ENGINEER - Vol. XXXXI, No. 04 pp. 22-31, 2008© The Institution of Engineers, Sri Lanka

Maintainability Risk Factors of Flat Roofs inMulti-Storey Buildings in Sri Lanka

Nayanthara De Silva, Malik Ranasinghe and R. Rameezdeen

Abstract: The research into the issue of maintainability of multi-storey buildings in Sri Lanka isstill in its adolescent stage. One of the critical building elements that requires immediate attention formaintainability is flat roof areas. The flat roofs are often subjected to alternate drying and wettingcycles under tropical conditions, causing many defects and subsequent deterioration when properdetailing related to design, construction and maintenance actions are lacking. The inherent risks ofmaintainability of the buildings can be identified by analyzing their defects causing factors. In thisresearch, 12 such risk factors related to the maintainability of flat roofs were identified. Further, ascoring system using Artificial Neural Networks (ANN) is developed to forecast the level ofmaintainability, which is projected, based on risk analysis. The level of maintainability of a typical flatroof is shown as 51%. The risk factors are also prioritized to give guidance.

Keywords: Maintainability, Risk analysis, Sensitivity analysis, Flat roofs, Neural Networks

1. Introduction

The concept of maintainability was initiated bymilitary services in USA in the 1950s [1] as wellas in UK in the 1960s [2-3] and then broadenedall over the world. Currently it generates asufficient portion of the total constructionoutput all over the world especially indeveloped countries. This is due to the fact thatthe maintenance budgets are increasedsignificantly due to various factors like poordesigns, inferior construction and poor planningfor maintenance. hi certain instances themaintenance cost resulting from the above factorscan even amount to a half of the total cost of theoriginal construction itself [4]. These increasedcosts are reported due to the growing complexityof buildings, lack of integration betweendifferent stages such as design; construction andmaintenance, and budget and time constraints[5-6]. According to the data available fromdepartment of Census and Statistics in SriLanka, the proportion of the maintenarp :e cost isaround 17% of the total expenditure on buildingconstruction related activities [7].

However, in general practice, there is t.:> budgetallocation for the maintenance in most of thebuildings in Sri Lanka. This becomes a seriousissue when more than one client is occupying asingle property (eg. apartments orcondominiums) because individual owners maybe reluctant to take responsibilities or use theirown money when a common area like roofincurs a problem. With the lack of fundingallocated for the maintenance, rapidly

mushrooming condominiums in Colombo, andits suburbs and other cities in Sri Lanka maysoon become a heavy strain to the facilitiesmanagers and the users if this is not properlyhandled.

In the Sri Lankan construction industry, theearly allocation of necessary provisions towardsfuture maintainability is not considered as acommon practice. This practice is likely to causeserious issues in the near future especially withrespect to the large numbers of apartmentbuildings, condominiums and shoppingcomplexes. As a result, these buildings are likelyto suffer from poor maintenance causingsignificant issues to their owners, users,managers and regulatory and safety authorities.

2. Implications ofMaintainability Practices

Poor

Many researchers have discussed the generatorsor causes of maintainability problems fromdeficiencies under three main divisions [5, 8-17]as,

Mrs. Nayanthara De Silva, BSc Eng., MSc (Building),Senior Lecturer, Dept. of Building Economics, University ofMoratuwaEng. (Prof.) Malik Ranasinghe, BSc Eng., MASc, PhD,CEng, FIE(Sri Lanka), Vice Chancellor, University ofMoratuwaDr. R Rameezdeen, BSc QS, MEng(Const.Mgt.) PhD,AA1QS, AIQS(SL), Head, Dept. of Building Economics,University of Moratuwa

ENGINEER 22

• Causes initiated during the design stage.• Causes initiated during the construction

stage.• Causes initiated during the usage stage (ad-

hoc maintenance practices)

It is however identified that all thesedeficiencies can be planned for avoidance orcontrol during the design stage. Owing to this,it requires taking an integrated approachthroughout the design, construction andmaintenance phases of a building [18].Unfortunately in the normal contract system,designers of buildings rarely have long terminterest in the buildings they produced; hence,they become divorced from the maintainabilityproblems that follow from their poorinvolvement. On the other hand, involvementof a facility manager or a maintenance managerduring the design process and the constructionprocess is almost negligible in the localindustry.

Therefore under these prevailingcircumstances, there are many existing risks.The implications of buildings, from such riskscould be economic implications, prestige,interest of public and users and, effect onmaterial. Economic implications can be causedfrom the increased maintenance budget due todeficiencies arising during the above threestages. A large portion of this budget may beused for repairs or replacements of componentsdue the rapid deterioration under theprevailing conditions. The rate of deteriorationwill depend on the sensitivity of the componentfor a deterioration agent (risks parameter) andthe quantity of a deterioration agent at thelocation of a component [19]. In this research,721 defects/problems related to flat roofs of 50buildings arisen from above causes wereobserved and listed into 16 major types (Table1).

3. Research Method3.1 Field SurveyPhase 1: Fifty multi-storey buildings (above 4storeys) were randomly selected to identifymaintainability problems of flat roofs. Thesecompromised residential buildings (46%),commercial buildings (30%) and officebuildings (24%). The age group of the buildingsample is spanned from 5 years to 25 years. Acondition survey was carried out to explore theexisting defects and their extent. In this survey,the conditions were clearly defined with the aidof photographs to illustrate very severe, severe,moderate, law, very low conditions of buildingdefects. This survey was conducted during the

monsoon season; June-July, in order to capturemore problems caused under the tropicalenvironment.

The defects identified from the conditionsurvey were analyzed in detail to identify theircauses. The maintenance teams' knowledge andexperiences in this field were elicited in thisexercise. However, for those buildings in whicha permanent maintenance team was notavailable, this exercise was conducted with thehelp of residents and experts from otherbuildings. Further, a clear picture of theproblem was shown to these experts with theaid of photographs and in certain cases, withsmall video clips of the sites.Phase 2: Residence managers or maintenancemanagers were interviewed to elicit the existingrisk conditions of their buildings. Structuredsurvey forms were used in this exercise.Phase 3: The actual maintenance costs of thesebuildings were worked out from the rates ofcleaning and operational costs obtained fromthe maintenance managers. Further, differentrates for retrofitting of those identified defectswere obtained from individual quantitysurveyors and several organizations in order tocalculate retrofitting costs of building wherethese values were not available.

3.2 Prototype Scoring SystemThis model is developed on the basis ofmaintainability risk analysis. Themaintainability risk is evaluated through thecauses of problems and defects occurringduring the maintenance stage. The impact ofthese problems and defects are analyzed basedon the loss of performance of the building bymany authors [20-22]. Similarly, someresearchers have measured the magnitude ofbuilding maintenance by using cost as thecriterion [23-24]. Further, Chew et al [25] havedeveloped a prototype scoring system toevaluate the level of maintainability ofbuildings at the design stage, combining thesetwo approaches.

In this research, the framework of the scoring systemdeveloped by Chew et al., [25] was used to develop aprototype model for flat roof maintainability. Chew'sframework was used here due to its relevance undertropical conditions. Artificial Neural Networktechnique was used to develop the model.Defect/problem causing factors identified from thefield survey were taken as inputs (i.e. maintainabilityrisk factors) for the model. The output of the modelwas worked out based on the deviation from theoptimum level of maintainability, which wasmathematically expressed by Chew et al., [25].

23 ENGINEER

Table 1: Flat Roof Defects

o201

0203040506

07

08

09

10

111213141516

tn•t-»o•a0)D

Splitting

Blistering of paint filmWater poundingSlippageCrocodiling or AlligatoringGravel scouringBlistering of waterproofing -membranesystemBulging effect of sealantAdhesive failure of sealant frombiological growth

Joint failure

Crack on roof slabDeflection of roof slabChokageCrack at roof drainageWater leakage through the slabMechanical damage

.133QJD

*

*

*

*

*

*

*

*

Con

stru

ct!

on

*

****

*

*

*

**

**

Mai

nten

ane

e

*

*

*

*

****

*

>w-wcds*

*

*

*

*

*

*

*

*

*

M_i C/l

6*O VJ-<1-7 QJ^ -0

64

5856203350

34

32

32

43

665866383833

^£

OSPi

3

4616127

11

14

14

8

1

419912

4. Flat Roof Maintainability RiskFactors

Maintainability of buildings is more or less alocalized issue that depends on several localfactors like climatic conditions, architecturalpractices, construction materials and techniquesetc. and hence it is important to identifysignificant risk factors confined to existingconditions. However, the maintainability risks offlat roofs could be looked at under three mainstages:

1. Design Risk- Causes initiated during the design stage -water-tightness, stress accommodation,water flow control, accessibility,environment related and materials'performance, etc.,

2. Construction Risk- Causes initiated during the constructionstage - quality of construction

3. Maintenance Risk- Causes initiated during the maintenancestage - quality of maintenance

Under above three stages, 12 significant riskfactors were identified for roof's maintainability.

These risk factors were evaluated using 18parameters (Table 2).

5. Risk Analysis

Once significant risk factors have been identified,they can be analysed in depth. Such risk analysisis concerned with the severity of each riskattribute to maintenance life cycle cost. Over thepast years, several formal techniques have beendeveloped to facilitate the risk analysis. They areranged from simple techniques such aselementary risk analysis up to more sophisticatedtechniques that involve modern concepts likeapplication of expert systems and neuralnetworks [26]. However, the inherent complexityof the maintainability, sophisticated expertsystems like neural networks are useful tools tomodel the underlying function to grade the levelof maintainability confined in a design [27-29].

In this model, risks parameters are converted intoa homogeneous scale using a discrete scale(rating) between 1 and 5 to represent the level ofrisk for each factor. A value closer to 1 signifiesgreat risk while a value closer to 5 represents anegligible risk.

ENGINEER 24

Table 2: Flat Roof Maintainability Risk Factors(1) Risk Factors

1 . Detailing of concrete slab

2. Waterproofing systemselection

3. Waterproofing detail

4. External drainage systemselection

5. External drainage systemdetails

6. External drainage systemmaintainability

7. Material selection

8. Material detail

9. Material maintainability

10. Construction quality

11. Maintenance practice

12. Building profile

(2) Parameters

1. Strength

2. Concrete cover

3. Joint detailing

4. Suitability

5. Reliability

6. Design/ Application detail -joints, penetrations,projections, etc.,

7. Type of system

8. Sufficient capacity

9. Extent of maintenance workrequirements.

10. Durability

11. Sustainability

12. Cleanability

13. Inspection

14. Workmanship quality

15. Method16. Efficiency

17. Age18. Height

(3) Elements

•3* Adherence to typical design /and detailing

* Type of protection mechanism

•I* Adherence to typical design /application considerations anddetailing

Life span

Testing requirements

Cleaning requirement

Inspection requirement

Financial

Technical

Human resource

Quality management

Maintenance inspections,equipments and chemical usedfor cleaning

* Frequencies of inspection andcleaning cycles

5.1 Risk of Roof Slab Failures

Failure to achieve dense, sound andhomogeneous concrete slab could lead to manydefects increasing maintainability problems of flatroofs. Owing to this, the selection of properconcrete mix design, concrete cover, jointdetailing can be considered during the designstage to achieve a good performing structure.Having said that, the grading for the performanceof a slab (1st risk factor) was establishedconsidering the above parameters (i.e. concretemix design, concrete cover, joint detailing, etc.,),using 1-5 scale.

5.2 Risk of Waterproofing Failures

The failure to achieve the water tightness is themost significant problem in flat roofsmaintainability. The water tightness of a flat roofrelies mainly on the integrity of the structure,waterproofing design detailing including theadequacy of waterproofing over penetrations,projections, and detailing over joints. For theselection of waterproofing systems determines thedegree of protection required while the method ofconstruction limits the type of protection systemsto be chosen (2nd risk factor). The quality ofapplication also determines the eventual degree ofwater tightness attained. With the proper

25 ENGINEER

selection of the waterproofing system,appropriate design detailing and soundworkmanship during application, the watertightness in flat roof could be improved andprolonged.

The waterproofing detailing is another criticalfactor in achieving the water tightness. In thisregard, joints are vulnerable points prone to wateringress and should be paid greater attention inproviding detailing over joints, penetrations, andeven projections. Proper detailing in the designwill minimize the risk of such lines of weaknesswhile inadequate detailing in the design willcause great risk and results in repetitiveoccurrence of defects at these critical points (3rdrisk factor).

5.3 Risk of External Drainage System Failures

Although the roof structure must be designed toprevent any ingress of surface runoff, this can notbe completely avoided. As such, there is a needfor a good flow drainage system to ensure that theroof is protected from this external source ofwater (4th risk factor). The surface drainagesystem should adhere to the local designguidelines for efficient removal of the rain waterrun off from the structure (5th risk factor). Being atropical country with frequent rainstorms, theroof requires more regular care and maintenanceto the drainage systems to prevent pipe clogging,overflows, leaks and discharge interferences.Therefore the extent of maintainability of suchsystems would be equally important as its details.In this regard, accessibility for maintenance isidentified as the most important consideration (6th

risk factor).

5.4 Risk of Material

With good workmanship, appropriate materialdetailing and with proper maintenance, a selectedmaterial should perform up to its expectedlifespan. However, the actual lifespan andperformance can be undermined by theinteractions with other materials in the structure.Further, careful consideration is needed to ensurethe durability of the selected material under itsprevailing environment, so that the overalllifespan would not be compromised due topremature failures (7th risk factor).

The selected roof materials should have the abilityto resist defects resulting from external factorssuch as moisture, rain, chemicals, etc., in order toperform its intended function through its design

life. The inherent properties of materialsdetermine the level of sustainability of theirfunctional purposes under various exposureconditions. Therefore performance of materialsunder such conditions can be assessed throughtesting and approvals for properties pertinent tomaintainability (8th risk factor). Here, materialscan be tested or checked under local standards,manuals or reputable international standards forcommon consequences given by the tropicalclimate as, (1) resistance to cracking, (2)resistance to dampness, (3) resistance to chipping,(4) resistance to staining such as biologicalgrowth, sealant, and staining caused byincompatibility of material interfaces, and (5)resistance to chemical attacks.

During the service life, the frequency ofmaintenance cycles that has to be required yearlyfor maintaining selected materials for thestructure represents a large proportion of themaintenance budget that would be incurred bythe building owner. Therefore, the maintainabilityof such materials in order to meet aesthetical andfunctional performance requirements should beeasy with lesser requirements of cleaning andinspections (9th risk factor).

5.5 Risk of Construction Quality

The construction quality, also known as theworkmanship, has been identified as one of themost crucial causes in building failures by manyindustry professionals (10th risk factor). Findingsfrom interviews carried out for this studyhighlighted the main issues on unskilledlabourers and cost-driven contractors which ledto negligence in workmanship. Further, thecontractors should be encouraged to participateon the scheme so as to improve their quality andcompetitiveness to enhance the firms' image andprofessionalism. Clients and developers shouldalso be encouraged to specify the use ofaccredited contractors for their projects to ensurethe level of construction quality achieved.Therefore construction quality can be graded as,

Constructbn QualityScore=

(1)

Where CQ

CQTech

CQHR-

financial quality

technical quality

human resource quality

ENGINEER 26

CQ - quality management

5.6 Risk of Maintenance Practice

All buildings deteriorate through aging undervarious exposure conditions. However, they haveto maintain their performance throughout theirdesign lives. The management of this processthrough corrective maintenance needs extensivecapital expenditure. But, this should beminimized whilst ensuring maximumperformance of buildings throughout its life. Thisinvolves the development of routine andpreventive maintenance policies and proceduresto plan for the repair and replacement of buildingcomponents. In general, proper maintenanceprocedures should be able to upkeep thedurability and the performance of the material.Therefore, flat roofs having a plannedmaintenance strategy that includes regularinspections, cleaning and corrective repair andreplacement works can be awarded with a highergrading (11th risk factor).

5.7 Risk of Building Profile

A building's profile such as its age and heightaffects its maintainability. Low-rise buildings areassumed to take less effort in maintaining thantall buildings. The age of the building on the otherhand may dictate the frequency of occurringdefects. In addition, the durability of flat roofelements that have been repaired, in olderbuildings may require additional effort, comparedwith new buildings. Therefore, the height and ageof a building should be factored in and adjustedin the determination of flat roof maintainability(12th risk factor).

6 Risk Management

Risk management in the maintainability can beexpressed as controlling of the LCC associatedwith the maintenance. In the maintainability riskmanagement, avoiding or minimizing byadopting safety measures in the design,construction or even maintenance practices couldcontrol the LCC. In other words, the inherentmaintainability risk should be carefully analysedand controlled to minimize the LCC associatedwith maintenance. Therefore, a model which canbe used to indicate or predict the maintainabilityof buildings can be designed as a riskmanagement tool. In this research, a model isdeveloped to predict the level of maintainabilityof flat roofs which is estimated by analysing the

existing risk conditions identified from thesurvey.

6.1 Outline of the Proto-type Model

To assess the level of maintainability, a modelusing multi-layered perceptron neural networkswhich is known by the name of the algorithmused to train it, back propagation, was developed.The model's framework was designed to computethe level of maintainability as an output inresponse to input parameters that reflect theseverity of inherent risk of flat roofmaintainability. Mathematically, it can beexpressed as a mapping function:

fE : SR i-» M (2)

where domain 9? is a set of nth dimensional

vector [r},r2,..rn] of risk inputs representing a

risk domain of the maintainability, and M is a setof scalars providing points to an underlyingmaintainability function.

6.2 Input Risk Parameters

In this model, 12 significant risk factors of flat roofmaintainability as illustrated in Table 1 weretaken as inputs to the network. The equation 4summarises the quantification of risk factors.

Supposed that K sets of sub-grades denoted by

[Fn]k,m = l,...M,k = \,..K is established to set

the alternatives of each factor. The overall level of

seriousness of a risk R, is then modeled by,

R, =N

(3)

where Fm denotes k * sub-grade of mth

attribute (element)Am is the area of coverage by m to elementN is the number of parameters of ;th risk

factor

6.3 Level of Maintainability - The Output of theModel

The level of the maintainability was worked outbased on the cost associated with maintenanceactivities such as repair, replacement and cleaningcosts. The optimum level of maintainability wasevaluated based on the minimum life cyclemaintenance costs at a certain period [25],

27 ENGINEER

MS = \- (4) —(D -O}2

2P(6)

where m is the current number ofdefects/problems in a building, M is totalnumber of defects/problems in the building at thebase period (Table 1), RRDC^is the respective

repair and replacement costs of I t h

defect/problem, CC is actual cleaning cost andBCC is the basic cleaning cost derived fromrequired cleaning cycles.

6.4 Network Design and Training

Designing of the network is determined by threeparameters; number of neurons in the input,hidden and output layers. However, the numberof neurons in the input and output layers are pre-determined by the size of the input and outputvectors respectively. In this network, the numberof nodes in these two layers were set as 12 and 1respectively. Even though there is no exact rulefor determining the optimum number of hiddennodes, it was found from the trial and errormethod that it should be between the average andsum of the input and output nodes. Furthermore,it is found that, the size of the hidden layer can beset according to the rule of thumb; 75% of the sizeof the input layer [30-31]. Among variousarchitectures and paradigms, the most simplestand practicable network; backpropagation whichhas been used to perform higher level humantasks such as planning, scheduling andforecasting, etc., was used for the model. A set of50 examples was fed into this feed forwardnetwork for training. This learning algorithmperforms simple gradient descent, propagatingthe differences, or error ( E ) , back through thenetwork to contribute a small adjustment to theeach weight or bias parameters as,

9 P (5)

where P is the number of training samples, Dtj is

the desired output value of the z'th processingelement (neuron) to the j th training example and

Oj is the corresponding output value.

