Author: Rainer Pfluger, Sascha Hammes, Jan Steiger, Wolfgang Hasper

Aim and scope of the PHI’s “Verified Performance” programme is to verify the performance of a new build or deep retrofit in the field as a standard measure of final approval. If carried out by an independent party that both customer and supplier have previously agreed upon, the process may offer an unbiased analysis to both. To this end it must focus on the building's efficiency, this is the proportion of use and effort. Use in this context means the living quality in the conditioned space, expressed as thermal comfort and indoor air quality; effort relates to the energy spent.From the quality assured design process a third-party-certified energy model of the building is available, that can be used to predict the building’s performance under known boundary conditions. During the design stage, standardised values must be assumed for climate, occupancy, electricity consumption by users and other factors. In hindsight, however, most of these variables can be known from measurements and be used to update the energy model for the period in question. If a good agreement of the expected consumption, as predicted by the updated energy model, and the measured consumption, is found, the probability of major flaws in the building’s fabric, technical systems or their commissioning is low. Due to inevitable measuring uncertainty, unknown details of boundary conditions and building use, the match cannot be expected to be perfect. Uncertainty analysis can, however, point to the band of results which still indicate agreement. Careful selection of sensor specifications will ensure this band is sufficiently narrow to achieve a meaningful result.

The process also provides important learning opportunities for future projects and planning.

In addition to energy efficiency requirements, there are also target criteria for thermal comfort and indoor air quality. While energy efficiency can be easily assessed using one parameter, the quantitative assessment of health and comfort can be complex and requires the evaluation of several parameters. Indicators such as thermal comfort and indoor air quality must be considered separately as, despite their interdependence, they differ fundamentally both in their effects and in the measures taken to improve them.

Even if individual indicators can be standardized in their unit, many parameters must be evaluated individually in the respective application context and the existing framework conditions. Decisions on threshold values and comparisons of variants only appear appropriate if all indicators show the same trend. An objective comparison of variants is more difficult if individual indicators show opposing trends. At this point, it would then be necessary to combine the individual indicators into a single parameter, which in turn requires weighting. Weighting only proves to be scientifically justifiable if it has been comprehensively validated on the basis of subject studies, which can prove difficult due to the high level of user diversification. The Fanger model is a validated model for assessing thermal comfort (content of the ISO 7730 standard). Health indicators are standardized using the unit DALY (disability-adjusted life years) for various health-damaging effects. EN 16798-1 defines further requirements for indoor air quality, the thermal and visual indoor environment and acoustics. These are sometimes used as a basis for planning, especially for system dimensioning in buildings and for energy efficiency calculations.

However, the validity of individual validated and statistically based indicators can be limited when examining smaller buildings with a limited number of people. Especially if age, gender or behavior are unevenly distributed. This can falsify the assessments of the building envelope and building services. A case-by-case assessment would therefore be appropriate.

Individual behaviors have a significant impact on IAQ and energy consumption [Lopez 2021, Hong 2017]. Knowledge of user-centric targets in planning and operation can therefore help to improve IAQ and energy efficiency by considering the impact of user behavior. To move from existing concepts at an aggregate level to user-centered performance indicators, Han et al. identify requirements in Resolution of performance indicators by building type, Uniform scaling of performance indicators and Stronger quantification for performance quality assessments [Han 2020]. Living Quality Indicators are suitable for quantification and individual assessment, e.g., CO₂ concentration, temperature and relative humidity for the indoor climate.

In Passive House/EnerPHit standards, a high level of airtightness is required to keep heat losses to a minimum. Efficient mechanical ventilation systems are therefore required in this context to minimize the risks of overheating, high humidity and air pollution. Indoor environmental quality (IEQ) is becoming increasingly important in buildings with high energy standards. However, there is currently no general consensus on the measurement, limitations of the assessment classes and weighting of individual categories for the assessment of IEQ [Heinzerling 2013, Han 2020]. Due to the resulting variability in the assessment procedure, there are currently many degrees of freedom for the building assessment. This also applies to indoor air quality (IAQ) parameters. Although the CO₂ concentration, as well as the concentration of particles and volatile organic compounds, are good indicators of IAQ, there is no general standard for the quantitative assessment of air quality. However, the CO₂ concentration is often used as a quality indicator for IAQ and the calculation of air pollution [Persily 2017, Belmonte 2019]. Correlations can be demonstrated between the CO₂ concentration and certain gaseous compounds and bioaerosols [Lopez 2021]. EN 13779 sets specifications for the time in defined concentration ranges for IAQ assessments, e.g., high IAQ for CO₂ concentrations below 750 ppm or particularly low IAQ for concentrations above 1200 ppm.

Living Quality Indicators (LQI) are directly or indirectly influenced by the building envelope and building services. Despite a partial overlap, a distinction is made between health and comfort-related indicators. If people are exposed to inadequate comfort conditions in the building over the long term, this can have health consequences. This article is limited to temperature and humidity data for the statistical analysis of overheating frequencies and summer comfort.

