Towards a universal framework for heat pump monitoring at scale

1.1 IMPORTANCE OF HEAT PUMP MONITORING

The decarbonisation of heat in buildings is likely to heavily feature electrically powered heat pumps (IEA 2021). Many European countries have targets for heat pump roll out in homes, and the European Union (EU) is targeting 50–60 million heat pumps installed by 2030 (EHPA 2024; Eurofound 2024). While the technology is mature (Barnes et al. 2024), and there are European performance standards for heat pumps based on laboratory tests (EHPA 2025), in-situ (i.e. in-home) heat pump performance has been shown to be highly variable (Carroll et al. 2020; O’Sullivan & O’Donovan 2021). Collecting good-quality data on how heat pumps actually operate and perform as they are rolled out across building stocks is crucial to understand household energy bills, electrical power demand, carbon emissions, indoor conditions in homes, as well as system efficiency in a range of building types with different user behaviours. This in turn is key in both building confidence in the technology and identifying areas for improvement (Bradford & Byrne 2013).

Collecting good-quality data can, however, be very challenging and resource intensive (Domoney et al. 2023; O’Donovan & O’Sullivan 2021a, 2021b). First, there is no standard method. This creates work for every project, e.g. researchers in Fraunhofer undertook ‘intensive analysis and evaluation of the available monitoring technology’ before being able to undertake their intended research (Gunther et al. 2020). Second, monitoring systems can require multiple parties to be on-site to install and test them (Gunther et al. 2020), and the equipment and data access can be expensive (Domoney et al. 2023). Third, many problems have been observed with data quality (Colclough et al. 2022; Love et al. 2017; Lowe et al. 2017), including incorrect installation of sensors, sensor calibration, degradation or power issues, and data-transmission issues. Brudermueller et al. (2025) recently collected data from more than 1000 heat pumps of a single brand across Europe using sensors already installed by the heat pump manufacturer (see Section 1.2). However, the authors did not have access to the sensor specifications to calculate the uncertainty on the measurements or to compare their results with those from other manufacturers. All this means that collecting reliable data from large numbers of heat pumps, to gain a representative picture of operation and performance across different building and occupant characteristics and different heat pump brands, is costly and difficult.

In response to this, the aims of this paper are twofold:

• To explore in depth the challenges involved in collecting good-quality heat pump data through two field-monitoring projects.

• To propose a solution to overcome these challenges: a universal framework for heat pump monitoring at scales beyond bespoke installations or projects and potentially applicable to all newly installed heat pumps.

1.2 BACKGROUND TO IN-SITU HEAT PUMP MONITORING

Key variables to measure heat pump performance consist of electricity consumption and heat output, in order that the coefficient of performance (COP) of the system at a particular operating boundary can be measured (Zottl et al. 2011). Additionally, outdoor temperature is important for contextualising the results, and indoor temperature, hot water temperature, flow and return temperatures can help explain performance metrics. Finally, a range of other parameters can be ascertained for a deeper understanding of heat pump behaviour, such as operating mode and compressor modulation. Data are normally collected as a time-series and different temporal resolutions may be needed to produce various metrics or study different effects. Important work was undertaken over a decade ago (Zottl & Nordman 2011) to standardise heat pump efficiency metrics by introducing the ‘SEPEMO boundaries’ (seasonal performance factor and monitoring), which define what heat outputs and electrical inputs are included in each metric, but do not specify how the measurements should be carried out.

There are two main ways of collecting monitored heat pump data, which will be referred to here as third-party monitoring and onboard monitoring. Third-party monitoring consists of hardware, software and data storage purchased from sources other than the heat pump manufacturer. The hardware can be fitted onto the heat pump at the time of installation by the heat pump installer, or afterwards. Onboard monitoring refers to data collected by the heat pump’s internal systems and stored either within the heat pump or sent to manufacturers’ servers. Onboard monitoring takes place for diagnostic purposes and to show the householder the status or performance of the heat pump. It has also recently begun to be used for large-scale heat pump performance assessment (Brudermueller et al. 2025).