In this process, the difference cf output activitypattern moves nearer to its global minima as,

where Dis the desired output and Ois thecomputed output

The accuracy of the network was tested until thebest network structure was established. The bestresults (Equation 7) were obtained from thenetwork of one hidden layer that consists of 9processing elements (PE) (Figure 1). The totalnumber of 14,419 network cycles (epochs) wererun to converge to the given target (10~3).

6.5 Network Performance and Analysis

Low approximation error (0.001) combined withlow generalization error (0.0270) in the networkensures the best performance of the network.Such a paradigm ensures a good approximationwith a smooth fitting over the training data andthat smooth fitting leads to ensure bestforecasting (generalization) of new data.

In the process of grading maintainability of abuilding, the network determines the level ofmaintainability associated to a particular designby analyzing its risk environment (i.e. input riskparameters) from "known" risks conditions thatare derived from examples (cases) shown to thenetwork during its training cycles.

The network was used to predict themaintainability of flat roofs in Sri Lanka. A typicalcase (Table 3) which represents a typical flat roofcondition was tested. The results showed that themaintainability level was 51%. This indicates thatlots more could be done to improve flat roofs'performance to achieve high maintainable roofs inmulti-story buildings in Sri Lanka. In general,water leaking problems due to lack ofmaintenance together with lack of waterproofingdesign detailing and poor performance of theconcrete slab cause many maintainabilityproblems. The need of an effective drainagesystem was too highlighted by many maintenancemanagers. Even with good drainage systems,they were not being properly maintained due tolack of allocation or no allocation of maintenancebudget.

Misfortune of traditional construction process ofmany multi-story buildings in Sri Lanka is thatniether maintenance person is employed norbudget is allocated for regular maintenance of aproperty.

ENGINEER 28

r Slab performance

[Waterproofing selection I

IWaterpronfinp detail

I Drainage selection

I Drainage details

W

Drainagemaintainability

I Building profile

I Material selection

I Material detail

Material maintainability

I Construction quality

Maintenance nractice

. Output

Figure 1: Architecture of the Network

farther, trained network was used to extractknowledge for the users to gain a betterunderstanding of how the networks solve theproblem. Pruning techniques uses to eliminateless important connections (nodes) from thenetworks is identified for this task [32-34].

According to this technique, the maximum error-£ value indicates the highest sensitivity of the

Table 3: Risk Conditions of a Typical RoofStructure

Risk Parameter

Slab Performance

Waterproofing System SelectionWaterproofing System Detail

External Drainage System Selection

External Drainage System DetailExternal Drainage SystemMaintainability

Building Profile

Material Selection

Material Detail

Material Maintainability

Construction Quality

Maintenance Practice

RiskCondition

3.50

3.62

3.293.44

3.42

3.20

4.03

3.58

2.93

3.10

3.27

3.26

parameter to its desired accuracy. Having set that,observing the sensitivity of each factor in thetraining dataset can be determined theimportance of the factor to function of themaintainability. Mean squared errors ( £ ) of roofrisk parameters obtained according the algorithm,showed all of 12 factors were significant to the flatroof maintainability. Among these factors,performance of the slab detailing have shownhighest sensitivity to the network followed bywaterproofing detailing, external drainagemaintainability, etc., (Table 4)

7 Conclusion

The total number of randomly selected 50buildings was used to evaluate themaintainability problems of flat roofs. Theresearch revealed that the awareness ofmaintainability concept of buildings is very lowand hence, maintainability issues related tobuilding design, construction and maintenanceare not yet taken up by the practitioners in theindustry, in order to enhance the maintainabilityof buildings. However, this ignorance has createdan escalating number of building defects andproblems with increased number of multi-storeybuildings. In flat roofs of 50 buildings, 721defects/problems were observed. Among themcracks, chockages in down pipes and splittingwere ranked as first, second and third defectsrespectively.