Sensor selection proves to be fundamental for living quality monitoring and therefore for monitoring success and quality assurance. In accordance with ISO/DIS 7730, the analytical determination and interpretation of thermal comfort (PMV and PPD indices and local thermal comfort criteria) is based in particular on radiation temperature, air temperature, relative air velocity and humidity. If the air temperature and radiation temperature do not differ significantly, these can be formulated as the operative temperature using the arithmetic mean. For occupied rooms, the requirements for a high level of comfort (PPD < 6 %) are to be determined in particular by low fluctuations in the operative temperature (max: ±0.8 K), low draught risk (<0.08 m/s), low radiation temperature asymmetry (ceiling/floor < 5 K) and low vertical air temperature difference between head and foot when a person is seated (< 2 K). The influence of temperature is of the greatest importance here. The EnerPHit standard also ensures good airtightness with a maximum n50 ⩽ 1.0 h-1 and is systematically validated with pressurisation and depressuriation tests in accordance with ISO 9972. This minimizes the risk of draughts.

In order to fully map the IAQ in addition to thermal comfort, the measurement of the indicators relative humidity and carbon dioxide concentration is used alongside temperature. However, in order to obtain reliable evaluations based on this very limited set of parameters, the measurement uncertainty must not exceed narrow limits [outPHit D.6.5]. Mobile sensors and local data storage with evaluation in post-processing are suitable for practical use, especially for temporary operation.

A before-and-after comparison is required to monitor success. For this reason, monitoring should be carried out for a few weeks in various locations and apartments in the unrenovated state of the building. Longer periods are necessary in order to take into account the influences of user behavior and individual targets in particular, as the validity of statistical target values may be limited, especially in smaller buildings with a limited number of people.

In the outPHit research project, a wireless system based on the low-power wireless network protocol LoRaWAN (Long Range Wide Area Network) has prevailed. However, there are several suitable approaches that need to be evaluated against the respective framework conditions. The aim should be to make installation easy for users and as non-invasive as possible for residents. Data series of max. 10-minute intervals ensure both a long service life for wireless systems, without battery replacement and thus maintenance interventions, which can have a disruptive effect on comfort, and at the same time this ensures a sufficiently precise resolution for data evaluation. This is because temperature developments are generally associated with a certain degree of inertia.

Performance indicators are generally limited to clearly defined areas and categories via standards and guidelines. With regard to the air quality parameter CO₂, the effects are primarily limited to health and comfort during occupancy periods. In order to quantify how much and for how long certain environmental parameters in a building exceed defined limit values, the relative threshold deviation (RTD) is used. A distinction is first made between the heating period and the non-heating period. When planning based on the Passive House Planning Package (PHPP, based on the monthly method EN13790/ISO52016), a detailed energy balance model is created. The calculation results can be used to derive the typical heating and non-heating periods based on the climate data set. The measured parameters, e.g., CO₂ concentration, relative humidity and temperature in the room, are plotted as a cumulative distribution function for the evaluation period and only for hours with occupancy. The area between the limit value and this function shows the duration and extent to which the limit value was exceeded. This integral limit value deviation is set in relation to the maximum permissible deviation. It should be noted that a combination of parameters is not possible if different scales exist.

The RTD method is best used for seasonal assessment of LQI, as building use and air change rates vary greatly depending on the time of year. In winter, windows are usually closed, while in summer they are opened more often, which affects indoor air quality and energy efficiency. Airborne mold spores can pose a health risk depending on the type and quantity present, see [outPHit D6.10]. Buildings that are renovated to the EnerPHit standard rely on mechanical ventilation with heat recovery (MVHR) in winter, which ensures the air flow required for hygienic reasons at all times and regardless of the weather. Moisture-related risks, such as mold, are eliminated. Good insulation prevents low surface temperatures and moisture build-up.

The following Figure 1 show examples of the RTD as a basis for evaluating the building situation with regard to the temperature and humidity parameters. If there are excessive deviations from the comfortable corridor, this indicates a need for action. RTD as part of monitoring is an effective method of monitoring success.

In addition to measuring temperature, humidity and CO₂, microbiological sampling is useful for assessing indoor air quality (e.g., via impaction, filtration or isokinetic sampling for ventilation systems), as it provides a detailed understanding of the types and concentrations of microorganisms in indoor air.

These can affect the health of occupants, for example through respiratory infections, allergies, asthma or sick building syndrome, and can also damage the building fabric. Bioaerosols in indoor spaces are caused by occupant activities, contaminated building materials or outside air. The proportion of bacteria and fungi in the air differs significantly between indoor and outdoor areas, as indoor areas often offer better conditions for growth. The growth of fungi and bacteria is linked to specific temperatures and usually a relative humidity of at least 60 %. In the context of microbiological assessments, simplified methods are required for an initial, rough assessment [Seifert 2002]. These refer to the difference in the number of colony-forming units per sample volume (CFU/m³) in the outdoor air compared to the respective indoor air count.

