Jump to content

Alejandro Rodríguez

  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by Alejandro Rodríguez

  1. Hello Xiaodong, In fact, this error may have two different causes: or the section of your coil is not constant or you have points in the lines defining your trajectory. Since coil is defined through surface extrusion, I guess you are in the second case: the solution is to define new cross-section faces relying this points I think that you should be able to define your non-meshed coils in a way that their geometry will be nearly the same you have in your meshed coils. Actually, depending on the application and the specific geometry, differences in coil’s shape could lead to important differences in magnetic behaviour (in fact, they can play a major role in local effects). In conclusion, in a general case, you should not neglect differences in geometry, especially in the areas where coil is near ferromagnetic components. I am sorry, but for further help you have to send us your project in order to check it. Best regards,
  2. Hello Natto, Actually, it really depends on your application definition. In a general way, resistances in stranded coil conductor components associated to face/volume regions are automatically evaluated during resolution taken into account coil area, length, material resistivity at working conditions, fill factor and number of turns (see image). Additionally, specific simulations can be performed to estimate resistance in working conditions (temperature and current distribution may play a major role in resistance values). Since you will obtain all electrical values you can evaluate real resistance during postprocessing. Anyway, for most of applications, a critical variable for coil resistance calculus is the fill factor that is mostly estimated through practical measures. Best regards, Alejandro
  3. Hello Louis, You will find Berttoti coefficient’s excel file in this path: [FLUX INSTALLATION DIRECTORY]\flux\Flux\DocExamples\Tools\BertottiLossesCoefficients\BertottiLossesCoefficients.xls By default, [FLUX INSTALLATION DIRECTORY] is C:\Program Files\Altair\2019_1 for version 2019.1. Best regards, Alejandro
  4. Hello Ruud, Sorry for this late answer. Actually, the easiest way to perform your calculus with movement is to use a support, typically a path or a 2D grid. In both cases you can associate a mechanical set to the support. If I have understood correctly your case, you can define your path in the mechanical set of the rotor. Once the project have been solved you can directly represent the normal (or tangential) component to this path. The “Curve -> 2D curve (Path)” menu allows to choose the normal component of your vector quantity (e.g., flux density) directly, no need to define an specific coordinate system. Hope this helps. Best regards,
  5. Hello Tehran, Actually, the multi-static kinematic model, the moving part of the device is not moving. The computation of the electromagnetic field is carried out for various relative positions of moving and fixed parts, chosen in the scenario. This model performs a set of Magneto Statics computations. Therefore, computation of the electromagnetic field is not the same used in real moving mechanical sets, since it does not take into consideration the dynamic equation. Use this model is equivalent to run a parameterized study where the position of the moving part is a varying parameter, that means that the variation in rotor position should be established in the scenario definition -> control of parameters. The torque calculated in this way is equivalent to the torque experimented by the rotor as a function of its position (ANGPOS_ROTOR ), so the concept is the same that using a rotating mechanical set at imposed speed and the results should be quite close. Anyway, I strongly advise you to use imposed speed instead, since dynamic equations are more accurate that magneto static ones when movement is involved. Best regards,
  6. Hello Sandesh, I am not completely sure if I understand your questions: I think that you want to compute global radial force over the stator for each rotor position and for each current, is this right? In this case the simplest and more straightforward solution is to use a sensor defined as you in the image attached where you should choose as lines of interest the borders between the teeth and the airgap (as shown). CYLIN_REF makes reference to a centred cylindrical-coordinates systems as it is shown in the second picture (i.e., it will established your reference to calculate the radial part of the force). If, on the contrary, you want to know the local forces over the stator border with the airgap (which is usually the case to do NVH analysis) I advise you to follow Flux supervisor’s example that you will find in 2D, Open example->Multiphysics->Vibroacustic analysis via OptiStruct . Best regards.
  7. Hello Tehran, Actually, you have several possibilities depending of your physics and your specific needs. The most simple one, only for linear problems with no magnetic saturation (i.e., rotor and stator modelled as ideal materials), is to use an specific sensor which measures inductance (it divides the total flux embraced by the coil between the current carried by this same coil, which is, in fact, an accurate definition of inductance for magnetic linear problems). You can find more accurate and sophisticated way to calculate this value for electric machines in our tutorial in 2D: Brushless IPM motor (embedded magnets), section 7.3. You will find it in the supervisor : Open example -> Technical tutorials -> Brushless IPM motor (embedded magnets) ->Computation of inductances and static torque. Another approach to simulate inductance in electric machines is to use the macro ComputeInductanceMatrix, you can load it into your project from Macros_Flux2D_Postproc. This is an advanced approach which uses frozen permeability techniques to take into account material saturation caused by the magnets where the machine is in working conditions. Please, notice that this macro can take some time to compute since it runs several simulations to accurately represents this physical phenomena. I hope the proposed documentation helps you with your inductance calculus. Please, do not hesitate to return to me in case you need further information about these topics. Best regards.
