monitoring and targeting - southcorner...

In-depth management guide

Monitoring and targetingTechniques to help organisations control and manage their energy use

2Monitoring and targeting

Preface

Reducing energy use makes perfect business sense; it saves money, enhances an organisation’s reputation and helps everyone in the fight against climate change.

The Carbon Trust provides simple, effective advice to help organisations take action to cut emissions. One of the simplest ways to do this is to use energy more efficiently.

This technology guide explores monitoring and targeting techniques, and shows how organisations can adopt an appropriate level of monitoring and targeting in a way which will help them to save energy and cut costs.

Menu


Even for the more advanced aspects, an appreciation of the basic physical principles behind the use of energy at your place of work and a basic grasp of maths is all that is needed.

M&T is an extensive topic and this guide can only provide an introduction to the more complex analysis methods and tools that can be used.

Users should contact the Carbon Trust or an expert consultant for more detail or further advice on applying these techniques within their organisation.

2. Quantify the savings achieved by any and all of your energy projects and campaigns in a manner that accounts fully for variations in weather, levels of production activity and other external factors. Many users cite this as the most valuable result of M&T.

3. Identify fruitful lines of investigation for energy surveys. Rather than starting a survey with no clear agenda, you can go prepared with specific questions to ask, prompted by observed erratic or unexpected patterns of consumption.

4. Provide feedback for staff awareness, improve budget setting and undertake benchmarking.

This guide presents M&T from two perspectives. One is routine use (on a weekly cycle, for example). Routine M&T as explained here is quick and simple and requires no particular expertise on the part of the user. The other perspective is target-setting and diagnosis, an aspect that will appeal to users who wish to analyse data in more depth. Although this aspect needs to be addressed when first setting up an M&T scheme, it then becomes optional once the system is up and running.

The purpose of monitoring and targeting (M&T) is to relate your energy consumption data to the weather, production figures or other measures in such a way that you get a better understanding of how energy is being used. In particular, it will identify if there are signs of avoidable waste or other opportunities to reduce consumption.

Data collection may be manual, automated, or a mixture of the two. Once an M&T scheme has been set up, its routine operation should be neither time-consuming nor complex. An M&T scheme will provide essential underpinning for your energy management activities, allowing you to:

1. Detect avoidable energy waste that might otherwise remain hidden. This is waste that occurs at random because of poor control, unexpected equipment faults or human error, and which can usually be put right quickly and cheaply (or, indeed, at no cost). Intercepting and rectifying such problems should more than cover the cost of operating the M&T scheme. See the next section for some examples.

What is monitoring and targeting?Monitoring and targeting is an energy management technique that can be applied in any type and size of organisation, whether commercial, industrial or public sector.

Menu


The meaning of target

Monitoring and targeting can be focused on invoice checking, contract tariff negotiation and financial budgeting, or on the physical performance of the organisation’s buildings, processes and vehicles. It is the latter interpretation that we will use in this guide.

Often, targets are set without consideration of practical application or achievability. This guide focuses on achievable operational targets. An operational target:

Is calculated rationally to reflect known achievable •performance

Is discussed and agreed•

Is applied to individual monitored streams •of consumption

Is continually amended to reflect improvements•

Does not need to be widely disclosed•

Has performance reviewed monthly at most, •commonly every week, or even daily or per shift

Is beneficial for day-to-day cost and •environmental management.

Hidden avoidable energy loss happens all the time, usually because of some minor control failure or the inappropriate action of occupants or maintenance staff. This guide will help you detect and remedy this waste, and disclose other opportunities to save energy and money.

Examples of waste and saving that have been identified by employing M&T include:

Losses that would have equalled £3,500 a year •were incurred when a limit switch came loose at a waste-water treatment works, causing some machinery that should have been running intermittently to run continuously.

Gas consumption at a council depot doubled when •a maintenance contractor left the heating system running 24 hours a day.

The front steps of an office block were found to be •costing £5,000 a year because the control had failed on the de-icing heaters (which the owners did not even know they had).

Frost-protection systems are a prime cause of waste. •One energy manager cut the electricity consumption of his HQ building by 40% when he found that all the electric frost protection pre-heaters were running on his air handling units.

Hundreds of pounds a year were lost when someone •left a bypass valve open on a steam trap in the basement of a paper mill.

Further information

Fact sheetsenergy management (GIL136)

understanding your energy consumption (CTL001)

Assessing the energy use at your industrial site (CTL002)

Assessing the energy use in your building (CTL003)

Menu

http://www.carbontrust.co.uk/Publications/pages/publicationdetail.aspx?id=GIL136

http://www.carbontrust.co.uk/Publications/pages/publicationdetail.aspx?id=CTL001







Figure 1 An example of an overspend league tableAn overspend league table is a simple and very effective report format to ensure instances of exceptional consumption are detected. It can be applied in any circumstances: by large and small users, for buildings, processes and vehicles, and at any chosen interval. It can be implemented in spreadsheets or by any good proprietary M&T software, and requires no special skills or knowledge to interpret it.

An overspend league table is simply a list of variances ranked in descending order of cost. ‘Variance’ is the difference between the actual quantity used and

the corresponding expected quantity. Methods for estimating expected consumption are covered later in this guide.

Reporting exceptionsOne of the key functions of a monitoring and targeting scheme is to alert the user to instances of exceptional excess consumption of energy.

Assessment intervals

This guide refers to weekly assessment intervals. Smaller businesses may prefer to operate M&T schemes at monthly intervals, while ambitious energy-intensive firms may opt for daily or even per-shift reporting.

Longer intervals mean more delay before exceptions are flagged up, and at short intervals there is more risk of spurious alerts, even if all the data can be collected at the required frequency. experience shows that weekly analysis is a good compromise, at least as a starting point.

The main point to bear in mind is that equal time intervals should be used.

Menu


This form of exception report has many advantages. First and foremost, the most significant items are always at the top of the list, so it does not matter whether the report contains 10 items or 10,000. Second, it will be immediately evident how much remedial effort is warranted; indeed in some weeks there may be no problems that are sufficiently costly to merit investigation. Third, as exceptions are judged on cost, non-energy commodities can be included in the same report. By including other utilities and consumables such as water, effluent discharge volumes, chemicals and so on, the overspend league table can provide valuable benefits beyond energy savings alone.

The final advantage of the overspend league table is that it requires no special expertise to interpret it. All the user needs to do is note which consumption streams appear significantly overspent, verify the related input data, and if the problem seems genuine, raise the issue with whoever operates or understands the item in question.

Routine monitoring by means of an overspend league table is so simple that it should take no more than a few minutes a week – unless, of course, additional time is warranted by the occurrence of a significantly costly problem. Using it is a robust procedure, since almost anyone can deputise for the principal user in his or her absence.

Although this guide goes on to describe various associated analytical and charting techniques, these can be thought of as optional. Once an M&T scheme has been set up, the overspend league table may be all that is needed for simpler examples.

By including other utilities and consumables, the overspend league table can provide valuable benefits beyond energy savings alone

Top tip

It is essential to ensure that variations in input data are not caused by faulty measuring equipment or sensors. Meters and sensors should be regularly maintained and calibrated.

Case studyA borough council

The energy manager at a borough council applied M&T on a monthly basis to all council buildings, using in-house manual meter readings. Some months after implementing the scheme he detected increased water consumption, and in some cases increased electricity consumption, at several of their public conveniences.

On investigation it transpired that the gents’ urinal flush controls had been replaced. new infrared motion sensors had been fitted in place of the old hydraulic controls and were responding to homeless people sleeping in the toilets overnight. Increased electricity use was mainly due to the rough sleepers jamming the hand-dryer pushbuttons with matchsticks, although in one case the lavatory attendant was living in a store cupboard and had wired in an electric heater.

Menu


Precedent-based targets, used with caution, may be the best method when consumption is seasonal, but unrelated to any measurable driving factor. Retail premises, for example, may exhibit peaks of consumption during the pre-Christmas period and New Year sales, and general electricity demand in schools, which otherwise might be steady through the year, should fall during holidays.

Generally, however, precedent-based targeting models can be too simplistic, and organisations may want to consider activity-based targeting. This is particularly appropriate when there are clear drivers for changing energy consumption, for example, changes in production throughput. See page 18 for examples of how data can be compared using automatic meter readings.

Precedent-based targeting

Precedent-based targeting models are most commonly used in monthly monitoring schemes, when expected consumption can be deduced from what was used in the corresponding month a year before. One weakness of this procedure is that it assumes that conditions (especially the weather) were comparable in the two months. This is not always the case, although in some circumstances the differences may not be too significant.

A more problematic issue is what happens when energy waste has occurred. The resulting excessive consumption erroneously raises the expected quantity a year later. To work effectively, discount abnormal months to prevent them being used for target setting. At the other end of the scale, some automatic monitoring and targeting schemes attempt to compare consumption at very short intervals (e.g. half an hour), with a target template derived from previous similar days. The same caveat applies: abnormal consumption patterns must be filtered out from the pool of data used as precedents to make this effective.

Methods for calculating expected consumption fall into two categories. There are those based on precedent (comparison with previous periods), and activity-based methods that relate expected consumption to its driving factors (weather, production throughput, mileage, etc.). For the sake of brevity we will refer to any procedure for calculating expected consumption as a ‘targeting model’.

estimating expected consumptionThe crux of the overspend league table – and of monitoring and targeting in general – is having good estimates of expected consumption for comparison with what is actually used.

Menu


1 Limited figures are available free of charge. Commercial data providers can be found on the web and it is also possible to calculate figures from local measurements.

Activity-based targeting

Activity-based targeting models calculate expected consumption by reference to its driving factors – the measurable things that cause consumption to vary. Examples of these are given in Table 2.

Table 1 Driving factors

Consumption driven by the weather

With very few exceptions, energy users of all kinds are dealing with weather-related energy consumption, usually for heating but increasingly also for cooling. Here, external air temperature is the dominant influence. For convenience, regional or local air temperature data can be converted into ‘degree-day’ values, which provide an index of how cold (or hot) the weather was in a given place in any particular week or month1.

Fuel consumption for space heating will normally exhibit a straight-line relationship to heating degree days (Figure 2), while electricity consumption should relate to cooling degree days (Figure 3) if any chilling plant exists on the metered circuit.

In either case, it is possible to identify a performance characteristic line which typifies the relationship. For a given degree-day value, the characteristic line enables energy users (or, more usually, the M&T software or spreadsheet) to ‘read off’ the expected consumption (see Figure 4). This is explained more fully later.

Energy use Possible driving factor

Space heating Outside temperature

Air conditioning Outside temperature; possibly also humidity levels

Steam raising Quantity of steam produced

Production process

Production quantity

exterior lighting Hours of darkness

Drying Quantity of water removed from product

3,500

3,000

2,500

2,000

1,500

1,000

500

00 20 40 60 80 100 120 140

Laboratory Electricity

Cooling degree days

Performance characteristic line(s)

kWh

Figure 2 Example of relationship between fuel consumption and heating degree days

1,800,000

1,600,000

1,400,000

1,200,000

1,000,000

800,000

600,000

400,000

200,000

00 10 20 30 40 50 60 70

Building 14 gas

Heating degree days


kWh

Figure 3 Example of electricity consumption related to cooling degree days

Menu


In Figure 4, the expected consumption where degree days = 28 is 1,250,000kW/h. The intercept (energy = 900,000kW/h where degree days = 0) shows the consumption where no heating is required. The values on the graph at this point could represent gas being used for production or hot water etc. The scatter points can indicate various things, including how well controlled the heating system is, but they can be influenced by other variables such as varying production levels in a factory.

Consumption driven by production throughput

In some (but by no means all) production processes, energy consumption can be related to production throughput using the same straight-line basis that was explained above. Figure 5 illustrates just such a case for gas used in a commercial bread oven.

If a simple straight-line relationship is not applicable, other methods can be used. However, as there are

operational advantages in being able to represent performance as a straight-line characteristic on a scatter diagram, methods that manipulate the data to make that possible are generally to be preferred. It is beyond the scope of this guide to deal with these in detail, but some hints are given below. Suffice to say that any appropriate method can be used to infer expected consumption from other independent measurements.

Figure 4 The performance characteristic line superimposed on the scattered data points

Figure 5 A performance characteristic line for a simple process

1,800,000

1,600,000

1,400,000

1,200,000

1,000,000

800,000

600,000

400,000

200,000

00 10 20 30 40 50 60 70

Building 14 gas

Heating degree days

Performance characteristic line(s)kW

h250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200

Process ovens – gas

Output (tonnes)

kWh

Note

The characteristic line will usually only be needed for computations within the normal range of data. The position of the intercept must be interpreted with caution if it is extrapolated well beyond this range. Where degree days are the driving factor, the intercept will also be sensitive to the choice of base temperature.