In this study, both third-party and onboard monitoring are examined concerning their suitability for heat pump monitoring at scale.

2.1 DESCRIPTION OF TWO FIELD-MONITORING PROJECTS

The Irish monitoring was carried out as a pilot study within the Monitoring and Comparing of Air Source Heat Pumps (MacAirH) project (SEAI 2024), co-led by University College Dublin (UCD) and Munster Technological University (MTU) in Ireland; this project will be referred to as ‘MacAirH’ throughout. The overall aim of the wider MacAirH project was to monitor, compare, model and analyse the in-use performance of a large sample set of retrofitted heat pumps in different types of buildings. In order to do this, 40 air-to-water and exhaust-air heat pumps across Ireland were monitored in domestic (and a small set of non-domestic) buildings.

This paper focuses on the pilot stage, in which the research team selected a subset of six homes containing different brands of heat pumps to evaluate different monitoring strategies. Several third-party monitoring systems were installed and compared with onboard monitoring systems in terms of their ease of installation, cost and data quality. In order to extract data from the onboard systems, extra hardware and communications devices had to be installed, as described further in the results section. Participants did not receive financial incentives for participation, but were offered a summary of their heat pump’s performance.

The British field-monitoring project was carried out as part of the Energy Demand Observatory and Laboratory (EDOL) project (EDOL 2023), led by University College London (UCL) and the University of Oxford, as a pilot study to determine the best type of heat pump monitoring to undertake later on in the project. EDOL used a different approach to MacAirH, recruiting heat pump users who had themselves set up different types of monitoring—in some cases multiple types of monitoring for the same heat pump—and working with them to understand their monitoring set-ups and any differences in the quality, accessibility and costs of different types of monitoring. Six participants were recruited from a UK-based online forum hosted by the company Open Energy Monitor by responding to a forum post advertising the study. Most of the participants had purchased heat pump monitoring solutions from the organisation. The six participants each received £100 in shopping vouchers in exchange for sharing data with the research team, who analysed the data and discussed the findings with the participants.

Multiple heat pump brands and types of third-party monitoring were present within the two projects; the aim of this paper is not to compare brands but to highlight general findings. Typically the same heat pump manufacturers serve the Ireland and UK markets, rendering the two field-monitoring projects similar in their scope. However, domestic heat pumps are set up in slightly different ways in different countries, e.g. in Ireland it is less common to put antifreeze in air-source heat pump fluid than in the UK. The Irish homes studied were smaller than those in the UK sample, leading to more challenges in installing monitoring kit due to space constraints. The UK project contained one ground-source heat pump; apart from this, all heat pumps in the two projects were air-to-water. Data collection took place between October 2023 and December 2024 for the MacAirH pilot; EDOL participants provided historic and current data from August 2022 to December 2024.

3.1 DATA AVAILABILITY

‘Data availability’ refers to what data exist. This differs from how easily data can be retrieved, which is discussed in Section 3.2 on ‘data accessibility’.

In general, both third-party and onboard monitoring were found to produce data that can be used to calculate heat pump performance and interpret it to a degree, while onboard monitoring contains a wider set of variables which can allow further interpretation.

Third-party monitoring normally focuses on key performance-related variables such as heat output (measured using flow rate, flow and return temperatures) and electricity input. High resolution data can be produced: EDOL and MacAirH both encountered systems giving sub-minutely data. This allows for key effects to be understood, for example, in start-up and shut-down. MacAirH initially encountered a third-party monitoring system not specifically tailored for heat pump analysis that provided data in 15-min intervals. It was found that this was not high enough resolution to fully understand system performance. However, it was also found to be extremely challenging to increase the monitoring time resolution to 1 min. This highlights that monitoring systems not specifically tailored for heat pump analysis or the study’s requirements can be far less effective or useful, even if they are highly accurate and costly.

