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.
[Comsol 3.5a Structural movements] problem to obtain correct displacements
Posted 20.10.2009, 04:34 GMT-4 1 Reply
Please login with a confirmed email address before reporting spam
Hello to all
I try to impose a displacement to a structural block but I have some difficulties to obtain correct displacements. My block has the left bottom corner with coordinate of (0,0) and the block height is of 30 cm.
I declare the constant A=5 cm, after I insert the function, in scalar expression, that is s=sin(2*pi*t[1/s]). Finally i open subdomain setting, i set prescribed displacement, and in this window i write: A*s*(30[cm]-y[cm])/(30). In this way i should obtain a block rotation where the rotational axis is on the top of the block and the max amplitude of rotation is on the bottom of the block, but when i display the results, i find a rotation of the block, with a sinusoidal aspect but the maximum value of displacement of the bottom part is not 5 cm but less about 3.5cm(the length of simulation is 1 second and the time step is 0.01 second).
The second question is about the correct use of dimensionless parameter, for example if I want a displacement I create the constant A and I assign the value 5[cm], so this is my max amplitude displacement, but I want also a sinusoidal displacement and, to create it, i must write sin(2*pi*t).Now I put all in subdomain setting, but to obtain a correct unit of measurement(that is a length) i must put[1/s] after t, in other way comsol underlines to me the unit in red. My question is :is it [1/t] the correct way to exploit time development without the variation of measurement unit? and more in general is the use of [1/ ] the correct way to exploit the variation of dimensional parameter without trouble in the unit change?
Thank you
Luca
I try to impose a displacement to a structural block but I have some difficulties to obtain correct displacements. My block has the left bottom corner with coordinate of (0,0) and the block height is of 30 cm.
I declare the constant A=5 cm, after I insert the function, in scalar expression, that is s=sin(2*pi*t[1/s]). Finally i open subdomain setting, i set prescribed displacement, and in this window i write: A*s*(30[cm]-y[cm])/(30). In this way i should obtain a block rotation where the rotational axis is on the top of the block and the max amplitude of rotation is on the bottom of the block, but when i display the results, i find a rotation of the block, with a sinusoidal aspect but the maximum value of displacement of the bottom part is not 5 cm but less about 3.5cm(the length of simulation is 1 second and the time step is 0.01 second).
The second question is about the correct use of dimensionless parameter, for example if I want a displacement I create the constant A and I assign the value 5[cm], so this is my max amplitude displacement, but I want also a sinusoidal displacement and, to create it, i must write sin(2*pi*t).Now I put all in subdomain setting, but to obtain a correct unit of measurement(that is a length) i must put[1/s] after t, in other way comsol underlines to me the unit in red. My question is :is it [1/t] the correct way to exploit time development without the variation of measurement unit? and more in general is the use of [1/ ] the correct way to exploit the variation of dimensional parameter without trouble in the unit change?
Thank you
Luca
1 Reply Last Post 20.10.2009, 09:25 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi Luca
My understanding (and use) if the [1/..] is to ensure correct units, and to cancel out unwanted unit dimensions, as you pointed out. If "t" is defined in [min] then t[min]*[1/sec] => t*60[1] (without units) and that is what I would use within a sinus function.
That COMSOL flags something red in V3.5 does not necesarily mean an error, neither than it will calculate wrongly, it's just a warning of possble incoherence in units. Among other integration coupling constants do not have their units following them correctly, I expect this to be better in the new V4, coming soon.
Now back to your second calculation of rotation, if you are in 2D (I assume so) normally "s" is defined internally, s is going from 0-1 along an adge, but the direction is not obvious, check the edge "arrow" in edge view. You might have nameing conflict there, s has no unit =[1] , I would call your sinus function something else perhaps "si".
Finally your expression "*(30[cm]-y[cm])/(30)" is this "*(1-y/(30[cm]))" you mean ?, which for me is dimensionless but different to the first expression (I assume "y" is a length related someway to [m] this might be wrong, see below).
The easiest way to check I know about, is to set up, in 2D structural, a rectangular geometry block of 1[m] side, ands olve as is, all defaults.
Then postprocess on an edge and type "y", "y[cm]" and finally "y[1/cm]", check the value and the dimensions of the reply.
Mostly you will find the replies within COMSOL, and one must first "isolate" the issues to something unambiguous and simple.
So I assume there might be a few items that could explain the "red" flag of COMSOL
Good luck
Ivar
My understanding (and use) if the [1/..] is to ensure correct units, and to cancel out unwanted unit dimensions, as you pointed out. If "t" is defined in [min] then t[min]*[1/sec] => t*60[1] (without units) and that is what I would use within a sinus function.
That COMSOL flags something red in V3.5 does not necesarily mean an error, neither than it will calculate wrongly, it's just a warning of possble incoherence in units. Among other integration coupling constants do not have their units following them correctly, I expect this to be better in the new V4, coming soon.
Now back to your second calculation of rotation, if you are in 2D (I assume so) normally "s" is defined internally, s is going from 0-1 along an adge, but the direction is not obvious, check the edge "arrow" in edge view. You might have nameing conflict there, s has no unit =[1] , I would call your sinus function something else perhaps "si".
Finally your expression "*(30[cm]-y[cm])/(30)" is this "*(1-y/(30[cm]))" you mean ?, which for me is dimensionless but different to the first expression (I assume "y" is a length related someway to [m] this might be wrong, see below).
The easiest way to check I know about, is to set up, in 2D structural, a rectangular geometry block of 1[m] side, ands olve as is, all defaults.
Then postprocess on an edge and type "y", "y[cm]" and finally "y[1/cm]", check the value and the dimensions of the reply.
Mostly you will find the replies within COMSOL, and one must first "isolate" the issues to something unambiguous and simple.
So I assume there might be a few items that could explain the "red" flag of COMSOL
Good luck
Ivar