Pilot project: split unit in the Passive House building in Darmstadt Kranichstein

This article is based on a paper published in the 62nd volume of the Research Group for cost-effective Passive House buildings on the topic of “ Cost-Effective Refurbishments”. The original article by Wolfgang Feist, Jürgen Schnieders is availabe in German in the proceedings, which can be accessed here.

Since 2016, an entire residential unit on the west side of the Passive House building in Darmstadt Kranichstein has been heated solely using a “mini-split unit”.

Figure 1: South-west view of the Passive House building in Darmstadt-Kranichstein (left) and the internal part of the unit enclosed in the measurement channel (top right)

The unit was installed in the dining room (ground floor north) on the west wall (see Figure 2; a detailed description can be found in [Feist 2022a]). Measurements of the unit were continuously recorded for 6 years, a publication on the results can be found in [Feist 2022b]. In summary: this solution works very well, both in summer and winter, and the power consumption values of the mini-split unit are extremely low in the Passive House, making it a very cost-effective heating option. Pictures of the indoor and outdoor units are shown in Figure 3.

Figure 2: Ground floor plan showing the location of the split air conditioning unit (red: “Klima” (internal part of the air conditioning)); indoor and outdoor units are connected by refrigerant pipes through the west wall in the dining room (ground floor north). The radiators shown in are no longer in operation.

Figure 3: Installation of the split unit: outdoor unit with evaporator (heating mode) and compressor (left), insulated refrigerant line (centre) and fan coil unit (right).

Due to the very low heating load in a Passive House, it was assumed that a single split unit of this type would be capable of heating the entire three-storey (!) residential unit from the ground floor via the open stairwell. This assumption stems from the buoyancy-induced air exchange in the stairwell (see Figure 4) [Peppes 2001]. The measurement campaign from 2017 to 2025 was designed to verify whether this expectation could be met. Figure 5 shows the measurement technology specially developed for this purpose and installed in the building: a permanently installed measurement channel in which the inner part of the split unit is installed and which acts as a heat meter for air flow; the accuracy achieved, including the measurement deviations of the sensors, is 4.6%; this already permits a valid evaluation of such a heating system.

Figure4: A split unit on the ground floor heats the entire building due to natural convection via the open stairwell.

Figure 5: The volume flow measurement channel (left) is the basis for a heat meter for air.

Figure 6 shows example results of measurements with the channel: the blue (upper) curve represents the measured total heat flow for heating the building, while the red (lower) curve represents the electrical power measured at the same time for operating the device (this includes control, fans and compressor, as well as any power required for de-icing the outdoor unit). On the cold winter day shown here, periods with the compressor running continuously and periods with rather erratic switching between different compressor outputs can be seen. The latter is unfavourable for operating performance – there is significant potential for improving the control regime for the heating operation of split units. In the medium term, further efficiency improvements can therefore be expected for these systems. The unit used here provides economically viable heating operation today already, and further development will make devices of this type even more attractive.

Figure 6: Results from the measurement channel during a cold winter week (left in the picture: the measurement channel). Unnecessarily 'nervous' control behaviour can be seen here (discussed in [Feist 2022b]).

The main purpose of the measurement channel test was to clarify whether comfortable indoor conditions could be achieved with such a solution. For this purpose, the indoor temperatures measured during the winter period were used (for details, see [Feist 2022b]). Figure 7 shows the time course of the average indoor temperature over an entire heating period in which the split unit was the only heating system used – note the very high resolution of the temperature scale. The red symbols represent the measured values, the solid black line represents the values calculated in a Dynamic building simulation with DYNBIL based on the measured boundary conditions [Feist 2023]. The measured values remain within the comfort range throughout; the approximately 0.4 K temperature drops due to night-time shutdown, increased temperatures during public holidays (New Year's Eve/New Year's Day), and some temperature increases during sunny periods (15 January to 21 January 2020) can be seen. Overall, it is clear that the split unit can easily manage the task of heating the entire flat. In the comparison between the measurements and the simulation, the margin of error is around 0.3 K. The DYNBIL model used here can therefore be seen as a validated solution.

