Energy Efficiency

, Volume 3, Issue 3, pp 189–201

Strategic homogenisation of energy efficiency measures: an approach to improve the efficiency and reduce the costs of the quantification of energy savings


    • Energy Institute at the Johannes Kepler University Linz
  • Andrea Kollmann
    • Energy Institute at the Johannes Kepler University Linz

DOI: 10.1007/s12053-009-9060-z

Cite this article as:
Reichl, J. & Kollmann, A. Energy Efficiency (2010) 3: 189. doi:10.1007/s12053-009-9060-z


With the ongoing efforts on the European level to promote energy efficiency, the need for the development of harmonised evaluation criteria for energy efficiency measures arises. Such criteria will allow extensive comparisons of the success or failure of the implementation of energy efficiency measures throughout Europe and will support the development of a first–best strategy for the realisation of energy savings targets in Europe. Two fundamental evaluation possibilities exist: bottom-up and top-down quantifications of energy savings. Bottom-up calculations give a more detailed view of the impact of energy efficiency measures but are much more costly and time consuming than top-down calculations. In our opinion, this effort can be reduced without losing precision in the savings calculations by the homogenisation of these energy efficiency measures. In this paper, we develop a framework specifying how such a homogenisation could look.


Energy efficiency measureStrategic measure homogenisationQuantification of energy savings


The targets of large-scale energy efficiency programmes, such as those of the Directive 2006/32/EC (European Commission 2006) of the European Parliament and of the Council on energy end-use efficiency and energy services (henceforth ESD), are not achieved by the implementation of a small number of energy efficiency measures. In the context of this paper, the terms energy efficiency programme, energy efficiency measure and energy efficiency action are defined as follows: An energy efficiency programme initiates a number of energy efficiency measures and provides the (legal) framework for their implementation. An energy efficiency measure is a single project that motivates a target group to realise certain energy saving actions, such as the installation of energy-efficient heating devices in households.

Large-scale energy efficiency programmes, such as the ESD, can only succeed if energy efficiency measures covering all sectors of energy consumption are carried out repeatedly to reach a large percentage of all energy consumers. In this regard, the costs for monitoring and quantifying these energy efficiency measures are high. In our opinion, this effort can be reduced for measures addressing the energy consumption of households by the homogenisation of these energy efficiency measures. This is carried out by incorporating the similarities of specific energy efficiency measures already in the conceptual phase of the measures, such as of energy audits focusing on household electricity consumption.


Whenever bottom-up calculations of the achieved energy savings of a certain programme are part of the fulfilment of the programme itself, these energy-saving calculations must meet the requirements concerning quality and comprehensiveness demanded in this programme. In this respect, the precision and the database of energy savings calculations cannot be chosen by the party carrying out the energy efficiency measure alone, but (usually) need to meet some standard defined as part of the programme. As legally binding energy efficiency targets (or indicative targets in the case of the ESD) are usually not formulated for one party alone, the requirements on methodology, precision and data used for the calculations require serious efforts of the parties participating in the programme, in order to achieve comparable results. In case of the ESD, the 27 member states of the European Union are required to reduce their end-use energy consumption by 9% in the period 2008 to 2016. The requirements for the member states for reporting the achieved energy savings therefore need to guarantee comparable results of their energy efficiency improvement efforts. These requirements, especially concerning the data that need to be provided, may induce high costs for the evaluation process. Especially for small-dimensioned and regional energy efficiency measures, these costs may exceed reasonable levels and may thus put the implementation of the measures in question.

The basis of the determination of energy savings is the difference of the energy consumption in the business-as-usual (BAU) scenario and the energy consumption after the energy efficiency action has become effective, which we refer to as the reporting period (RP). The parameters of the RP energy consumption are monitored or can be estimated with respect to the configuration of the energy efficiency measure (this could, e.g., be the minimum requirements to qualify for housing subsidies). In contrast, the determination of the energy consumption in the BAU scenario is more problematic as BAU never occurs and is therefore unobservable. In many cases, a comprehensive definition of the BAU scenario involves costly surveys of the determinants of the energy consumption of the energy efficiency measure’s subject. Examples of such determinants are the average real-life energy consumption of refrigerators over their lifetime1 (not the power input according to manufacturers’ data), the average specific heat demand per square metre of single-family dwellings built between 1970 and 1980 or the development of the level of equipment ownership of mobile air conditioners.

