a methodology for surveying domestic water consumption

Upload: kelly-fontoura

Post on 14-Apr-2018

223 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    1/12

    A Methodology for Surveying Domestic Water ConsumptionB y K. EDWARDS, BSc, CEng, MICE (Member) and L. MARTIN, MA, FG S (Member)*

    ABSTRACTThis paper describes the experimental design andengineering of the largest contemporary survey ofdomestic water consumption in the UK. Results arepresented for the first full year of data co llection andare discussed in term s of ex planatory variables, relia-bility and usefulness. Sign ificant findings include (a)the range o f p e r capita consumption, (b) relationshipsto explanatory variables, (c) componen ts of consump-tion including legitimate night flows, and (d) the con-tributiyns of specific dom estic appliances.Ke y words: Consumption; demand; domestic use; leakage;metering; night flows; tariffs; water consumption.

    INTRODUCTIONSince privatization, a key element of water com-panies concerns has been the charging method whichshould be adopted for domestic customers. Optionsfor charging depend upon a detailed knowledge ofcustomer water use, accurate forecasting of futuredemands and the temporal and spatial variability,together with their causes, which underlie thesepatterns.

    Reviews of existing studies confirmed the opinionthat the Ang lian region was unique w ithin the UK interms of climatic and, above all, socio-economicparameters. No current analysis provided the neces-sary data which the company required; therefore afive-year study was sanctioned to examine both thecompon ents and the variability o f domestic consump-tion within East Anglia. This paper examines theengineering behind this survey and presents results forthe first full twelve months of data. The survey isknown by its acronym SODC ON.

    BACKGROUNDOTHER TUDIESIn the late 1980s, many studies were addressing, bothdirectly and indirectly, the issue of dom estic meteringand domestic consumption. The most well-known

    This paper was presented for discussion at a meeting of the EastAnglian Branch, held on 11 May 1994.*Customer Services Support Manager (and SODCON ProjectManager) and Marketing Development Manager, respectively,Anglian Water, Huntingdon, Cambridgeshire, UK .

    programme was that of the National MeteringTrials()which took place between 1989 and 1993 intwelve locations, mostly in southern and centralEngland.Other long-term studies had been undertaken bySouth West Water(*) nd S evern Trent Wated3).How-ever, none of these was able to provide Anglian Waterwith the specific information required because,although these studies may have been representativeof their own area, they failed to provide data whichcould be reliably transferred to the region of EastAng lia. The Na tional Metering Trials were not estab-lished to be studies of d omestic consumptionper se,and the regional studies in South West and SevernTrent Water were only statistically relevant to theirrespective regions.REGIONALPECIFICS

    The East Anglian region is unique in the UK.Clima tically, it has the most continental climate of theU K its soils are mostly derived from quaternarydepo sits with a high percentag e of glacial gravels andsands, and there is a n unusual mixture of heavy claysadjacent to very light sands/sandy loams. From asocio-eco nom ic perspective, the region is o ne of highgrowth potential (albeit latent in the 198Os/early1990s) and net p opulation migrational influx.As such,East Anglia does not possess the same econo-geog raphical structure as any other part of the UK. Nowater consumption studies were available whichcould exp lain the likely pattern of d om estic water usein this part o f he coun try. It was therefore decided thatthe investm ent, which was required in establishing afull survey of domestic consumption in the EastAng lian region, would be justified. In 19 91, approvalwas given to the undertaking of the comprehensivesurvey of domestic water consumption, known asSODCON.OBJECTrVES

    The m ain purpose o f he su rvey was to aid the selec-tion of A nglian Waters future charging method and toenable the com pany to manage its water resources tomeet future demands. SODCON was thereforedesigne d to prov ide the fo llowing information:(i) An explanation of the factors that determine un-measured water demand;(ii) Details of the patterns of water consumption acrossdifferent household, soil, climatic and socio-

    economic types;JCIWEM, 1995,9, October 411

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    2/12

    EDWARDS AN D MARTIN ON(iii) Estimates of demand responses to different tariffstructures, analysing the effects arising from differentcharging options;(iv) Detailed costs of metering;(v) Aggregate estimates of per capita consumption foruse in demand forecasting and investment appraisal;and(vi) Estimates of leakage, legitimate night use and thedevelopment of an economic leakage control model.As an added benefit, the company would obtainvaluable experience in the field of public relations andgain a substantial database of socio-economic infor-mation in respect of its domestic customers.