Figure 7: Measurement results (red) compared to the DYNBIL simulation (solid black) in winter 2019/2020 of the Passive House building heated exclusively using the split unit [Feist 2024]

The original reason for installing split units today is predominantly cooling. This option was also investigated in field tests in Darmstadt Kranichstein. Here is a brief summary of the results:

The assumption that active cooling leads to high electricity consumption and thus high operating costs in summer is still widespread – there were even fears of an extreme impact on the power grid load (this has a real basis, as the widespread use of split cooling units in the Italian power grid has repeatedly led to peak load problems¹⁾ ). However, measurements at the system in Kranichstein invalidates these concerns: in buildings that have adequate thermal insulation (using EnerPHit as a reference here) and adequate solar shading, the active cooling loads in summer are very low (less than 500 watts of electricity per residential unit, which is significantly less than other loads) and the cooling electricity consumption in summer also remains low (between 0.5 and 3 kWh/(m²a)), see the typical summer situation in Figure 8. Additional calculations were carried out to determine how this demand could develop as climate change progresses: even assuming a +2 K increase, the demand for cooling remains significantly lower than that for heating. It should also be noted that the cooling electricity demand falls within the period of electricity surpluses from photovoltaic generation – this is favourable for the operation of the electricity grid (in Passive House buildings and EnerPHit renovations, the systems can be preferentially operated during periods of PV surplus, when the building can undertake the storage function for up to 18 hours). Reservations with regard to active summer cooling are therefore unfounded under the given boundary conditions, provided that the buildings have undergone an energy retrofit according to the EnerPHit concept (see also [Feist 2025a]).

Figure 8: Is cooling with a split unit a power guzzler

Figure 9 shows the measured room temperatures in four rooms of the building. Cooling is provided solely by the split unit installed on the ground floor on the north side (dining room). The temperature could be kept below 26 °C also on the top floor (yellow curve, “2nd floor”), even during heat waves, while the ground floor remained at 22 °C. This is perfectly acceptable during the summer for normal residential operation. If desired, an even more uniform spatial and temporal temperature distribution in summer would also be possible if another split unit were installed in the upper part of the home – this would still be a cost-effective solution for heating and air conditioning.

Figure 9: Effectiveness of cooling using the split unit on the ground floor (measurement results to clarify this question; cf. [Feist 2022a]).

An ecological assessment of the operation of the split unit in the Kranichstein Passive House building using the PER method (primary energy renewable, [PER PHI 2023]) is reliable. Since exclusively electrical energy is supplied in this case, the electricity consumption values are decisive for the assessment. Figure 10 summarises the measured values for the operating year from October 2019 to September 2020. The total domestic electricity consumption for all appliances (excluding heating and DHW appliances) in the home is shown in green (predominant, almost constant throughout the year at around 250 kWh/month). The electricity consumption for heat recovery ventilation is shown in yellow, the blue pattern represents electricity for hot water generation (see following sections), red represents the consumption for operating the split unit for heating, and solid blue represents the operation of the split unit for cooling. This results in a total of around 33 kWh/(m²a), which is less than the 36 kWh/(m²a) (light yellow area) provided by renewable energy generation (PV on the building and share of electricity generated by the wind turbine owned by the residents). The building can therefore be classified as a net zero energy building. However, even in this case, the consumption values are slightly higher than the renewable generation available in November, December and January due to the winter heating load. This results in a net electricity consumption of around 480 kWh per year for the entire home – in the long term, this energy would have to be provided through seasonal energy storage. Since this is a very small amount in absolute terms, equivalent to around 50 litres of heating oil (or m³ of natural gas), it can be covered, for example, through available seasonal energy stored in the form of biomass ²⁾ that is assignable to the building (for comparison: cooking oil consumption in Germany is currently around 15.7 litres/person per year; consumption of biogenic fuels is around 38.4 litres/person per year; 3 people live in this building).

Figure 10: Annual total electricity consumption in the building compared to renewable electricity generation (yellow area) from systems operated by the owners' association.

The building described in [Schnieders 2022], an end-of-terrace house (or semi-detached house) with a living area of 120 m² will be used in this study (see Figure 11).

Figure 11: Model building of an end-of-terrace house for the simulation of the “existing building”, “GEG” and “EnerPHit” standards; basis for examining the use of split units for heating support.

The model building can be simulated using different insulation standards, e.g. in accordance with an EnerPHit Retrofit Plan (ERP), see Table 1.

