ANSYS Maxwell 16.02
Jerrica Ingles
After importing the geometry into DM and generating the model, I can see the global variables and I am able to export them to Workbench as input paramters to use them in a parameteric analysis to determine the change in capacitance with the electrode dimensions. I'm also able to connect this to Maxwell, set up my simulation conditons and export output parameters (capacitance matrix) to the WB.
The issue that I'm seeing with my parametric analysis is that only DP1 updates and all remaining design points seem to show the same solution. I've noticed that the geometry in Maxwell does not refresh/update after the first design point and the following error shows up in Maxwell
[error] Link parameter 'P3@DS_SQ' in source system 'Geom' is mapped to constant '1' in target system. Map the link parameter to a design variable or project variable in the target system to enable synchronization. (4:09:29 PM Aug 15, 2018)
It seems that SolidWorks 2017 is not supported in PlugIn Mode with Workbench 17.2 as per the documentation. So ideally first step would be downgrading your SolidWorks version to 2015 or 2016 to work with R17.2. Alternatively, you can also upgrade your ANSYS Workbench version to anything from R18.0 to R19.1 to work with SolidWorks 2017 in the plugin mode.
The issue is not with the Solidworks version but rather with the fact that Maxwell requires you to define local variables for all your geometry parameters from Design Modeler so that you can parametrize them from WB
hi wallygator88,I have met the same issue as you have , and i am really happy to know that you have found the answer , i want to know if i should define local variables name in maxwell exactly the same as the ones in SW? like you issue, the parameter 'DS_SQ' should be define in maxwell? even though it has no meaning in maxwell?
Hey i dont know if it really helps, but i have a similar problem. Look into page 1333 of the Maxwell Help Release 2020. When you click on your imported geometry, you go to properties and there u put a new local variable into the value of the imported parameter.
Do you know how to make the imported parameter and the maxwell parameter automatically equal when running an optimization? In your picture for example your design point 0 is P1 dist=0.5 and P3 dist=0.5. When I start an optimzation, ansys put new design points for example -> P1 dist=0.3 and P3 dist=0.65, but they need to be equal.
Join us for this insightful webinar to discover how HPC and cloud computing unlocks the full potential of engineering simulation. Whether you are an engineer, IT manager, researcher, or technology enthusiast, this webinar provides the knowledge and tools to enhance your simulation workflows and drive innovation in your industry.
It is fascinating to compare experimental and simulation test results for a tensile test specimen. The ability to conduct tests virtually, rather than physically, utilizing virtual simulations for testing not only increases efficiency but also leads to significant cost savings. In this crosstalk, we will address some of the challenges typically encountered in uniaxial tensile simulations.
Subscribe to the Ansys Blog to get great new content about the power of simulation delivered right to your email on a weekly basis. With content from Ansys experts, partners and customers you will learn about product development advances, thought leadership and trends and tips to better use Ansys tools. Sign up here.
current effects on. For my simulation set up I have a balloon boundary around the model. For the excitation I set the red (primary winding) and blue (secondary winding) as coils with the appropriate polarity. I then grouped them as windings and add the current excitation. I add all my winding to the matrix so the solver will output the inductance matrix. I'm trying to find the leakage inductance in the model and I have 2 methods to do so. The first is through using equation derived from the 2x2 inductance matrix. I show the derivation below. The total leakage inductance reflected to the primary is L11 - (M^2)/L22. The other method I am using is energy. I short the secondary winding and make sure that there is no energy in the core so the only energy left in the simulation is energy being stored in the leakage inductance. Then I take the energy output from the convergence tab of maxwell or by taking the volume integral of the energy density in the region around my winding and set it equal to .5*(Lktotal)*Ip^2 (this equation also refers to the primary side). The problem is that both methods give me very different results, with the energy method having a leakage value 10 to 20 times higher than the equation method. When I compare equation method and energy method in magnetostatic solver the results are very close. Am I setting up something wrong in my eddy current simulation or doing something else wrong? Any insight would be great.
However as 3rd option to use stored energy, assign 1A in each electrical turn of the winding to which the leakage inductance will be referred (pri or sec). Put equal and opposite ampere-turns in the other winding (sec or pri). Leakage inductance = 2 * energy.
Do you see what error I'm making that results in this difference? I attached a picture showing that there is no energy in my core. Taking the volume integral of the region with only my winding gives me 2 uH leakage energy.
In the past I have been using other materials to calculate values of the mutual inductance because I did not believe this would make a significant difference in the values achieved, but I know think otherwise. I have tried to remedy this by creating a material with near 0 (10^(-10)) value for the relative permeability, since a true superconductor would have a value of 0 here, but Maxwell doesn't seem to be able to simulate something with a relative permeability of 0, and a large value for the conductivity (it starts throwing errors around 10^12). A relative permeability of 0 would give a magnetic susceptibility value of -1, but I have not been able to find a location that I can specific the chi value. The built in material 'perfect conductor' does not have the parameters that a true superconductor has, and also Maxwell does not allow conduction paths to be made out of this material. I'm mostly curious about a way that I can confidently represent the true mutual inductance of my chip and get around the restrictions maxwell has put on the material settings.
I also have the issue of the problem region I am solving over. Since I have a current entering and exiting the chip, I am only able to make the problem region flush with two of the sides, and then extend some distance on the others. I'm not sure what effect the problem region has with the values, but they seem to fluctuate significantly as I adjust the size of the region that I am solving over. But this still restricts me to have a gap of 0 on two of the sides, which seems like it is causing issues. What is the proper way to create a problem region to give me accurate results.
Regarding the region and excitation if I understood the question correctly, you can make a short conduction loop in your design and make an intersection and apply the current that way. that way you are not adding extra length to your design and region. Did I understand it correctly?
Can you explain the short loop a bit further? does this mean having a line of conductor, but having a small loop at some point within the line and then applying it in that circle? would this require creating a insulating boundary condition on the line, and would that also represent the current flowing through the wire in the correct way?
My main question with the region is that I am unsure of how large I should make it. Before i was partially limited due to the current excitations, but on the sides that were not if I increased the region size, it would increase the value calculated for the mutual inductance up to some point where it peaked. So I'm unsure how large to make the region to get an accurate number for how large the actual mutual inductance is. Should it be as large as possible for it to converge, as small as can fit around the model, or somewhere in between?
b37509886e