International Journal of Sustainable Energy Planning and Management Vol. 23 2019 69 *Corresponding author - e-mail: k.jemmad@gmail.com International Journal of Sustainable Energy Planning and Management Vol. 23 2019 69–82 ABSTRACT As key elements of energy planning, ISO 50001 recommended inter alia identifying appropriate indicators to monitor and measure energy performance. Benchmarking can be a helpful tool to establish energy efficiency or performance indicators. While we agree that is hard to get an absolutely universal indicator aggregating several physical indicators defined in differing units; it is however possible to expand the area of cases covered or improve its characteristics such as accuracy, representativeness and simplicity. In this paper, we developed an aggregated dimensionless “Indicator for Energy Benchmarking” (IEB) to enhance the range of models of indicators dedicated to the engineering field. The systems targeted are low and middle level systems of the energy indicators pyramid. We built the proposed indicator based on specific characteristics: process decomposition-oriented, increasing when energy consumption decreases, dimensionless, with limited threshold value to 1. Consequently, the indicator provides many advantages in comparison to simple metrics and complex indicators such as: direct detection of energy use failure processes, creating interdependence between benchmarked systems scores, no need for data history to start benchmarking of a multisystem. IEB can be implemented as an integral part of many energy management or energy efficiency standards, methodologies or tools such as EN 16231:2012 and ISO 50001:2018. In last section, we calculate the indicator for 2 central sterile service departments of 2 university hospitals in Morocco to show its potential and operating mode. 1. Introduction Over the last decades, energy sustainability was on the public agenda as a main pillar of sustainable develop- ment. In response, researches has been allocated over different subjects like environment, policy, energy supply, energy use, energy security, energy transitions, etc. [1, 2, 3, 4]. In particular, energy planning and management was recognized as a central element for achieving solu- tions to energy sustainability by contributing to more efficient use of available energy sources, improving com- petitiveness and reduction of greenhouse gas emissions and other related environmental impacts [1, 5, 6]. As key elements of energy planning, ISO 50001, recommends inter alia identifying appropriate energy performance indicators to monitor and measure energy performance.[6] For that purpose, Energy efficie- ncy benchmarking is listed as an instrument in the energy management systems standard ISO 50001. Energy efficiency benchmarking can assist the plan- ning of energy targets and the review of energy effi- ciency progress. [7] External benchmarking may be used to establish a range of energy performance indicators for an installa- tion/facility or a specific product/service in the same field or sector. [8] Developing an aggregate metric to measure and benchmarking energy performance Kamal Jemmad*, Abdelhamid Hmidat and Abdallah Saad Laboratory of Energy and Electrical Systems, National Higher School of Electricity and Mechanics, Hassan II University, Route El Jadida, km7, Casablanca, Morocco Keywords: Energy performance; Energy efficiency; Benchmarking; Indicator; Central Sterile Service Department URL: http://doi.org/10.5278/ijsepm.3383 mailto:k.jemmad@gmail.com http://doi.org/10.5278/ijsepm.3383 70 International Journal of Sustainable Energy Planning and Management Vol. 23 2019 Developing an aggregate metric to measure and benchmarking energy performance used for an output and the output itself [12]. Metrics can be indicators (figures that indicate something) or index (several indicators combined into one) [17]. However, the two terms are widely used interchangeably in literature. Indicators can fall into four groups: Thermodynamic, Physical-thermodynamic, Economic-thermodynamic, and Economic [15, 18, 19, 20, 21]. Although economic indicators are useful at an aggregate level (i.e. the energy efficiency of the entire economy or the industrial sector as a whole), at a disaggregated level physical indicators give more insight into actual differences in energy efficiency levels. [18] Thermodynamic indicators are more suitable to power plants and technical equipment such as rotating machines, boilers, etc. They are expressed as a ratio of 2 energy units such as kWh or joules representing the consumption versus the output as a heat content or work potential. By using conversion rates, these metrics values can be expressed in percentage as for electrical efficiency. [22, 23]. But still, Patterson mentions that consumers, of course, do not value the end use service on the basis of its heat content or work potential. Therefore, physical- thermodynamic indicators have the added advantage that they directly reflect what consumers are actually requiring in terms of an end use service [18]. In that sense, they are more useful compared to thermodynamic indicators in engineering applications in industry, tertiary and residential sectors. Typically, indicators can be either: output divided by energy consumption called “energy productivity” – traditionally presented as “energy efficiency” [14, 24] – or energy consumption divided by output called “energy intensity” [19, 24, 25]. The Specific Energy Consumption (SEC) is a basic indicator which is a common indicator of energy intensity at the process level. [26]. It is sometimes also called the Unit Energy Consumption (UEC) or Physical Energy Intensity (PEI) [13]. The specific energy consumption (SEC) is a mixed physical– thermodynamic intensity indicator. [26] The denominator value (output) can be determined in terms of dominant parameter or unit of activity. [10] The dominant parameters can be of different types: − Dimensional parameters (area, volume). In commercial buildings, surface (in m2) is commonly used where energy use is primarily tied to plug loads, lighting and HVAC systems. [27] Although many relevant indicators were developed and used in their dedicated field. We estimate that there is still a need for development of new energy performance indicators due to many restrictions found in existing metrics as detailed in discussion section. To overcome these restrictions, we suggest to construct an indicator based on following conception elements: a) Process oriented decomposition of the system b) Aggregation of physical indicators with different units. c) Creation of interdependence between scores of benchmarked systems. d) The use of a dimensionless number. e) The higher threshold value of the indicator is limited to 1. f) The indicator must increases when energy consumption decreases In this study, we propose an aggregated dimensionless “Indicator for Energy Benchmarking” (IEB) to enhance the range of models of indicators dedicated to the engineering field. IEB is a process-oriented energy performance indicator to be used as a tool for assessing and benchmarking energy performance of low and middle level systems -according to energy indicators pyramid-. Aware of existing differences and difficulty in defining some energy use related concepts such as efficiency, con- servation, savings and performance. [9, 10, 11] We deal with energy performance indicators issues without dis- cussing differences between these concepts. Fortunately but strangely enough according to IEA, there are fewer problems in defining the concept of energy efficiency indicators. [10] 2. Literature review In this study the term “energy performance indicator” means the same as “Energy performance indicator” (EnPI) or (EPI), “Energy Efficiency Indicator (or Index)” (EEI), “Energy Use Indicator” (EUI), “Energy Intensity Index” (EII), “Measuring Energy Efficiency Perfor- mance” (MEEP) or others -used in literature as equiva- lent too. [12, 13, 14, 6, 10, 15, 16] 2.1. Energy performance indicators Basically, energy performance indicators are metrics intended mainly to assess how well the energy is used to provide the output. It consists of a ratio between energy https://www.linguee.fr/anglais-francais/traduction/interdependence.html International Journal of Sustainable Energy Planning and Management Vol. 23 2019 71 Kamal Jemmad, Abdelhamid Hmidat and Abdallah Saad bottom-up classification is: equipment/device, facility/ factory, sector, national economy. Energy performance indicators used for benchmarking at the down levels of the pyramid are more representative and gives more insight about energy use. However the quantity of data required at this level can be limiting. [14, 15, 19, 35] 2.3. Benchmarking Energy performance indicators - state of art: Data analysis is a substantial step of the energy benchmarking methodology which consist of: assess current performance levels, produce tables, charts and graphs. [8] Metrics are a main tool used in practice to perform this task. Approaches distinguished can be classified into 2 categories: 1st category: Comparison of actual Specific energy consumption SECact with a reference Specific energy consumption SECref. These energy intensity/SEC-based models are often used due to their simplicity and acceptable accuracy. [36] Many types can be used to determine a reference Specific energy consumption SECref depending on the aim of analysis: − Average. [13, 37] − Best plant based on an extensive survey of literature and exchange of information within the network during those years. Especially countries that are generally considered to be among the most efficient. [38] − Best practice observed. the complete production plant with the lowest specific energy consumption that already is in full operation; − Best practical means. the production plant with the lowest specific energy consumption that can be realized using proven technology at reasonable costs; − Best available technology: the production plant with the lowest specific energy consumption that can be realized using proven technology [13] 2nd category: Calculate a benchmarking Energy per- formance indicator or Index. Generally, a dimension- less index or indicator is constructed by aggregating system decomposition based on metrics cited in 1st category. Many approaches are proposed in literature especially for industrial sector. We note following methods: BEST: Benchmarking and Energy Savings Tool is developed by Worrell & Price and Lawrence Berkeley − Not properly technical (number of personnel, clients, rooms, beds, etc.). [28] − Properly technical: for instance, in industry the physical production corresponds to a dominant output of the branch and is usually measured in ton (e.g. crude steel, cement, clinker). [20] − Combination of many types: The output can also be expressed by a multi-parameters function: e.g. Bakar et al. proposes a new energy index where the output is expressed as: Area (m2) x number of occupants (person) in kWh/m2/ person. [29] Consequently, metrics that don’t relate useful output to energy consumed are out of the scope. For instance, The Power Usage Effectiveness PUE developed by the non-profit organization of IT professionals “Green Grid” is widely used by the IT industry as an energy efficiency indicator for data centers [30, 31, 32]. Defined by PUE= Total Facility Power/IT Equipment Power, it is clear -as confirmed by its developer and other experts - that PUE is an infrastructure energy efficiency not a data center productivity metric and therefore does not provide guidance about energy use by IT equipment. [30, 31, 32] 2.2. Benchmarking Energy performance Many definitions are available. The definition of EN 16231:2012 standard is a good one: Benchmarking is the process of collecting, analyzing and relating performance data of comparable activities with the purpose of evaluating and comparing performance between or within entities. [8] Energy benchmarking is useful for understanding energy use patterns, identifying inefficiencies in energy use, estimating potential for energy conservation, and designing policies to improve the energy economy. [33] Different types of benchmarking exist: Internal: compares performance against internal baseline or benchmark. External: compares performance against a metric “outside” of the organization identifies “Best in Class” performance. Quantitative: data-driven; compares actual numbers. Qualitative: based on best practices; compares actions. [34] Energy performance benchmarking is processed at many levels of aggregation, generally known as the energy performance indicators pyramid. An example of 72 International Journal of Sustainable Energy Planning and Management Vol. 23 2019 Developing an aggregate metric to measure and benchmarking energy performance The reference energy use represents the amount of energy an industrial sector would have used if no improvements in energy efficiency had taken place with respect to a certain base year (in this case 1995). The reference energy use is therefore also referred to as ‘frozen-efficiency’ energy use. The reference energy is based on the physical production of products of an industrial sector and the specific energy consumption for these products in the base year 1995: in which SECi,j,0 is the specific energy demand for energy demand type j to produce product i in the base year (e.g. in GJ per tonne of product) and Pi,k the physical production of product i in year k. [39] ENERGY STAR score The U.S. Environmental Protection Agency (EPA) supported the development of ENERGY STAR Energy Performance Indicators program (ES-EPI) for benchmarking energy performance of industrial facilities and ENERGY STAR Commercial Buildings Program for commercial buildings. Energy performance Indicators EPI score ranges from 1 to 100. According to Boyd et al.: The EPI is a statistical benchmarking tool that provides a “birds-eye” view of sector specific plant-level energy use via a functional relationship between the level of energy use and the level and type of various production activities, material input’s quality, and external factors, e.g. climate and material quality. The EPI uses stochastic frontier regression to estimate the lowest observed plant energy use, given these factors. [40] The physical production indicator: developed by Farla and Blok (2000) especially for the country-level analysis: where Ei,0 is the energy consumption of (sub-)sector i and subscript 0 refers to the base-year of the analysis. PPIi is the physical production index of sector I expressed by: , ,0 , reference , actual ,0 , ,0 , SEC SEC = ∑ ∑ i j i ki j k j i j i ki P E E P ( ) ,0 1 ,0 E Physical production indicator PPI PPI m i icountry i i=   = ×     ∑ ( ) 1 PPI P n i x x x w = = ×∑ National Laboratory (LBNL) based on decomposition of the entity to processes. [33, 36] The aggregated EEI “energy efficiency index” is calculated as follows: Where: EEI = energy efficiency index n= number of process steps to be aggregated EIi= actual energy intensity (EI) of process step i EIi,B= benchmark energy intensity (EI) of process step i Pi= production quantity for process step i Etot= total actual energy consumption for all process steps The Energy Efficiency Index (EEI): developed by Phylipsen et al. (2002) for the Netherlands. In which EEIa is the energy efficiency index for sector a, SECa the specific energy consumption for sector a, SECref;a the reference specific energy consumption for sector a, Ei the energy consumption for product i; mi the production quantity of product i; SECi the specific energy consumption of product i; SECref;i a reference specific energy consumption of product i; Ea the energy consumption in sector a, and i the products 1–n made in sector a. [38] The Energy efficiency indicator (EEI): developed by Neelis et al. (2007a) in which k is the year of analysis with 0 denoting the base year 1995, j the type of energy demand (electricity, fuels/ heat, non-energy use), EEIj,k the energy efficiency indicator for type of energy demand j in year k, Eactual,j,k the actual energy use from energy statistics for type of energy demand j in year k and Ereference,j,k the reference energy use for type of energy demand j in year k. 1 1 1 100 100* * , ⋅ = = ⋅ ⋅ ∑ ∑ ∑ n i i i= n n i i i i , i= o i= t t P EI EEI P E P E I B EI B ( ), ,, SEC 100 100 SEC SECSEC ii ia i a a ref a i ref ii ref i ii ii E m E EEI mm m = = = ∑ ∑ ∑∑ ∑ actual , , reference , = j kj k j k E EEI E International Journal of Sustainable Energy Planning and Management Vol. 23 2019 73 Kamal Jemmad, Abdelhamid Hmidat and Abdallah Saad Pérez-Lombard et al. (2012) consider that initial value judgments for the definition and qualification of service output become essential as a first step in the energy efficiency Indicators construction process. [11] However, definition of process/output intrinsic charac- teristics is out of the scope of this study. While some metrics are well known some are hidden/complex. In this study, we are not concerned by the determination of a specific metric for each process. Our work consists of how to use metrics to build an aggregated dimensionless indicator for benchmarking and assessment of energy use/performance. It is supposed in next sections that the output/process values to be used in equations are already clearly defined within the professional community related to systems in study. 3.3. Process indicator In the methodologies based on energy intensity -applicable to energy productivity too-, determining the benchmarks consists primarily of establishing the benchmark intensities for each of the sub-processes. [28] Then, as recommended by ISO 50001, indicators should be compared to a reference. [6] So we divide S E by a reference ref S E       , which by the way ensure a dimensionless value. So we get process indicator (PI) in the form: ref S EPI S E =       Until then, the use of ref S E       as a reference ratio is the common method as detailed in section 3. Now, we suggest another procedure: Decompose ref S E       into 2 reference values Sref and Eref. That is to say ref ref ref SS E E   =    It follows ref ref S EPI S E = (3); or in another way (4) ref ref ES PI S E = × Let consider a set of m comparable systems; each system composed of n processs of different or identical units as presented in Table 1. The process j belonging to system i is represented by the pair (Sij, Eij). Let Emin,j be the minimum value of the energy consumptions of all processes j in the set. Emin,j = min {E1 j, E2 j, ..., E nj} = mini E ij (5) (1) (2) PPIi,0 refers to the base-year of the analysis Px=the physical production of product x; wx= the weight of product x in the index. The aggregate SEC of each product (in a specific base-year) SECagg,i,0=Ei,0/PPIi,0 is chosen as the weight wx in the physical production index. Comparable results will be obtained if a best-practice SEC for a specific product would be considered as the weighting factor. Unlike the SEC of a specific product, the aggregate SEC will not have a meaning in itself, but will serve only to indicate the relative development of (physical) energy intensity in time. [26] 3. Methodology - aggregated dimensionless indicator for energy benchmarking In this section, we will present a developed Energy performance indicator as a contribution to enhancing the range of models of indicators available especially in the engineering field. 3.1. Construction basis The indicator proposed is intended for use especially in the engineering fields at the bottom and middle level of the energy performance indicator pyramid. That is to say, at the level of: company, factory/facility, and equip- ment/appliance. The systems targeted are mainly those powered by electrical energy. However it could be exploited for others energy sources. 3.2. Basic metric selection The relation between energy and output is generally represented by 2 basic forms: S/E “energy productivity” and E/S “specific energy consumption”. And as reported by Chang and Hu 2010, each represents identical measures from different perspectives, and they are used interchangeably in traditional literature. [24] We need to check that the increase in energy performance is equivalent to the decrease in consumption ∆E<0. So, we choose to use S/E ‘’energy productivity‘’ as the basic form to construct the indicator. 2 2 2 1 1 1 1 1 2 1 2 2 1 ; indicator decreases 2 ; indicator increases < ⇒ < < ⇒ < st nd E E E case: E E S S S S S S case: E E E E E 74 International Journal of Sustainable Energy Planning and Management Vol. 23 2019 Developing an aggregate metric to measure and benchmarking energy performance − It raises the bar higher: i.e. it is hard -however not impossible- to get , , ,0 max j ref min j SS E E   =    α . − Creates interdependence between systems scores. If one system make a positif step in one of the two drivers (E, S), the score of the others will get lowered if they don’t follow. This characteristic can help induce diligence and assiduity even among the best of class. 3.4. Aggregating processes indicators One of the major problems in creating indicators is the aggregation of different output units. Sij can be mass, volume, bulk, km, etc. The construction of a dimensionless process indicator in section 3.3 simplifies now the aggregation. The proposed aggregation of processes indicators to system-level indicator is a weighted arithmetic mean WAM. Since energy saving is the ultimate goal behind the use of energy performance indicators; the weight chosen is energy consumption. In other words, the indicator must expose the effect of big energy consumers performance on the system energy saving. The aggregated dimensionless indicator for energy benchmarking (IEB) of a system i is represented by Eq. (11): Eq. (11) can be simplified to ( ) , ,01 , 1 = = × = ∑ ∑ n ij min jj max j ni ijj S E S IEB E α 12; however it is more insight- ful and instructive to calculate process indicatorsn (PI)ij (11) ( ) ( ) , ,0 1 1 , 1 1 = = = = × × × = = ∑∑ ∑ ∑ n ij min j n ijj ij ijj max j ij n ni ij ijj j S E E E PI S E IEB E E α We note Emin,j,0 the first recorded value of Emin,j . And Smax,j is the maximum value of the outputs of all processes j in the set. Smax,j=max{S1 j, S2 j, …, Snj } = maxi Sij (6) α is defined according to the analyst estimation in order to keep Eref,j invariant for many years of benchmarking. E.g. if Emin,j can fulfill this condition, then α=1 and Eref,j= Emin,j,0. From (4) and (7), we define the j process indicator of a system i (PI)ij as: By definition, (PI)ij is dimensionless. From (8) and (9); we (PI)ij≤1 deduce that (10) When using , , ,0 max j ref min j SS E E   =    α , each time numerator or denominator changes, all the systems will have their scores changing. Which keep the users in a continued quest of energy performance improvement. The ratio , , ,0 max j min j S Eα is different from other references examples such as ref S E      max as it presents additional advantages: ( ), , ,0 , , Then, we set 7 ; where 0 1ref j min j ref j max j E E S S = < ≤ = α α (8) ( ) , ,0 , ij ref ij min j ij ref ij max j ij S E S E PI S E S E = × = × α (9) , , , , ,0 , We have , and , 1 then 1 ij max j min j min j ij min j ij ij max j i j S S E E E E E S S ∀ ≤ ≤ ≤  ≤    ≤  α α Table 1: Example of decomposition of m systems to n processes of different units per system Process 1 … Process j … Process n (kg, kWh) (m3, kWh) (nbr of pieces, kWh) System 1 (S11, E11) … (S1j, E1j) … (S1n, E1n)  ⁞ ⁞ ⁞ ⁞ ⁞ System i (Si1, Ei1) … (Sij, Eij) … (Sin, Ein)  ⁞ ⁞ ⁞ ⁞ ⁞ System m (Sm1, Em1) … (Smj, Emj) … (Smn, Emn) International Journal of Sustainable Energy Planning and Management Vol. 23 2019 75 Kamal Jemmad, Abdelhamid Hmidat and Abdallah Saad α, Emin,j,0 are fixed by definition. Ei is decreasing, so 1 Ei is increasing. Sij is increasing. If Sij