Installation

OpenFlowsheet needs Python 3.13 or newer and runs on Linux (x86-64 and aarch64 are the tested platforms).

python -m pip install openflowsheet

This installs the simulator, the command line openflowsheet and the base dependencies: NumPy, SciPy, PyYAML, jsonschema and CasADi, all pinned to exact versions.

Installing for exact replay

A run bundle records the hash of the package's dependency lock only when the installed runtime dependencies are exactly at the lock, including jsonschema's own dependencies (attrs, jsonschema-specifications, referencing, rpds-py), which a plain install leaves free to upgrade. To install at the lock, constrain the install by the release's requirements.lock (in the repository at the release's tag, and in the source distribution):

python -m pip install -c requirements.lock openflowsheet

Without it, a bundle written where one of those four differs records no lock, and replay names the distributions that differ.

CasADi's licence

CasADi is licensed under the LGPL-3.0-or-later. It is a separate package that pip obtains from PyPI; OpenFlowsheet uses it through its Python API only and distributes none of its bytes. See Licence and attributions.

Optional extras

Extra	Install	What it enables	Licences of what it adds
coolprop	pip install "openflowsheet[coolprop]"	The CoolProp pure-fluid basis (coolprop-heos-v1): one of 123 fluids per flowsheet, for the thermal.* units	CoolProp (MIT), inventoried in docs/s07-coolprop-licences.json
server	pip install "openflowsheet[server]"	The HTTP server (openflowsheet serve-http, with the optional web shell) and the MCP server (openflowsheet serve-mcp)	mcp (MIT), Starlette and Uvicorn (BSD-3-Clause) and their dependencies, all permissive; the closure is inventoried in the repository (docs/t07-server-licences.json)
nlp	see below	The general NLP bridge of the studies layer (Pyomo and Ipopt through cyipopt)	Pyomo (BSD-3-Clause), cyipopt and Ipopt (EPL-2.0); the audited environment's binaries are inventoried in docs/m03-ipopt-audit.md
lca	pip install "openflowsheet[lca]"	The offline Brightway adapter that builds unit-score tables for the LCA from a database you have prepared (see Cost estimation and LCA); assessments evaluate without it	Brightway's bw2data and bw2calc (BSD-3-Clause) and their dependencies, inventoried in docs/e01-lca-licences.json
dev	pip install -e ".[dev]"	The test and lint tools, for working on OpenFlowsheet itself	—

The nlp extra cannot be installed from PyPI alone: PyPI has no cyipopt wheel for CPython 3.13. It installs into a conda-forge environment that provides cyipopt and the Ipopt libraries; the repository's scripts/build-m03-ipopt-env.sh builds exactly the audited one.

From source

git clone https://github.com/computational-chemical-engineering/openflowsheet.git
cd openflowsheet
make venv        # .venv with the project and its dev extra, from the pinned lock
make check       # ruff, mypy --strict and the test suite

A source checkout also carries the registered benchmark cases.

Verify the installation

python -c "import openflowsheet; print(openflowsheet.__version__)"
openflowsheet --help

Then solve the quickstart's flowsheet: Quickstart. With the coolprop extra, python -c "import CoolProp" should succeed as well.