Menu


Alternatives to straight-line targets

Production processes are often more complex to set targets for. One common situation is where several products with different energy intensities are made on the same plant. Here, a tabular form of calculation is necessary to establish expected consumption.

Suppose there are three products: A, B and C, which respectively require 10kWh, 20kWh and 50kWh per unit to manufacture. Suppose also that there is a fixed overhead of 3,000kWh per week. These values would all be constant from week to week, but actual production throughputs would vary.

Take a week when production of A, B and C was 400, 500 and 600 units respectively. Expected consumption would then be given by multiplying the quantity produced by the kWh per unit:

Another way to view these situations is to work out an energy-weighted equivalent output. Five hundred units of Product B, for example, is the equivalent (in energy demand) of 1,000 units of product A (because its energy intensity is double). In the example shown here, the mixed production is the equivalent of 400 + (500 x 20/10) + (600 x 50/10) = 4,400 units of Product A. Reducing the week’s bundle of production to a single equivalent output in this way enables a straight-line characteristic to be drawn.

Activity Quantity in week kWh per unit Total kWh

Product A 400 10 4,000

Product B 500 20 10,000

Product C 600 50 30,000

Fixed demand 1 3,000 3,000

Grand total expected for the week Product A 47,000

Degree days

Heating degree days are a measure of the severity and duration of cold weather. The colder the weather in a given month, the larger the degree-day value for that month. They are, in essence, a summation over time of the difference between a reference or ‘base’ temperature and the outside temperature.

Further informationDegree days for energy management (CTG004)

Menu

http://www.carbontrust.co.uk/Publications/pages/publicationdetail.aspx?id=CTG004


2 Or other chosen interval.

Cusum analysis

Cusum analysis can help find the lowest sustainable position for the performance characteristic line. It can also be used to diagnose changes in performance.

The M&T spreadsheet or software will provide a tabulation of weekly consumption and a relevant driving factor from which to draw a scatter diagram chart. Figure 6 shows energy consumption on the vertical axis and driving factor values on the horizontal axis.

The next step is to superimpose a line of best fit on the scattered points. The spreadsheet will have a regression function that will do this. Note that the energy figures are the dependent variable or y-axis data, while the driving factor figures are the independent variable or x-axis data.

Key M&T charting techniques that users should be aware of are:

How to fix straight-line targets at the lowest •sustainable level

How they can help diagnose abnormal performance.•

In order to apply these techniques, users must be familiar with cusum analysis. This technique is relatively simple, but very effective. If the energy performance for a building or process is consistent every week, its actual consumption will be roughly equal to expected values (however calculated). In some weeks actual will exceed expected and in others it will be less, but in the long term the positive and negative variances cancel out and their cumulative sum (‘cusum’) will remain roughly constant. If, however, a problem occurs that causes persistent energy waste, even if the problem is minor, positive weekly variances will outweigh the negative and their cumulative sum will increase. The cusum chart would switch from a horizontal to a rising trend, as will be shown later in Figure 15.

Key charting techniquesIn addition to creating a straight-line target, it can also be important to know additional techniques when operating a monitoring and targeting scheme.

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200

Site E gas – 45 unit

Output (tonnes)

kWh


Figure 6 Energy consumption related to driving factors

Menu


The software will identify the intercept – the point where the characteristic crosses the vertical (energy) axis – and the slope of the line3. In the analysis below, the intercept is denoted by c and the slope by m.

This then provides enough information to construct the cusum calculation. There are six columns (including date stamps) in the following spreadsheet:

Note that the formulae in column D refer to c and m (the intercept and slope of the current performance characteristic line) stored in cells $B$2 and $B$3 respectively. These are constants, although possibly subject to amendment from time to time as the view of what is achievable changes.

Figure 8 An example of a cusum calculation

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200


Output (tonnes)

kWh


Column A B C D E F

Contains Input data Input data Input data Formula Formula Formula

Date of end of period

Quantity consumed (kWh)

Driving factor, such as tonnes or degree days

Expected consumption (m x col. C) + c

Difference between actual and expected (col. B – col. D)

Running total of column E

3 In MS Excel this is termed the x-coefficient.

Figure 7 Superimposing a line of best fit

Menu


250,000

200,000

150,000

100,000

50,000

0

27/0

2

33/0

2

39/0

2

45/0

2

51/0

2

5/03

11/0

3

17/0

3

23/0

3

30/0

3


Week

kWh

Co-plotted actual and expected

Figure 9 A time-series view of expected consumption

60,000

40,000

20,000

0

-20,000

-40,000

-60,000

27/0

2

33/0

2

39/0

2

45/0

2

51/0

2

5/03

11/0

3

17/0

3

23/0

3

30/0

3

Week

kWh

Site E gas – 45 unitDeviation from expected

Figure 10 Plotting the difference

Mathematically, a straight line is denoted by y = mx + c, where:

y is the dependent variable (in our case energy •used or expected to be used).

x is the driving factor or independent variable, •which could be degree days or production throughput, for example.

m is a constant indicating the slope of the line. •

in energy monitoring it shows how many extra •units of energy are used for each extra unit of driving factor.

c is a constant coefficient representing the •intercept on the y (energy) axis. It is the amount of energy predicted to be used when the driving factor is 0. This is sometimes called the base load or fixed load.

Figure 9 shows the time-series view of expected consumption – Column D, as a green continuous line – and actual consumption – Column B, plotted as points – together on the same graph. It provides a subjective view of how the two relate. It also shows any periods when they diverge.

Figure 10 is a plot of column E (the difference), and it shows more clearly the history of the variance. This chart is the equivalent of a control chart in the quality-monitoring technique statistical process control. It can have limits of expected or allowable variation superimposed, as illustrated by the broken blue line. It clearly shows whether or not the deviation between observed and expected consumption is within the normal behaviour of the process.

Figure 11 is the cusum chart. It plots the values from Column F. In the example, it can be seen that the cusum has both upward and downward sloping sections. Upward gradients signify a persistent tendency to use more energy than expected, and downward gradients show sustained better-than-expected performance.

250,000200,000150,000100,00050,000

0-50,000

-100,000-150,000-200,000-250,000

27/0

2

33/0

2

39/0

2

45/0

2

51/0

2

5/03

11/0

3

17/0

3

23/0

3

30/0

3


Week

kWh

Cumulative sum of deviations

Figure 11 An example of a cusum chart

Menu


Fixing a ‘tough but achievable’ target characteristic

In order to detect – and thus rectify – accidental energy waste, target performance characteristics need to stretch people without breaking them. Set the target too leniently, and opportunities will be missed because some problems will not show up as overspends. Set it too aggressively4, and everyone will become demoralised because they will be unable to avoid adverse reports.

The clue to best achievable performance is in the cusum chart. There may be periods when it slopes downwards for weeks at a time, indicating a persistent tendency to use less energy than predicted by the trial characteristic. These favourable periods present an opportunity for further analysis, as illustrated in Figure 12.

Note: Figure 12 is based on a real case where performance had changed from time to time. Often, it is not possible to discover the reasons for past adverse changes; however, this does not prevent the user identifying when the best performance occurred.

If we identify the same hand-picked data points on the scatter diagram (Figure 13), we see that they lie towards the lower edge of the scatter. This is no surprise, but until we drew the cusum we did not know that these low consumptions were consistently achievable. In general, just arbitrarily picking low points on a scatter diagram is not reliable.

250,000

200,000

150,000

100,000

50,000

0

-50,000

-100,000

-150,000

-200,000

-250,00027/02 33/02 39/02 45/02 51/02 5/03 11/03 17/03 23/03 30/03


Week

kWh


Figure 12 Identifying best achievable performance

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200


Site E gas – 45 unit production

kWh


Figure 13 Best achievable performance on the scatter diagram

4 Setting extreme targets is an alternative technique which may be applied in some circumstances, but such methods are outside the scope of this guide. The Carbon Trust can offer additional advice in this area. Call 0800 085 2005 for further help.

Menu


By fitting a new performance characteristic line through the points identified as representing best achievable performance, this becomes the new operational target (Figure 14).

Diagnosing adverse changes in performance

The cusum chart can help to diagnose long-term adverse performance. Figure 15 shows a cusum chart that has developed a rising gradient towards the end (indicating, as previously explained, an adverse change in the way the monitored plant uses energy).

This time, pick the points on the rising section of the cusum to select adverse weeks.

120,000

100,000

80,000

60,000

40,000

20,000

00 50 100 150 200 250 300

Site E rolls production

kWh

Site E gas – rolls etcPerformance characteristic line(s)

Figure 16 Comparing adverse characteristics with what is achievable

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200


kWh

Site E gas – 45 unitPerformance characteristic line(s)

Figure 14 Finding a new operational target

Figure 15 An example of a cusum chart with a rising gradient

200,000

150,000

100,000

50,000

0

-50,000

-100,00036/02 42/02 48/02 2/02 8/02 14/03 20/03 26/03 33/03 39/03

WeekkW

h

Site E gas – rolls etcCumulative sun of deviances

Menu


By looking at the same points on the scatter diagram, compare the adverse characteristic with what is achievable. This should enable someone with technical knowledge of the process to infer the nature of the fault (in this example, the symptom is fixed extra consumption unrelated to production throughput).

Furthermore, the analysis indicates the magnitude of the excess consumption. The timing was already shown on the cusum chart. This is a powerful set of evidence to aid in identifying the cause of the deviation and resulting excess cost and carbon emissions.

Persistent abnormal performance

The techniques described so far are designed to detect waste when performance changes, but they can sometimes provide clues about performance that is persistently poor and has a consumption characteristic that looks wrong.

Figure 17 shows the electricity demand pattern of a college campus (blue line) compared with what it should be in theory (red line). The weather-related demand was unexpected – it was traced to students using portable electric heaters in the halls of residence.

Figure 18, meanwhile, shows electricity demand in a log chipper feeding a pulp mill. The blue line shows that demand barely changes with throughput. The red line represents a more rational characteristic that the mill could achieve by improving motor control during idle periods.

Figure 17 Comparing actual consumption with theoretical demand

Figure 18 Identifying a more rational characteristic

70,000

60,000

50,000

40,000

30,000

10,000

20,000

00 10 20 30 40 50 60 70 80 90 100

College Q main electricity

Heating degree days

kWh


14,000

12,000

10,000

8,000

6,000

2,000

4,000

00 1000 2000 3000 4000 5000 6000 7000

Woodyard chip conveyor to pulp mill

kWh

Chipper powerPerformance characteristic line(s)

Further information

Advanced metering for SMes (CTC713)

Menu

http://www.carbontrust.co.uk/Publications/pages/publicationdetail.aspx?id=CTC713


Automatic meter readingAutomatic meter reading ensures bills are based on actual, rather than estimated consumption, and avoids the need for manual readings, which can be impractical and unreliable.

The increasing availability of automatic meter readings makes it possible to monitor energy consumption in detail. Suppliers are obliged to install half-hour meters at larger consuming sites. These ‘Code 5’ fiscal electricity meters in the UK record data at 30-minute intervals. Smaller sites are able to pay for half-hourly meters through independent meter providers. How often a reading is taken will depend on the supplier, although customers may request regular readings. If you have a half-hourly meter, ask your supplier or service provider for a regular copy of the data. They may even provide web tools to help you analyse your data.

This fine-grained data can be used both to visualise demand patterns and to assess energy performance. The examples that follow mainly discuss electricity, as that is the most common utility to be measured at short intervals.

Tax incentives

enhanced Capital Allowances (eCAs) enable businesses to buy energy efficient equipment using a 100% rate of tax allowance in the year of purchase. Businesses can claim this allowance on the investment value of energy efficient equipment, if it is on the energy Technology List. The procedure for claiming an eCA is the same as for any capital allowance. For further information please visit www.eca.gov.uk or call the Carbon Trust on 0800 085 2005.

Further information

Metering technology overview (CTV027)

Menu

www.eca.gov.uk

http://www.carbontrust.co.uk/Publications/pages/publicationdetail.aspx?id=CTV027


Visualisation

Most organisations have buildings with a pattern of occupation that is repeated from one week to the next, with a consistent profile for each day of the week.

Examining the daily profile can reveal anomalous performance. For example, in Figure 19 we see the 24-hour profile of electricity supply to a department store. The blue trace is Wednesday the 15th of August 2007. The purple and green traces are for the subsequent two days and show unusual high overnight consumption. The cause was found to be the air conditioning, which had been left running continuously after being maintained.

