← Back to team overview

yade-users team mailing list archive

Re: [Question #670862]: Applying normal strain rates in PeriTriaxController engine

 

Question #670862 on Yade changed:
https://answers.launchpad.net/yade/+question/670862

Bruno Chareyre proposed the following answer:
For the record:

- despite above statements I confirm that the strain rate is simply
imposed by maxStrainRate independently of sample size.

 - there is no "ramping" at any point (ramping is linear and defined with a specific slope). Simply: if you want to go from 0 to N by steps S, and if N is not a multiple of S then the last step needs to be less than S. It causes the apparent ramping but it happens only at one specific iteration and it is not linear in general, hence not really a ramp in the classical sense.
For instance if O.dt=1, (target-current) = 0.1 and maxStrainRate=0.3, the sequence will be {0.3, 0.3, 0.3, 0.1, 0} 
If O.dt!=1 the sequence will different.

- clearly if one wants to impose rate of strain for a number of
iterations, instead of a final target strain, the absolute value of
target strain is arbitrary but it must be large enough. I would set it
to e.g. +/-1e10  depending on desired direction.

@Mehdi
- All this should not matter at all even for small strain probing. If I understand correctly you think smaller strain increments can be achieved with fewer iterations, but it is not really the case. If you reach a given strain in less than, say, 100 iterations then the final state is probably not equilibrated enough to get a relevant measure of anything. What you get is just numerical noise.
On the other if you go from one state to another in 1000+  iterations then the fact that the last iteration has a smaller rate is probably not an issue.

- your apparent size dependency is due to using different particle
sizes, hence different O.dt.

-- 
You received this question notification because your team yade-users is
an answer contact for Yade.