Getting Started with Maxwell 3D (ANSOFT)

[emredemir]

Ok Gerbi;


i really understand you are bussy but when you have free time Can you turn me for this topic.
also i can say my coil made by rectangular so i have to use rectangular for turns ,arent i? So i draw lots os rectangular and merge them . but i think draw one rectangular also useful for me,isnt it? because i will be take same results from programme.

For example one turn with current is 500 A. Other type is 5 turn with 100 A. they are the same things F= N*I . so Fuxue said and show me that i have to give in windings part Current= 100*sin(2*pi*20000*time) and solid . so it will solve eddy current and skin effect i think so...

 

[Fuxue Jin]

Give me some time and I will prepare some pictures to show how. I just finished a similar project.

Ok Gerbi;


i really understand you are bussy but when you have free time Can you turn me for this topic.
also i can say my coil made by rectangular so i have to use rectangular for turns ,arent i? So i draw lots os rectangular and merge them . but i think draw one rectangular also useful for me,isnt it? because i will be take same results from programme.

For example one turn with current is 500 A. Other type is 5 turn with 100 A. they are the same things F= N*I . so Fuxue said and show me that i have to give in windings part Current= 100*sin(2*pi*20000*time) and solid . so it will solve eddy current and skin effect i think so...

 

[Fuxue Jin]

Let's start with what's in the real world and then take a look at what can be done in Maxwell that is close to the real world.

The follow picture is a ferrite RM6 core with 24 turns winding. This is a real inductor model, made by SolidWorks. I don't think Maxwell can create this model.

For this 3D model, after imported into Maxwell, it takes about 20GB memory to generate mesh, and may take a few days for a single CPU machine to finish the solution (given the core material is BH curve, not linear). So for you, it is not practical.

The following is a simplified model.

It still has 24 turns. But each turn is represented by a single ring. In Maxwell, all these single ring will be added into one winding. So electrically, it is equivalent to the previous real model. And this is much easier for Maxwell to generate mesh and finish the solution. On top of that, the 3D model can be split into 1/8 of it and reduce the simulation time by 1/8.

(to be continued)

 

[Fuxue Jin]

Before we talk about your project, and I assume you understand what is skin effect, here let's review it again.

In the following pictures, the circle represents the cross section of a solid round wire, with 2mm diameter. An AC current at 1KHZ, 100KHZ, and 1MHZ are applied to the wire. The color represents the current density distribution.

1KHZ

100KHZ

1MHZ

So, due to skin effect, the whole cross section area of the wire is not fully used at high frequency, resulted in high resistance and hence higher power loss comparing DC current.

(to be continued)

 

[Fuxue Jin]

Now let's talk about your project.

First, since your main interest is the power loss in the copper, so you may just ignore the core loss, and then you don't have to have core in your project. A simple air coil will be good enough.

Second, if you don't need to have a real model representing a real transformer or inductor, you may simply using a circle to represent a round wire, and you can use Maxwell 2D instead of 3D for your project. A circle with symmetry around Z axis in 2D represent a ring in 3D. You can start with only one ring for the winding (one turn winding), and add more rings later (into one winding) to the extend that your machine can handle.

The resistance of a single ring can be calculated by the length and cross section area, so you can compare your simulated results easily.

Hope this helps.

 

[emredemir]

So Fuxue you explain me lots of think but why you don't say me to draw this coil on solid i have lots of machine engineer friends and they could draw me easily. anyway i want to send you my project if you accept by mail or other.

Firstly my first point is calculate power loss on a coil. The secon one is observe that increased the frquency and its effects on power loss.


So Which type of solutiion do you prefer to me? ...

 

[Fuxue Jin]

My suggestion is to draw a ring in Maxwell, assign copper to the material property, and go from there, to calculate the copper loss, stranded and solid, so you can get used to Maxwell.

After that, you can work on your real model, whether created in Maxwell or imported from SolidWorks.

So Fuxue you explain me lots of think but why you don't say me to draw this coil on solid i have lots of machine engineer friends and they could draw me easily. anyway i want to send you my project if you accept by mail or other.

Firstly my first point is calculate power loss on a coil. The secon one is observe that increased the frquency and its effects on power loss.


So Which type of solutiion do you prefer to me? ...

