Version 2.4.0

• Introduces flexible components of the Conversion class. This allows for conversion components that can choose their input or output from different commodities. The share of each commodity can be limited by a flowShare. Emissions can be assigned by emissionFactors specific to a commodity.
• Allow intermittent pathway parameters (e.g., capacityFix). Pathway parameters can now be defined only for single investment periods, while being None for others.
• Fixes bug in models using gurobi.
• Improves data type checking.

Version 2.3.8

• Fixes bug for multi period data for Transmission components. Error message has been improved.
• Fixes bug in Storage init. Handle empty dict for operationTimeSeries if no timeseries data is set by setting it to None.
• Fixes bug for capexIfBuilt in optimization summary. Previously it displayed NPV, now it is TAC.

Version 2.3.7

• Fixes a bug in the serialization to datasets of timeseries data in Transmission components
  • Treating the processing of Transmisison components separately when creating xrds input data in addDFVariablesToXarray
  • Updating addTimeSeriesVariableToDict to allow operationRates for Transmisison components
• Adds solver_io="python" to the execution of Gurobi if gurobipy is installed
• Adds the dimension parameter 2dim in checkAndSetInvestmentPeriodTimeSeries in transmission.py to fix operation time series for transmission components
• Deletes doubled function in utils.py

Version 2.3.6

• Allow for python>=3.10<3.13, limit dependencies to the next major version
• Add dependencies to pyproject.toml for pip install
• Update community guidelines, documentation, examples

Version 2.3.5

Version 2.3.5 adds the functionality to pass maximum, minium and fixed values to the commissioning variables and growth rates for comissioning constraints. Set CommissioningMin for a lower bound, CommissioningMax for an upper bound and CommissioningFix to set the lower and upper bounds to the same value.

Version 2.3.4

Version 2.3.4 introduces the following functionality:

• operationRateMin for components

Furthermore, it introduces the following changes:

• Installation
  • Remove the pip install from requirements.yml. It has to be run manually now.
• Testing
  • Run Pytest in parallel with pytest-xdist
  • Add test workflows on GitHub actions to test the installtion on Windows, Mac and Ubuntu Linux
• Documentation
  • Deprecate the news feed in the documentation in favor of GitHub release notes
• Requirements
  • Fix Python version to 3.10
  • Restrict scikit-learn>=1.2 because of not backward compatible changes in that version
  • Remove GDAL since it is a secondary requirement
  • Adaptions to Pandas 2.2 in preparation of version 3.0, which will introduce breaking changes

Version 2.3.3

ETHOS.FINE release 2.3.3 provides the following changes:

Functionality

• Add a the possibility to set a minimal operation rate in the Conversion class

Maintenance

• Use relative paths in tests to prevent errors when using different working directories
• Update scikitLearn in the requirements and adapt breaking changes in version >1.2

Version 2.3.2

IMPORTANT: The name of the package folder has been changed from FINE to fine in this release. If you still see a FINE folder locally after pulling the latest change, you might need to clone the repository to a new folder.

Further, FINE release 2.3.2 provides changes in the requirements:

• Pin GDAL to version 3.4.3 because version 3.4.1 is not compatible with the latest Fiona versions.
• Change the repository of gurobi-logtools from pypi to conda-forge.

Version 2.3.1

ETHOS.FINE release 2.3.1

• Adds a performance summary as attribute EnergySystemModel.performanceSummary. The performance summary includes Data about RAM usage (assesed by the psutil package), Gurobi values (extracted from gurobi log with the grblogtools package) and other various parameters such as model buildtime, runtime and time series aggregation paramerters.
• Fixes a bug in the stochastic optimization example.
• Makes subclass conversionPartLoad usable again. The nSegments parameter has to be set manually depending on the form of the non-linear function.
• Drops the constraint on the version of pandas to also work with versions lower than 2.

Version 2.3.0

FINE release (2.3.0) provides new major functionalities:

• Representation of multiple investment periods in a transformation pathway (perfect foresight) or for single year optimization (stochastic optimization)
• Consideration of CO₂ budgets for the full transformation pathway
• Consideration of stock including techno-economic parameters depending on commissioning date
• Variable efficiencies for conversion components depending on commissioning date and operation time
• Additional or lowered costs for components which are not present for full investment periods

The ConversionPartLoad class is not supported in this release due to the deprecated package GPyOpt.
The installation method has been changed from setup.py to pyproject.toml.