pythermalcomfort 3.9.8

pythermalcomfort is a comprehensive toolkit for calculating thermal comfort indices, heat/cold stress metrics, and thermophysiological responses. It supports multiple models, including PMV, PPD, adaptive comfort, SET, UTCI, Heat Index, Wind Chill Index, and Humidex. The package also includes thermophysiological models like the two-node (Gagge) and multinode (JOS-3) models to estimate physiological responses such as core temperature, skin temperature, and skin wettedness.

pythermalcomfort

pythermalcomfort is a Python toolkit for computing thermal comfort indices, heat/cold stress metrics, and thermophysiological responses. Its implementations adhere to international standards and peer-reviewed research, offering researchers, engineers, and building scientists reliable, standards-compliant calculations without the burden of implementing them manually.

Important

When pythermalcomfort informs published work, please cite it as:

Tartarini, F., Schiavon, S., 2020.
pythermalcomfort: A Python package for thermal comfort research.
SoftwareX 12, 100578.
https://doi.org/10.1016/j.softx.2020.100578

Key Features

• Thermal Comfort Models – PMV/PPD, adaptive comfort assessments, SET, and more bundled into a single API surface.

• Heat & Cold Stress Indices – UTCI, Heat Index, Wind Chill, Humidex, and other commonly-referenced metrics.

• Thermophysiological Modeling – two-node (Gagge) and multinode (JOS-3) models for estimating core/skin temperatures and skin wettedness.

• Standards Compliance – Calculations based on ASHRAE 55, ISO 7730, EN 16798, and supporting references.

• Vectorized Inputs – Accepts scalars, lists, or NumPy arrays; most functions broadcast across input arrays automatically.

• Pythonic API – Simple, documented entry points that plug into analysis workflows and pipelines.

• Rich Documentation – Tutorials, examples, and reference guides for each supported model and index.

• Active Development – Frequent releases, new features, and responsive issue resolution.

• Open Source – MIT licensed and developed transparently on GitHub.

Why Choose pythermalcomfort?

• Precision – Accurate evaluations of comfort and stress that engineers can trust.

• Efficiency – Eliminates repetitive code so teams can focus on insights, not implementation details.

• Versatility – Useful in building science, HVAC design, biometeorology, sports science, and thermal physiology.

• Evidence-Based Decisions – Supports data-driven HVAC sizing, occupant comfort strategies, and performance benchmarking.

Installation

Install from PyPI:

pip install pythermalcomfort

For alternative installation instructions, including development builds and optional dependencies, see the official docs.

Requirements

• Python 3.10+

• NumPy, SciPy, Numba, setuptools (installed automatically)

• Optional: pandas, Matplotlib, or other plotting libraries for examples and visualizations

Quick Start

A few lines are all you need to get started:

from pythermalcomfort.models import pmv_ppd_iso, utci

# Calculate PMV and PPD using ISO 7730 standard
result = pmv_ppd_iso(
    tdb=25,   # dry-bulb temperature in °C
    tr=25,    # mean radiant temperature in °C
    vr=0.1,   # relative air speed in m/s
    rh=50,    # relative humidity in %
    met=1.4,  # metabolic rate in met
    clo=0.5,  # clothing insulation in clo
    model="7730-2005",
)
print(f"PMV: {result.pmv}, PPD: {result.ppd}")

# Calculate UTCI for heat stress assessment
result = utci(tdb=30, tr=30, v=0.5, rh=50)
print(result.utci)

# Most functions also accept arrays for bulk calculations
result = utci(tdb=[28, 30, 35], tr=[28, 30, 35], v=0.5, rh=50)
print(result.utci)

For a full list of models and indices, see the API reference.

Support pythermalcomfort

Maintaining an open-source scientific package takes time. You can help by:

• Sponsoring on GitHub

• Submitting code, docs, or tests via a pull request

• Reporting reproducible bugs or feature requests in the issue tracker

• Assisting with testing, translations, or PR reviews

• Starring or sharing the project to raise awareness

Contributions

We welcome all contributions. Please read the contributing guide before you start.

Quick checklist

• Open an issue when planning large changes to align on scope.

• Fork the repo and create a feature branch.

• Add or update tests for new behavior.

• Run linters/formatters and fix the reported issues.

• Update docs or the changelog when the public API changes.

• Submit clear, focused PRs with related issues linked.

