Extended version of Modelica.Magnetic.FluxTubes (Revision 7955)

[modelica-trac-importer]

Reported by jziske on 5 Dec 2014 17:13 UTC
Uploaded extended version of Modelica.Magnetic.FluxTubes for testing [7955].

The Extension of the Modelica.Magnetic.FluxTubes library contains new elements for modeling ferromagnetic (static) and eddy current (dynamic) hysteresis effects and permanent magnets. The FluxTubes.Material package is also extended to provide hysteresis data for several magnetic materials. These data is partly based on own measurements. For modeling of ferromagnetic hysteresis, two different hystereses models have been implemented: The simple Tellinen model and the considerably more complex Preisach hysteresis model. The following packages have been added:

• FluxTubes.UsersGuide.Hysteresis
• Fluxtubes.Examples.Hysteresis
• FluxTubes.Shapes.HysteresisAndMagnets
• FluxTubes.Material.HysteresisEverettParameter
• FluxTubes.Material.HysteresisTableData

Testing of the new models and constructive feedback is highly appreciated. The extensions are intended for inclusion into MSL after testing. Care should be taken with initialization of the rather complex Preisach hysteresis model.

---

Migrated-From: https://trac.modelica.org/Modelica/ticket/1620

[modelica-trac-importer]

Comment by christiankral on 16 Dec 2014 19:24 UTC

Cuboid vs Generic Models

The names of the models included in FluxTubes.Shapes.HysteresisAndMagnets suggest a rectangular shape of the area of cross section. However, the geometry parameters are implemented inconsistently and the models do not really assume rectangular shapes.

A) Area of cross section defined by a and b such that area = a*b

• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystTellinenSoft
• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystTellinenEverett
• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystTellinenTable
• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystPreisachEverett

B) Area of cross section defined by one parameter

• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystTellinenHard
• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystTellinenPermanentMagnet
• FluxTubes.Shapes.HysteresisAndMagnets.CuboidLinearPermanentMagnet

No generic models are included. Why? How shall a user consider other than rectangular areas of cross sections? Determine equivalent lengths of a rectangle?

I think it makes a lot more sense to only have generic models, and omit the rectangular structures, since rectangular areas of cross sections can be calculated by area = a*b without stressing a user.

Model parameter names must be unified

I also wonder why these model distinguish different names for the area of cross section

• Apm
• Afe
  and the lengths
• l
• Lfe
  There is no need for different names. On important argument on unified parameter names is the parameter compatible exchange of models.

The names shall be unified to

• A or
• area
  The length names shall be unified to
• l

SI units and/or documentation strings are not used in models

In some models variables are declared as Real even though they shall be SIunits.

Models affecty by this issues (as far as I identified them):

• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystTellinenPermanentMagnet
  • MagRel, dHyst, hystR, hystF,...
• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystPreisachEverett
  • H1,H2,H3,H4,i_delAsc,i_delDesc,i_del,i_init,i_changeAsc,v,...
• FluxTubes.Shapes.HysteresisAndMagnets.CuboidHystTellinenEverett
  • H2,H3,P2,P3,H_lim,...
• FluxTubes.Material.HysteresisEverettParameter.BaseData
  • ... all parameters except for sigma
• FluxTubes.Material.HysteresisTableData.BaseData
  • tabris, tabfal

Missing experiment annotation

• FluxTubes.Examples.Hysteresis.InductorWithHysteresis
• FluxTubes.Examples.Hysteresis.SinglePhaseTransformerWithHysteresis1
• FluxTubes.Examples.Hysteresis.SinglePhaseTransformerWithHysteresis2

It is also preferred to use generic experiment annotations instead of Dymola specific annotations.

Compatibility with OpenModelica

The following examples does not run in OpenModelica.

• SinglePhaseTransformerWithHysteresis1
• SinglePhaseTransformerWithHysteresis2
• ThreePhaseTransformerWithRectifier

Please check library code for compatibility with Modelica Specification 3.2.

