What is the best software for district heating modelling?
It depends on the decision. For screening where a network makes sense: THERMOS, Citiwatts, City Energy Analyst or nPro. For deciding what to build across heat, electricity and storage: an optimiser such as Sympheny, PyPSA, Calliope, oemof, energyPRO or nPro. For pipe hydraulics: pandapipes, STANET, NetSim, sisHYD or Fluidit. For dynamics and controls: IDA Districts or Modelica. For operation: energyPRO, Gradyent or Optit.
Which software can size district heating pipes?
nPro, oemof DHNx and THERMOS size or route pipes at planning stage. pandapipes, STANET, NetSim, sisHYD, Fluidit and IDA Districts verify them hydraulically.
Which tools support multi-year investment optimisation?
Among others: Sympheny (multi-stage runs), PyPSA, Calliope, oemof-solph, ETHOS.FINE and SpineOpt.
Which open-source tools can model electricity and heat together?
PyPSA, Calliope, oemof-solph, ETHOS.FINE, SpineOpt and urbs optimise them jointly. pandapipes with pandapower couples them in network simulation.
Do I need a commercial solver?
No. Free solvers work, especially for linear problems. On large mixed-integer problems, commercial solvers such as Gurobi still solve more problems and solve them faster in public benchmarks, which matters for project time and cloud costs.
What is the difference between optimisation and simulation?
A simulation tells you how a design you specify will perform. An optimisation searches many designs and finds the best one for your objective under the model's assumptions. They answer different questions and are often used one after the other, sometimes repeatedly.
Does Sympheny do hydraulic network simulation?
No. Sympheny models thermal networks as links between hubs with capacity, losses and cost. That suits concept-stage comparisons but does not replace pipe-level engineering, which is done in a dedicated tool.
Can I use Sympheny with AI assistants?
Yes. Sympheny works well with AI assistants: a connector for Claude is available, and an official MCP server is coming soon. The optimisation itself remains a deterministic MILP.