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.

Design optimization for iron loss minimization in induction machine (IM)

Please login with a confirmed email address before reporting spam

Hi all,
I have a model of an induction machine, which is attached in the file. In this model, besides stator and rotor laminations there are still holes located in the stator lamination that I try to make to change the rotational iron loss in the machine. These holes may be for ventilation system. Assuming that, across area of holes is kept constant. But the shape and position of these holes will influence the magnetic flux density as well as magnetic field strenght, then resulting changes on iron losses of the stator core. Therefore, to find an optimal position, we need an optimization method in the comsol model.

Based on this, I have some questions that I need your help.
1- How to build up a iron loss model (or hysteresis loss model) in the induction motor model? So far I do not know how to obtain iron loss distribution in the model. I want to compare iron loss distribution with and without holes.
2- Is it possible to adopt an optimization method to obtain optimal location of holes, which offers a minimum iron loss. Genetic algorithm can be used or developed in Comsol? Is there any references about this?

I believe that there are some guys interested in this problem and having a lots experiences on it with Comsol. I appreciate your help or useful comments that we can learn together.
Thank you so much.

Best,
TUNG


13 Replies Last Post 14.03.2015, 06:34 GMT-4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27.10.2011, 10:39 GMT-4
Hi Phan Van Tung:

I am also interest in this kind of problem. But I do not know much in this area. So I can only provide some superficial opinion.

First, however, I cannot open your file, neither 4.2 nor 3.5a.

Second, about the iron loss. What is your study type? Frequency domain?
I think in comsol there is not a term represent the loss. But it may be add a body heat source which is a function of the local B and frequency. This function can be pre-calculated or from experiment. (I am not very sure)

Vincent
Hi Phan Van Tung: I am also interest in this kind of problem. But I do not know much in this area. So I can only provide some superficial opinion. First, however, I cannot open your file, neither 4.2 nor 3.5a. Second, about the iron loss. What is your study type? Frequency domain? I think in comsol there is not a term represent the loss. But it may be add a body heat source which is a function of the local B and frequency. This function can be pre-calculated or from experiment. (I am not very sure) Vincent

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27.10.2011, 11:15 GMT-4
Hi Vincent,
I also dont know why it is impossible to open with 4.2. I will try to create a model with 4.2 version. But it can open with 4.0 because I created it with 4.0 version. If you can open it you can see its area. To make it clear, I attach a picture of the model.

Iron loss is composed of hysteresis loss and eddy-current loss. So I think this problem may be related to hysteresis issus that can be done using AC-AC capability of Comsol. It's my idea.

Now I just try with stationary type to check magnetic flux density. How is the time dependent or frequency domain? Which should be used? As I said I don't know how to investigate iron loss distribution in the model so far. Can you say more about the body heat source that is the function of local B and frequency?

The position of holes (near or far from the shaft center) will affect magnetic flux density in the model. I want to find an optimal position to obtain a minimum iron loss in the stator core. That's something I try to do.

Thank you.

Tung
Hi Vincent, I also dont know why it is impossible to open with 4.2. I will try to create a model with 4.2 version. But it can open with 4.0 because I created it with 4.0 version. If you can open it you can see its area. To make it clear, I attach a picture of the model. Iron loss is composed of hysteresis loss and eddy-current loss. So I think this problem may be related to hysteresis issus that can be done using AC-AC capability of Comsol. It's my idea. Now I just try with stationary type to check magnetic flux density. How is the time dependent or frequency domain? Which should be used? As I said I don't know how to investigate iron loss distribution in the model so far. Can you say more about the body heat source that is the function of local B and frequency? The position of holes (near or far from the shaft center) will affect magnetic flux density in the model. I want to find an optimal position to obtain a minimum iron loss in the stator core. That's something I try to do. Thank you. Tung


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27.10.2011, 20:42 GMT-4
Hi Phan:

Sorry I did not mention in the first reply that what I said is based on the build in "inductive heating" branch and I thought iron loss you mean is hysteresis loss. Because this branch(combine magnetic and heat transfer branch) already considers eddy-current loss as Joule heat source. You can read the document of "Inductive Heating of a Copper Cylinder" in the model library.