Common commands

# clone your fork and add upstream remote
git clone [e-mail]:your-username/pythermalcomfort.git
cd pythermalcomfort
git remote add upstream [e-mail]:pythermalcomfort/pythermalcomfort.git
git fetch upstream

# create a branch and work on it
git checkout -b Feature/awesome-feature
tox  # run the full matrix (slow)
tox -e py312  # run a single env
pytest -k test_name_fragment

# fix linting/formatting
ruff check --fix
ruff format
docformatter --in-place --wrap-summaries 88 --wrap-descriptions 88 pythermalcomfort/*.py

# commit and push
git add .
git commit -m "feat: short description of change"
git push origin Feature/awesome-feature

Release process

Releases are tag-driven and published via GitHub Actions Trusted Publishing (OIDC — no PYPI_API_TOKEN or TEST_PYPI_API_TOKEN secret is required).

The standard cycle is:

1. Develop and test a release candidate on development → TestPyPI.

2. Merge development → master via pull request.

3. Finalize the version on master → PyPI.

Step 1 — pre-release on development (TestPyPI)

git checkout development
git pull --ff-only
git fetch --tags --prune

# Start the RC cycle for the next patch release (e.g. 3.9.8 → 3.9.9rc1):
bump-my-version bump patch

# Push the bump commit and tag — this triggers tests + TestPyPI deploy:
git push
git push --tags

If the RC needs additional fixes, make the commits then create another RC:

bump-my-version bump pre_n    # e.g. 3.9.9rc1 → 3.9.9rc2
git push
git push --tags

Step 2 — open and merge a pull request from development to master

CI runs the full test suite on the PR. Once it passes, merge via GitHub.

Step 3 — finalize on master (PyPI)

git checkout master
git pull --ff-only
git fetch --tags --prune

# Strip the rc suffix to produce the stable version.
# Replace X.Y.Z with the target version (e.g. 3.9.9):
bump-my-version bump --new-version X.Y.Z patch

# Push the bump commit and tag — this triggers tests + PyPI deploy:
git push
git push --tags

Rules:

• All RC tags (vX.Y.ZrcN) must be pushed from development.

• All stable tags (vX.Y.Z) must be pushed from master after merging.

• Tag format: vX.Y.Z for stable, vX.Y.ZrcN for pre-release.

• Do not push a stable tag before the corresponding development → master PR has been merged; the CI deploy job will reject it.

Getting Help

• Open an issue on GitHub with a minimal reproduction in the issue tracker.

• Ask questions in PR comments for implementation guidance.

• Review the contribution guidelines for testing, documentation, and changelog expectations.

• Consult the API reference and examples at https://pythermalcomfort.readthedocs.io/en/latest/

Changelog

A full list of changes per release is available in the CHANGELOG.

License

pythermalcomfort is released under the MIT License.

Stats

Changelog

3.9.8 (2026-05-25)

• Added optional round_output parameter to adaptive_ashrae and adaptive_en to control rounding of output values.

• Added limit_inputs parameter to ankle_draft and vertical_tmp_grad_ppd, consistent with other model functions.

• ankle_draft and vertical_tmp_grad_ppd now raise UserWarning when inputs exceed model applicability limits.

• Fixed compliance attribute being included in non-ASHRAE PMV model outputs; it is now only returned by pmv_ppd_ashrae.

• Fixed UTCI stress category mapping when units="IP"; categories were incorrectly mapped before IP unit conversion.

3.9.3 (2026-05-01)

• Maintenance release: internal CI pipeline improvements and dependency updates. No user-facing changes.

3.9.2 (2026-04-14)

• Updated sports_heat_stress_risk so risk_level_interpolated now uses 1.0-4.0 instead of 0.0-3.0.

• Updated sports_heat_stress_risk to enforce the sport-specific minimum air speed (sport.vr).

3.9.1 (2026-02-25)

• Improved speed of PHS model.

3.9.0 (2026-02-03)

• Added sports_heat_stress_risk function to assess heat stress risk for athletes during outdoor sports activities based on environmental conditions. Addresses issue #237.

3.7.0 (2025-10-28)

• Added machine learning model to predict skin and rectal temperature ridge_regression_predict_t_re_t_sk.

3.6.1 (2025-10-07)

• Fix issue with disc calculation in the two_nodes_gagge model and limiting its value to 6. Close #251

• Improve documentation for the disc function.

