Applied Energy 88 (2011) 612–621

Contents lists available at ScienceDirect

Applied Energy

journal homepage: www.elsevier .com/ locate/apenergy

Methodology for characterising domestic electrical demand by usage categories

R.A.R. Kilpatrick *, P.F.G. Banfill, D.P. JenkinsSchool of the Built Environment, Heriot-Watt University, Edinburgh EH14 4AS, UK

a r t i c l e i n f o

Article history:Received 15 October 2009Received in revised form 11 June 2010Accepted 2 August 2010Available online 17 September 2010

Keywords:DomesticElectrical demandPowerAppliancesEnergyDemand profiles

0306-2619/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.apenergy.2010.08.002

* Corresponding author. Tel.: +44 (0) 131 451 4637E-mail address: rark4@hw.ac.uk (R.A.R. Kilpatrick)

a b s t r a c t

Electricity consumption in the United Kingdom is continually growing with demand from the domesticsector a potential/major contribution to this increase in consumption. Although demand is increasing, lit-tle information exists on the domestic components that contribute to an increase in domestic energy con-sumption. Thus, a greater understanding on what is contributing to the increase in domestic energy usageis a pre-requisite to understand how it can be reduced in the future or, if not reduced, contained at itscurrent level.

This article discusses a separation filter designed for disaggregating domestic electrical demand datainto different appliance categories. The filter is applied to a real time domestic electrical dataset spanning1 year, and trends in standby, cold, heating element spikes and residual demand are identified. Severalreasons to account for each of the trends are discussed. Additionally, the filter is applied to synthetic databoth to confirm the accuracy of the separation filter and to finely adjust the filter for future application.The results indicate an increase in occupancy-related demand consumption during the winter monthsand an increase in cold consumption during the summer months. Furthermore, the results demonstratethat in contrast to changes observed in occupancy-related demand and cold consumption, there is littlevariation in standby and heating element spike consumption throughout the year. Finally, the potentialadvantage of incorporating a tailored separation filter into domestic smart meters is discussed.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Domestic energy accounts for a significant proportion of totalenergy consumption both in industrialised and developing coun-tries; estimates suggest that domestic consumption accounts for31% of UK energy consumption [1]. Homes are where people spenda sizeable portion of their lives, where they eat, sleep and relax. It istherefore not surprising that domestic energy consumption resultsin a large impact on the national grid of industrialised and develop-ing countries. According to Wood and Newborough [2], the UK’speak power requirement increased from 37.7 GW in 1968 to50 GW in 2000. Wood argued that domestic energy use was themain driving factor for this rise in power usage. Furthermore, Bal-aras et al. estimated that building stock in the EU accounted for40% of total energy consumption, with 63% of that value beingassociated with residential use [3]. In the UK, electricity consump-tion of a household can vary from 2000 to 6000 kW h/year [2]. Theaverage domestic electrical energy consumption for a single house-hold in Scotland suggested is to be 4792 kW h/year [1] and theaverage household consumption in Great Britain is 4628 kW h/year[1]. The UK is one of the top four most energy consuming countrieswithin the EU, in terms of residential and tertiary energy use [3].

ll rights reserved.

..

Several studies have been undertaken to investigate domestic elec-tricity use within the UK ([4,5] for example) but little informationis given on how total domestic energy use is affected by the sea-sons or occupancy. In addition, interpretation of domestic usagedata is complicated by the fact that several methods have beenused to analyse energy demand data. Methods used have rangedfrom computer simulation and analysis programs to basic numericcalculation.

To gain a better insight into electricity demand, it is essential tobreak energy usage into its contributing components. Few investi-gations have shown annual trends for domestic buildings, or theannual trends within each of the appliance categories. This lackof basic information makes it difficult to advocate practical mea-sures to reduce energy usage at the individual domestic levelwhich could ultimately have a positive impact on national energyusage.

This study had two primary aims; the first was to design anddevelop a separation method that was capable of dividing a totalenergy demand profile into smaller individual energy categories.The second aim was to: (a) use the designed method on real-timedata, (b) study the different energy categories, (c) determine howthey changed throughout the year, with possible explanations toaccount for the observed changes. Although the described method-ology will be applicable to other industrialised and developingcountries this study will focus on the UK. A universal methodology

R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621 613

is important as it must be able to accommodate specific differencesin factors that contribute to domestic energy consumption. Housesin other countries (particularly those with domestic air-condition-ing) will have different electrical demand characteristics thanthose in the UK, where air-conditioning is not normally availablein domestic dwellings. In addition, houses with electrical demandsthat have a weaker relationship with external temperature (i.e.buildings without electrical heating or electrical cooling) are likelyto be more suitable for the approach chosen in this study.

