Skip to content
Snippets Groups Projects
Commit 1c80eb9c authored by Christophe Geuzaine's avatar Christophe Geuzaine
Browse files

up

parent 94cbc397
No related branches found
No related tags found
No related merge requests found
Pipeline #
......@@ -6,7 +6,7 @@
- Parameters shared by Gmsh and GetDp, and Onelab parameters
- FunctionSpaces for the 2D vector potential formulation
To compute the solution in a terminal:
To compute the solution in a terminal:
getdp electromagnet -solve MagSta_a
getdp electromagnet -pos Map
......@@ -15,29 +15,29 @@
Run (button at the bottom of the left panel)
------------------------------------------------------------------- */
/* Electromagnetic fields expand to infinity.
/* Electromagnetic fields expand to infinity.
The corresponding boundary condition can be imposed rigorously
by means of a gometrical transformation that maps a ring (or shell) of finite elements
to the complementary of its interior.
to the complementary of its interior.
As this is a mere geometric transformation,
it is enough in the model description to attribute a special jacobian
to the ring region ("AirInf"). See Jacobian{} section below.
With this information, GetDP is able to deal with the correct transformation
of all quantities involved in the model.
With this information, GetDP is able to deal with the correct transformation
of all quantities involved in the model.
The special jacobian "VolSphShell" has parameters.
The special jacobian "VolSphShell" has parameters.
There are 2 parameters in this case, "Val_Rint" and "Val_Rext",
which represent the inner and outer radii of the transformed ring region
and whose value must match those used
and whose value must match those used
in the geometrical description of the model (.geo file).
This is a typical case where Gmsh and GetDP must consistently share parameter values.
This is a typical case where Gmsh and GetDP must consistently share parameter values.
To ensure consistency in all cases, common parameters are defined
is a specific file "electromagnet_common.pro",
which is included in both the .geo and .pro file of the model.
is a specific file "electromagnet_common.pro",
which is included in both the .geo and .pro file of the model.
Besides sharing parameters between Gmsh and GetDP,
it is also useful to share some parameters (not all) with the user of the model,
i.e., to make them editable in the GUI before running the model.
i.e., to make them editable in the GUI before running the model.
Such variables are called Onelab variables (because the sharing mechanism
between the model and the GUI uses the Onelab interface).
Onelab parameters are defined with a "DefineNumber" statement,
......@@ -50,12 +50,12 @@ Group {
Air = Region[ 101 ]; Core = Region[ 102 ];
Ind = Region[ 103 ]; AirInf = Region[ 111 ];
Surface_ht0 = Region[ 1100 ];
Surface_ht0 = Region[ 1100 ];
Surface_bn0 = Region[ 1101 ];
Surface_Inf = Region[ 1102 ];
/* Abstract regions :
/* Abstract regions :
The purpose of abstract regions is to allow a generic definition of
the FunctionSpace, Formulation and PostProcessing fields
with no reference to model-specific Physical regions.
......@@ -80,51 +80,51 @@ Group {
Function {
mu0 = 4.e-7 * Pi;
murCore = DefineNumber[100, Name "Model parameters/Mur core",
Help "Magnetic relative permeability of Core"];
Help "Magnetic relative permeability of Core"];
nu [ Region[{Air, Ind, AirInf}] ] = 1. / mu0;
nu [ Core ] = 1. / (murCore * mu0);
NbTurns = 1000 ;
Current = DefineNumber[0.01, Name "Model parameters/Current",
Help "Current injected in coil [A]"];
Help "Current injected in coil [A]"];
Js_fct[ Ind ] = -NbTurns*Current/SurfaceArea[];
/* The minus sign is to have the current in -e_z direction,
so that the magnetic induction field is in +e_y direction */
/* The minus sign is to have the current in -e_z direction,
so that the magnetic induction field is in +e_y direction */
}
/* In the 2D approximation, the magnetic vector potential A and the current density Js
are not scalars, but vectors with a z-component only:
are not scalars, but vectors with a z-component only:
A = Vector [ 0, 0, az(x,y,t) ]
Js = Vector [ 0, 0, jsz(x,y,t) ]
In order to compute derivatives and apply geometric transformations adequately,
In order to compute derivatives and apply geometric transformations adequately,
GetDP needs this information.
Regarding discretization, now, A is node-based, whereas Js is region-wise constant.
Considering all this, one ends then up with the FunctionSpaces
"Hcurl_a_Mag_2D" and "Hregion_j_Mag_2D" as they are defined below.
Regarding discretization, now, A is node-based, whereas Js is region-wise constant.
