Written and maintained by CASRAI Editorial Board
Published
On this page
OpenSees (the Open System for Earthquake Engineering Simulation) is an open-source software framework for simulating how structural and geotechnical systems respond to hazards such as earthquakes. It was built as the computational platform for performance-based earthquake engineering research at the Pacific Earthquake Engineering Research Center (PEER), and researchers drive it through scripts and plain-text input files rather than a graphical interface.
Short answer
OpenSees is a finite element framework, not a single fixed program. Its official documentation describes it as a software framework for simulating the dynamic response of structural and geotechnical systems, originally the research platform for performance-based earthquake engineering at PEER and now used for fire, wind, earthquake and wave-action (tsunami or storm surge) problems. It is free for educational, research and non-profit use under the University of California licence, with separate permission needed to put it into commercial products. For research practice, the points that matter are which version and interpreter produced a result, which licence route applies, and whether the input scripts are archived with the data.
Definition and scope
The OpenSees wiki describes the software as an object-oriented, open-source framework for finite element analysis that can run as serial or parallel applications. Its core is written mainly in C++, with Fortran and C numerical libraries. Initial core funding came from the PEER Center, supported by the US National Science Foundation; the wiki states that no current support is provided from that source.
Because it is a framework, an analyst does not click through menus. Users work through interpreter applications that accept a scripting language, and a model is defined either in plain-text input files or in programs that call the framework. That design is why OpenSees appears in research papers as a method component, in the same way a statistical package does, and why the exact input script is part of the scientific record. For the field context, see our overview of what civil engineering is, and for the community and standards side of simulation work, the guide to NAFEMS.
What the framework models
The official documentation page names structural and geotechnical systems, and the PEER-originated purpose of simulating response to earthquakes. Its stated hazard coverage now also includes fire, wind, and wave action from tsunami or storm surge. The documentation is organised into user-facing sections (OpenSees Interpreters, Installation, Command Manual, Examples, Questions, Bugs and Feature Requests, and Copyright and License) and developer-facing sections (Source Code, Building Application, References, Extending Application, Contributing Code and Documentation, and Issues).
How OpenSees is used in practice
The Berkeley OpenSees site presents the software as a framework for developing applications, with the source code on GitHub, a bug tracker, a wiki, a community message board and a section on parallel computing. The GitHub repository holds the source for all revisions since version 2.3.2, with build documentation for Windows, macOS and Ubuntu. A version 3.8.0 binary release was announced on the Berkeley site in February 2026.
In day-to-day research the workflow has a recognisable shape. A researcher writes an input script that defines the model, runs it through an interpreter, and reads the output files. The documentation’s Command Manual and Examples sections are the reference for what each command does. Because the framework is scripted, a published analysis can in principle be re-run by someone else who has the same version and the same script. That is a computational reproducibility property, and it only holds if the script, the version and the interpreter are recorded. Our dictionary entry on computational reproducibility sets out the general concept.
Serial and parallel runs
The wiki states that OpenSees supports both serial and parallel applications, and the Berkeley site has a parallel computing section. This matters for reporting: a result produced by a parallel build should say so, because the application used is part of the method description.
Interpreters and related concepts
An interpreter is the application that reads your script and drives the OpenSees core. The documentation has an OpenSees Interpreters section for this. One interpreter route is OpenSeesPy, which its own documentation describes as a Python 3 interpreter of OpenSees. It is distributed on PyPI for Windows, Linux and macOS; its documentation lists version 3.8.0.0 (including macOS ARM64) and states that Python 3.12 is required. The documentation’s minimal script imports the module, clears any previous model, defines a two-dimensional model with three degrees of freedom per node, and prints it:
import openseespy.opensees as opsops.wipe()ops.model('basic','-ndm',2,'-ndf',3)ops.printModel()
OpenSeesPy organises its commands into groups: model, analysis, output, utility, FSI, sensitivity, reliability, parallel, and pre- and post-processing. The Berkeley site also mentions SimCenter tools for uncertainty quantification as part of the wider ecosystem around the framework.
Framework, interpreter and source are different things
The data table below compares three routes into OpenSees as the official pages describe them: the core application and its scripting interpreters, OpenSeesPy, and building from the GitHub source. Each has its own documentation, and the licence wording is not identical across them, which is the part most often overlooked.
Licence and citation
The OpenSees copyright page states that the software is copyright The Regents of the University of California. It grants educational, research and non-profit entities permission to use, reproduce, modify and distribute the software for noncommercial purposes only, without charge and without a written licence agreement. Other entities may use, reproduce and modify it for internal purposes only. Permission to incorporate the software into commercial products must be obtained from the University of California, which the page directs to its technology licensing office. The copyright notice and the paragraphs that follow it must appear in all copies and modifications of the software and documentation.
The same page states that the software and documentation are supplied as is, without accompanying services, that the university does not warrant uninterrupted or error-free operation, and that it disclaims implied warranties of merchantability and fitness for a particular purpose. The practical consequence for researchers is that validation of a model is the analyst’s responsibility and should be documented as such.
OpenSeesPy adds its own statement: it is free for research, education and internal use, while commercial redistribution, including cloud services that use import openseespy, requires a commercial licence from UC Berkeley. If your group runs analyses as a service for third parties, read that clause before building on it.
Citing OpenSees
The official documentation page we read does not give citation instructions. One secondary source, a DesignSafe user guide surfaced in search results, reportedly asks users of its hosted OpenSees to cite OpenSees and a 2017 paper by Rathje and Dawson; check the current instructions on the platform you actually run on before copying that wording. General software-citation practice is covered by the software citation working group entry in our dictionary.