1.1. Literature review

Load profile analysis in necessary to understand why energyusage is increasing and where it is being used in the domestic set-ting. Total energy consumption figures are useful for billing data,but it does not give an insight into how electricity is used, norhow consumer behaviour influences electrical use. To betterunderstand electrical usage (and to analyse load profiles), totalelectrical demands can be broken down into various categories ofusage. Wood and Newborough [2] believed that domestic energyusage could be described in terms of predictable, moderately pre-dictable and non-predictable usage. Firth et al. [4] believed energyusage could be broken down into ‘standby’, ‘continuous’, ‘cold’ and‘active’. It can be argued that both Wood et al. and Firth et al. def-initions of energy use are quite similar. ‘Continuous’ power is theresult of appliances being left on throughout the day. Such appli-ances include alarm clocks and home alarm systems. ‘Standby’ isvery similar to continuous, but is generally associated with thelow power mode of the appliance. Thus they consume electricityat a value between zero and the rated power of the appliance.Appliances in the standby category include TVs, Hi-Fis and otherhome entertainment systems. ‘Cold’ power includes any appliancethat uses a compressor refrigeration system, such as fridges andfreezers. ‘Active’ power is considered to result from appliancesbeing turned on for a period of time. This active component isthought to represent the remainder of the electrical use and canbe associated with appliances in operation such as cookers, show-ers, kettles, TVs and lights. In this study, we have considered,standby and continuous power as one ‘‘standby” component.

Several methods exist for analysing and disaggregating demanddata. One method is a software based system discussed by Farinac-cio and Zmeureanu [6]. This approach is a ‘top-down’ method thatuses previous data or assumptions to derive energy profiles forindividual types of energy demand. The Farinaccio method has sev-eral stages to separate out energy data. It uses pre-existing knowl-edge, gained during a training phase, to create a hypothesis aboutthe appliance or profile. The change in electric demand due to theappliance being turned on or off, the Herin ‘initial signal’, can beused to recognise appliances within the total load profile. Thismethod effectively uses predetermined appliance signatures andattempts to identify them with signatures in the profiles. A morestatistical approach was discussed in Aydinalp et al. [7], and in-volved taking into account the dwelling, number of occupants,number of children and whether the dwelling is owned or rented.

A calculative approach, using numerical calculations to deter-mine the individual components of energy usage was describedby Firth et al. [4]. This method involves multiplying the powerusage with the time intervals to obtain the energy use. The totalpower consumed, PTOTAL, is determined by the summation of cold,active and continuous and standby (C&S) power components.

If PTOTAL is defined, it is then possible to calculate the otherpower values contributing to PTOTAL. The first calculation is deter-mining the C&S power component. Firth assumes that the mini-mum value on the total load profile will represent the standby/continuous component.

The cold appliance consumption is more complicated to define,due to the nature of the cycling pattern in energy consumption. Inthis study we have assumed a constant minimum value, a constantmaximum value, and that the pattern does not change throughoutthe day. Firth et al. [4] suggest that taking data from 1 am to 4 amwill ensure an indicative cold power activity profile, as it is as-sumed that there will be no other activity than the refrigerationcontributing to energy usage. It should be noted that this also as-sumes that there is no home security lighting or courtesy lightingthat would affect the outcome of the monitoring period.

The last step is to determine the active portion of the energyconsumption. This portion accounts for the active appliances andthe active standby appliances (TVs or DVD players). If the stand-by/continuous and the cold power components have been calcu-lated, they can then be subtracted from the total powerconsumption to give an indication of active power. The definitionsof power, energy and duration for each of the different power cat-egories used in Firth et al.’s investigation can be found in Table 1.

Further research into domestic electricity consumption was car-ried out by Sidler [8]. As part of the ECODROME project, 20 house-holds were studied over a 2 year period. Each plugged load andlighting system was monitored with plug in meters that sent dataautomatically to a storage PC. The system was designed to be a ‘in-stall and forget’ system that requires no input or attention from thehouse occupants. The study provided detailed electricity demanddata of each household’s plugged load and lighting, that was ableto create a total electricity demand profile.

Constructing an algorithm that can disaggregate loads wouldhave a clear application in the area of smart metering of domesticenergy usage. There has been an increase in legislation across Eur-ope, regarding the importance of smart metering [9] and how itshould be used. In particular, it is often suggested that smart me-ters will be more effective if they provide the user with somegraphical interface showing how much energy they are using[10]. The methodology described in this study aims not just topresent the energy used by the occupant, but also to provide a vi-sual method of displaying the type of energy being consumed bythe occupant. It is suggested that information of this type indicat-ing, for example, whether standby loads are too high or energy useduring occupancy is excessive, will provide the user with an extralevel of information on which they can modify their behaviour inrelation to activities that consume energy.

The methodology discussed in this paper aims to extend obser-vations made in previous research. The methodology discussed inFarinaccio et al. aimed at breaking down recorded energy data intoindividual appliance usage, based on their unique energy signa-ture. There is one important point highlighted in this work: the‘initial signal’, used to identify appliances, can be confused bytwo or more appliances that have a similar ‘initial signal’. Equallythe system can be confused with two appliances being turned onat the same time, a problem that is evident with any load recogni-tion software [6]. It appears that this method is designed to sepa-rate demand data into the contributing appliances, rather thandivide the demand data into standby, cold and active.