Table 1: Characterisation of thermal insulation levels in the scenarios considered

One reason for installing a split air conditioning unit may be the desire to keep the indoor climate acceptable in at least one room during heat waves, given the rising summer temperatures due to climate change. The ground floor (living room, dining room and kitchen) was ideal for this purpose. We placed the indoor unit of the split system centrally on the ground floor (see Figure 14 below).

A modern (reversible) unit of this type can also be used to support space heating. It can be operated in addition to the existing central heating system, but in a typical existing building this certainly would not be able to cover the building's heating demand. For this case of bivalent parallel operation (with priority given to the air-to-air heat pump), Figure 12 shows the percentages covered by the two heat generators for heating. Surprisingly, even in this case, more than half of the heating demand (52%) can be met by the split unit, leaving 72 kWh/(m²a) for fuel based heating. In the event of a serious natural gas supply crisis, the split unit also makes it possible to keep the rooms on the ground floor heated in winter at least – the electricity consumption required for this is quite significant at around 3000 kWh in winter, but remains affordable.

Figure 12: A split unit on the ground floor of a moderately insulated building if used as a supplementary heating system – with the existing radiator heating system and central heat generator continuing to be operated – guarantees comfortable indoor conditions. The share of the energy for heating that the split unit can provide is surprisingly high at over 50%, even in existing buildings.

In principle, it is quite possible to heat a typical existing building using split units alone – however, this would require a fan coil unit (interior part of the split unit) to be installed in every main living area. That's at least three such units (see Figure 13); the investment costs for this usually wouldn't be any less than the cost of replacing the central heat generator in this property. More importantly, however, the operating costs in such a case would remain very high (we will discuss the total costs later in Life cycle assessment and cost comparison) and the poor state of insulation means that the surfaces of windows and exterior walls would continue to be cold. This would result in high radiation temperature asymmetry and a high level of air movement in the rooms. In buildings with little insulation, thermal comfort from a heating system based solely on convectors is actually reduced in comparison with heating systems with higher radiation levels (however, in well-insulated Passive Houses and EnerPHit-renovated buildings this is no longer the case, since higher temperatures of the interior surfaces are automatically achieved here).

Figure 13: Three (!) split units in the “existing building” case as the sole heating system. The level of comfort achievable is borderline. This is also because the surfaces of the building components remain very cold and the level of air movement is high. Operating costs also remain high – therefore this option on its own usually doesn't solve the problem.

The decisive factor now is that, as part of an EnerPHit retrofit plan (ERP), the building is gradually improved in terms of the thermal protection of the exterior components, and improvement of the windows and ventilation system. The better the standard achieved in this step-by-step way, the

• higher the coverage percentage that the split unit can achieve,
• lower the power consumption for the split unit and
• thus disproportionately lower the remaining fuel consumption of the old heating system.

Even in a moderately insulated building, the attempt to heat the entire building using only one split unit fails, as shown by the temperature profile in Figure 14. Acceptable temperatures are only achieved on the ground floor – the bathroom on the upper floor becomes warmer for a short time in the morning due to high internal heat gains (green). The living areas on the upper floor sometimes drop below 16 °C.

It therefore makes sense to operate the split unit in this building only to support the heating system and to continue operating the existing heating system (radiators with oil or natural gas-fired heat generators) in the rooms on the upper floor, at least until the existing heating system becomes due for renewal.

Figure 14: One split unit on the ground floor of the moderately insulated building and no other heating system. Comfortable conditions can only be achieved on the ground floor with this.

After all the construction measures of the step-by-step EnerPHit renovation have been implemented, the heating demand and the maximum heating load are significantly reduced. Due to this, the ratio of the thermal coupling of the interior spaces with each other to the thermal coupling towards the outside shifts, resulting in the indoor climate becoming more uniform (purely passively). Accordingly, the temperature differences between the ground floor and the rooms on the upper floor now remain below 3 K, even during a cold spell. The single split unit on the ground floor can almost cope with heating of the upper floor rooms as well in this way (Figure 15). Additional heating energy can now be provided by the old central heating system, which may still be in place (although this becomes less and less likely as the system ages), or via individual small electric convectors or infrared radiators that are only switched on when needed. The coverage percentage of the air-to-air heat pump would be 80 to 90% in this case. However, a better (and overall more cost-effective) solution is to install an additional split air-to-air heat pump on the upper floor, which we will discuss in the following section.