Quite a good way to visualise many weeks’ data is a false-colour contour plot. Figure 20 shows demand in a department store over the Christmas and New Year period. Many features can be picked out: the regularity of demand (late Thursday opening, for example); higher demand during the day in the run-up to Christmas; extended closures during the public holidays; and the fact that overnight demand is slightly higher before midnight than after. This latter feature is attributable to window display lighting, and the chart discloses that the lighting control fails or is overridden from time to time.

Both of these examples illustrate cases where a demand threshold was exceeded – quiet-hours demand in the first case, and daily peak demand in the second.

Setting automatic alarms against such limits is one way to detect waste, although it may result in numerous spurious alerts of relatively low financial value. Smarter ways of using fine-grained data and integrating them into the overspend league table, as examined next, can reduce spurious and trivial alerts.

In this false-colour contour plot, consumption rates are represented on a grid in which each column represents one day (midnight to midnight, top to bottom) and the days are arranged left to right. The cells of the grid are colour-coded to indicate the short-term power level. See Appendix A for suggestions as to generating these charts.

Figure 19 24-hour demand profiles

400

350

300

250

200

100

50

150

000:00 04:00 08:00 12:00 16:00 20:00

Retail store W electricity

Time of day

Po

wer

(kW

h)

16–Aug-07 abnormal nightime load


15–Aug-07 normal day

00:00

90-100

80-90

70-80

60-70

50-60

40-50

30-40

20-30

10-20

0-10

12:00

24:00

01/1

2/00

05 /1

2/00

09/1

2/00

13/1

2/00

17/1

2/00

21/1

2/00

25/1

2/00

29/1

2/00

02/1

2/01

0612

/01

1012

/01

1412

/01

1812

/01

Power percent

Figure 20 Store W electricity demand profile

Menu


Assessing performance with fine-grained data

To assess performance requires estimates of expected consumption. As mentioned earlier, energy uses with repetitive daily or weekly profiles can be assessed by reference to preceding periods (provided that unrepresentative days are excluded from the pool of precedents).

In theory, it would be possible to check every half-hour period against a precedent-based template profile, and to raise an alarm if the variance exceeds a certain limit. However, except in rare cases, this is not a practical strategy. False alarms may occur purely because demand is differently distributed during the day: the same (correct) quantity of energy could be used, but at different times, causing over-consumption in one interval to be balanced by under-consumption in another.

It is the cumulative deviation that is important, and this can be arrived at by measuring total actual consumption over a week, for example, and comparing it with what it would have been with the demand pattern represented by the target template profile.

Aggregating deviations over a week brings two benefits:

1. The ‘smoothing’ effect reduces spurious alerts.

2. The resulting estimate of excess consumption can be used in the overspend league table so that any detected problems can be assigned a sensible priority.

In practice, most managers say they prefer to deal with energy as a concentrated weekly task rather than being continually interrupted by alerts in real time, some of which may be insignificant or spurious. This is not to dismiss automatic meter reading as valueless: quite the opposite, since the stored fine-grained consumption data can be viewed, when required, as an aid to diagnosing abnormal performances.

The strategy of aggregating fine-grained data for weekly review, and only examining the detail as required, also accommodates those situations where demand patterns are not recurrent, notably in industrial processes. Here a form of activity-based targeting can be employed, using either physical sensors or a production scheduling database, to infer what the pattern of demand ought to be.

Figure 21 for example shows a pair of charts relating to a batch production plant in a factory. It compares the theoretical demand pattern (derived from a production scheduling database) with actual consumption.

00:00

1171101029588807366595144372922157

06:00

12:00

18:00

24:00M T T F S S M T T F S SWW M T T F S S M T T F S SWW

00:00

06:00

12:00

18:00

24:00

Theoretical pattern of demand Actual pattern of demand

kW

00:00

1171101029588807366595144372922157

06:00

12:00

18:00


00:00

06:00

12:00

18:00

24:00


kW

Figure 21 Batch production plant demand

Menu


As before, by aggregating a week’s worth of theoretical demand, one obtains an estimate of expected consumption that enables excess consumption to be included in the overspend league table.

Even more complex solutions can be employed when the risk of undetected loss is sufficiently high. In one particularly advanced application, process control data such as temperatures and flow rates are being collected from a set of distillation columns and analysed every 20 minutes (during steady-state operation) to calculate the theoretical heat demand from first principles.

The theoretical heat demand is added up over the course of a week and compared with metered steam consumption. Again, this is done using an overspend league table to detect and quantify any energy losses that may occur, for example as the plant’s internals begin to foul up or disintegrate.

To avoid data overload and possible spurious alerts, do not try to assess performance in real time. Have a weekly review, for example, and then drill down into the detail of any cases that require investigation. Use data visualisation techniques to help you and your colleagues find the causes of excess consumption.

Case StudyChemical works

A chemical works applied M&T to the steam consumption in some distillation columns, carrying out a weekly assessment of consumption against theoretical demand calculated from 20-minute-interval data on material flows and temperatures. It found a number of instances of avoidable waste, such as a case where the operators had bypassed a heat recovery unit and another where steam was leaking through a faulty trap.

In the latter case the fault was already known about, but its cost implications were not. The company also discovered (by chance) that it was possible to operate slightly outside the stipulated ‘reflux ratio’ and achieve higher thermal efficiency. Manual control was replaced with an automatic loop by reprogramming the plant control system, and the more efficient settings were adopted as the norm.

The factory’s process managers found the M&T scheme useful for comparing the relative thermal efficiencies of their distillation columns, enabling them to schedule production on a least-energy basis. It also gave them the ability to optimise maintenance shutdowns, as internal fouling and disintegration can now be detected through a deteriorating energy efficiency. They also welcomed the fact that operators no longer have to try to assimilate dozens of flow and temperature readouts, nor guess whether any deviations are significant or not; the overspend league table gives them a simple weekly summary of where, if anywhere, waste is occurring.

Menu


Automatic meter reading is beneficial when meters are inaccessible, too remote, or too numerous for manual reading to be an option. Meters with pulse outputs or serial communications interfaces are needed, along with additional components such as:

Data loggers (when the meter has no recording •capability of its own)

Data concentrators (when data from multiple logging •devices need to be marshalled)

Gateways (for passing data between networks using •different protocols or isolated by a firewall)

Software to interrogate the devices and record results •in a database.

Meter readings should be read in-house. It is best not to rely on invoice data, as suppliers are under no obligation to provide accurate readings. Nor do suppliers read private submeters, which are an important source of data. Furthermore, in-house meter readings can be scheduled at appropriate intervals (such as once a week). On half-hourly metered electricity supplies the data from the supplier will be more reliable, but unless the user makes other arrangements, the data will usually only be available monthly in arrears.

Implementing monitoring and targetingRolling out a monitoring and targeting scheme will require several factors to be in place, but once the scheme is up and running routine maintenance is simple.

In order to operate an M&T scheme effectively, the following three components must be in place:

1. Consumption data

2. Driving-factor data

3. Methods of calculating expected consumption.

Consumption data may come from meters (manually or automatically read), from delivery and stock-level figures, or from proxy measures such as run-hours counters or ammeters. The critical task is to ensure that data are synchronised as closely as possible with the required assessment intervals. Repeatability of measurements is more important than accuracy. Systematic bias in readings is of little consequence: it will affect the calculated consumption volume, but the unexpected change that caused it will still be evident.

Menu


The required driving-factor data will often include degree-day statistics, and according to the user’s circumstances may also comprise production figures, mileages, hours of darkness, or whatever other data helps to determine how much energy should have been used.

Two important points to note are that first, driving-factor data, like consumption data, need to be synchronised with the assessment intervals; and second, where production data are used, they should certainly be gross (rather than saleable) volumes, since it will often take as much energy to produce a unit of unsaleable scrap as a unit of good product.

Production volumes may need to be measured at intermediate points in the process in order to provide a meaningful guide to the energy requirements of different production stages. For example, in a paper-making machine, the paper goes through a drying stage after dewatering and pressing. The steam requirement for a tonne of discarded paper that gets as far as the reeler is the same as for a saleable tonne, but it is nil for a tonne of discarded paper that comes off at the wet end.

Finding appropriate indicators of production activity can be a challenge. Avoid complexity at the outset and try a simple approach – it may work. Refine the model later if necessary. Be pragmatic: if consumption X varies with driving factor Y, use that relationship as a targeting model even if X does not directly depend on Y. Unlike with rigorous statistical analysis, there does not need to be a causal link, merely that the relationship between X and Y should normally be consistent.

Having secured a supply of consumption and driving-factors data, the final task is to relate one to the other, establishing tough but achievable operational targets. These may be precedent-based or activity-based, using straight-line relationships or any other appropriate method of calculating expected consumption. However they are to be done, it is sensible to gather sufficient historical data and set some preliminary targets at the outset. Operational targets can and will be reviewed and revised as time goes by, and will not be widely publicised within the organisation, so it does not matter too much if they are not absolutely right first time.

Nor does it matter if initial coverage is incomplete. It is better to have the ability to detect waste in some areas than to remain in the dark until every single item is covered.

Further information


www.carbontrust.co.uk/mandt

energy management strategy (CTV022)

Practical energy management (CTV023)

Advanced metering for SMes (CTC713)Menu





http://www.carbontrust.co.uk/Publications/pages/publicationdetail.aspx?id=CTC713


Routine operation

Once the M&T scheme has operational targets set for at least some of the consumption streams, routine operation is very easy. At the chosen assessment interval (say once a week):

Acquire the necessary data (consumption •and driving-factors)

Display the overspend league table•

Check the data behind any significant overspends•

Ask for explanations.•

In some cases there will be a good explanation for the excess consumption. Otherwise, it is likely that there is some avoidable waste that could easily be remedied. Note that it will be easy to counter excuses (extreme weather, high production output, and so on) because most of these influences will already have been taken into account.

It may be unnecessary to do anything more by way of analysis. Conventional investigations can be carried out and remedial work can be put in hand (where economically justified) in the knowledge that failure will be exposed in future reports.

Only in a minority of cases – if at all – will it be necessary to carry out further analysis. However, fine-grained data may prove useful where available, and if the problem persists, it will ultimately become possible to diagnose the new consumption patterns using cusum-assisted regression analysis.

From time to time it will pay to look at the foot of the overspend league table, where all the ‘underspent’ streams accumulate. Persistent underspending suggests a target that is too lenient. Cusum analysis can then be used to identify when the improved performance started, and a new target can be set accordingly.

Choosing M&T software

Although there are many proprietary M&T software packages available, not all of them are designed for the purposes described in this guidance. Some software supplied with automatic meter reading hardware may do little more than display demand-profile charts, for example, while energy accounting software may lack the analytical and target-setting functions found in products with more of a bias towards physical performance.

To give the full benefits of waste-avoidance analysis, software needs to be able to support the principles and procedures explained in this guide. Appropriate products will have these hallmarks:

An emphasis on the assessment of physical •performance (as distinct from energy accounting, although they may offer this as well).

Facilities for setting up calculations of expected •consumption, at least using straight-line relationships with single driving factors and preferably other methods as well.

Generate the overspend league table as defined •above, either as a standard built-in report (preferred) or as a user-defined report template.

Provide scatter, deviation and cusum charts at least, •and employ them actively in the target-setting process by enabling selective analysis of hand-picked data points.

It is also possible to build a scheme in-house with spreadsheets. The overspend league table, in particular, is very easy to emulate, and all the required charts described earlier can be readily defined. The more advanced steps (such as selective regression analysis in particular) would be more of a challenge, but not beyond the more capable spreadsheet user. Examples of techniques, tips and tricks needed for energy analysis can be found on the web.

Menu


Step 3. See what systems are in place

Analyse what meters are currently on site and the data collection and recording techniques. Is enough data available from the current systems to properly assess the site? You may want more information than your current meters are giving you.

Step 4. Install sub-meters

Large users or multi-site organisations could benefit from installing sub-meters. Look at the strategies for sub-metering, and install based on the data needed.

Next stepsEffective M&T provides the basis for energy management. Follow the steps below to understand the energy consumption on the site and identify where more information is needed.

Step 1. understand the industry

Making sense of the complex industries of energy (electricity and gas) and water helps in understanding how energy is delivered to the site, how it is billed, and who has responsibility for supply and the hardware of utilities.

Step 2. Analyse energy bills

Invoices give a lot of information about supply, tariff and consumption. Understanding the information given on them can lead to large cost savings as well as provide details for your energy management strategy.

Further information



Menu




Activity-based targeting The process of estimating expected consumption volumes by reference to production throughput, prevailing weather and other driving factors. Comparison between expected and actual consumption reveals randomly-occurring accidental avoidable waste.

Advanced meteringSee Smart meter.

Assessment intervalThe period between exception reports; commonly weekly for industrial plants but often monthly for dispersed estates of buildings. Daily, per-shift and even ‘real-time’ assessments may be worthwhile in some circumstances. Returns of consumption and related driving factors must at least be synchronised with the assessment interval (although more frequent measurements may be of some diagnostic value).