The methodology and results discussed in Firth et al. [4] ap-peared to be more focused on producing an understanding of an-nual energy consumption. A calculative approach was used todetermine annual active, standby and cold energy consumption,and could be used to determine daily standby, active and activeenergy consumption. It does not appear that this method couldbe applied to daily demand data to produce separate energy useprofiles. The methodology discussed in Sidler [8] was a differentapproach to the previously discussed research. Individual appli-ances were monitored to create demand profiles. This methodwas based on a ‘‘bottom-up” approach rather than a ‘‘top-down”approach. The method discussed in [8] provided accurate demand

Table 1Components of power in relation to power consumed, energy consumed andduration.

Component Power consumed(kW)

Energy consumed(kW/h)

Duration(min)

Standby Low Low LongCold Low Low LongHeating

elementHigh Low Short

Residual Medium High Long

614 R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621

data, but required a large number of sensors and could only pro-cess data that has been collected from the monitored house. Con-versely, the disaggregation methodology discussed in this paperuses total demand data and breaks it down into appliance categoryusage. The discussed methodology may not provide detailed indi-vidual appliance usage, but it can be applied to older electrical de-mand data.

2. Methodology

The method used in this investigation utilised several stages toseparate standby, cold and active components from a total demandprofile. This section discusses each of these stages in turn. It shouldbe noted that the developed disaggregating methodology wasbased on UK domestic electricity data. Ideally, this discussed meth-odology can be applied to a variety of electric demand data, thoughfurther research would be needed into determine the influences on

Fig. 1. Data flow diagram, show

total electrical demand of space heating or air-conditioning used incooler or hotter climates. A basic overview of the separation meth-odology is shown in Fig. 1.

2.1. Standby removal

The first stage in the separation technique is removing thestandby component. The standby component is recognised ashaving a value equal to that of the lowest point on the load profile.‘‘Standby Power use typically describes the power consumption ofappliances when they are switched off or not providing theirprimary services” [11]. In order to remove standby power fromthe demand data, the assumption is made that the standby powerconsumption remains constant throughout the profile. It is thisconstant value that is subtracted from the total demand profile.A simple algorithm can be written to identify the lowest valueand subsequently remove it from the profile. Clearly, if theassumption that the minimum value is equal to the standby com-ponent is not correct, then this will have an impact on the follow-ing separation of the cold and active components.

2.2. Cold removal

The second stage of separation is removing the cold component.Cold power is characterised by a cycling on/off pattern and is easilyrecognisable in the load profile, shown in Fig. 2. Following theremoval of the standby component, the cold pattern should startfrom zero (C0), increase to a set point (C1) then decrease back tozero after a predetermined time (TN). This pattern then repeats

ing each filter and output.

R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621 615

after another preset time (TF). The assumption that the patterndoes not change throughout the period of the load profile wasmade.

A similar method for removing the standby component is ap-plied to the cold element removal. First, key parts of the patternare identified and values recorded. These key points are as follows:

� C0 (W)� C1 (W)� TN (min)� TF (min)

An initial approach was to use a filter to remove any value in theprofile that had a value equal to, or less than the on value, C1. Inaddition, the hypothesis is made that, if the load profile pointwas above the C1 value, then C1 value is subtracted from the loadprofile point.

This approach has two disadvantages. The first disadvantage isthat the approach will remove any other pattern that is less thanthe C1 value, including power spikes and other valuable data. Thesecond disadvantage is that the filter will also remove non-cold en-ergy from the large spikes (those that last for 30 min or longer)hence resulting in misleading data.

A more practical approach is to construct a cold profile span-ning an entire day, then to subtract this profile from the total loadprofile. The cold profile, as defined in Section 1, can be constructedusing the information gained from the previous analysis stage. Intheory, if the cold pattern remains constant throughout the year,then a standardised cold profile can be constructed and appliedto each single day. This concept would be viable for achieving gen-eric cold appliance filter. Thus it is important to establish the var-iability in the cold power throughout the day.

2.3. Active removal

The remaining component of the load profile is the active por-tion, as defined as Section 1. To further understand energy use, this

Fig. 2. Example cold profile, showing key values used

active portion can be broken down into two further categories;heating element spikes and residual demand. The heating elementspikes are a result of an appliance that utilises a heating elementbeing turned on, such as toasters, kettles, and electric ovens orhobs. These appliances cause large spikes (several kilowatts) andcan last for several minutes. The residual demand is the remainingarea of active portion once the heating element spikes have beenremoved. This component, usually the result of lighting and homeentertainment, is very important as it can represent a large per-centage of the Active energy consumption.

