District heating · Dispatch optimisation · Energy systems
A five-step linear programming model for district-heating dispatch using annual demand profiles, electricity-price data, storage assumptions and cost/emissions objectives to compare heat-source scheduling decisions.
Evidence dashboard
How should a Stockholm-style district heating system dispatch CHP, woodchip, geothermal heat, heat pumps and storage when demand, cost, emissions and electricity-market coupling are all active constraints?
A Python/PuLP linear programme was built stepwise: base dispatch, 2030 demand, storage integration, electricity-price-linked operation and emissions minimisation. Each step isolates one modelling assumption instead of mixing every intervention at once.
The input data combine five annual heat-demand profiles, E.ON electricity price data and solar-thermal profile data. The stepwise model tests how dispatch changes when storage, price signals and emissions weighting are introduced.
Representative dispatch signal: woodchip 600 MWh, geothermal 200 MWh, CHP approximately 612 MWh, heat pump up to 50 MWh.
The model simplifies seasonal plant constraints and did not fully activate CHP cogeneration in the base case. Its value is methodological: it shows how heat-source dispatch, storage and market signals can be compared in a reproducible decision-support model.
District heating is a strong fit for urban areas, but dispatch planning becomes difficult when demand varies seasonally and the asset mix includes technologies with very different costs, efficiencies and carbon intensities. The project set out to build a structured optimisation model for Stockholm-style district heating and test how results change as demand grows, storage is added and the objective shifts from cost to emissions.
The model was developed step by step so that each stage builds on the previous one and the effect of each design choice can be isolated.
The local project evidence includes the PRO2 brief, five Python model files, an annual demand workbook with 8761 rows including the header, an E.ON electricity price CSV with 8760 hourly values and a solar-thermal profile dataset. This matters because the project was not a static spreadsheet comparison; it was an optimisation workflow built around time-indexed operational data.
Three core heat generation technologies were parameterised from industry data:
Load 5 (stable commercial demand with low variation) was chosen alongside the residential profile because stable demand allows technologies to operate consistently, avoiding efficiency losses from frequent start-ups and shutdowns and reducing the need for backup capacity.
Woodchip and geothermal operate at their upper limits continuously, providing the stable renewable base. CHP fills the remaining demand gap, ranging between 572 and 651 MWh depending on period. Heat pumps contribute 0–50 MWh intermittently, deployed only when economically or operationally advantageous. No electricity purchases from the grid were required, confirming a high degree of self-reliance in the modelled configuration.
CHP electricity generation remained at 0 MWh during the analysed base-case periods, suggesting that cogeneration was not triggered under base-case cost conditions. This is a key finding: the optimiser prioritised heat output over electricity co-production, pointing to an opportunity in Step 4 where spot-price-linked CHP dispatch could unlock additional value.
The base-case dispatch established a clear priority order: renewable base load (woodchip, geothermal) first, flexible dispatchable CHP second, heat pumps as marginal support. This pattern held even as the objective function changed from cost to emissions in Step 5, because the lowest-cost technologies (woodchip, geothermal) are also zero-carbon.
The cost–emissions trade-off only becomes sharp when CHP — which carries 180 kg CO&sub2;/MWh — is competing with higher-cost but cleaner alternatives. Storage integration (Step 3) reduces dependence on CHP peaks by allowing surplus renewable heat to be shifted in time, which simultaneously reduces costs and emissions.
Relevance
This project is a direct example of process and energy-systems optimisation work: demand modelling, technology cost and emissions parameterisation, linear programming dispatch, and scenario comparison. The five-step structure demonstrates how to isolate the effect of individual design choices — storage, cogeneration, electricity price signal and objective function — in a single modelling campaign rather than building separate models.
The finding that the cost-optimal and emissions-optimal dispatches nearly coincide (because the cheapest technologies are also zero-carbon) is a practically important result: it means the case for renewable base load does not depend on a carbon price in this system configuration. That kind of insight is transferable to industrial decarbonization R&D, where heat pumps, storage, electrified heating and market signals must be assessed together rather than as isolated technologies.