COMPREHENSIVE MAINTAINABILITY SCORING SYSTEM (COMASS) FOR COMMERCIAL BUILDINGS IN TROPICAL CLIMATE OF SINGAPORE SUTAPA DAS NATIONAL UNIVERSITY OF SINGAPORE 2008 COMPREHENSIVE MAINTAINABILITY SCORING SYSTEM (COMASS) FOR COMMERCIAL BUILDINGS IN TROPICAL CLIMATE OF SINGAPORE SUTAPA DAS (B.Arch (Hons.), JU; M.Tech, IIT Madras) A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY DEPARTMENT OF BUILDING NATIONAL UNIVERSITY OF SINGAPORE 2008 ACKNOWLEDGEMENTS This thesis is dedicated to my husband Abhijit Chaudhuri for his unconditional love, sacrifice and patience COMASS is an outcome of persistent effort and a great deal of commitment It has drawn intellectual support and generous help from experts from various fields The list is endless and also the contributions It is like a beautiful pearl string where each pearl is equally precious and unique I take this opportunity to express my sincere thanks to everyone, who has been with me in this ‘Immense journey in search of excellence’ First of all I would like to specially mention about my supervisor and mentor Prof Michael Yit Lin Chew; my thesis committee members Dr Wong Nyuk Hien and Dr Evelyne Teo Without them nothing was possible Dr Poh Kim Leng from Dept of Systems & Industrial Engineering provided an extraordinary guidance in a critical part of this research I am deeply indebted to the following treasured personalities for their help in one way or other: ● Ms Nayanthara De Silva (University of Moratowa) ● Dr Kang Kok Hin (Institute of Facilities Management) ● Other professors and staffs of Dept of Building, National University of Singapore ● Industry experts from several prestigious facility management organizations ● My students: Adeline, Fong Yee, Hoe Kiat, Nur Hafizah, Tsz Shan and Wen Tirng ● My colleagues: Benson, Bin, Jovan, Shams, Swei Mei and many others I pay my gratitude to my family members - my husband, mom and mom-in-law for providing uncompromising moral strength to sail through all ups and downs The last but not the least mention is my late father – whom I could not even bid a farewell He wanted me to fulfil my academic responsibility and silently left his blessings i SUMMARY Growing complexity of building systems and higher user requirements have prioritized maintainability over maintenance especially for commercial building sector in order to attract and retain clients A thorough literature review indicated that unlike sustainability, there is no comprehensive maintainability scoring system (COMASS) to predict maintainability potential of buildings In spite of good design-construction-maintenance guidelines, recurrent defects keep buildings under constant maintenance, putting users’ health and safety at stake apart from affecting economy and system performance This paradox has been attributed to dearth of: (1) knowledge database of defects; (2) system selection framework and (3) communication Proposed COMASS addresses this knowledge gap It is a decision-enhancement tool aimed for part or whole of a new or existing building for selecting the best strategy to ‘design out’ defects Research objectives was set as deliverables: (1) Defect Library (DL) to improve the knowledge on building defects; (2) Maintainability Handbook to benchmark designconstruction-maintenance practices along with maintainability score (MS); and (3) integration of building elements and life cycle phases into COMASS Commercial buildings of Singapore for its key elements under central facility management were studied Qualitative FMECA (Failure Mode Effects and Criticality Analysis) was selected for defect grading Building was divided into nine major subsystems grouped into two main systems: (1) civil-architectural or C&A (basement, facade, wet-area and roof) and (2) mechanical-electrical or M&E (sanitary-plumbing, HVAC, elevator, electrical and fireprotection) From 14 detailed case studies and interview with 34 facility managers (FM), 319 defects pertaining to 62 major components of these subsystems were identified Their causes were analyzed in terms of (1) design/ specification; (2) construction/ installation; (3) maintenance; and (4) external factors About 45.5% defects were found critical based on their frequency and severity (impacts on economy, system performance and health-safety-comfort) ii