Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Parametric Segregated solving, Quick question
Posted 29 nov 2010, 21:28 GMT-5 5 Replies
Please login with a confirmed email address before reporting spam
Hello,
I have a quick question about Parametric Segregated solving--
I'm solving an off-shoot of the 3D Cantilever beam (with different geometries, etc.) with three coupled applications:
1) Electrostatics (es)
2) Solid, Stress-Strain (smsld)
3) Moving Mesh (ALE)
When solving this model, it seems that I must use Parametric Segregated groups and solve the three applications sequentially (which apparently causes ease of solution). What I'm wondering, is, how exactly do I solve using Parametric Segregated solving, should I solve it like this:
Method A
1) (es)
2) (es) + (smsld)
3 (es) + (smsld) + (ALE)
-or-
Method B
1) (es)
2) (smsld)
3) (ALE)
Basically, in the "Solve for" tab of the Solver manager, I'm wondering if I should solve my parametric segregated solution using Method A or Method B.
Thanks.
Best regards,
Kevin
I have a quick question about Parametric Segregated solving--
I'm solving an off-shoot of the 3D Cantilever beam (with different geometries, etc.) with three coupled applications:
1) Electrostatics (es)
2) Solid, Stress-Strain (smsld)
3) Moving Mesh (ALE)
When solving this model, it seems that I must use Parametric Segregated groups and solve the three applications sequentially (which apparently causes ease of solution). What I'm wondering, is, how exactly do I solve using Parametric Segregated solving, should I solve it like this:
Method A
1) (es)
2) (es) + (smsld)
3 (es) + (smsld) + (ALE)
-or-
Method B
1) (es)
2) (smsld)
3) (ALE)
Basically, in the "Solve for" tab of the Solver manager, I'm wondering if I should solve my parametric segregated solution using Method A or Method B.
Thanks.
Best regards,
Kevin
5 Replies Last Post 30 nov 2010, 02:47 GMT-5
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
basically you are asking if onne are looping:
a) once through the segregated solver (Metod A)
b) or many times, for which method B should do best.
Good question !
I had the impression it was "b)", but it is worth to have a closer look, unfortunately I do not know how to turn on a verboose solver log, could be handy some times ;)
--
Good luck
Ivar
basically you are asking if onne are looping:
a) once through the segregated solver (Metod A)
b) or many times, for which method B should do best.
Good question !
I had the impression it was "b)", but it is worth to have a closer look, unfortunately I do not know how to turn on a verboose solver log, could be handy some times ;)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
so you're saying that I should go with Method B?
Because Method A basically represents solving electrostatics alone, then (es) + (smsld), then all three together, while Method B represents solving each one independently and separately in a sequential manner.
Basically, I'm new to solving with Parametric Segregated solutions, so it would be correct to solve each application separately and independently, one by one? In this manner, would my final solution be the one after I solve part 3 (moving mesh ale)? I'm looking for the deflection of a thin membrane in an applied electric field (similar to 3D cantilever beam), so would this final solution of the deflection be correctly found by using Method B after finally solving for the moving mesh ALE?
Thanks for your help, as always.
Best regards,
Kevin
Because Method A basically represents solving electrostatics alone, then (es) + (smsld), then all three together, while Method B represents solving each one independently and separately in a sequential manner.
Basically, I'm new to solving with Parametric Segregated solutions, so it would be correct to solve each application separately and independently, one by one? In this manner, would my final solution be the one after I solve part 3 (moving mesh ale)? I'm looking for the deflection of a thin membrane in an applied electric field (similar to 3D cantilever beam), so would this final solution of the deflection be correctly found by using Method B after finally solving for the moving mesh ALE?
Thanks for your help, as always.
Best regards,
Kevin
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
as I said if the sequence is done only once it cannot converge in case of b), but if its looping around until the final sequence is fully converging the solution b) is workng and is the simples as only one physics is used per step
do you agree ?
--
Good luck
Ivar
as I said if the sequence is done only once it cannot converge in case of b), but if its looping around until the final sequence is fully converging the solution b) is workng and is the simples as only one physics is used per step
do you agree ?
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Okay, that makes sense because method b) only uses one physics application per step, so it would necessarily be the simplest and require the least memory, etc.
But how can I program this solver to solve in loops until the solution converges? Would I use the Solver Manager dialog box and solve 1, 2 and 3 each independently and store their solutions, somehow? Then run it through a loop?
I've never done this procedure of solving in loops.
Kevin
But how can I program this solver to solve in loops until the solution converges? Would I use the Solver Manager dialog box and solve 1, 2 and 3 each independently and store their solutions, somehow? Then run it through a loop?
I've never done this procedure of solving in loops.
Kevin
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
in v4 it sets iitself up and goes by itself, but I'm no longer sure how it was exactly in v3.5a.
By the way the latest patch v4.1.0.112 is in my view now better and as stable as 3.5a apart perhaps from the still missing "report", but I save the models as "m" files for my tracing, if you do a reset before storing you can use a string compare utility between versions to identify changes
--
Good luck
Ivar
in v4 it sets iitself up and goes by itself, but I'm no longer sure how it was exactly in v3.5a.
By the way the latest patch v4.1.0.112 is in my view now better and as stable as 3.5a apart perhaps from the still missing "report", but I save the models as "m" files for my tracing, if you do a reset before storing you can use a string compare utility between versions to identify changes
--
Good luck
Ivar