While the expected outcome of energy efficiency measures on the industry, community or large buildings sector justifies the effort for metering and/or extensive data collection, the effort for metering and data collection for an individual household may exceed a reasonable level compared to the expected savings. Therefore, a mechanism has to be developed that makes saving calculations for this sector less costly but still guarantees meeting the quality standards asked for in the corresponding programme. Such a solution appears important to the authors, as otherwise, the energy savings from such measures (as the effort does not permit calculation of the savings according to the data requirements) are not creditable to the energy savings target of the corresponding programme and will therefore probably not be carried out. One solution to this problem is the use of deemed savings. This approach follows an attractive concept regarding the reduction of the evaluation effort, namely to apply fixed values to the energy savings of energy efficiency actions (mainly on the household level). This means that a predefined value for the energy savings of a certain energy efficiency action (e.g. installation of an A+ + refrigerator) is credited without the need for a collection of specific data on the energy efficiency action and on the variables influencing the actual savings (e.g. the energy efficiency class of the replaced refrigerator). The concept of deemed savings is well established in the USA for analyses of the savings potential of certain measures (see, e.g. New York State Department of Public Service 2004). In Europe, an example of this approach directly related to the ESD are the methods currently developed by the Austrian Energy Agency2 as a temporary solution for the evaluation of the achieved energy savings until the European Commission approves its “harmonised bottom-up model” announced in the ESD.3 The deemed savings are derived from literature, expert opinions and especially from metering of representative devices and appliances. As this approach can give estimates on the dimension of the so far achieved savings, it appears as an appropriate temporary solution to the authors. Nonetheless, the use of literature values for the quantification of energy savings leads to results with very uncertain accuracy. Especially in case of the ESD and the large structural heterogeneity of the 27 member states, the concept of deemed savings may seriously bias the results.

In this paper, we present a further development of the application of deemed savings and fixed values that (a) allows one to meet comprehensive requirements on data and precision even for small-dimensioned energy efficiency measures, (b) allows one to keep the costs of the quantification of the single measure reasonable and (c) that allows one for ex-ante approximations of the energy savings that will be realised under a certain setting of a projected energy efficiency measure. The approach is directly related to the implementation of the ESD and the ongoing definition of a standard on the methodology and data requirements for the national energy efficiency action plans (NEEAPs). Even though this approach has been developed for the evaluation of a regional pre-implementation of the ESD4 based on its draft version from 2003 (European Commission 2003), in the authors’ point of view, the approach presented in this paper is adaptive for all large energy efficiency programmes that aim to achieve their targets by carrying out a number of individual measures.

The rest of the paper is organised as follows: the section “International efforts for calculation rules for energy savings” provides a summary of the major points of energy savings calculations. This includes a discussion on the ongoing work of the European Committee for Standardisation on the development of harmonised methods for the calculation of energy savings. In the chapter “Strategic homogenisation of energy efficiency measures”, our approach is presented and an example of a possible application is given.

International efforts for calculation rules for energy savings

The need for guidelines for the calculation of energy savings is increasing. As the number of national and international agreements and programmes on the reduction of the energy demand is growing, methods are needed to evaluate the achievements of these activities guaranteeing comparable results. Particularly, in case of the ESD, the relevance of the need for a common standard becomes apparent, where the energy saving achievements of 27 member states have to be compared. The European Commission has announced a harmonised bottom-up model for the calculation of energy savings before January 1, 2008. Already in early 2007, the Commission realised that this deadline was set too ambitiously.5 As a consequence, the European Committee for Standardisation (CEN) has implemented a Task Force to develop both a top-down model and a harmonised bottom-up model. Since March 2007, this CEN TF 190 has been working on these issues intensively. An important basis for the work of the TF is the project Evaluation and Monitoring for the EU Directive on Energy End-Use Efficiency and Energy Services (EMEEES). This EU-funded research project started in late 2006, in order to develop 20 examples of harmonised energy savings calculations, and presented its final results in 2009. The CEN TF 1906 considers them as a reference for the development of their standard.

A key feature of the EMEEES methodology is that the effort of the calculation varies with the knowledge about the analysed energy efficiency measure and the desired accuracy of the results. Depending on these factors, three levels for the data requirements have been defined: Level 1 represents a deemed savings approach on the European level, level 2 refines this approach to country-specific values (which have to be provided by the countries themselves) and level 3 requires comprehensive data collection of the analysed measure (Vreuls et al. 2009). Beside the EMEEES project, additional literature exists providing guidance for the calculation of energy savings. Among others, highly elaborated documents are the Model Energy Efficiency Program Impact Evaluation Guide (Dietsch 2007), the Impact Evaluation Framework for Technology Deployment Programs (Reed et al. 2007), the International Performance Measurement and Verification Protocol, IPMVP (Kromer 2007) and the Evaluation Guidebook Volume I (Vreuls 2005a) and Volume II (Vreuls 2005b). Further more, the Californian Database for Energy Efficient Resources, DEER (DEER 2008) provides rich information on costs and impacts of a large number of energy efficiency programmes and measures.

