T3 automates the development of detailed chemical kinetic models. Given a set of initial species and conditions, it produces a validated model with high-fidelity thermochemistry and rate parameters by iteratively combining RMG for model generation, Cantera for sensitivity analysis, and ARC for quantum mechanical calculations.
- Iterative refinement: runs RMG, identifies sensitive species and reactions via SA, computes accurate thermo/kinetics with ARC, and feeds the results back into the next iteration
- Multiple SA adapters: constant TP/HP/UV, PFR, JSR, and brute-force ignition delay time (IDT)
- Equivalence-ratio sweeps: fuel/oxidizer/diluent role taxonomy with automatic concentration computation across a grid of temperatures, pressures, and equivalence ratios
- Uncertainty analysis: local and global UA via polynomial chaos expansion
- Restart support: T3 can resume from any interrupted iteration
git clone https://github.com/ReactionMechanismGenerator/T3.git
cd T3
make install
conda activate t3_envThis clones and installs ARC, RMG-Py, RMG-database, and all external dependencies. See the full installation guide for manual installation and troubleshooting.
Create a YAML input file and run T3:
conda activate t3_env
python T3.py input.ymlA minimal input file looks like:
project: my_project
t3:
options:
max_T3_iterations: 10
sensitivity:
adapter: CanteraConstantTP
top_SA_species: 10
top_SA_reactions: 10
rmg:
database:
thermo_libraries: ['primaryThermoLibrary', 'BurkeH2O2']
kinetics_libraries: ['primaryH2O2']
species:
- label: H2
smiles: '[H][H]'
concentration: 0.67
- label: O2
smiles: '[O][O]'
concentration: 0.33
- label: OH
smiles: '[OH]'
SA_observable: true
reactors:
- type: gas batch constant T P
T: 1000
P: 1
termination_conversion:
H2: 0.9
termination_time: [5, 's']
model:
core_tolerance: [0.05, 0.01]
qm:
adapter: ARC
level_of_theory: CBS-QB3See the input reference for all available options.
See the full documentation for:
make test # Run tests with coverage
make test-main # Run main unit tests only
make test-functional # Run functional (integration) testsWe welcome contributions!
- Found a bug or have a feature request? Open an issue
- See the contributing guide to get started
T3 is released under the MIT License.