Figure 15: A split unit on the ground floor of an EnerPHit-renovated building, used here as the sole heating source. The level of comfort achieved is acceptable, but not outstanding.

The installation of a second fan coil unit on the upper floor brings a noticeable improvement, especially for summer cooling; based on natural convection, this can then provide cooling loads in summer more easily compared to a single unit on the ground floor. Here, we positioned the unit in the corridor on the upper floor – Figure 16 shows that in this way comfortable conditions can be achieved throughout the home, even in the coldest periods.

This can be implemented with two mini-split units or with a multi-split, which will then only have one outdoor unit. In the latter case, the cost of installing two indoor units (for the north and south sides) is not very high, due to which the control characteristics can be improved even more.

Figure 16: Two split units in an EnerPHit renovated building, used here as the sole heating system. The level of comfort achieved is good – good conditions can thus be guaranteed, even in summer.

As we have seen in use in renovated buildings, air-to-air heat pumps can supply the space heating energy, but how will hot water be provided with such a solution?

One option is to use decentralised instantaneous water heaters or electric compact storage heaters – as already discussed also for central heat pumps for the corresponding case without existing central hot water distribution (see Hot water [Schnieders 2025]). However, this solution is neither particularly cost-effective (direct electricity!) nor environmentally friendly. The effort required in the flats is also not as small as it might appear at first glance: due to the high connected load of the instantaneous water heaters, it is usually necessary to install a three-phase current line (400 V) in the bathroom.

Decentralised DHW heat pumps offer a better solution: compact DHW heat pumps that use air as a heat source are available on the market in various designs (120 to 300 litre storage tanks) and are comparatively inexpensive. Installation costs are also low, and operation with 230 V AC voltage is possible, meaning that a three-phase current line does not have to be laid.

Figure 17 shows an example solution from the EnerPHit pilot project in Darmstadt-Arheilgen, where a DHW heat pump with outside air as the heat source was installed. The picture shows the construction work: the exhaust air duct is already insulated, while the outside air duct is recognisable as a spiral duct (this duct must still be insulated, as the cold outside air would otherwise lead to condensation over long periods of time).

Figure 17: How can a solution for hot water generation using air-to-air heat pumps be implemented

The PHI conducted an accompanying measurement campaign for this project as well; the setups for the measurements on the water side are shown at the bottom right of the figure. This measurement has only been active for six months, and the seasonal coefficient of performance determined from this for the winter half-year is around 2.2. This is not a bad value for hot water production, given the requirements existing for the temperature level. The measurement results for summer operation will also become available and made accessible in due course.

Yet another unconventional approach was chosen by Benjamin Krick for the Seeheim-Jugenheim pilot project (see Figure 18 ): here, the hot water heat pump with an 80-litre water tank and R290 refrigerant is operated using room air as a heat source. The supply air for the integrated evaporator is taken from the stairwell right at the top, and the (cold!) exhaust air from the unit is distributed into the bedroom via a duct. In summer, this provides welcome (and virtually free) cooling, but in winter the approximately 300 watts of heat source output needed by the heat pump's evaporator must be provided by the space heating system in addition. This increases the heating demand – in winter, this only makes sense with regard to energy efficiency if heating is provided in a highly efficient manner – which is the case in this building thanks to the already installed air-to-air heat pumps. Winter operation with this solution is thus based on a heat pump cascade which, as we expect, will still have a significantly better overall COP than “1” which is achieved for the alternative direct electric solution. This design variant is also being monitored by the PHI, but results are not yet (shortly after installation in summer 2025) available.

Figure 18: Case study 2: Hot water heat pump operation with room air as a heat source (!) in the Seeheim-Jugenheim pilot project (Dr. B. Krick). In this case, the source heat for the hot water heat pump must also be provided by the split units for heating in winter – however, this heat pump cascade still has a significantly lower electricity requirement than hot water generation using instantaneous water heaters.

Finally, Figure 19 shows domestic hot water storage tanks for use with multi-split air source heat pumps which are also available on the market. These have an integrated refrigerant condenser and are connected to the outdoor unit via refrigerant lines. Additional fan coil units for heating can be supplied from the same outdoor unit at the same time.

Figure 19: Case study 3: Another solution with a multi-split unit: here, there is only one outdoor unit that supplies several indoor convectors for space heating (and cooling) and a hot water heater.

