Simple Finite Element Analysis in Python
An easy to use elastic 3D structural engineering finite element analysis library for Python.
- 3D static analysis of elastic structures.
- P-Δ analysis of frame type structures.
- Member point loads, linearly varying distributed loads, and nodal loads are supported.
- Classify loads by load case and create load combinations from load cases.
- Produces shear, moment, and deflection results and diagrams for each member.
- Tension-only and compression-only elements.
- Spring elements: two-way, tension-only, and compression-only.
- Spring supports
- Quadrilateral plate elements (based on an isoparametric formulation).
- Rectangular plate elements (based on a 12-term polynomial formulation).
- Basic meshing algorithms for some common shapes and for openings in rectangular walls.
- Reports support reactions.
- Rendering of model geometry, supports, load cases, load combinations, and deformed shapes.
- Generates PDF reports for models and model results.
As I've gotten into the structural engineering profession, I've found there's a need for an easy to use open-source finite element package. I hope to help fill that need by prioritizing the following:
Accuracy: There are no guarantees PyNite is error free, but accuracy and correctness are a priority. When bugs or errors are identified, top priority will be given to eliminate them. PyNite's code is frequently reviewed, and its output is tested against a suite of textbook problems with known solutions using continuous integration (CI) anytime a change to the code base is made. If you do happen to find an error, please report it as an issue.
Simplicity: There are other finite element alternatives out there with many more capabilities, but they are often lacking in documentation, written in outdated languages, or require extensive knowledge of finite element theory and/or element formulations to use. PyNite is not intended to be the most technically advanced solver out there. Rather, the goal is to provide a robust yet simple general purpose package.
Improvement: I plan to continue supporting PyNite for many years to come. There are a lot of pieces I'd like to add to PyNite going forward. There's a lot of potential to create extensions as well to solve all kinds of engineering problems. There are more problems to solve than I have time for, so some priorities will have to be made. The plan is to keep PyNite mainstream, adding core functionality first. Occasionally however I may just add what interests me at the time.
Collaboration: The intent is to keep PyNite free and open source. This will encourage future development and contributions. Keeping it open source will allow anyone to inspect and improve the code it runs on. If you see an area you think you can help PyNite improve in you are encouraged to contribute. I'd like to get PyNite doing a lot more. Don't be offended if I'm a little slow to accept your contributions. FEA is a very technical subject and accuracy is extremely important to me. Sometimes I'm a little slow understanding both FEA and Python and it takes some time for me to comprehend what's being proposed. I also have a young family to take care of that takes first priority.
Whether you just need help getting started with PyNite, or are looking to build something more complex, there are a few resources available:
- The examples in the "Examples" folder in this repository cover a variety of simple problems. The comments in the examples provide additional guidance on how PyNite works.
- The wiki in this repository provides basic guidance on how to use PyNite, although it has fallen behind the development of the program.
- If you're looking for more direct guidance on using PyNite, or for help coding a project, I am available on a private consulting basis. You can reach out to me directly at Building.Code@outlook.com to discuss options.
PyNite depends on the following packages:
- numpy: used for matrix algebra and dense matrix solver
- scipy: used for sparse matrix solver to improve solution speed
- matplotlib: used for plotting member diagrams
- PrettyTable : used to format tabular output
- VTK: Used for visualization - Note that VTK is a little picky about which version of Python you are running. You must run a 64 bit installation of Python, rather than a 32 bit version. VTK is published by Kitware. I've noticed Kitware takes a little time updating VTK to be compatible anytime a new version of Python is released. If you're having trouble installing VTK, you can see which versions of Python are supported by visiting https://pypi.org/project/vtk/#files. VTK does not need to be installed if you don't plan to use the visualization tools built into PyNite.
- PDFKit: Used for generating pdf reports. In order to generate pdf reports, PDFKit requires you to have wkhtmltopdf installed on your computer. This is a free program available for download at https://wkhtmltopdf.org/downloads.html. Once installed, you'll need to help PyNite find it. On Windows, this can be done by setting your PATH environment variable to include the path to "wkhtmltopdf.exe" after installation. For example, mine is installed at "C:\Program Files\wkhtmltopdf\bin"
- IPython: Used for displaying screenshots from VTK.
- jinja2: Used for templating reports into HTML prior to HTML-to-pdf conversion.
- jupyterlab: Only needed if you want to view the derivations used to build PyNite.
- sympy: Only needed if you want to view the derivations used to build PyNite.
- Major speed boost. A lot of time was being wasted reading/writing to a SciPy
lil_matrix format. The stiffness matrix is now assembled as a
coo_format first, and then switched to lil format later on to facilitate slicing.
- Refactored node names to match PEP8 style guide. Instead of
jNode, etc. PyNite now uses
j_node etc. This should only affect you if you were using the keyword arguments for node names.
- Removed test modules from package distribution. These files were unnecessary for users to have installed on their computers.
v0.0.48 - Global Load Directions
- Global load directions can now be used for member loads now. Use capital notation to apply the member load in the global X, Y, or Z direction (e.g. 'FX', 'MY', 'FZ'...). Use lower case notation to apply the member load in the member's local x, y, or z direction (e.g. 'Fx', 'My', 'Fz'...).
- Bug fix for rendering models without plates/quads (special thanks to tamalone1 for this pull request!)
v0.0.47 - Bug Fix After Refactoring
- Fixed a bug identified in Issue #98 caused by refactoring in v0.0.44.
v0.0.46 - Rendering Improvements
- The size of the scalar bar for plate contours is now a little more predictable. It used to be based off of the window size. Now it only resizes when the window gets really small.
- The user now has more control over screenshots. If a screenshot is requested, it will be captured when
the user closes the render window.
v0.0.45 - Rectangular Mesh Openings
- Openings can now be added to rectangular meshes. An example of a shear wall with openings has been added to the
Examples folder to help you get started.
v0.0.44 - PEP8 Style Updates
- Many method and function names have been renamed to match the PEP8 style guide for Python. For now PyNite is backwards compatible, but deprecation warnings have been implemented for function names that will be disappearing in a future release. If you see deprecation warnings when you run your code, it's recommended to move your code over to the new method and function names given in the deprecation warnings. This is a big change that will affect most users. I don't plan to make changes this big very often, but PyNite was getting large and complex enough that I had to do it now or never. When I started PyNite I didn't know much about Python coding styles. This change brings the code into better conformance with normal conventions in Python.
ClearLoads method (
delete_loads going forward) wasn't deleting plate and quad loads. Fixed this issue.
Here's a list of projects that run on PyNite: