Effective Feedback Of Whole-Life Data to The Design Process

Transcription

1 Effecve Feedback Of Whole-Lfe Daa o The Desgn Process Mohammed Kshk 1*, Assem Al-Hajj 1, Rober Pollock 1 and Ghassan Aouad 2 1 The Sco Suherland School, The Rober Gordon Unversy, Garhdee Road, Aberdeen AB10 7QB, UK 2 School of Consrucon and Propery Managemen, The Unversy of Salford, Salford M7 9NU, UK * Correspondng Auhor Tel: Fax: Emal m.kshk@rgu.ac.uk 1

2 Key acves of he mplemenaon of whole-lfe cosng (WLC) as a managemen ool durng he occupancy sage are oulned. Then, he logc of a novel whole-lfe managemen approach s desgned around wo recenly developed generc WLC daabases. The proposed approach allows he sysemac daa collecon of he runnng coss of occuped buldngs and provdes he necessary lnk beween occuped buldngs and he desgn process. Besdes, employs a generc cos-sgnfcan relaon ha auomacally denfes he cos-sgnfcance of he buldng componens on he elemens, acves and cos ems levels over he analyss perod. Keywords: Cos sgnfcance, Whole-lfe cosng, Whole-lfe managemen. INTRODUCTION In a recen paper (Kshk e al., 2003), exsng applcaons ha provde whole-lfe cosng (WLC) suppor have been revewed. Ths revew revealed ha hese applcaons vary from free smple spreadshee models o sophscaed, commercal sand-alone applcaons. Besdes, hree man caegores can be denfed. In he frs caegory, he applcaon s used only as a fnancal ool o calculae he whole-lfe cos of a sngle alernave. Obvously, he usefulness of hs caegory s lmed. In he second caegory, ncludng mos of he applcaons, an applcaon s manly used as a decson-makng ool o denfy he deal alernave from a number of compeng alernaves. In he hrd caegory, he applcaon s used as an asse managemen sysem. Typcally, s manly a daabase manager ha has he capably o record, modfy, analyse and manage WLC daa for an asse. Applcaons whn hs class are commercal, general-purpose sysems ha would requre exensve ranng of users. Anoher lmaon of almos all exsng applcaons s ha he cos breakdown srucure (CBS) s bul manually by he user and s mosly non-elemenal. Besdes, varous faces of uncerany n WLC daa are no effecvely handled. Furhermore, none of he exsng applcaons provdes he necessary lnk beween occuped buldngs and smlar subsequen buldngs. 2

3 In anoher paper (Al-Hajj e al., 2001), he auhors have proposed a framework for mplemenng whole-lfe cosng (WLC) n he desgn model whn he OSCON negraed envronmen developed a Salford Unversy (Aouad e al., 1997). The man feaure of hs framework s ha employs wo daabases: a resource daabase and a projec daabase. The resource daabase houses daa for several opons for every buldng elemen whle he projec daabase accommodaes WLC daa for he opmum se of opons seleced a he desgn sage o be used hroughou he buldng lfe cycle. In a subsequen paper (Kshk e al., 2002a), he resource daabase has been desgned usng an elemenal cos breakdown srucure (CBS). In anoher paper (Kshk e al., 2002b), a compable srucure of he projec daabase has been desgned usng he concep of he buldng objec ha enables manpulang varous buldng componens n a praccal and convenen manner whn a CAD applcaon. The objecve of he research work ha underpns hs paper s o exend he mplemenaon of WLC o he occupancy sage. Ths wll be acheved by exendng he resource and projec daabases o allow collecng and recordng acual performance and cos daa of occuped buldngs and he developmen of a whole-lfe cosng managemen (WLCM) applcaon. In he nex secon, he basc requremens for an effecve WLC nformaon managemen durng he occupancy sage are oulned. Then, he projec and resource daabases are exended. Ths s followed by dervng a number of cos mercs o be used by he applcaon. Nex, he logc of he WLCM applcaon s nroduced. Fnally, he research work s summarsed and drecons for furher research are nroduced. KEY WLCM REQUIREMENTS The man objecve of whole-lfe cosng managemen (WLCM) s o assess and conrol coss hroughou he whole-lfe of he buldng o oban he greaes value for he clen (Flanagan e al., 1983; Seeley 1996). A relaed acvy called whole-lfe cosng plannng (WLCP) s o plan he mng of work and expendure on he buldng, akng no accoun he effecs of performance and qualy. A hrd acvy s whole-lfe cosng analyss (WLCA) ha ams o relae runnng coss and performance daa and o provde feedback o he desgn eam abou he runnng coss of occuped buldngs (Flanagan e 3

4 al., 1989). Thus, an effecve whole-lfe managemen sysem should have four capables: (1) recordng he acual performance and cos hsory of he buldng; (2) analyss of he recorded hsory feedng he accumulaed experence back o he desgn sage; (3) assessmen and conrol of coss hroughou he whole-lfe of he buldng; and (4) plannng he mng of work and expendure. These processes are dscussed n he followng subsecons. Recordng Performance and Cos Daa The man role of he projec daabase durng he occupancy sage s o record he acual performance and cos daa of he buldng,.e. o work as an asse regser. Two caegores of hs daa can be denfed: elemenal daa and non-elemenal daa. The frs caegory ncludes daa ha can be lnked o specfc elemens of he buldng;.e. daa represenng he hsory of varous acves of he buldng elemens, e.g. cleanng coss of exernal walls. Accordng o Ashworh (1996, 1999), he value of any recorded elemenal daa s lmed whou s conex nformaon as defned by a number of daa characerscs. He lsed some useful characerscs such as he manenance polcy beng appled, he cause of any componen falure, mng dsorons and he effec of delayed manenance acves. Thus, he elemens hsory should reflec no only he change n he predefned acves bu he daa of any new acves ha mgh emerge durng he occupancy sage as well due o change of he manenance polcy and/or he premaure falure or unsasfacory performance of some elemens. The second caegory ncludes oher daa ems ha canno be lnked o specfc physcal buldng objecs or nvolve he neracon of more han one objec lke energy consumpon. Obvously, almos all hese cos ems are relaed o operang he facly under consderaon such as energy, general raes, nsurance and secury coss. Energy coss of buldngs depend heavly on he use and hours of buldng sysems operaons, weaher condons, he performance level requred by owners, he buldng s desgn and nsulaon provsons, locaon, sze and poson (Krk and Dell Isola, 1995; Boussabane e al., 1999; Bordass, 2000). Ashworh (1999) lsed some facors affecng he raeable values of buldngs ncludng he locaon, sze, and amenes avalable. He poned ou also ha safey facors such as ype of srucure, maerals used and class of rade affec he nsurance premums and secury coss. 4

5 Analyss of Recorded Daa The recorded hsory of varous elemens of he buldng should be analysed o predc fuure acves and her assocaed coss whn he occupancy sage of he buldng (.e. feed-forward of nformaon) and o nform he desgn sage of oher projecs (.e. feedback of nformaon). Obvously, he feed-forward and feedback nformaon should be sored n he projec and resource daabases, respecvely. Durng each feed-forward process, fuure mngs and coss of varous acves are predced. In dong so, eher sascal or arfcal neural neworks (ANNs) me seres analyss models can be used. In he frs few years of he occupancy sage, he forecasng process wll rely manly on he orgnal daa se used n he desgn of he buldng. Evenually, he process wll rely on he daa ha has been generaed whn he projec self. The feedback process ams o provde he lnk beween occuped buldngs and smlar subsequen buldngs whereby performance and cos daa colleced durng he occupancy sage can be fed back o he desgn sage. In hs way, he desgn of new buldngs wll be based on real nformaon leadng o nformed WLC-based decson makng durng he desgn process. Whou hs feedback loop, however, hs process wll reman enrely based on arbrary assumpons or expecaons. Assessmen and Conrol of Coss To oban he greaes value for he clen, he coss and performance of varous acves need o be assessed and conrolled durng he whole-lfe cycle of he buldng. In dong so, wo man seps may be denfed. In he frs sep, areas n whch he coss of usng he buldng as dealed by he WLCA are denfed (Flanagan e al., 1989). Nong ha coss lke mos hngs - are unevenly dsrbued, Al-Hajj (1991) and Al- Hajj and Horner (1998) proposed o focus on a small number of cos ems known as he cos sgnfcan ems (CSIs). They developed a mehod o denfy CSIs usng a sascal approach based on he Pareo s 80/20 rule. Alhough her pon was o use he CSI dea o develop smple cos models o predc he runnng coss of buldngs, he dea can be used o denfy hgh cos acves as poenal sudy areas. However, a new defnon of hese ems s needed for elemenal ems. Rnala e al. (2001a, 2001b) 5