The third-party monitoring kit requires dedicated installation, and MacAirH documented availability issues in terms of finding installers with tools and components to place temperature sensors in the centre of the fluid flow, especially for newer Alu-Plex pipes which require expensive bespoke power tools to install them correctly. Furthermore, MacAirH also found plumbers not allowing flow sensor installation due to a fear of pressure drop causing a fault in the heat pump system for which they would be liable, and electricians unable to retrofit submetering onto the heat pump circuit due to the heat pump electrics not being installed according to regulations. Finally, measuring the heat output of the heat pump while operating in domestic hot water (DHW) mode in a split heat pump system was not possible with the third-party monitoring systems available to the MacAirH project. This was due to the entire DHW circuit being contained within the indoor unit of the heat pump. The flow meter from the third-party monitoring system could not be installed inside the heat pump unit as it would void the manufacturer’s warranty and due to space constraints. The only option available in these circumstances was to install the flow meter on the space heating circuit, thus losing the heat output measurement on the DHW circuit.

Onboard monitoring, as mentioned above, was found to consist of a wider set of variables. As well as heat, electricity, and flow and return temperatures, other temperatures around the system (including room temperature and outdoor temperature, refrigerant temperature), heat pump operation mode (e.g. Legionella cycle mode, defrost mode, hot water mode), and number of cycles are logged by the heat pump, with some differences in available variables between brands. Variables were found in EDOL and MacAirH to be available at different time resolutions between one day and 30 s, depending on the set-up, and with no standard resolution amongst brands. Onboard monitored data were found in EDOL and MacAirH to be unavailable in some cases if occupants switched heat pumps off over the summer; it was not always clear when this had happened as opposed to the monitoring system failing. EDOL also found a case in which a heat pump factory reset led to historic onboard monitored data being lost.

3.2 DATA ACCESSIBILITY

Accessible data are important for occupants, researchers and third parties elsewhere in the energy system (e.g. demand response aggregators) to understand the heat pump’s behaviour and performance.

Third-party monitoring equipment is designed to give accessible data, while this is not always the case for onboard monitoring. Overall, onboard monitored data accessibility was found in EDOL and MacAirH to vary by heat pump brand. All brands allowed customers to view energy data (and some showed heat pump performance) on the heat pump controller. There were also apps available for viewing data (and controlling heat pumps), although some manufacturers charged for this, and some brands did not have internet connection incorporated as standard and so required the purchase of a gateway by the customer.

In terms of researcher access to data, some brands were found to make onboard monitored data available via an open protocol, i.e. modbus. However, accessing this generally required additional equipment which was not installed by default. Other brands used proprietary protocols; these data could again be retrieved by installing additional equipment (e.g. a microcontroller that converts the data to a different protocol). Some types of heat pump included a USB port or SD card slot where onboard data could be logged locally; however, this requires a person to download the data before they are overwritten, which can be as soon as six months.

In the EDOL study, expert participants had put in significant work to extract the onboard monitored data in time-series form, which required expert knowledge, coding ability and access to heat pump manufacturer documentation, which had proved not always straightforward to obtain. In the MacAirH study, extra hardware (including to connect the heat pumps to the internet) was installed by researchers to retrieve onboard monitored data in time-series form.

3.3 DATA QUALITY

In general, it was found that while both third-party and onboard monitoring can have data quality issues, the system boundaries and measurement uncertainties on third-party monitoring are known, whereas in onboard monitoring, they are not set out by heat pump manufacturers.

Third-party monitoring equipment is by definition specified independently of the choice of heat pump and the purchaser has the option of buying ‘MID approved’ meters, referring to conformity with the Measurement Instruments Directive (2014/32/EU) (European Union 2014) and requiring accuracy within certain limits.¹ This should lead to high-quality data as long as the equipment is correctly installed. The need for in-situ installation, as opposed to factory installation of onboard monitoring, increases the risk of installation error.

For example, EDOL found one case in which the flow temperature was above the return temperature when the compressor was on but below it when off (Figure 1). This could be explained by the return temperature sensor being plumbed in while the flow temperature sensor was strapped on. This does not reflect the proper procedure for installing third-party heat metering but was also found in a previous UK field trial (Lowe et al. 2017).