    EXPERIMENTAL DESIGNSAMPLINGFor the survey to be statistically robust it had to iden-tify all the characteristics which were likely to explainvariations in consumption. A sample of approximatelytwo thousand households was chosen for the 1 1 1 sur-vey. They all had meters and data loggers installedexternally, as detailed below. This is a relatively smallsample in relation to the 1.4 million unmeasuredhouseholds in the region, and it was recognized that arandom distribution would not give sufficient confi-dence. A structured, stratified approach was thereforeselected. It was decided to sample households in clus-ters, divided into divisional sub-samples according tcrthe balance of population in each, and stratified byACORN (A Classification Of Residential Neighbour-hoods). ACORN is a geodemographic classificationsystem for small aread4),which classifies enumeratedistricts (EDs) according to simitarity across selectedcensus variables.The sampled clusters were allocated to 50 publicwater supply zones picked as areas within which thesehouseholds were located. Each zone was checked toensure that all climatic and soil types, typical of theregion as a whole, were represented. This segment-ation would allow for the significance of gardenwatering to be measured.A Similar check was made tolocate sites with pipework of differing 5ges and buriedin differing soil types.In the Anglian region it was determined that therewere three climatic zones where rainfall tended to beheavy, medium or light. Similarly there were three soiltypes; heavy (clay), medium (chalk) or light (sandy).The age of the network was classified as pre-war,1950s, 1970s and modern. It was felt that therewould be sufficient correlation between age andpipeline materials, therefore no fiuther subdivisionswere made.The ACORN classifications which were used forthe sample were consolidated into eight groups:

    (u) Housing in agncultural areas;(b) Modem family housing, higher income bracket;(c) Older housing of intermediate status;47 8

    (d) Older terraced housing;(e),0,g) Council housing (three categories);(71) High-status non-family areas;(i) ffluent suburban housing; and0) Better-off retirement areas.SPECIFICTUDIES

    Within the sample of 2000 households a sub-set ofone hundred was selected for special investigation.These were known as the Golden 100 and eachproperty therein had, in addition to its external meter,every separate water-using appliance monitored indi-vidually at its point of use.To further develop Anglians understanding of theeffects of metering on demand, a further 1000 cus-tomers were recruited. These customers were alreadypaying for their water on a measured basis, by meter.They undertook the same socio-economic interviews,but their meters were read manually at a monthly fie-quency and their patterns of consumption were com-pared with the SODCON database.

    PROJECT ENGINEERINGTIMESCALEThe overall project has been planned to run for fiveyears. The first year was used for setting up and engin-eering the project, and the remaining four years arebeing used for collecting data and carrying outanalysis. Table I lists important dates and actions.

    TABLE I. KEYDATESN SODCON PROGRAMMEAction I DatesPilot2000 volunteers1855 meter units installed in145 internal meters installed63 mains meters installed for leakagiInstall meters and telemetry toSocio-economic questionnaireProperty and appliance surveyBuild input and output data

    highway

    G100completedcompletedprogrammes

    Jul, Aug 1991Oct, Nov, Dec 1991;Jan,Feb, Mar 1992Nov, Dcc 1991; Jan, Feb,Mar 1992Feb, Mar, Apl 1992May, Jun, JuI199 2May, Jun, Aug, Sept 1992May, Jun, Jul 1992May, June, Jul, Aug, Sept1992Oct, Nov, Dec I99 1; Jan,Fcb, Mar, Apl, May, Jun,Jul, Aug, Sept 1992

    MANPOWERSeven full-time staff are employed on the projectand two contractors are engaged to read meters anddata loggers and change meters when necessary. Inaddition, they also carry out any plumbing mainten-ance i-equired in the so-called Golden 100 houses.The staff comprise a market research coordinator,

    based at AWS regional headquarters, and four marketresearch officers, one in each Division. The latter areJCIWEM, 1995,9, October