• PMV ASHRAE model returns the compliance boolean value with the ASHRAE 55:2023 standard. Close #253

• Improve formatting of models outputs to the console.

3.6.0 (2025-09-22)

Warning

breaking change for the phs function: - the phs function now returns all the outputs needed to use the outputs of a previous calculation as inputs for a new calculation. - removed outputs: water_loss, water_loss_watt - added outputs (with units):

• sweat_loss_g [g] — cumulative sweat mass per person (not area‑normalised)

• sweat_rate_watt [W·m⁻²] — instantaneous evaporative heat flux at skin

• evap_load_wm2_min [W·min·m⁻²] — accumulated evaporative load for chaining

Migration:

# <= 3.5.x
grams = res.water_loss
w_m2 = res.water_loss_watt
# >= 3.6.0
grams = res.sweat_loss_g
w_m2 = res.sweat_rate_watt
carry = res.evap_load_wm2_min  # for multi‑segment runs

3.5.1 (2025-09-15)

• Improved documentation on how to contribute to the project

3.5.0 (2025-09-10)

• Added the scale_winds_speed_log function to scale wind speed.

3.4.3 (2025-07-31)

• fix: wind chill temperature was not imported in pythermalcomfort.models

3.4.2 (2025-07-22)

• fixed unit of sweat_rate in the PHS model

3.4.1 (2025-07-14)

Warning

removed support for Python 3.8 and 3.9 pythermalcomfort now requires Python 3.10 or higher. Added support for Python 3.13

• fixed some typo in the documentation

• better formatted the code

3.4.0 (2025-06-08)

• Added the work_capacity_dunne.

• Added the work_capacity_hothaps.

• Added the work_capacity_iso.

• Added the work_capacity_niosh.

3.3.0 (2025-06-05)

• Added the two_nodes_gagge_ji function to calculate the two-node model for older individuals

• Added the Temperature-Humidity Index (THI).

3.2.0 (2025-05-20)

• Added the two_nodes_gagge_sleep function to calculate the two-node model for sleeping individuals

• Added the ESI function to calculate the Environmental Stress Index.

3.1.0 (2025-04-28)

• Updated the PHS model in compliance with the ISO 7933:2023 standard

  • Added default‑kwarg overrides for 2023 mode (f_r, t_re, t_cr_eq)

  • removed unused variable round from default_kwargs

• Included test cases according to the ISO 7933:2023 standard

• Added AutoStrMixin to provide a formatted __str__ representation for result classes

Note

By default the ISO 7933:2023 standard will now be used for the PHS model. The 2004 standard can still be specified as an optional parameter

Warning

The third ISO 7933:2023 test case is currently failing and has been marked xfail while the discrepancy is investigated.

3.0.1 (2025-04-14)

• allow np.float and np.int as inputs to all functions

• fixed documentation for phs - met units

3.0.0 (2025-02-03)

Warning

pythermalcomfort version 3.0.0 introduces some breaking changes.

How functions return results: as the functions now return dataclass instances with the calculation results. This change enhances the structure and accessibility of the results. For example:

from pythermalcomfort.models import pmv_ppd_iso

result = pmv_ppd_iso(
    tdb=[22, 25], tr=25, vr=0.1, rh=50, met=1.4, clo=0.5, model="7730-2005"
)
print(result.pmv)  # [-0.  0.41]

This update aims to make the package more user-friendly and to provide a more organized way to access all calculation results.

Moved functions Moved all the functions that were in the psychrometrics.py file to the utilities.py file.

Changed function names All the PMV functions have been renamed using the following format: pmv_XXX where XXX is the standard or the model name.

PMV function The pmv_ppd function now has been split into two functions: pmv_ppd_iso and pmv_ppd_ashrae.

Note

We have updated all functions to accept Numpy arrays as inputs, allowing you to pass multiple values at once for faster results. Single values are still accepted, and the functions will return results as before. Additionally, we have synchronized the tests with the R comf package to ensure consistent calculation results across both packages.

Other improvements include:

• Enhanced documentation with more examples.

• Better described the models.

• Added more tests to ensure calculation accuracy.

• Implemented input validation to ensure inputs are within model applicability limits.

• Harmonized input names across all functions.

• Added surveys to assess thermal comfort to the documentation.