Maintainability guidelines comprised of 731 defect-mitigating checklist factors was developed from literature and DL They were grouped into life-cycle phases and further subgrouped under components of subsystem Factors’ relative weights (RW) are their ability to mitigate both critical and non-critical defects Weighted sum of the MS of each factor indicates MS for individual sub-system, where a higher score means higher maintainability For integration of subsystems into COMASS w.r.t both objective and subjective parameters, AHP (analytic hierarchy process) using interview and questionnaire was critically selected Eleven consistent results were used to determine 12 sets of RWs for subsystems for various location-height combinations Seamless matching of objective result with experts’ subjective opinion proved the integration process logical and comprehensive Predictive accuracy of COMASS was found satisfactory through operational validity and sensitivity test via Monte Carlo simulation A prototype multi-tenant office tower at CBD was modelled COMASS is the first attempt in building maintainability for holistic integration of phases of building lifecycles and components From existing defects, COMASS evaluated the entire decision making process of building life cycle and reflected back the same on performance Hence COMASS was able to bridge the knowledge gap between theoretical guidelines and their real life implication This research further highlighted that performance, not cost was the main governing factor in facility management The standard of performance based on both objective and subjective parameters imposes different emphasis of different building components This decision–enhancement tool empowered with a user friendly, performance based online version (www.hpbc.bdg.nus.edu.sg) aspires to improve the quality of building industry significantly The generic method is applicable to other building types and climates Further refinement with real-life testing, consideration of chain effect of defects to the fullest extent and time-dependent decision making were identified as the scope of future research Keywords: AHP, Benchmarking, Defects, FMECA, Maintainability, Scoring system iii TABLE OF CONTENTS Acknowledgements ………………………………………………………………… i Summary ………………………………………………………………… ii Table of Contents ………………………………………………………………… iv List of Tables ………………………………………………………………… ix List of Figures ……………………………………………………………………… xiv List of Acronyms …………………………………………………………………… xvii CHAPTER 1.1 1.2 1.3 1.4 1.5 INTRODUCTION ……………………………………………… Background …………………………………………………………………… 1.1.1 The concept of maintainability and maintenance ………………… 1.12 Significance of maintainability …………………………………… 1.1.3 Research problem – A paradoxical situation in Singapore………… 1.1.4 Rationale of the study ……………………………………………… 1.1.4.1 Dearth of knowledge database of defect ……………… 1.1.4.2 Dearth of system selection framework ………………… 1.1.4.3 Lack of communication ………………………………… Research guideline …………………………………………………………… 1.2.1 Knowledge gap ………………………………………………….… 1.2.2 Aim and Objectives ……………………………………………… 1.2.3 Hypothesis ………………………………………………………… 1.2.4 Scope of Research ………………………………………………… 1.2.5 Knowledge Contribution ………………………………………… 1.2.6 Practical Implication ……………………………………………… Definition of terms …………………………………………………………… Organisation of the thesis …………………………………………………… Summary ……………………………………………………………………… 1 2 4 6 7 7 9 10 10 12 CHAPTER 2.1 2.2 2.3 2.4 2.5 2.6 LITERATURE REVIEW ……………………………………… 13 Introduction …………………………………………………………………… Building components in terms of maintainability …………………………… Building maintenance ………………………………………………………… 2.3.1 Objectives of maintenance ………………………………………… 2.3.2 Decision support frameworks in maintenance ……………… …… Maintainability ……………………………………………………………… 2.4.1 Objectives of maintainability ……………………………………… 2.4.2 Maintainability in building research ……………………………… Maintainability tools in system engineering ………………………………… 2.5.1 Fault Tree Analysis (FTA) ………………………………………… 2.5.2 Fishbone Diagram …………………………………………… … 2.5.3 Failure Mode Effect and Criticality Analysis (FMECA) ………… 2.5.4 FMECA in building sector ……………………………………… Existing building grading systems …………………………………………… 2.6.1 Classification of grading system ………………………………… 2.6.1.1 First