The first stage of separation of the active component is remov-ing the heating element spikes. This was achieved by designing andimplementing a five stage filter. The multi-stage filter is based onthe assumption that the majority of heating spikes only last for5 min or under. The filter analyses the active portion of the demanddata and subsequently removes any spikes that coincide with apredetermined criterion or condition. A ‘spike factor’ (S.F) is de-fined, that compares power at any given time, Pt, with a point someother time, Pt+i, where i is the duration between these two points(normally 1–5 min). The S.F can be adjusted to increase or decreasethe spike removal, with a high S.F value only removing the largerpower spikes. The spike factor is represented in the equationbelow:

S:F ¼ Ptþi

Pt

The selected time intervals were i = 5, 4, 3, 2, and 1 min(s).With the heating elements removed, the remaining data can be

classed as the residual demand. No further separation was carriedout on this component as it tends to be a relatively smooth profileresulting from several different appliance signatures (such as TVs,computers and lighting, as well as the electric pump associatedwith a boiler). The residual demand profile itself tends to be a goodindicator of occupancy in the dwelling, so energy use in this profilecan be considered to be ‘‘occupancy-related”.

for data analysis (time axis in minutes:seconds).

616 R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621

3. Analysis

To investigate the accuracy of the obtained results a comparisonis needed. Ideally, data with known variables, i.e. known equip-ment usage, duration of usage and energy consumption corre-sponding to usage, are required as a comparison. The problemwith using total electrical demand data is that it is very difficultto positively identify appliances from the profile. This can resultin separation errors, causing inaccurate Standby, Cold and Activevalues. One solution is to use composite data, i.e. an electrical de-mand profile made up of known individual appliance profiles, as analternative to real-time data.

The advantage of using composite data is that the load profilesare constructed from the summation of known data. Although thesummation of the data is artificial, (albeit based on typical occu-pancies), the appliance signatures used in the profile are real-timeappliance signatures [12]. The user sums the real-time appliancesignatures over an entire day to construct the synthetic profile,as shown in Fig. 3. The end result is a load profile graph withknown values for standby, cold and active portions.

The constructed synthetic load profile was subjected to the sep-aration filters described in Section 3. To determine how successfulthe separation was, the results from the filters were compared tothe known variables, as discussed previously.

It should be noted that one disadvantage of using the syntheticprofile is that the standby portion of energy use is not a separatecomponent. Instead it is part of each real time signature. This doesnot affect the separation technique, but will affect the comparisonresults. To overcome this, the standby component is removed, asper the methodology, but is then added to the residual demand.This coincided with how the synthetic data is combined.

3.1. Cold comparison

Fig. 4 compares the actual cold profile from the synthetic (origi-nal) data with the cold profile predicted by the cold profile filter ofSection 2.2. The two profiles, though similar, are not an identicalmatch. The profiles appear to be matched up to 0700 h but afterthat the profiles become unsynchronised. The likely reason for thisis that the synthetic profile may change during the day resulting ina non-symmetric profile. If this is the case, the designed separationfilter will not successfully remove the entire cold profile. To suc-cessfully remove the cold profile a ‘‘fine tuning” system needs to

Fig. 3. Example of synthetic data

be incorporated into the separation method. Each time the Cold fil-ter is applied, the resulting active profile has to be checked to en-sure the cold profile has been successfully removed.

3.2. Heating element spike comparison

Fig. 5 compares the known heating element spikes (from kitch-en use) from the synthetic data with those predicted by the filter-ing process of Section 2.3. The graph demonstrates that themajority of heating element spikes are removed by the filter. Thereare several spikes that are either not removed or not completely re-moved by the filter. Examples of these problems are shown at0730 h, 0815 h, 1700 h and 2130 h. Initially the filter removed only20% of the spikes, with the spike factor set at 2. With the spike fac-tor adjusted to 1.15, the filter removed approximately 80% of thespikes. The problem with adjusting the spike factor to this valueis that the filter then also removes unwanted features such as partsof the cold appliance signature. This is noticeable during the hoursfrom 1000 h to 1530 h. However this only represents a small por-tion of the total demand profile and is considered an acceptable er-ror margin.

3.3. Residual demand comparison

Fig. 6 shows the residual demand for both the synthetic dataand that following the use of the filter. The synthetic residual de-mand data is the summation of lighting power and home enter-tainment power. The filtered profile has a similar shape to thesynthetic data, but there are several discrepancies with the profilematching. Several spikes do not match the synthetic profile. Thiscould simply reflect the fact that the spike filter has not success-fully removed all the heating element spikes. In theory the spikefactor could be adjusted to remove these additional spikes. Anotherproblem is that from 0900 h to 1500 h the filtered profile appearsto resemble a cold profile pattern. This could be due to errors inremoving the cold profile in the earlier filter stage. Incorrect datamay have been removed accidentally by the Cold filter resultingin the unmatched profiles.

3.4. Synthetic test results

The synthetic test is useful in determining how successful thediscussed methodology is in separating load profile data. It can

representing one 24 h period.

Fig. 4. Comparison of synthetic data cold profile (original), and synthetic filtered cold profile (filter).

0

0.5

1

1.5

2

2.5

3

3.5

4

00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00Time

Pow

er (k

W)

Heating Element(Original)Heating Element(Filter)

Fig. 5. Comparison of synthetic heating element spikes (original) and heating element spikes from filtered synthetic data (filter).