Glossary

Base temperatureIn relation to the calculation of degree days, the assumed lowest outside air temperature at which a particular building can maintain comfort without artificial heating (or the highest temperature at which cooling is not required). In the UK a base temperature of 15.5ºC is common for heating assessment.

Benchmarking Identifying the things that enable good performance, either by critically comparing the performance of similar installations, or (if no comparable cases exist) identifying past periods of better performance.

Code 5Code 5 users are sites which already have electricity consumption monitored half-hourly. Sites with peak consumption exceeding 100kW for three consecutive months are classified as ‘Code 5’ and suppliers collect actual consumption data into the Balancing and Settlement Code for accurate billing. These accurate data are likely to be available for energy management purposes and should be requested from the energy supplier.

Consumption dataFigures, collected at equal intervals, representing the quantities of energy, water, or other consumable resources used. May be derived directly either from meters or from stock-level and delivery information; or may be computed by arithmetic combining two or more flows, or inferred from indirect measurements such as data-logged electric current.

Menu


Control limit A tolerance band on the deviation chart indicating the level of deviation which is considered to be significant.

Cumulative deviationSee Cusum

Cusum Cumulative sum of the difference between actual and target consumption in successive monitoring intervals. Usually presented as a chart whose most important property is that it manifests a horizontal trend as long as consumption remains close to target.

Degree days A measure of how cold (or hot) the weather has been (relative to a stated base temperature) measured over a regular monitoring interval, usually weekly or monthly.

Deviation chart A time-series chart showing the difference between actual and expected consumption in each successive monitoring period. May have control limits superimposed. Precursor to the cusum chart which shows the cumulative sum of deviation values.

Driving factor An independently measurable factor (e.g. production throughput, mileage, degree-day value) that determines the required quantity of energy or other consumable resource.

Expected consumption Theoretical quantity of energy, water, etc., against which actual consumption can be gauged. Can be calculated in various ways ranging from precedent (same period the year before) to rigorous mathematical modelling from first principles, but most commonly calculated using a simple empirical straight-line relationship between past consumption and corresponding values of an appropriate driving factor.

False-colour contour plotGraphical presentation of fine-grained (e.g, half-hourly) data in which consumption rates are represented on a grid in which, typically, each column represents one day midnight to midnight and the cells of the grid are colour-coded to indicate the short-term power level. Useful for visualising repetitive daily and weekly patterns. Sacrifices precision in the data but can display several months’ data on a single page.

Fixed demand The ‘base load’ consumption that is incurred regardless of prevailing weather, production output, etc.; as distinct from the variable component of demand.

Gross production The favoured measure of production activity in energy-intensive manufacturing, as distinct from net or saleable production. Reflects the fact that it may take as much energy to make unsaleable product as saleable. In thorough implementations, it may be necessary to record gross throughput at each significant stage in a process, to recognise the fact that product may either be diverted to scrap between stages, or else held in buffer storage.

Historical baseline The characteristic performance of a building, vehicle, or manufacturing process, when first assessed at the outset of an energy management programme.

League table A report consisting of a list of items ranked in order of significance, for example according to the gross quantity or cost of energy used. See especially overspend league table.

Menu


Precedent-based targeting Targeting method in which, usually, monthly consumption is gauged against the same month a year before. Simplistic because it assumes (a) that conditions were indeed comparable in the precedent month and (b) that no waste had occurred which would inflate the target for the period being assessed. Precedent-based targets can also be applied to half-hourly or other high-frequency data, usually by defining a profile ‘template’ on the basis of historical performance.

Quiescent threshold A simple exception-reporting method for high-frequency data in which out-of-hours consumption is monitored to ensure that it stays below some chosen level. Higher-than-expected consumption often indicates items left running unnecessarily.

Regression analysis A statistical technique for determining the constant and coefficients in a multi-variate targeting model of the form:

E = k0 + k1.D1 + k2.D2 + ... + kn.Dn

Overspend league table Key weekly (or daily, etc) reporting technique in which monitored streams of consumption are listed in descending order of their apparent unaccountable excess costs. Provides a rational view of where best – if anywhere – to direct investigations and remedial action. Conveniently accommodates any number of streams, whether of energy, water, or other resources, in a single concise summary that requires no specialist knowledge to produce or interpret.

Performance characteristic line A line, usually straight and diagonal, superimposed on the scatter diagram of consumption volume versus driving factor. Represents the idealised relationship between the two and enables the expected consumption to be estimated when the value of the driving factor is known. The performance characteristic line can be set to show the target, in which case it will occupy the lowest justifiable position on the chart; or the standard, in which case it represents current average performance and would be used for forward budget estimation; or the historical baseline representing average performance in the base year.

Limit, control Margin of error allowed in the estimation of expected consumption and used to indicate deviations from target that are significant compared with normal variability.

Moving annual total A method of reporting in which the most recent 12 months (or 52 weeks) of consumption are stated, regardless of the time of year. When applied to budget tracking, provides a more stable estimate of end-of-year out-turn than can be obtained by projecting from results for the year to date.

Norm chart A time-series chart in which actual consumption volumes are co-plotted with the corresponding expected values (inferred from the driving- factor values).

Operational target Values for expected consumption in each monitored stream, against which actual consumption can be compared for the purposes of detecting adverse changes in performance. Operational targets are based on a rational analysis of achievable performance and it is accepted that they may be refined at any time as more evidence is gathered.

Menu


Risk of undetected loss A formal method of evaluating the cost-effectiveness of expenditure on additional metering. Consumption is disaggregated according to where it is used, and differing percentage losses are assumed according to the nature of the application. Systems with low load factors are presumed to have more scope for undetected waste than those which have to operate continuously close to their maximum rating.

Scatter diagram An x-y plot of consumption versus driving factor, both measured at regular intervals (typically weekly or monthly).

Smart meterA meter with data logging and two-way communications, allowing, for example, data to be transmitted electronically to a meter operator; a supplier to disconnect a customer remotely; or costs to be displayed on the basis of real-time price information.

Specific energy ratioThe ratio between energy (etc) used and its presumed driving factor. An unreliable method of reporting, suitable for high-level management presentations but usually of little value for active management control.

Standard Current average performance characteristic (any wasteful use included). Usually used to project budgets by reference to future expected activity levels and weather.

Stream A measurable flow of energy, water, etc: typically that taken through an individual submeter but would also include consumptions arrived at by difference (between a main meter and its downstream submeters, say) or by adding two or more flows (such as the oil and gas used in a dual-fuel boiler). A stream need not necessarily be metered: it could be computed from changes in stock level, or estimated from a proxy measure such as hours run. Some practitioners treat driving factors (production, degree day histories, etc) as ‘streams’ as well.

SubmeterUsually any consumption meter downstream of a main supply meter; typically used to measure a branch flow to a particular building zone, or item of process equipment, etc.

Targeting modelMathematical procedure for calculating expected consumption from independent measurements of driving-factor data, from first principles, or on the basis of precedent, etc.

Variable component of demand That portion of demand that varies in direct proportion to the relevant driving factor, as distinct from the fixed (i.e. purely time-related) component.

X-Y scatter Graph in which a stream’s consumption is plotted against the relevant driving factor, say on a weekly basis, and usually with a straight line superimposed to represent the achievable target. A standard performance line or the historical baseline can also be superimposed, the former being used for budget projection.

Year to date An inferior method of reporting in which consumption etc. are reckoned from the start of the accounting year, discarding information from the year before.

Menu


Step 2

Highlight the data (including date and time labels) and from the Excel menu select Insert, Chart. Pick the ‘surface’ option and click Finish.

Step 1

Arrange the half-hourly values in a table, with one column per day.

Appendix A

Constructing a false-colour contour plot

A false-colour contour plot can be constructed

in Microsoft Excel by one of two methods. Advanced users can use macro code to colour the spreadsheet cells according to their relative values. Another method is to use Excel’s surface chart format as follows:

Menu


Step 3

The data will be plotted as a three-dimensional surface. To convert it to a contour map, right-click on the chart and choose 3-D view. Set both the rotation and elevation to 90º, and perspective to 0º.

Step 4

Finally, tidy up the display by editing the range-colours to give an orderly spectrum.

You may need to change the y-axis major interval to increase the number of steps and you will probably need to increase the size of the chart, add titles and so on.

Menu


Energy accountable centres

In larger organisations, energy accountable centres (EACs) can be set up so that managers can see and respond to the energy used within their respective jurisdictions. A monitoring and targeting scheme can be programmed to calculate the necessary EAC totals, which may include apportioning the consumption registered on shared meters.

However, in order to detect and diagnose waste effectively, it is preferable to monitor and target the whole consumption through each individual meter. It follows that EAC totals should only be used for summary reporting, and not for management control.

Appendix C

Performance indicators

Historically, it has been common to express the performance of industrial processes as specific energy ratios (SER), kWh per unit of output. A monitoring and targeting scheme can be programmed to calculate SERs, but they should be used with caution because they usually vary with product output (falling as throughput increases), and are affected by shifts in the balance between different product grades. They may also be affected by the weather. In other words, they cannot be relied upon as a measure of energy efficiency. They should therefore only be used for summary reports (if at all) and not for day-to-day management control.

The counterpart in buildings is the normalised performance indicator (NPI), the weather-adjusted kWh per square metre of floor area. These can be used to provide comparisons between buildings, or against published yardstick values, as a rough-and-ready benchmarking aid. A monitoring and targeting scheme can be programmed to calculate NPIs provided that appropriate data on building floor area, occupancy pattern and type of use are available.

The techniques described in this guide provide, in effect, a method of self-benchmarking to complement benchmarking on the basis of NPIs or SERs.

Appendix B

Menu


Carbon footprint calculator – Our online calculator will help you calculate your organisation’s carbon emissions.

www.carbontrust.co.uk/carboncalculator

Interest free loans – Energy Efficiency Loans from the Carbon Trust are a cost effective way to replace or upgrade your existing equipment with a more energy efficient version. See if you qualify.

www.carbontrust.co.uk/loans

Carbon surveys – We provide surveys to organisations with annual energy bills of more than £50,000*. Our carbon experts will visit your premises to identify energy saving opportunities and offer practical advice on how to achieve them.

www.carbontrust.co.uk/surveys

Action plans – Create action plans to implement carbon and energy saving measures.

www.carbontrust.co.uk/apt

Case studies – Our case studies show that it’s often easier and less expensive than you might think to bring about real change.

www.carbontrust.co.uk/casestudies

Events and workshops – The Carbon Trust offers a variety of events and workshops ranging from introductions to our services, to technical energy efficiency training, most of which are free.

www.carbontrust.co.uk/events

Publications – We have a library of free publications detailing energy saving techniques for a range of sectors and technologies.

www.carbontrust.co.uk/publications

Need further help? Call our Customer Centre on 0800 085 2005

Our Customer Centre provides free advice on what your organisation can do to save energy and save money. Our team handles questions ranging from straightforward requests for information, to in-depth technical queries about particular technologies.

* Subject to terms and conditions.

Go online to get moreThe Carbon Trust provides a range of tools, services and information to help you implement energy and carbon saving measures, no matter what your level of experience.

Menu

www.carbontrust.co.uk/carboncalculator

www.carbontrust.co.uk/loans

www.carbontrust.co.uk/surveys

www.carbontrust.co.uk/apt

www.carbontrust.co.uk/casestudies

www.carbontrust.co.uk/events

www.carbontrust.co.uk/publications

The Carbon Trust is a not-for-profit company with the mission to accelerate the move to a low carbon economy. We provide specialist support to business and the public sector to help cut carbon emissions, save energy and commercialise low carbon technologies. By stimulating low carbon action we contribute to key UK goals of lower carbon emissions, the development of low carbon businesses, increased energy security and associated jobs.

We help to cut carbon emissions now by:

providing specialist advice and finance to help organisations cut carbon•

setting standards for carbon reduction.•

We reduce potential future carbon emissions by:

opening markets for low carbon technologies•

leading industry collaborations to commercialise technologies•

investing in early-stage low carbon companies.•

www.carbontrust.co.uk 0800 085 2005

ACT ON CO2 is the Government’s initiative to help individuals understand and reduce their carbon footprint. Visit http://actonco2.direct.gov.uk for more information.

The Carbon Trust receives funding from Government including the Department of Energy and Climate Change, the Department for Transport, the Scottish Government, the Welsh Assembly Government and Invest Northern Ireland.