Inputs from these project reports are used to develop the harmonised bottom-up model, in order to meet the requirements for measure evaluations in the light of the ESD. These project reports give guidance on monitoring and evaluation on a general and conceptual level.

Beside these supranational efforts, some European countries have developed their own bottom-up energy efficiency impact evaluation methodologies. The need for these methodologies emerged from the implementation of white certificates schemes7 in Italy (2005), France (2006) and the UK (2002).8 The methodologies used in these countries for the evaluation of the savings achieved by the participating parties are summarised in Capozza (2006), including a detailed description of the Italian evaluation methodology. Further country-specific descriptions of the evaluation practice are presented in Angioletti (2005) for France and in Ofgem (2005) for the UK. A comparison of these methods (including the Austrian methodology mentioned in the section “Background”) reveals the similarities and differences of the applied approaches: The Technical Guidance Manual I (Ofgem 2005) explains the quantification process of energy efficiency measures with regard to the UK white certificate scheme. Common household measures are quantified following a very detailed deemed savings approach (Ofgem 2007), providing, e.g. 51 different values for loft insulation in residential buildings, depending on the characteristics of the building and the quality of the insulation. In Italy, three evaluation approaches with different levels of accuracy are in use; these are (a) deemed savings (~70% of certified savings9), (b) engineering calculations based on default values and some on-field measurements (~20%) and (c) ex-post evaluation on the base of detailed energy monitoring plans (~10%). Each energy efficiency measure is quantified by the use of one of these approaches, depending on how well the energy savings of a certain measure are understood and, therefore, whether metered measure-specific values are necessary or not. The Italian and the UK deemed savings values are regularly updated to follow market transformation. In the French white certificates scheme, energy savings are evaluated ex-ante according to data concerning technologies and sales of equipment and corrected after receipt of the certificates. For the household sector, standardised methods for the quantification of 32 energy efficiency actions are available, which permit the calculation of savings without the need for measure-specific metering. The Austrian methodology for energy saving calculations with regard to the ESD is mainly a deemed savings approach with the possibility to account for measure-specific values, which differ from the default values. The framework and data requirements for the adoption of these measure-specific values have not yet been determined. The deemed savings calculations in the Austrian approach are based on straightforward engineering algorithms, whereas default values are given for all variables. These default values were taken from the corresponding literature and documents regarding evaluation methods for CO2 reductions (Bundesministerium für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft 2006).

All national approaches from the previous paragraph offer easily applicable methods for the most common energy efficiency measures in the household sector. The level of detail of these methods differs noticeably, from the very detailed approach of the UK to the collection of literature values in Austria. Besides this, another difference between these methods is the way how less common measures are handled or higher accuracy is achieved, where needed. In this regard, the Italian “three levels of accuracy” model appears very flexible and is comparable with the “three levels of harmonisation” approach of the EMEEES Project (Vreuls et al. 2009). The European Commission follows the approach of the EMEEES project in their draft harmonised bottom-up methodology; however, a final document is still not available. In the light of these conceptual references, the technical framework for energy saving calculations is given in the subsequent section.

Framework for energy saving calculations

In terms of this paper, we define the aim of an energy savings calculation as the change in the energy consumption initiated exclusively by the evaluated measure; in other words: to assess what the energy consumption would have been without the energy efficiency measure. To achieve this, the actual energy demand is compared with the unobserved BAU demand that represents the development of the energy consumption including all influencing factors such as energy prices, changes in population, climate, etc., but without the impact of the energy efficiency measure. In this regard, it is obvious that the determination of the BAU demand is both difficult to estimate with sufficient precision and crucial for the outcome of the creditable savings.

