Ivar KJELBERG
COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Please login with a confirmed email address before reporting spam

Posted:
1 decade ago
Mar 15, 2010, 11:22 a.m. EDT

Hi

For me you have two questions in there:

1) change the temperature at given times: yes why not try

T = T0*(t<t0)+T1*(t>=t0)*(t<t1)*...*Tn*(t>=tn)

or otherwise simpler: define a piecevise function fT(t) and say T=fT(t)

2) couple the temperature between two items, or model at given times

Well why at given times ? why not express a continously linked physics and solve them together ?

There are a few examples in the doc about the weak constraints and boundary integrations, probably this is a good example for the non-ideal weak contraits (single way link), even an example somewhere about a thermal case with the temperature of one border linking around onto the other.

If not I would try to play with some integration variables over the domain to measure and then use the equation above for the controlled domain, but one must add some "*at(ti)" where "ti" are some of the PREVIOUS solved time steps to get it to work. But I'm not sure this (the "at()" operator) would even work during the simulation, you would have to test it out carefully, never done something like that

Hope this helps

Good luck

Ivar

Hi
For me you have two questions in there:
1) change the temperature at given times: yes why not try
T = T0*(t=t0)*(t=tn)
or otherwise simpler: define a piecevise function fT(t) and say T=fT(t)
2) couple the temperature between two items, or model at given times
Well why at given times ? why not express a continously linked physics and solve them together ?
There are a few examples in the doc about the weak constraints and boundary integrations, probably this is a good example for the non-ideal weak contraits (single way link), even an example somewhere about a thermal case with the temperature of one border linking around onto the other.
If not I would try to play with some integration variables over the domain to measure and then use the equation above for the controlled domain, but one must add some "*at(ti)" where "ti" are some of the PREVIOUS solved time steps to get it to work. But I'm not sure this (the "at()" operator) would even work during the simulation, you would have to test it out carefully, never done something like that
Hope this helps
Good luck
Ivar

Please login with a confirmed email address before reporting spam

Posted:
1 decade ago
Mar 16, 2010, 8:41 a.m. EDT

Thanks, I will give it a shot.

The reason I don't want to continuously couple them is because I am modelling a "pseudo-assembly line", where the "subjects" move at a set time. Basically, the model has warm boxes placed at one point, and then these move down the line. And I want to investigate the heat transfer around them. I thought about a moving mesh for this, but with 9 moving boxes I thought this might be too many calculations, and I thought this way might be simpler.

Thanks, I will give it a shot.
The reason I don't want to continuously couple them is because I am modelling a "pseudo-assembly line", where the "subjects" move at a set time. Basically, the model has warm boxes placed at one point, and then these move down the line. And I want to investigate the heat transfer around them. I thought about a moving mesh for this, but with 9 moving boxes I thought this might be too many calculations, and I thought this way might be simpler.

Please login with a confirmed email address before reporting spam

Posted:
1 decade ago
Mar 17, 2010, 6:57 a.m. EDT

Where would you enter the piecewise definition? I entered "80*(t<60) + 900*(t>=60)" (as a simple example) in the Subdomain Expression field for the relevant subdomain, and the solver produced a duplicate variable error for T.

Where would you enter the piecewise definition? I entered "80*(t=60)" (as a simple example) in the Subdomain Expression field for the relevant subdomain, and the solver produced a duplicate variable error for T.