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    3/12

    A METHODOLOGY FOR SURVEYINGDOMESTIC WATER CONSUMPTIONresponsible for most of the activities in their divisionsand manage their stand-alone computer databases. Asystems analyst programmer and a principal engineerspecializing in telemetry systems are also members ofthe team. Economic consultants have been engaged toadvise on the project design and to analyse the data.PILOT TUDY

    A pilot study was designed to gauge customer reac-tion to an in-depth study of domestic water consump-tion and two questionnaires were used, one in the formof a diary of water use and the other for socio-economic information. The pilot study was carried outin the Northampton area for a period of six weeksduring the months of July and August 1991.Fifty households who were already paying theirwater charges on a metered basis took part in thestudy. The volunteers meters were read manuallyfour times each day for the duration of the pilot, thenthe diary and socio-economic data were matched tothe meter readings to establish if there was any corre-lation of water use.The following lessons were learned from the pilotstudy:

    There was sufficient evidence to justify carrying outthe full survey;People generally were interested in how water wasused and were willing to join in the survey. Of thepeople contacted to take part in the pilot study, 50%agreed to participate;Volunteers were not aware of the project in advance,and a substantial amount of time was spent explainingwhat the survey was all about;Customers were far more receptive to talking to staffin uniform and using a clearly marked Anglian Watervehicle than to staff not in uniform and usingunmarked private vehicles;Volunteers were reluctant to give details of their earn-ings because the questions were too specific; andDiary forms were not completed accurately becausevolunteers had to fill in too much detail.

    The following action was taken to address the prob-lems which were identified in the pilot study:Information leaflets about the project were preparedand sent out with a mail shot in advanceof callingonpropective volunteers;A decision was made that the market research officerswould wear a uniform and use clearly marked AnglianWater vehicles;The diary form was made more user-friendly, as a ticksheet, to improve accuracy; andThe socio-economic questionnairewas altered to givewider earning bands.

    MAINODCON STUDYMeters and D ata Collection

    It should be remembered that the survey comprisedvolunteers who pay their water charges by rateablevalue but have agreed to have meters installed to theirJ.CIWEM, 1995,9, October

    water services. The meters have been mainly sited inthe highway at the boundary to the properties but,when shared Services were identified, some meterswere installed at the point where the services splitwithin the boundaries of the property-mostly in thegardens. However, 145 meters were installed insideproperties because it was impossible to fit these satis-factorily outside.Pulsed meters which provided a pulse for every litreof flow were chosen for the project. Three types ofdomestic meters were chosen:

    (i) Schlumberger P40s (installed outside properties);(ii) Kent pulsed output meters (installed inside propertiesat the stop-tap position); and(iii) Andrae Leonberg turbine (used inside homes at everypoint of use).The loggers which were used to collect the meter-reading data from the Schlumberger and Kent meterswere Newlog Universal loggers supplied by Tech-nolog Ltd. These loggers record the flow of water (inlitres) every 15 mins and can hold up to three monthsdata. The loggers are down-loaded to disk usingToshiba computers by the meter readers, then thedisks are sent to headquarters for processing. Data-validation rules have been established to determinewhether the retrieved data can be accepted automatic-ally or if the reviewer has to examine it first.For the Golden 100 properties, a Kent pulsed out-put meter was installed at the internal stop-tap locationas a reference meter, and Andrae Leonberg metershave been fixed in positions where they are at leastobtrusive to each appliance. There are, on average,fourteen meters in each household.It was important that the meters used in the Golden100 households did not have registers that the volun-teers could read and, as a result, influence the way theywould normally use water. All these meters arescanned by telemetry every 15 mins using MicroMedina outstations to hold the information, hen theyare downloaded using PSTN lines to Anglians head-quarters during the night. All the same data-validationrules apply to these households as to the other 1900; inaddition, the summation of all the small meters arechecked against the reference meter on a daily basis.

    Household SurveysTwo types of survey have been carried out with allthe volunteers; a property and appliance survey and asocio-economic survey. The property and appliancesurvey was carried out by the market research officerslisting and numbering all rooms in the property and allwater-using appliances in each room. In addition, theareas of garden which could be cultivated (or were setto lawns) were measured, as this would have an influ-ence on garden watering.A detailed socio-economic questionnaire wascompiled and carried out by the M OM organization.