The authors are convinced that methods for a harmonised bottom-up evaluation are unmistakeable if two paradigms are incorporated: (a) methods should be developed for actions, not for measures. This means that, e.g. it is not the energy efficiency measure discount tickets for energy efficient heating systems itself that is quantified but the action initiated by offering the discount tickets, such as the installation of a condensing boiler. The second paradigm should be that (b) saving calculations follow the gross-to-net principle being widely discussed in literature (see, e.g. Dietsch 2007). According to this principle, in the first step, the gross energy savings induced by a certain action are determined and are then corrected for savings that cannot be assigned to the action itself to obtain the net savings. Each savings figure serves a different purpose. Gross energy savings are, e.g. used for forecasting, while net energy savings are better suited for programme evaluations (for a more detailed discussion, see, e.g. Vine 2008). The basis for the determination of the gross savings is the difference between the energy consumption in the BAU scenario and in the RP. This figure is then adjusted for energy consumption factors that are independent from the energy efficiency action. The general formula for the calculations of gross savings, therefore, is
$$ \begin{array}{lll} \textnormal{gross energy savings}_{t} &=& {\rm EC}_{{\rm BAU},t} - {\rm EC}_{{\rm RP},t} \\ &&\pm\; {\rm adjustments}_{{\rm gross},t}, \end{array} \label{equation-subscript_t} $$
where ECBAU refers to the energy consumption in the BAU-scenario and ECRP refers to the energy consumption in the RP (that is, after the energy efficiency action has become effective) in the year t. The variable adjustmentsgross accounts for factors of the energy consumption that are independent from the energy efficiency action, such as weather or number of people in a household.

The energy consumption in the BAU scenario is fictional to the point that it actually has never occurred and is, therefore, unobservable. Its determination heavily relies on assumptions regarding what “Would have occurred had the program not been implemented” (Dietsch 2007). For example, subscript t in formula 1 denotes that the BAU energy consumption and the energy consumption in the RP are not necessarily constant from the year of the implementation of the energy efficiency measure to the last year in which the measure is effective. As an example, consider an energy efficiency measure intended to replace an old atmospheric gas boiler before the end of its physical lifetime by a highly efficient condensing boiler. In this example, the condensing boiler would have been installed anyway, but the energy efficiency measure ensures that this installation is carried out 5 years earlier than initially intended. In this case, the highly efficient boiler replaces an old inefficient boiler in the first 5 years after the implementation of the measure, while from the sixth year on, the highly efficient boiler replaces a (other) highly efficient boiler and, therefore, does not save any additional energy compared to the situation where no energy efficiency measure was carried out (see also Reichl and Kollmann 2009). In practice, it is impracticable to determine whether the new device is a “replacement on burnout” or an “early retirement” for every single case. In this regard, a standard assumption—dependent on the configuration of the energy efficiency measure—is desirable.

For the net savings, the gross savings are corrected for savings that cannot be attributed to the energy efficiency measure. The most prominent examples of such factors are free riders, multipliers and rebound effects. Free riders are participants of an energy efficiency measure (that is, e.g. offering discount tickets to households for the purchase of an energy-efficient refrigerator) who would have implemented the energy efficiency action (the use of an efficient refrigerator) independent from the measure at the same time and with the same outcome anyway (the household would have purchased and used an equivalently efficient refrigerator even without the offered discount ticket). Multipliers are the opposite of free riders. A multiplier purchases and uses the efficient refrigerator from the previous example without the benefit of the energy efficiency measure. He does so as he has, e.g. experienced the performance of the new refrigerator through a measure participant and now gets an equivalent device himself. A rebound effect is observed if the reduced energy costs per unit of consumed energy service (e.g. the costs per degree room temperature) lead to an increased number of consumed units (increased room temperature) due to the reduced overall costs.10 The formula for net energy savings, therefore, is
$$ \begin{array}{lll} \textnormal{net energy savings} &=& \textnormal{gross energy savings} \\&&\pm\; {\rm adjustments}_{{\rm net}}, \end{array} $$
where adjustmentsnet refer to the savings corrections discussed in the last paragraph. Figure 1 illustrates the process from the difference of energy consumptions (BAU to RP) to the gross energy savings and, finally, to the net energy savings.
Fig. 1

Process of stepwise adjustments from the difference in energy consumption (ECBAU − ECRP) to the final net energy savings