First, those species that can typically reach high concentrations in the outdoor air are compared. Then the sum of all CFU/m³ belonging to genera that indicate increased indoor air humidity is evaluated for increased occurrence. In the third step, individual species within the aforementioned set of genera are examined. Each of the three steps is evaluated. To provide a good overview at a glance, radial diagrams (radar plots) with color coding according to the traffic light scheme are provided for each assessment category. The overall situation can only be categorized as free of findings if all “green” markings are awarded. This is shown as an example in the following illustration.

Indoor air samples and surface temperature measurements can be used to retrospectively check whether a refurbishment measure was successful and whether the building meets the desired performance targets.

Microbiological samples allow the identification and quantification of specific bacterial and fungal species and their concentrations, identifying potential health risks that conventional sensors for temperature, humidity, CO2 and VOC levels do not detect. These samples are an important part of indoor air quality assessment and provide valuable information on the health and safety of indoor environments. They also support the development of effective operational and remediation strategies. Mold, which is often caused by high humidity due to inadequate ventilation, is a significant health risk. The outPHit research project was able to highlight the importance of mechanical ventilation and high energy efficiency standards to prevent mold growth and meet comfort and health requirements (see D6.12 report on microbiological assessment of indoor air quality in case study projects [outPHit D6.12]).

For the post occupancy evaluation to work well, it is recommended for the project to meet the following pre-requisites:

• A careful building energy efficiency design with third-party quality assurance makes a detailed and reliable energy model (e.g. from [PHPP10]) available. As the analysis relies on this model in order to normalise results its validity is a key point.
• Careful site supervision ensures compliance with design.
• Qualified experts and tradespersons are tasked with the remaining work.
• The airtightness has been measured for pressurisation and depressurisation (EN ISO 9972, NA Germany) and the residual leakage is very low (n50 ≤ 1.0 h-1 for retrofits or ≤ 0.6 h-1 for new builds).
• Systematic commissioning, including flow balance adjustment, of the ventilation system guarantees air change rates according to the design of the ventilation system.
• The energy balance calculation of the project (e.g. with[PHPP10]) has been updated to the finally executed project specifications.
• With the completion of the renovation project, a third-party verification (e.g. a building certification according to the Passive House or EnerPHit standard) confirms the fulfilment of the criteria of the efficiency standard.

The Passive House Institute has developed a monitoring platform that can be used via any web browser. A user login ensures that access to project information is granted to registered users and only to the specific projects they are assigned to. The platform can collect measured data for a project, visualise and statistically analyse it. At the quantitative analysis stage the data can be automatically aggregated into monthly mean values and exported into the PHPP energy balance calculation of the project. Results are finally available as a pdf report. For privacy reasons the report can omit all individual data but only present building-wide averages.

For every certified building most relevant data is already available in the PHI database. A monitoring configuration can then be easily added. For other buildings a building object can be created with data from a PHPP v10 or higher. To set up the project the entire TFA of the building is first subdivided into zones, that relate to flats or units. In themselves these zones comprise rooms into which the individual sensors can be assigned. As both zones and rooms are characterised by an area, this data lends itself to automatically calculate area-weighted mean values in order to derive building-wide conditions. Sensor properties such as measuring uncertainty are configured and represented in the data plots, but also used in the quantitative assessment.

Data can be automatically imported from the PHI's LoRa data acquisition system. Alternatively, csv data can be imported manually. The interface can be configured to suit a wide range of time stamp and data formats.

Transmitting analogue signals for long distances increases the noise and introduces additional causes for faults and errors. Digitizing the values as closely as reasonably possible to the location of the measurement helps minimise these undesired effects. Technology to transmit the digital data is not vital, almost any digital data acquisition system can work as the data rate is low, compared with other IT systems. GreenPHY power line communication appeals as it uses existing infrastructure (the power lines) and also provides power supply to the sensors at the same time, but unfortunately was found to be unsatisfactory in old multi-branch electrical systems. The alternative recommendation is to use LPWAN radio technology, specifically LoRaWAN (referred to as LoRa in this document for brevity) which operates within the EU at 868 MHz and hence in a frequency band that is not too easily attenuated within a building. The technology is easy to deploy temporally, runs on battery power for about 2-5 years and requires no fixed infrastructure or otherwise lost investment. Moreover, it can be delivered on site pre-configured and be easily deployed by non-expert personnel. Equipment is available at low cost, which is crucial for economic viability of the programme. One central receiver is installed per building in order to log the messages from all sensor nodes and to relay them to a central data base. Sensors with high gain (2 dBi) external antennas are preferred for robust radio transmission. A wide range of sensors, meters, pulse counters and weather stations are available in the LoRa ecosystem, with a small number of devices meeting demanding specifications. In cases where digital/wireless data acquisition systems already exist for other reasons it can be a consideration to use/upgrade them the monitoring purposes discussed here.