Figure 1

Consistency amongst installers of third-party monitoring equipment presented an issue in MacAirH, e.g. in the location of the flow meter. In some cases, due to space constraints or lack of awareness, flow meters were installed close to a bend or filter on the circuit, resulting in turbulent flow being present close to the flow meter which is likely to lead to inaccurate data (BEIS 2016). There was also one case of a heat meter being installed on the wrong circuit within a building and another case in which the installer failed to install current transducers on the three-way valve leading to difficulties in interpreting whether the heat pump was in space or water heating mode. Furthermore, there was one case in which different electricians had different opinions about whether there was a back-up heater (and as a result which SEPEMO system boundary the measurement took place at).

Onboard monitoring, on the other hand, should have more consistent sensor placement for a given brand of heat pump, since sensor installation is carried out in a factory. However, the locations of sensors vary between brands, may include different heat and electricity sources, and the uncertainty on each measurement is usually not stated by the manufacturer.

One consequence of this is that the reported performance (e.g. seasonal performance factor—SPF) cannot be compared between different brand of heat pumps, as the system boundaries for each brand are unknown. EDOL and MacAirH both attempted to investigate what is included in the heat and electricity measurements in onboard monitoring. This investigation was found to rely on the cooperation of manufacturers. EDOL had one occurrence of a manufacturer reporting they were unable to disclose the location of the sensors and algorithms to calculate energy consumption for commercial reasons, and other occurrences of no engagement from manufacturers upon asking about this. MacAirH’s experience of assistance from manufacturers was also variable, with some manufacturers providing sensor locations and calculation methods, some the former only, and some unresponsive to requests.

To illustrate this issue, Figure 2 shows daily heat pump electricity consumption from one EDOL participant during a period in summer in which the heat pump was only providing hot water. On some days the onboard and third-party monitoring aligned closely, but on other days the former gave higher readings than the latter; it was not possible to determine the cause (which manifested as extra consumption included in the onboard monitoring) due to lack of knowledge about how and where measurements are carried out by the onboard sensors.

Figure 2

Since onboard and third-party monitoring may differ in what electricity and heat they include, and since the uncertainty in onboard monitored data is not regulated, it is likely that onboard and third-party monitoring give different estimates of heat pump performance. Both EDOL and MacAirH were able to obtain a limited amount of data from a small number of sites with concurrent onboard and third-party monitoring. Only two EDOL participants had managed to retrieve onboard electricity and heat: one using extra hardware and communications, and another using a USB stick in the heat pump controller. A third had attempted using extra hardware, but the onboard monitored data had not logged. In MacAirH, four sites were available with concurrent data from the two approaches. Unfortunately only two sites were available with one full year of data; obtaining longer datasets is the focus of further work. Therefore, the term ‘seasonal performance factor’ will be avoided in the following analysis and ‘performance ratio’ will be used instead to emphasise that the dates and durations of monitoring differ between heat pumps. Equation (1) defines the performance ratio:

Table 1 summarises the discrepancy in reported heat, electricity and performance ratio between onboard monitoring and third-party monitoring for six sites. The third-party monitoring used the SEPEMO H2 boundary, whilst in the onboard monitored data it was not fully clear what the system boundary was, although H2 was likely to be the closest.

Table 1 shows that onboard reported performance ratio can either be higher or lower than the third-party measurement, and that the absolute values of the discrepancy vary significantly: within the six sites absolute discrepancies of 2–34% were observed. The mean absolute discrepancy was 17%; for context, a 17% discrepancy on a third-party reported performance ratio of 3 would yield an onboard monitored value of 2.5 or 3.5 depending on the sign of the discrepancy.

Table 1 also shows that the absolute discrepancy in reported heat tends to be larger than that in reported electricity. The temperature measurements made by the onboard sensors are likely to be a key reason for this. First, the location and calibration of the onboard sensors are unknown. This can lead to effects such as shown in Figure 3 where the onboard monitored flow temperature increases with respect to the third-party monitored equivalent only at certain times (in this case, when the flow temperature is raised). Second, there is no requirement for onboard temperature sensors used for heat calculations to be matched pairs.