R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621 617

be seen from Figs. 4–6 that the developed filters are reasonably butnot totally effective in separating out the different power compo-nents. The results highlight that the disaggregation methodologyis suitable; however the set-up parameters do require adjustmentto achieve optimum results. The correct selection of the ‘spike fac-tor’, and careful design of the cold profiles, improves the accuracyof the separation technique. Additionally, the correct set-upparameters could be adjusted when real-time data is introduced.

4. Results and validation

The separation method is applied to total electric demand data,with one-minutely temporal precision, for an entire year (data ta-ken from Peacock and Newborough [13]) for a single dwelling in1994, see Fig. 7 for an example of monthly power consumption.Little information is known about the building, i.e. size, age, occu-pancy, but this is not required for the separation (with the filtersactually being designed for use with such a ‘‘blind” dataset). Dueto time constraints, it is not possible to apply the filters to all365 days. Instead, 8 days from each month (each Wednesday andSaturday in the week) are chosen, totalling 96 days and the results

scaled up for the entire year. It is important to select a weekendday because the energy profile can be significantly different to aweekday. Generally, more energy is consumed at the weekenddue to the occupants staying in the house for longer periods. Thetotal annual consumption of the house is 5900 kW h, whereasthe total calculated from the 96 design days (and extrapolatedfor the whole year) is 6060 kW h confirming that the chosen designdays were suitable for extrapolation.

The result of applying the separation algorithms for all designdays is shown in Fig. 8, giving monthly breakdown by energy usecategory. For the results section, we will define spring months asMarch, April and May, and summer months as June, July and Au-gust. Autumn months are also defined as September, October andNovember, and lastly winter months as December, January andFebruary.

4.1. Standby results

Fig. 8 shows that the average standby energy consumption var-ies slightly throughout the year (with a maximum of 5.1 kW h/dayin August to a minimum of 2.7 kW h/day in February). If both May

Fig. 6. Comparison of synthetic residual demand (original) and residual demand from filtered synthetic data (filtered).

Fig. 7. Example of power demand for the studied dwelling over a 1 month period (July 1994).

618 R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621

and August months are removed, there is little change between thesummer months and the winter months. In December the con-sumption was 4.12 kW h/day compared with July’s consumptionwhich was 4.18 kW h/day.

Previous research [8,11,14,15], investigated standby power con-sumption, with varying detail. From such work, it can be estab-lished that the average standby power consumption in the UK is32 W, resulting in 277 kW h/year or alternatively 8% of totaldomestic electricity demand [14]. The average EU yearly standbyenergy consumption figure is 439 kW h/year [8], which equatesto over 50 W of standby. A worse case scenario could see the stand-by at 81 W resulting in a consumption of 647 kW h/year [15]. Themaximum standby power, minimum standby power and yearlystandby energy consumption figures of the studied building were212 W, 90 W and 1383 kW h/year, respectively. When comparedwith the EU average it would appear that the studied house hasabnormally high standby power, due either to poor energy man-agement or to having a large amount of appliances.

The marginal rise in standby energy during the summer monthscould be due to a variety of causes. The main reason could be theoccupants watching more television during the winter monthsthan the summer months. The standby power consumption of aTV will decrease when it is turned on because it is now identifiedas consuming active power. The other reason for the increase inStandby power during the summer months may be due to moreappliances being used in the summer than in winter, perhaps re-lated to outdoor activities. Furthermore, the standby energy canalso contain the standby power associated with the refrigerationappliances which consume more energy in the summer due tohigher external temperatures.

4.2. Cold results

Fig. 8 also shows the cold profile energy consumption. The re-sults reveal that there is a difference in consumption betweenthe summer months and winter months (with the maximum

Fig. 8. Standby, cold, heating element and residual demand results for a single dwelling over a 12 month period.

R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621 619

consumption of 2.91 kW h/day in September and the minimumconsumption of 2.01 kW h/day occurring in March). If Septemberis removed, a slight rise in cold energy consumption can be seenduring the summer months. The results show that the averageconsumption of cold energy during the spring months rises from2 kW h/day to 2.5 kW h/day, the summer months see a drop inconsumption from a peak of 2.8 kW h/day to 2.6 kW h/day. Au-tumn also sees a drop in cold energy consumption which peakedat 2.9 kW h/day and fell to 2.6 kW h/day. Lastly the cold energyconsumed during the winter period remains at a fairly constant2 kW h/day. This difference in consumption is likely to be due tothe refrigeration devices working harder during the summermonths. If there is a rise in ambient temperature, the fridge or free-zer will require more power, either as a change in peak cyclingpower or a change in the time period of each cycle. Meier [16] dis-covered during researching refrigerator energy use that energyconsumption of a fridge doubled when kitchen ambient tempera-ture increased from 17 �C to 28 �C. Another reason for this differ-ence between summer and winter may be due to another Coldappliance being used during the summer months. This appliancecould be a wine cooler, beer fridge or an alternative cooling device.

