Simulations Show CO2 Could Keep a Munich Geothermal Heat Battery From Clogging

Beneath Munich, a deeply buried layer of limestone and dolomite already delivers roughly 400 megawatts of thermal power through 42 operating geothermal wells. Now researchers have modelled whether the same formation could do something more ambitious: store summer’s surplus heat underground and hand it back in winter. A new simulation study, published in the journal Geothermal Energy, examines the chemical consequences of running such a geothermal battery in the Malm reservoir of the South German Molasse Basin, and concludes that the main chemical hazard, carbonate scaling, can be managed with carbon dioxide added to the water before it is injected.

During summer, when Munich’s demand for heat falls well below what the geothermal wells can supply, water is extracted at the natural reservoir temperature of about 80 degrees Celsius, heated to 135 degrees, and pushed into a dedicated storage well roughly 2,000 metres below the surface. In winter the flow reverses: the stored water is produced, cooled to 60 degrees in heat exchangers, and reinjected through a separate injection well, while the production wells continue to supply geothermal heat directly. The authors use geothermal battery as an umbrella term for heat stored in geological media, synonymous with geological thermal energy storage.

The chemistry is the difficult part. Heating Malm thermal water to 135 degrees drives carbonate minerals far into supersaturation, because calcite and dolomite are retrograde soluble: they dissolve less readily as temperature rises. Without treatment, the authors’ simulations show that both minerals would precipitate near the storage well, cutting porosity there by up to 2.2 percentage points over ten simulated years of operation. Since the thermal water carries less than one gram per litre of dissolved salts, the expected clogging agents are precisely these carbonates. The team therefore conditioned the water with about 12 millimoles of CO2 per kilogram, roughly eight times the dose reported for an ATES project at Middenmeer, which brings the calcite saturation index to zero and keeps the gas fully dissolved at plant operating pressures.

With CO2 inhibition in place, the simulated picture reverses. Near the storage well, calcite dissolves while dolomite precipitates, but because dolomite reacts about ten times more slowly, the net effect is a porosity increase of up to 0.9 percentage points. The one exception is a scenario in which the reservoir’s high-permeability zone is entirely dolomitic: there, with no calcite available to dissolve, dolomite precipitation produces a slight porosity reduction of less than 0.2 percentage points, although the authors consider it unlikely that such a zone would occur or materially reduce productivity. In the mixed calcite-dolomite reservoir they modelled as the base case, the chemical changes stay essentially confined to within about 50 metres of the wellbore.

The study pays particular attention to the reservoir’s architecture. The Malm is a karstified carbonate formation up to 600 metres thick, in which flow is dominated by discrete zones of enhanced permeability associated with karstification. The model assigns a 20-metre-thick such zone within 180 metres of net reservoir, carrying 70 percent of the total transmissibility of 800 darcy-metres, with the remaining matrix at 1.5 darcy and a porosity of 7.7 percent. Because no field-scale measurements of reactive surface area exist for karstified reservoirs, the team bracketed the problem with two end members: a laboratory-derived ratio of 800 square centimetres per litre representing the matrix, and an idealized karst channel with a half-metre diameter giving only 40 square centimetres per litre.

That choice matters because reaction rates scale with the surface area available to the fluid. In the high surface-area case, chemical equilibrium is reached within half a year at 81 degrees; in the low case, equilibrium is not even approached over the same period. Yet the DZEP still sees the strongest chemical alteration, because the sheer volume of water flowing through it overcompensates for its lower reactivity. In the matrix, where flow is slower and surface area higher, significant porosity changes extend only a few metres from the well. A sensitivity test in which the surface-area ratio was raised by an order of magnitude changed porosity outcomes by less than 0.1 percentage points, a result the authors attribute to calcite’s rapid kinetics dominating the system.

The largest chemical effects of all appear not at the storage well but at the injection well during winter discharge. Water that has equilibrated with calcite at 120 to 135 degrees inside the reservoir still carries dissolution potential when it is cooled to 60 degrees and reinjected. Over ten years, the model predicts about 1,050 cubic metres of dissolved rock in this scenario, compared with roughly 100 cubic metres for a conventional geothermal doublet injecting at the same temperature year-round. Porosity near the injection well rises by up to 2.7 percentage points, against 0.7 percentage points in the conventional case, even though the battery’s injection well operates only six months per year. Distributed across the reservoir volume, this amounts to an average near-wellbore porosity increase of less than 0.3 percentage points, from 7.7 to about 8 percent.

Is that dissolution dangerous? The authors argue not. A thousand cubic metres corresponds to a cube with ten-metre edges, and uncollapsed cavities of comparable size are documented from other Malm wells. Moreover, the simulation deliberately does not couple porosity changes back to permeability, because karstified carbonates often lack a clear porosity-permeability relationship; enlarging existing decimetre-scale conduits is unlikely to change their flow behaviour, while the finer matrix pores are more sensitive but also hardened by the reservoir’s long diagenetic history. The team also notes that water-filled caves in carbonate reservoirs have been shown to remain mechanically stable at burial depths far exceeding 2,000 metres.

The study is explicitly a modelling exercise, and the authors are candid about its limits. The grid is coarse, with 10-metre cells at the well, so processes such as wormholing cannot be resolved; the single-layer representation of the high-permeability zone is a simplification; and the thermal water composition had to be calibrated to exact equilibrium with calcite and dolomite to avoid masking the induced effects. Porosity was calculated from a depth correlation, and the fluid composition derives from a measured sample in northern Munich, corrected for degassing and adjusted for equilibrium. The results should therefore be read as an estimate of the magnitude of chemical change and a process-level understanding, not as a field-validated prediction.

Even so, the direction of the findings is clear. In the simulations, CO2 inhibition prevents formation damage at the storage well, drives modest porosity increases at both storage and injection wells, and keeps all significant chemical alteration within the near-wellbore region. Previous work cited in the paper suggests the storage well could recover 68 percent of injected heat in its first cycle and 87 percent after ten years, with recovery temperatures above 110 degrees, although the mixed production temperature would still require supplementary heating to meet district heating demand. The authors also flag that CO2’s corrosiveness raises long-term well-integrity questions, which they plan to address in a separate publication. Within those caveats, the chemical de-risking presented here supports the case for testing geothermal batteries as a seasonal storage option for decarbonized district heating.

Subject of Research: Reactive transport simulation of chemical interactions during seasonal geothermal heat storage in the Malm reservoir near Munich

Article Title: Chemical interactions of a geothermal battery in the Malm reservoir of Munich

Article References: Hörbrand, T., Nitschke, F., Zosseder, K., & Kohl, T. (2026). Chemical interactions of a geothermal battery in the Malm reservoir of Munich. Geothermal Energy, 14(1), Article 27. https://doi.org/10.1186/s40517-026-00402-8

Image Credits: AI Generated

DOI: 10.1186/s40517-026-00402-8

Keywords: geothermal energy, geothermal battery, Malm reservoir, Munich, seasonal heat storage, reactive transport simulation, CO2 inhibition, carbonate scaling, dolomite, calcite, porosity, South German Molasse Basin