 

[emredemir]

Fuxue you said me that i have to draw a ring but how? for example draw ring by torus is usefull for 3D?Fuxue i want to send you my project and you can easily find what is my problem ,İf you have enough time. Furthermore, it doesn't take your time really. Because my problem very simple and you understood what i want. May it is very easy to describe me something.You don't think that i don't want from you to do my work . Only analyze my work and tell me about What i did wrong or i didnt use


Thanks Fuxue again...

 

[Fuxue Jin]

I don't have access to Maxwell until next week so cannot look at your project.

You can draw a ring in a few ways. You can also import a ring from SolidWork if you have someone else help you in SolidWork.

In Maxwell, you can draw a torus. You can also draw a circle and sweep around an axis to make a ring.

Fuxue you said me that i have to draw a ring but how? for example draw ring by torus is usefull for 3D?Fuxue i want to send you my project and you can easily find what is my problem ,İf you have enough time. Furthermore, it doesn't take your time really. Because my problem very simple and you understood what i want. May it is very easy to describe me something.You don't think that i don't want from you to do my work . Only analyze my work and tell me about What i did wrong or i didnt useThanks Fuxue again...

 

[emredemir]

Fuxue :


i want to insert a variable like that but i don't know Is it true or false?

Current1= 400*sin(2*pi*2000*time)


i inserted this variable in Winding part but when i make analyses i didnt see any sinusoidal figure on the Loss map. So is it a true or false situation for me? I saw just the same things when i inserted the dc current. There was nothing unchanged.

I wonder Current1 variable is true or not inserted.

 

[Fuxue Jin]

The current is correct.

You can plot input current vs time to verify it is sine wave. Loss is not sine wave, you need take average value on the loss vs time curve. Make sure you plot solidloss if your copper is solid.

Fuxue :


i want to insert a variable like that but i don't know Is it true or false?

Current1= 400*sin(2*pi*2000*time)


i inserted this variable in Winding part but when i make analyses i didnt see any sinusoidal figure on the Loss map. So is it a true or false situation for me? I saw just the same things when i inserted the dc current. There was nothing unchanged.

I wonder Current1 variable is true or not inserted.

 

[emredemir]

Fuxue i create coil by polyline is it solid or not? i inserted the solid winding current so that it is true? furthermore i see losses from ''Create Transient REport'' part on Results,arent i? So i want to send some you some picture of my project . may be you can understand easily i hope. but i don't have i send photos you from this forum


İn Addition Do you know What is the ohmic loss and What Do we doo in Field overlays?

 

[Fuxue Jin]

Polyline is a line, not sure how you create a coil with that. Anyway, a coil needs to be a 3D object with volume.

Yes, you will be able to plot a rectangular plot for loss vs time in "Create Transient Report".

Ohmic loss only appears in calculator, I think. That is the I^2R loss in a conductive object.

Field overlay allows you to create B or J filed on certain plane.

You need to be a little patience to learn the basic. Maxwell is not a car you jump into drive, but it's an aircraft.

Fuxue i create coil by polyline is it solid or not? i inserted the solid winding current so that it is true? furthermore i see losses from ''Create Transient REport'' part on Results,arent i? So i want to send some you some picture of my project . may be you can understand easily i hope. but i don't have i send photos you from this forum


İn Addition Do you know What is the ohmic loss and What Do we doo in Field overlays?

 

[emredemir]

hello Fuxue:

i m annoyed to you again so thet i have some problems. firstly i want to calculate a power. this power is on a coil. but there is no excitation. two line is coupled with this coil and so excitations were given by this lines. for example one of these have 540 A AC and the other one is the same which inductively effects coil and make up current and a power on coil. So my teacher wants from me to calculate power and current on this coil. Is it impossible or not?

 

[Fuxue Jin]

In Maxwell, excitation is applied to an area which is a 2D object. Excitation cannot be on a line.

Calculate power is possible depending on how complicate your model will be.

Take a screen shot of your model and post it here. Let's take a look at your model.

hello Fuxue:

i m annoyed to you again so thet i have some problems. firstly i want to calculate a power. this power is on a coil. but there is no excitation. two line is coupled with this coil and so excitations were given by this lines. for example one of these have 540 A AC and the other one is the same which inductively effects coil and make up current and a power on coil. So my teacher wants from me to calculate power and current on this coil. Is it impossible or not?