The energy balance based analysis requires the temperature and relative humidity as well as the pressure of the ambient air to be known. The tolerable uncertainty of ± 1 hPa can be met by most sensors of reasonable qualtiy.
Further, short wave solar radiation (global radiation) on the horizontal plane (GHI) must be available. A total uncertainty on the order of ± 5 % of the daily sum is an adequate compromise of cost and useful data. For the later building performance analysis global radiation data is required not only for the horizontal but also for the vertical plane in four cardinal directions. This is achieved by measuring global horizontal radiation with a good grade irradiation sensor (Pyranometer) and treatment of the data within a mathematical sky model [Perez/Ineichen]. A different approach would be to directly measure all five desired components using calibrated and temperature-compensated PV-cells. Either method is subject to its particular set of uncertainties but both seem useful with regard to the required overall uncertainty. The PHI's monitoring platform integrates a sky model and thus works based on just GHI.

Outdoor air temperature and relative humidity are indispensable, tolerable uncertainty is ± 0.2 K and 2% rh. All meters for energy used within the thermal envelope of the building should be sampled. Depending on local conditions some sub-metering can be useful, e.g. for heat-pump systems, solar thermal hot water systems etc. To standardise the data acquisition, it can be useful to either use meters with pulse output as are available for electricity, heat, gas or water. If meter data beyond the energy count is desired, the Meter bus (M-Bus as per EN 13757) can be used to interface the more comprehensive meter data sets; Gateways for LoRa integration exist, or a stand-alone M-bus network could be operated.
Fortunately, an increasing number of meters can generically provide data wirelessly over LoRa. For any heat pump systems a sub-meter for electricity (input) and a heat meter (output) are strongly recommended in order to determine the heat pump performance. If cooling via water wells or ground probes is used, a heat meter is also very useful here. For utility meters the measuring uncertainty is normally below ± 2 % for electricity and gas, while heat meters read to ± 5 %. Care should be taken to not exceed this value.

Weather and indoor air parameters shall be sampled at regular intervals, both mean and instantaneous values are possible as long as the interval is not too long. The interval shall be the same for weather and indoor data sampling and shall not exceed 20 min. Otherwise the dynamics of the situation is lost. Less than 10 min. will normally be useless and, on the contrary, make the dataset unnecessarily large and pose high demands on the scaling functionality of a data base. Meters shall be sampled at least monthly, on the beginning of the 1^st day of the month. If automated equipment is used an interval of 3 hours is reasonable but not mandatory. Since the number of meters is usually low, it is also possible to apply the same standard interval as for room and weather sensors.

For small buildings up to 30 dwelling units each dwelling unit shall be measured with at least one sensor station located in a central spot that is representative of the whole. For larger buildings it is useful to follow the same rule. However, a statistically significant result might be achievable with less than 100 % coverage, if at the cost of reduced robustness. Measured dwelling units must nevertheless be evenly distributed across the entire volume of the building in order to represent the average conditions and the impact of variations in solar exposition, user behaviour, fabric heat loss, and other variations in the best possible way. For buildings with more than 30 dwelling units the required minimal number of measured dwelling units may be estimated for a total number of dwelling units t with the following formula. n = 30 + 1/3 * (t-30) Any decimal fractions are always rounded up to full integer numbers. In addition to this core equipment one sensor should be placed in each stairwell located within the thermal envelope, at half the height of the thermal envelope. In unheated basements one auxiliary sensor shall be placed in a representative spot. For basements within the thermal envelope one sensor shall be placed for each 100 m² of basement floor area.

All areas within the thermal envelope of non-residential buildings are divided into zones that are distinct by storey, orientation, solar exposition, usage, occupancy, internal heat gains and regular room temperature. For each zone, as a rule, 3 sensors shall be placed evenly spaced in order to sample the representative mean conditions. The number of sensors may be reduced for small areas. There shall never be less than 1 sensor per 100 m². In addition to this core equipment one sensor shall be placed in each stairwell located within the thermal envelope, at half the height of the thermal envelope. In unheated basements or parking spaces one auxiliary sensor shall be placed in a representative spot. For basements within the thermal envelope one sensor shall be placed for each 100 m² of basement floor area.