29 ENGINEER

Table 4: Sensitivity Analysis of Risk InputsRisk Factorsperformance of the detailing ofthe slabwaterproofing detailingexternal drainage systemmaintainabilitybuilding profilematerial maintainabilitymaintenance practicesexternal drainage systemwaterproofing systemmaterial selectionconstruction qualitymaterial detailingexternal drainage detailing

MSE ( s ) - maximum8.7

6.74.6

3.93.62.92.72.62.21.91.41.2

Rank1

23

456789111213

This model which is developed based onmaintenance cost approach would eventually beused to guide the maintenance requirements offlat roofs during the design stage. Further, as apredictive tool, this system would enable owners,designers, facility managers, contractors and anyother party with an interest in achieving the mostfavourable maintainability level right from thedesign/planning stage.

The accuracy of this ANN model developed forflat roof was tested. The low network error andthe generalization error indicated the goodperformance of the network and its capability ofpredicting levels of maintainability for newdesigns. The trained network error was used toevaluate the maintainability level of a typical casestudy. A fifty one percent level was observed andtherefore more consideration is needed toenhance the level of maintainability.

Further, it is useful to identify importantconsiderations in roof maintainability. Therefore,the sensitivity of all risk parameters towardsaffecting level of maintainability for flat roofs wasevaluated. The extraction of the knowledgeembedded in a trained network is acomprehensive way for designers to understandthe function through the sensitivities of theinputs. An algorithm which was established usingthe pruning concept was used to identify thesensitivity of these individual inputs to thefunction of maintainability. This sensitivityanalysis showed that the maintainability of flatroofs relies very much on its,

o Performance of structural elementso Detailing of waterproofing being appliedo Maintainability of external drainage

systemo Building profile

o0

Materials performance suchmaintainability and durability and,Maintenance practicesConstruction quality

as

References

1. De Silva, N., Dulaimi, M.F., Ling, F.Y.Y and Ofori, G.(2004) Improving the maintainability of buildingsin Singapore, Building Research & Information,39(10), 1243-1251.

2. Seely, I. H. (1994) Building Maintenance, 2nd ed.Macmillan Press Ltd: London.

3. Wood, B. (2005) Towards innovative buildingmaintenance, Structural Survey, 23(4), 291-297.

4. Boussabaine, A.M. and Kirkham, R.J. (2004)Simulation of maintenance costs in UK localauthority sport centers, Construction Managementand Economics, 22(1), 1011-1020.

5. Al-Hammad, A., Assaf, S. and Al-Shihah, M. (1997)The effect of faulty design on buildingmaintenance, Journal of Quality in Maintenance, 3(1),29-39.

6. Shohet, I.M., Puterman, M. and Gilboa, E. (2002)Deterioration patterns of building claddingcomponents for maintenance management,Construction Management and Economics, 20, 305-314.

7. www.statistics.gov.lk, accessed on 18th July 2007.8. Gambardella, L M and Moroni, M (1990) Expert

systems application to building pathologydiagnosis: methodology, Proceeding of the SecondEuropean Conference on Application of ArtificialIntelligence and Robotics to Building, Architecture andCivil Engineering, EuropIA 90, Liege, Belgium, 252-258.

9. Honstede, W.V. (1990) Research into the quality ofhousing stocks in the Netherlands, BuildingMaintenance and Modernization Worldwide, 2, 659-668.

10. Assaf, S., Al-Hammad, A. and Al-Shihah, M. (1995)The Effect of Faulty Construction on BuildingMaintenance, Building Research and Information, 23,175-81.

ENGINEER 30

1 1 , Olubodun, O. F. (1996) An empirical approach to thei-valuation of factors in local authority housingmaintenance requirements in the city of Manchester,Unpublished PhD Thesis, University of Salford.