 

[emredemir]

This is my model and you may not understand there is two line which have current excitations and they are colored grey (iron). also red one is ferrit model and brown one is coil ( copper). my work is give the lines current excitation and calculate the current and power on coil. This is possible?

http://https://www.physicsforums.com/attachment.php?attachmentid=47762&stc=1&d=1338311715

 

[Fuxue Jin]

I am not sure how much you understand about a transformer. Just by looking at the model, a few questions here;

You have an E core, but only half. Normal transformer need two E cores mating together to close the magnetic path without gap. Only half E core means a huge air gap.

You have orange color coil around center core leg, which is good. But that is what we have been talking about so much previously. This coil may represent a multi-winding turns which you cannot use for solid power loss simulation.

Two grey wires in the model are strange, totally wrong to me. You never have a transformer made that way.

There is another mistake. You don't have region defined, so that is the reason you have those error messages.

I am really sorry but I have to say, based on these, you don't have good understanding about transformer and Maxwell. Your assignment may be over your capability and you need much more time to get the basic.

This is my model and you may not understand there is two line which have current excitations and they are colored grey (iron). also red one is ferrit model and brown one is coil ( copper). my work is give the lines current excitation and calculate the current and power on coil. This is possible?

http://https://www.physicsforums.com/attachment.php?attachmentid=47762&stc=1&d=1338311715

 

[emredemir]

Fuxue i erased the region when i attached the photo. i now transformer model but i use different type of this. i want to show you real image of model. may be you can understand simply. but i want to ask you can i draw this iron tracks by cylinder? also how can i calculate power on coil? and how can i see the current and voltaj on this coil? the current which on these tracs coupling with coil and generate current on this coil. are you understand what i say? also you can easily say me what i done wrong . it is not important for me. :)

 

[Fuxue Jin]

In the picture, I don't see coils on the core center leg. Do you have windings on the center leg?

I assume those two black long bars are what you called "iron tracks". If so, yes, you can drawing them by cylinder and assign voltage or current to them. And then obtain power loss in Maxwell.

Fuxue i erased the region when i attached the photo. i now transformer model but i use different type of this. i want to show you real image of model. may be you can understand simply. but i want to ask you can i draw this iron tracks by cylinder? also how can i calculate power on coil? and how can i see the current and voltaj on this coil? the current which on these tracs coupling with coil and generate current on this coil. are you understand what i say? also you can easily say me what i done wrong . it is not important for me. :)

 

[emredemir]

fuxue i draw new things and i send you now. but which analyses type do you suggest me? transient or magnetostatic? in addition iin transient analyses: do i give these tracks coil terminal and define winding on this tracs ? also winding you say on coil is excitation one or mean turn which one i didnt understand. and finaly haw can we calculate power on this coil:)

 

[emredemir]

Fuxue ;

i have already calculated the induced current on coil but i want to calculate power on this coil now. do you know anything else ?*

 

[Fuxue Jin]

You need go to Results, then plot loss vs time.

Fuxue ;

i have already calculated the induced current on coil but i want to calculate power on this coil now. do you know anything else ?*

 

[emredemir]

hi fuxue:

i want to ask you that. how can we solve this error do you know?
error: object 'part 1' and 'box1' intersect.

Do you know anythin about this.? in addition i import the file on this project. may be it caused this error or not?

 

[Fuxue Jin]

As the message says, two objects intersect each other. They cannot overlap.

hi fuxue:

i want to ask you that. how can we solve this error do you know?
error: object 'part 1' and 'box1' intersect.

Do you know anythin about this.? in addition i import the file on this project. may be it caused this error or not?

 

[emredemir]

thanks fuxue;

but i have a new error but this is killl me

ERROR: unable to create simulation working directory within temp dir:...

Do you know anything about thet.??

 

[Petr McDuck]

Hello,

is there anybody who know how to change the default value of stator winding conductivity (resistivity) in RMxprt design in case of 3 phase induction motor...?
As I know the default material of winding is copper...but i need to change it...

