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
Apr 17, 2011, 3:36 a.m. EDT
Hi
if you search the Forum, you will find different explanations around these facts, in 2 word (or a bit more;):
all functions wants unitless variable calls, "t" the time has a unit of seconds, to render it unitless we write t[1/s] and expressed in seconds (this will cancel the "orange writing" warning signal rom COMSOL
"s" is an internal variable defined in 2D and running normally from 0 to 1 along EACH Edge=Boundary in 2D.
It also defines the direction of the edge/boundary see the arrows in edge mode. THe direction is important, as the boundary normal (up-normal) points OUTWARDS from the Domain w.rt. the right hand rule i.e. a vertical Edge with an arrow up (+Y) has its normal to the left (-X) direction hence the Domain it belongs to is to the right !
Then s*(1-s) is a parabola starting and finishing at "0" for s=0 and s=1, this is tyspically the velocity flow of a "non-slip" wall condition. If you integrate s*(1-s) between 0 and 1 you will get 1/6 so U_mean*6*s*(1-s)*step1(t[1/s]) gives you the average velocity at step=1.
THe maximum of s*(1-s) is for s=1/2 with a value of 1/4, hence U_peak*4*s*(1-s)*step1(t[1/s]) is the expression for the o-peak or amplitude value of the velocity, when you know thta one.
Note: in 3D it becomes less evident, "s" no longer exist, it's "s1 s2", they do no longer go from 0 to1 as this depends on the boundary shape. In 3D you must revert to normalised absolute values to express your velocity. Often using the cylindrical coordinates eases the expression.
Finally, I (almost) always add inital conditions for my flow, and always start with a little flow, even if the first step might be with v=0, AS WELL as I add alittle pressure drop to initialise the vectors, this has saved me hours of time waiting for the solver to converge, so I find it very useful.
To check your initial conditions, as these become rapidly complex forskewed geoemtries, you can always right-click on the Solver ... Dependent Variables node and select Compute to Selected. This will load the initial conditions from its default values (or a stored solutionif you chain solvers) and then you have a "Data Set" to probe and analse
Take a look at the models and the explanations in the doc, AND in the Model Library (do not forget to update it with the "update command"
--
Have fun Comsoling
Ivar
Hi
if you search the Forum, you will find different explanations around these facts, in 2 word (or a bit more;):
all functions wants unitless variable calls, "t" the time has a unit of seconds, to render it unitless we write t[1/s] and expressed in seconds (this will cancel the "orange writing" warning signal rom COMSOL
"s" is an internal variable defined in 2D and running normally from 0 to 1 along EACH Edge=Boundary in 2D.
It also defines the direction of the edge/boundary see the arrows in edge mode. THe direction is important, as the boundary normal (up-normal) points OUTWARDS from the Domain w.rt. the right hand rule i.e. a vertical Edge with an arrow up (+Y) has its normal to the left (-X) direction hence the Domain it belongs to is to the right !
Then s*(1-s) is a parabola starting and finishing at "0" for s=0 and s=1, this is tyspically the velocity flow of a "non-slip" wall condition. If you integrate s*(1-s) between 0 and 1 you will get 1/6 so U_mean*6*s*(1-s)*step1(t[1/s]) gives you the average velocity at step=1.
THe maximum of s*(1-s) is for s=1/2 with a value of 1/4, hence U_peak*4*s*(1-s)*step1(t[1/s]) is the expression for the o-peak or amplitude value of the velocity, when you know thta one.
Note: in 3D it becomes less evident, "s" no longer exist, it's "s1 s2", they do no longer go from 0 to1 as this depends on the boundary shape. In 3D you must revert to normalised absolute values to express your velocity. Often using the cylindrical coordinates eases the expression.
Finally, I (almost) always add inital conditions for my flow, and always start with a little flow, even if the first step might be with v=0, AS WELL as I add alittle pressure drop to initialise the vectors, this has saved me hours of time waiting for the solver to converge, so I find it very useful.
To check your initial conditions, as these become rapidly complex forskewed geoemtries, you can always right-click on the Solver ... Dependent Variables node and select Compute to Selected. This will load the initial conditions from its default values (or a stored solutionif you chain solvers) and then you have a "Data Set" to probe and analse
Take a look at the models and the explanations in the doc, AND in the Model Library (do not forget to update it with the "update command"
--
Have fun Comsoling
Ivar