6 nvesgaed analogy-based esmaon (ABE) as a poenal approach o denfy CSIs. However, hey concluded ha ABE couldn be used a ha sage because s esmaon performance was poor. They arbued hs o he nadequacy and nconssency of he daa used n her sudy and emphassed he need for sysemac daa collecon. The second sep s o monor he denfed areas (elemenal and non-elemenal) and reassess her coss agans benchmarks and/or arges. The monorng and benchmarkng processes help o keep coss on arges or o modfy hese arges f necessary n he lgh of he accumulaed experence hrough he feed-forward and feedback processes. Work and Expendure Profles All WLC calculaons and plannng a he desgn sage are based on wo assumpons. Frs, elemenal manenance acves wll follow a planned manenance program o mee he sandards of manenance and lves of varous elemens of he buldng as rereved from he resource daabase. Secondly, all non-elemenal coss are based on characersc values from comparable buldngs or se by approprae expers. The acual performance and cos daa of he buldng should be recorded and analysed hroughou he occupancy sage o predc and plan for fuure acves. Falure o do so may resul n escalaed replacemen and manenance coss as well as premaure obsolescence (Krk and Dell Isola, 1995). Thus, s crucal o updae he nal manenance and replacemen profles and he esmaed projec cash flows esmaed a early sages of he projec. EXTENDING DATABASES The Projec Daabase The projec daabase has been exended o faclae recordng he acual daa of he occuped buldng under consderaon and o allow he feed-forward process dscussed earler. The orgnal srucure of he projec daabase allows elemenal daa o be recorded n a sraghforward way on he acves and cos ems levels usng wo ables. The frs able sores he hsory of he elemenal acves. The frs able sores nformaon ha unquely denfy each elemenal acvy and maps o a specfc buldng objec. Besdes, records wheher he acvy has been done or rescheduled, 6

7 he assocaed acual or rescheduled me and he reason behnd ha. Lkewse, he cos ems hsory able sores he acual cos of he elemenal acves. To allow sandardsaon of non-elemenal daa collecon and recordng, a codng sysem for hese coss was needed. Because non-elemenal coss are no assocaed wh specfc buldng objecs and are manly occupancy coss, he codes used n he BMI propery-occupancy-cos analyss form seems o be a logcal choce. Besdes, hese coss depend on he hours of use and he occupancy profle of he buldng especally for publc buldngs such as hospals and schools (Hobbs, 1977; Flanagan e al., 1989). Thus, wo ables have been ncluded n he projec daabase o sore non-elemenal daa and he occupancy profle of he buldng. Fgure 1 shows he srucure of he exended projec daabase. Fgure 1: The srucure of he exended projec daabase. The Resource Daabase The resource daabase has been also exended o allow he feedback process. Ths has been faclaed by he ncluson of hree ables. The frs able sores he man characerscs of buldngs whose cos and performance nformaon are fed-back o he resource daabase. For each buldng, a record s used o sore he buldng sze, hegh, locaon, age, and ype. The oher wo ables sore he fed-back non-elemenal and 7

8 elemenal daa, respecvely, as well as nformaon ha unquely denfy each record and he buldng ha generaed. Fgure 2 shows he srucure of he exended resource daabase. More deals of he exended resource and projec daabases wll be repored n wo separae papers. Fgure 2: The srucure of he exended resource daabase. COST SIGNIFICANT ITEMS The Cos Sgnfcance Relaon Based on earler work on cos sgnfcance (Poh and Horner, 1995; Horner and Zakeh, 1996), Al-Hajj and Horner (1998) defned a cos sgnfcan em as ha whose cos s greaer han he mean em value. For n cos ems, I, hs sgnfcance condon can be expressed as S = I C > n j= 1 C j n (1) 8

9 where S s he se of sgnfcan ems and C are coss of he ems. In her work, he denfed ses of sgnfcan ems from 5 ses of daa consued beween 75% and 89% of he oal coss n hese ses. Because hs defnon provdes a smple procedure o denfy he sgnfcan ems n any bach of cos daa, wll be adoped heren wh he followng slgh modfcaon. S = I C n j= 1 C j n (2) Ths modfed relaon ndcaes ha a cos sgnfcan em s ha whose cos s greaer han or equal he mean em value. I s mporan o noe ha hs s no o descrbe an average cos as a sgnfcan cos. Raher, hs modfcaon s necessary because f all he ems have he same cos, no em would be denfed as sgnfcan usng relaon 1. The modfed relaon, however, needs o be generalsed such ha sgnfcan acves, elemens and objecs can be denfed as well. Dvdng boh sdes n he above nequaly by n C j j = 1, relaon can be expressed as S = I SR ST (3) where SR are he ems cos raos gven by SR = n C j = 1 C j (4) and ST s a sgnfcance hreshold gven by 1 ST = (5) n The sgnfcance relaon 3 ndcaes ha an em n a gven daa se s sgnfcan f s cos rao exceeds a hreshold gven by he recprocal of he number of ems of ha se,.e. he cos rao gven by equaon 4 may be regarded as a sgnfcance measure. Thus, relaon 5 can be used o denfy he sgnfcan acves of gven elemen, he 9

10 sgnfcan elemens of a gven objec and he sgnfcan objecs of he buldng by comparng her cos raos wh he sgnfcance hresholds of he elemen acves, he objec elemens and he buldng objecs, respecvely. The Cos Measure Anoher ssue has o be addressed: wha s he cos measure o be employed n he calculaon of cos raos? In her mehodology, Al-Hajj and Horner (1998) used a deflaed cos measure whereby all cos daa were deflaed o a common year usng he nflaon ndces publshed by he BMI. In hs way, he effec of nflaon was accouned for. They dd no, however, dscoun coss o accoun for he me value of money. Ths seems reasonable because he man purpose of her work was o develop models o predc he annual runnng coss of buldngs and he dscounng process mgh have no effec on her mehodology because all coss ncurred a he same year wll be dscouned wh he same facor. Because he man purpose of he WLC managemen of buldngs s o oban he greaes value for he clen, he choce of a whole-lfe cos measure, e.g. he ne presen value, s logcal. Besdes, should reflec any change of he sgnfcance of an em over he analyss perod. Two measures ha sasfy boh requremens can be denfed: he cumulave WLC conrbuons and he remanng WLC conrbuons. The cumulave WLC measure, cash flows CWLC, of a cos em C, represened by a number of C, s he summaon of all dscouned cash flows whn he perod from he nal me, = 0, o he presen me, p,.e. CWLC = p = 0 PWS C (6) where PWS s he dscoun facor a me gven by PWS 1 = (7) (1 + r) The use of he CWLC measure o calculae he sgnfcance rao of an em (Eq. 5) would reflec he pas sgnfcance of an em. 10

11 The remanng WLC measure, RWLC, of an em s he summaon of all dscouned cash flows whn he perod from he presen me, p, o he end of he analyss perod, T,.e. T RWLC = PWS C (8) = p The use of hs measure o calculae he sgnfcance rao of an em (relaon 5) would reflec he fuure sgnfcance of an em. Ths s n lne wh he man objecve of he plannng process. Thus, he RWLC wll be employed o calculae he sgnfcance rao as SR = n RWLC j= 1 RWLC j = T PWS C = p n T PWS C j= 1 = p j (9) DESIGN OF THE APPLICATION The WLCM applcaon has been developed n hree modules: WLCPre, WLCPro and Feedback as shown n fgure 3. Resource Daabase Projec Daabase WLCPre A WLCPro Feedback Fgure 3: Smplfed process flow dagram of he WLCM applcaon. 11

12 The WLCPre Module The WLCPre module rereves he recorded daa from he projec daabase and generaes he nformaon requred for he WLCPro and Feedback modules as follows: A connecon o he projec daabase s esablshed. The objec elemens, acves and cos ems daa are rereved. Each cos em s expressed as a number of cash flows (CFs) sored n a wo dmensonal CF marx. The characerscs of each cos em are sored n a marx for laer use. For an elemenal em, hs daa ncludes he em s descrpon, recurrence, rae, un and generc codes and s objec, elemen and acvy numbers. For a non-elemenal cos em, hs daa ncludes he em s recurrence, generc and BMI codes. The elemenal and non-elemenal hsory daa are rereved o updae he cos ems CF marces hrough a me seres model. The updaed elemenal and non-elemenal marces are combned no wo elemenal and non-elemenal cash flow marces, ECF and NECF, respecvely, whch are hen carred forwards o he WLCPro and Feedback modules. A smplfed process flow dagram of he WLCPre module s shown n fgure 4. Projec Daabase characerscs cos daa cos daa characerscs CF marces CF marces hsory daa Updaed ECF marx Updaed NECF marx hsory daa A Fgure 4: A smplfed process flow dagram of he WLCPre module. 12

