Flexible Heat: Cutting District Heating Costs with mFRR Ancillary Services
– Reducing the Cost of Heat through Flexible Load and Thermal Storage

Written by Jeppe Urup Byberg and Jorrit Wronski · 18th of May 2026

Why Read This Article?

With more variable renewable generators participating in the electricity markets, the energy grid becomes increasingly more dependent on flexible load. This flexible load can be provided by heat suppliers consisting of heat pumps or electric boilers coupled with thermal-storage tanks. This article investigates the effects of offering ancillary services and estimates that the cost of heat can be lowered by over 50%. Specifically, this analysis investigates the impact of offering 1 MW electricity mFRR ancillary service using a heat pump with 10 MW thermal power capacity with a COP of 3.

Introduction

Electrified district heating systems can play an active role in the electricity market. With optimized production plans, heat suppliers can shift electricity consumption without compromising heat delivery. Shifting electricity consumption enables participation within the ancillary service markets.

When the actual energy production or consumption deviates from the plan, the Transmission System Operator (TSO) activates the ancillary service to balance the grid. Among the different ancillary services, the manual Frequency Restoration Reserve (mFRR) market presents an attractive opportunity for heat pump-based systems. Because mFRR operates on a 15-minute market resolution and allows a longer activation time than faster reserve products, heat pump systems with more limited ramping capabilities can still participate. This broadens the range of eligible technologies, as even larger ammonia heat pump systems may be able to adjust their load sufficiently within the required activation window.

Offering mFRR can be embedded into the CleanAction framework, such that it is possible to make the production plan with ancillary service offerings in mind. This article investigates the economic and operational impact of participating in the mFRR market for a district heating system equipped with a heat pump and thermal storage.

Market Opportunities

The production plan for tomorrow is submitted to the TSO at 12:00 each day. This plan is referred to as the day-ahead (DA) plan, which includes the expected production for every 15 minutes from midnight to midnight. Submitting the DA plan each participant must estimate their consumption or production up to 36 hours into the future, and with some participants being renewable based, this can yield a high level of uncertainty. As forecasts are updated and uncertainties materialize, deviations from the DA plan can be corrected through intraday trading or balanced through ancillary service markets.

This is where the offering of mFRR is relevant for the heat suppliers. When using the CleanAction framework, the knowledge of power consumption, heat production and State-of-Charge (SoC) of heat storage facility is optimized based on forecasts on energy prices and weather. The CleanAction optimization ensures a heat production plan that is within the physical constraints of the facility, whilst minimizing the expected costs of production. With the insights from CleanAction, the effects of activation from the ancillary service market in up or down regulation direction is embedded into the framework. An activation can be interpreted as shifting the expected SoC-trajectory based on which direction the regulation is activated. Upward regulation can be interpreted as electricity grid needing more energy, hence when activated at upward regulation the electrical power usage of the heating assets must be decreased, yielding less thermal power. On the other hand, downward regulation corresponds to withdrawing more energy from the grid. To do so, the electrical heating assets must have capacity to withdraw the extra amount of electrical power.

For both types of regulation, the SoC trajectory must always be within the limits, i.e. be greater than 0% and less than 100%. To be activated in the ancillary service market, the market clearing price must be greater than or equal to the bid offered for the ancillary service.

Simulation and Methodology

To evaluate the effects of ancillary services a case study is investigated. This involves a district heating system that annually supplies 20 GWh heat. This is done with a heat pump with 10 MW thermal power capacity, connected to a 70 MWh heat storage facility corresponding to 1500 m³.

To get a proper estimate of the effects of mFRR, data from 2025 have been used in the CleanAction framework with perfect foresight.

The mFRR activation is purely algorithmic, with very few assumptions. The mFRR bids have a granularity of 1 MW electricity, such that the regulation performed must be 1 MW, 2 MW, etc. In this simulation only 1 MW bids are submitted, with a COP of 3, this corresponds to 3 MW heat. The algorithm to decide if a bid in the ancillary service market is submitted is firstly checking if an increase or decrease in electricity by 1 MW is within the capacity.

With the CleanAction insights the trajectory of the State-of-Charge (SoC) for all time steps in production plan is obtained. Cost minimization programs will utilize the storage facility, such the SoC occasionally will be at 0% or at 100%. An activation of the mFRR shifts the anticipated SoC of the storage facility. The mFRR activation algorithm is then secondly checking whether the SoC-trajectory is within the boundaries of the system at all time steps, i.e. the activation will yield that all future SoC values will be between 0% and 100%. Both checks are based on information that is available with the CleanAction insights.

If the activation algorithm allows for ancillary services, and the market clearing price is greater than the bid price, then the service is activated if the market needed regulation for that specific time. The algorithm is not optimizing which time step to offer ancillary services but will offer the service if the system can deliver it and the energy grid needs the activation.