Figure 3

Conversely, the EDOL project had one case where flow assumptions introduced a large error; in this case, the onboard estimated flow rate was incorrect by a factor of 3 in hot water mode. It was hypothesised in this case that flow rate was not measured with a flow meter but calculated from the circulation pump speed, and that the conversion factor was significantly wrong in this particular installation.

A further issue that could affect both third-party and onboard monitored estimations of heat output is unknown levels of antifreeze in the central heating system, as this affects heat capacity and thus recorded heat output. MacAirH found that central heating water changes could change the glycol ratio, and introduction of air to the heating system over the course of the monitoring period impacted onboard and third-party flow meters differently.

Regarding onboard electricity consumption reporting, it was found that only some heat pump brands included MID-approved electricity meters. Of those without, some had an electricity meter and some did not, instead estimating electricity consumption from a different variable such as circulation pump speed.

3.4 DATA COST

Data cost is an important consideration for both householders seeking to understand the performance of their own heat pump and for research projects seeking to collect data on a large scale.

For households, accessing real-time onboard monitored performance data was possible either free of charge on the heat pump controller or via an app if the heat pump was already connected to the internet. In some cases the latter was provided as standard with the app being either free or having a small annual fee; in other cases it required an upfront cost for an internet gateway.

For research use, due to the unknown uncertainties with onboard monitored data discussed in Section 3.3 as well as difficulties extracting them in time-series form, onboard monitored data are often not appropriate to use in standard format. Either third-party systems with characterised accuracies are required, or onboard systems are supplemented with extra hardware in order, for example, to provide an MID-approved electricity meter, along with extra hardware and software to extract and store if the data are not provided by the heat pump manufacturer.

The MacAirH project investigated the costs of two such upgraded onboard monitored solutions, and five third-party solutions, which provided good-quality data for research purposes.² It is important to note that these were all investigated once the heat pumps had already been installed, therefore the installation costs are expected to be significantly higher than would have been the case if the monitoring were installed at the same time as the heat pump, whereas the hardware costs would not be expected to change.

The costs of the above monitoring solutions installed in Irish homes are presented in Table 2. Capital costs (hardware and its installation) for retrofitted metering ranged from around €500 to €2700 per heat pump. Onboard monitoring was not exempt from hardware and installation costs. In one case, this was because the team added an MID-certified electricity meter into the onboard monitoring set-up, due to not being able to find information about the accuracy of the pre-installed electricity meter nor what was included in its measurements. In the other case, this was not possible, but there were still costs of wi-fi-enabling components and SD cards incurred for data transmission.

Annual fees were similar across all monitoring solutions shown in Table 2, and all included wi-fi (assumed to be €10/month). As many MacAirH participants did not have this, €120 can be subtracted from the annual fees row in Table 2 if wi-fi is already available. Some solutions also incurred annual software costs such as access to online dashboards which managed the incoming data.

It can be concluded that all options for fitting third-party monitoring or supplementing onboard monitoring, once the heat pump is installed/about to be installed, present significant costs, and that these costs are a barrier to heat pump monitoring at scale.

The time cost for monitoring for research purposes can also be high. MacAirH found that for both third-party and (upgraded) onboard monitoring, coordination across numerous stakeholders was necessary for on-site installation and troubleshooting. Stakeholders included facility managers, technical managers, tenants, electricians, plumbers, university staff, monitoring contractors, remote assistance from monitoring providers and heat pump suppliers. Furthermore, heat pump technology lacks a standardised installation pattern for distribution pipes within systems. In MacAirH the installation approach was found to vary based on the installer or contractor and the fact that each building can have unique characteristics that require a customised monitoring plan for each site. This customised monitoring approach often led to longer than expected time required for installers to complete the install and results in higher cost overall of the monitoring.