Whilst reasonable steps have been taken to ensure that the information contained within this publication is correct, the authors, the Carbon Trust, its agents, contractors and sub-contractors give no warranty and make no representation as to its accuracy and accept no liability for any errors or omissions. Any trademarks, service marks or logos used in this publication, and copyright in it, are the property of the Carbon Trust. Nothing in this publication shall be construed as granting any licence or right to use or reproduce any of the trademarks, service marks, logos, copyright or any proprietary information in any way without the Carbon Trust’s prior written permission. The Carbon Trust enforces infringements of its intellectual property rights to the full extent permitted by law.

The Carbon Trust is a company limited by guarantee and registered in England and Wales under Company number 4190230 with its Registered Office at: 6th Floor, 5 New Street Square, London EC4A 3BF.

Published in the UK: May 2010.

© Queen’s Printer and Controller of HMSO.

CTG008v2

In-depth management guide

Monitoring and targetingTechniques to help organisations control and manage their energy use


Preface

Reducing energy use makes perfect business sense; it savesmoney, enhances an organisation’s reputation and helps everyonein the fi ght against climate change.

The Carbon Trust provides simple, effective advice to help organisations take action to cut emissions. One of the simplestways to do this is to use energy more effi ciently.

This technology guide explores monitoring and targeting techniques, and shows how organisations can adopt an appropriate levelof monitoring and targeting in a way which will help them to saveenergy and cut costs.


xsada sa dsada•

Even for the more advanced aspects, an appreciation of the basic physical principles behind the use of energy at your place of work and a basic grasp of maths is all that is needed.

M&T is an extensive topic and this guide can only provide an introduction to the more complex analysis methods and tools that can be used.

Users should contact the Carbon Trust or an expert consultant for more detail or further advice on applying these techniques within their organisation.

2. Quantify the savings achieved by any and all of your energy projects and campaigns in a manner that accounts fully for variations in weather, levels of production activity and other external factors. Many users cite this as the most valuable result of M&T.

3. Identify fruitful lines of investigation for energy surveys. Rather than starting a survey with no clear agenda, you can go prepared with specifi c questions to ask, prompted by observed erratic or unexpected patterns of consumption.

4. Provide feedback for staff awareness, improve budget setting and undertake benchmarking.

This guide presents M&T from two perspectives. One is routine use (on a weekly cycle, for example). Routine M&T as explained here is quick and simple and requires no particular expertise on the part of the user. The other perspective is target-setting and diagnosis, an aspect that will appeal to users who wish to analyse data in more depth. Although this aspect needs to be addressed when fi rst setting up an M&T scheme, it then becomes optional once the system is up and running.

The purpose of monitoring and targeting (M&T) is to relate your energy consumption data to the weather, production fi gures or other measures in such a way that you get a better understanding of how energy is being used. In particular, it will identify if there are signs of avoidable waste or other opportunities to reduce consumption.

Data collection may be manual, automated, or a mixture of the two. Once an M&T scheme has been set up, its routine operation should be neither time-consuming nor complex. An M&T scheme will provide essential underpinning for your energy management activities, allowing you to:

1. Detect avoidable energy waste that might otherwise remain hidden. This is waste that occurs at random because of poor control, unexpected equipment faults or human error, and which can usually be put right quickly and cheaply (or, indeed, at no cost). Intercepting and rectifying such problems should more than cover the cost of operating the M&T scheme. See the next section for some examples.

What is monitoring and targeting?Monitoring and targeting is an energy management technique that can be applied in any type and size of organisation, whether commercial, industrial or public sector.


The meaning of target

Monitoring and targeting can be focused on invoice checking, contract tariff negotiation and fi nancial budgeting, or on the physical performance of the organisation’s buildings, processes and vehicles. It is the latter interpretation that we will use in this guide.

Often, targets are set without consideration of practical application or achievability. This guide focuses on achievable operational targets. An operational target:

Is calculated rationally to refl ect known achievable • performance

Is discussed and agreed•

Is applied to individual monitored streams • of consumption

Is continually amended to refl ect improvements•

Does not need to be widely disclosed•

Has performance reviewed monthly at most, • commonly every week, or even daily or per shift

Is benefi cial for day-to-day cost and • environmental management.

Hidden avoidable energy loss happens all the time, usually because of some minor control failure or the inappropriate action of occupants or maintenance staff. This guide will help you detect and remedy this waste, and disclose other opportunities to save energy and money.

Examples of waste and saving that have been identifi ed by employing M&T include:

Losses that would have equalled £3,500 a year • were incurred when a limit switch came loose at a waste-water treatment works, causing some machinery that should have been running intermittently to run continuously.

Gas consumption at a council depot doubled when • a maintenance contractor left the heating system running 24 hours a day.

The front steps of an offi ce block were found to be • costing £5,000 a year because the control had failed on the de-icing heaters (which the owners did not even know they had).

Frost-protection systems are a prime cause of waste. • One energy manager cut the electricity consumption of his HQ building by 40% when he found that all the electric frost protection pre-heaters were running on his air handling units.

Hundreds of pounds a year were lost when someone • left a bypass valve open on a steam trap in the basement of a paper mill.

Further information

Fact sheetsEnergy management (GIL136)

Understanding your energy consumption (CTL001)

Assessing the energy use at your industrial site (CTL002)

Assessing the energy use in your building (CTL003)


Figure 1 An example of an overspend league tableAn overspend league table is a simple and very effective report format to ensure instances of exceptional consumption are detected. It can be applied in any circumstances: by large and small users, for buildings, processes and vehicles, and at any chosen interval. It can be implemented in spreadsheets or by any good proprietary M&T software, and requires no special skills or knowledge to interpret it.

An overspend league table is simply a list of variances ranked in descending order of cost. ‘Variance’ is the difference between the actual quantity used and

the corresponding expected quantity. Methods for estimating expected consumption are covered later in this guide.

Reporting exceptionsOne of the key functions of a monitoring and targeting scheme is to alert the user to instances of exceptional excess consumption of energy.

Assessment intervals

This guide refers to weekly assessment intervals. Smaller businesses may prefer to operate M&T schemes at monthly intervals, while ambitious energy-intensive fi rms may opt for daily or even per-shift reporting.

Longer intervals mean more delay before exceptions are fl agged up, and at short intervals there is more risk of spurious alerts, even if all the data can be collected at the required frequency. Experience shows that weekly analysis is a good compromise, at least as a starting point.

The main point to bear in mind is that equal time intervals should be used.


This form of exception report has many advantages. First and foremost, the most signifi cant items are always at the top of the list, so it does not matter whether the report contains 10 items or 10,000. Second, it will be immediately evident how much remedial effort is warranted; indeed in some weeks there may be no problems that are suffi ciently costly to merit investigation. Third, as exceptions are judged on cost, non-energy commodities can be included in the same report. By including other utilities and consumables such as water, effl uent discharge volumes, chemicals and so on, the overspend league table can provide valuable benefi ts beyond energy savings alone.

The fi nal advantage of the overspend league table is that it requires no special expertise to interpret it. All the user needs to do is note which consumption streams appear signifi cantly overspent, verify the related input data, and if the problem seems genuine, raise the issue with whoever operates or understands the item in question.

Routine monitoring by means of an overspend league table is so simple that it should take no more than a few minutes a week – unless, of course, additional time is warranted by the occurrence of a signifi cantly costly problem. Using it is a robust procedure, since almost anyone can deputise for the principal user in his or her absence.

Although this guide goes on to describe various associated analytical and charting techniques, these can be thought of as optional. Once an M&T scheme has been set up, the overspend league table may be all that is needed for simpler examples.

By including other utilities and consumables, the overspend league table can provide valuable benefi ts beyond energy savings alone

Top tip

It is essential to ensure that variations in input data are not caused by faulty measuring equipment or sensors. Meters and sensors should be regularly maintained and calibrated.

Case studyA borough council

The energy manager at a borough council applied M&T on a monthly basis to all council buildings, using in-house manual meter readings. Some months after implementing the scheme he detected increased water consumption, and in some cases increased electricity consumption, at several of their public conveniences.

On investigation it transpired that the gents’ urinal fl ush controls had been replaced. New infrared motion sensors had been fi tted in place of the old hydraulic controls and were responding to homeless people sleeping in the toilets overnight. Increased electricity use was mainly due to the rough sleepers jamming the hand-dryer pushbuttons with matchsticks, although in one case the lavatory attendant was living in a store cupboard and had wired in an electric heater.


A chemical works applied M&T to the steam consumption in some distillation columns, carrying out a weekly assessment of consumption against theoretical demand calculated from 20-minute-interval data on material fl ows and temperatures. It found a number of instances of avoidable waste, such as a case where the operators had bypassed a heat recovery unit and another where steam was leaking through a faulty trap.

In the latter case the fault was already known about, but its cost implications were not. The company also discovered (by chance) that it was possible to operate slightly outside the stipulated ‘refl ux ratio’ and achieve higher thermal effi ciency. Manual control was replaced with an automatic loop by reprogramming the plant control system, and the more effi cient settings were adopted as the norm.

The factory’s process managers found the M&T scheme useful for comparing the relative thermal effi ciencies of their distillation columns, enabling them to schedule production on a least-energy basis. It also gave them the ability to optimise maintenance shutdowns, as internal fouling and disintegration can now be detected through a deteriorating energy effi ciency. They also welcomed the fact that operators no longer have to try to assimilate dozens of fl ow and temperature readouts, nor guess whether any deviations are signifi cant or not; the overspend league table gives them a simple weekly summary of where, if anywhere, waste is occurring.


Precedent-based targets, used with caution, may be the best method when consumption is seasonal, but unrelated to any measurable driving factor. Retail premises, for example, may exhibit peaks of consumption during the pre-Christmas period and New Year sales, and general electricity demand in schools, which otherwise might be steady through the year, should fall during holidays.

Generally, however, precedent-based targeting models can be too simplistic, and organisations may want to consider activity-based targeting. This is particularly appropriate when there are clear drivers for changing energy consumption, for example, changes in production throughput. See page 18 for examples of how data can be compared using automatic meter readings.

Precedent-based targeting

Precedent-based targeting models are most commonly used in monthly monitoring schemes, when expected consumption can be deduced from what was used in the corresponding month a year before. One weakness of this procedure is that it assumes that conditions (especially the weather) were comparable in the two months. This is not always the case, although in some circumstances the differences may not be too signifi cant.

A more problematic issue is what happens when energy waste has occurred. The resulting excessive consumption erroneously raises the expected quantity a year later. To work effectively, discount abnormal months to prevent them being used for target setting. At the other end of the scale, some automatic monitoring and targeting schemes attempt to compare consumption at very short intervals (e.g. half an hour), with a target template derived from previous similar days. The same caveat applies: abnormal consumption patterns must be fi ltered out from the pool of data used as precedents to make this effective.

Methods for calculating expected consumption fall into two categories. There are those based on precedent (comparison with previous periods), and activity-based methods that relate expected consumption to its driving factors (weather, production throughput, mileage, etc.). For the sake of brevity we will refer to any procedure for calculating expected consumption as a ‘targeting model’.

Estimating expected consumptionThe crux of the overspend league table – and of monitoring and targeting in general – is havinggood estimates of expected consumption for comparison with what is actually used.


1 Limited fi gures are available free of charge. Commercial data providers can be found on the web and it is also possible to calculate fi gures from local measurements.

Activity-based targeting

Activity-based targeting models calculate expected consumption by reference to its driving factors – the measurable things that cause consumption to vary. Examples of these are given in Table 2.

Table 1 Driving factors

Consumption driven by the weather

With very few exceptions, energy users of all kinds are dealing with weather-related energy consumption, usually for heating but increasingly also for cooling. Here, external air temperature is the dominant infl uence. For convenience, regional or local air temperature data can be converted into ‘degree-day’ values, which provide an index of how cold (or hot) the weather was in a given place in any particular week or month1.

Fuel consumption for space heating will normally exhibit a straight-line relationship to heating degree days (Figure 2), while electricity consumption should relate to cooling degree days (Figure 3) if any chilling plant exists on the metered circuit.

In either case, it is possible to identify a performance characteristic line which typifi es the relationship. For a given degree-day value, the characteristic line enables energy users (or, more usually, the M&T software or spreadsheet) to ‘read off’ the expected consumption (see Figure 4). This is explained more fully later.

Energy use Possible driving factor

Space heating Outside temperature

Air conditioning Outside temperature; possibly also humidity levels

Steam raising Quantity of steam produced

Production process

Production quantity

Exterior lighting Hours of darkness

Drying Quantity of water removed from product

3,500

3,000

2,500

2,000

1,500

1,000

500

00 20 40 60 80 100 120 140

Laboratory Electricity

Cooling degree days


kWh

Figure 2 Example of relationship between fuel consumption and heating degree days

1,800,000

1,600,000

1,400,000

1,200,000

1,000,000

800,000

600,000

400,000

200,000

00 10 20 30 40 50 60 70

Building 14 gas

Heating degree days


kWh

Figure 3 Example of electricity consumption related to cooling degree days


In Figure 4, the expected consumption where degree days = 28 is 1,250,000kW/h. The intercept (energy = 900,000kW/h where degree days = 0) shows the consumption where no heating is required. The values on the graph at this point could represent gas being used for production or hot water etc. The scatter points can indicate various things, including how well controlled the heating system is, but they can be infl uenced by other variables such as varying production levels in a factory.