  8. Hello Andoni, In fact all these tutorials are included as examples in Flux Supervisor, you have all the documents (.ppt and .pdf) and the associated projects there. Best regards.
  9. Hello Ramesh, In my opinion, it is not surprising that end-ring resistance and inductance values have an influence over the machine electromagnetic torque. In fact, they have an impact over your rotor circuit and, therefore, over the currents/voltages flowing through it and these electrical variables will directly affect your torque. On the other hand, estimate accurate values for end-ring inductance is really tricky and measures are generally necessary. This is also true for the resistance but, in this case, you can always get an approach taking into account the length of the end-ring sections, its crossing area and its material’s resistivity. If you want some simulation examples regarding induction motors with squirrel cage (including end-ring resistance and inductance values), you will find them in Flux supervisor. More precisely: Open example ->Technical Tutorials -> Induction motor_1 and Open example ->Application notes -> Induction motor_2 for 2D and Open example ->Application notes -> Induction motor for 3D (see image attached). Best regards,
  10. Hello Samesp, Actually, you are facing this problem because the formulation required to solve the project has made some hypotheses than must be fulfilled, in particular about the simple connectivity of your domain. In other words, no “holes” are allowed in your domain. Please, find attached the files Connectivity_Problems.pdf which explains this problems a little more in detail. To solve it you have a dedicated menu that you can find in the data tree: Physics-> Regions -> Magnetic circuit cut. There, you can create the required cuts by yourself or let Flux to create them automatically, I advise you to choose the second option . Please, find attached an image showing this menu. Once the cuts have been created you should be able to solve your problem normally. Best regards. Connectivity_Problems.pdf
  11. You are welcome. In fact, this is a good possibility for a “new feature request”. Today, your best option if one commands takes a long time is to kill your Flux project manually and take advantage of the automatically generated python file to reproduce all the non-blocking steps that have been done after the last project opening. You will find this python file in your working directory under the name Flux2D_log.py (or Flux3D_log.py). Of course, you can edit it as any usual Python file, it is also worthwile to notice that only the finish commands are recorded there (i.e., the blocking command will not be there). Best regards.
  12. Hello. Unfortunately, the commands are treated one by one, so a new command cannot be run until the previous one was finished, this remains true if the commands is introduced through the GUI or a python instruction/file. In conclusion, you cannot choose to stop a command depending on how much time he spends to finish. Best regards.
  13. Hello Andoni, In fact, from the point of view of the excel file these parameters are not inputs but outputs. Their initial values are just an initial point in order to launch the optimization process managed by excel. On the other hand, from Flux point of view, you can find their meaning in page 7 in the document "ComputationIronLosses.pdf" I have sent you some weeks ago. Please, take into account that for the same losses inputs (usually taken from material datasheet) you can find different set of parameters that fit well with the losses curve of the material. The set of parameters you will obtained will depend mainly on the initial point (i.e., the initial values of k1, k2, k3 and a1, a2, a3). Sometimes is useful to play with these initial point to find the best set of Berttoti’s coefficients. It is also frequent that you want to fix some parameters (tipically the three exponents a1, a2 and a3) you can do so excluding these exponents from the excel. Additionally, you can established maximum and minimum value for these parameters during the fitting process. Please, find attached a .pdf which explains the meaning of these coefficients and their calculus process in the excel file (DetermineBertottiCoefficientsForIronLossesComputation.pdf) Hope this helps. Best regards. DetermineBertottiCoefficientsForIronLossesComputation.pdf
  14. Hello Andoni, Please, could you explain a little more how are you using the losses formula and what formula is it ? Is it the one which appears in page 7 in the pdf I have sent you some weeks ago? Are you applying it node by node or are you doing some assumptions? If Flux is delivering convincing values and they diverge from your approximations it seems that some disagreement between both models (numerical and analytical) exists. Best regards.
  15. Hello Holo, In fact, your problem is that you have made a mistake in the input power calculus. You have assumed that, as far as each current source provides 2100 W (mean value) the three will provide three times this quantity (this is, 6300 W). This is not the case, as far as you have a neutral cable (four cable connexion, taking into account the ground) the sources are consuming some of the energy produced by the others, in other words some of this power is coming back to the grid. If you plot the sum of the active power of the three phases in each instant you will obtain the next curve below, with a mean (ignoring the transitory part) of 4235 W. This match much better with your results. Anyway, I advise you to simulate some additional rotor positions in order to assure global accuracy. So you will have a power equilibrium as follows: P_input = 4235 W P_mec=4170 W (mean torque equal to 26.55 Nm) P_Joule= 13+45=58 W P_Berttoti=2.56*4 W=10.24 W The rule is: any quantity calculated directly from the regions should be multiplied by the number of portions in your periodicity but the quantity calculated from the circuit are not multiplied since the circuit already takes into account the periodicity in order to obtain the electric variables (i.e., currents and voltages). On the other hand, if you want to avoid the problem of the neutral in your circuit I advise you to try this connexion: As you can see, the third current source is no longer necessary as far as the other two are already imposing the current to the third branch. If, anyway, you want to represent it (e.g., because you are interested in its behaviour) you should do as follows: The resistance above should be a very high one, since this branch is only there to avoid that small differences between the current sources (i.e., numerical noise) leaves to inconsistent equations . For example a value for 1E7 is fine. Hope this helps. Best regards,
  16. Hello Holo, In fact, Berttoti losses are calculated taken into account the device length, since Berttoti approach deals with losses in the ferromagnetic volumes. However, losses are only calculated in the simulated part of the device: for example, if you are simulating only one pole of your four-poles machine and using periodicities your losses are equal to: 4200+4*400=5800W (not 4600 W, as you have said). I do not know what is your problem with “Power Balance”. Cand you send me this macro in order to check it? Best regards.