Thank you very much for you reply,

Petr

 

[pourmoosa]

Dear all
I am Working with Maxwell 3d, to simulate Linear induction motor,my email is aliasgharpourmoosa@gmail.com
When setting up a transient solver problem in Maxwell 3D, the following error occurred when running the simulation:

"Could not find the face. There might be an isolated hole in the band"

Does anyone know what this means? When I use the magnetostatic solver the simulation runs fine.
sorry for my English
Thank you
Ali

 

[GenB]

Hi, I'm trying to learn the basics of this program (V13) and haven't had luck with the user's guide. Can anyone explain how I might model the magnetic field of a wire with a current of 1 A in Maxwell 3D? It would be greatly appreciated!

 

[msorensen]

hello
I want to simulate an electric motor. I have one broken down and I know what they look like individual parts. It is important to perform simulation of the Maxwell 3D, not the "Rmxprt" because I have a lot of ideas for different rotor designs, and I sketched them in 3D. For three months I look through the documentation and tutorials with ansoft. In no case is similar to the embodiments of my model.
I need your help.
It shows in the Annex, where is my problem:
http://www.scribd.com/doc/102721866/2012-0810-2340

Generally, I know that I need to simulate: magnetic-transient or transient Electric. I suspect that the two simulations to connect in some way. Maybe someone will tell me if I'm wrong?

In the case of transient simulation Electric-like state on the ends of the coil voltage. The input impedance and resistance is needed. Where can I calculate this data? I can generally count on a paper but I doubt that all engineers as the problem in this way.

Does anyone have any suggestions on how to start the simulation?
Regards.
Sorensen

 

[annalisemalta]

Hi,

I am using Maxwell 13 and I am trying to simulating a 3 phase 4 pole permanent magnet machine, so far I have managed to get reasonable results however I am still not sure about certain things.

#1 problem that I have is that I managed to excite the winding with a 3 phase current input however I am getting results that do not make sens when I tried to excite the winding with a 3 phase voltage input.

#2 another question that I have is how to determine weather the machine is extracting or absorbing power? ie what is the software's conventional equivalent circuit of a generator/ motor?

Thanks a lot
Annalise

 

[Fuxue Jin]

For the voltage excitation, you need input the coil resistance. The simulation will not model the resistance if the conductor coil is specified as stranded.

Hi,

I am using Maxwell 13 and I am trying to simulating a 3 phase 4 pole permanent magnet machine, so far I have managed to get reasonable results however I am still not sure about certain things.

#1 problem that I have is that I managed to excite the winding with a 3 phase current input however I am getting results that do not make sens when I tried to excite the winding with a 3 phase voltage input.

#2 another question that I have is how to determine weather the machine is extracting or absorbing power? ie what is the software's conventional equivalent circuit of a generator/ motor?

Thanks a lot
Annalise

 

[annalisemalta]

Hi,

i did fill the coils resistance and inductance, and the voltage as a sin input with each phase shifted by 120 degrees... However the current waveforms include kind of a DC offset and I got two of the phases shifted upward and another one shifted downwards. I tried to invert the excitation of the winding shifted downwards to at least get the same DC shift in all the phases but did not work out as well...

I thought maybe I need to set other parameters that maybe are disabled when the coils are excited with current...

 

[gerbi]

Hello
annalisemalta, this phenomena You are observing is transient state in electromagnetic circuit.
This is pretty similar to inrush current in transformer. Voltage and flux waves in core are shifted by 90 deg. Now imagine what happends when You switch on voltage. If voltage waveform isn't at it's peak (flux is then at zero crossing - current too) current will have some disorder component in it. It will decline in time with some time constant, and only main component of waveform will remain.

You can check if it's true - just set up voltage excitation in one phase to be at it's peak in t=0 (other phases shifted +/- 120deg).

 

[annalisemalta]

Yes there is kind of a transient but even after almost 0.5s the waveforms are very far away from a normal 3 phase without a dc shift. I attached my current results for the time I simulated

 

[gerbi]

Rearrange excitation like I said. See if it helps or not. If not.. maybe try with external circuit from Maxwell Circuit Editor ? There are some more possibilities with it.