In some cases, the outdoor units of such air source heat pumps are intrusive or are perceived as unsightly in the urban setting. For such cases, solutions for the outdoor unit have been developed at the University of Innsbruck that can be completely housed in the insulated façade of a Passive House building or an EnerPHit renovation – the approx. 25 cm of façade insulation there provides sufficient space to accommodate small compressors and the associated air/refrigerant heat exchangers (see Figure 20, [Monteleone et al 2025]). The outside can then be designed with adapted intake covers to match the respective façade. In EnerPHit building renovations, the space demand for this is acceptable since only correspondingly low outputs are required from these heat sources. Other concepts are being followed in Chinese projects, where the mounting brackets for the outdoor units of the split systems are integrated into the architectural design of the façade.

Figure 20: Case study 4: The outdoor unit of an air-to-air heat pump can be housed in the thermal insulation of the exterior wall in energy-efficient buildings. This integration allows for a façade design that is adapted to the respective situation, eliminating the need for conspicuous outdoor units.

Here we will present the results for decentralised heating solutions in EnerPHit renovated buildings in a comparison. The capital costs for the air-to-air heating systems as well as the associated domestic hot water systems are included (each with a useful life of 20 years, effective real annual interest rate of 2%, electricity costs of 0.30 €/kWh). The costs for residential ventilation are also taken into account – the structural renovation costs are the same in all cases considered here and are not added up for the sake of clarity. Elsewhere, we have shown that under the given boundary conditions, implementation of the measures proposed for EnerPHit here is economically viable [Theumer 2025]. We will initially base our calculations on four dwelling units (DU) of the apartment building under consideration, between which the investments in central building services may be distributed (the influence of the number of DU will be discussed later).

The remaining PER demand (renewable primary energy demand) is a suitable indicator for the ecological assessment (especially climate protection aspects). Figure 21 shows this after the results given in [Schnieders 2025] for the central solutions listed here (first 5 column stacks); these result in average values of around 38 kWh_PER /(m²a) and differ only slightly from one another – only the variants with a 4-pipe system and the electric instantaneous water heaters stand out. The decentralised solution with instantaneous water heaters is also comparable to the latter. The remaining four decentralised solutions discussed here have a PER demand of around 35 kWh_PER /(m²a), which is around 9% cheaper than the centralised solutions; this is outside the margin of error of this calculation – the decentralised solutions have lower distribution and supply losses, and the annual performance factors of the air source heat pumps are comparable. It is interesting to note that, in terms of PER, the outdoor air heat pump in the EnerPHit building hardly differs from the indoor air-based solution (however, the comfort aspect still needs to be checked; we expect results on this to be presented at the Passive House Conference 2026); it can be seen that the heating demand increases slightly with the indoor air solution, but the hot water heat pump works more efficiently. The use of a Passive House compact heat pump unit has also been included as Case H) – these units combine ventilation, heating and hot water generation in one device. They have proven highly effective in Passive House buildings, but for use in an EnerPHit renovation, heating using the supply air from the ventilation system on its own is not sufficient, and the existing radiators, for example, must be used as a supplementary heating. This results in slightly higher PER demand values for the heating part, but these are still lower than those of the central variants.

Figure 21: Comparison of scenarios: the PEr demand for different solutions for heating and hot water supply in an EnerPHit renovated building.

For the economic assessment, the present value of the total life cycle costs per m² of living space is used; the boundary conditions are the same as in [Schnieders 2025]. Figure 22 shows the results for the case with four residential units in a building section; with 356 €/m² (+/- 8), the central systems are all approximately at the same level (including the case with the instantaneous water heater). The decentralised solutions also differ only slightly from each other at 311 €/m² (+/- 13), but on average they are around 12% cheaper than the centralised solutions for four residential units. This means that besides more flexible application (step-by-step over time) and lower costs, decentralised solutions also result in lower financial burdens; however, it should be noted that these solutions are only practicable for EnerPHit renovated buildings. In buildings with a poorer level of insulation, the investment cost benefits are lost (more split units are required), operating costs increase significantly (higher air temperatures, longer running times), and comfortable operation is more difficult to achieve.

Figure 22: Comparison of scenarios: life cycle costs over 20 years for different solutions for heating and hot water supply in an EnerPHit renovated building.

