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.
Applying no-standard boundary loads in structural mechanics
Posted 28.11.2015, 08:27 GMT-5 Structural Mechanics Version 5.1 7 Replies
Please login with a confirmed email address before reporting spam
Dear COMSOL community,
I would like to perform a stress analysis on a Synchronous Reluctance Machine in order to investigate the stress and displacement on the iron bridges. The difference in my study with respect to the "standard" stress analysis is that I need to impose boundary layer load which are not standard, besides the trivial rotating frame load which takes into account for the centrifugal force.
In my case the forces I need to impose as load have been computed analytically in an eccentricity study in terms of magnetic forces operating on the rotor at the airgap. Therefore now I need to take the force computed (which is a numerical function with a force value for each angle along the whole periphery of the rotor) and impose it as a boundary load.
Is there a way to establish a boundary load from a set of values? Do I need to split the boundary using several nodes and put one by one my load point by point or there is a way?
I was also thinking to decompose in Fourier series my numerical set of forces along the rotor periphery and assign an analytical expression in terms of Fourier series using the Fourier coefficient. But, what if I have too many coefficients in order to be precise in defining the Boundary load?
I thank you in advance if you have advices about how I could perform this.
I would be really grateful.
Sincerely,
Nicola
I would like to perform a stress analysis on a Synchronous Reluctance Machine in order to investigate the stress and displacement on the iron bridges. The difference in my study with respect to the "standard" stress analysis is that I need to impose boundary layer load which are not standard, besides the trivial rotating frame load which takes into account for the centrifugal force.
In my case the forces I need to impose as load have been computed analytically in an eccentricity study in terms of magnetic forces operating on the rotor at the airgap. Therefore now I need to take the force computed (which is a numerical function with a force value for each angle along the whole periphery of the rotor) and impose it as a boundary load.
Is there a way to establish a boundary load from a set of values? Do I need to split the boundary using several nodes and put one by one my load point by point or there is a way?
I was also thinking to decompose in Fourier series my numerical set of forces along the rotor periphery and assign an analytical expression in terms of Fourier series using the Fourier coefficient. But, what if I have too many coefficients in order to be precise in defining the Boundary load?
I thank you in advance if you have advices about how I could perform this.
I would be really grateful.
Sincerely,
Nicola
7 Replies Last Post 02.12.2015, 10:47 GMT-5
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hi Nicola
I do not see any issue, as in COMSOL any field in the GUI can accept a value, a Parameter a Scalar or a full field over f(x,y,z) or even an equation linked to something the user defines in a global or local equation/constraint ...
So if you have a phenomenological equation, or a look up table from some measurements, you can include these and refer to them. The only issue might be that you would need another reference frame, i.e. a cylindrical for a rotary machine, bit that you define in the Definitions - Coordinate Systems. and you get also all the matrix values fro the tensor transformation back and forth (check the equation view)
There are a few examples in the application library, take a look
--
Good luck
Ivar
I do not see any issue, as in COMSOL any field in the GUI can accept a value, a Parameter a Scalar or a full field over f(x,y,z) or even an equation linked to something the user defines in a global or local equation/constraint ...
So if you have a phenomenological equation, or a look up table from some measurements, you can include these and refer to them. The only issue might be that you would need another reference frame, i.e. a cylindrical for a rotary machine, bit that you define in the Definitions - Coordinate Systems. and you get also all the matrix values fro the tensor transformation back and forth (check the equation view)
There are a few examples in the application library, take a look
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Dear Ivan,
thanks a lot. So let's say I have a look up table in which I possess force load values point by point along the periphery of my rotor both in the radial and azimuthal direction (I defined a cylindrical coordinate system in r and phi). I imagine I have to exploit the global parameters definition and the interpolation function. Now do I have to define one different interpolation function for each component: radial and tangential in my case, as int1 will be f_r(phi) and int2 will be f_t(phi), and then call them in the boundary force loading frame?
Did I understand correctly? Or there is a way to include both the tangential component and radial component as function of the phi angle as a unique global variable as interpolation for calling it once defined?