Placement on the roof on a mast clear of shading objects/above ridge height and as far as practically possible from the (hot) surface of the roof, of airsource heat pumps, AC units or other heat exchangers. The Pyranometer/irradiation sensor must be carefully levelled and mounted free of any obstructions in its field of view (hemisphere with unobstructed horizon). If a mast in the instrument’s field of view cannot be avoided it must be oriented towards the near pole (north in the northern hemisphere), in order to minimise the impact on the instrument’s readings. A full horizon view shall be photographically documented from nearby the sensor’s location. It must be noted that the irradiation sensor, regardless its type, requires regular maintenance in the form of cleaning from dust and debris. Typically a monthly schedule is adequate, but this may depend on local conditions. Safe access is, therefore, required. For practical reasons the temperature/rel. humidity sensors within their radiation shield (e.g. with fan aspirator) shall be mounted nearby the Pyranometer. However, a location in a shaded spot on the building’s side toward the near pole (north in the northern hemisphere) is ideal. The air pressure and CO₂ sensors may be integrated in the data sampling/radio transmitter electronics housing, provided, that the enclosure is adequately ventilated to the outside.

Placement of room sensors shall be chosen clear of direct sun light at any time of the year, if possible at half room height / 1.10 m from the floor, on an internal wall. A proven location is by the light switches next to the door. In a room oriented towards the far pole (south in the northern hemisphere) this height may, however, be too low to exclude solar exposition in the winter and a greater height is advised in such cases (e.g. 1.8 m. The altitude of the sun at summer solstice can be estimated as 90°- Latitude +23.5° at winter solstice as 90°- Latitude -23.5°, neglecting atmospheric refraction). In highly insulated buildings temperature stratification at times without solar energy input is known to be very low (< 0.5 K over the entire height of a normal room), hence height is not overly critical. A careful documentation of the location, with photographic evidence including a yardstick to indicate the height above the finished floor level, is required for each sensor.

In most cases meters are provided and owned by the utility for billing purposes. The building design shall as far as possible anticipate the subsequent monitoring for performance verification and place the meters at the balancing boundary whenever possible. If any meters also measure meaningful amounts of energy that is used outside the thermal envelope, sub meters must be fitted to establish the respective amounts. This holds in particular for electricity meters that also measure energy for e.g. vehicle charging or powerful outdoor / garage lighting. Sub-metering can be dispensed with where the energy use outside the thermal envelope is less than 1 % of the total. A heat pump shall always be fitted with a sub-meter for real power (high inductive loads from the compressor motor) for the electricity input to the heat pump system, comprising the compressor, controls, auxiliary pumps and fans as well as a heat meter for the output. If individual electricity consumption is of no concern any privacy issues can be overcome by installing a central electricity meter in the root of the system, e.g. with current transformers and voltage measurment connection to all three live conductors. In this configuration a summary value for the entire building is obtained. While this meter is a costly bit of kit it saves installing many individual meter pickups for each flat.

During the data acquisition phase, that normally lasts for two years, data is nevertheless useful for continual evaluation of the indoor environment and metered energy. The monitoring platform provides tools to visualise the data and automatically performs a first statistical analysis. This includes the identification of minimum and maximum values, quartiles, median, arithmetic mean and standard deviation. Moreover results are binned to illustrate the frequency distribution. A very fast and intuitive check is possible based on this data and assists evidence-based commissioning. Ongoing commissioning work means that the data from this first period is, however, disturbed by various issues in the plant. In the second year of operation such flaws have been identified and ironed out. Undisturbed data is thus available for the final, quantitative analysis. The following examples are taken from the database and illustrate the procedure and potential, based on an exemplary data set from one bedroom in a demonstration project.

Lines plot of room temperature for a year.

Boxplot of elementary statistical characteristics for the same data set.

Histogram of binned values from the same data set.

All available data is converted into monthly representations and serves to automatically update the energy balance calculations (by [PHPP] methods in the MONI worksheet) with measured boundary conditions for weather and usage. The boundary conditions data set is handed to the calculation engine and results are returned in the form of another data set, for display, documentation and further use in the database. As a result the predicted energy use for the prevailing boundary conditions from the period under consideration can be compared to the metered values. For reference also the design value, based on standard climate data and usage, is shown.

Short wave solar radiation is measured as global horizontal radiation and can be handled with a sky model by [Perez/Ineichen] with regard to air pressure as a correction for air mass as well as considering air temperature and air relative humidity. The procedure follows a proven implementation in [pv_lib] and results are aggregated into monthly integrals of global irradiation on the vertical plane in the cardinal directions in addition to the measured value on the horizontal plane. Down welling long wave radiation is estimated from the dew point of the air and considered in the form of the monthly mean sky temperature. Air temperature and relative humidity are considered as monthly mean values respectively. Ideally, the outdoor CO₂ concentration is used as a summand in fine temporal grain for IAQ assessment in order to determine the concentration increase in interior spaces with regard to the outdoor concentration. As the latter is subject to considerable temporal variations, particularly in urban areas, it is valuable to also have the outdoor concentration measured. However, this makes sense only in cases where very high quality sensors are used, with a practical measuring uncertainty well below 50 ppm. Inexpensive sensors do not provide this level of accuracy, but auto-calibrate to the lowest observed values over a week, thus indirectly providing a reference to the outdoor value. Nonetheless, any short-term (diurnal) fluctuations are lost. Given the cost constraints that apply to the suggested monitoring approach this is considered tolerable, however.

