Read a GMSH model by physical groups¤

This live script is written as a guided walkthrough for a post-processing workflow. It focuses on retrieving, organizing, and visualizing model or response data after an OpenSees analysis. Read the text cells first, then run each code cell in order so that the variables, model state, and recorded results are available for the later sections.

This example demonstrates how to read a GMSH model by physical groups and convert it to an OpenSeesPy model using the Gmsh2OPS class.

This example is based on GMSH Example t15.

msh file can be found in .

clc; clear;

opsMAT = OpenSeesMatlab();
ops = opsMAT.opensees;

g2o = opsMAT.pre.Gmsh2OPS;

g2o.readGmshFile('utils/t15.msh');

Output

Info:: 5 Physical Names.
Info:: 166 Nodes; MaxNodeTag 166; MinNodeTag 1.
Info:: 571 Elements; MaxEleTag 571; MinEleTag 1.
Info:: Geometry Information >>>
43 Entities: 17 Point; 18 Curves; 7 Surfaces; 1 Volumes.


Info:: Physical Groups Information >>>
3 Physical Groups.
Physical Group names: Boundary, Load, Volume


Info:: Mesh Information >>>
166 Nodes; MaxNodeTag 166; MinNodeTag 1.
567 Elements; MaxEleTag 571; MinEleTag 5.

In the example above, we defined the following physical groups for converting OpenSees elements. Volume 1 is used to generate elements, while the boundary consists of the bottom 1 surface, 4 lines, and 4 points!

gmsh.model.addPhysicalGroup(dim=0, tags=[1, 2, 9, 13], tag=1, name=”Boundary”) # points
gmsh.model.addPhysicalGroup(dim=1, tags=[1, 8, 13, 17], tag=2, name=”Boundary”) # lines
gmsh.model.addPhysicalGroup(dim=2, tags=[18], tag=3, name=”Boundary”) # surface
gmsh.model.addPhysicalGroup(dim=2, tags=[27], tag=4, name=”Load”) # surface load
gmsh.model.addPhysicalGroup(dim=3, tags=[1], tag=4, name=”Volume”) # volume

physicalGroups = g2o.getPhysicalGroups();
disp(physicalGroups.keys());

Output

{'Boundary'}    {'Load'}    {'Volume'}

ops.wipe()
% Initialize a basic 3D model with 3 degrees of freedom per node
ops.model("basic", "-ndm", 3, "-ndf", 3)

% Define an elastic isotropic material
% Material ID: 1
% Elastic modulus: 3e7
% Poisson's ratio: 0.2
% Density: 2.55
matTag = 1;
ops.nDMaterial("ElasticIsotropic", matTag, 3e7, 0.2, 2.55)

% Create OpenSeesPy node commands based on all nodes
g2o.createNodeCmds();

% Create OpenSeesPy element commands for specific entities
% FourNodeTetrahedron elements
%
eleTags = g2o.createElementCmds(...
    "FourNodeTetrahedron", ...  % OpenSeesPy element type
    OpsEleArgs={matTag}, ...  % Additional arguments for the element (e.g., mat tag)
    PhysicalGroupNames="Volume");

% fixed nodes
fix_node_tags = g2o.getNodeTags(PhysicalGroupNames="Boundary");
for i = 1: numel(fix_node_tags)
    ops.fix(fix_node_tags(i), 1, 1, 1);
end

% If there are too many geometries on the boundary, you can iterate through and extract all lines and points on a geometry using the following commands:
boundary_dim_tags = g2o.getBoundaryDimTags(DimEntityTags=[2, 18], IncludeSelf=true);
disp(boundary_dim_tags);

Output

disp(physicalGroups("Boundary"));

Output

opsMAT.vis.plotModel();

Output

[OpenSeesMatlab] Model summary
  Nodes: 166
  Solid elements: 473

Eigen analysis:

tag = 1;
opsMAT.post.saveEigenData(tag, 10, solver='-genBandArpack');
eigenData = opsMAT.post.getEigenData(odbTag=tag);

opts = opsMAT.vis.defaultPlotEigenOptions;
opts.color.useColormap = true;
% opts.color.colormap = jet(256);
modeTags = 1:4;
cmps = "parula";
for i = 1:4
    subplot(2,2,i)
    ax = gca();
    opsMAT.vis.plotEigen(modeTags(i), eigenData, opts=opts, ax=ax);
    axis off;
    colormap(ax, cmps);
end

We can apply the load by extracting the elements belonging to surface with tag 27: First, we convert the element to which the surface load is applied to SurfaceLoad Element, and then we can apply the surface load to it:

pressure = -1;
load_ele_tags = g2o.createElementCmds(...
    "TriSurfaceLoad", ...  % OpenSeesPy element type
    OpsEleArgs={pressure}, ...  % Additional arguments for the element
    PhysicalGroupNames="Load");

Output

[OpenSees] TriSurfaceLoad element - Written: J. A. Abell (UANDES). Inspired by the makers of SurfaceLoad

load_ele_tags = num2cell(load_ele_tags);

ops.timeSeries("Linear", 1)
ops.pattern("Plain", 1, 1)
ops.eleLoad("-ele", load_ele_tags{:}, "-type", "-surfaceLoad")

ops.constraints("Transformation")
ops.numberer("RCM")
ops.system("BandGeneral")
ops.test("NormDispIncr", 1.0e-12, 6, 2)
ops.algorithm("Linear")
ops.integrator("LoadControl", 0.1)
ops.analysis("Static")
ops.analyze(10)