3.5 SUMMARY OF THE RESULTS

A summary of the key results regarding data availability, accessibility, quality and cost is given in Table 3. The information regarding onboard monitoring is given for six heat pump brands encountered in EDOL and MacAirH and common in the EU and UK markets. Third-party products are represented in one column as they were found to be fairly similar to each other.

Table 3 shows that all the heat pump brands represented here report either SPF or its underlying heat output and electricity consumption to the customer, over the controller and/or an app. However, the uncertainties about these values are not stated, in some cases there is no electricity meter, there is no requirement for the heat calculation to be accurate, and it is not known what uses of electricity or heat the measurements include. Extracting the onboard data for research was also difficult, requiring access to properties to install extra hardware. Third-party monitoring equipment was designed to a stated accuracy, although it was costly and relied on correct installation. Thus, it is concluded that both onboard and third-party monitoring present different, but significant, challenges.

There are several important implications of this for heat pump monitoring at scale. First, for cost reasons, only a minority of heat pumps have reliable monitoring, meaning most customers do not have an accurate indication of the performance of their unit. Second, large-scale studies that use third-party monitoring are very costly to carry out, whereas large-scale studies that rely on onboard monitoring are prone to error and have difficulty comparing brands due to different sensor locations and accuracies (which are often unknown) across heat pump brands.

4.1 SCALABLE

It is important first to define the scale of the proposed framework. The authors propose that every new heat pump sold in Europe would be manufactured equipped with the monitoring solution described below. This is to give individual households confidence that the data reported by their heat pump’s manufacturer are reliable.

An alternative implementation would be a sample approach, in which self-monitoring is fitted onto a representative sample of heat pumps produced by each manufacturer. This would be less costly for manufacturers in terms of extra hardware and communications, while still providing a large dataset for research use. However, customers would be in the same situation as at present: most would not have a reliable indication of the performance of their heat pump. Therefore, it is recommended here that all heat pumps contain the same monitoring and communications capability.

The inclusion of monitoring in every heat pump leads to a need to design a monitoring framework that is low cost and has standard methods and formats for data extraction, storage and access.

4.2 STANDARDISED

The term ‘standardised’ in the universal framework means that the variables measured, sensor locations and specifications, and minimum data frequency are the same for every heat pump, irrespective of brand and model. This will almost certainly require monitoring to be onboard as opposed to third party in order to ensure consistency of sensor placement through factory installation.

It is recommended that the monitoring scope aligns as closely as possible to an existing SEPEMO boundary. Appendix 1 in the supplemental data online shows the range of different system boundaries and components that require monitoring in each case. The authors’ recommendation is that the universal framework aligns with SPF H3; a proposal for implementing this is given in Appendix 2 online.

It is proposed that the monitoring includes two ‘levels’:

• A set of variables required to calculate SPF: electrical input to the heat pump (and back-up heaters controlled by the heat pump); and heat output of the heat pump. Note that electricity consumption is also an important metric in itself due to its link to running cost.

• A limited set of variables enabling understanding of these key metrics, e.g. flow and return temperatures from/to the heat pump, heat pump mode, indoor and outdoor temperatures.

Finally, the timestamp associated with all logged measurements should be in a standardised format that works across time zones and summer-time changes, and the resolution of the monitoring should be standardised to allow data to be easily imported to a centralised database (see Section 4.4).

Although not applicable at a European level, a precedent for this is the standardisation of electric vehicle charging data collection and communications capability brought in by the UK government in 2022 (UK Government 2021). This regulation is applicable to any charger, dictating a maximum electrical measurement uncertainty, and stating the time resolution of the measurements.

4.3 ACCURATE

The monitoring framework should state a maximum allowed uncertainty on SPF. This does not require specification of the maximum individual uncertainty in the flow sensor, flow and return temperatures, and electrical input measurements. However, it may be that a limit is specified for electrical consumption uncertainty, as onboard monitored electricity consumption is used for billing purposes for some European customers.³ EN 50470 (CEN 2018) and the MID are the European standards that relate to electricity metering; these could be used in the framework.