Sincerely,
Nicola
thanks a lot. So let's say I have a look up table in which I possess force load values point by point along the periphery of my rotor both in the radial and azimuthal direction (I defined a cylindrical coordinate system in r and phi). I imagine I have to exploit the global parameters definition and the interpolation function. Now do I have to define one different interpolation function for each component: radial and tangential in my case, as int1 will be f_r(phi) and int2 will be f_t(phi), and then call them in the boundary force loading frame?
Did I understand correctly? Or there is a way to include both the tangential component and radial component as function of the phi angle as a unique global variable as interpolation for calling it once defined?
Sincerely,
Nicola
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hi
well I might not have understood you question correctly either ;)
But if you have two interpolation functions (r: radial and t: tangential): f_r(phi) and f_t(phi) you can define a boundary load in Cylindrical coordinates, with both functions, something like on the image attached
--
Good luck
Ivar
well I might not have understood you question correctly either ;)
But if you have two interpolation functions (r: radial and t: tangential): f_r(phi) and f_t(phi) you can define a boundary load in Cylindrical coordinates, with both functions, something like on the image attached
--
Good luck
Ivar
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Dear Ivar,
very thanks. Yes that was exactly what I meant. Right now I am coping with another issue. Or maybe it's not as in the beginning of my thread. Nevertheless my doubt is the following:
I load my model only with centrifugal forces using Rotating body frame. Imposing different speeds in rpm I get different values of Von Mises stress mapped all over the domain. However the solid displacement seems to be exactly the same. As you can see from the figures attached at 10 rpm and 1000 rpm the displacement of the iron rotating structure is exactly represented in the same way in terms of dimensions.
Does this representation have to be trusted? To be more precise I am doing an elastic simulation only, therefore the displacement should show some kind of proportionality with respect to the rotating velocity applied in my poor opinion. Is there a way to get the maximum displacement out of the COMSOL result after the structural mechanics simulation?
Sincerely,
Nicola
very thanks. Yes that was exactly what I meant. Right now I am coping with another issue. Or maybe it's not as in the beginning of my thread. Nevertheless my doubt is the following:
I load my model only with centrifugal forces using Rotating body frame. Imposing different speeds in rpm I get different values of Von Mises stress mapped all over the domain. However the solid displacement seems to be exactly the same. As you can see from the figures attached at 10 rpm and 1000 rpm the displacement of the iron rotating structure is exactly represented in the same way in terms of dimensions.
Does this representation have to be trusted? To be more precise I am doing an elastic simulation only, therefore the displacement should show some kind of proportionality with respect to the rotating velocity applied in my poor opinion. Is there a way to get the maximum displacement out of the COMSOL result after the structural mechanics simulation?
Sincerely,
Nicola
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Dear Jeff,
thanks. I have another problem right now. In particular I would like to perform a sweep parametric analysis changing for each "sweep step" the loads I am applying. For each one of them I would like to compute the maximum displacement and store it. I can see the displacement, using the rescaling, but I cannot find a way to get it out of post processing. How can I do that? Do you advice me some pre-made comsol model available online. I cannot find something which is simular to what I would like to do unfortunately.
Sincerely,
Nicola
thanks. I have another problem right now. In particular I would like to perform a sweep parametric analysis changing for each "sweep step" the loads I am applying. For each one of them I would like to compute the maximum displacement and store it. I can see the displacement, using the rescaling, but I cannot find a way to get it out of post processing. How can I do that? Do you advice me some pre-made comsol model available online. I cannot find something which is simular to what I would like to do unfortunately.
Sincerely,
Nicola
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hi Nicola,
If it's solely for postprocessing purposes: use Derived Values.
If it's for pre-processing as well: use a Component coupling.
Best,
Jeff
PS: To keep the Discussion Forum orderly and easier for others to search, we ask that you please create a new thread when you move on to a different question.
If it's solely for postprocessing purposes: use Derived Values.
If it's for pre-processing as well: use a Component coupling.
Best,
Jeff
PS: To keep the Discussion Forum orderly and easier for others to search, we ask that you please create a new thread when you move on to a different question.