Electric field plot

Please login with a confirmed email address before reporting spam

Hi, I am using the scattering field formulation for my model. I am using an x-polarized negative z-propagated plane wave as my background electric field. My structure is a box consisting of an air domain(top) and cancer domain(below). I used scattering boundary conditions. After the simulation, I realized that 1. emw.normE = abs(emw.Ex) 2. but abs(emw.Ez) is not = 0. if 1 is true, then how is 2 possible(since i am propagating the z direction anyways). or i am not plotting it correctly. I have attached my file. Thanks



15 Replies Last Post 2 ott 2024, 14:07 GMT-4
Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 16 ago 2024, 20:02 GMT-4
Updated: 3 months ago 16 ago 2024, 20:07 GMT-4

Your example, at least as provided, is geometrically simple, but computationally very large, due to the small wavelengths involved relative to the model geometry. Your "cancer" material is defined as having a relatively high dielectric constant and also a relatively high conductivity. As such, it is very reflective to a direct incident RF wave at the specified frequency. I prepared a modified model simply to get it to execute. I made that model less wide in the two lateral dimensions and used linear elements and a user-specified mesh to keep the model computationally small enough. I noticed you did not take advantage of Comsol's built in tool for specifying plane wave RF incidence, so I changed your model to use that, and I encourage you to take a look at that. Frankly, this is still not a very good model for the purpose of investigating such a high frequency (500 GHz) wave interacting with your target. Computed penetration of such a 500 GHz wave is surely quite minimal into this material, at least if it really has the properties (dielectric constant and conductivity) that you specified, at that extreme frequency. The model also shows this more or less, but I still wouldn't trust the computed details for the penetrated field quantitatively here, since there are likely numerical noise issues. Perhaps you could redo this model with a much thinner layer for the "cancer" (since barely any field is getting into it anyway) and then could mesh that "cancer" domain more finely, if you really want to study it in detail. You could also try a 2D model, since that would help enormously in reducing your computational size. I hope that helps. Good luck.

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
Your example, at least as provided, is geometrically simple, but computationally very large, due to the small wavelengths involved relative to the model geometry. Your "cancer" material is defined as having a relatively high dielectric constant and also a relatively high conductivity. As such, it is *very* reflective to a direct incident RF wave at the specified frequency. I prepared a modified model simply to get it to execute. I made that model less wide in the two lateral dimensions and used linear elements and a user-specified mesh to keep the model computationally small *enough*. I noticed you did not take advantage of Comsol's built in tool for specifying plane wave RF incidence, so I changed your model to use that, and I encourage you to take a look at that. Frankly, this is *still* not a very good model for the purpose of investigating such a high frequency (500 GHz) wave interacting with your target. Computed penetration of such a 500 GHz wave is surely quite minimal into this material, at least if it *really* has the properties (dielectric constant and conductivity) that you specified, at that extreme frequency. The model also shows this more or less, but I still wouldn't trust the computed details for the penetrated field quantitatively here, since there are likely numerical noise issues. Perhaps you could redo this model with a *much thinner* layer for the "cancer" (since barely any field is getting into it anyway) and then could mesh that "cancer" domain more finely, if you really want to study it in detail. You could also try a 2D model, since that would help enormously in reducing your computational size. I hope that helps. Good luck.


Please login with a confirmed email address before reporting spam

Posted: 3 months ago 3 set 2024, 11:25 GMT-4

Thank you very much for your help. I am however confused with some results. If I plot the background field in the xy-plane, I am supposed to get a fairly constant value. lets say I expect to get real(emw.Ebx) = -0.45 V/m. but I get real(emw.Ebx) = {-0.39 to -0.503}. I know theres is going to be computational errors and noise. Is there a way to the results as close to what I expect as possible?

Thank you very much for your help. I am however confused with some results. If I plot the background field in the xy-plane, I am supposed to get a fairly constant value. lets say I expect to get real(emw.Ebx) = -0.45 V/m. but I get real(emw.Ebx) = {-0.39 to -0.503}. I know theres is going to be computational errors and noise. Is there a way to the results as close to what I expect as possible?

Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 3 set 2024, 12:44 GMT-4

The real part of any non-zero component of the background field will exhibit spatial oscillations with phase, just as it should. If you want to sanity-check the magnitude, plot abs of the field component of interest. E.g., plot abs(emw.Ebx), not real(emw.Ebx).

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
The real part of any non-zero component of the background field will exhibit spatial oscillations with phase, just as it should. If you want to sanity-check the magnitude, plot *abs* of the field component of interest. E.g., plot abs(emw.Ebx), not real(emw.Ebx).