Consumption driven by production throughput

In some (but by no means all) production processes, energy consumption can be related to production throughput using the same straight-line basis that was explained above. Figure 5 illustrates just such a case for gas used in a commercial bread oven.

If a simple straight-line relationship is not applicable, other methods can be used. However, as there are

operational advantages in being able to represent performance as a straight-line characteristic on a scatter diagram, methods that manipulate the data to make that possible are generally to be preferred. It is beyond the scope of this guide to deal with these in detail, but some hints are given below. Suffi ce to say that any appropriate method can be used to infer expected consumption from other independent measurements.

Figure 4 The performance characteristic line superimposed on the scattered data points

Figure 5 A performance characteristic line for a simple process

1,800,000

1,600,000

1,400,000

1,200,000

1,000,000

800,000

600,000

400,000

200,000

00 10 20 30 40 50 60 70

Building 14 gas

Heating degree days

Performance characteristic line(s)kW

h

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200

Process ovens – gas

Output (tonnes)

kWh

Note

The characteristic line will usually only be needed for computations within the normal range of data. The position of the intercept must be interpreted with caution if it is extrapolated well beyond this range. Where degree days are the driving factor, the intercept will also be sensitive to the choice of base temperature.


Alternatives to straight-line targets

Production processes are often more complex to set targets for. One common situation is where several products with different energy intensities are made on the same plant. Here, a tabular form of calculation is necessary to establish expected consumption.

Suppose there are three products: A, B and C, which respectively require 10kWh, 20kWh and 50kWh per unit to manufacture. Suppose also that there is a fi xed overhead of 3,000kWh per week. These values would all be constant from week to week, but actual production throughputs would vary.

Take a week when production of A, B and C was 400, 500 and 600 units respectively. Expected consumption would then be given by multiplying the quantity produced by the kWh per unit:

Another way to view these situations is to work out an energy-weighted equivalent output. Five hundred units of Product B, for example, is the equivalent (in energy demand) of 1,000 units of product A (because its energy intensity is double). In the example shown here, the mixed production is the equivalent of 400 + (500 x 20/10) + (600 x 50/10) = 4,400 units of Product A. Reducing the week’s bundle of production to a single equivalent output in this way enables a straight-line characteristic to be drawn.

Activity Quantity in week kWh per unit Total kWh

Product A 400 10 4,000

Product B 500 20 10,000

Product C 600 50 30,000

Fixed demand 1 3,000 3,000

Grand total expected for the week Product A 47,000

Degree days

Heating degree days are a measure of the severity and duration of cold weather. The colder the weather in a given month, the larger the degree-day value for that month. They are, in essence, a summation over time of the difference between a reference or ‘base’ temperature and the outside temperature.

Further informationDegree days for energy management (CTG004)


2 Or other chosen interval.

Cusum analysis

Cusum analysis can help fi nd the lowest sustainable position for the performance characteristic line. It can also be used to diagnose changes in performance.

The M&T spreadsheet or software will provide a tabulation of weekly consumption and a relevant driving factor from which to draw a scatter diagram chart. Figure 6 shows energy consumption on the vertical axis and driving factor values on the horizontal axis.

The next step is to superimpose a line of best fi t on the scattered points. The spreadsheet will have a regression function that will do this. Note that the energy fi gures are the dependent variable or y-axis data, while the driving factor fi gures are the independent variable or x-axis data.

Key M&T charting techniques that users should be aware of are:

How to fi x straight-line targets at the lowest • sustainable level

How they can help diagnose abnormal performance.•

In order to apply these techniques, users must be familiar with cusum analysis. This technique is relatively simple, but very effective. If the energy performance for a building or process is consistent every week, its actual consumption will be roughly equal to expected values (however calculated). In some weeks actual will exceed expected and in others it will be less, but in the long term the positive and negative variances cancel out and their cumulative sum (‘cusum’) will remain roughly constant. If, however, a problem occurs that causes persistent energy waste, even if the problem is minor, positive weekly variances will outweigh the negative and their cumulative sum will increase. The cusum chart would switch from a horizontal to a rising trend, as will be shown later in Figure 15.

Key charting techniquesIn addition to creating a straight-line target, it can also be important to know additionaltechniques when operating a monitoring and targeting scheme.

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200


Output (tonnes)

kWh


Figure 6 Energy consumption related to driving factors


The software will identify the intercept – the point where the characteristic crosses the vertical (energy) axis – and the slope of the line3. In the analysis below, the intercept is denoted by c and the slope by m.

This then provides enough information to construct the cusum calculation. There are six columns (including date stamps) in the following spreadsheet:

Note that the formulae in column D refer to c and m (the intercept and slope of the current performance characteristic line) stored in cells $B$2 and $B$3 respectively. These are constants, although possibly subject to amendment from time to time as the view of what is achievable changes.

Figure 8 An example of a cusum calculation

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200


Output (tonnes)

kWh


Column A B C D E F

Contains Input data Input data Input data Formula Formula Formula

Date of end of period

Quantity consumed (kWh)

Driving factor, such as tonnes or degree days

Expected consumption (m x col. C) + c

Difference between actual and expected (col. B – col. D)

Running total of column E

3 In MS Excel this is termed the x-coeffi cient.

Figure 7 Superimposing a line of best fi t


250,000

200,000

150,000

100,000

50,000

0

27/0

2

33/0

2

39/0

2

45/0

2

51/0

2

5/03

11/0

3

17/0

3

23/0

3

30/0

3


Week

kWh

Co-plotted actual and expected

Figure 9 A time-series view of expected consumption

60,000

40,000

20,000

0

-20,000

-40,000

-60,000

27/0

2

33/0

2

39/0

2

45/0

2

51/0

2

5/03

11/0

3

17/0

3

23/0

3

30/0

3

Week

kWh

Site E gas – 45 unitDeviation from expected

Figure 10 Plotting the difference

Mathematically, a straight line is denoted by y = mx + c, where:

y is the dependent variable (in our case energy • used or expected to be used).

x is the driving factor or independent variable, • which could be degree days or production throughput, for example.

m is a constant indicating the slope of the line. •

in energy monitoring it shows how many extra • units of energy are used for each extra unit of driving factor.

c is a constant coeffi cient representing the • intercept on the y (energy) axis. It is the amount of energy predicted to be used when the driving factor is 0. This is sometimes called the base load or fi xed load.

Figure 9 shows the time-series view of expected consumption – Column D, as a green continuous line – and actual consumption – Column B, plotted as points – together on the same graph. It provides a subjective view of how the two relate. It also shows any periods when they diverge.

Figure 10 is a plot of column E (the difference), and it shows more clearly the history of the variance. This chart is the equivalent of a control chart in the quality-monitoring technique statistical process control. It can have limits of expected or allowable variation superimposed, as illustrated by the broken blue line. It clearly shows whether or not the deviation between observed and expected consumption is within the normal behaviour of the process.

Figure 11 is the cusum chart. It plots the values from Column F. In the example, it can be seen that the cusum has both upward and downward sloping sections. Upward gradients signify a persistent tendency to use more energy than expected, and downward gradients show sustained better-than-expected performance.

250,000200,000150,000100,00050,000

0-50,000

-100,000-150,000-200,000-250,000

27/0

2

33/0

2

39/0

2

45/0

2

51/0

2

5/03

11/0

3

17/0

3

23/0

3

30/0

3


Week

kWh


Figure 11 An example of a cusum chart


Fixing a ‘tough but achievable’ target characteristic

In order to detect – and thus rectify – accidental energy waste, target performance characteristics need to stretch people without breaking them. Set the target too leniently, and opportunities will be missed because some problems will not show up as overspends. Set it too aggressively4, and everyone will become demoralised because they will be unable to avoid adverse reports.

The clue to best achievable performance is in the cusum chart. There may be periods when it slopes downwards for weeks at a time, indicating a persistent tendency to use less energy than predicted by the trial characteristic. These favourable periods present an opportunity for further analysis, as illustrated in Figure 12.

Note: Figure 12 is based on a real case where performance had changed from time to time. Often, it is not possible to discover the reasons for past adverse changes; however, this does not prevent the user identifying when the best performance occurred.

If we identify the same hand-picked data points on the scatter diagram (Figure 13), we see that they lie towards the lower edge of the scatter. This is no surprise, but until we drew the cusum we did not know that these low consumptions were consistently achievable. In general, just arbitrarily picking low points on a scatter diagram is not reliable.

250,000

200,000

150,000

100,000

50,000

0

-50,000

-100,000

-150,000

-200,000

-250,00027/02 33/02 39/02 45/02 51/02 5/03 11/03 17/03 23/03 30/03


Week

kWh


Figure 12 Identifying best achievable performance

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200



kWh


Figure 13 Best achievable performance on the scatter diagram

4 Setting extreme targets is an alternative technique which may be applied in some circumstances, but such methods are outside the scope of this guide. The Carbon Trust can offer additional advice in this area. Call 0800 085 2005 for further help.


By fi tting a new performance characteristic line through the points identifi ed as representing best achievable performance, this becomes the new operational target (Figure 14).

Diagnosing adverse changesin performance

The cusum chart can help to diagnose long-term adverse performance. Figure 15 shows a cusum chart that has developed a rising gradient towards the end (indicating, as previously explained, an adverse change in the way the monitored plant uses energy).

This time, pick the points on the rising section of the cusum to select adverse weeks.

120,000

100,000

80,000

60,000

40,000

20,000

00 50 100 150 200 250 300


kWh


Figure 16 Comparing adverse characteristics with what is achievable

250,000

200,000

150,000

100,000

50,000

00 200 400 600 800 1000 1200


kWh

Site E gas – 45 unitPerformance characteristic line(s)

Figure 14 Finding a new operational target

Figure 15 An example of a cusum chart with a rising gradient

200,000

150,000

100,000

50,000

0

-50,000

-100,00036/02 42/02 48/02 2/02 8/02 14/03 20/03 26/03 33/03 39/03

Week

kWh

Site E gas – rolls etcCumulative sun of deviances


By looking at the same points on the scatter diagram, compare the adverse characteristic with what is achievable. This should enable someone with technical knowledge of the process to infer the nature of the fault (in this example, the symptom is fi xed extra consumption unrelated to production throughput).

Furthermore, the analysis indicates the magnitude of the excess consumption. The timing was already shown on the cusum chart. This is a powerful set of evidence to aid in identifying the cause of the deviation and resulting excess cost and carbon emissions.

Persistent abnormal performance

The techniques described so far are designed to detect waste when performance changes, but they can sometimes provide clues about performance that is persistently poor and has a consumption characteristic that looks wrong.

Figure 17 shows the electricity demand pattern of a college campus (blue line) compared with what it should be in theory (red line). The weather-related demand was unexpected – it was traced to students using portable electric heaters in the halls of residence.

Figure 18, meanwhile, shows electricity demand in a log chipper feeding a pulp mill. The blue line shows that demand barely changes with throughput. The red line represents a more rational characteristic that the mill could achieve by improving motor control during idle periods.

Figure 17 Comparing actual consumption with theoretical demand

Figure 18 Identifying a more rational characteristic

70,000

60,000

50,000

40,000

30,000

10,000

20,000

00 10 20 30 40 50 60 70 80 90 100

College Q main electricity

Heating degree days

kWh


14,000

12,000

10,000

8,000

6,000

2,000

4,000

00 1000 2000 3000 4000 5000 6000 7000

Woodyard chip conveyor to pulp millkW

h

Chipper powerPerformance characteristic line(s)

Further information

Advanced metering for SMEs (CTC713)

120,000

100,000

80,000

60,000

40,000

20,000

00 50 100 150 200 250 300


kWh



Automatic meter readingAutomatic meter reading ensures bills are based on actual, rather than estimated consumption, and avoids the need for manual readings, which can be impractical and unreliable.

The increasing availability of automatic meter readings makes it possible to monitor energy consumption in detail. Suppliers are obliged to install half-hour meters at larger consuming sites. These ‘Code 5’ fi scal electricity meters in the UK record data at 30-minute intervals. Smaller sites are able to pay for half-hourly meters through independent meter providers. How often a reading is taken will depend on the supplier, although customers may request regular readings. If you have a half-hourly meter, ask your supplier or service provider for a regular copy of the data. They may even provide web tools to help you analyse your data.