The horizon for which the SoC-trajectory must be within bounds is a modelling choice. Within the CleanAction framework the horizon for optimization is 3 days, the shorter the SoC-trajectory is within bounds the algorithm will choose to participate more often. For this simulation the SoC-trajectory must be within bounds for 2 days. For clarity, a timeline for the markets and the CleanAction procedures are visualized. Activations in the ancillary service market must be feasible with the production plan for the coming two days. The ancillary service market is operating every 15 minutes, and feasibility of this market must also be checked simultaneously.

For this simulation setup the ancillary service market is investigated under different scenarios. Up regulation corresponds to lowering the electricity consumption of the heat pumps according to the production plan. The objective of the production plan optimization is to minimize the operational expenses, hence buying electricity at the cheapest hours. With this observation offering up regulation may increase the operational expenses. Because of this a setup with only offering down regulation is investigated. The set of scenarios are listed:

• Reference case: CleanAction optimizes the production plan without offering ancillary services.
• Offering up and down regulation: CleanAction optimizes the production and offers down and up mFRR regulation.
• Offering only down regulation: CleanAction optimizes the production plan and offers only down mFRR regulation.

With the ancillary service implemented into the CleanAction toolbox, it is possible to analyze the effects of offering ancillary services. It is assumed that participation in the mFRR ancillary service market is purely as price-taker, with no influence on the cleared market price. When participating in the market, the bid submitted is at 0 €/MWh.

From these results it is evident that when offering ancillary services, the cost of heat is halved from 16.62 €/MWh in the reference case to the 8.30 €/MWh when offering both up and down regulation. The case with only offering down regulation showcases, that the electricity expenses can be lowered by roughly 50,000€ annually when, corresponding to a Cost of Heat at 9.17 €/MWh. Only offering down regulation is a more conservative strategy, as this does not put as much trust into the demand forecasts, and will only increase the SoC when activation occurs. Although this is more conservative, the setup reduces the annual cost of heat 44%.

Introducing regulations to CleanAction reduces the cost of heat, however, the CleanAction production plan optimization is initially minimizing the operational costs of the district heating facility, and then participation in the ancillary service market is performed algorithmically. When minimizing the cost of the operation in the production plan, the SoC-trajectory will at time steps be 0% or 100%, such that no activations are possible. It is investigated whether DA-production plan should be constrained to be between 20% and 80%, such that the more mFRR activations are possible.

• Offering up and down regulation with SoC limits: CleanAction optimizes the production plan ensuring that the SoC at every time step is between 20-80%, allocating storage for both up and down regulation.
• Offering only down with SoC limits: CleanAction optimizes the production plan ensuring that the SoC at every time step is between 20-80%, allocating storage for only down regulation.

These cases yield the following results compared to the reference case:

The results indicate that when the production plan is optimized to allow for participation in the ancillary service market, the cost of heat can be decreased by roughly 60% when offering both up and down regulation. When optimizing the production plan with SoC limits, the expenses in the DA market are approximately equal to the optimization without the SoC limits, however the mFRR revenue increases when optimizing with capacity for regulation.

In this article the SoC limits have been chosen to be 20%. This has been to make a fair comparison between the cases across setups and display the modelling possibilities. Only offering down regulation and having a lower SoC limit is constraining the model to not utilize the lower 20% of capacity in the storage facility. A future article will analyze the sensitivity of the setups.

Key Findings

• Offering mFRR ancillary services can significantly reduce the cost of heat for district heating systems with electric heat production.
• A 10 MW thermal heat pump connected to a 70 MWh thermal storage facility can provide meaningful flexibility to the electricity grid.
• Offering both up and down regulation reduced the annual cost of heat from 16.62 €/MWh to 8.30 €/MWh in the first simulation setup.
• Optimizing the production plan with SoC limits increased mFRR revenue and reduced the cost of heat to 6.75 €/MWh.
• Only offering down regulation is a more conservative strategy, but still provides substantial cost reductions.

Conclusion

This article presents the effects offering ancillary services, manual Frequency Restoration Reserve (mFRR), for a district heating plant. The mFRR service is considered, because with a longer response time, a broader range of heat pump technologies can participate.

To illustrate the impact of the ancillary services a case study consisting of a 10 MW thermal power heat pump with a 1500 m³ thermal energy storage is used to supply the annual heat demand of 20 GWh. The production plan is generated using the CleanAction toolbox, after which it is algorithmically determined whether mFRR bids can be submitted without violating operational constraints.

The analysis compares setups where the CleanAction production plan is allowed to use the full storage capacity with setups where the storage state of charge is constrained between 20% and 80% at all time steps. All setups involving ancillary services reduce costs compared to the reference case. When the production plan is optimized to increase the plant’s ability to participate in mFRR, the cost of heat is reduced by approximately 60% compared to the reference case, reaching a final cost of heat of 6.75 €/MWh.

Even a more conservative approach, where heat production is not reduced solely to enable ancillary-service participation, reduces the cost of heat to 7.72 €/MWh when SoC limits are included in the CleanAction production plan.