The life cycle costs of centralised solutions depend heavily on the number of residential units that can be supplied from the same central system, as the costs of centralised installations that are not output-dependent only increase slightly with the number of residential units; the costs of centralised solutions therefore decrease as the number of residential units increases, while the costs of decentralised solutions remain at roughly the same level (if the average size of the residential units does not change significantly). Figure 23 shows the total life cycle costs of four selected variants depending on the number of residential units (each with 80 m²). It can be seen that air-to-air heat pump solutions can be particularly attractive in single-family homes and multi-family homes with fewer than 5 residential units, while for more than 10 residential units the costs for the various solutions are barely any different.

Figure 23: Scenarios – Influence of the number of residential units (RU) in the EnerPHit building on life cycle costs.

• Modern split units are efficient, quiet and reversible: they can also be used for heating.
• They can be used in a step-by-step manner, e.g. starting with summer cooling for one room.
• Their use for heating is also efficient and practical in well-insulated buildings (e.g. in an EnerPHit renovated building).
• Tests in practice show that one unit is sufficient in a Passive House, and 1 to 2 units per apartment in EnerPHit renovations are enough.
• Measurements have confirmed the results of the simulation [Feist 2024].
• Hot water can also be provided on an apartment-by-apartment basis using hot water heat pumps.
• There are several implementation options for all these approaches (including wall-integrated outdoor units).
• These solutions are advantageous in economic terms and ecologically sustainable in an EnerPHit renovated building (very low PER demand and low maximum load).

The associated potential solutions have already been demonstrated in practical projects. At the 28th International Passive House Conference, the experiences gained in these projects were discussed in more detail.

[Feist 2022a] Feist, Wolfgang: Installation of a split unit, accessed on 04.07.2025

[Feist 2022b] Feist, W.: Heating with a split unit? Passive House in Darmstadt Kranichstein – experiment on heating and cooling from a spatially concentrated source (German) Innsbruck University Press, 2022, ISBN 978-3-99106-078-9., accessed on 04.07.2025

[Feist 2023] Feist, W.; Hasper, W.; Huneke, S.; Peper, S.; Schnieders, J.: Passive House Institute: InSituVerification/IEA Annex 71 – Development of a method for determining the energy efficiency of buildings based on optimised in situ measurements; sub-project: Users – Technology – Building envelope: Factors influencing the operating behaviour of buildings. Part 1: Validation of algorithms for thermal building simulation based on field measurement results

[Feist 2025a] Feist, W.:Should active cooling be avoided? Passipedia .

[Feist 2025b] Feist, W.: Room air conditioners for heating (only German available) , accessed on 04.07.2025

[Monteleone et al 2025] Monteleone, W.; Pinotti, R.; Ochs, F.; Dermentzis, G.: Hygrothermal analysis of prefabricated façades with active components for minimally invasive renovations. Journal of Building Engineering, 107, 2025. https://doi.org/10.1016/j.jobe.2025.112735

[Peppes et al 1997] Peppes, A. A.; Santamouris, M.; Asimakopoulos, D. N.: Measurement and CFD modelling of airflow through stairwells, Bibinf UK, Air Infiltration and Ventilation Centre. Proceedings of “Ventilation and Cooling”, 18th Annual Conference, Athens, Greece, 1997, Volume 2, pp. 513-522

[PER PHI 2023] Renewable primary energy – PER. Passipedia.

[PHPP]Passive House Planning Package. Passive House Institute, Darmstadt (1998-2020), here: 9th edition 2017

[Schnieders 2022]Schnieders, J.: Mini split unit as support for space heating(German available) Passive House Institute, May 2022. , accessed on 04.07.2025

[Schnieders 2025] Schnieders, J.: Overview: Solutions for heating and cooling In Protocol Volume No. 62 of the Research Group on Cost-effective Passive Houses, Passive House Institute, Darmstadt

[Theumer 2025] Theumer, S.: Introduction: Cost-effective renovation In Protocol Volume No. 62 of the Research Group on Cost-effective Passive Houses, Passive House Institute, Darmstadt

Additional example for heating with a minisplit: Straw Passive House(German)

Cost-effective retrofits – Summary and outlook

Research Group 62 - Cost-Effective Refurbishments

Refurbishment with Passive House components

Building Services