(like 1 milli sec) at the start of the analysis, so that i can get proper result (optimum time for the explicit analysis), for example if a load of F N is applied at the free end of the cantilever, I can apply the load in T1 sec or for T2 sec, how to calculate that time.
So you want to estimate the critical (or stable) time step size? You can find estimates e.g. in Belytschko et al.'s book "Nonlinear Finite Elements for Continua and Structures", or in the manuals of explicit finite element codes, say LS-DYNA:
http://www.dynasupport.com/tutorial/ls-dyna-users-guide/time-step-size
Regards
My understanding is that you want to know the time period for which to analyze the problem or when to stop it. It depends on applications. Generally it would be the users discretion unless a standard is already defined for the particular case scenario. For eg, if it is impact problem, its logical to run the simulation till contact between the object and the projectile is present. If it is crash studies, you would probably give a velocity/acceleration curve to bring the speed to zero during the crash. The curve (the point of zero velocity/acceleration) will tell you when the simulation should stop.
Having said that, in explicit solvers (eg. LS-Dyna), if you have initially given a smaller time, you can restart the calculation from the end point and run for extra time that you think you need or if you have given more time, you will realize it through parallel post-processing (you can check results as the calculation is running) and hence you will know if the analysis is good enough to stop.
Hope it helped you! All the best!
Considering load duration, the load duration time (Delta_T) can be estimated by the following empirical rule:
Delta_T > 50 * delta_t_Courant (1)
where delta_t_Courant = delta_x / Cp, and Cp is the fastest bulk wave of the media, Cp=sqrt((lambda + 2*mu)/ro) with lambda and mu are Lame's constants and ro is the material density. If inequality (1) is violated you'll generally get visible oscillations behind the wave front.
Similar estimate can be applied to periodic loadings. In this case the empirical estimate to avoid high frequency noise is the following:
omega < 2*pi /(50 * delta_t_Courant) (2)
Thanks for you responses
I had another question
case-1:If we are applying load very slowly(time is very high), the dynamic (inertia) will not come in picture-(static analysis).
case-2:Where as if the load is applied in less time dynamic effects will come in to picture (inertia)-(dynamic analysis),
How to calculate the time (time in case-2) so that we will know that inertia effects are needed to be considered?
Let L be a typical dimension of the region, where dynamic effects should be studied, or L can be a typical dimension of the model. Then if load variation time T is smaller than (or comparable with) L/Cp, the dynamic effects will be prominent. Note, that instead of Cp you can use other velocities Cs or Rayleigh velocity Cr, depending on the wave nature you analyze.
Critical time increment should be as:
Δ𝑡𝑐𝑟≤𝑙/𝑐𝑑
where 𝑙 is the dimension of the smallest discrete element in the model and 𝑐𝑑 is dilatation speed of material, defined as:
𝑐𝑑=√2𝐺(1−𝜈)/(1−2𝜈)≈√𝐸/𝜌
- Badges
- Science topic