In general, allowable measurement uncertainties depend on the applications (JCGM 2023). Here, the application is to allow a heat pump customer to identify whether their heat pump is performing well, averagely or poorly. A proposal for the maximum uncertainty on SPF is given in Appendix 3 in the supplemental data online.

4.4 ACCESSIBLE YET SECURE

The data need to be accessible to a set of core end users, who are the heat pump customer, heat pump manufacturer, government and regulators. Additional end users could be researchers and external parties who wish to create models of heat pump operation to develop products.

Households should have the capability to share their heat pump data easily and to revoke consent when desired. The data could be stored within each country on shared national platforms, which include publicly available aggregated reporting of heat pump performance. A proposal is given in Appendix 4 in the supplemental data online; however, to ensure successful implementation of this architecture, detailed specifications should be determined collaboratively by a consortium comprising stakeholders from industry, regulatory bodies, research institutions and consumer groups.

4.5 COST-EFFECTIVE

In order for the monitoring framework to be included with every new heat pump, the cost to heat pump manufacturers must be as low as possible, yet households should not incur costs to access data on an ongoing basis.

It is argued below that the monitoring approach outlined in Section 4.3 will not be prohibitively expensive for heat pump manufacturers. Table 4 shows that the retail cost of buying one set of equipment that is able to reliably monitor SPF for 10 years is currently around €200.

The actual cost to manufacturers would be far less than that shown in Table 4 for two reasons. First, economies of scale arising from manufacturers purchasing tens of thousands of sets of equipment would significantly reduce unit costs, Second, many heat pump manufacturers already include several of the components shown in Table 4, or similar components with higher measurement uncertainty, and already incur server costs. Therefore, for most manufacturers, the increase in cost is expected to be far less than the upper bound without economies of scale given in Table 4.

Onboard monitoring would also minimise the time cost of monitoring highlighted in Section 3.4, as sensors would already be present and not require physical access to properties for installation. It would also prevent every government or academic research project investing time researching the most appropriate method of monitoring.

4.6 LIMITATIONS AND AREAS FOR FURTHER DEVELOPMENT

Given its requirement to be inexpensive enough to be scalable, the proposed universal framework will necessarily have limitations in its scope. As an example of the trade-off, the proposed monitoring set-up shown in Appendix 2 in the supplemental data online can highlight underperformance, but would not provide enough information to diagnose all underlying causes, which will likely require additional testing and checks. It also does not monitor the entire heating system, since components not linked to the heat pump controller are not monitored under the framework. Whilst an important aim of the framework is to measure performance within a real heating system as opposed to during lab testing, possible issues, such as high circulation pump energy consumption, will not be flagged. Once a basic monitoring framework is agreed, it can evolve over time as low-cost Internet of Things (IoT) sensing becomes more common and it is easier to incorporate more auxiliary components of the heating system into the monitoring.

The proposal also currently allows monitoring (which is a one-way information flow), but not control (two-way flow). This can be considered and expanded in future as more energy system services require interfacing with heat pumps.

Another important consideration is how to ensure sensors remain calibrated over time. It is not clear what level of calibration is currently carried out on existing sensors used for heat pump control purposes. Regarding monitoring, in the UK, non-domestic properties receiving Renewable Heat Incentive subsidies are required to have 10-yearly calibration checks of all heat meters (Ofgem 2025). This could be a suitable frequency of calibration check for the universal monitoring framework. Sensors could either be calibrated or swopped out after 10 years during annual servicing. However, more research is needed to determine the most cost-effective means of ensuring that sensors remain calibrated over time.

4.7 IMPLEMENTATION

The next steps towards implementing the framework require involvement from industry and the European Committee for Standardization (CEN). A detailed specification should be developed, and its placement within or in addition to existing EN standards should be agreed.

Once developed, governments can then adopt the standard into their national regulations in a way that ensures consistency between countries. As well as producing consistent datasets across European countries, this will simplify the required changes for heat pump manufacturers who sell their products in European markets.