• Added a detailed section about clothing insulation.

2.10.0 (2024-03-18)

• allow n-dimensional arrays for pet_steady and speedup p_sat calculation

2.9.1 (2024-01-19)

• Fixed error calculation of mass sweating in PET mode, the unit was incorrect

2.9.0 (2024-01-15)

Warning

pythermalcomfort 2.9.0 is no longer compatible with Python 3.8

• The PHS model accepts arrays as inputs

2.8.11 (2023-10-26)

• wrote more test and improved code

2.8.11 (2023-10-26)

• fixed issues with the documentation and sorted the models in alphabetical order

2.8.7 (2023-10-23)

• Adaptive ASHRAE now returns a dataclass

2.8.6 (2023-10-09)

• re-structured and linted the code

2.8.4 (2023-09-20)

• calculation of cooling effect in pmv (standard=’ashrae’) triggered only when v>0.1 m/s

2.8.3 (2023-09-14)

• general improvements in the JOS3 model

2.8.2 (2023-09-04)

• general improvements in the JOS3 model

• fixed error when e_max == 0

2.8.1 (2023-07-05)

• pythermalcomfort needs Python version > 3.8

• fixed issue in Cooling Effect calculation

2.8.0 (2023-07-03)

• allowing the cooling effect to range from 0 to 40

• fixed PHS documentation

• improved JOS3 documentation

2.7.0 (2023-02-16)

• changed coefficient of vasodilation in set_tmp() to 120 to match ASHRAE 55 2020 code

• slightly modified value in validation tables

2.6.0 (2023-01-17)

• max sweating rate can be passed to two node model

• max skin wettedness can be passed to two node model

• rounding w to two decimals

• use_fans_heatwave function accepts arrays

• fixed typos unit documentation

2.5.4 (2022-10-12)

• PHS model accepts all required inputs to be run on a minute by minute basis

• fix error check compliance PHS model

2.5.0 (2022-06-13)

• Added the adaptive thermal heat balance (ATHB) model

2.4.0 (2022-06-10)

• Added e_pmv model - Adjusted Predicted Mean Votes with Expectancy Factor

• Added a_pmv model - Adaptive Predicted Mean Vote

2.3.0 (2022-06-01)

• Added discomfort index

2.2.0 (2022-05-17)

• Implemented a better equation to calculate the mean radiant temperature

2.1.1 (2022-05-17)

• Fixed how DISC is calculated

2.1.0 (2022-04-20)

• Added Physiological Equivalent Temperature (PET) model

• In PMV and PPD function you can specify if occupants has control over airspeed

2.0.2 (2022-04-12)

• UTCI accepts lists as inputs

2.0.0 (2022-04-07)

Warning

Version 2.0.0 introduces some breaking changes. Now the default behaviour of most of the function is that they return a np.nan if the inputs are outside the model applicability limits.

For most functions we are no longer printing Warnings. If you want the function to return a value even if your inputs are outside the model applicability limits then you can set the variable limit_input = False. Please note that you should refrain from doing this.

Note

Starting from Version 2.0.0 of pythermalcomfort now most of the functions (see detailed list below) accept Numpy arrays or lists as inputs. This allows you to write more concise and faster code since we optimized vectorization, where possible using Numba.

• Allowing users to pass Numpy arrays or lists as input to the pmv_ppd, pmv, clo_tout, both adaptive models, utci, set_tmp, two_nodes

• Changed the input variable from return_invalid to limit_input

• Increased speed by using Numba @vectorize decorator

• Changed ASHRAE 55 2020 limits to match new addenda

• Improved documentation

1.11.0 (2022-03-16)

• Allowing users to pass a Numpy array as input into the UTCI function

• Numpy is now a requirement of pythermalcomfort

• Improved PMV, JOS-3, and UTCI documentation

• Testing PMV, SET, and solar gains models using online reference tables

1.10.0 (2021-11-15)

• Added JOS-3 model

1.9.0 (2021-10-07)

• Added Normal Effective Temperature (NET)

• Added Apparent Temperature (AT)

• Added Wind Chill Index (WCI)

1.8.0 (2021-09-28)

• Gagge’s two-node model

• Added WBGT equation

• Added Heat index (HI)

• Added humidex index

1.7.1 (2021-09-08)

• Added ASHRAE equation to calculate the operative temperature

1.7.0 (2021-07-29)

