Dear Elmer community,
Unfortunately, I have no previous experience with Elmer, but in the course of my phD thesis I came up with a challenge that I hopefully can handle with Elmer. So I would liketo describe my task and ask the experienced Elmer users if this can be done with Elmer in principle. If I know that this can be done, then I can start highly motivated to learn all the details needed.
So here's what I would like to do: For my overall project I would like to investigate the propagation of different sources of vibration via structure-borne noise. One of these vibration sources is a double-fed induction generator, so I would like to know what kind of excitation forces or other excitation mechanisms originate from the generator in nominal operation. In similar projects 2D transient electromagnetic simulations were performed, mostly using Ansys Maxwell 2D or something similar. Since I myself am a mechanical engineer, unfortunately I have hardly any electromagnetic experience to build on. Generally speaking, I would like to calculate the forces acting on the structure by the generator operation, so that I can calculate a frequency spectrum of this excitation at the end, because that is what I need for my actual structure-borne sound simulation.
It would be really great for me, if someone could confirm that my plan could be done in principle with ELMER. The biggest help I could wish for would be a basic tip about a possible approach, which solver or maybe a thematically fitting tutorial or something similar. What I have found are this tutorials ( https://www.researchgate.net/publicatio ... e_Tutorial and https://www.researchgate.net/profile/Pa ... -Elmer.pdf ) but since my Elmer and electromagnetic understanding isn't that broad yet, I am not sure whether theese approaches covers fully what I need to get the vibration excitation due to an doubly-fed induction machine. So I am really really thankful for every kind of input.
I'm really excited and looking forward to the answers.
Thank you very much and best regards,
Marc
Vibration Excitation due to an double-fed indcution machine
-
Marceltov14
- Posts: 3
- Joined: 21 Feb 2019, 15:51
- Antispam: Yes
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Vibration Excitation due to an double-fed indcution machine
Hi Marc
I guess you are targeting 2D problems. Hence the tutorials by Pavel offer valid starting point. After running successfully the induction machine you can compute the nodal forces.
I would recommend only hierarchic coupling to elastic displacements within same transient simulation. Such coupled simulations can be done with Elmer. Can you make the same 2D assumption for elasticity solver?
Note that you may require quite many timesteps (~1 deg/timestep). If you're looking to accurately model the higher harmonics this may blow up the simulation time.
When you have vibrations you could make a FFT for them and use them as boundary sources for acoustic simulations. Or perhaps solve directly a transient wave equation.
Yes, this could probably be done in 2D. It is not an easy problem. Starting from scratch you could easily spend a few months on it. The full workflow has not been demonstrated. There are tests going from induction machine to nodal forces. Nodal forces to displacement. And harmonic displacements to pressure field of Helmholtz solver. In 3D the computational cost and possible challenges in convergence raise the bar even higher.
If I was given this problem I would work out the workflow from some simple (non-rotating) eloctromechanics toy model to the acoustic fields, and only then extend it to the 2D machine. It would also help if you had a person familiar with electrical machine design to guide you.
-Peter
I guess you are targeting 2D problems. Hence the tutorials by Pavel offer valid starting point. After running successfully the induction machine you can compute the nodal forces.
I would recommend only hierarchic coupling to elastic displacements within same transient simulation. Such coupled simulations can be done with Elmer. Can you make the same 2D assumption for elasticity solver?
Note that you may require quite many timesteps (~1 deg/timestep). If you're looking to accurately model the higher harmonics this may blow up the simulation time.
When you have vibrations you could make a FFT for them and use them as boundary sources for acoustic simulations. Or perhaps solve directly a transient wave equation.
Yes, this could probably be done in 2D. It is not an easy problem. Starting from scratch you could easily spend a few months on it. The full workflow has not been demonstrated. There are tests going from induction machine to nodal forces. Nodal forces to displacement. And harmonic displacements to pressure field of Helmholtz solver. In 3D the computational cost and possible challenges in convergence raise the bar even higher.
If I was given this problem I would work out the workflow from some simple (non-rotating) eloctromechanics toy model to the acoustic fields, and only then extend it to the 2D machine. It would also help if you had a person familiar with electrical machine design to guide you.
-Peter
-
Marceltov14
- Posts: 3
- Joined: 21 Feb 2019, 15:51
- Antispam: Yes
Re: Vibration Excitation due to an double-fed indcution machine
Hi Peter,
thank you very much for your quick and helpful answer. You are totally right, I am targeting 2D electromechanical simulation.
Regarding your words, I think working with the tutorials offered by Pavel will be one of my next steps in this case.
I am not sure, whether my initial post was quite clear: My project scope for an elmer model contains the electromechanical model to evaluate the forces acting on the rotor and stator due to the electromechanical operation. The "rest" of the workflow, dynamic structure-borne sound and acoustic radiation is implemented in an other modeling step which is already implemented in my model, which will be not connected to this elmer project. The elmer model i am hoping for, should just provide dynamic load information in terms of load spectra etc.. Therefor, I "only" need the exciation coming from the generator during operation in form of force spectra or something else.
So for this approach, did i get you right, that you would couple the electromechanics with the elastic solver? For this purpose I would definitely be fine with an 2D representation of the relevant structural element rotor and stator. I am interest in resulting vibration excitations upto 2000 Hz. Thats the range where my project experience the most interesting effects.
In othe words, I am mostly in need for the step "indcution machine to nodal forces". I am note quite sure yet, in which extent the step "nodal forces to displacement" is significant for the typical induction machine excitation characteristics. Since I a already implemented the structural behaviour of the induction machine in my "struture-borne" sound model, maybe I do not need them in this simulation. Otherwise, maybe the displacements effects the electromechanical behaviour and have influence to the electromechanical excitation effects. Or was your advise for hierachic coupling meant, that I should model an 1-way couling (electro-mechanics influence displacement, but displacment have no influence on electromechanics) or would you recommend 2-way coupling (both sides, electromechanics and displacments, influence the other side).
With the new specification of what I am aiming at, are you still the same level of optimistic that this could be done with Elmer, or maybe a little bit more or less?
I would like to thank you again. I hope I was able to make my thougth at least a little bit understandable and I would be really thankful for another hint and advise from you in the topics I just mentioned. Thank you very much and have nice rest of the weekend and a nice start to the new week ahead.
Best regards,
Marc
thank you very much for your quick and helpful answer. You are totally right, I am targeting 2D electromechanical simulation.
Regarding your words, I think working with the tutorials offered by Pavel will be one of my next steps in this case.
I am not sure, whether my initial post was quite clear: My project scope for an elmer model contains the electromechanical model to evaluate the forces acting on the rotor and stator due to the electromechanical operation. The "rest" of the workflow, dynamic structure-borne sound and acoustic radiation is implemented in an other modeling step which is already implemented in my model, which will be not connected to this elmer project. The elmer model i am hoping for, should just provide dynamic load information in terms of load spectra etc.. Therefor, I "only" need the exciation coming from the generator during operation in form of force spectra or something else.
So for this approach, did i get you right, that you would couple the electromechanics with the elastic solver? For this purpose I would definitely be fine with an 2D representation of the relevant structural element rotor and stator. I am interest in resulting vibration excitations upto 2000 Hz. Thats the range where my project experience the most interesting effects.
In othe words, I am mostly in need for the step "indcution machine to nodal forces". I am note quite sure yet, in which extent the step "nodal forces to displacement" is significant for the typical induction machine excitation characteristics. Since I a already implemented the structural behaviour of the induction machine in my "struture-borne" sound model, maybe I do not need them in this simulation. Otherwise, maybe the displacements effects the electromechanical behaviour and have influence to the electromechanical excitation effects. Or was your advise for hierachic coupling meant, that I should model an 1-way couling (electro-mechanics influence displacement, but displacment have no influence on electromechanics) or would you recommend 2-way coupling (both sides, electromechanics and displacments, influence the other side).
With the new specification of what I am aiming at, are you still the same level of optimistic that this could be done with Elmer, or maybe a little bit more or less?
I would like to thank you again. I hope I was able to make my thougth at least a little bit understandable and I would be really thankful for another hint and advise from you in the topics I just mentioned. Thank you very much and have nice rest of the weekend and a nice start to the new week ahead.
Best regards,
Marc
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Vibration Excitation due to an double-fed indcution machine
Hi Marc
Having another code do part of the work does not change much the overall conclusion. I guess it would then suffice if you get time dependent nodal (or disctributed) forces. That could involve minor coding or scripting to get those passed to your vibration code.
If you're targeting 2000 Hz, I guess your timesteps should be around 1/20000 s or less. The number of timesteps over a few cycles could then be quite large.
Basically there could be a feedback from displacement fields due to magnetostriction or changes in geometry. However, I would leave those out if that's not particularly your line of research. They should have minor effect anyways.
-Peter
Having another code do part of the work does not change much the overall conclusion. I guess it would then suffice if you get time dependent nodal (or disctributed) forces. That could involve minor coding or scripting to get those passed to your vibration code.
If you're targeting 2000 Hz, I guess your timesteps should be around 1/20000 s or less. The number of timesteps over a few cycles could then be quite large.
Basically there could be a feedback from displacement fields due to magnetostriction or changes in geometry. However, I would leave those out if that's not particularly your line of research. They should have minor effect anyways.
-Peter
-
Marceltov14
- Posts: 3
- Joined: 21 Feb 2019, 15:51
- Antispam: Yes
Re: Vibration Excitation due to an double-fed indcution machine
Hi Peter,
thank you very much for your input and advise. I think I have to try my best with this challenging task and look where does it get me.
Thanks for your help.
BR,
Marc
thank you very much for your input and advise. I think I have to try my best with this challenging task and look where does it get me.
Thanks for your help.
BR,
Marc