A building object in the data base is subdivided into zones (e.g. flats) and rooms within these zones. Each room is characterised by its area and can be assigned a number of sensors. Mean indoor conditions for the energy assessment of the building are derived as area-weighted means (TFA) of the room/zone values, both for temperature and relative humidity.

An estimate of the internal heat gains prevailing in the evaluation period is derived from the metered electricity consumption and estimated occupancy, considering PV generation and the amount of energy delivered to the grid (exported from the balancing boundary). Deducted is any noteworthy consumption outside thermal envelope, particularly electric vehicle charging, underground parking lights/ventilation, external lighting in general as well as estimates for evaporation and drain losses.

No measurement is ever totally accurate. In building monitoring a whole range of measured values each have a measuring uncertainty and all individual uncertainties combine with regard to the final result. An evaluation of uncertainties involved in energy balancing for buildings has been documented in [Johnston e.a. 2020]. Based on that study an abbreviated and adapted approach is given in the following section.

Any calculation can only be as accurate as the respective input data. It is in the nature of buildings and construction that deviations from the design data (e.g. effective insulation thickness, thermal conductivity) can be regularly observed. Such deviations result, by uncertainty propagation, in a limited accuracy of the calculated heating energy demand of a building. The magnitude of the resulting uncertainty shall be estimated in the following discussion.

• The effective insulation thickness in a building component can vary by up to 10 mm due to construction tolerances, partially compressed insulation, deviations in the spacing of fixtures/dowelling; the nominal values for the thermal conductivity of officially approved insulation materials are, as a rule, considerably higher than the actual value and only in rare individual cases a deviation in the opposite direction is possible (e.g. caused by soaking). A typical uncertainty on the parameters of the insulation can thus be estimated by a variation in effective insulation thickness of 15 mm, which will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 0.7 kWh/(m²a).
• Deviations in operation of MVHR systems. The effective heat recovery rate in the field can be affected by air leakage, disbalance, condensation and other effects. A typical uncertainty on the parameters of the MVHR can be estimated by a variation of the effective heat recovery rate of 5 %, which will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 1.2 kWh/(m²a).
• Measuring uncertainty of the airtightness test. The leakage flow as well as the reference volume (or area) affect the accuracy of the n₅₀-value. Under practical constraints of the test (e.g. influence of wind) and building preparation both can only be determined with limited accuracy. A typical uncertainty on the parameter airtightness can be estimated by a variation of the n₅₀ value of 0.05 h^-1, which will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 0.24 kWh/(m²a).
• Deviations in the thermal properties of windows. U-Values of windows can be determined with limited accuracy even when using the DIN EN 10077 methods. The effective glazing U-value is affected by variations in the noble gas fill concentration; dimensions of all the window components are subject to tolerances, particularly the distance from the glazing edge of spacer and edge seal. A typical uncertainty on the window parameters can be estimated by a variation of the glazing U-value of 0.05 W/(m²K) and 0.02 W/(m²K) for the frame, and 0.005 W/(mK) for the linear thermal bridge effects, which will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 0.6 kWh/(m²a).

The above items define the construction parameters of a Passive House building with regard to the annual space heating demand to 90 to 95 %. The impact of further effects is not very large (e.g. the absorption coefficient of external surfaces). The major effects discussed above are independent, hence the total effect can be lumped together with a quadratic uncertainty propagation, where x is the parameter under consideration and Δx the respective standard deviation: Using the respective values derived in the above discussion yields a propagated uncertainty of ± 1.5 kWh/(m²a) for about 90 % of the building-related parameters. For any calculation of the annual space heating demand, therefore, a total uncertainty of about ± 1.6 kWh/(m²a) must be considered, due to inherent uncertainties in a building’s construction parameters. This refers to any calculation, regardless how sophisticated and physically accurate the model might be.

The impact of user behaviour and usage in general on the measured space heating demand is very important. Hence, the most relevant parameters reflecting the usage need to be measured if a comparison of measured vs. projected consumption is desired. In order to estimate the uncertainty margin for usage parameters three quantities matter in particular:

• The mean room temperature. The propagated uncertainty for an uncertainty in the mean room temperature of ± 0.3 K will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 4 % or 0.6 kWh/(m²a).

Measuring this quantity with high grade sensors of lower measuring uncertainty than ± 0.3 K may reduce this contribution. With the ± 0.2 K sensor specification it is suggested to adjust the total uncertainty from this factor to 0.5 kWh/(m²a) for the outPHit verified performance scheme.

• The effective air change rate. It has been observed in intensely monitored buildings that in most cases windows are opened in the main heating season only occasionally and for short times. Typical effective additional air change rates (daily average) from windows and doors have been found ranging from 0.003 h^-1 and 0.09 h^-1 (with occasional deviations). In most cases this suggests to neglect the effect. The uncertainty in the usage-related air change rate can be estimated as ± 0.012 h^-1 which will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 1 kWh/(m²a).