• Implemented function to calculate the if fans are beneficial during heatwaves

• Fixed error in the SET equation to calculated radiative heat transfer coefficient

• Fixed error in SET definition

• Moved functions optimized with Numba to new file

1.6.2 (2021-07-08)

• Updated equation clo_dynamic based on ANSI/ASHRAE Addendum f to ANSI/ASHRAE Standard 55-2020

• Fixed import errors in examples

1.6.1 (2021-07-05)

• optimized UTCI function with Numba

1.6.0 (2021-05-21)

• (BREAKING CHANGE) moved some of the functions from psychrometrics to utilities

• added equation to calculate body surface area

1.5.2 (2021-05-05)

• return stress category UTCI

1.5.1 (2021-04-29)

• optimized phs with Numba

1.5.0 (2021-04-21)

• added Predicted Heat Strain (PHS) index from ISO 7933:2004

1.4.6 (2021-03-30)

• changed equation to calculate convective heat transfer coefficient in set_tmp() as per Gagge’s 1986

• fixed vasodilation coefficient in set_tmp()

• docs changed term air velocity with air speed and improved documentation

• added new tests for comfort functions

1.3.6 (2021-02-04)

• fixed error calculation solar_altitude and sharp for supine person in solar_gain

1.3.5 (2021-02-02)

• not rounding SET temperature when calculating cooling effect

1.3.3 (2020-12-14)

• added function to calculate sky-vault view fraction

1.3.2 (2020-12-14)

• replaced input solar_azimuth with sharp in the solar_gain() function

• fixed small error in example pmv calculation

1.3.1 (2020-10-30)

• Fixed error calculation of cooling effect with elevated air temperatures

1.3.0 (2020-10-19)

• Changed PMV elevated air speed limit from 0.2 to 0.1 m/s

1.2.3 (2020-09-09)

• Fixed error in the calculation of erf

• Updated validation table erf

1.2.2 (2020-08-21)

• Changed default diameter in mean_radiant_tmp

• Improved documentation

1.2.0 (2020-07-29)

• Significantly improved calculation speed using numba. Wrapped set and pmv functions

1.0.6 (2020-07-24)

• Minor speed improvement changed math.pow with **

• Added validation PMV validation table from ISO 7730

1.0.4 (2020-07-20)

• Improved speed calculation of the Cooling Effect

• Bisection has been replaced with Brentq function from scipy

1.0.3 (2020-07-01)

• Annotated variables in the SET code.

1.0.2 (2020-06-11)

• Fixed an error in the bisection equation used to calculated Cooling Effect.

1.0.0 (2020-06-09)

• Major stable release.

0.7.0 (2020-06-09)

• Added equation to calculate the dynamic clothing insulation

0.6.3 (2020-04-11)

• Fixed error in calculation adaptive ASHRAE

• Added some examples

0.6.3 (2020-03-17)

• Renamed function to_calc to t_o

• Fixed error calculation of relative air speed

• renamed input parameter ta to tdb

• Added function to calculate mean radiant temperature from black globe temperature

• Added function to calculate solar gain on people

• Added functions to calculate vapour pressure, wet-bulb temperature, dew point temperature, and psychrometric data from dry bulb temperature and RH

• Added authors

• Added dictionaries with reference clo and met values

• Added function to calculate enthalpy_air

0.5.2 (2020-03-11)

• Added function to calculate the running mean outdoor temperature

0.5.1 (2020-03-06)

• There was an error in version 0.4.2 in the calculation of PMV and PPD with elevated air speed, i.e. vr > 0.2 which has been fixed in this version

• Added function to calculate the cooling effect in accordance with ASHRAE

0.4.1 (2020-02-17)

• Removed compatibility with python 2.7 and 3.5

0.4.0 (2020-02-17)

• Created adaptive_EN, v_relative, t_clo, vertical_tmp_gradient, ankle_draft functions and wrote tests.

• Added possibility to decide with measuring system to use SI or IP.

0.3.0 (2020-02-13)

• Created set_tmp, adaptive_ashrae, UTCI functions and wrote tests.

• Added warning to let the user know if inputs entered do not comply with Standards applicability limits.

0.1.0 (2020-02-11)

• Created pmv, pmv_ppd functions and wrote tests.

• Documented code.

0.0.0 (2020-02-11)

• First release on PyPI.