To be moved to MSL

The models

• FluxTubes.Utilities.Delay (block, based on Modelica delay operator)
• FluxTubes.Utilities.FloatingAverage
  shall rather be moved to Modelica.Blocks.Math, since these functions are to generic to be located in Modelica.Magnetic.FluxTubes.Utilities

[modelica-trac-importer]

Comment by ahaumer on 21 Sep 2015 20:54 UTC
There are two issues left (regarding the version sent by mail 2015-09-16):

1. Modelica_Magnetic_FluxTubes.Examples.Hysteresis.HysteresisModelComparison and Modelica_Magnetic_FluxTubes.Examples.Hysteresis.ThreePhaseTransformerWithRectifier have a double "experiment" annotation - please remove one of them.
2. Some models do not handle units in a clean way (although Dymola doens't complain about that), especially:
   Shapes.HysteresisAndMagnets.GenericHystTellinenSoft
   Shapes.HysteresisAndMagnets.GenericHystTellinenHard
   Shapes.HysteresisAndMagnets.GenericHystTellinenEverett
   Shapes.HysteresisAndMagnets.GenericHystPreisachEverett
   Let's have a look at the first one:

  parameter SI.MagneticFluxDensity Js = 1.8 "Saturation polarization";
  parameter SI.MagneticFluxDensity Br=0.9 "Remanence";
  parameter SI.MagneticFieldStrength Hc=120 "Coercitivity";
  parameter Real K=1 "mu_0 multiplier";
  extends Interfaces.PartialGenericHysteresisTellinen(mu0=K*mu_0);
protected 
  parameter SI.MagneticFluxDensity eps=Js/1000;
  parameter Real H0=0.5*Modelica.Math.log((1 + Br/Js)/(1 - Br/Js));
  parameter Real M=H0/Hc;
equation 
  hystR = Js*tanh(M*Hstat - H0) + mu0*Hstat - eps/2;
  hystF = Js*tanh(M*Hstat + H0) + mu0*Hstat + eps/2;

To my understanding, this should be changed to something like that:

  parameter SI.MagneticFluxDensity Js = 1.8 "Saturation polarization";
  parameter SI.MagneticFluxDensity Br=0.9 "Remanence";
  parameter SI.MagneticFieldStrength Hc=120 "Coercitivity";
  parameter Real K=1 "mu_0 multiplier";
  extends Interfaces.PartialGenericHysteresisTellinen(mu0=K*mu_0);
protected 
  parameter SI.MagneticFluxDensity eps=Js/1000;
  parameter SI.MagneticFieldStrength H0=0.5*Modelica.Math.log((1 + Br/Js)/(1 - Br/Js));
  parameter Real M=H0/Hc;
  parameter SI.MagneticFieldStrength unitH=1;
equation 
  hystR = Js*tanh((M*Hstat - H0)/unitH) + mu0*Hstat - eps/2;
  hystF = Js*tanh((M*Hstat + H0)/unitH) + mu0*Hstat + eps/2;

[modelica-trac-importer]

Comment by ahaumer on 23 Sep 2015 13:12 UTC
Previous issue resolved, 2 left:

All initial conditions in Exmaples have to bet set uniquely, but in Examples.Hyteresis.HyteresisModelComparison protected discrete non-states {aSavI, bI, bSavI, hmaxI} are not.

Every component has to be tested at least in 1 example. Is this true for {GenericHystTellinenHard, GenericHystTellinenEverett, GenericHytTellinenPermanentMagnet, GenericLinearPermanentMagnet}?

[modelica-trac-importer]

Comment by beutlich on 2 Nov 2015 19:04 UTC
78ba858 went to MSL trunk.

[modelica-trac-importer]

Comment by thomasboedrich on 22 Dec 2015 14:14 UTC
Replying to [comment:3 ahaumer]:

Previous issue resolved, 2 left:

All initial conditions in Exmaples have to bet set uniquely, but in Examples.Hyteresis.HyteresisModelComparison protected discrete non-states {aSavI, bI, bSavI, hmaxI} are not.

The abovementioned issue has already been resolved in 78ba858 uploaded to MSL trunk.

Every component has to be tested at least in 1 example. Is this true for {GenericHystTellinenHard, GenericHystTellinenEverett, GenericHytTellinenPermanentMagnet, GenericLinearPermanentMagnet}?

Test models for Magnetic.FluxTubes (for ModelicaTest) will be provided in January 2016.

[modelica-trac-importer]

Comment by thomasboedrich on 11 Jan 2016 16:56 UTC
A test library for Magnetic.FluxTubes has been added to ModelicaTest (5bfb31b).

[modelica-trac-importer]

Modified by thomasboedrich on 11 Jan 2016 16:57 UTC