Web products, APIs, data applications and automation in Python, built with architecture, testing and well-planned operations in mind.
Energy System Modelling | Anexia Digital Engineering

Energy system modelling with PyPSA & pandapower.
How do technical limits, costs and time series affect your energy or power system? Anexia Digital Engineering maps scenarios, investment decisions and grid states in open-source Python models and makes the results traceable for the question you need answered.
Two tools, clear roles
Energy and grid questions work on different time horizons. For expansion planning, dispatch and sector coupling over long time series, PyPSA is the right framework. For load flow, short circuit and state estimation in a detailed grid model, it is pandapower. We choose according to the question and combine both where a plan also has to hold up in the grid.
Networks & scenarios
Buses, generators, loads, lines, links, storage, snapshots and investment periods as a consistent network built from your data.
Optimisation & expansion
Economic dispatch, linear optimal power flow and capacity expansion for generation, storage, conversion and transmission.
Grid calculation
AC and three-phase load flow, short-circuit currents to IEC 60909, state estimation and optimal power flow in a detailed grid model.
Reproducible runs
Versioned assumptions, tested Python workflows and checked result exports that your team can repeat on its own.
Energy system modelling in Python
Whether a storage asset pays off, whether a line can cope with the loss of another, or how a higher share of renewable generation changes a grid are not questions you can answer by instinct. Technical limits, costs and time series interact. Energy system modelling makes these relationships computable, provided model, data and assumptions are out in the open. Anexia Digital Engineering works with two open-source Python tools with clearly separated roles. PyPSA models and optimises energy and power systems across time series and scenarios, while pandapower calculates electrical grids element by element.
Both tools are Python packages and fit into existing data workflows. Input data can be prepared with pandas, results plotted with Matplotlib or made usable for domain experts in a Streamlit app. Because the model exists as code, it can be versioned, tested and repeated. That sets it apart from spreadsheet models whose logic is scattered across countless formulas, and it makes assumptions readable for engineering, management and auditors alike.
Capacity expansion & dispatch with PyPSA
The framework represents generation, grids, storage, time series, coupled energy carriers and technical constraints, including linearised power flows in AC and DC networks. From your domain and input data we build a pypsa.Network with buses, generators, loads, lines, links, storage units, snapshots and investment periods. On top of that we optimise economic dispatch or linear optimal power flow (LOPF) with network constraints, so variants can be compared. In capacity expansion planning, the model decides on investment in generation, storage, conversion and transmission. N-1 line outages are covered in a security-constrained LOPF (SCLOPF), which makes security assumptions visible in the model.
Using n.optimize() and Linopy we add variables, objective functions and project-specific constraints, for example expansion limits, minimum shares or emission targets that follow from your question. After every run we check energy and power balances before interpreting anything. An optimum that violates a balance or sits on a forgotten limit points to a data or modelling error, not to a result.
Grid calculation with pandapower
For the static analysis of electrical grids we model buses, lines, switches, transformers, loads and generators as a maintained grid model. On that basis we run AC and three-phase load flow, short-circuit calculations to IEC 60909 and state estimation from measurement data. We configure optimal power flow with costs, flexibilities and constraints and examine grid structures with graph search. Results for each grid element are prepared so that states can be compared and assessed, and remain available for further analysis in Python.
A load flow tells you which voltages settle at the nodes and how heavily lines and transformers are loaded. The AC calculation is a non-linear system of equations solved iteratively. When it fails to converge, the cause often lies in the data, such as wrong line parameters, unclear switching states or implausible loads. We work through such cases systematically rather than loosening tolerances, because a grid model is only as good as its topology and parameters.
The right tool for each question
Long-term energy system and capacity planning belongs with PyPSA, particularly for large grids, long time series or cross-sector systems. Element-based electrical grid calculation for transmission, sub-transmission or European three-phase distribution grids belongs with pandapower. Many projects need both in sequence. An expansion scenario from the system optimisation can be recalculated in detail for selected grid states to check whether voltages and loading stay within permitted limits. We work out the right combination from your question, without you having to commit to a stack in advance.
Tools, data & limits

A scenario is only as robust as the assumptions you can read in it.
Runs, operation & handover
Reproducible energy system modelling
We develop reproducible Python workflows for data import, time series preparation, solver integration, testing and result export. Input data and assumptions are versioned, package and solver versions pinned and scenarios stored as configuration. Small reference networks with known results serve as automated tests, and domain reference cases from your organisation serve as acceptance criteria. If a run is repeated a year later, it either produces the same result or shows clearly which change explains the difference.
We define the result variables together with you before anything is calculated. That prevents an optimisation from producing hundreds of figures of which none answers the actual question. Result exports are checked, documented and structured so they can flow into reports, dashboards or further analysis.
Computing in operation
Larger models need computing time and memory that a laptop cannot offer. We plan where and how runs take place, whether on your infrastructure or in the Anexia Group’s own data centres and cloud infrastructure in Europe. That covers recurring runs, storing results with a reference to data and model version, and the question of which solver licences apply in which environment. The Anexia Group’s certified management systems under ISO 27001, ISO 9001 and ISO 14001 include software development in their scope.
You bring the target picture, data sources and decision criteria, and together we turn them into an agreed model state with versioned assumptions and clear result variables. The outcome is reproducible runs, checked exports and a clear handover for further work in your stack. Anexia Digital Engineering has been developing software since 2006, with more than 40 specialists in Austria and Germany and over 100 Python projects. In the energy sector, these include a Python backend for a solar panel configurator that processes input from the configuration tool. Our wider approach to software for the energy industry is described on the Industry, Automotive & Energy page.
The technical assessment on this page is based on the official project sites and documentation of PyPSA, including its source repository, and of pandapower with its technical documentation. Product names and logos belong to their respective owners, and naming them implies no partnership or certification.
Questions on energy system modelling
Related topics
- Python Development
- Machine Learning Development
Models built with scikit-learn, TensorFlow, PyTorch and Keras that hold up against a baseline and reach your application as a tested inference interface.
- Data Analysis with Python
Verifiable analyses and data apps built with pandas, NumPy, SciPy, Matplotlib and Streamlit, delivered as repeatable pipelines rather than one-off notebooks.
- Python Desktop Applications
Desktop interfaces with PyQt, PySide, Tkinter, PySimpleGUI and Kivy, connected to your Python business logic and packaged for your target platforms.
- Backend Development
Java and Spring Boot, .NET and C#, Python with Django and FastAPI, Node.js with TypeScript and PHP with Laravel - dependable backends and APIs.
- Data & AI Technologies
TensorFlow, PyTorch, Apache Spark, Kafka, PostgreSQL, MongoDB, Elasticsearch, Power BI and Tableau - technologies for data and AI.
Quantify. Your energy question deserves an open model.
Tell us about your question, the data you have and the decision the model should support. Together we work out which tool fits and how the first runs become a model your team can operate itself.
