Timesteps

Altair Forum User

Hey, 

 

how is Radioss defining the timesteps for solving my problem? For a solved problem what have a duration of 2 milliseconds radioss produced 134 frames. Is it possible to define my own timesteps, framenumber or frequency of writing results? Especially if I estimate problems they need a minute or so that I can define that Radioss should write a resultfile every second.

 

Thank you very much.

 

Best,

Max  

Altair Forum User

I found it at the book 'Crash Analysis With RADIOSS' (page 181 'Solid Element Time Step') so I assumed there is no way to define the timestep because it depends on the length and some other material parameters (Length of elements, speed of sound, bulk modulus and Lame moduli)? Is it right?

 

Best,

Max  

Altair Forum User

Hi Max,

From the time step equations, we can see that both material modulus and density control the speed of sound. The solution time step is inversely proportional to the speed of sound. So, if the material is stiffer (higher modulus of elasticity), the speed of sound increases resulting in a lower solution time step. Based on these parameters, solver will define the time step for the run.

The user can impose a time step for the run. In order to impose user should use /DT/NODA/CST in the engine file specifying the Tmin, which is the target time step. But in the other hand, this will lead to adding mass to the system in order to maintain the specified time step. Another option is using Advanced Mass Scaling technique where Advanced Mass Scaling (AMS) allows setting the time step to a higher value than the usual one. AMS does not modify the global mass so that the global momentum of the related nodes is conserved

I recommend you to go through https://altair-2.wistia.com/medias/o0bfml9ah5 link which describes about checking time step of a run and imposing a time step to a run.

In order to increase the frequency of writing animation files user can provide a smaller value of /ANIM/DT, which is specifies the frequency of writing animation files.

 

Altair Forum User

Thank you very much. 

 

SO, if I understand it right you have a critical timestep and for estimation my timestep isn't allowd to be smaller than my critical timestep right? But for me it is more important to increase the timestep, what means that I can save cpu time. 

For example my timestep is estimated as 5e-008 so and i want want to increase the step up to 5e-006, what would save cpu time, right?

Sorry for that confusing questions /emoticons/default_wink.png' srcset='/emoticons/wink@2x.png 2x' title=';)' width='20' /> 

 

Best,

 

Max  

Altair Forum User

Hi Max,

You can increase the timestep to save run time. But in the other hand it will increase the global mass of the system, if you are drastically increasing the timestep. Since mass is directly proportional to KE,it will be increasing and the results will not be reliable.

User should monitor the mass added to the system in the engine out (_0001.out) file. It will be indicated in percentage. Please see the attached image.

MAS.ERROR, dM/M column in the below image shows the mass error. The mass error shows is 0.8e-3 which indicates that increase in mass by 0.8% . For a model which is assigned with the optimum time step the mass error should be less than 3%, that is .03. 

<?xml version="1.0" encoding="UTF-8"?>

Altair Forum User

Hi George, 

 

thanks for all the details. That makes totally sense. But is there a way to fix this mass error, if i increase my timestep? And just a small question dm/m = 0.8e-3 these are 0.08% or (just to know the meaning of the values)?

 

Thanks you a lot.

 

Best,

Max  

Altair Forum User

Hi Max,

There are no methods to fix it, but choose or impose an optimum timestep and thereby make sure that the mass added is very less to the system.

Mass error (dM/M) of 0.8e-3 shows that 0.08% mass has been added to the system for maintaining the user imposed time step.

Altair Forum User

Hey George, 

 

thanks for the explanation. but now I'm confused, because in my model the Dm/m started with 1.0353E+00 and it was processed with RADIOSS calculated timestep. For my thinking then it would be better to decrease the timestep, right? 

 

Best,

Max 

Unable to find an attachment - read this blog

Altair Forum User

Yes Max, use a smaller value of timestep.

Altair Forum User

Thank you very much you helped me a lot! 

 

Best,

Max 

Altair Forum User

Hey, 

 

just a short last question. Is it for the Mass Error ok (attached txt)?

 

Or has it stay constant all the time, because during the estimation the error is increasing.

 

From: 0.3241E-03 at the beginning to 0.5592 at the 386000th cycle.   

 

Best,

Max  

Unable to find an attachment - read this blog

Altair Forum User

Hi Max,

You can see that in the last cycles of the out file the time step is increasing and the solver is adding more mass to the system to keep the timestep to the user specified value.

In the final cycles, mass error is getting more. It is always advised to keep the mass error less for reliable results.