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.
Influence of weak constraints on displacement ? Fixed-Fixed beam problem
Posted 30 dic 2010, 23:43 GMT-5 3 Replies
Please login with a confirmed email address before reporting spam
Dear All,
I have a simple 2D fixed beam of i =1000um, b = 10um and h = 50um. While computing the eigenfrequency with the weak constraints as ideal or non ideal, I get the correct displacement in terms of nm. But if I switch off the weak constraints option I get unrealisitic values like 2.2m for the displacement at resonance.
Could anyone pls explain why this happens? When to select weak constraints and when not to select? I am not so conversant with the math behind FEM. So pls explain it in layman terms.
Best regards,
Ashwin Simha.
I have a simple 2D fixed beam of i =1000um, b = 10um and h = 50um. While computing the eigenfrequency with the weak constraints as ideal or non ideal, I get the correct displacement in terms of nm. But if I switch off the weak constraints option I get unrealisitic values like 2.2m for the displacement at resonance.
Could anyone pls explain why this happens? When to select weak constraints and when not to select? I am not so conversant with the math behind FEM. So pls explain it in layman terms.
Best regards,
Ashwin Simha.
3 Replies Last Post 6 gen 2011, 06:31 GMT-5
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
this is not related to the weak constraints, in my view. An eigenfrequency analysis does not have any normalsation fixing the amplitude: 2m just as 2km (even for a 1mm long part) are all valid replies, it depends on the numerical normalisation used. This is also to be understood because an eigenfrequency analysis is not bound on input energy, nor in damping.
In the latest 4.1 version you can select different normalisation modes (eigenfrequency solver sub node) and select the rms or mass participation factor normalisation. The later normalises the different modes such that their square max amplitude adds up to the total mass, this is very handy as you can check if you are considering sufficient modes, or if you are forgetting some important ones. Unfortunately with COMSOL today, only displacement modes are given (u,v,w), and not (yet?) the three rotational modes (for which the mass participation factor adds up to the inertia terms)
If you want to look at amplitudes you can use an frequency sweep with an fixed excitation amplitude and some damping to dissipate energy, but often the latter (structural damping) is a delicate issue and little good material or fixing related data exist
--
Good luck
Ivar
this is not related to the weak constraints, in my view. An eigenfrequency analysis does not have any normalsation fixing the amplitude: 2m just as 2km (even for a 1mm long part) are all valid replies, it depends on the numerical normalisation used. This is also to be understood because an eigenfrequency analysis is not bound on input energy, nor in damping.
In the latest 4.1 version you can select different normalisation modes (eigenfrequency solver sub node) and select the rms or mass participation factor normalisation. The later normalises the different modes such that their square max amplitude adds up to the total mass, this is very handy as you can check if you are considering sufficient modes, or if you are forgetting some important ones. Unfortunately with COMSOL today, only displacement modes are given (u,v,w), and not (yet?) the three rotational modes (for which the mass participation factor adds up to the inertia terms)
If you want to look at amplitudes you can use an frequency sweep with an fixed excitation amplitude and some damping to dissipate energy, but often the latter (structural damping) is a delicate issue and little good material or fixing related data exist
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello Ivar,
Many thanks. But its still not very clear to me when to use weak constraints and when not to do so. Here is a summary of the results in short
WEAK CONSTRAINTS = OFF : z-displacement = 2.2 m
WEAK CONSTRAINTS = IDEAL/NON-IDEAL : z-displacement = 9.9um --> which looks realistic for the dimensions stated in my previous post (1000uX50uX10u beam).
Why does such a deviation exist? I have left the mass damping factor and stiffness damping factor (used in structural damping) at the default values in the subdomain settings.
Regards,
Ashwin Simha.
Many thanks. But its still not very clear to me when to use weak constraints and when not to do so. Here is a summary of the results in short
WEAK CONSTRAINTS = OFF : z-displacement = 2.2 m
WEAK CONSTRAINTS = IDEAL/NON-IDEAL : z-displacement = 9.9um --> which looks realistic for the dimensions stated in my previous post (1000uX50uX10u beam).
Why does such a deviation exist? I have left the mass damping factor and stiffness damping factor (used in structural damping) at the default values in the subdomain settings.
Regards,
Ashwin Simha.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
perhaps it looks realistic, but pls verify how you excite, an eigenfrequency analysis is not energy input limited, hence 2m or 2um are just as good results, it's a question of normalization, pure numerical values. You are obviously changing the normalization with the weak constraints. Apart if you have something else in there.
In V4 you can select the type of normalization and i.e. use the participation mass values that gives you in-site in the type of mode and their RELATIVE energy content per DoF
--
Good luck
Ivar
perhaps it looks realistic, but pls verify how you excite, an eigenfrequency analysis is not energy input limited, hence 2m or 2um are just as good results, it's a question of normalization, pure numerical values. You are obviously changing the normalization with the weak constraints. Apart if you have something else in there.
In V4 you can select the type of normalization and i.e. use the participation mass values that gives you in-site in the type of mode and their RELATIVE energy content per DoF
--
Good luck
Ivar