4.3. Heating element spike results

In the case of heating element spike consumption, the resultsillustrate that there is no apparent trend between this energy con-sumption and the varying seasons. The maximum heating elementspike consumption of 3.88 kW h/day occurs in October and theminimum consumption of 2.03 kW h/day occurs in July. If the re-sults are further analysed, and a trend line added, it is possible todiscern a slight rise in energy consumption towards the wintermonths (with summer having an average of 2.29 kW h/day andwinter having an average of 2.48 kW h/day). It is difficult to explainthis rise in demand purely as a function of appliance usage. As dis-cussed before, the main appliances that use a heating element arekettles, showers, toasters and electric cookers or hobs, as well asappliances that heat water, such as washing machines.

One reason for the rise in this energy use during winter could bethat the electric cooker is working harder in winter than insummer. If the ambient temperature was low, the cooker woulduse more power to maintain the desired temperature. Alterna-tively, the cooker could be used more during winter because ofmore holiday meals being cooked (e.g. Christmas and New Yearperiod). Another reason for a winter rise in consumption couldbe due to the residents boiling a kettle more often i.e. having morecups of tea/coffee during the winter months, or even increased useof electric heaters. However, the seasonal difference appears to bevery small.

4.4. Residual demand results

Residual demand energy consumption is seen to drop duringthe summer months and then rise during the winter months. Thepeak value of 12.26 kW h/day occurs in December and the mini-mum value of 4.95 kW h/day occurs in March. This rise in wintercan be seen further if we look at the individual seasonal residualenergy consumption. The spring sees an average consumption of5.6 kW h/day whereas the summer has an average consumptionof 6.3 kW h/day. The rise in energy consumption is apparent be-tween autumn and winter months with averages of 9.08 kW h/day and 10.24 kW h/day, respectively. From this information, it issuggested that the trend in residual demand influences the trendin active power consumption.

The trend in residual energy consumption seen in Fig. 8 couldsimply relate to the increase of lighting. As it becomes darker ear-lier in the winter, lighting will be used earlier in the day and be onfor longer when compared with the summer. Lighting can accountfor 10% [15], 12% [15] and 20% [5,17] of total electricity demanddepending on energy category. This equates to 605.9 kW h/year,727 kW h/year and 1211 kW h/year, respectively in the studiedhouse. If this is compared with the average UK of 715 kW h/year[18], it falls between 10% and 12% lighting ranges. Thus any in-crease in the use of lighting above ‘‘normal” will have a significantimpact on the total Active consumption for that household. With

620 R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621

the information unavailable, it is difficult to accurately determineif the rise in energy consumption during winter months is solelydue to the increased usage of lighting, but it is reasonable to sug-gest that this is a major factor. If the monitoring techniques usedin Sidler [8] were applied instead of the separation filter, theassumption that lighting is a major factor in increasing residual de-mand could be answered. However, as discussed in Section 2, thiswould require a longer research time, and the added cost of buyinglarge quantities of monitoring equipment. Another reason for theincrease in power during the winter months is the use of homeentertainment systems such as televisions, PCs, computer consoles,HI-FIs and DVD players. It is likely that residents will stay insidemore during the winter months due to, for example, bad weatherconditions, increasing their use of home entertainment systems.This confirms the previous asserted point that the residual demandprofile is essentially an occupancy driven profile. The rise in energyuse during the winter could again be a result of space heaters beingused (albeit smaller systems that do not show up as large powerspikes). Based on the available information, it will be assumed thatthe Active component of the load profile is purely the summationof lighting power and home entertainment.

4.5. Overview of annual consumption

The energy consumed in each energy component for the se-lected dwelling over the year is shown in Table 2. From this dia-gram it can be assumed that the studied dwelling has relativelyhigh energy consumption. According to Firth et al. [4], the high,medium and low consumption classifications can be determinedby the standby, cold and active percentages (in relation to total en-ergy consumption). Table 2 demonstrates the three energy con-sumption groups and the corresponding component percentages.Additionally, the standby, cold and active results from this investi-gation can also be found in Table 2. From this table, it can be con-firmed that the studied house falls within the high energy group. Asecond confirmation that the studied dwelling is a ‘‘High EnergyGroup” building is that the total energy consumption of4841 kW h/year is around 1300 kW h less than that of the totalfound in this investigation.

5. Discussion

The aim of this investigation was to create a usable and genericseparation technique that when applied to domestic load datawould identify and separate individual appliance signatures thatcontribute to total load. To accomplish this objective the followingspecifications were defined for the developed filters that would beused on total energy data. The filters would: (a) be able to separatestandby, cold and active components of energy use, (b) be applica-

Table 2Energy group classification and criteria [4] plus results from this study in bold.