12, Bourker, K. and Davies, H. (1997) Factors affectingthe service life predictions of buildings: adiscussion paper, Laboratory Report, BuildingResearch Establishment, Garston, UK.

13, Meier, J. R. and Russell, J. S. (2000) Model processfor implementing maintainability, Journal ofConstruction Engineering and Management, 126(6),440-450.

!•!. Stephen, J.H., (2002). Building Services Maintenance-The Forgotten Discipline. Aha ManagementPublications; www.aha.com.au/energyl.htm,accessed on 25th August 2007

15. Chew, M.Y.L. and De Silva, N. (2003)Maintainability problems of wet areas in high-riseresidential buildings, Building Research &Information, 31(1), 60-69.

16. Chew, M.Y.L. and Tan, P.P. (2003). Facades stainingarising from design features, /. Construction andBuilding Materials, 17(3), 181-187.

17. Newton, L.A. and Christian, J. (2006) Impact ofquality on building costs, Journal of InfrastructureSystems, 12(4), 199-206.

18. Shabha, G. (2003) A low cost maintenance approachto high rise flats, Facilities, 21(13/14), 315-322.

19. Hermans, M.H. (1995) Deterioration Characteristicsof Building Components: A Data Collection Modelto Support Performance Management, Eindhoven,TU Eindhoven.

20. Straub, A. (2003) Using a condition-dependentapproach to maintenance to control costs andperformance, Journal of Facilities Management, 1(4),380-95.

21. Chew, M.Y.L. and De Silva, N. (2004) Factorialmethod for performance assessment of buildingfacades, /. Construction Engineering andManagement, 130(4), 525-533.

22. Teo, E. A. and Harikrishna, N. (2006) A quantitativemodel for effective maintenance of plastered andpainted facades, Construction Management andEconomics, 24,1283-1293.

23. Tesfaye, A. (1997) A holistic approach towards lowmaintenance buildings: a case study of sampleresidential buildings in Botswana, Proceedings of 5th

International Conference on Inspection, Appraisal,Repairs and Maintenance of Buildings and Structures,15-16 May 1997, Singapore.

24. Chanter, B. and Swallow, P. (2000) BuildingMaintenance Management, Blackwell, Oxford.

25. Chew, M.Y.L., De Silva, N. and S.S. Tan (2003)Maintainability of buildings in the tropics, CIBInternational Workshop on Management of Durabilityin the BuildingProcess, Politecnico di Milano, Italy.

26. De Silva, E.N.D. (2000) Allocation of Contingency andits Management for Construction Costs in Singapore: AProbabilistic Approach Using ANN, UnpublishedMaster of Science Thesis, University of Singapore.

27. Lisboa, P.G.J. (1992) Neural Networks CurrentApplications, Chapman & Hall, London.

28. Ian, F. and Nabil, K. (1998) Artificial Neural Networksfor Civil Engineers: Advanced Features andApplications, ASCE, Reston, VA.

29. Haykin, S. (1999) Neural Networks: A ComprehensiveFoundation, Second Edition, Prentice Hall, UpperSaddle River.

30. Bailey, D. and Thompson, D. (1990) How to developneural network applications, AI Experts, 5(6), 38-47.

31. Goh, B H (2000) Evaluating the performance ofcombining neural networks and genetic algorithmsto forecast construction demand: the case study ofthe Singapore residential sector, ConstructionManagement and Economics, 18(2), 209-217.

32. Zurada, J.M., Malinowski, A. and Cloete, I. (1994)Sensitivity of minimization of input datadimension for feedforward neural networks, IEEEInternational Symposium on Circuits and Systems,London.

33. Rudy, S. (1997) Extracting rules from neuralnetworks by pruning and hidden-unit splitting,Neural Computation, 9(1), 205-225.

34. Chew, M.Y.L., De Silva, N. and S.S. Tan (2004) "Aneural network approach to assessing buildingfagade maintainability in the tropics," ConstructionManagement and Economics, 22(6), 581-594.

31 ENGINEER