 

[eddyrooney87]

Hi everyone,

Need you guys help with Ansoft Maxwell 13. I am currently working on electromagnetic linear actuator as a vehicle suspension system. My design should look like below:
_____
| s |
| h |_____ ____ ______
| a | N | | C | | S |
| f |_____| |___| |_____|
| t |

It will be in cylindrical shape. Hope you understand.
N = north pole (magnet material, Neodymium)
S = south pole
C = coil winding

I already design the geometry in Ansoft. But I do have some question related to Maxwell:

1. How to assign magnetic pole (N/S) to my design.
2. How to assign coil polarity during current excitation.
3. Can we create external force to excite the shaft downward or upward?
4. What type of solver should be used for this kind of design?

Sorry for my english. Appreciate your help.
Eddy

 

[gerbi]

1. How to assign magnetic pole (N/S) to my design.
2. How to assign coil polarity during current excitation.
3. Can we create external force to excite the shaft downward or upward?
4. What type of solver should be used for this kind of design?

Ad. 1: Assign material (permeant magnet) to object. In it's properties You can select how magnetic coercivity vector is oriented
Ad. 2: by altering current direction. Coil polarity depends on direction of current
Ad. 3: Yes, there is such a posibility (assigning a band with mechanical transient)
Ad. 4: Transient (Magnetic)

One more thing, I suggest reading some tutorials and searching Maxwell On-Line Help for vital information about preparing simulation. If You do not have tutorials write me a PM with your e-mail adress and I will send them.

 

[annalisemalta]

gerbi,

is it possible to send me some tutorials on annalisexuereb@gmail.com

Thanks
Annalise

 

[Fuxue Jin]

You don't necessarily input extra inductance. The simulation will use the modeled inductance of the coil. Induced current will have 90 degree phase shift from the input voltage.

Hi,

i did fill the coils resistance and inductance, and the voltage as a sin input with each phase shifted by 120 degrees... However the current waveforms include kind of a DC offset and I got two of the phases shifted upward and another one shifted downwards. I tried to invert the excitation of the winding shifted downwards to at least get the same DC shift in all the phases but did not work out as well...

I thought maybe I need to set other parameters that maybe are disabled when the coils are excited with current...

 

[eddyrooney87]

Hi gerbi,

Ad. 1: Assign material (permeant magnet) to object. In it's properties You can select how magnetic coercivity vector is oriented

It is we change the magnitude value to negative(-)/positive(+). Not really understand what you are suggesting.

Ad. 3: Yes, there is such a posibility (assigning a band with mechanical transient)

I did try to assign the shaft as a band but error come out. "Invalid object selected for band. No objects within the band". How could I assign the shaft in mechanical transient?

Appreciate if you could share with me some tutorials for Maxwell 3D for my reference. Please send to dey_ad@yahoo.com.

 

[gerbi]

You can change Magnitude but X-Y-Z components aswell. Field with orientation (1x,0y,0z) will be perfect opposite to (-1x,0y,0z).

About band.. You need to close moving object inside other object to which You'll asign a band.

Please, read the tutorials (especialy about Toyta Prius motor) - all basics are explained there.

 

[annalisemalta]

"You don't necessarily input extra inductance. The simulation will use the modeled inductance of the coil. Induced current will have 90 degree phase shift from the input voltage."


I know that... my problem is that the shift is a dc shift between the currents of the phases not between the voltage and the current, as in the attached screenshot

 

[annalisemalta]

Hi,

regarding the transient plots in Maxwell... What is the Moving1.Toqrue? and when exciting winding by current what is the voltage that is available for plotting? the induced voltage or the terminal voltage?

Thanks

 

[cimopata]

Hy Everyone!

I would like to ask for your help. There is a transformer transient tests on which I am doing.
I want to calculate or measure the amount of heat loss in [W] occurs in the core.

How do I do this?
Thanks
Bálint

 

[gerbi]

What is the Moving1.Toqrue?

Don't know. There is no info on it in help. Try plotting it and see how it behaves.

when exciting winding by current what is the voltage that is available for plotting?

it's voltage induced by the current, across the considered element

I want to calculate or measure the amount of heat loss in [W] occurs in the core.

In transient solver there is a possibility to define core loss model. Go to material properties, from menu choose "core loss type" -> "electrical steel". Some more coefficients are now avaible in your window. At the bottom from the menu choose "calculate properties for" -> "core loss at one frequency". In new window enter data You have (frequency, thickness etc.), You should have it from core material manufarcturer and transformer design data. If You don't know sheet material conductivity, which is not so easy to obtain, go with ~2 MS/m.

Ok, with material properties setup ready go to Maxwell2D/3D -> Excitations -> Set Core Loss. Pick core element (if material props are fine, there should be check in right column).
Make mesh refinement in core (accuracy of losses calculation depends on calculated flux distribution) and solve model.
For results plot core loss vs time in results -> create transient report-> rectangular plot.

 

[cimopata]

Don't know. There is no info on it in help. Try plotting it and see how it behaves.

it's voltage induced by the current, across the considered element


In transient solver there is a possibility to define core loss model. Go to material properties, from menu choose "core loss type" -> "electrical steel". Some more coefficients are now avaible in your window. At the bottom from the menu choose "calculate properties for" -> "core loss at one frequency". In new window enter data You have (frequency, thickness etc.), You should have it from core material manufarcturer and transformer design data. If You don't know sheet material conductivity, which is not so easy to obtain, go with ~2 MS/m.

Ok, with material properties setup ready go to Maxwell2D/3D -> Excitations -> Set Core Loss. Pick core element (if material props are fine, there should be check in right column).
Make mesh refinement in core (accuracy of losses calculation depends on calculated flux distribution) and solve model.
For results plot core loss vs time in results -> create transient report-> rectangular plot.

Thanks for your help!

Yes, I found this options and I adjusted the core losses in 25°C(by the "power ferrit" core loss type).
Actually I'm using N87 ferrit core. I already can measure the core loss. The results is very nice :)
The next step what I would like to do is the "Thermal modifier".
By the core loss type I have Power ferrite. There is 3 value (Cm, X, Y). Every value have "Edit thermal modifier" option. In this window how can I adjust the loss for example 100°C?

Thanks!
Best Regards!
Bálint

 

[cimopata]

Other quoestion:

The simulation is very slow. 50us transient (0.5us step) takes about 20 hour. I would like to speed up. My toroids dimension is about 50mmx16mm. What is the recomended mesh size (faster simulation/precise caltulation).

My compure:
Intel core i7
6GB 1333mHz RAM

Thanks!

 

[gerbi]

[..] how can I adjust the loss for example 100°C?

I'm sorry, but I have no expierence in this field. It appears that You'll need to read some articles about material properties of power ferrite. Maxwell help contain no detailed information on this one.

The simulation is very slow. 50us transient (0.5us step) takes about 20 hour. I would like to speed up. My toroids dimension is about 50mmx16mm. What is the recomended mesh size (faster simulation/precise caltulation).

It depends. For some crude calculations (to get familiar with model, results etc) mesh shouldn't be very fine.
For this first few simulations and 3D model about few thousand elements ? Thats my guess. For more detailed calculations few times more, but do not create monstrous mesh because it won't help accuracy much.

Some more hints on mesh:
1) Observe calculated energy error in solution data. If it is below 0,5% You can assume it is quite accurate (very accurate below 0,1%).
2) Do some simulations in static solvers (like Eddycurrent). Mesh is done adaptively then. You can observe which areas have increased meshing and do similary in transient solver manually.
3) Observe results (mostly flux distribution). Field details like increased density (if present) should be visible and look natural.
3) Meshing is one of most important things in FEM, it's better to learn how good mesh looks like.

 

[eddyrooney87]

hi gerbi,

i have difficulties on assigning current to my coil as it is wounded on cylinder core. When I want to assign current, this error occur "Excitation 'Current 1':Cannot apply Current Excitation to non planar entity". Any tips and help from you.

Thanks

 

[cimopata]

Some more hints on mesh:
1) Observe calculated energy error in solution data. If it is below 0,5% You can assume it is quite accurate (very accurate below 0,1%).
2) Do some simulations in static solvers (like Eddycurrent). Mesh is done adaptively then. You can observe which areas have increased meshing and do similary in transient solver manually.
3) Observe results (mostly flux distribution). Field details like increased density (if present) should be visible and look natural.
3) Meshing is one of most important things in FEM, it's better to learn how good mesh looks like.

Thanks gerbi!

I am going to try it!

Maybe have you any experience the distributed mode computing between computers?
I tried, but there is something error message:
"Error: Unable to contact AnsoftRSMService, Reason: ANS_CANNOT_CONNECTTO_ANSOFTRSMSERVICE
Testing Completed.
Error: Some tests have failed.