Annual consumption (kW h)

Total Standby andcontinuous

Coldappliance

Activeappliance

All dwellings [4] 3100(100%)

601 (19%) 620 (20%) 1879 (61%)

Low energy group[4]

1770(100%)

297 (17%) 535 (30%) 938 (53%)

Medium energygroup [4]

2689(100%)

402 (25%) 577 (21%) 1710 (64%)

High energygroup [4]

4841(100%)

1104 (23%) 747 (15%) 2990 (62%)

Results from thisstudy

6059(100%)

1383 (23%) 906 (15%) 3770 (62%)

ble for automatic application to any given load and (c) be versatilewith regard to their application. Initial testing of the developed fil-ters proved successful in separating out total domestic usage intoindividual components and achieved the desired outcome of iden-tifying and isolating energy usage at the ‘‘appliance” level. In addi-tion the developed filters were capable of disaggregating electricityusage into standby, cold, heating element spikes and residual de-mand components with a reasonable degree of accuracy.

Application of the filters to the 96 days dataset, and when ad-justed for an entire year, indicated that the heating componentthat contributed to total energy usage, was primarily related todomestic occupancy. Although active appliance usage was also re-lated to occupancy, there did not appear to be a seasonal variationrelationship with the heating component as it only varied from2 kW h/day to 3.8 kW h/day throughout the year. In contrastalthough residual demand was also consistent with an occupancyprofile, seasonal effects were apparent; consumption was greaterin winter at 10. kW h/day compared with a summer’s average of6.3 kW h/day.

It has already been discussed in Section 4.5 that this change inenergy use could be related to a greater requirement for lighting bythe occupants due to it getting darker outside in winter or alterna-tively less lighting being required due to longer daylight in thesummer months. The filters also highlighted that both the coldand standby components of total energy usage were only margin-ally affected by seasonal variation, with an increase from 3.4 kW h/day (average) in winter to 4.28 kW h/day (average) in summer forthe standby component of energy usage. The filters also identifieda slight rise for cold appliance usage in summer, with an average of2.7 kW h/day compared with the average winter consumption of2.1 kW h/day. The standby, cold and active components identifiedwith the developed filters suggest that when compared to the datapublished by Firth (see Table 2), the houses contributing to thedataset used in this study were in the ‘‘high energy” group. Thisview is further supported by the fact that the households in thisstudy used in excess of 6000 kW h/year.

An important specification of the filters was that they had to beversatile. In addition to being able to identify and separate standby,cold and active components of total energy usage, it was also arequirement that the filters should be able to monitor applianceenergy usage at the level of the individual appliance type. Thusthe ultimate goal of developing the filters was that they shouldbe able to be applied retrospectively to historical data in additionto being applied in a ‘‘real life” situation in a domestic setting. Thusa key advantage of the developed filters is that they can be appliedto actual measured energy consumption data that has been previ-ously separated into appliance usage data, as described by [6,8].This disaggregated appliance power usage data can be combinedby energy use type, to form accurate standby, cold, heating ele-ment spike and residual demand data. This information generatedonce could then be compared to the results from applying the filterto the same data, similarly to the synthetic data test discussed inSection 3.

The ultimate aim will be to extend the specifications of the fil-ters to achieve the goal to develop automatic separation filters thatcan be applied to ‘‘real time” energy usage. To achieve this objec-tive further refinement of the filters will be required. One develop-ment would be a detection-and-mapping system for the cold filter.If a load profile is recorded when there is least activity, ideally be-tween 1 am and 4 am [4], the cold profile could be mapped andrecreated. The recreated cold profile could then be removed fromthe total profile, allowing further separation of the other powercomponents. A further required enhancement to the filters wouldbe the ability to automatically adjust the ‘spike factor’ frequentlyobserved in the analysed dataset. Instead of the user manuallyadjusting the ‘spike factor’, the filter would modify the ‘spike fac-

R.A.R. Kilpatrick et al. / Applied Energy 88 (2011) 612–621 621

tor’ based on the output. Essentially the system would be based on‘trial and error’ to determine the optimum value. The last requiredenhancement to the current filter would be to extend the numberof heating element spike removal filters. The current filter systemhas five sub-filters allowing the removal of up to 5 min spikes. Thislimitation of the current filter system means that only spikesappearing within that time frame will be removed. Appliances suchas electric cookers, showers and washing machines tend to haveheating times longer than 5 min. Thus, the current spike removalfilter would be unable to remove these spikes fully, leading to po-tential inaccurate results. One solution would be to increase thenumber of sub-filters from five, to approximately thirty. With thisnumber of sub-filters, projections suggest that the majority ofheating element spikes associated with electric cookers/ovenswould be removed.