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 3 set 2024, 18:46 GMT-4

Thank you sir. Again, is it possible to get Electric field component in the z direction while the plane wave is propagating in the z direction?

Thank you sir. Again, is it possible to get Electric field component in the z direction while the plane wave is propagating in the z direction?

Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 4 set 2024, 09:42 GMT-4

Are you asking a computational question or a physics question? A z-directed plane wave has no z component of E, but if it scatters/interacts with a material or boundary, a non-negligible z component might (or might not) arise, depending on the material and the details of the interaction. If a plane wave is normally incident (i.e., perpendicular to a flat surface) upon an isotropic material, then I would not expect any z component of E to arise, since there would be no change to the symmetry.

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
Are you asking a *computational* question or a *physics* question? A z-directed plane wave has no z component of E, but if it scatters/interacts with a material or boundary, a non-negligible z component might (or might not) arise, depending on the material and the details of the interaction. If a plane wave is normally incident (i.e., perpendicular to a flat surface) upon an isotropic material, then I would not expect any z component of E to arise, since there would be no change to the symmetry.

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 4 set 2024, 11:23 GMT-4

I was asking a physics question to relate it to computational. So for example, the model that you just helped me with, when i plot the abs(emw.relEz), i get 0.26 V/m amplitude while the abs(emw.relEx) gives 0.78 V/m. This makes the Ez component quite significant here. This is the reason for my question. can this be a simulation noise or something? If so, is there anything you recommend can help fix this?

I was asking a physics question to relate it to computational. So for example, the model that you just helped me with, when i plot the abs(emw.relEz), i get 0.26 V/m amplitude while the abs(emw.relEx) gives 0.78 V/m. This makes the Ez component quite significant here. This is the reason for my question. can this be a simulation noise or something? If so, is there anything you recommend can help fix this?

Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 4 set 2024, 11:58 GMT-4
Updated: 3 months ago 4 set 2024, 11:59 GMT-4

The scattering boundary conditions on the sidewalls are not really correct for this configuration. If you replace them with PECs on the walls normal to the x axis, and PMCs on the walls normal to the y axis, then those will cleanly fit the example incident plane wave and you will see the Ez component nearly vanish. If you are interested in exploring a non-normal angle of incidence, you may have to introduce some other boundary condition, such as perfectly matched layers, or even appropriately periodic conditions, such as Floquet boundaries.

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
The scattering boundary conditions on the sidewalls are not really correct for this configuration. If you replace them with PECs on the walls normal to the x axis, and PMCs on the walls normal to the y axis, then those will cleanly fit the example incident plane wave and you will see the Ez component nearly vanish. If you are interested in exploring a non-normal angle of incidence, you may have to introduce some other boundary condition, such as perfectly matched layers, or even appropriately periodic conditions, such as Floquet boundaries.

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 4 set 2024, 12:23 GMT-4

So you mean the boundaries on the X-axis should be PEC and the boundaries on the Y-axis should be PMC.? Is this because the plane wave is x polarized or the PEC and PMC on the boundaries does not matter.

So you mean the boundaries on the X-axis should be PEC and the boundaries on the Y-axis should be PMC.? Is this because the plane wave is x polarized or the PEC and PMC on the boundaries does not matter.

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 4 set 2024, 13:13 GMT-4

I have added two models , which one is the correct setup

I have added two models , which one is the correct setup


Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 4 set 2024, 13:40 GMT-4

Electric field lines terminate on PECs, while magnetic field lines terminate on PMCs. As I explained, use the "PECs on the walls normal to the x axis." When discussing a vector quantity, "normal to" means "perpendicular to." This corresponds to your Example2.mph, for incident polarization (E) parallel to the x axis (and thus the PEC surfaces must be perpendicular to E).

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
Electric field lines terminate on PECs, while magnetic field lines terminate on PMCs. As I explained, use the "PECs on the walls normal to the x axis." When discussing a vector quantity, "normal to" means "perpendicular to." **This corresponds to your Example2.mph**, for incident polarization (E) parallel to the x axis (and thus the PEC surfaces must be perpendicular to E).

Please login with a confirmed email address before reporting spam

Posted: 3 months ago 4 set 2024, 14:42 GMT-4

This makes sense to me. Thank you very much. And so if I should make the boundaries PML, it should solve the scattering problems I faced.

This makes sense to me. Thank you very much. And so if I should make the boundaries PML, it should solve the scattering problems I faced.