Measuring the added air change rate by recording and evaluating all window opening events involves a very high number of window contacts -which are notorious to fail- and the related monitoring equipment. The evaluation is also not easy and almost fails entirely as soon as a single window’s status is not accurately logged. It is hence considered impractical to regularly measure the effective air change rate in the scope of a simplified monitoring approach. Here lies the most impactful uncertainty on the overall result and grossly unusual user behaviour may cause significant differences in measured vs. projected energy consumption. Hence, this contribution cannot be reduced and even larger amounts must be kept in mind as a possibility when evaluating monitoring results.

• The internal heat gains. Monitoring results from a large number of German Passive House buildings point to effective internal heat gains of 2 W/m² (± 0.3). The figure represents the balance of positive and negative contributions. The remaining uncertainty of ±0.3 W/m² will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 1.3 kWh/(m²a).

Primarily, the total amount of IHG can be verified for the individual case using the actual number of users and the electrical energy use from measured values. The uncertainty in internal heat gains may be also somewhat reduced, however, this effect is limited due to the unknown utilisation factor for electrical energy (e.g. drain losses from dishwashers, washing machines etc. can only be estimated by empirical factors). Hence it is suggested to remain with the ± 0.3 W/m² or 1.3 kWh/(m²a) for the outPHit Verified Performance scheme.

The three main influence factors discussed above account for about 70 % of all usage-related uncertainties. If it is accepted to consider them largely independent the total uncertainty can be lumped together with a quadratic uncertainty propagation again: Using the respective values derived in the above discussion yields a propagated uncertainty of ± 1.7 kWh/(m²a) for about 70 % of the building-related parameters. For the calculation of the annual space heating demand, therefore, a total uncertainty of about ± 2.5 kWh/(m²a) must be considered, due to uncertainties in a building’s usage parameters. The figure can be reduced to about ± 2.4 kWh/(m²a) if the above improvements for the outPHit Verified Performance scheme discussed above are factored in.

With a radiation shield outdoor air temperature can be measured with a high grade sensor to an uncertainty of ± 0.15 K and good instruments permit measuring global horizontal solar radiation to ± 5 %. This will yield an impact on the space heating demand of a typical Passive House building in the magnitude of 4.5 % or 0.7 kWh/(m²a).

The three categories of uncertainty discussed above, namely construction parameters, usage parameters and weather data uncertainty can regarded as independent. Therefore, the total uncertainty can be lumped together with a quadratic uncertainty propagation and yields a result of ± 3 kWh/(m²a). This value can be confirmed for the outPHit Verified Performance scheme as only slight improvements over the already ambitious assumptions of the original investigation are practical. Even with an accurate energy balance model, careful inputs and correct usage of the model the available accuracy of input data does not permit a lower uncertainty in the results, regardless how sophisticated the method is (e.g. dynamic building simulation with high spatial and temporal resolution). The remaining uncertainty amounts to a low absolute value that does not materially change with regard to a building’s energy efficiency standard. However, with decreasing energy demand the proportion of the uncertainty in the energy consumption figures naturally increases. This, on the other hand, is still of limited significance, given the very low figures e.g. in a Passive House building. Particularly for the deep retrofit context in the outPHit project measured and calculated values pre- and post-refurbishment will still be amply accurate to establish the successful implementation of the chosen measures and the resulting energy savings. For different energy efficiency standards the absolute uncertainty of ±3 kWh/(m²a) implies the relative uncertainty according to the table. The outPHit demonstration projects are expected to meet the [EnerPHit] heating demand criterion of 25 kWh/(m²a) with a relative uncertainty of 12 %.

Despite good grade sensors, determination of inputs is subject to uncertainties. For highly energy efficient buildings a comparatively high relative uncertainty results from the largely absolute nature of the uncertainties involved: Matching the calculated results will never be possible with less than about ±1.6 kWh/(m²a) that are inherent to construction properties uncertainties. Additional uncertainty results from usage parameters, the amount depends partially on the grade of sensors. User behaviour in terms of additional air change via windows is, and remains, the most significant unknown. To measure it a complete suite of sensors for all openings were required that also distinguish tilt vs turn as well as the turn angle, and each sensor would present a single point of failure for the entire effort. A serious approach would multiply the cost for sensors as well as the complexity of the evaluation. Hence it is considered impractical, but also dispensable, in the vast majority of cases within the scope of the outPHit verified performance programme. The PHPP monitoring data evaluation yields a plausible range for results, taking measuring uncertainties into consideration (as uncertainty limits). Located within the band lies the range of ±3 kWh/(m²a) derived above. If household electricity consumption exceeds a certain limit (e.g. 15 kWh/(m²a)), feedback to the building users shall be triggered and guidance on efficient use of electricity be given. While monthly balance calculations are valuable and can support informed performance optimisation, they may contain some bias due to seasonal capacity effects (e.g. sensible and latent heat absorbed into or released from the building structure), slightly distorting the monthly values in the spring and autumn. Therefore, it might be considered to base any certification only on the annual values. This way, the vast majority of capacitive effects can be assumed to cancel each other. First-year effects, such as drying of concrete, screed and other building materials can have a considerable impact. Systematic evaluation should, therefore, use the first year for error detection and building services optimisation, while only in the second year a robust evaluation of energy performance can be made. For refurbishment projects the old condition of the building also presents a useful reference. If billing information or meter readings are available for all relevant energy carriers they are the best possible source. In other cases, a variant of the energy balance calculation with PHPP may be performed for the unrefurbished condition, taking a reduction factor for partial heating/reduced average indoor temperatures into account. With regard to this reference a global reduction in space heating demand of 75 % should be achieved.

