A Python FEM implementation.

N dimensional FEM implementation for M variables per node problems.

Viewer

FEMViewer

Installation

Use the package manager pip to install AFEM.

pip install AFEM

From source:

git clone https://github.com/ZibraMax/FEM
cd FEM
python -m venv .venv
python -m pip install build
python -m build
python -m pip install -e .[docs] # Basic instalation with docs

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

Full Docs

Tutorial

Using pre implemented equations

Avaliable equations:

• 1D 1 Variable ordinary diferential equation
• 1D 1 Variable 1D Heat with convective border
• 1D 2 Variable Euler Bernoulli Beams
• 1D 3 Variable Non-linear Euler Bernoulli Beams
• 2D 1 Variable Torsion
• 2D 1 Variable Poisson equation
• 2D 1 Variable second order PDE
• 2D 1 Variable 2D Heat with convective borders
• 2D 2 Variable Plane Strees
• 2D 2 Variable Plane Strees Orthotropic
• 2D 2 Variable Plane Strain
• 3D 3 variables per node isotropic elasticity

Numerical Validation:

• 1D 1 Variable ordinary diferential equation
• 1D 1 Variable 1D Heat with convective border
• 1D 2 Variable Euler Bernoulli Beams
• 1D 3 Variable Non-linear Euler Bernoulli Beams
• 2D 1 Variable Torsion
• 2D 1 Variable 2D Heat with convective borders
• 2D 2 Variable Plane Strees
• 2D 2 Variable Plane Strain

Steps:

• Create geometry
• Create Border Conditions (Point and regions supported)
• Solve!
• For example: Example 2, Example 5, Example 11-14

Example without geometry file (Test 2):

import matplotlib.pyplot as plt #Import libraries
from FEM.Torsion2D import Torsion2D #import AFEM Torsion class
from FEM.Geometry import Delaunay #Import Meshing tools

#Define some variables with geometric properties
a = 0.3
b = 0.3
tw = 0.05
tf = 0.05

#Define material constants
E = 200000
v = 0.27
G = E / (2 * (1 + v))
phi = 1 #Rotation angle

#Define domain coordinates
vertices = [
        [0, 0],
        [a, 0],
        [a, tf],
        [a / 2 + tw / 2, tf],
        [a / 2 + tw / 2, tf + b],
        [a, tf + b],
        [a, 2 * tf + b],
        [0, 2 * tf + b],
        [0, tf + b],
        [a / 2 - tw / 2, tf + b],
        [a / 2 - tw / 2, tf],
        [0, tf],
]

#Define triangulation parameters with `_strdelaunay` method.
params = Delaunay._strdelaunay(constrained=True, delaunay=True,
                                                                        a='0.00003', o=2)
#**Create** geometry using triangulation parameters. Geometry can be imported from .msh files.
geometry = Delaunay(vertices, params)

#Save geometry to .json file
geometry.exportJSON('I_test.json')

#Create torsional 2D analysis.
O = Torsion2D(geometry, G, phi)
#Solve the equation in domain.
#Post process and show results
O.solve()
plt.show()

Example with geometry file (Example 2):

import matplotlib.pyplot as plt #Import libraries
from FEM.Torsion2D import Torsion2D #import AFEM
from FEM.Geometry import Geometry #Import Geometry tools

#Define material constants.
E = 200000
v = 0.27
G = E / (2 * (1 + v))
phi = 1 #Rotation angle

#Load geometry with file.
geometry = Geometry.importJSON('I_test.json')

#Create torsional 2D analysis.
O = Torsion2D(geometry, G, phi)
#Solve the equation in domain.
#Post process and show results
O.solve()
plt.show()

Creating equation classes

Note: Don't forget the docstring!

Steps

1. Create a Python flie and import the libraries:

   from .Core import *
   from tqdm import tqdm
   import numpy as np
   import matplotlib.pyplot as plt

   • Core: Solver
   • Numpy: Numpy data
   • Matplotlib: Matplotlib graphs
   • Tqdm: Progressbars

2. Create a Python class with Core inheritance

   class PlaneStress(Core):
       def __init__(self,geometry,*args,**kargs):
       #Do stuff
       Core.__init__(self,geometry)

   It is important to manage the number of variables per node in the input geometry.

3. Define the matrix calculation methods and post porcessing methods.

   def elementMatrices(self):
   def postProcess(self):

4. The elementMatrices method uses gauss integration points, so you must use the following structure:

   for e in tqdm(self.elements,unit='Element'):
       _x,_p = e.T(e.Z.T) #Gauss points in global coordinates and Shape functions evaluated in gauss points
       jac,dpz = e.J(e.Z.T) #Jacobian evaluated in gauss points and shape functions derivatives in natural coordinates
       detjac = np.linalg.det(jac)
       _j = np.linalg.inv(jac) #Jacobian inverse
       dpx = _j @ dpz #Shape function derivatives in global coordinates
       for k in range(len(e.Z)): #Iterate over gauss points on domain
           #Calculate matrices with any finite element model
       #Assign matrices to element

A good example is the PlaneStress class in the Elasticity2D.py file.

Roadmap

1. 2D elastic plate theory
2. Transient analysis (Core modification)
3. Non-Lineal for 2D equation (All cases)
4. Testing and numerical validation (WIP)

Examples

References

J. N. Reddy. Introduction to the Finite Element Method, Third Edition (McGraw-Hill Education: New York, Chicago, San Francisco, Athens, London, Madrid, Mexico City, Milan, New Delhi, Singapore, Sydney, Toronto, 2006). https://www.accessengineeringlibrary.com/content/book/9780072466850

Jonathan Richard Shewchuk, (1996) Triangle: Engineering a 2D Quality Mesh Generator and Delaunay Triangulator

Ramirez, F. (2020). ICYA 4414 Modelación con Elementos Finitos [Class handout]. Universidad de Los Andes.

License

MIT