Considering all this, one ends then up with the FunctionSpaces
"Hcurl_a_Mag_2D" and "Hregion_j_Mag_2D" as they are defined below.
The function space "Hregion_j_Mag_2D" provides one basis function,
and hence one degree of freedom, per physical region in the abstract region "Vol_Js_Mag".
The constraint "SourceCurrentDensityZ" fixes all these dofs,
The constraint "SourceCurrentDensityZ" fixes all these dofs,
so the FunctionSpace "Hregion_j_Mag_2D" is fully fixed and has no FE unknowns.
One could thus have replaced it by a simple function
One could thus have replaced it by a simple function
and the Galerkin term would have been
Galerkin { [ Vector[ 0,0,-Js_fct[] ] , {a} ]; In Vol_Js_Mag;
Jacobian Vol; Integration Int; }
instead of
instead of
Galerkin { [ - Dof{js} , {a} ]; In Vol_Js_Mag;
Jacobian Vol; Integration Int; }
Thechosen implementation below is however more effeicient
Thechosen implementation below is however more effeicient
as it avoids evaluating repeatedly the function Js_fct[] during assembly.
*/
Constraint {
{ Name Dirichlet_a_Mag;
{ Name Dirichlet_a_Mag;
Case {
{ Region Sur_Dir_Mag ; Value 0.; }
}
......@@ -164,7 +164,7 @@ FunctionSpace {
}
Include "electromagnet_common.pro";
Include "electromagnet_common.pro";
Val_Rint = rInt; Val_Rext = rExt;
Jacobian {
{ Name Vol ;
......@@ -179,11 +179,11 @@ Integration {
{ Name Int ;
Case { {Type Gauss ;
Case { { GeoElement Triangle ; NumberOfPoints 4 ; }
{ GeoElement Quadrangle ; NumberOfPoints 4 ; }
{ GeoElement Quadrangle ; NumberOfPoints 4 ; }
}
}
}
}
}
}
Formulation {
......@@ -215,26 +215,31 @@ Resolution {
PostProcessing {
{ Name MagSta_a_2D; NameOfFormulation Magnetostatics_a_2D;
Quantity {
{ Name a;
Value {
Local { [ {a} ]; In Dom_Hcurl_a_Mag_2D; Jacobian Vol; }
{ Name a;
Value {
Local { [ {a} ]; In Dom_Hcurl_a_Mag_2D; Jacobian Vol; }
}
}
{ Name az;
Value {
{ Name az;
Value {
Local { [ CompZ[{a}] ]; In Dom_Hcurl_a_Mag_2D; Jacobian Vol; }
}
}
{ Name b;
Value {
{ Name b;
Value {
Local { [ {d a} ]; In Dom_Hcurl_a_Mag_2D; Jacobian Vol; }
}
}
{ Name h;
Value {
{ Name h;
Value {
Local { [ nu[] * {d a} ]; In Dom_Hcurl_a_Mag_2D; Jacobian Vol; }
}
}
{ Name js;
Value {
Local { [ {js} ]; In Dom_Hcurl_a_Mag_2D; Jacobian Vol; }
}
}
}
}
}
......@@ -242,7 +247,7 @@ PostProcessing {
e = 1.e-5;
h = 0.02;
p1 = {e,h,0};
p2 = {0.25-e,h,0}; // horizontal cut through model, just above x-axis.
p2 = {0.25-e,h,0}; // horizontal cut through model, just above x-axis.
PostOperation {
......@@ -252,7 +257,10 @@ PostOperation {
"View[l].IntervalsType = 1;",
"View[l].NbIso = 40;"],
File "tmp.geo", LastTimeStepOnly] ;
Print[ a, OnElementsOf Dom_Hcurl_a_Mag_2D, File "a.pos" ];
Print[ js, OnElementsOf Dom_Hcurl_a_Mag_2D, File "js.pos" ];
Print[ az, OnElementsOf Dom_Hcurl_a_Mag_2D, File "az.pos" ];
Print[ b, OnElementsOf Dom_Hcurl_a_Mag_2D, File "b.pos" ];
Print[ b, OnLine{{List[p1]}{List[p2]}} {50}, File "by.pos" ];
}
}
......
This directory contains ONELAB tutorials.
See the corresponding wiki (http://gitlab.onelab.info/doc/tutorials/wikis/home)
See the corresponding wiki at http://gitlab.onelab.info/doc/tutorials/wikis/home
as well as comments in the .pro and .geo files for more information.
Additional models can be found on http://gitlab.onelab.info/doc/models.
Additional examples highlighting various physical models and numerical
techniques can be found on http://gitlab.onelab.info/doc/models/wikis/home.
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment