ČSN ISO 50006 - Systémy managementu hospodaření s energií - Hodnocení energetické hospodárnosti pomocí ukazatelů energetické hospodárnosti a výchozích stavů spotřeby energie

3.1.18 static factor

identified factor that significantly impacts energy performance (3.1.9) and does not routinely change

Note 1 to entry: Significance criteria are determined by the organization.

EXAMPLE Facility size, design of installed equipment, number of weekly shifts, range of products.

[SOURCE: ISO 50001:2018, 3.4.8]

####

Abbreviated terms

CDD

cooling degree day

CUSUM

cumulative sum

EnB

energy baseline

EnMS

energy management system

EnPI

energy performance indicator

HDD

heating degree day

SEC

specific energy consumption

SEU

significant energy use

Overview of EnPIs, EnBs and energy performance

An organization establishes EnPIs and EnBs to measure and monitor energy performance and demonstrate energy performance improvement.

EnPIs provide relevant energy performance information to interested parties (e.g. internal users, supply chain), to understand energy performance and take actions to control and improve energy performance.

EnPI values quantify the energy performance of the entire organization or its various parts (e.g. facilities, equipment, systems or energy using processes). Potential EnPIs need to be analysed to decide if they are appropriate before being selected. EnPIs can be expressed by using an energy model and can be reported in units of energy consumption (e.g. GJ, kWh) or energy efficiency (e.g. km/l).

Energy consumption of an organization can be significantly affected by relevant variables such as weather, production, etc. If the organization has data which indicates that relevant variables significantly affect energy performance, normalization should be carried out to enable comparison of energy performance. Normalization is used to account for the changes in the relevant variables to monitor and evaluate energy performance, and evaluate and demonstrate energy performance improvement.

Energy targets are set by the organization and may be based on identified and planned energy performance improvement opportunities.

Figure 1 illustrates an example of the relationship between energy performance improvement, EnPIs, EnBs, EnPI values and energy targets. Figure 1 also illustrates how energy performance improvement is achieved when an EnPI value improves compared with the EnB, whether or not energy targets are met.

The process to develop, use and update EnPIs and EnBs is described in detail in Clauses 5 to 10. This process helps the organization to monitor and evaluate energy performance and demonstrate energy performance improvement. The processes within the EnPI and EnB planning are presented in Annex A.

Key

X

time

Y

energy consumption

NOTE The trend of changing energy consumption indicates that there is (are) relevant variable(s) and normalization is required.

Figure 1 — Example of conceptual relationship between energy performance, EnPIs, EnBs, EnPI values and energy targets

Obtaining relevant energy performance information

Initial-energy-performance-related information

Organizations should identify current types of energy uses and evaluate current and past energy consumption and energy efficiency based on measurement and other data. Significant energy uses (SEUs) are identified by analysing this information together with factors that affect energy performance.

This process helps to identify the SEUs and prioritize opportunities for energy performance improvement.

NOTE This process is defined in ISO 50001:2018, 6.3 as “energy review”.

Determining users of energy performance indicators

EnPIs should be developed to meet the needs and expectations of different users and should be easily understandable.

Multiple EnPIs can be required to meet user needs. Aligning the EnPI boundaries with functional roles can ensure that the EnPIs meet user needs and that responsibility for managing the EnPI can be effectively assigned.

EnPIs can be developed for internal or external users. Internal users can use EnPIs for a wide variety of purposes such as, but not limited to, maintenance, operation and energy performance evaluation. External users typically use EnPIs to meet information requirements derived from legal requirements and other requirements (e.g. sustainability reports).

NOTE EnPIs and EnBs required for external purposes, such as those for government reporting, are not always sufficient for managing energy performance improvement under ISO 50001 or for organizations wishing to understand their actual energy performance improvement.

Table 1 describes some common EnPI users.

Table 1 — EnPI users

Types of EnPI users

Typical needs

Top management

Top management needs information from EnPIs to understand the energy performance of the organization and to support energy performance improvement actions.

Energy management team

Group who supports the organization, including top management in: a) setting up an EnPI, b) maintaining an EnPI, c) monitoring EnBs, current EnPI values, values of all relevant variables in predetermined intervals, d) setting energy targets and calculating extent of achievement of energy target, e) conducting normalization and comparison of current EnPI values with EnBs and energy target, f) reporting of EnPI values and deviations, and g) interpreting the results.

Plant or facility management

Typically controls resources within the plant or facility and is responsible for results. The plant or facility manager should understand both planned energy performance and investigate and respond to significant deviations in energy performance and in financial terms. Plant or facility managers may use all of the EnPIs in their plant or facility including the EnPIs regarding their SEUs, and comparable EnPIs from other sites for benchmarking purposes.

Operation and maintenance personnel

Responsible for using EnPIs to control and ensure efficient operation by taking actions for significant deviations in energy performance, eliminating energy waste and undertaking preventive maintenance. Operation and maintenance personnel may use the EnPIs relevant to the process or equipment for which they have responsibility.

Engineers

Plan, execute and evaluate an energy performance improvement action using suitable EnPIs including the method(s) used to evaluate energy performance improvement.

External users

External users such as regulatory bodies, professional and sector associations, EnMS auditors, customers or other organizations can need information from EnPIs to feed into their relevant processes.

EnPI owner

Person who is responsible for monitoring, analysing and reporting an EnPI and its values.

Defining the energy performance indicator boundaries

To measure energy performance, suitable measurement boundaries for each EnPI should be specified. When specifying an EnPI boundary the organization should consider the user needs (see 5.2) and also:

— organizational responsibilities in relation to energy management, including the level of control and/or influence which the organization has over its energy performance;

— the SEUs;

— facilities, equipment, systems or energy-using processes that the organization wishes to isolate and manage;

— the ease of isolating the EnPI boundary by measuring energy consumption and relevant variables;

— the EnMS boundary;

— available data for energy consumption and relevant variables.

The three primary EnPI boundary levels are individual, system and organizational as described in Table 2.

Additional information on EnPI boundaries can be found in Annex B.

Table 2 — The three EnPI boundary levels

EnPI boundary levels

Description and examples

Individual (facility/equipment/ energy-using process)

The EnPI boundary can be defined around the physical perimeter of a facility, equipment or energy-using process which the organization wishes to isolate and manage.

EXAMPLE 1 The energy use of steam production equipment separate from other energy uses.

System

The EnPI boundary can be defined around the physical perimeter of a group of facilities/equipment/energy-using processes interacting with each other that the organization wants to control and improve.

EXAMPLE 2 The steam production and the steam using equipment, such as a dryer.

Organizational

The EnPI boundary can be defined around the organization also taking into account the responsibility in energy management of individuals, teams, groups or business units as designated by the organization.

EXAMPLE 3 Steam purchased for a factory(ies), or a department of the organization.

Defining and quantifying energy flows

The organization should identify energy flows across the boundary. The organization can use a diagram (e.g. see Figure 2) to determine the energy information required to establish EnPIs. The diagram shows energy flows within and across the EnPI boundaries. They can also include additional information, such as metering points and product flows which are important for energy analysis and establishment of EnPIs.

The organization should measure energy flows across each EnPI boundary. This includes delivered and on-site generated energy. Consideration should be given to energy which crosses the EnPI boundary and is stored.

Key

M

measurement

Figure 2 — Energy, raw material and product flow diagram

Defining and quantifying variables related to energy performance

Organizations should determine relevant variables for each EnPI boundary. For example:

a) using more electricity to produce some products compared to another, the organization should consider using product mix as a relevant variable;

b) using more natural gas in the winter, the organization should consider using heating degree days as a relevant variable.

The factor that affects the amount of energy consumption required should be considered for relevant variables.

Table 3 presents considerations that are used to identify variables.

Table 3 — Considerations used to identify variables

Aspects

Description

Inputs

Input-related variables are based on the quality and/or quantity of inputs which enter the boundary (e.g. milk entering a pasteurization process).

Process

Process-related variables relate to the activities within the boundary. An example can be the different process temperatures and residence times required to complete process steps. In a building, a process related variable can be occupancy.

Outputs

Output-related variables are outputs which exit the boundary. In a manufacturing process, an output related variable can be the quantity of product produced.

Environment

Environment-related variables are based on the external environment (e.g. heating degree days, cooling degree days, relative humidity).

The variables may be directly measured or derived from measurements (e.g. production is directly measured whereas heating degree days is derived from measurements of outside environment temperature and base temperature).

Analysis of data collected by the organization can indicate relevant variables. A methodology to determine which variables are relevant is described in Annex D.

If the determination of relevant variables within a selected boundary is difficult, the boundary can be adjusted (e.g. subdivided).

Collecting data

Data collection

The organization should specify and collect the data related to energy consumption and relevant variables for each EnPI. It should plan the access to the collected data, the timing of collection, the process of collection and storage, and any pre-analysis cleaning or manipulation of the data.

It is possible that an organization will find that some of the EnPIs which were previously identified are not measurable due to data limitations or other barriers. In such cases, the organization should assess and consequently revise the EnPIs or introduce additional meters, measurement, or modelling methods.

If the expenditure towards installation of new meters, sub-meters and/or sensors to collect data on the required variables is justified by the improvement in its energy performance, the organization should specify such metering in its data collection plans.

In cases where more detailed energy consumption data are not available, energy bills can be used.

Consideration should be given to variation in the billing period between bills.

Table 4 describes examples of challenges in gathering data.

Table 4 — Energy data collection challenges

Scenarios

Description and examples

Lack of detailed measured data from energy suppliers

When an organization does not have detailed measured data from energy suppliers, they may consider additional options for measurements provided by themselves or through their energy supplier.

Lack of data on relevant variables

When an organization does not have data for certain facilities, equipment, systems or energy using processes, they may add measuring instruments to acquire these data, or use external sources such as official weather data. A proxy variable can be used for relevant variables for which data are not directly available (e.g. electricity consumption representing elevator usage as a proxy for building occupancy).

Data quality

The quality, precision and accuracy of the data collected to calculate EnPIs needs to be considered if the calculated results are to be meaningful. Prior to calculating EnPIs and their corresponding EnBs, the organization should review the set of measured energy consumption and relevant variables to determine the data quality.

Ensuring that data used are of appropriate quality and completeness can help increase the robustness of the determined EnPI value and ensure that they meet the needs of the organization. Factors to consider in determining the appropriate quality of data may include the following:

— the method of collection, i.e. manual or automatic;

— the source of data, e.g. third-party weather station data;

— the frequency of data collection, i.e. covering all shifts, hourly, daily, monthly, working hours and seasons;

— the accuracy of meters and measuring equipment;

— precision (measurement uncertainty regarding bias, linearity, resolution, etc.);

— repeatability of data from the data source;

— validation of the data.

Measurement

Measurements can be taken continuously (e.g. using data from a supervisory control and data acquisition system or a data acquisition and handling system), on a temporary basis (e.g. using data loggers) or on a spot basis (e.g. using mobile/portable meters). If continuous measurement is not possible, the organization should ensure that spot or temporary measurements are made during periods that are representative of typical operating conditions (see 5.6.2).

Energy consumption may be measured by using permanent or temporary, meters or sub-meters, or may be estimated by other means such as engineering calculations or modelling. Wherever possible, permanent meters should be installed for measurement. Temporary meters may be used during energy audit or where continuous measurements are not required. The organization should be aware of the accuracy and repeatability of the measuring equipment and should consider the relative importance of the decisions being made as a result of the data collected from these measurement instruments.

In many cases, the quantity of energy consumed can only be measured indirectly. This can require measuring a flow, volume or mass of fuel supplied, and can vary with factors such as composition, outdoor temperature, pressure and other factors. Multipliers or factors are commonly applied to the actual measured flow of gas or liquid fuel to calculate the quantity of energy contained in the fuel. These should be based on verifiable sources.

Data collection frequency

The data collection period and frequency should be sufficient to capture a range of operating conditions and provide an adequate number of data points for analysis. The organization should select an appropriate data collection frequency (e.g. hourly, daily, weekly) for the energy consumption and relevant variables included in each EnPI and the corresponding EnB.

NOTE Data collection frequency is typically based on available data (e.g. monthly energy bills).

The data collection frequency may be much higher than the frequency of reporting in order to measure and understand the impact of relevant variables on energy performance. For example, hourly, daily or weekly data collection can be needed at the operational level to address significant deviations. Such energy consumption and relevant variables should be aggregated for periodic review (e.g. monthly reviews at the organizational level).

For statistical analysis, it is necessary that energy consumption and associated relevant variable data have the same time intervals.

Even if the data collection period is the same (e.g. monthly), the data periods for energy and the relevant variables can be different. In such cases, the data periods should be adjusted so that the data period for energy and relevant variable data are aligned.

EXAMPLE Energy consumption is metered on the 20th day of each month and provided as an energy bill. The relevant variable (production) is measured at the end of each month. The organization decides to unify the timing of metering to the 20th day of every month and to estimate the data of production.

Identifying and analysing outliers

Faulty metering, faulty data capture or unusual operating conditions can produce significant outliers. Before excluding an outlier, investigations should be carried out to determine if there is a legitimate reason for the outlier. If some outliers are excluded, care should be taken to ensure that this does not introduce bias into the EnPI value or its corresponding EnB.

EXAMPLE An annual plant shutdown can result in a significant variation in energy consumption.

Outliers may be identified by appropriate methods (e.g. scatter diagram, trend line). Data points more than a pre-determined number of standard deviations from the expected value of the trend line or function may be outliers.

Determining energy performance indicators

General

While selecting appropriate EnPIs, the effects of relevant variables and the needs of users of the information are key factors to be considered.

If appropriate, organizations should determine EnPIs including at least one EnPI for each SEU.

There are many other types of indicators that are used to monitor other parts of the EnMS as defined by the organization (e.g. control of SEUs, increase employee awareness of energy, benchmarking equipment or processes). Care should be taken in using these indicators as EnPIs as they do not always appropriately monitor energy performance or appropriately represent measures of energy performance improvement.

While choosing EnPIs, the organization should consider its existing measurement and monitoring capabilities, related to energy consumption and relevant variables.

Annex C provides additional information on the selection of EnPIs.

When the organization’s objectives include reduction of greenhouse gas emissions, it should consider using additional indicators with CO2 emission factors. See Annex G for additional information.

EnPIs can be used for a variety of purposes such as:

— understanding the energy performance of facilities, equipment, systems or energy-using processes;

— communicating information and engaging the organization in issues related to energy performance;

— tracking progress towards energy targets;

— managing and controlling SEUs;

— monitoring and measurement of energy performance;

— evaluating and demonstrating continual energy performance improvement.

EnPI values may be available from measurements or calculations.

Expressing energy performance indicators

Statistical model

General

The organization should normalize (see Clause 8) its energy consumption or energy efficiency using an appropriate energy model. An energy model can be used to calculate the expected energy consumption or the expected energy efficiency.

One relevant variable

General

In cases where there is only one relevant variable, a simple linear regression or a nonlinear regression energy model for energy consumption or energy efficiency can be used.

A simple linear regression energy model for energy consumption can be expressed by Formula (1):

Y = mx + C (1)

where

 

Y

is the energy consumption;

 

m

is the energy consumption per unit of the relevant variable;

 

x

is the value of the relevant variable;

 

C

is the base load energy consumption, not related to the relevant variable.

Special cases of the linear regression energy model are described in 6.2.1.2.2 and 6.2.1.2.3.

Simple metric

In the specific case where m = 0, the energy model can be expressed by Formula (2):

Y = cE (2)

where

 

Y

is the energy consumption;

 

cE

is the constant energy consumption.

A simple metric can be used as an EnPI where there are no relevant variables affecting energy consumption. Whether a simple metric is appropriate or not can be established by observing a trend of energy consumption over time. This means that the daily, weekly or monthly energy consumption varies within an acceptable range as established by the organization. If Y is not constant, or within an acceptable range established by the organization, then this indicates that there can be relevant variables and normalization is required.

Ratio

In the specific case where c = 0, the energy model can be expressed by Formula (3):

Y = mx (3)

where

 

Y

is the energy consumption;

 

m

is the energy consumption per unit of the relevant variable;

 

x

is the value of the relevant variable.

In this specific case where the base load is zero, ratio of energy consumption per unit of the relevant variable (m) gives an appropriate energy model. This is usually known as specific energy consumption (SEC).

A ratio can be used as an EnPI when there is one relevant variable affecting energy consumption and there is no base load energy consumption.

Multiple relevant variables

In cases where there is more than one relevant variable, a multiple linear regression or a multivariable regression energy model can be used. A multiple linear regression energy model for energy consumption can be expressed by Formula (4):

Y = m1x1 + m2x2 + … + mnxn + c (4)

where

 

Y

is the energy consumption;

 

m1, m2, … mn

are the energy consumption per unit of relevant variables;

 

x1, x2, … xn

are the relevant variables;

 

c

is the constant value.

In practice, an energy model with multiple relevant variables is most common.

Aggregated models

An aggregated energy model can be calculated by combining different energy models.

Condition-based models are also aggregated models. In this case, different energy models are applied on either side of a threshold value (N) of a relevant variable. A condition-based model can be expressed by Formulae (5) and (6):

Y = f (x1, x2, … xn) –– if xi > N (5)

Y = g(x1, x2, … xn) –– if xi <= N (6)

where

 

Y

is the energy consumption;

 

f

is the energy model considering the relevant variables, when the relevant variable xi is above the threshold value (N);

 

g

is the energy model considering the relevant variables, when the relevant variable xi is below or at the threshold value (N);

 

x1, x2, … xn

are the relevant variables.

EXAMPLE The state of a plant includes not only “in operation” but also “part-load” and “standby”. A condition-based model can be used if it cannot be treated as an outlier.

Engineering model

Engineering models are often described by physical or empirical laws (e.g. equation relating fluid resistance and flow velocity to pump power consumption).

Engineering models can be used for calibrated simulation to assess the energy performance of simple and complex facilities, equipment, systems or energy-using processes.

NOTE Calibrated simulation is a simulation that adjusts parameters of the energy model so that the actual energy consumption and the simulation result (expected energy consumption) are equivalent.

The organization can use existing engineering models (e.g. for buildings). However, creating an engineering model can require particular expertise.

Establishing energy baselines

Concept of EnB

The EnB is used for the comparison of energy performance. Comparison is made for monitoring energy performance and demonstrating energy performance improvement.

The following steps should be taken to establish an EnB:

— determine the specific purpose for which the EnB will be used;

— determine a suitable data period;

— collect the data;

— analyse the data to develop a method of normalization (if applicable);

— determine and evaluate the EnB.

Determining baseline period

When establishing the EnBs the organization should determine a suitable period considering its energy goals and targets along with the nature of its operations. The baseline period should be long enough to ensure that the variability in operating patterns is accounted for by the EnPI and EnB (seasonality in production, weather patterns, etc.).

The frequency with which an organization acquires data can be a consideration in determining a suitable baseline period.

Table 5 presents typical baseline periods to be considered.

If an organization wishes to monitor EnPIs every day, even where a baseline period is one year, daily data are required for the EnB. In this case, the EnB is set for one year of daily data.

Table 5 — Typical baseline periods to be considered

Typical periods

Description and examples

One year

The most common baseline period is one year. It can capture a full range of weather conditions or business operating cycles.

Less than one year

A shorter period may be used where energy consumption is seasonal (e.g. a vegetable canning factory, ski resort).

Short EnB durations can also be necessary for situations in which there is an insufficient quantity of reliable, appropriate or available historical data.

More than one year

Seasonality and business trends can combine to make a multi-year EnB optimal (e.g. a winery wants to track energy performance only during the crushing and fermentation period of each year; however, over multiple years).

Normalization

Concept of normalization

During the operation of any facility, equipment, system or energy-using process the relevant variables routinely change. As a result, the energy performance, energy consumption and energy efficiency appear to fluctuate. Normalization is used to calculate the EnB to account for changes in the relevant variables.

The organization should establish an EnB for each EnPI using values of energy consumption and relevant variables during the baseline period. The EnB should be normalized if the organization has data indicating that certain variables significantly affect energy performance.

The organization should measure its actual energy consumption during the reporting period and compare it versus the expected energy consumption. An example of how to monitor and measure energy performance is shown in Figure 1.

Energy performance improvement is evaluated by the organization by comparing the EnPI value against the corresponding EnB.

A stepwise procedure is presented in Annex D as a guidance to carry out normalization.

A numerical example of the application of this procedure is given in Annexes E and F.

Uncertainty of model

When developing EnPIs and EnBs organizations should consider the uncertainty of measurements and of the energy model. Additional resources should be considered for higher accuracy.

Organizations should select an energy model that will result in EnPI values with the appropriate uncertainty for each comparison purpose. In the operation of facilities, equipment, systems or energy-using processes, EnPI value is compared with the operation criteria (e.g. upper and lower limits) and the energy target.

Maintaining energy performance indicators and energy baselines

General

When changes to facilities, equipment, systems or energy-using processes occur, energy efficiency, energy consumption and associated relevant variables can be impacted. The organization should ensure that the current EnPIs, the corresponding boundaries and EnBs are still appropriate and effective in measuring energy performance. If they are no longer appropriate, the organization should review or develop new EnPIs and corresponding EnBs.

There are several tests for determining whether the EnPI and EnB are still appropriate or valid including:

a) the relevant variables used to determine the expected energy consumption from the energy model should fall within one of the following:

1) within the range of relevant variables used in the model;

2) not exceeding a pre-determined number of standard deviations from the mean of the relevant variable data;

b) identifying major changes in static factors which can invalidate the determination of energy performance under equivalent conditions.

The baseline period can be revised (e.g. shifted to a different time period), or energy performance can be calculated without changing the baseline period.

Table 6 illustrates circumstances that can require organizations to revise EnPIs and corresponding EnBs.

Table 6 — Examples of circumstances that may require revision of EnBs and EnPIs

Common changes

Description and examples

Static factors

If a static factor (see 9.2) changes, the related EnB may be revised. In some cases, it can be necessary to develop a new EnPI and EnB. Statistical tests may establish whether an organization should develop a new EnB or EnPI. For example, these can include major production processes added or stopped and or changes to the number of production shifts or substantial modifications to the building structure and building equipment.

Relevant variables

When the relevant variable significantly changes and operates significantly outside of the range upon which the baseline was established, a new EnB and associated EnPI should be established.

Energy type

When an organization changes the types of energy it is using, it may need to modify what is tracked (EnPIs) and how those factors are weighted in its EnB.

Data availability

Improvements to the facility’s metering and data collection system can result in better quality data becoming available or new relevant variables coming to light. It can be necessary to revise EnPIs and EnBs.

Data frequency

If data are collected at more regular intervals or at a higher frequency, this can enable more effective management with a new EnPI and EnB.

Baseline period

Organizations may wish to update the baseline period to lock in accomplishments to date and focus on improving against the current energy performance instead of a past period. A strategic decision of such a nature can necessitate the updating of the baseline period to a more recent period (such as the last year) to serve as the new reference point.

According to a predetermined method

The organization can find it useful to identify conditions in advance that can require a revision of EnPIs and corresponding EnBs. For example, many organizations update their EnBs annually.

Static factor changes

Static factors need to be considered if they change, and if that change affects the relationship between energy consumption and relevant variables.

Table 7 describes examples when changes in static factors require revising the EnPIs or EnBs.

Table 7 — Static factor changes that require revision of EnPIs or EnBs

Scenarios

Description and examples

Change in product type

A plant produces a consistent set of products. The introduction of a new product that changes the set of products should be considered as a static factor.

Change in shifts per day

A plant has a fixed number of production shifts per day. If the number of shifts increases or decreases, then this should require maintenance.

Change in building occupancy

A building has a relatively stable number of occupants. If the number of occupants significantly increases or decreases due to new leases, then this should require maintenance.

Change in floor area

A building has a fixed floor area. If the organization significantly expands the building, or sells or rents out part of it, then this should require maintenance.

Monitoring and reporting of energy performance and demonstrating energy performance improvement

General

Energy performance can be monitored using EnBs and EnPIs for the following purposes:

— to ensure that operational control of processes is effective;

— to demonstrate energy performance improvement;

— to monitor progress towards achievement of energy targets.

Energy performance should be presented to users based on their needs and roles.

Monitoring and reporting

Energy performance can be monitored by comparing the actual energy consumption (EnPI value) against the expected energy consumption (EnB) on an hourly, daily, weekly or monthly basis. Monthly comparison can be adequate in the initial stages of developing these concepts.

Several tools and techniques are used to monitor and report energy performance, based on the energy model, including:

— monitoring the difference between the actual and the expected energy consumption using a trend chart of EnPIs (and relevant variables);

— monitoring the cumulative sum (CUSUM) of the difference between the actual and the expected energy consumption using a trend diagram;

— comparing the difference between the actual and the expected energy consumption with an energy target (energy target can be calculated as a targeted percentage reduction from the expected energy consumption);

— monitoring the energy consumption and the production by using a scatter diagram.

In each case, the information can be represented graphically or in tables.

The process of monitoring energy performance using EnPIs is routine. If an unexpected result is observed, the cause should be investigated by:

— investigating operational control of the equipment/systems to establish the cause of the deviation;

— if the deviation is causing excessive energy consumption, taking corrective action to prevent the deviation from occurring again;

— if the deviation is a result on unexpectedly low consumption, establishing the cause and trying to embed this action into normal operations;

— ensuring that the data are accurate;

— considering if a static factor has changed.

The results of EnPI monitoring can also be reported in summary or in detail.

Demonstrating energy performance improvement

Organizations can need to demonstrate energy performance improvement.

Improvement in energy performance should be evaluated by comparing EnPI values against the corresponding EnB(s).

This can be done at facility level, SEU level, process level, etc.

Table 8 illustrates some common approaches to monitor and report on energy performance improvement.

Table 8 — Monitoring, reporting and demonstrating energy performance improvement

Method

Description and examples

Formula

Difference

Difference between the reporting period EnPI value (R) and the EnB (B).

The cumulative sum of R – B is a useful technique for monitoring and for demonstrating improvement.

R – B

Per cent change

Difference between the reporting period EnPI value (R) and the EnB (B), expressed as a percentage of the EnB.

[(R – B)/B] × 100

Ratio

Ratio of the reporting period EnPI value (R) and the EnB (B).

(R/B)

Index

Ratio of the reporting period EnPI value (R) and the EnB (B), expressed as a percentage.

(R/B) × 100

Key

B is the EnB.

R is the reporting period EnPI value.

(informative) SEQ aaa \h SEQ table \r0\h SEQ figure \r0\h EnPI and EnB planning process

Figure A.1 — EnPI and EnB planning process

(informative) SEQ aaa \h SEQ table \r0\h SEQ figure \r0\h Examples of EnPI boundaries

During the process of measuring, monitoring, analysing and evaluating energy performance, and demonstrating energy performance improvement, it is important to find the most inefficient part of the production system. An EnPI boundary can be used effectively to focus on this part by narrowing the boundary. As a first step, the EnPI boundary is the entire organization. In such cases, the target boundary should be divided into several EnPI boundaries. As next steps, the EnPI boundaries should be narrowed down to the SEU level for identifying areas in which energy performance can be improved. Figure B.1 shows the EnPI boundary division process.

Figure B.1 — EnPI boundaries division process

When dividing the EnPI boundaries, organizations should consider that:

a) the number of divisions should be minimized;

b) it is recommended that the boundary is first divided into two parts such as SEU and other;

c) facilities that work in the same way should be categorized together;

d) the facility can be divided on the basis of process (e.g. process for product X, process for product Y and utilities);

e) the EnBs can be established for each operational status of the EnPI boundary.

The operational status refers to production ramp-up, normal operation, production hold, production stop, etc. As a minimum, it is recommended that organizations establish at least two EnB operational status conditions: under production conditions, and under non-production conditions.

(informative) SEQ aaa \h SEQ table \r0\h SEQ figure \r0\h Examples of energy performance indicators

Table C.1 provides descriptions about the expression of EnPIs, as well as examples of their applications. It outlines the various EnPI calculation methods as well as when an organization should choose each method. All methods should be regularly maintained to ensure valid results.

Table C.1 — Expression of EnPIs — Applications and examplesN IF "x_+3" " N " N 0 ,N COMPARE N 0 ) !Syntax Error, , = 1 " QUOTE "" " IF N N ">" " " QUOTE "" " IF "x_-3" "N IF N " N ">" " ""

Categories

Typical uses

Examples of EnPIs and applications

Observations

No relevant variable (simple metric, see 6.2.1.2.2)

— Measuring reduction in absolute consumption of energy.

— Meeting regulatory requirements based on absolute savings.

— Understanding of trends in energy consumption.

— Monitoring and determining energy performance improvement in cases where there are no relevant variables which affect consumption.

— Energy value as a basic EnPI to understand real energy use and calculate other EnPIs.

EnPIs:

— Does not take into account the effects of relevant variables, giving misleading results for most applications.

— Can be obtained from a utility meter/bill.

— energy consumption (kWh, GJ);

— electricity consumption for lighting (kWh);

— fuel consumption for a boiler (GJ).

One relevant variable (ratio, see 6.2.1.2.3)

— Expressing the energy efficiency of a piece of equipment or a system, particularly under standard or equivalent conditions.

— Monitoring the energy efficiency of systems that have only one relevant variable and no base load.

— Meeting regulatory requirements based on energy efficiency.

EnPIs and applications:

— Can be appropriate where there is one relevant variable and no base load and is misleading in other cases.

— Metrics of energy performance improvement SEC type EnPI value calculations should be avoided unless otherwise required for legal requirements or other requirements. Where SECs are required, it is good practice to include the underlying assumptions when reporting these values.

— Statistical tests can be required to confirm that there is no base load and that other relevant variables do not exist.

— kWh/t of production;

— GJ/quantity of product;

— l/100 km.

Statistical model

— System with one or more relevant variables and significant base load energy consumption.

— At the system and organizational level.

— Determining energy performance improvement within the relevant variable ranges during the baseline period.

— Evaluation of potential energy performance improvement opportunities, e.g. by evaluating different modes of operation of equipment or processes.

— Verification of energy savings from an implemented energy performance improvement opportunity.

EnPIs:

— Linear regression models with one or more relevant variables are commonly used and held as a best practice by many practitioners when monitoring energy performance and determining energy performance improvement for a facility.

— If the boundary is complex, the boundary should be divided according to the organization, activity type, etc, before applying the statistical model. This division reduces the number of relevant variables and thus makes it easier to check and maintain the appropriateness of the model.

— modelled energy-based unit.

Applications:

— model of energy performance of a production facility with one or more production types;

— model of energy performance of a hotel with variable occupancy rate and outside temperature;

— model of the relationship between the energy consumption of a pump/fan and the flow rate.

Engineering model

— To supplement energy flow measurements, and in some cases used to effectively infer or calculate energy flows from relevant variables.

— At a design phase to conceptually optimize energy performance or estimate energy performance improvement for a specific energy performance improvement action (EPIA) before investments are made.

Applications:

— Sometimes referred to as simulations, engineering models can capture the energy performance of simple or complex systems and facilities.

— Engineering models can encapsulate a large number of relevant variables and provide insights regarding process and/or system transient operation when properly calibrated to measured performance.

— If well calibrated, engineering models can provide a normalized basis by which to determine energy performance improvement.

— Engineering-based modelling can also be used to directly calculate energy losses or gains, e.g. additional waste heat recovery.

— Engineering modelling accounts for changes in boundaries and static factors over time.

— Details of engineering modelling are outside the scope of this document.

— Building modelling, including calibrated simulation, is another example of calculating an EnPI value from engineering models.

— Model of a petroleum refinery.

— Engineering models can be used even if relevant variables are not independent of each other (e.g. temperature and pressure).

— Model of an electric arc furnace: in addition to the measured electrical and gas flows, carbon powder is added to the batch to adjust the chemistry of the steel. This carbon also adds combustion energy to the batch, and although the number of bags is typically tracked, the process model is often used to calculate the energy contribution.

(informative) SEQ aaa \h SEQ table \r0\h SEQ figure \r0\h Example of normalization stepwise process

Preparation for normalization

The organization should use spreadsheets or software with data analysis capabilities for statistical calculations and the normalization process. Specialized statistical analysis packages can also be used when available. Depending on the physical processes, the organization may choose a linear or nonlinear model.

The following items apply to the evaluation of validity of the model:

— The adjusted coefficient of determination (R2) can be used to choose between the different models. Typically, it can be considered that a higher adjusted R2 value represents a better model.

NOTE Standard error can also be used to determine the most suitable model. A lower standard error represents a better model.

— An F-test is used to evaluate the overall statistical significance of a regression model. The regression model is considered statistically significant when the F-test value is less than 0,1, indicating that at least one of the potentially relevant variables used in the model has a “significant” effect on energy consumption.

— The P-value criterion is typically used to determine if a variable significantly affects energy consumption. A P-value less than, for example, 0,1 or 0,05 is often used as the significance criterion. These figures indicate that there is a 90 % or 95 % chance that the variable has a systematic impact and is thus significant.

Step 1 — Collection of data for the baseline period

The potentially relevant variables are selected based on brainstorming or by considering variables related to inputs, outputs, process and environment (see 5.5).

Considering the EnPI boundary, the data are collected for energy consumption and the selected relevant variables and tabulated for further analysis.

Step 2 — Correlation test

The purpose of step 2, which is optional, is to get a preliminary analysis of the relationships between the energy data and the variables.

A correlation test on all energy types and all potentially relevant variables should be carried out in order to obtain values for the coefficient of determination R2 for each single relation.

NOTE The correlation test is used to evaluate the association between two or more variables.

Step 3 — Regression analysis

Regression analysis is carried out to quantify the effect of each potentially relevant variable on the energy consumption. The results of the regression analysis are examined to establish which variables are relevant.

If more than one variable does not meet the P-value criterion, the one with the highest P-value is eliminated and the regression analysis is repeated with the remaining variables. There needs to be a technical understanding of why a variable is being eliminated. It can be due to inaccurate data or poor operational control. This process continues until all variables have a P-value of less than 0,1. This is the resultant energy model and the coefficients of each relevant variable, and the intercept are used as the baseline model.

After checking the significance of the coefficients, collinearity, which can lead to distorted results of the analysis, can be checked using various tests and indicators such as the variance inflation factor (VIF).

(informative) SEQ aaa \h SEQ table \r0\h SEQ figure \r0\h Example of normalization

General

The following example is for a heating system of a building. It outlines the basic steps taken to develop an energy model that can be used to determine and monitor energy performance and demonstrate energy performance improvement.

NOTE Gas consumption is normally measured in m3 but also can be expressed in kWh. The conversion is made using the formula: gas consumption (kWh) = gas consumption (m3) × calorific value × volume correction factor / unit conversion factor.  

Boundary and collection of energy data

An EnPI boundary was established for the whole building including all energy uses within. Monthly gas consumption data for this EnPI boundary was collected from available sources for the baseline period. The baseline period was established to be 1 January 2020 to 31 December 2020 as this time period fully captures all operational conditions during different seasons. The data was entered into a spreadsheet (see Table E.1).

Table E.1 — Monthly as consumption of the baseline period

Year

Month

Measured gas consumption kWh

2020 (baseline period)

January

1 414

February

916

March

1 007

April

921

May

475

June

260

July

218

August

252

September

635

October

607

November

1 038

December

1 267

Preliminary analysis of the data

A graph of monthly energy consumption data versus corresponding dates of the baseline period was generated (see Figure E.1), which indicates that the amount of natural gas consumed varies seasonally over the year.

Key

X

month (January to December)

Y

gas consumption, in kWh

measured gas consumption, in kWh

Figure E.1 — Gas consumption versus corresponding dates of the baseline period 2020

The consumption in the summer, where the heating unit is turned off, shows that there is a demand for heating energy (here for hot water, for washing hands, etc.) that is not dependent on the season (temperature) but can be seen as a base load throughout the year. These observations indicate that a simple metric and ratio are not suitable to use when determining energy performance and demonstrating energy performance improvement for the building. Accordingly, normalization is required.

Given that the natural gas consumed within the EnPI boundary is used by a heating system, heating degree days (HDD) data in kelvin days (Kd) were collected as a potentially relevant variable (see Table E.2). If the HDD are statistically significant, then it is chosen as a relevant variable.

Table E.2 — Consumption data and heating degree days

Year

Month

Measured gas consumption kWh

Heating degree days kelvin days (Kd)

2020 (baseline period)

January

1 414

598

February

916

397

March

1 007

325

April

921

345

May

475

95

June

260

0

July

218

0

August

252

4

September

635

168

October

607

210

November

1 038

393

December

1 267

476

The HDD data was entered into the spreadsheet and a scatter diagram of HDD data against natural gas consumption data was developed. Visual analysis confirms a linear relationship between natural gas consumption and HDD exists (see Figure E.2).

Key

X

heating degree days, in Kd

Y

gas consumption, in kWh

Figure E.2 — Scatter diagram of monthly gas consumption versus heating degree days in 2020

Regression analysis

The method of linear regression was selected to create an energy model. To create the linear regression energy model, a trend line (line of best fit) was created with the energy consumption and HDD data (see Figure E.2). The resulting energy model is expressed as discussed in Formula (1) (see 6.2.1.2).

The gas consumption is function of HDD, and the baseload is represented by the constant which is typically expressed as gas consumption (kWh/month) = 1,980 5 (kWh/HDD per month × HDD per month) + 253,88 (kWh/month).

The R2 of 97 % indicates a reasonable correlation.

NOTE Models with lower R2 values can still be used to generate meaningful results. A general threshold for R2 cannot be given as it is highly dependent on the variance in the analysed data.

Monitoring and reporting of energy performance and demonstrating energy performance improvement

This energy model enables calculation of expected energy consumption during a reporting period to monitor energy performance and determine energy performance improvement. For this, HDD data from January 2021 was obtained and inserted into the spreadsheet-based energy model to calculate the expected consumption for January 2021. The expected energy consumption (i.e. EnB) for January 2021 (1 1466 kWh) was then compared with the actual energy consumption for January 2021 (1 343 kWh) and the difference of −123 kWh indicates an energy performance improvement (see Table E.3).

Each individual month is monitored in the same way as with January 2021, by inserting the monthly HDD data into the energy model to calculate an expected energy consumption (i.e. EnB). The expected energy consumption is compared to the actual energy consumption and a difference is calculated. The difference can be positive or negative depending on the actual energy performance in the month.

A cumulative sum total of the difference between each month’s actual and expected energy consumption is developed to create a cumulative sum of differences (CUSUM). A downward trend in the CUSUM indicates sustained energy performance improvement (see Table E.3 and Figure E.3).

Table E.3 — Data including cumulative sum of differences (CUSUM)

Year

Month

Measured gas consumption

Heating degree days

EnB (expected consumption calculated using the model)

Energy performance change (measured gas consumption – calculated gas consumption)

CUSUM energy performance improvement

kWh

Kd

kWh

kWh

kWh

2021 (reporting period)

January

1 343

612

1 466

−123

−123

February

968

365

977

−9

−131

March

957

345

937

19

−112

April

750

298

844

−94

−206

May

451

75

402

49

−157

June

247

0

254

−7

−164

 

Key

X

month (December to June)

Y

CUSUM energy performance improvement, in kWh

Figure E.3 — Demonstration of energy performance improvement using CUSUM

Over the six months of 2021, the reporting period, the CUSUM indicates an energy performance improvement of 164 kWh.

Furthermore, a trend of the actual energy consumption (including the baseline period) with the expected energy consumption based on the energy model highlights the model fit and can be used to identify irregularities in the reporting period.

(informative) SEQ aaa \h SEQ table \r0\h SEQ figure \r0\h Example of normalization — Multivariate–analysis

Process flow and background information

The following information is considered in the example of normalization:

a) the company is in the food sector with large cooling requirements;

b) there are three primary products designated as product 1, product 2 and product 3;

c) part of the quantity produced of each of the products 1 and 2 are further processed to produce as product 3, i.e. some of the output of products 1 and 2 are raw materials for product 3.

Step 1 — Collection of data for the baseline period

The EnPI boundary is concerned with electricity consumption in the whole facility taking account of four potentially relevant variables: the quantity of each of the three products (production in metric tons) and the outside air temperature represented by cooling degree days with a base temperature of 5 °C (CDD5, in Kd).

Table F.1 shows the data for electricity consumption and potentially relevant variables. These data are from the historical baseline period, typically the previous 12 months.

Table F.1 — Electricity consumption data

Month

Outside temperature (CDD5)

Product 1

Product 2

Product 3

Actual electricity consumption

 

Kd

t

t

t

kWh

01/11

26

329,17

2 963,26

1 388,18

3 062 456

02/11

49

361,77

3 052,9

1 389,96

2 900 987

03/11

83

425,11

3 248,11

1 514,3

3 121 863

04/11

209

339,17

2 850,36

1 385,05

2 933 486

05/11

290

419,35

3 370,76

1 599,93

3 385 952

06/11

346

471,32

3 062,72

1 561,18

3 385 400

07/11

396

429,44

3 132,3

1 587,08

3 342 738

08/11

486

481,97

3 058,92

1 501,05

3 641 060

09/11

402

420,67

2 892,72

1 528,71

3 492 160

10/11

229

304,43

2 925,89

1 429,84

3 068 278

11/11

122

413,14

3 134,95

1 523,18

3 065 000

12/11

23

360,37

2 328,19

1 277,08

2 861 263

Step 2 — Correlation test

Table F.2 shows the coefficients of determination (R2) between the electricity consumption and the potentially relevant variables. In this case, it shows that there is a correlation between all of the potentially relevant variables and the electricity consumption. It is notable that the highest impact on the electricity consumption of this production process is CDD5 with a coefficient of determination of 0,79.

There is also a correlation between product 3 and products 1 and 2 of 0,51 and 0,65, respectively. This is possibly because product 3 is a further processed version of products 1 and 2.

In cases where variables are strongly correlated with each other, there is a possibility of collinearity which can distort the results of regression analysis although a variable can be relevant. If this occurs, it will become apparent during the regression analysis steps.

NOTE If other energy data are available, such as gas consumption for the same period, they can also be included in this test to show the correlation of the variables with gas consumption.

Table F.2 — Coefficients of determination

R2

CDD5

Product 1

Product 2

Product 3

Electricity consumption

CDD5

1

 

 

 

 

Product 1

0,41

1

 

 

 

Product 2

0,09

0,19

1

 

 

Product 3

0,44

0,51

0,65

1

 

Consumption

0,79

0,61

0,20

0,57

1

Step 3 — Regression analysis

Regression analysis was carried out to establish the energy model to be used. The results of the regression analysis indicate an adjusted R2 of 0,813 and coefficients and P-values for each potentially relevant variable.

Table F.3 shows the results of regression analysis using electricity consumption as the y-value and of the four potentially relevant variables as the x-values. If a P-value is greater than 0,1, the variable will be eliminated.

Table F.3 — Results of regression analysis

 

Coefficients

P-value

Intercept

2 057 245,67

0,020 404 26

CDD5

1 025,488 25

0,016 560 71

Product 1

1 368,220 95

0,173 102 76

Product 2

113,976 063

0,679 101 14

Product 3

13,129 039 8

0,989 536 54

Product 3 has the highest P-value which is greater than 0,10, and this indicates that it is not a relevant variable and should be eliminated from the regression analysis and a further regression analysis carried out without product 3.

If a potentially relevant variable is to be eliminated from the regression analysis, there should be a technical understanding of the reason for this elimination. In this case, the technical reason is indicated by the correlation of product 3 to both products 1 and 2. The cause of this correlation is that product 3 is a downstream process whose raw material includes some of both product 1 and product 2. Thus, product 3 is collinear with both products 1 and 2 which can lead to distortion of regression results. The potentially relevant variable product 3 is eliminated and the share of electricity consumption related to product 3 is included in products 1 and 2.

(Multi) collinearity can be checked by using various tests and indicators such as the variance inflation factor (VIF) and should be based on technical knowledge.

The results of completing the regression analysis a second time without product 3 is shown in Table F.4.

Table F.4 — Results of regression analysis

Regression statistics

 

Adjusted R2

0,836 415 11

 

 

Coefficients

P-value

Intercept

206 471 8,81

0,000 721 99

CDD5

1 028,119 02

0,003 154 67

Product 1

1 373,749 14

0,105 592 49

Product 2

117,008 3

0,402 144 43

Based on this second test, product 2 has P-value greater than 0,1 and needs to be eliminated from the regression analysis.

The cause of the high P-value has technical or operational considerations. After investigation, it was found that operational control in the product 2 department was not adequate. Electricity was being consumed even when no production was occurring. This is a very common cause of high P-value of low R2 in single variate analysis. The cause of this low P-value needs to be corrected.

The results of completing the regression analysis a third time without products 2 and 3 is shown in Table F.5.

Table F.5 — Results of regression analysis

Regression statistics

 

Adjusted R2

0,840 366 12

 

 

Coefficients

P-value

Intercept

2 326 964,59

6,242 8E-06

CDD5

1 034,570 76

0,002 168 14

Product 1

1 594,663 29

0,049 204 61

In this case, the P-values of both variables are below 0,1, indicating that there is more than a 90 % chance that both variables are relevant to electricity consumption. R2 is the highest of the three tests. This gives a high level of confidence in the baseline formula for the expected energy consumption.

The resulting energy model is expressed as discussed in Formula (1) (see 6.2.1.2).

The expected electricity consumption is a function of two relevant variables, namely the cooling degree days and product 1.

Based on this, the baseline model in this case is the expected electricity consumption (kWh/month) = 1 595 (kWh/tonne of product 1) + 2 326 965 (kWh/month) + 1 035 (kWh/CDD5).

This formula is then used in the reporting period to compare expected electricity consumption with actual electricity consumption each month.

The trends in Figure F.1 show a comparison of the actual and the expected electricity consumption during the baseline period. This trend can give confidence about the accuracy of the baseline formula.

Key

X

month (January to December)

Y

electricity consumption, in kWh

actual electricity consumption

expected electricity consumption

Figure F.1 — Actual versus expected electricity consumption in the baseline period 2020

NOTE 1 In the case illustrated above, operational control was improved in the product 2 department during 2021 and the baseline regression analysis was repeated in early 2022. The product 2 P-value had decreased, and it became a relevant variable in the 2021 baseline model used as the baseline for monitoring performance each month in 2022.

NOTE 2 The steps above can be used to develop a statistical model at the whole facility level as demonstrated in this annex as well as at SEU, building or process levels. The regression analysis is the same.

(informative) SEQ aaa \h SEQ table \r0\h SEQ figure \r0\h Reporting aggregated information

Using statistical or engineering models for calculating EnBs and corresponding energy targets in order to compare those with the measured energy consumption has the advantage that all changes in energy performance have the same units (e.g. kWh) and can therefore be summed up over different types of processes and systems, resulting in a total change in energy performance over all SEUs.

In addition, changes in energy costs and CO2 emissions can be calculated and aggregated. An example is given in Table G.1.

Regardless of the form of the mathematical function linking energy consumption to relevant variables, the results need to be in energy consumption units such as kWh or GJ. These resultant energy consumption figures can be added together to give overall results. For example, the energy performance change of several SEUs in energy consumption terms can be added together to give the total change in energy performance.

Table G.1— Example of an aggregation of energy performance indicator valuesN IF "x_+3" " N " N 0 ,N COMPARE N 0 ) !Syntax Error, , = 1 " QUOTE "" " IF N N ">" " " QUOTE "" " IF "x_-3" "N IF N " N ">" " ""

Facility/ equipment/ process/ system

Energy type

Change in energy performance

Change in costs

Change in greenhouse gas emissions

Change in energy performance

Primary energy factors by type

Change in energy performance relate to primary energy

Energy unit costs by type

Change in energy cost

CO2e emission factors by type

Equivalent in CO2e

MWh

MWh

$/MWh

$

t/MWh

t

Facility 1

Gas

−730

1,1

−803

50

−36,500

0,18

−131,4

Electricity

10

2,5

25

140

1,400

0,61

6,1

Process 1

Electricity

−468

2,5

−1 170

140

−65,520

0,61

−285,48

Non-SEUs

Gas

−12

1,1

−13

50

−600

0,18

−2,16

Electricity

−21

2,5

−53

140

−2,940

0,61

−12,81

Whole site within boundary (aggregated)

 

−1 221

 

−2 014

 

−104,160

 

−425,75

Bibliography

[1] ISO 50001:2018, Energy management systems — Requirements with guidance for use