Another point is that if you want to study how much haet produce when this induction machine work and the temperature distribution, I guess you should first use frequency -- stationary study type. Frequency is for magnetic branch -- the induction machine works under a specific frequency; stationary is for heat transfer.

Combined what I said before, comsol can solve the eddy-current problem in your model, but it has not built the hysteresis loss in. Thus you can add a body heat source density term in heat transfer to represent hysteresis loss. The hysteresis loss is related to frequency and B, higher frequency, lower loss; higher B(in frequency domain this is the max value in a period), higher loss. I suggest you read the paper on hysteresis loss because what I said is by speculating. My research direction is inductive heating -- opposite of yours and now I do not consider hysteresis loss (it is less significant to eddy-current loss in my problem), but I also hope one day I can add the hysteresis loss in my model.

Last, about optimization, I am not familiar. So I can only recommend you read the user's guild of optimization module and model in the library.


Vincent



Hi Phan: Sorry I did not mention in the first reply that what I said is based on the build in "inductive heating" branch and I thought iron loss you mean is hysteresis loss. Because this branch(combine magnetic and heat transfer branch) already considers eddy-current loss as Joule heat source. You can read the document of "Inductive Heating of a Copper Cylinder" in the model library. Another point is that if you want to study how much haet produce when this induction machine work and the temperature distribution, I guess you should first use frequency -- stationary study type. Frequency is for magnetic branch -- the induction machine works under a specific frequency; stationary is for heat transfer. Combined what I said before, comsol can solve the eddy-current problem in your model, but it has not built the hysteresis loss in. Thus you can add a body heat source density term in heat transfer to represent hysteresis loss. The hysteresis loss is related to frequency and B, higher frequency, lower loss; higher B(in frequency domain this is the max value in a period), higher loss. I suggest you read the paper on hysteresis loss because what I said is by speculating. My research direction is inductive heating -- opposite of yours and now I do not consider hysteresis loss (it is less significant to eddy-current loss in my problem), but I also hope one day I can add the hysteresis loss in my model. Last, about optimization, I am not familiar. So I can only recommend you read the user's guild of optimization module and model in the library. Vincent

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28.10.2011, 04:40 GMT-4
Hi Vincent,
Thanks a lot your useful comments about this issue. Based on what you said, I understand that the model sould be solved under multiphysics problem using magnetic and heat transfer.

Currently, I would like to investigate the influence of holes geometry and its quantity on the iron losses of the machine. After that, using Matlab to optimize objective function with GA technique. To this end, we need an integration between Matlab and Comsol, which requires much more experiences that we need to share.

The optimization method is used to obtain maximum heat flux and minimum iron losses in the model. If anyone who has exprienced some similar projects, please give me a help.

Thank you.
Tung
Hi Vincent, Thanks a lot your useful comments about this issue. Based on what you said, I understand that the model sould be solved under multiphysics problem using magnetic and heat transfer. Currently, I would like to investigate the influence of holes geometry and its quantity on the iron losses of the machine. After that, using Matlab to optimize objective function with GA technique. To this end, we need an integration between Matlab and Comsol, which requires much more experiences that we need to share. The optimization method is used to obtain maximum heat flux and minimum iron losses in the model. If anyone who has exprienced some similar projects, please give me a help. Thank you. Tung

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 28.10.2011, 06:31 GMT-4
Hi

do not forget that the COMSOL solver is an optimisation solver, so most optimisation can be done internally !

It's easier with the optimisation module, as there it's pre-cooked for you ;)

--
Good luck
Ivar
Hi do not forget that the COMSOL solver is an optimisation solver, so most optimisation can be done internally ! It's easier with the optimisation module, as there it's pre-cooked for you ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28.10.2011, 06:48 GMT-4
Hi Ivar,
With my problem, it should be used with Comsol and Matlab using genetic algorithm or optimisation module of Comsol? Is there any example model for optimisation module? May I have a detailed advice about this issue? Is it possible to use Jiles-Atherton hysteresis model to analyse iron loss in this case? I think this problem is also related to geometry optimisation.

