An easy to use elastic 3D structural engineering finite element analysis library for Python.
The easiest way to install Pynite is with pip: pip install PyniteFEA
For a more detailed discussion on installation options and dependencies see https://pynite.readthedocs.io/en/latest/installation.html
- 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.
- Automatic handling of internal nodes on frame members (physical members).
- Tension-only and compression-only elements.
- Spring elements: two-way, tension-only, and compression-only.
- Spring supports: two-way and one-way.
- Quadrilateral plate elements (DKMQ formulation).
- Rectangular plate elements (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 difficult 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: PyNite is getting better at what it does. Each new feature provides leverage to build upon previous features in more elaborate ways. Key to improvement is (a) maintaining the core features that other features rely on, and (b) adding new features that are solid stepping stones to other features. Improvement most often happens by getting the small and simple things right incrementally, rather than with sweeping overhauls all at once.
-
Collaboration: If you see a way to improve Pynite, you are encouraged to contribute. There are many simple ways to contribute that don't take much effort. Issue reports and pull requests can be very helpful. One easy way to contribute is to add to or improve the library of simple example problems in the
Examples
folder. Please keep them relatively simple. Most users learn Pynite from following these simple examples. Another way to help Pynite without having to know too much about its internal workings is to help with the documentation files in thedocs\source
folder of this repository. These files help new users learn Pynite. If you are able to make a bigger commitment, and would like to become a regular contributor to the project, please reach out about becoming a collaborator.
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.
- Documentation is a work in progress and can be found on readthedocs here: https://pynite.readthedocs.io/en/latest/index.html.
- 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 [email protected] to discuss options.
Here's a list of projects that use PyNite:
- Building Code (https://building-code.herokuapp.com/) - This one is my personal side project.
- Standard Solver (https://www.standardsolver.com/)
- Civils.ai (https://civils.ai/1/free-3D-finite-element-structural-analysis)
- Phaenotyp (https://github.com/bewegende-Architektur/Phaenotyp) (https://youtu.be/shloSw9HjVI)
v0.0.97
- Fixed physical member load and deflection diagrams. Physical members are a newer feature. Member internal results were being reported correctly, but the diagrams for these members had not been revised to plot correctly. The old method for plain members was still being used. Physical members were not considering that a physical member was made from multiple submembers, and results for each span needed to be combined to get the whole plot.
- Switched some commonly used python libraries to be installed by default with
Pynite
. MostPynite
users will want these libraries installed for full-featured use ofPynite
. These libraries help withPynite
visualizations, plotting, the sparse solver, andJupyter Lab
functionality. This is just easier for new python users. I was getting a lot of questions about how to set up libraries, and this takes the guesswork away. This is part ofPynite's
objective to stay easy to use.
v0.0.96
- Changed quad elements from MITC4 formulation to DKMQ formulation. This greatly improves plate results at corners and increases the speed with which the plate's stiffness matrix is assembled. MITC4 element code has been retained as legacy code, but is no longer used by the program.
- Breaking Changes: Implemented snake-case for dictionary names (e.g.
FEModel3D.Nodes
is nowFEModel3D.nodes
). These changes were made to preparePynite
for a v1.0 release that is consistent with thePEP8
style guide forpython
. - Bug fix for tension/conpression-only member internal results. While global results were correct, member internal results were showing results from the first tension/compression only iteration.
- Member results arrays can now be customized to pick up user defined points. Member results arrays generate results much faster now too.
v0.0.95
- Bug fix for rendering negative point loads via
Pyvista
. They were being rendered as positive loads. The analysis was not impacted by this bug.
v0.0.94
- Added rendering via
Pyvista
. This greatly simplified the rendering code and provided a fresh look to the rendereings. Renderings in jupyter are now interactive. Global axes are also now shown in rendereings. To usePyvista
instead ofVTK
, use the newRendering
library rather than the oldVisualization
library. Rendering viaVTK
directly is still available. - Bug fix for member self-weight. The program was throwing exceptions instead of calculating member self-weight. Added a unit test to help prevent this issue from occuring again as code changes.
- Refactored
material
to bematerial_name
in the code. The prior naming convention caused confusion which led to the self-weight bug.
v0.0.93
- Fixed phantom reactions showing up at unsupported nodes. If there was a support defined at a node, the program was summing reactions for all directions at the node, rather than just the supported directions. This caused the program to report "extra" reaction directions at any supported node (if the user queried them). Element forces/stresses were not affected as this was a post-processing reaction summing issue. Reactions for supported directions were summed correctly, except in the case of nodes with both spring supports and other supports. Only unsupported directions, and nodes with both spring supports and other supports, were showing phantom reactions. This bug also caused statics checks to fail from time to time.
- Reorganized physical member code to match member code more consistently.
v0.0.92
- Added member self-weight calculations via
FEModel3D.add_member_self_weight()
. This only applies to members. This feature does not calculate self-weight for plate and quad elements.
v0.0.89-0.0.91
- Migrating visualizaton code from VTK to PyVista. PyVista greatly simplifies the rendering code, and simplifies adding new features to the renderings. This feature is only partially complete and partially functional.
v0.0.88
- Reorganized physical member code to match member code more consistently.
v0.0.87
- Fine-tuned P-$\delta$ effects. P-$\delta$ effects are now included in member internal slope and deflection calculations.