v1.5.0: assess a site from an installed package

An installed package can now assess a new site from a TOML file.
borethm-new-site creates a starter or copies a built-in site with --from;
borethm-evaluate --site-file checks or evaluates it. The format describes
temperature profiles and linear stress cases, including a stress rule for a
known faulting regime. load_site and layered_geotherm join the public
Python interface. Exported results can record the input file's path and
SHA-256.

Reports and JSON include an opening reading of the checks that determine
the status. Documentation gains a searchable website and fast checks
against retained evidence. Study reproduction commands write to a separate
output folder. Calculations, numerical inputs and saved results for the
built-in sites are unchanged.

Required site selection

borethm-evaluate requires --site or --site-file. Calls without either
option no longer select Soultz automatically.

TOML site template

The packaged templates/site_template.toml replaces
examples/site_template.py. Python users can still construct a
SiteProfile and call evaluate.

Explicit coupling override

evaluate(..., coupled=False) now switches coupling off even when the site's
own flag is set. The default behaviour and built-in reference results are unchanged.

Install with python -m pip install borethm==1.5.0.

Documentation | New-site guide

Validation: 690 tests passed, including 104 slow tests, in 5,986.93 seconds. All 12 OS/Python combinations and wheel installations on Linux, Windows and macOS passed. CI result.

v1.4.2: technical accuracy of model documentation

v1.4.2

Corrects the model documentation to match the implementation: circulation exposure floors, drilling-rate branches and face heat, water-property limits and numerical guards, and absolute and gauge pressure conventions. The validation guide distinguishes fitted parameters and shared observations, including the limits on the Soultz and Newberry cross-checks and the FORGE tool-temperature comparison.

Calculations, numerical inputs, public signatures and saved results are unchanged. Docstring edits change source fingerprints; stored fingerprints and goldens are preserved.

Validation: GitHub CI passed 404 fast tests on each of Python 3.11, 3.12, 3.13 and 3.14, plus the locked-version and wheel-installation jobs. The local fast suite passed 404 tests in 24.64 seconds, with 106 slow tests excluded. Documentation links, API generation, distribution validation and fresh wheel installation passed. Full physical-regression evidence is inherited from v1.4.0; calculation-code and input equivalence were checked for this documentation patch.

Release record · Documentation · PyPI

python -m pip install borethm==1.4.2

v1.4.1: documentation and reader guidance

v1.4.1

This release revises BoreTHM's reader documentation and clarifies the assumptions behind the drilling results. It distinguishes target-depth stability from whole-hole fluid windows, updates the study summary and PDF, makes the API reference self-contained, and improves the user guides and data notes. README links now work from PyPI, and documentation checks cover the data guides and absolute repository URLs.

Calculations, numerical inputs, public signatures and saved reference results are unchanged. Benchmark dataset 1.0 has refreshed guide text and provenance with unchanged numerical content and tolerances.

Validation: GitHub CI passed 404 fast tests on each of Python 3.11, 3.12, 3.13 and 3.14, along with the locked-version and wheel-install jobs. The local fast suite also passed 404 tests, with 106 slow tests excluded. API and link checks, benchmark assembly, distribution validation and a fresh v1.4.1 wheel installation passed. The complete v1.4.0 physical-regression run is inherited evidence; executable-code and input equivalence were verified for this documentation patch.

Release record · Documentation · PyPI

python -m pip install borethm==1.4.1

BoreTHM v1.4.0

BoreTHM now installs as the borethm Python package, with a documented public interface and command-line tools. The release reorganises the code and packaged data, adds contribution and support guides, and provides a separate reference benchmark dataset.

No physics, input, default or result changes. Every existing numerical regression reproduces within its unchanged tolerance.

Install

python -m pip install borethm
borethm-water-table
borethm-evaluate --site newberry --json newberry.json

Requires Python 3.11 or later. Install "borethm[plot]" for figures. Reproduce reference calculations with Python 3.14.4 and the lock file, as described in the README.

Documentation and data

• README
• Public Python interface
• Benchmark dataset 1.0: six reference cases with inputs, expected outputs, units and existing tolerances. Some geometry and geotherm definitions remain code references, as explained in the guide.
• Release record

Validation

The full CI run passed 507 tests, including all 106 slow regressions, in 2 hours 3 minutes 48 seconds. The Python 3.11 to 3.14 fast-test matrix and clean wheel installation also passed. Calculation source, input data, tests and the lock file match the tested candidate, abdb56c; release integration updates the date and documentation.

Code is Apache 2.0. Documentation and reference results are CC BY 4.0; third-party data retain their source licences and credits.

Citation

Use the software DOI for the project citation. The v1.4.0 archive identifies this release specifically. CITATION.cff on main includes the software DOI and author ORCID iD.

v1.3.6

Publishes the breakout-timing results from the existing failure-delay experiment, lever L7, with paired zero-delay and one-hour calculations and a fixed delay scan.

• At Soultz, a one-hour delay lowers the most demanding target drilling requirement from 1.3129 to 1.1753 SG-equivalent. The first open sampled whole-hole drilling window occurs at 20 minutes, bracketed by a closed sample at 15 minutes.
• At United Downs, the corresponding target requirement falls from 1.2667 to 1.1397 SG-equivalent. The first open sampled window occurs at 15 minutes, bracketed by a closed sample at 10 minutes.

The complete programme remains NO-GO at both sites: trip-fluid windows remain closed and tool-temperature limits are exceeded. These calculations test assumed breakout formation delays; they provide no measurement supporting a particular delay.

The scientific page explains the criterion and reproduction commands. The saved results distinguish target SG-equivalent requirements from surface-density window bounds.

No model, input or default changes. Earlier results and existing goldens are unchanged. The summary and its PDF are unchanged.

Validation: 357 fast tests and 106 slow tests passed. The slow suite emitted 12 runtime warnings in existing reference calculations; all assertions passed.

v1.3.5

v1.3.5: the BoreTHM name and the Apache 2.0 licence