Thank you.
Tung
Hi Ivar, With my problem, it should be used with Comsol and Matlab using genetic algorithm or optimisation module of Comsol? Is there any example model for optimisation module? May I have a detailed advice about this issue? Is it possible to use Jiles-Atherton hysteresis model to analyse iron loss in this case? I think this problem is also related to geometry optimisation. Thank you. Tung

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28.10.2011, 13:07 GMT-4
Hi all,
Is it possible to use Jiles-Atherton model to calculate or even minimize iron losses in the model of an induction motor? I would like to hear feasible advice. Thank you
Hi all, Is it possible to use Jiles-Atherton model to calculate or even minimize iron losses in the model of an induction motor? I would like to hear feasible advice. Thank you

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 28.10.2011, 15:25 GMT-4
Hi

probaly not "genetic type " optimisation, but fairly advanced optimisation, yes. Check the doc, and the optimisation module. Finally, COMSOLs mathematical "heart" is a minimising algorithm with many different solvers, no ?
--
Good luck
Ivar
Hi probaly not "genetic type " optimisation, but fairly advanced optimisation, yes. Check the doc, and the optimisation module. Finally, COMSOLs mathematical "heart" is a minimising algorithm with many different solvers, no ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 31.10.2011, 00:23 GMT-4
Thank Ivar. Your advice is useful and I will check it. But I am still studying on how integrate Matlab and Comsol so that I can apply GA technique in my application. This is due to the requirement of the project. Any helps are appreciated.
Thank you
Thank Ivar. Your advice is useful and I will check it. But I am still studying on how integrate Matlab and Comsol so that I can apply GA technique in my application. This is due to the requirement of the project. Any helps are appreciated. Thank you

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 01.11.2011, 02:11 GMT-4
Anyone knows how to display iron loss in the model? Thank you so much.

Tung
Anyone knows how to display iron loss in the model? Thank you so much. Tung

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 01.11.2011, 07:26 GMT-4
Hi Tung,

i have asked the same question, the answer is still no.

how and what will you see on the core area if it frequently being magnetized and unmagnetized? BH-closed loop area are depens on material properties not on position in some area. So i hope i can see a curve hysteresis loss (watt) vs time, but i have not yet.

but eddy current loss you can see it, directly. but dont forget to define resistivity/conductivity of core material. :) happy end, right?

regards
akmal
Hi Tung, i have asked the same question, the answer is still no. how and what will you see on the core area if it frequently being magnetized and unmagnetized? BH-closed loop area are depens on material properties not on position in some area. So i hope i can see a curve hysteresis loss (watt) vs time, but i have not yet. but eddy current loss you can see it, directly. but dont forget to define resistivity/conductivity of core material. :) happy end, right? regards akmal

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 01.11.2011, 13:56 GMT-4
Hi Akmal,

I agree that eddy current loss is dependent on geometry and conductivity of the stator core. Therefore, in my problem, the eddy current loss should be investigated taking into account influence of holes in the core.

Regarding B-H curve problem, I already sent you a detailed discussion on it. Please take time to check it and talk more. I think it is an interesting topic. Back to hysteresis loss, I think it is dependent on both magnetic flux density (B) and magnetic field intensity (H). And in induction machines, B is also affected by variations of core area. I think so. I do not really know this in permanent magnet synchronous machines in which the flux is generated by the permanent magnetic field.

What are your other opinions on this problem? Please share or discuss here. Thank you.

Best,
Tung
Hi Akmal, I agree that eddy current loss is dependent on geometry and conductivity of the stator core. Therefore, in my problem, the eddy current loss should be investigated taking into account influence of holes in the core. Regarding B-H curve problem, I already sent you a detailed discussion on it. Please take time to check it and talk more. I think it is an interesting topic. Back to hysteresis loss, I think it is dependent on both magnetic flux density (B) and magnetic field intensity (H). And in induction machines, B is also affected by variations of core area. I think so. I do not really know this in permanent magnet synchronous machines in which the flux is generated by the permanent magnetic field. What are your other opinions on this problem? Please share or discuss here. Thank you. Best, Tung

Please login with a confirmed email address before reporting spam

Posted: 10 years ago 14.03.2015, 06:34 GMT-4
plz help me
how to measure
1)torque
2)phase current
in this model
plz help me how to measure 1)torque 2)phase current in this model

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.