The determination of adjustmentsgross is outlined in the IPMVP (Kromer 2007). These adjustments are carried out by using existing statistical data or by data collected simultaneously with the implementation of the energy efficiency measure. To adjust gross energy savings for factors such as free riders, rebound effect, etc., data are needed that cannot be retrieved from statistical resources. Each of these factors needs to be surveyed with respect to the specific energy efficiency measure it is needed for, which adds considerable effort to the measure evaluation. In many cases, a comprehensive savings calculation cannot disregard the adjustment for these factors, as their influence can be dramatic. Sathaye (2006) finds the percentage of free riders to be between 90% and 100% for energy efficiency measures regarding the building envelope, while the percentage of free riders for heating, ventilation and air conditioning (HVAC) is less than 10% in his analysis of the US housing sector. Similar findings are reported by PennFuture (2005). Whether the effort for the determination of these factors is essential depends on the purpose of the evaluation and the requested accuracy of the results. If the evaluation is conducted for an individual measure and rough estimates on its achievements are sufficient, a calculation of the savings without these adjustments can be appropriate in the authors’ opinion. In contrast to saving calculations where the results are used for programme internal purposes only, neglecting the incorporation of these adjustment factors appears to be problematic if measures carried out by different parties and/or in different countries need to be compared based on their saving achievements. In the authors’ opinion, energy-saving calculations in consequence of the ESD should clearly aim for sufficient precision, in order to distinguish between “real” energy efficiency improvements and “illusiveness” improvements that may be induced by neglecting the above factors. An important point is that the results of the evaluated energy efficiency measures will be used to revise the previous measures, thereby enhancing their effectiveness. In this regard, neglecting factors such as free riders or double counting may facilitate inefficient allocation of the existing administrative and financial resources.

Only in some cases and for some of these factors, methods different from customer surveys among participants can be applied. The Homes Energy Efficiency Database (HEED) of the British Energy Savings Trust contains detailed data of about 5 million homes (Amato 2008; Staniaszek 2005) in the UK (status: January 2008). These data include records about energy efficiency measures on an individual address level and by supplier, which allows for high-precision detection of instances of double counting (Energy Saving Trust 2004). Besides this, the assessment of these factors is usually carried out by surveys asking participants about their motives for the implementation of the energy efficiency action and how the energy efficiency measure has influenced this process. The elicitation of such factors from surveys poses two practical problems: (a) the uncertainty of the results from this type of survey is relatively high, which leads to the consequence that (b) sample size and survey techniques need to account for this circumstance. For a survey using highly elaborated and time-costly techniques in combination with a large sample size, the evaluation effort can increase to an unreasonable level compared to the effort for the energy efficiency measure itself.11

The effort can even exceed this level, when it comes to measures for which the realised actions are unobservable for the executing party. This could, e.g. be a measure that promotes energy audits for households. In this case, it might be impossible (due to limited resources) to monitor the implemented actions of the households. According to the evaluation principle (a), only energy efficiency actions in place of energy efficiency measures are quantified. To calculate the savings from these energy audits with respect to this principle, the executing party needs to determine the kind and number of the initiated actions and provide all necessary parameters for the calculation.

As discussed earlier, this high effort for data collection of the energy efficiency actions and the adjustment factors may mean that reporting for a number of energy efficiency measures fails to meet the requirements, with the result that the improvements in energy efficiency from these measures are not credited to the savings target. In particular, this problem is serious for energy efficiency measures focusing on a relatively small number of participants. These are, e.g., energy audits for households offered by a commune or by a local electric utility.

Strategic homogenisation of energy efficiency measures

Following the last section, the following variables are required for the quantification of the energy savings achieved by the implementation of a specific energy efficiency measure:
  • The energy efficiency actions initiated by the measure

  • The energy consumption in the BAU scenario

  • The energy consumption in the RP

  • The \(\textnormal{adjustments}_{\rm gross}\)

  • The \(\textnormal{adjustments}_{\rm net}\)

Each of these variable groups is determined with respect to the specific energy efficiency measure intended for the evaluation (in a comprehensive approach) or the values of these variables are taken from literature (as, e.g. in the temporary solution of the Austrian Energy Agency). Measure-specific variable determination is desirable as energy efficiency measures implemented independently from each other are usually heterogeneous regarding their configuration and the conditions under which they are carried out. This may result in energy savings that are hardly comparable with each other. As an example, the authors have evaluated three energy efficiency measures for the promotion of highly efficient refrigerators conducted by three local electric utilities in 2007.12 While the targets of these three measures have been identical, the instruments used in these measures differed to a serious degree. The measures reached from subsidies for the purchase of A+ + devices and additional payments for the verified disposal of the old devices to a subsidy scheme lacking a clear specification of the requirements for eligible devices. These differences in the configuration of the measures resulted in estimates ranging from 230 kWh per device for the measure strictly promoting A+ + devices to 99 kWh per device for the less specified measure (of which about the half of the overall new installed devices could not be credited as energy efficiency actions at all due to missing information on the energy efficiency labels of the new devices).