This fi ne-grained data can be used both to visualise demand patterns and to assess energy performance. The examples that follow mainly discuss electricity,as that is the most common utility to be measured at short intervals.

Tax incentives

Enhanced Capital Allowances (ECAs) enable businesses to buy energy effi cient equipment using a 100% rate of tax allowance in the year of purchase. Businesses can claim this allowance on the investment value of energy effi cient equipment, if it is on the Energy Technology List. The procedure for claiming an ECA is the same as for any capital allowance. For further information please visit www.eca.gov.uk or call the Carbon Trust on 0800 085 2005.

Further information



Visualisation

Most organisations have buildings with a pattern of occupation that is repeated from one week to the next, with a consistent profi le for each day of the week.

Examining the daily profi le can reveal anomalous performance. For example, in Figure 19 we see the 24-hour profi le of electricity supply to a department store. The blue trace is Wednesday the 15th of August 2007. The purple and green traces are for the subsequent two days and show unusual high overnight consumption. The cause was found to be the air conditioning, which had been left running continuously after being maintained.

Quite a good way to visualise many weeks’ data is a false-colour contour plot. Figure 20 shows demand in a department store over the Christmas and New Year period. Many features can be picked out: the regularity of demand (late Thursday opening, for example); higher demand during the day in the run-up to Christmas; extended closures during the public holidays; and the fact that overnight demand is slightly higher before midnight than after. This latter feature is attributable to window display lighting, and the chart discloses that the lighting control fails or is overridden from time to time.

Both of these examples illustrate cases where a demand threshold was exceeded – quiet-hours demand in the fi rst case, and daily peak demand in the second.

Setting automatic alarms against such limits is one way to detect waste, although it may result in numerous spurious alerts of relatively low fi nancial value. Smarter ways of using fi ne-grained data and integrating them into the overspend league table, as examined next, can reduce spurious and trivial alerts.

In this false-colour contour plot, consumption rates are represented on a grid in which each column represents one day (midnight to midnight, top to bottom) and the days are arranged left to right. The cells of the grid are colour-coded to indicate the short-term power level. See Appendix A for suggestions as to generating these charts.

Figure 19 24-hour demand profi les

400

350

300

250

200

100

50

150

000:00 04:00 08:00 12:00 16:00 20:00

Retail store W electricity

Time of day

Po

wer

(kW

h)



15–Aug-07 normal day

00:00

90-100

80-90

70-80

60-70

50-60

40-50

30-40

20-30

10-20

0-10

12:00

24:00

01/1

2/00

05 /1

2/00

09/1

2/00

13/1

2/00

17/1

2/00

21/1

2/00

25/1

2/00

29/1

2/00

02/1

2/01

0612

/01

1012

/01

1412

/01

1812

/01

Power percent

Figure 20 Store W electricity demand profi le


Assessing performance with fi ne-grained data

To assess performance requires estimates of expected consumption. As mentioned earlier, energy uses with repetitive daily or weekly profi les can be assessed by reference to preceding periods (provided that unrepresentative days are excluded from the pool of precedents).

In theory, it would be possible to check every half-hour period against a precedent-based template profi le, and to raise an alarm if the variance exceeds a certain limit. However, except in rare cases, this is not a practical strategy. False alarms may occur purely because demand is differently distributed during the day: the same (correct) quantity of energy could be used, but at different times, causing over-consumption in one interval to be balanced by under-consumption in another.

It is the cumulative deviation that is important, and this can be arrived at by measuring total actual consumption over a week, for example, and comparing it with what it would have been with the demand pattern represented by the target template profi le.

Aggregating deviations over a week brings two benefi ts:

1. The ‘smoothing’ effect reduces spurious alerts.

2. The resulting estimate of excess consumption can be used in the overspend league table so that any detected problems can be assigned a sensible priority.

In practice, most managers say they prefer to deal with energy as a concentrated weekly task rather than being continually interrupted by alerts in real time, some of which may be insignifi cant or spurious. This is not to dismiss automatic meter reading as valueless: quite the opposite, since the stored fi ne-grained consumption data can be viewed, when required, as an aid to diagnosing abnormal performances.

The strategy of aggregating fi ne-grained data for weekly review, and only examining the detail as required, also accommodates those situations where demand patterns are not recurrent, notably in industrial processes. Here a form of activity-based targeting can be employed, using either physical sensors or a production scheduling database, to infer what the pattern of demand ought to be.

Figure 21 for example shows a pair of charts relating to a batch production plant in a factory. It compares the theoretical demand pattern (derived from a production scheduling database) with actual consumption.

00:00

1171101029588807366595144372922157

06:00

12:00

18:00


00:00

06:00

12:00

18:00

24:00


kW

00:00

1171101029588807366595144372922157

06:00

12:00

18:00


00:00

06:00

12:00

18:00

24:00


kW

Figure 21 Batch production plant demand


As before, by aggregating a week’s worth of theoreticaldemand, one obtains an estimate of expected consumption that enables excess consumption to be included in the overspend league table.

Even more complex solutions can be employed when the risk of undetected loss is suffi ciently high. In one particularly advanced application, process control data such as temperatures and fl ow rates are being collected from a set of distillation columns and analysed every 20 minutes (during steady-state operation) to calculate the theoretical heat demand from fi rst principles.

The theoretical heat demand is added up over the course of a week and compared with metered steam consumption. Again, this is done using an overspend league table to detect and quantify any energy losses that may occur, for example as the plant’s internals begin to foul up or disintegrate.

To avoid data overload and possible spurious alerts, do not try to assess performance in real time. Have a weekly review, for example, and then drill down into the detail of any cases that require investigation. Use data visualisation techniques to help you and your colleagues fi nd the causes of excess consumption.


A chemical works applied M&T to the steam consumption in some distillation columns, carrying out a weekly assessment of consumption against theoretical demand calculated from 20-minute-interval data on material fl ows and temperatures. It found a number of instances of avoidable waste, such as a case where the operators had bypassed a heat recovery unit and another where steam was leaking through a faulty trap.

In the latter case the fault was already known about, but its cost implications were not. The company also discovered (by chance) that it was possible to operate slightly outside the stipulated ‘refl ux ratio’ and achieve higher thermal effi ciency. Manual control was replaced with an automatic loop by reprogramming the plant control system, and the more effi cient settings were adopted as the norm.

The factory’s process managers found the M&T scheme useful for comparing the relative thermal effi ciencies of their distillation columns, enabling them to schedule production on a least-energy basis. It also gave them the ability to optimise maintenance shutdowns, as internal fouling and disintegration can now be detected through a deteriorating energy effi ciency. They also welcomed the fact that operators no longer have to try to assimilate dozens of fl ow and temperature readouts, nor guess whether any deviations are signifi cant or not; the overspend league table gives them a simple weekly summary of where, if anywhere, waste is occurring.


Automatic meter reading is benefi cial when meters are inaccessible, too remote, or too numerous for manual reading to be an option. Meters with pulse outputs or serial communications interfaces are needed, along with additional components such as:

Data loggers (when the meter has no recording • capability of its own)

Data concentrators (when data from multiple logging • devices need to be marshalled)

Gateways (for passing data between networks using • different protocols or isolated by a fi rewall)

Software to interrogate the devices and record results • in a database.

Meter readings should be read in-house. It is best not to rely on invoice data, as suppliers are under no obligation to provide accurate readings. Nor do suppliers read private submeters, which are an important source of data. Furthermore, in-house meter readings can be scheduled at appropriate intervals (such as once a week). On half-hourly metered electricity supplies the data from the supplier will be more reliable, but unless the user makes other arrangements, the data will usually only be available monthly in arrears.

Implementing monitoring and targetingRolling out a monitoring and targeting scheme will require several factors to be in place, but once the scheme is up and running routine maintenance is simple.

In order to operate an M&T scheme effectively, the following three components must be in place:

1. Consumption data

2. Driving-factor data

3. Methods of calculating expected consumption.

Consumption data may come from meters (manually or automatically read), from delivery and stock-level fi gures, or from proxy measures such as run-hours counters or ammeters. The critical task is to ensure that data are synchronised as closely as possible with the required assessment intervals. Repeatability of measurements is more important than accuracy. Systematic bias in readings is of little consequence: it will affect the calculated consumption volume, but the unexpected change that caused it will still be evident.


The required driving-factor data will often include degree-day statistics, and according to the user’s circumstances may also comprise production fi gures, mileages, hours of darkness, or whatever other data helps to determine how much energy should have been used.

Two important points to note are that fi rst, driving-factor data, like consumption data, need to be synchronised with the assessment intervals; and second, where production data are used, they should certainly be gross (rather than saleable) volumes, since it will often take as much energy to produce a unit of unsaleable scrap as a unit of good product.

Production volumes may need to be measured at intermediate points in the process in order to provide a meaningful guide to the energy requirements of different production stages. For example, in a paper-making machine, the paper goes through a drying stage after dewatering and pressing. The steam requirement for a tonne of discarded paper that gets as far as the reeler is the same as for a saleable tonne, but it is nil for a tonne of discarded paper that comes off at the wet end.

Finding appropriate indicators of production activity can be a challenge. Avoid complexity at the outset and try a simple approach – it may work. Refi ne the model later if necessary. Be pragmatic: if consumption X varies with driving factor Y, use that relationship as a targeting model even if X does not directly depend on Y. Unlike with rigorous statistical analysis, there does not need to be a causal link, merely that the relationship between X and Y should normally be consistent.

Having secured a supply of consumption and driving-factors data, the fi nal task is to relate one to the other, establishing tough but achievable operational targets. These may be precedent-based or activity-based, using straight-line relationships or any other appropriate method of calculating expected consumption. However they are to be done, it is sensible to gather suffi cient historical data and set some preliminary targets at the outset. Operational targets can and will be reviewed and revised as time goes by, and will not be widely publicised within the organisation, so it does not matter too much if they are not absolutely right fi rst time.

Nor does it matter if initial coverage is incomplete. It is better to have the ability to detect waste in some areas than to remain in the dark until every single item is covered.

Further information



Energy management strategy (CTV022)

Practical energy management (CTV023)

Advanced metering for SMEs (CTC713)


Routine operation

Once the M&T scheme has operational targets set for at least some of the consumption streams, routine operation is very easy. At the chosen assessment interval (say once a week):

Acquire the necessary data (consumption • and driving-factors)

Display the overspend league table•

Check the data behind any signifi cant overspends•

Ask for explanations.•

In some cases there will be a good explanation for the excess consumption. Otherwise, it is likely that there is some avoidable waste that could easily be remedied. Note that it will be easy to counter excuses (extreme weather, high production output, and so on) because most of these infl uences will already have been taken into account.

It may be unnecessary to do anything more by way of analysis. Conventional investigations can be carried out and remedial work can be put in hand (where economically justifi ed) in the knowledge that failure will be exposed in future reports.

Only in a minority of cases – if at all – will it be necessary to carry out further analysis. However, fi ne-grained data may prove useful where available, and if the problem persists, it will ultimately become possible to diagnose the new consumption patterns using cusum-assisted regression analysis.

From time to time it will pay to look at the foot of the overspend league table, where all the ‘underspent’ streams accumulate. Persistent underspending suggests a target that is too lenient. Cusum analysis can then be used to identify when the improved performance started, and a new target can be set accordingly.

Choosing M&T software

Although there are many proprietary M&T software packages available, not all of them are designed for the purposes described in this guidance. Some software supplied with automatic meter reading hardware may do little more than display demand-profi le charts, for example, while energy accounting software may lack the analytical and target-setting functions found in products with more of a bias towards physical performance.

To give the full benefi ts of waste-avoidance analysis, software needs to be able to support the principles and procedures explained in this guide. Appropriate products will have these hallmarks:

An emphasis on the assessment of physical • performance (as distinct from energy accounting, although they may offer this as well).

Facilities for setting up calculations of expected • consumption, at least using straight-line relationships with single driving factors and preferably other methods as well.

Generate the overspend league table as defi ned • above, either as a standard built-in report (preferred) or as a user-defi ned report template.

Provide scatter, deviation and cusum charts at least, • and employ them actively in the target-setting process by enabling selective analysis of hand-picked data points.

It is also possible to build a scheme in-house with spreadsheets. The overspend league table, in particular, is very easy to emulate, and all the required charts described earlier can be readily defi ned. The more advanced steps (such as selective regression analysis in particular) would be more of a challenge, but not beyond the more capable spreadsheet user. Examples of techniques, tips and tricks needed for energy analysis can be found on the web.


Step 3. See what systems are in place

Analyse what meters are currently on site and the data collection and recording techniques. Is enough data available from the current systems to properly assess the site? You may want more information than your current meters are giving you.