    Questions included: numbers and ages of people in thehousehold, their qualifications and income, and mat-479

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    4/12

    EDWARDS AND MARTIN ONters of hygiene (for example, how many baths didpeople in the household tend to have each week). Theresponse was excellent: 94% of the householdsresponded to the interview without the need for furtherrequests.In addition to the surveys, all householders exceptthe Golden 100 were asked to com plete a diary oftheir water use for one week. This w as required to tryand m atch the water use against the m eter recordings.Clipboards with appropriate proformas w ere given tothe volunteers by the market research officers whenthey were carrying out the property and appliance sur-vey for each room in the house w hich had water-usingappliances. At the time the volunteers were asked tocomplete a w ater-use diary for one week starting thefollowing Monday. The market research officersexplained the reasons why the diaries were requiredand answ ered any queries which had been raised bythe volunteers.Computer Systems

    The computer system has been built in two parts: aninput system is used at division level and the centraloutput system is operated at regional headquarters.A personal computer (PC) based data system withoffice-based and portable PCs was set up at the start ofthe project to handle the market research officersrecords, and this has been extended to allow networktransfer of the data to headq uarters. The inpu t systemwas developed by a contract firm (Logica) and testedwith live data in March 1992. Data collection frommeter loggers is taken by portable PCs, and the systemhas been developed to allow collected data to beuploaded onto the office-based PCs and validated. Italso reviews and validates data collected by telemetry.The disk space required for SODCON is about 6.5gigabytes; approximately 3500 points are pro-cessed each day, wh ich amounts to 3500megabytes ofdata.Customer Care

    It was recognized at an early stage that the wholeproject would depend on customer care. Accordingly,it was decided that the m arket research officers wou ldbecome the volunteers contact for any m atter associ-ated with Anglian Water, not just SODCON. Theywould become a name and, more importantly, a face tothe volunteers. Each market research officer had atelephone and answering machine installed in theirhomes. At the o utset, when the mail shot and pamphlet(explaining the project) was sent out to the prospec-tive volun teers, the m arket research officers businesscard (with their office and home numbers) wasincluded, with the request that the vo lunteers contactthem at any time at the office or at their home s. About60% of the volunteers are only contactable out ofnormal working hours, therefore it is essential thatmarket research officers work flexible hours. Eachofficer is provided with a laptop computer whichallows them to work from their homes and input infor-480

    mation to the databases without travelling to theiroffices.The market research officers have been trained tocarry out socio-economic and social-class classifi-cation interviews, and it was essential that all inter-viewing was carried out in a professional manner. Allquestionnaire forms are hand-delivered by the m arketresearch officers so that face-to-face contact wit hth evolunteers is maintained. Payments are made eachyear to the volunteers in recognition of their co-operation and continued support.

    RESULTSAt present, data from April 1992 to O ctober 199 3 arebeing used for analytical purposes. Although therewill never be com plete data for the whole sam ple of2000 properties because of mechanical failures andsom e attrition, the samp le is of sufficient size to givecomplete confidence in the data accuracy andreliability.The results, covering the period to October 1993,represent the first fully validated set of data. As su ch,there are several areas of work, notably leakage,whe re analysis is (as yet) inco mp lete, but the m ainfindings are given below.PATTERNSF USE

    SODC ONs data for the distribution of pe r capitaconsumption (PCC) shows an average value of 145lihead. d, but, as wou ld be ex pected, there is a consid-erable range (Fig. 1 ). Mo nthly variations in total con-sumption are presented in Fig. 2, and the expecteddiurnal consum ption patterns have bee n recorded; anexample of this is presented in Fig. 3 . Weekly flowpatterns are show n in Fig. 4.A detailed analysis of flows has determined thatlegitimate night-time consumption (strictly speakinghouseholdnight use, provided that underground sup-ply pipe losses and plumbing losses can be excluded)in the reg ion is ap proxim ately 2.5 Uproperty. h, and isnormally codsistent throughout the week (Fig. 5) .RELATIONSHIPS

    Analysis of S ODC ON data have provided interest-ing data o n the following relationships between PCCand a num ber of variables:(a) Household size;(6) Housing type (ACORN groups);(c) Socio-economic groupings (income bands); and(d) Rateable value of properties;and these relationships are presented in Figs. 6-9.

    DETAILEDATTERNSF CONSUMPTIONAn analysis of the data provided by the Golden100 volunteer households has comm enced, and the

    JCIWEM, 1995,9, October

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    5/12

    A METHODOLOGY FOR SURVEYING DOMESTIC WATER CONSUMPTION

    Per capha consumption(1lhead.d)Fig. 1. Distribution of daily per capita consumption

    1504

    140I1204

    100Oct Nov Dec Jan Feb Mar Apr May Jun Jul Au g Sep

    MonthFig. 2. Daily per capita consumption (October 1992 to September 1993)

    '"I , ' '48J U W E M , 1995,9, October

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    6/12

    EDWARDS AND MARTIN ON30i

    O k0 6 12

    Time of day (houn)Fig.3. Diurnal pattern of water use

    18

    200

    150

    m 100A

    50

    0SUN MO N TUE WED THU FRI SAT

    Day of weekFig. 4. Mean daily PCC

    overall pattern of appliance use is shown in Fig. 10. Manyvariations underlie this overall pattern, as can be seen bythe comparison of two individual households (Fig. 11).domestic consumption is that of quality control. SouthWest Water have one of the country's longest runningcontinuous surveys of demand, and they have esti-mated(5) hat 15% of their data set significantly altersevery year, in terms of household size, occupationbands, or socio-economic status. The need to ensurethat data on the sample set are sufficiently up-to-date isan ongoing and paramount requirement, and there is acontinuous recruitment drive for new volunteers. Todate the current number of SODCON participants is1929.

    DISCUSSIONQUALITYONTROLOne of the most critical aspects of all surveys of482 JCIWEM, 1995,9 ,October

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    7/12

    A METHODOLOGY FOR SURVEYING DOMESTIC WATER CON SUMPTION3

    2.5

    2cEx 1.5e5

    1

    0.5

    0Sun Mon Tue Wed Thu Fri Sat

    DayFig. 5. Night-time consumption per property

    Naturally there has been some attrition to theSODCON sample. Approximately 7% of the samplehave already moved house and there have been birthsand deaths. In addition, a number of households haveelected to pay their water charges by volume, but newhouseholds will continue to be included to maintain thesample of approximately 2000 households. Regularup-date questionnaires will be asked of the volunteers

    to maintain accuracy of PCC and socio-economictrends.OF USE

    The figures for per capita consumption show theexpected wide range of values about the mean of 145l/head.d. All observations have been included in the

    250

    200

    150U1C

    100

    50

    0 I1 2 3 4 5 6 7 8Numberin householdFig. 6. Mean PCC by household size

    J.CIWEM, 1995,9,October 48 3

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    8/12

    EDWARDS AND MARTIN ON180

    160

    140

    120-Uu 10030 80al

    P60

    40

    20

    0A C D E F I J K

    ACORNFig. 7. Mean PCCby ACORN classification

    160

    155

    150

    U1 145f

    140

    135

    130LO W

    Household incomeFig. 8. Mean PCC by household income

    High

    I

    analysis, although some are clearly outliers. The comparisons with other studies. SODCON was origin-average value for PCC accords with the estimate for ally designed to meet Anglians needs alone, but it isdomestic consumption used by Anglian Water and reassuring that the data also appears to be represent-aligns with other companies data but, for the reasons ative of other areas.given earlier, it would be wrong to draw any close However, it must be recognized that SODCON pro-484 J.CIWEM, 1995,9,OLhber

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    9/12

    A METHODOLOGY FOR SURVEYING DOMESTIC WATER CONSUMPTION

    250-

    200-

    Lowest HighestRV bands

    Fig. 9. Mean annual household consumption by RV band

    vides actual consumption data, and these patterns areinfluenced by dynamic variables such as weather andthe economy.No-one would expect the period October 1992 toSeptember 1993 to be typical. The economic reces-sion will have had an impact on domestic demand, bothin terms of absolute levels of consumption and the tim-ing of water use. The summer of 1993 was spectacu-

    Toilets

    larly wet and dull. Very little, if any, garden wateringwas carried out during the first period of the SODCONsurvey. Rainfall in the periodJuly 1992 to March 1994has been 129% of the long-term average in East Anglia- a period of increased wetness of the same returnperiod (16&200 years) as the drought which precededit; therefore no inferences should be drawn about long-term domestic consumption from these data.

    Showers Baths4% 13%

    Externa l use3% Washing

    Toilet32%

    machine12%,p p ; yLuxury

    Miscellaneous35%Bathshowersw7%nd

    SODCON DATANOV 92 - OCT 93Fig. 10. Composition of water use

    LCIWEM, 1995,9, October

    WATER FACTS1992

    485

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    10/12

    EDWARDS AND MARTIN ONPROPERTYA

    Average peak flow(W

    PROPERTY B

    Timing of peak

    Washing machinehot2%

    Dishwasher mld*I

    19%Bath mid5% 2%

    TOTAL FLOW. 18 0 m3 TOTAL FLOW: 127 m3Fig. 11. Examples of water use by individual propert ies (November 1992-October 1993)

    TABLEII. AVERAGEOMPOSITIONFNIGHT-TIMESE (OlO(M400 HOURS)Washing machineDishwasherToiletBathShowerioutside tap

    49.9%29.9%19.5%0.6%Trace

    The higher-than-expected evel of legitimate night-time consumption led to hrther analysis. Although thestandard deviation of these results was high, as wouldbe expected (since some houses use no water at night,and others use a considerable amount), the largesample size (1624 for this exercise) means that thestandard error of the estimate is low(0.06Vproperty. h),and the sample mean lies within 5% of the populationmean with 95% confidence. An analysis of applianceuse at night shows that washing machines use half ofall water used (Table 11).An analysis of peak hourly flow indicates that theonly appliance which has its peak flow at night is thedishwasher, which may indicate a correlation betweenownership of these appliances and use of economyelectricity tariffs. Other appliances have their peakflows as might be expected from behavioural patterns(Table 111).RELATIONSHIPS

    The relationship between per capita consumptionand the various parameters presented in Figs. 6-9 areproviding useful infomation towards an understand-486

    ing of consumption patterns. The most significantfactor is household size, which also affects the vari-ability of per property consumption more than anyother single parameter. ACORN groupings indicatemaximum PCC values in classes A (agricultural vil-lages) and J (affluent suburban houses), and the lowestPCC is in ACORN class F (council)- all of which isnot surprising. The patterns of PCC against income andproperty RV also exhibit the expected trends, thehigher consumption in the lowest RV band (less than&40pa) being due to the relatively highper capita con-sumption in small, low-RV properties occupied by asingle person.EFFECTF METERINGN CONSUMPTION

    There is a wealth of opinion available on the effectof metering on consumption, but there is very little evi-dence. Overseas studies offer a wide range of possibil-ities but their relevance to UK, let alone the Anglianregion, is quesiionable. The National Metering Trialsreported that the effect ranged from a slight increase to

    TABLEIU. PEAKAPPLIANCELOWS ND THEJRIMINGApplianceTotal flowWashing machineDishwasherBathShowerToiletOutside tap

    32.09.85.210.15.412.61.2

    08:1510:4503:1518:4507:30OR0013:30

    ~~ ~

    JCIWEM, 1995,9, October

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    11/12

    A METHODOLOGY FOR SURVEYING DOMESTIC WATER CONSUMPTIONa fall of 17%; the average reduction in use was1 %().Evidence from SODCON has provided AnglianWater with a unique opportunity to develop robust esti-mates of the demand effect. It compares the consump-tion of approximates 1000 households who pay on ameasured tariff with the consumption of the unmea-sured SODCON households. The reported results (ofan implied estimate of 15%) should be regarded as pre-liminary, although they represent a considerableimprovement upon estimates from elsewhere, andfurther work will be carried out as more informationbecomes available.

    WAY FORWARDSODCON is already providing a vast amount of datafor analysis, but some areas still need to be developed.There is a need to examine distribution losses, andSODCON can provide, through its water delivereddata, excellent information on part of the water lossbalance. Further work needs to be carried out toensure compatibility between the meter readings onboth research meters and the individual propertymeters.Having successfully measured domestic water con-sumption, monitoring sewage flows from properties isone logical step forward. A complex prototypesampler/monitor has been developed and tried byAnglian Water, but it now needs to be installed in suf-ficient locations to provide representative data on netproperty water use.More applied research is needed to investigate thecomplicated inter-relationships between consumptionand such factors as weather, soil type, distributionwater pressure and geographical location, and theseanalyses will be continued during the remaining yearsof the project.

    CONCLUSIONS1. SODCON has already proved its worth in the qual-ity of data which it has been able to provide.Domestic consumption levels, patterns and par-ameters which influence them have been identified.Specific details such as night-time flows, key-appliance uses and combined flow-rates have beenmonitored and new areas of research have beenindicated.2 . The survey is also contributing significantly to pro-viding a better service to Anglians customers. Theinformation is being used as a tool to educate cus-tomers as to how water is used in the home and, by.implication, how it can be conserved. As an under-standing of components of waste and supply pipeand plumbing losses is being improved, advice canbe given to customers to improve the performanceof their supply systems. Information from SOD-

    CON can also help customers to influence their ownwater bills.3. The projects greatest asset, however, is undoubt-edly the enormous data base of (as yet) untappedinformation, which will be analysed and investi-gated in future years, assisting in such tasks as long-term forecasting, tariff setting and demandmanagement.

    REFERENCES(1) NATIONAL METERJNGRIALS ORKINGROUP.Water MeteringTrials: Final Rep ort. (Eds M. Hall and I. Kalsi.) WSA, WCA,OFWAT, WRc and DOE, 1993.(2) HOOPER,. A survey of domestic water consumption in Devonand Comwall. Proc. Royal Statistical Society C o n f , April1979.(3) THACKRAY,. E., COCKER, V. , AND ARCHIBALD, G. G. TheMalvem and Mansfield studies of domestic water usage. Proc.Instn. Civ. Engrs., 1978,64, (l),3741.(4) CACI NFORMATIONERVICES.981.(5 ) MOORE,C. J. Personal communication.

    DISCUSSION(Abridged)

    Mr R. Cook (Consultant), opening the discussion, asked ifreference had been made to the zonal metering studieswhich had been carried out by Anglian Water during1975-76, and whether there was any merit in revisiting thoselocations to compare consumptions.Mr B. Hayes (Consultant) said that, in his opinion, themain benefit to Anglian Water would be the reduction inconsumption of 15%, which would delay the need for capitalschemes. He enquired what the benefit to the customerwould be, and wondered whether the company would beconsidering differential tariffs for night-time water use.Mr S. Whipp (WRc Swindon) asked if SODCON hadprovided any information on the return period probability ofspecific levels of water consumption from groups of proper-ties, which might place peak demands on the distributionsystem.Mrs C. Ridgewell (Essex and Suffolk Water) requestedan outline of the methodology used to weight the SODCONsample results to provide a regional average figure for waterconsumption.Mr V. Ewan (Ewan Associates) asked whether the ong-inal premise of SODCON (i.e. that regional specifics neces-sitated a separate Anglian investigation of consumption)meant that the authors now believed that all water companiesshould undertake similar studies, or whether the Angliandata were transportable to other parts of the country.Mr M. Pocock (Three Valleys Water) questioned theallowance made for underground supply pipe losses on theTariff 1000 sample, which were used to derive the 15%reduction in consumption.Mr D. Evans (NRA)enquired when the authors would beable to provide estimates of consumptive use of water withinthe home, by means of their sewage flow monitor.Mr J. Smithson (Mott MacDonald) wondered whetherthe increased level of night flow would mean that AnglianWater would change its policy of mains rehabilitation as aresult of this alteration to the economic balance.Mr E. Butt (Water Quality Management Ltd) asked if

    487.CIWEM, 1995,9, October

  • 7/27/2019 A Methodology for Surveying Domestic Water Consumption

    12/12

    A METHODOLOGY FOR SURVEYING DOMESTIC WATER CONSUMPTIONdetailed consumption information was given to the Golden100households if requested, and also whether the 2000 vol-unteers would be charged for their water on a measured basisat any stage in the future, to examine the effect, if any, ofsuch a change on domestic consumption.Mr D. Harri s (Anglian Water) enquired if the authorswere satisfied with the accuracy of the meters in use, partic-ularly when summated in the Golden 100 properties, andMr W. Davidson (also from Anglian Water) wondered if thedetailed data available from SODCON could throw any lighton problems of meter under-recording.Mr M. Woolgar (W S Atkins Ltd) asked if the Golden100 properties were fully representative of the larger sam-ple. He also enquired whether the difference in percentagewater use by washing machines in the Anglian area, com-pared to Waterfacts data for the country as a whole, wasrelated to ownership or usage variations.AuthorsReplyIn reply to Mr Cook, the authors said that they had not yetattempted any measurements at the old zonal metering sites,largely due to their concern that the population base withinthe zones would undoubtedly have changed radically overthe past 20 or so years. The turnover of volunteers in theSODCON study was about 7% per annum, therefore theapplicability of the original locations and their data may notbe great.Responding to Mr Hayes, they confirmed that any defer-ment of capital schemes would allow, ceterispparibus,for afavourable effect on charges, which would certainly benefitthe customer. They commented that intelligent meters,capable of differential charges according to the time of day,were being investigated by Anglian Water in a separate studyto the SODCON project.In reply to Mr Whipp, the authors explained that they hadyet to cany out this particular analysis. In answering MrsRidgewell, they said that the general technique was to exam-ine the contribution to the variation of the dependent variable(consumption) provided by each of the independent vari-ables (e.g. the socio-economic,household size, and ACORNparameters which exhibited multi-collinearity) using step-wise multiple regression. Once the effect of each indepen-dent variable was known, the results could be weighted totake into account variations in characteristics of the samplewhich would vary through time, to arrive (hopefully) at astable estimate of PCC across the Anglian region as a whole.Responding to Mr Ewan, the authors felt that the detail of

    the data vindicated the use of the SODCON exercise withinAnglian Water. Moreover, because of this level of detail, andin particular the knowledge of the inherent variability of fac-tors affecting consumption, they believed that it should bepossible to apply the SODCON database to other parts of theUK and provide consumption information with a high degreeof confidence.The authors repeated that no allowance had been made forunderground supply pipe losses on the 1000 properties in thetariff study, and acknowledged to Mr Pocock that such lossesmight, if undetected, affect the reduction in consumptionthus estimated. It would, however, be relatively straight-forward to carry out a simple test to verify this matter.The authors explained that Mr Evans would have to waitseveral months for sufficient samples to be collected to beable to answer his query. Given the probable higher degreeof variability in water returned to sewer compared to waterdelivered, it was possible that the standard error of the esti-mate would be high. Replying to Mr Smithson, the authorsdoubted whether this increase in night flow, although great inpercentage terms, would materially alter the rehabilitationprogramme, firstly since the effect on water delivered wasnot yet proven and secondly because the prime driver forrehabilitation schemes was water quality, not quantity.Replying to Mr Butt, the authors confirmed that, onrequest, details of consumption patterns were sent to theGolden 100 volunteers. The transfer of customers fromunmeasured to measured charges did occur, if SODCONvolunteers elected to be charged on this basis; therefore thiseffect could be monitored using the existing database, ifnecessary.The meters used in the survey were Class D(Schlumberger P40s and Kent PSMTs) or Class B (AndraeLeonberg) instruments, and the authors assured Mr Harristhat they were confident that individual and cumulativeaccuracieswere acceptable. The square wave characteristicof flows through most point-of-use meters certainly assistedmeasurements; there were relatively small volumes passingat low flows. However, at this stage, the authors were unableto provide Mr Davidson with any information on character-istics of under-recording of meters from the SODCON trials.In their reply to the final questions, the authors confirmedthat the Golden 100 data were indeed representative of thelarger sample. They considered that the greater use of wash-ing machines in East Anglia was related both to the largermarket penetration of the appliance in the region and thegreat propensity of use.

    488 J.CIWEM, 1995,9, October