With further development and enhancements to the existingfilters, the system has the potential to achieve the goal of a fullyautomatic separation filter system that can be applied to ‘‘realtime” energy usage in a domestic setting. Life style changes canbe made by the consumer and the impact of these changes is read-ily measurable. The direct effect of simply switching off TVs, lightsand computers, or switching to energy efficient appliances andlighting has on their energy demand can be instantly available.The latest smart meters discussed by the Energy Saving Trust[10], have displays that range from simple LCD numerical displays,to a more complex colour bar graph system, that moves from red togreen depending on the energy use (green representing low con-sumption and red representing high energy consumption). If thesame output format shown in Fig. 8 were to be used, customerscould see the bars increase or decrease as appliances are beingturned on or off. Such a system, although still only providing basicenergy consumption data to the customer, does make it possible toidentify areas of energy ‘wastage’ and then take steps towardsmore efficient usage of energy. In future, with the introduction ofenergy category recognition software, further domestic energy sav-ings can be obtained.

6. Conclusions and recommendations

A method of disaggregating short-time-resolution electricityconsumption profiles for dwellings has been developed. It allowsa user to enter or copy and paste load profile data into a spread-sheet where a series of filters separates it into four components –standby, cold, heating element spikes and residual demand. Thesystem has the potential to be used as an automatic filter indomestic smart meters, which could give householders detailedinformation not only on the total energy used in a period but alsothe appliances that are contributing to their total energy consump-tion and costs. This information would allow the impact of changesin behaviour to be immediately visible to the householder.

Applied to a real data set collected from a single dwelling over12 months the separation filter shows clear trends in the fourcomponents, with an increase in occupancy-related electricityconsumption during the winter and an increase in cold-related

consumption in summer. On the other hand there is little variationin standby and heating element spike consumption over the year.

Since demand for electricity will continue rising unless drasticmeasures are taken, the ability to identify the individual contribu-tors to total demand allows consumers to recognise and then con-trol their consumption. Therefore it is strongly recommended thatresearch effort be devoted to developing fully automatic patternrecognition software that could be integrated into the next gener-ation of smart meters.

Acknowledgement

The authors would like to thank David Kane for providing theappliance profiles as used in the synthetic data test.

References

[1] Energy consumption in the United Kingdom. Official Governmental Document,courtesy of the Department of Trade and Industry (National Statistics); 2008.

[2] Wood G, Newborough M. Dynamic energy consumption indicators fordomestic appliances: environment, behaviour and design. Energy Build2002;35(8):821–41.

[3] Balaras CA, Gaglia AG, Georgopoulou E, Mirasgedis S, Sarafidis Y, Lalas DP.European residential buildings and empirical assessment of the Hellenicbuilding stock, energy consumption, emissions and potential energy savings.Build Environ 2007;42:1298–314.

[4] Firth S, Lomas K, Wright A, Wall R. Identifying trends in the use of domesticappliances from household electricity consumption measurements. EnergyBuild 2007;40(5):926–36.

[5] Mansouri I, Newborough M, Probert D. Energy consumption in UK households:impact of domestic electrical appliances. Appl Energy 1996;54(3):211–85.

[6] Farinaccio L, Zmeureanu R. Using a pattern recognition approach todisaggregate the total electricity consumption in a house into the major endusers. Energy Build 1999;30(3):245–529.

[7] Aydinalp M, Ugersal VI, Fung AS. Modelling of the appliance, lighting and spacecooling energy consumptions in the residential sector using neural networks.Appl Energy 2002;71(2):87–110.

[8] Sidler O. DSM: major findings of an end use metering campaign in 400households of four European countries. In: European Council for an EnergyEfficient Economy (ECEEE) summer study proceedings; 2003.

[9] Directive 2006/32/EC of the European parliament and the council 5th April2006. Energy end-use efficiency and energy services and repealing CouncilDirective 93/76/EEC, EU Parliament Directive.

[10] The smart way to display: a summary report on consumer preferences forenergy display designs. Energy Saving Trust Publication; 2009. <http://www.energysavingtrust.org.uk/corporate/Global-Data/Publications/Corporate-publications-folder/The-smart-way-to-display-full-report> [accessed 07.06.10].

[11] Meier A, Lin J, Liu J, Li T. Standby power use in Chinese Homes. Energy Build2004;36:122–1216.

[12] Kane D, Newborough M. Estimating carbon savings for domestic base-loadmicro-CHP systems. World Renewable Energy Congress (WREC) IX, Florence;2006.

[13] Peacock A, Newborough M. Impact of micro CHP systems on domestic sectorCO2 emissions. Appl Therm Eng 2005;25(17–18):2653–76.

[14] International Energy Agency (IEA). Things that go Blip in the night: standbypower and how to limit it, courtesy of IEA Publications; 2001. <http://www.iea.org/publications/free_new_Desc.asp?PUBS_ID=1110> [accessed 07.06.10].

[15] Bonneville E, Rialhe A. Demand side management for residential andcommercial end users. Efficiency Eco-Des 2006;146(3):283–93.

[16] Meier A. Refrigerator use in the laboratory and in the field. Energy Build1995;22:233–43.

[17] Stokes M, Rylatt M, Lomas K. A simple model of domestic lighting. EnergyBuild 2004;36:103–16.

[18] Richardson I, Thomson M, Infield D, Delahunty A. Domestic lighting: a high-resolution energy demand model. Energy Build 2009;41:781–98.