Where it matters in research workflows
For a research office, OpenSees raises questions that are not about structural mechanics at all.
- Reproducibility. A result is only repeatable if the software version, the interpreter, and the input script are recorded. The Berkeley site and OpenSeesPy documentation both name version 3.8.0 as current on the dates we read them; other versions exist in the GitHub history back to 2.3.2.
- Licence fit. Academic research falls inside the noncommercial grant. A university spin-out, a consultancy or a product team falls under the “other entities, internal purposes” wording or the commercial-incorporation clause, which is a licensing question rather than a technical one.
- Data and script archiving. Simulation inputs and outputs belong with the experimental data they were compared against. The guide to the DesignSafe-CI and NHERI data repository covers the natural-hazards engineering cyberinfrastructure where such files are shared, and our engineering reports guide covers how the analysis gets written up.
- Funding history. The wiki records NSF-sponsored PEER Center support for the initial core and says no current support is provided from it, so sustainability of any given feature is a community and developer matter rather than a funded service.
Common misunderstandings
OpenSees is not only for earthquakes
The name points to earthquake engineering, and that is the origin, but the official documentation now lists fire, wind and wave-action hazards as well. The framework is a general finite element platform for structural and geotechnical response.
Open source does not mean unrestricted
The source is public and anyone can read it on GitHub, but the copyright page limits use and distribution by user type and purpose. “Free” here means no charge for the stated noncommercial and internal uses, not a licence to repackage the software.
OpenSeesPy is not a separate analysis engine
OpenSeesPy is described by its documentation as a Python 3 interpreter of OpenSees. It gives access to the same framework from Python; it is not an independent code, and its version number should be reported alongside the OpenSees version.
It has no graphical front end by itself
The wiki describes use through interpreter applications with scripting. Pre- and post-processing are command groups in OpenSeesPy and are otherwise handled by other tools; do not assume a visual model builder exists in the core framework.
Support is not guaranteed
The copyright page says the software is supplied without accompanying services and with no obligation to update it. Help comes through the community message board and the bug tracker described on the Berkeley site.
Frequently asked questions
What does OpenSees stand for?
It stands for the Open System for Earthquake Engineering Simulation, per the official OpenSees documentation.
Is OpenSees free to use?
For educational, research and non-profit entities, the copyright page grants use, reproduction, modification and distribution for noncommercial purposes without charge. Other entities may use it for internal purposes only, and incorporating it into commercial products needs permission from the University of California.
What language is OpenSees written in?
The wiki states the core is written mainly in C++, with Fortran and C numerical libraries.
What is OpenSeesPy?
It is a Python 3 interpreter of OpenSees, installable from PyPI, and its documentation lists version 3.8.0.0 and a Python 3.12 requirement.
Can OpenSees run in parallel?
Yes. The wiki describes serial and parallel applications, and the Berkeley site has a parallel computing section.
Where is the OpenSees source code?
On GitHub in the OpenSees organisation, which holds all revisions since version 2.3.2, with build documentation for Windows, macOS and Ubuntu.
Sources
- OpenSees, UC Berkeley: https://opensees.berkeley.edu/ (checked 2026-11-14)
- OpenSees Wiki, Main Page: https://opensees.berkeley.edu/wiki/index.php/Main_Page (checked 2026-11-14)
- OpenSees copyright and licence notice: https://opensees.berkeley.edu/OpenSees/copyright.html (checked 2026-11-14)
- OpenSees Documentation: https://opensees.github.io/OpenSeesDocumentation/ (checked 2026-11-14)
- OpenSeesPy documentation: https://openseespydoc.readthedocs.io/en/latest/ (checked 2026-11-14)
- OpenSees source repository: https://github.com/OpenSees/OpenSees (checked 2026-11-14)
Data: Three routes into OpenSees compared: interpreter, OpenSeesPy and GitHub source
Compiled from the OpenSees wiki, the OpenSees copyright page, the OpenSeesPy documentation and the GitHub repository, read 2026-11-14. Compares how each route is used, what the sources state about versions and requirements, and the licence wording attached to each.
| Route | How it is used (as stated) | Version or requirement stated | Licence wording stated |
|---|---|---|---|
| Core application with scripting interpreters | Interpreter applications with a scripting language; models in plain-text input files or programs; serial and parallel | Version 3.8.0 binary announced February 2026 on the Berkeley site | Educational, research and non-profit entities: noncommercial use, no charge; other entities: internal purposes only; commercial products need University of California permission |
| OpenSeesPy | A Python 3 interpreter of OpenSees, installed from PyPI for Windows, Linux and macOS | v3.8.0.0 listed (incl. macOS ARM64); Python 3.12 required | Free for research, education and internal use; commercial redistribution, including cloud services using import openseespy, needs a commercial licence from UC Berkeley |
| Build from GitHub source | Source repository with build documentation for Windows 10, macOS and Ubuntu | All revisions since version 2.3.2 are in the repository | Same copyright notice applies; notice must appear in all copies and modifications |
Sources
- OpenSees wiki main page (accessed 2026-11-14)
- OpenSees copyright page (accessed 2026-11-14)
- OpenSeesPy documentation (accessed 2026-11-14)
- OpenSees GitHub repository (accessed 2026-11-14)
- OpenSees home page (accessed 2026-11-14)
Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.