Step 4. Install sub-meters

Large users or multi-site organisations could benefi t from installing sub-meters. Look at the strategies for sub-metering, and install based on the data needed.

Next stepsEffective M&T provides the basis for energy management. Follow the steps below to understand the energy consumption on the site and identify where more information is needed.

Step 1. Understand the industry

Making sense of the complex industries of energy (electricity and gas) and water helps in understanding how energy is delivered to the site, how it is billed, and who has responsibility for supply and the hardware of utilities.

Step 2. Analyse energy bills

Invoices give a lot of information about supply, tariff and consumption. Understanding the information given on them can lead to large cost savings as well as provide details for your energy management strategy.

Further information




Activity-based targeting The process of estimating expected consumption volumes by reference to production throughput, prevailing weather and other driving factors. Comparison between expected and actual consumption reveals randomly-occurring accidental avoidable waste.

Advanced meteringSee Smart meter.

Assessment intervalThe period between exception reports; commonly weekly for industrial plants but often monthly for dispersed estates of buildings. Daily, per-shift and even ‘real-time’ assessments may be worthwhile in some circumstances. Returns of consumption and related driving factors must at least be synchronised with the assessment interval (although more frequent measurements may be of some diagnostic value).

Glossary

Base temperatureIn relation to the calculation of degree days, the assumed lowest outside air temperature at which a particular building can maintain comfort without artifi cial heating (or the highest temperature at which cooling is not required). In the UK a base temperature of 15.5ºC is common for heating assessment.

Benchmarking Identifying the things that enable good performance, either by critically comparing the performance of similar installations, or (if no comparable cases exist) identifying past periods of better performance.

Code 5Code 5 users are sites which already have electricity consumption monitored half-hourly. Sites with peak consumption exceeding 100kW for three consecutive months are classifi ed as ‘Code 5’ and suppliers collect actual consumption data into the Balancing and Settlement Code for accurate billing. These accurate data are likely to be available for energy management purposes and should be requested from the energy supplier.

Consumption dataFigures, collected at equal intervals, representing the quantities of energy, water, or other consumable resources used. May be derived directly either from meters or from stock-level and delivery information; or may be computed by arithmetic combining two or more fl ows, or inferred from indirect measurements such as data-logged electric current.


Control limit A tolerance band on the deviation chart indicating the level of deviation which is considered to be signifi cant.

Cumulative deviationSee Cusum

Cusum Cumulative sum of the difference between actual and target consumption in successive monitoring intervals. Usually presented as a chart whose most important property is that it manifests a horizontal trend as long as consumption remains close to target.

Degree days A measure of how cold (or hot) the weather has been (relative to a stated base temperature) measured over a regular monitoring interval, usually weekly or monthly.

Deviation chart A time-series chart showing the difference between actual and expected consumption in each successive monitoring period. May have control limits superimposed. Precursor to the cusum chart which shows the cumulative sum of deviation values.

Driving factor An independently measurable factor (e.g. production throughput, mileage, degree-day value) that determines the required quantity of energy or other consumable resource.

Expected consumption Theoretical quantity of energy, water, etc., against which actual consumption can be gauged. Can be calculated in various ways ranging from precedent (same period the year before) to rigorous mathematical modelling from fi rst principles, but most commonly calculated using a simple empirical straight-line relationship between past consumption and corresponding values of an appropriate driving factor.

False-colour contour plotGraphical presentation of fi ne-grained (e.g, half-hourly) data in which consumption rates are represented on a grid in which, typically, each column represents one day midnight to midnight and the cells of the grid are colour-coded to indicate the short-term power level. Useful for visualising repetitive daily and weekly patterns. Sacrifi ces precision in the data but can display several months’ data on a single page.

Fixed demand The ‘base load’ consumption that is incurred regardless of prevailing weather, production output, etc.; as distinct from the variable component of demand.

Gross production The favoured measure of production activity in energy-intensive manufacturing, as distinct from net or saleable production. Refl ects the fact that it may take as much energy to make unsaleable product as saleable. In thorough implementations, it may be necessary to record gross throughput at each signifi cant stage in a process, to recognise the fact that product may either be diverted to scrap between stages, or else held in buffer storage.

Historical baseline The characteristic performance of a building, vehicle, or manufacturing process, when fi rst assessed at the outset of an energy management programme.

League table A report consisting of a list of items ranked in order of signifi cance, for example according to the gross quantity or cost of energy used. See especially overspend league table.


Precedent-based targeting Targeting method in which, usually, monthly consumption is gauged against the same month a year before. Simplistic because it assumes (a) that conditions were indeed comparable in the precedent month and (b) that no waste had occurred which would infl ate the target for the period being assessed. Precedent-based targets can also be applied to half-hourly or other high-frequency data, usually by defi ning a profi le ‘template’ on the basis of historical performance.

Quiescent threshold A simple exception-reporting method for high-frequency data in which out-of-hours consumption is monitored to ensure that it stays below some chosen level. Higher-than-expected consumption often indicates items left running unnecessarily.

Regression analysis A statistical technique for determining the constant and coeffi cients in a multi-variate targeting model of the form:

E = k0 + k1.D1 + k2.D2 + ... + kn.Dn

Overspend league table Key weekly (or daily, etc) reporting technique in which monitored streams of consumption are listed in descending order of their apparent unaccountable excess costs. Provides a rational view of where best – if anywhere – to direct investigations and remedial action. Conveniently accommodates any number of streams, whether of energy, water, or other resources, in a single concise summary that requires no specialist knowledge to produce or interpret.

Performance characteristic line A line, usually straight and diagonal, superimposed on the scatter diagram of consumption volume versus driving factor. Represents the idealised relationship between the two and enables the expected consumption to be estimated when the value of the driving factor is known. The performance characteristic line can be set to show the target, in which case it will occupy the lowest justifi able position on the chart; or the standard, in which case it represents current average performance and would be used for forward budget estimation; or the historical baseline representing average performance in the base year.

Limit, control Margin of error allowed in the estimation of expected consumption and used to indicate deviations from target that are signifi cant compared with normal variability.

Moving annual total A method of reporting in which the most recent 12 months (or 52 weeks) of consumption are stated, regardless of the time of year. When applied to budget tracking, provides a more stable estimate of end-of-year out-turn than can be obtained by projecting from results for the year to date.

Norm chart A time-series chart in which actual consumption volumes are co-plotted with the corresponding expected values (inferred from the driving-factor values).

Operational target Values for expected consumption in each monitored stream, against which actual consumption can be compared for the purposes of detecting adverse changes in performance. Operational targets are based on a rational analysis of achievable performance and it is accepted that they may be refi ned at any time as more evidence is gathered.


Risk of undetected loss A formal method of evaluating the cost-effectiveness of expenditure on additional metering. Consumption is disaggregated according to where it is used, and differing percentage losses are assumed according to the nature of the application. Systems with low load factors are presumed to have more scope for undetected waste than those which have to operate continuously close to their maximum rating.

Scatter diagram An x-y plot of consumption versus driving factor, both measured at regular intervals (typically weekly or monthly).

Smart meterA meter with data logging and two-way communications, allowing, for example, data to be transmitted electronically to a meter operator; a supplier to disconnect a customer remotely; or costs to be displayed on the basis of real-time price information.

Specifi c energy ratioThe ratio between energy (etc) used and its presumed driving factor. An unreliable method of reporting, suitable for high-level management presentations but usually of little value for active management control.

Standard Current average performance characteristic (any wasteful use included). Usually used to project budgets by reference to future expected activity levels and weather.

Stream A measurable fl ow of energy, water, etc: typically that taken through an individual submeter but would also include consumptions arrived at by difference (between a main meter and its downstream submeters, say) or by adding two or more fl ows (such as the oil and gas used in a dual-fuel boiler). A stream need not necessarily be metered: it could be computed from changes in stock level, or estimated from a proxy measure such as hours run. Some practitioners treat driving factors (production, degree day histories, etc) as ‘streams’ as well.

SubmeterUsually any consumption meter downstream of a main supply meter; typically used to measure a branch fl ow to a particular building zone, or item of process equipment, etc.

Targeting modelMathematical procedure for calculating expected consumption from independent measurements of driving-factor data, from fi rst principles, or on the basis of precedent, etc.

Variable component of demand That portion of demand that varies in direct proportion to the relevant driving factor, as distinct from the fi xed (i.e. purely time-related) component.

X-Y scatter Graph in which a stream’s consumption is plotted against the relevant driving factor, say on a weekly basis, and usually with a straight line superimposed to represent the achievable target. A standard performance line or the v can also be superimposed, the former being used for budget projection.

Year to date An inferior method of reporting in which consumption etc. are reckoned from the start of the accounting year, discarding information from the year before.


Step 2

Highlight the data (including date and time labels) and from the Excel menu select Insert, Chart. Pick the ‘surface’ option and click Finish.

Step 1

Arrange the half-hourly values in a table, with one column per day.

Appendix A

Constructing a false-colour contour plot

A false-colour contour plot can be constructed

in Microsoft Excel by one of two methods. Advanced users can use macro code to colour the spreadsheet cells according to their relative values. Another method is to use Excel’s surface chart format as follows:


Step 3

The data will be plotted as a three-dimensional surface. To convert it to a contour map, right-click on the chart and choose 3-D view. Set both the rotation and elevation to 90º, and perspective to 0º.

Step 4

Finally, tidy up the display by editing the range-colours to give an orderly spectrum.

You may need to change the y-axis major interval to increase the number of steps and you will probably need to increase the size of the chart, add titles and so on.


Energy accountable centres

In larger organisations, energy accountable centres (EACs) can be set up so that managers can see and respond to the energy used within their respective jurisdictions. A monitoring and targeting scheme can be programmed to calculate the necessary EAC totals, which may include apportioning the consumption registered on shared meters.

However, in order to detect and diagnose waste effectively, it is preferable to monitor and target the whole consumption through each individual meter. It follows that EAC totals should only be used for summary reporting, and not for management control.

Appendix C

Performance indicators

Historically, it has been common to express the performance of industrial processes as specifi c energy ratios (SER), kWh per unit of output. A monitoring and targeting scheme can be programmed to calculate SERs, but they should be used with caution because they usually vary with product output (falling as throughput increases), and are affected by shifts in the balance between different product grades. They may also be affected by the weather. In other words, they cannot be relied upon as a measure of energy effi ciency. They should therefore only be used for summary reports (if at all) and not for day-to-day management control.

The counterpart in buildings is the normalised performance indicator (NPI), the weather-adjusted kWh per square metre of fl oor area. These can be used to provide comparisons between buildings, or against published yardstick values, as a rough-and-ready benchmarking aid. A monitoring and targeting scheme can be programmed to calculate NPIs provided that appropriate data on building fl oor area, occupancy pattern and type of use are available.

The techniques described in this guide provide, in effect, a method of self-benchmarking to complement benchmarking on the basis of NPIs or SERs.

Appendix B

The Carbon Trust is a not-for-profi t company with the mission to accelerate the move to a low carbon economy. We provide specialist support to business and the public sector to help cut carbon emissions, save energy and commercialise low carbon technologies. By stimulating low carbon action we contribute to key UK goals of lower carbon emissions, the development of low carbon businesses, increased energy security and associated jobs.

We help to cut carbon emissions now by:

providing specialist advice and fi nance to help organisations cut carbon•

setting standards for carbon reduction.•

We reduce potential future carbon emissions by:

opening markets for low carbon technologies•

leading industry collaborations to commercialise technologies•

investing in early-stage low carbon companies.•

www.carbontrust.co.uk0800 085 2005

ACT ON CO2 is the Government’s initiative to help individuals understand and reduce their carbon footprint. Visit http://actonco2.direct.gov.uk for more information.

The Carbon Trust receives funding from Government including the Department of Energy and Climate Change, the Department for Transport, the Scottish Government, the Welsh Assembly Government and Invest Northern Ireland.

Whilst reasonable steps have been taken to ensure that the information contained within this publication is correct, the authors, the Carbon Trust, its agents, contractors and sub-contractors give no warranty and make no representation as to its accuracy and accept no liability for any errors or omissions. Any trademarks, service marks or logos used in this publication, and copyright in it, are the property of the Carbon Trust. Nothing in this publication shall be construed as granting any licence or right to use or reproduce any of the trademarks, service marks, logos, copyright or any proprietary information in any way without the Carbon Trust’s prior written permission. The Carbon Trust enforces infringements of its intellectual property rights to the full extent permitted by law.

The Carbon Trust is a company limited by guarantee and registered in England and Wales under Company number 4190230 with its Registered Offi ce at: 6th Floor, 5 New Street Square, London EC4A 3BF.

Published in the UK: March 2010.

© Queen’s Printer and Controller of HMSO.

CTG008v2

monitoring and targeting - southcorner...

Documents