  17. You are welcome. Sorry but I misunderstood your initial question. Please, find attached some technical information explaining the formulas used to calculate iron losses through Berttoti method. You have also the possibility to use other estimation methods in order to apply Berttoti, but they are deprecated because of their lesser accuracy. Just one remark: losses are calculated taking into account all the nodes in your mesh and (if you are doing a transient simulation) for every instant of time. However, they are not calculated during the solving process but during the postprocessing. During the solving you are just stocking some variables in all the concerned nodes (i.e., the nodes appertaining to laminated regions) that then will be used to calculate iron losses using Berttoti. Best regards. ComputationIronLosses.pdf
  18. Hello. Unfortunately, there is no direct way to measure the angle between two lines. Depending on your project and your goals you can used different method to obtain this angle automatically: -Through a python file programming the trigonometric relation (since you have the coordinates of all the segment’s points). -You can also define a cylindric coordinate system centred in your line intersection. Best regards.
  19. Hello Imanol, In order to help you further, please, could you send me the ".log" and the ".report" files associated to this error? (you will file them in your working directory). Best regards,
  20. Hello Andoni, Actually, you have two examples in Flux supervisor about the way to simulate transformer. Within these examples complete documentation and the simulation process explained step by step are also included (image attached). Flux Supervisor (3D) -> Open Example -> Application notes -> Power transformer Flux Supervisor (3D) -> Open Example -> Application notes -> Three phases transformer Overlay Additionally, losses in transformer are not an easy matter since there are several different types: losses in the windings (Joule losses), losses in the magnetic core (iron losses) and also many kinds of “parasitic losses”, among them losses caused by induced currents in the tanks are specially relevant. Flux is able to calculate all this effect using different approaches: -From winding losses: They are Joule losses which are calculated taking advantage of the circuit context contained in Flux. The coupling between FEM simulation and circuit simulation allows to obtain the current density distribution within the coils and, using these values, obtain Joule losses. -From magnetic core losses: In the postprocessing options you have the possibility to calculate the iron losses using Berttoti (you should provide the material coefficients needed to use this approach). -Losses from induced currents in the tank: Induced currents are calculated during the magnetic simulation if the tank is defined as “magnetic conducting region”. From this currents the Joule losses can be evaluated using a sensor. For more information, please consult the supervisor examples I have talked you about. Hope this helps. Best regards.
  21. Hello, Holo. Anyway if you want more information about this kind of coupling you can find a complete example in Altair Flux supervisor, just go to 2D -> Open Example -> Brushless IPM motor -thermal analysis -> Magneto thermal Flux-Flux co-simulation (image attached) . It may be useful for you. Best regards.
  22. Hello Imanol, Unfortunately there is no command to clear the output window. Additionally, all the information that is displayed in this windows is also written in the ".log" project file. Anyway, I really think its length should not be a problem. More likely there exists another bug which is blocking your project. For further help, please send me the ".log" and the ".report" files associated with the project execution that crashes. Best regards.
  23. Hello Romain. Please, can you send me the log and report files corresponding with this instance of Flux? I hope once I got this information, I will able to help you. Best regards.
  24. Hello Twitzel, You cannot import directly a meshed current density because import at same time geometry, mesh and physics is not allowed. However, if I have understood you properly, you have several possibilities to perform the process you are thinking about. The simpler one is to use a non-meshed coil as source. It has the advantage to be simple to define, fast to calculate and it offers the possibility to define first all your geometry and do the coil just at the end, just as you have proposed. This method is specially suitable if you are interesting in your coil meanly as an external excitation; however, if you are studying important physical effects that take place inside the coil this is not the best approach. Another possibility is to define your geometry in Flux (or import it from another software) and then import coil’s geometry from a different CAD file. The limitation here is that both geometries must be unmeshed and you have to be careful in order to avoid possible collisions or undesired contact between the faces/volumes defined in different files. You can also define and mesh all your geometry in another software (Simlab or Hypermesh for example) and then import the geometry into Flux, where it will be treated. However, if you want to use this method it is necessary to define all your geometry at the same time, you cannot just import a meshed geometry a final step once you have defined previously the rest of your geometry (probable with its own mesh). As you see, you have plenty of choice, if depends of your project what of these approaches is the best for you. Hope this helps. Best regards.
  • Create New...