While measure-specific variable determination may be practically unfeasible for a considerable number of energy efficiency measures and for executing parties, the application of literature values for these variables may erode the reliability of the results. Therefore, the authors propose a different procedure, in order to determine these variables, which may increase the precision compared to retrieving the values from literature surveys and still allow to keep the effort on a reasonable level compared to the mostly impractical method (considering, e.g. a number of more than 150 million households affected by the ESD) of measure-specific variable determination. The focus point of our approach is to provide a procedure specifying how the information from the evaluation of one energy efficiency measure can be transferred to another measure, without starting each evaluation process from zero and, therefore, carrying out the full evaluation each time anew. In this respect, our approach is in line with the guidelines defined in the IPMVP (Kromer 2007). The IPMVP outlines the energy savings calculations for one individual measure, sometimes on the basis of relatively small stratified random samples of the contained energy efficiency actions. It therefore provides a procedure specifying how to transfer information from the evaluation of a subsample of energy efficiency actions to the whole number of actions within one measure. Our approach shows how the evaluated information of one energy efficiency measure (e.g. evaluated under the guidelines of the IPMVP) can be “reused” for the evaluation of further, similar measures, thus achieving a reduction of the effort. Appropriate sampling techniques are discussed in TecMarket Works et al. (2004) in detail.13

As a start, we define two different groups of variables that influence the outcome of an energy efficiency measure: We understand the configuration of an energy efficiency measure to be the bundle of activities set by the executing party, in order to induce the desired energy efficiency actions. The conditions of an energy efficiency measure are then the set of variables influencing the energy consumption that cannot be influenced by the executing party, such as geographical issues. For the rest of the paper, we make the assumption that the variables from the previous list are highly comparable even for energy efficiency measures that are carried out independently from each other, if
  1. 1.

    The configuration of the energy efficiency measures is equal.

  2. 2.

    The conditions under which the measures are carried out are equal.

  3. 3.

    The number of participants in each measure is sufficient, such that outliers do not significantly affect the result.


Following this assumption, the variables required for the determination of the energy savings do not have to be determined for each energy efficiency measure but are determined once for a reference measure and can then be reused (with limitations to the time) for the quantification of all measures complying with the requirements of the above list.

The concept of strategic homogenisation of energy efficiency measures (SHM), therefore, is to conceive a reference measure for each energy efficiency activity that is intended to be carried out more than once. The variables relevant for savings calculations of this reference measure are then determined with high precision. The following measures complying with the above requirements are then quantified by using the values determined for the reference measure instead of measuring these values for the new measures itself. This is carried out by the formulation of a homogenised protocol (HP). A HP consists of two-level information on the reference measure: (a) the first information level contains conditions on the measure-independent variables that need to be fulfilled to fit in the specific HP. These are variables such as the geographical area for which (relative) homogenous reactions on the energy efficiency measure among participants can be assumed. Additional conditions apply for the climate, the degree of equipment ownership of the hardware relevant for the specific measure or the usage pattern in the BAU-scenario. Level b of a HP defines precisely which activities are used and which incentives are offered to potential participants by the executing party. The reference measure is then conceived and carried out according to the HP along with a comprehensive monitoring. The detailed information on the variables then becomes part of the HP. For all new instances of the reference measure, the HP is used as a standard that clearly defines the configuration of the measure and its field of application. The concept of SHM is therefore located between the application of deemed savings or fixed values and the calculation of the energy savings on the basis of data collected specifically for each individual evaluated measure. Additionally, SHM clearly restricts the application of the saving values to a class of precisely described measures. The crucial point in this regard is that the variables determined for the reference measure can only be transferred to further instances of the measure if the configuration of these measures is selected to be equal. This includes the fact that the incentives offered to attract participants need to be the same in the reference measure, as well as in any further instances. In the context of discount tickets for the purchase of highly efficient white goods, this means that, beside the intensity of the subsidy (in percent of the price or absolute), the white goods that are eligible for the subsidy also need to be precisely specified. The determination of the requirement (1) therefore enables a precise calculation of the average energy consumption of the new white goods and (2) the constant intensity of the subsidy prevents from attracting different consumer groups or creating motives for structurally different purchases—e.g. a high subsidy could stimulate the purchase of an oversized refrigerator, whereas a small subsidy will probably less frequently cause this problem.

An example for a project that could easily be used as reference measure is the subsidy scheme for heat pump tumble dryers of the Swiss Agency for Efficient Energy Use (SAFE) in cooperation with the Zurich Municipal Electric Utility (ewz). In this measure, the purchase of a highly efficient heat pump tumble dryer is subsidised by about 250 € and the eligible devices are listed on the Swiss energy efficiency platform The Bern Municipal Electric Utility (ewb) has a very similar measure for the promotion of highly efficient heat pump dryers—including the same level of subsidy and the restriction to the same list of eligible devices. Even though statistical figures on the impact of the measures have not yet been published (see Bush and Nipkow 2006 for details), the equality of the incentives and requirements would make an evaluation of the savings effects of the ewz measure also conclusive for the ewb measure in the authors’ opinion.

In this example, the only necessary assumption for the application of SHM is that the pattern of use of households from Zurich and Bern are sufficiently similar. Under this assumption, the same HP could also be applied for instances of this measure conducted in Germany or France. If the subject of the energy efficiency measure was, e.g. a heating system—instead of a tumble dryer whose energy consumption is independent of the outdoor temperature—the application of the HP would be limited due to the different weather conditions in these regions. The geographical area and the subgroup of the population, which are appropriate for a joined evaluation, are therefore determined by the subject of the measure to a large extent.

In a classic deemed savings approach, the incentives used to attract measure participants are not incorporated in the evaluation of the achieved savings. In this respect, SHM can be interpreted as a special case of the deemed savings approach, as it is directly related to the evaluated measures and does not only give average values for some sector or subpopulation.

One crucial point for the successful application of SHM concerns the degree of homogenisation needed to achieve results with the desired precision. The diversity of the possible fields of application does not allow for a single answer to the question of how much homogenisation is sufficient. In the authors’ opinion, the crucial point is that the planning and evaluating institution is aware of the influencing factors and that the homogenisation of each of these factors is carefully balanced against the loss in precision if a factor is not homogenised (or the additional effort if this factor is surveyed for each individual measure). As another example for a field of application for SHM, we refer to the last section and point out energy audits offered by communes or local electric utilities. Energy audits for households pose several problems regarding a comprehensive monitoring if the concrete energy efficiency actions implemented by the households are unobservable. Unobservability in this regard is rather a resource problem than a measurement problem. Energy audits focusing on a subsector of the total energy consumption of a household, such as focusing on electricity only, are likely to initiate yearly energy savings (far) below 1 MWh.14 In this regard, the effort for the determination of every single implemented action is likely to exceed reasonable levels compared to the relative low savings. This is due to the fact that energy efficiency actions initiated by energy audits are usually implemented with delay and not all at once. To achieve a complete ascertainment, it can therefore become necessary to contact the participants several times. In case of communes or electricity utilities, SHM can be applied to avoid unreasonable monitoring effort and to achieve superior precision for the assessed energy savings compared to the use of fixed literature values. The executing parties cooperate on this topic and define their targets for the measures. According to the subject of the planned energy audits, a reference measure is conceived and the HP is set up with respect to the requirements of the executing parties.15 Costs for the monitoring of the reference measure are then borne by the executing parties instead of a single commune/utility. Besides the information on the area and period for which the specific HP is valid, the HP of the energy audit on the electricity consumption of households in this example contains information on:
  • Which energy efficiency actions are addressed by the audits

  • The frequency of occurrence of the implementation of the intended actions

  • The exact configuration of the audits, e.g. on-site, 30–60 min, electricity consumption meter explained and left to the participant...

  • The energy consumption of each addressed source in the BAU scenario and in the RP

  • The \(\textnormal{adjustments}_{\rm gross}\)

  • The \(\textnormal{adjustments}_{\rm net}\)

If all relevant points are included in the HP and the monitoring and quantification of the reference measure is carried out precisely, it can reasonably be assumed for each future instance of the reference measure (within a reasonable time period) that equal values are generated in the relevant variables. The effort for the quantification of a further instance of the reference measure is thus reduced to the collection of the non-homogenised variables, such as the number and specifics of the new installed devices. This means an essential reduction, as the most costly variables, e.g. the energy consumption in the BAU scenario, do not have to be determined again.

Therefore, we see two major advantages of energy savings calculations (additionally to the reduced quantification effort) according to the concept of SHM: As fewer resources have to be spent on the quantification of the instances, parts of the saved resources can be allocated to the monitoring of the reference measure. Consequently, the above mentioned variables can be determined with higher precision than if these variables were monitored for every single measure itself. Furthermore, the availability of many of the relevant variables as part of the HP allows forecasting achieved savings by an instance of the reference measure with relatively high accuracy. In the case of a long-term energy efficiency programme, such as the ESD, the possibility to forecast savings is of high relevance for the development of economically efficient energy efficiency strategies and action plans. Without accurate forecasts of the contribution of a certain energy efficiency measure to the overall savings target, together with the determination of the cost-effectiveness of the measure, it is likely that the selection of implemented energy efficiency measures does not represent the first-best solution.

Besides the positive preferences of SHM regarding effort and precision of energy efficiency measures and the ability to provide data for forecasting, the SHM concept has some restrictions. One of the major preconditions for the implementation of SHM is that all instances of the reference measure produce the same values in the variables relevant for quantification. This sets limits to the target groups for which SHM can be applied. Only groups for which one can reasonably assume high homogeneity in the reaction to a certain energy efficiency measure are suitable for SHM. This therefore requires a careful selection of the target group for which a HP is applied.


This paper addresses the calculation of energy savings achieved by energy efficiency measures for the household sector. Two approaches for the determination of the necessary variables for a bottom-up quantification are discussed: namely to determine these variables for each individual measure by metering and customer surveys and to apply fixed values obtained from literature. While the effort for comprehensive metering and representative surveys is high, the precision of the fixed values approach is assumed to be relatively low as these fixed values are not directly related to the specifics of the evaluated measure.

Therefore, we propose a concept in between these two approaches. In our opinion, the effort of a savings calculation can be reduced by the homogenisation of the energy efficiency measures. This is carried out by incorporating the similarities of specific energy efficiency measures already in the conceptual phase of the measures, such as of energy audits focusing on electricity consumption. When the conditions of the measures, as well as the activities of the executing party and the incentives offered to the participants, are defined equally for otherwise independent measures, it can be assumed that the variables relevant for quantification take the same values. The costly ascertainment of these variables then only needs to be carried out once and can consequently be reused for a number of similar energy efficiency measures implemented on groups of energy consumers that can reasonably be assumed to be homogeneous regarding their reaction to the energy efficiency measure.

Beside the reduced effort for the determination of energy savings achieved by the implementation of energy efficiency measures, the homogenisation of energy efficiency measures has some additional desirable preferences. Due to the a priori known values for some of the variables relevant for quantification, forecasts of the energy savings achieved by further instances of the homogenised energy efficiency measure can be carried out with high accuracy. This allows the planning of cost-effective and highly targeted energy efficiency strategies and action plans.


The lifetimes of energy efficiency measures for bottom-up quantifications are coded in CEN (2007) in a European context.


See, e.g. Adensam et al. (2008). These methods also allow for the inclusion of data specific for the evaluated energy efficiency action. In this case, the benefit of providing energy savings with a relatively low effort is lost.


The publication of this harmonised model was originally announced for 1 January 2008 by the European Commission. The authors expect the finalised paper for 2010.


In 2004, the regional government of Upper Austria created the energy efficiency programme Energie Star 2010 (Dell 2004). The authors were assigned with the calculation of the energy savings achieved within this programme and have thereby experienced the limitations of the existing approaches. Whether our approach will be applied for further energy saving calculations for the Energie Star 2010 programme will be decided when it is known if its results will be credited for the ESD savings target or not.


In 2007, the European Commission (DG TREN) confirmed to one of the authors that the deadline of January 1, 2008, would not be met and that the development of the harmonised bottom-up model would not be finished before 2010.


The leaders of the EMEEES subgroups on top-down and bottom-up calculations were chosen to also be the leaders of the subgroups of the CEN TF 190 on these topics to have excellent transfer of information.


For a description of the main principles of white certificate schemes, see Bertoldi and Rezessy (2008) and Oikonomou et al. (2007).


Dates in brackets refer to the first year for which compliance was obligatory. The regulatory framework was specified earlier, respectively.


See Labanca (2007).


Theoretically, it is also a rebound effect if the saved money from the reduced heating costs is used to finance an additional (energy-intensive) vacation trip. In this case, the rebound effect is nearly unobservable.


Vreuls et al. (2009) state that the costs for the determination of the adjustmentsnet can “easily reach 100,000 €”, whereas it is usually desired that evaluation costs do not exceed 3–5% (Dreessen and Langlois 2005) of the overall measure costs.


The evaluation was part of a study for the Austrian representation of electric utilities (VEÖ) in preparation for possible voluntary agreements in consequence of the ESD. The project report is not public.


There, the authors conclude that a stratified sample design is usually more appropriate than pure random sampling. One can usually find heterogeneous groups regarding their achieved savings—with highly different coefficients of variations. Simple random sampling is, therefore, not applicable without a significant loss in precision.


This represents electricity savings of ~20% of an average annual electricity consumption of 4,000 kWh per household.


It can be useful to have more than one reference measure, each with an associated HP.



The authors want to thank five anonymous reviewers for their valuable comments and input.

Copyright information

© Springer Science+Business Media B.V. 2009