13 The WLCPro Module The WLCPre module generaes varous plannng profles and denfes sgnfcan ems of he buldng. Ths s done n he followng seps (fgure 5): Equaon 7 s used o calculae he dscoun facors a varous years of he analyss perod. These facors are sored n a wo dmensonal marx DF. A profles DCF marces DCF marces profles sgnfcance DF marx sgnfcance Projec Daabase Fgure 5: A smplfed process flow dagram of he WLCPro module. The non-elemenal dscouned cash flow marx (NEDCF) s obaned by dscounng he NECF marx usng he DF marx. Then, equaons 6 and 8 are employed o calculae he non-elemenal CWLC and RWLC marces. Equaon 11 s used o calculae he non-elemenal sgnfcance facors from he non-elemenal RWLC marx. Then, he non-elemenal sgnfcan ems are denfed usng he sgnfcance relaon 5. Usng he elemenal characerscs marx, CF marces of he curren objec are consruced on he generc; elemen and acvy levels and her correspondng DCF are calculaed usng he DF marx. Then, equaons (8 and 10) are employed o calculae he CWLC and RWLC marces on he generc, elemen and acvy and cos em levels. Equaon 9 s used o calculae he elemenal sgnfcance facors on he generc, elemen and acvy and cos em levels from her correspondng RWLC marces. Then, he cos sgnfcan ems, acves and elemens are denfed usng he sgnfcance relaon 5. 13

14 The Feedback Module The Feedback module auomacally feeds back he buldng performance and cos daa o be used n oher projecs. Ths s done n he followng seps (fgure 6). Projec Daabase A Curren objec daa Buldng characerscs hsory daa hsory daa Feedback Non- feedback CF marces Resource Daabase Fgure 6: A smplfed process flow dagram of he Feedback module. The buldng physcal and general daa are rereved from he projec daabase. A connecon o he resource daabase s esablshed. The resource daabase s checked for an exsng record for he buldng. Oherwse, a new record s nsered. The records of elemenal coss for he curren year are rereved from he projec daabase. Each cos em s ransformed o a cos rae hrough nerpreaon wh he correspondng objec/buldng physcal daa. Then, a record s nsered n he correspondng able n he resource daabase. The records of he buldng non-elemenal daa for he curren year are rereved from he projec daabase. Then, a record s nsered n he correspondng able n he resource daabase. The connecon o he projec daabase s closed. The connecon o he resource daabase s closed. A complee process flow dagram of he applcaon s shown n fgure 7. 14

15 hsory daa Curren objec daa feedback Projec Daabase cos daa cos daa CF marces characerscs characerscs CF marces Updaed ECF marx sgnfcance sgnfcance Updaed NECF marx DCF marces profles profles DCF marces DF marx Resource Daabase Fgure 7: The complee process flow dagram of he applcaon. hsory daa Buldng characerscs feedback 15

16 SUMMARY AND THE WAY FORWARD Followng a bref revew of exsng applcaons ha provde whole-lfe cosng suppor, he basc processes of effecve whole-lfe managemen of buldngs have been explored. The frs process s o record he acual performance and cos hsory of he buldng. The second process s o analyse hs recorded daa o predc fuure acves and her assocaed coss whn he occupancy sage of he buldng (.e. feed-forward of nformaon) and o nform he desgn sage of oher projecs (.e. feedback of nformaon). The feed-forward and feedback nformaon are sored n he projec and resource daabases, respecvely. The hrd process s o assess and conrol coss whereby he man acvy s o denfy cos sgnfcan ems. The fourh process s o produce varous work and expendure plannng profles. Two recenly developed WLC daabases have been exended o faclae he frs wo processes. Then, he logc of a WLCM applcaon has been developed o auomae he oher wo processes. The proposed approach has wo unque mers. Frs, pus forward a sysemac framework for daa collecon and recordng usng a well-defned cos breakdown srucure and keepng s conex nformaon. Secondly, offers a sarng pon for a praccal feedback mechansm ha lnks occuped buldngs o he desgn process. The proposed approach, however, handles elemenal and non-elemenal coss ndependenly. I should be developed furher o cover he neracon of varous buldng elemens and operang coss of he buldng. Besdes, he ransformaon of acual coss o cos raes s done hrough a predefned code ha denfes whch objec/buldng physcal quany o use. In he fuure, a procedure should be ncluded o denfy whch physcal quanes have he greaes correlaon wh acual coss. ACKNOWLEDGEMENTS The research work ha underpns hs paper has been manly funded by he EPSRC gran number GR/M83476/01 underaken jonly by wo eams from he Rober Gordon 16

17 Unversy and he unversy of Salford. The auhors would lke also o hank wo anonymous referees for her helpful commens and suggesons REFERENCES Al-Hajj, A (1991) Smple cos-sgnfcan models for oal lfe-cycle cosng n buldngs. Unpublshed PhD hess, Deparmen of Cvl Engneerng, Unversy Of Dundee. Al-Hajj, A. and Horner, M.W. (1998) Modellng he Runnng Coss of Buldngs. Consrucon Managemen and Economcs, 16, Al-Hajj, A., Pollock, R., Kshk, M., Aouad, G., Sun, M. and Baks, N. (2001) On he requremens for effecve nformaon managemen n whole-lfe cosng whn an negraed envronmen. In: Kelly, J. and Huner, K. (Eds.), The Annual Conference of he RICS Research Foundaon, 3-5 Sepember, Glasgow Caledonan Unversy. RICS, Vol. 2, Aouad, G., Chld, T., Marr, F. and Brandon, P. (1997) OSCON fnal repor. Unversy of Salford, UK. Ashworh, A. (1996) Esmang he lfe expecances of buldng componens n lfe cycle cosng calculaons. Srucural Survey, 14 (2), 4-8. Ashworh, A. (1999) Cos sudes of buldngs. Longman, Harlow. Bordass B. (2000) Cos and value: fac and fcon. Buldng Research & Informaon, 28 (5/6), Boussabane, A.H.; Krkham, R.J. and Grew, R.J. (1999) Modellng Toal Energy Coss of Spor Ceners. Facles, 17(12/13), Flanagan, R., Norman, G. and Furbur, D. (1983) Lfe Cycle Cosng For Consrucon, Surveyors Publcaons. Flanagan, R., Norman, G., Meadows, J. and Robnson, G. (1989) Lfe Cycle Cosng - Theory and Pracce. Blackwell Scence Ld. ISBN: Hobbs, S. (1977) Collecon and use of buldng manenance cos daa. Buldng Research Esablshmen (BRE), Deparmen Of Envronmen, Curren Paper Ocober, Horner, M. and Zakeh, R. (1996) Characersc ems - a new approach o prcng and conrollng consrucon projecs. Consrucon Managemen and Economcs, 14, Krk, S. J., and Dell Isola, A. J. (1995) Lfe Cycle Cosng for Desgn Professonals. McGrew-Hll Book Company, New York. Kshk M., Al-Hajj A., Pollock R., and Aouad A. (2002) A generc daabase for wholelfe cosng applcaons n consrucon. Proceedngs of he 18 h Annual Conference of he Assocaon of Researchers n Consrucon Managemen (ARCOM 2002), Unversy of Norhumbra, 2-4 Sepember 2002, Vol. 1, Kshk M., Al-Hajj A., Pollock, R., Aouad A. and Baks, N. (2002) A whole-lfe cosng applcaon for he opmum desgn of consrucon asses. Proceedngs of he Annual Conference of he RICS Research Foundaon (COBRA 2002), Nongham Tren Unversy, 5-6 Sepember 2002, Kshk M., Al-Hajj A., Pollock R., Aouad A., Baks, N. and Sun, M. (2003) Whole-Lfe Cosng n Consrucon - A Sae of The Ar Revew. The RICS Research Paper Seres, Vol. 4, No

18 Poh, P. and Horner, R. (1995) Cos-Sgnfcan Modellng Is Poenal for Use n Souh-Eas Asa. Engneerng, Consrucon and Archecural Managemen, 2 (2), Rnala, K.; Kumar, B.; and MacLeod, I. (2001a) A framework for analogy based esmaon n buldng servces. Proceedngs of he Sxh Inernaonal Conference on he Applcaon of Arfcal Inellgence o Cvl and Srucural Engneerng, Sepember 19-21, Venna, Ausra, Rnala, K.; Kumar, B.; and MacLeod, I. (2001b) Analogy based esmaon n buldng servces. Proceedngs of he CIB-W78 Inernaonal Conference on IT n Consrucon n Afrca 2001, 30 May 1 June, 2001, Mpumalanga, Souh Afrca. Seeley, I. H. (1996) Buldng Economcs. Fourh Edon, Macmllan Press Ld., Basngsoke. 18