Software and hardware requirements

The input data files for the CLEWs-EU model are constructed in an Excel file with multiple worksheets. Each worksheet corresponds to a particular SET or Parameter in the OSeMOSYS model.

A Python package, Otoole, is used to handle all data manipulations required for running the model. Users should ensure they have:

  • the latest stable version of Python, and

  • the latest version of Otoole installed.

To use Otoole, a YAML configuration file is required. A project-specific YAML file has been prepared for CLEWs-EU and is available in the model_files folder of the model’s GitHub repository.

An optimisation solver is also required to run the model. Any of the following options may be used:

  • GLPSOL – mandatory, as it is part of the GLPK package and required to generate the model matrix. Note: the disaggregated CLEWs-EU model is computationally intensive; GLPSOL should not be used for the optimisation step.

  • GUROBI – proprietary (academic licence available)

  • CPLEX – proprietary (academic licence available)

  • CBC – open-source

As the CLEWs-EU model was developed in parallel modules, computational intensity varies depending on whether:

  • the fully integrated model is run, or

  • only an individual sectoral module is executed.

The primary limiting factor is the amount of available memory.

The disaggregated version of the integrated CLEWs-EU model is computationally intensive and is therefore not run with annual time steps; instead, 5-year steps are used. In contrast, annual resolution is maintained in each standalone sectoral model.

As indicated in Table 27, the disaggregated integrated CLEWs-EU model requires ~260 GB of memory to run. By comparison, the aggregated version requires only ~8 GB, despite using annual resolution, due to its significantly lower structural complexity, making it suitable for stakeholder engagement.

Table 27. Key hardware requirements to run the disaggregated version of the CLEWs-EU model and its sectoral components

CLEWs-EU sectoral model

Memory requirements (GB RAM)

Energy module

170 GB

Electricity supply

38 GB

Buildings

12 GB

Industry

34 GB

Transport

32 GB

Land & Water module

180 GB

Integrated CLEWs-EU model

260 GB

Model setup

The GitHub repository provides the following material:

  1. OSeMOSYS code This is the version of the OSeMOSYS code developed specifically for this project. Three separate files are provided, depending on which model is used (i.e., the individual sectoral models or the integrated CLEWs-EU model). These files are plain-text and include all parameters, variables, and constraints that describe the relationship between input data (parameters) and model outputs (variables).

  2. CLEWs-EU model data files Model data files are provided as Excel spreadsheets for:

    • the energy module and its sub-components,

    • the land and water module,

    • and the integrated CLEWs-EU model (energy + land + water).

    These files contain the full model structure in the form of Sets (e.g., years, fuels/commodities, technologies, emissions, etc.) as well as all associated techno-economic assumptions (e.g., energy service demand projections, costs, efficiencies, and so on).

  3. Supporting data This folder contains additional information needed by future CLEWs-EU users. The naming convention files are included here, along with more detailed input data and modelling assumptions for the final version of the model.

  4. Data preparation scripts These scripts automate the extraction of required data from external databases (e.g., JRC-IDEES 2021). They have been developed for the transport and industrial sectors and may be extended to other sectors in future releases. The scripts are included for transparency and to support potential future updates to model detail.

Using the CLEWs-EU model

The procedure to run the CLEWs-EU model is the same as for any other OSeMOSYS model. Specifically, the following steps should be followed:

  1. Using Otoole, convert the OSeMOSYS data file from Excel format to a text data file. Users may refer to the examples page of the Otoole documentation for the exact syntax.

  2. Use GLPK to create the LP version of the model (.lp file), using as input:

    • the OSeMOSYS data file (.txt), and

    • the OSeMOSYS code file (.txt).

    This step generates the problem matrix by loading the data into the OSeMOSYS equations.

  3. Optimise the .lp file using any of the recommended solvers. Once an optimal solution is found, write the solution file (.sol format).

  4. Use Otoole to export the results from the solution file into CSV result files.

  5. Open the CSV files to inspect results in tabular form. At a later stage, visualisation scripts will be developed to allow automatic and interactive result exploration.