[Belmonte 2019] Belmonte, J.F.; Barbosa, R.; Almeida, M. G. (2019) CO2 concentrations in a multifamily building in Porto, Portugal: occupants’ exposure and differential performance of mechanical ventilation control strategies. Journal of Building Engineering 23, pp. 114-126.

[EN16798-1] Energy performance of buildings - Part 1: Indoor environmental input parameters for the design and assessment of energy performance of buildings in terms of indoor air quality, temperature, light and acoustics - Modules M1-6; Committee 141 - Air-conditioning engineering, 2019.

[Han 2021] Han, L.; Wang, Z.; Hong, T. (2021) Occupant-Centric key performance indicators to inform building design and operations. Journal of Building Performance Simulation 14(6), pp. 814-842.

[Heinzerling 2013] Heinzerling, D.; Schiavon, S.; Webster, T.; Arens, E. (2013) Indoor environmental quality assessment models: A literature review and a proposed weighting and classification scheme. Building and Environment 70, pp. 210-222.

[Hong 2017] Hong, T.; Yan, D.; D’Oca, S.; Chen, C.-F. (2017) Ten Questions Concerning Occupant Behavior in Buildings: The Big Picture. Building and Environment 114, pp. 518-530.

[Johnston, D., e.a. 2020] Are the energy savings of the passive house standard reliable? A review of the as-built thermal and space heating performance of passive house dwellings from 1990 to 2018, Energy Efficiency (2020) 13:1605–1631, https://doi.org/10.1007/s12053-020-09855-7

[Lopez 2021] Lopez, M.; Guyot, G.; Golly, B.; Ondarts, M.; Wurtz, F.; Gonze, E. (2021) Relevance of CO2-based IAQ indicators: Feedback from long-term monitoring of three nearly zero-energy houses. Journal of Building Engineering 44, pp. 103350.

[outPHit D6.5] Steiger, J., Grove-Smith, J., Krick, B., Müller, L., Hasper, W.: outPHit D6.5 Description of a certification scheme on “verified building performance”, 2022.

[outPHit D6.10] Maier, M., Kirchmaier, M., Pfluger, R.: outPHit D6.10 Instructions for adequate monitoring equipment for living quality assessment, 2021.

[outPHit D6.11] Pfluger, R., Hammes, S., Steiger, J., Hasper, W.: outPHit D6.11 Report on living quality indicators before and after retrofit, 2023.

[outPHit D.612] Pfluger, R. Hammes, S.: outPHit D6.12 Report on microbiological assessment of indoor air quality in case study projects, 2024.

[Perez/Ineichen]Perez, R., Ineichen, P., Seals, R., Michalsky, J. and Stewart, R. (1990) Modelling Daylight Availability and Irradiance Components from Direct and Global Irradiance. Solar Energy, 44, 271-289

[Persily 2017] Persily, A. (2017) Indoor Carbon Dioxide as Metric of Ventilation and IAQ: Yes or No or Maybe? Is ventilation the Answer to Indoor Air Quality Control in Buildings? Do we Need Performance-Based Approaches? AIVC Workshop 2017, Brussels, Belgium.

[pv_lib] Holmgren, W., C. Hansen and M. Mikofski (2018). “pvlib Python: A python package for modelling solar energy systems.” Journal of Open Source Software 3(29): 884.

[Seifert 2002] Seifert, B. e.a.: Leitfaden zur Vorbeugung, Untersuchung, Bewertung und Sanierung von Schimmelpilzwachstum in Innenräumen, Umweltbundesamt, Berlin 2002.

[Wingfield 2009] Wingfield, J.; Bell, M.; Miles-Shenton, D.; South, T.; Lowe, R. J. (2009) Evaluating the Impact of an Enhanced Energy Performance Standard on Load-Bearing Masonry Construction – Final Report: Lessons from Stamford Brook – Understanding the Gap between Designed and Real Performance. Leeds Metropolitan University, Leeds, UK.

Refurbishment with Passive House components

outPHit - Deep retrofits made faster, cheaper and more reliable