Please login with a confirmed email address before reporting spam

Posted: 2 months ago 1 ott 2024, 13:21 GMT-4
  • Hello sir, I used your suggestion on my problem above. I investigated both PEC/PMC boundary conditions as well as PML. I realized that PEC/PMC gave me the results I expected but the PML results is questionable. I wonder if I did something wrong using the PML. My geometry will become complex in the future so I dont want to rely on PEC/PMC . Is there anything I should do? I have attached my files and results . I welcome your suggestions.Thank you.
* Hello sir, I used your suggestion on my problem above. I investigated both PEC/PMC boundary conditions as well as PML. I realized that PEC/PMC gave me the results I expected but the PML results is questionable. I wonder if I did something wrong using the PML. My geometry will become complex in the future so I dont want to rely on PEC/PMC . Is there anything I should do? I have attached my files and results . I welcome your suggestions.Thank you.


Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 months ago 1 ott 2024, 19:30 GMT-4

It seems to me that your objective here is to simulate a representative subsection of a larger space, perhaps even an array (in effect, if not necessarily deliberate). Of course, if your incident wave is at normal incidence, you can use the PEC & PMC sidewalls previously discussed. But if you want to consider cases where the incident wave is not at normal incidence, then you may wish to consider a model using Floquet boundary conditions. There are a number of examples in the Comsol Application library. See, among others, https://www.comsol.com/model/modeling-of-pyramidal-absorbers-for-an-anechoic-chamber-12129 . That application shows how to use periodic floquet boundaries, and also provides one example of the use of a PML, relevant to this kind of work. Perhaps the RF absorber in that model is analogous to the cancer tissue in your model? If so, studying that model might really help you. Good luck.

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
It seems to me that your objective here is to simulate a representative subsection of a larger space, perhaps even an array (in effect, if not necessarily deliberate). Of course, if your incident wave is at normal incidence, you can use the PEC & PMC sidewalls previously discussed. But if you want to consider cases where the incident wave is not at normal incidence, then you may wish to consider a model using Floquet boundary conditions. There are a number of examples in the Comsol Application library. See, among others, https://www.comsol.com/model/modeling-of-pyramidal-absorbers-for-an-anechoic-chamber-12129 . That application shows how to use periodic floquet boundaries, and also provides one example of the use of a PML, relevant to this kind of work. Perhaps the RF absorber in that model is analogous to the cancer tissue in your model? If so, studying that model might really help you. Good luck.

Please login with a confirmed email address before reporting spam

Posted: 2 months ago 2 ott 2024, 12:45 GMT-4

My wave is at normal incidence so the PMC/PEC walls was valid. However, my goal is to simulate a structure like seen in the attached file, called figure. Initially when I used SBC and PML, I was getting electric field components in the Z.which was what i did not understand since my plane wave z propagated. My initial guess was that, my structure might not be flat. This is why i decided to use a simple flat structure as I have shown in our previous exchange. I was getting Ez components which was solved when i used PEC/PMC walls. However, when I used PEC and PMC walls, on my real geometry (figure), the results gets worse. I even get Ey and same Ez component. The PML works better on my real structure than PEC/PMC. I am just trying to understand why I get Ez components regardless of what i use. And since my structure will get more complicated, I dont want to rely on PEC/PMC walls. Please refer to the attached files for confirmation. I couldnt get my .mph files to be less than 5MB, so if you need to review that, I can send it to you anyhow you prefer. Thanks.

My wave is at normal incidence so the PMC/PEC walls was valid. However, my goal is to simulate a structure like seen in the attached file, called figure. Initially when I used SBC and PML, I was getting electric field components in the Z.which was what i did not understand since my plane wave z propagated. My initial guess was that, my structure might not be flat. This is why i decided to use a simple flat structure as I have shown in our previous exchange. I was getting Ez components which was solved when i used PEC/PMC walls. However, when I used PEC and PMC walls, on my real geometry (figure), the results gets worse. I even get Ey and same Ez component. The PML works better on my real structure than PEC/PMC. I am just trying to understand why I get Ez components regardless of what i use. And since my structure will get more complicated, I dont want to rely on PEC/PMC walls. Please refer to the attached files for confirmation. I couldnt get my .mph files to be less than 5MB, so if you need to review that, I can send it to you anyhow you prefer. Thanks.


Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 months ago 2 ott 2024, 14:07 GMT-4
Updated: 2 months ago 2 ott 2024, 14:09 GMT-4

I'm sorry, but I just don't have time right now to review this topic and your models further. I encourage other contributors to this site to offer their suggestions.

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
I'm sorry, but I just don't have time right now to review this topic and your models further. I encourage other contributors to this site to offer their suggestions.

Reply

Please read the discussion forum rules before posting.

Please log in to post a reply.

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.