generation: pass-fail ……………………………… 13 14 16 16 17 21 21 21 23 23 24 24 27 28 28 30 iv 2.7 2.8 2.9 2.6.1.2 Second generation: simple additive …………………… 2.6.1.3 Third generation: weighed additive …………………… 2.6.1.3 Others ………………………………………………… Principles for weighing and aggregation of multiple parameters …………… 2.7.1 Equal weight ……………………………………………………… 2.7.2 Weights based on statistical models ……………………………… 2.7.3 Weights based on opinions: MCDA methods …………………… 2.7.4 Comparisons of the methods ……………………………………… Knowledge gap ……………………………………………………………… Summary ……………………………………………………………………… CHAPTER 3.1 3.2 3.3 3.4 3.5 3.6 30 30 32 34 35 35 36 39 40 41 RESEARCH METHODOLOGY ……………………………… 42 Introduction …………………………………………………………………… 3.1.1 Brief overview of research methodology ………………………… Phase 1: Conception ………………………………………………………… Phase 2: Individual maintainability scoring ………………………………… 3.3.1 Selection of defect grading strategy ……………………………… 3.3.2 Development of criticality parameters …………………………… 3.3.3 Selection of respondent and sampling frame ……………………… 3.3.4 Site Investigation ……………………………………… 3.3.5 Questionnaire design and pilot survey …………………………… 3.3.6 Main survey ……………………………………… 3.3.7 Proposed Defect Library (cause and criticality analysis) ………… 3.3.8 Maintainability Handbook and subsystem grading ……………… Phase 3: Integration of building subsystems into COMASS ………………… 3.4.1 Selection of AHP as suitable technique for integration …………… 3.4.2 Construction of hierarchy ……………………………………… 3.4.2.1 Goal ……………………………………… 3.4.2.2 Criteria ……………………………………… 3.4.2.3 Sub-criteria under location …………………………… 3.4.2.4 Sub-criteria under height ……………………………… 3.4.2.5 Mutual exclusiveness of sub-criteria …………………… 3.4.2.6 Alternatives ……………………………………… 3.4.3 Development of questionnaire …………………………………… 3.4.4 Selection of respondents and sample size ………………………… 3.4.5 Data collection – survey & interview ……………………………… 3.4.6 Data analysis and global weight (GW) calculation ……………… 3.4.6.1 Inconsistency ratio …………………………………… 3.4.6.2 Aggregation of results ………………………………… 3.4.6.3 Derivation of global weights from local weights ……… 3.4.7 Development of COMASS ……………………………………… Phase 4: Conclusion ……………………………………… 3.5.1 Checking the predictive accuracy ………………………………… 3.5.1.1 Validation ……………………………………… 3.5.1.2 Sensitivity analysis …………………………………… 3.5.1.3 The proposed testing method for COMASS …………… 3.5.2 Practical application ……………………………………… 3.5.3 Conclusion and recommendation ………………………………… Summary ……………………………………… 42 42 42 44 44 45 46 47 48 49 50 51 52 53 55 55 56 56 57 57 57 58 60 60 61 61 62 62 63 64 64 64 66 66 67 67 67 v CHAPTER 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 DEFECT ANALYSIS …………………………………… 68 Introduction …………………………………………………………………… General findings of questionnaire survey …………………………………… 4.2.1 Demographic information ………………………………………… 4.2.2 Significance of grading criteria and building subsystems ………… 4.2.3 Format of defect reporting ………………………………………… Defects in basement ………………………………………………………… 4.3.1 Criticality analysis of defects in basement ………………………… Defects in facade ……………………………………………………………… 4.4.1 Criticality analysis of defects in facade …………………………… Defects in wet area …………………………………………………………… 4.5.1 Criticality analysis of wet area defects …………………………… Defects in roof ………………………………………………………………… 4.6.1 Criticality analysis of defects in roof ……………………………… Defects in sanitary-plumbing system ………………………………………… 4.7.1 Criticality analysis of defects in sanitary-plumbing system ……… Defects in HVAC system …………………………………………………… 4.8.1 Criticality analysis of defects in HVAC system ………………… Defects in elevators (or lifts) ………………………………………………… 4.9.1 Criticality analysis of defects in elevators ………………………… Defects in electrical system …………………………………………………… 4.10.1 Criticality analysis of defects in electrical system ………………… Defects in fire protection system ……………………………………………… 4.11.1 Criticality analysis of defects in fire protection system …………… Comparison of causes and criticality of building subsystems ……………… Summary …………………………………………………………………… 68 68 68 70 70 71 74 75 83 85 89 89 94 95 101 102 107 108 114 116 122 123 128 129 131 CHAPTER 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 MAINTAINABILITY SCORING FOR BLDG SUBSYSTEMS 132 Introduction …………………………………………………………………… General format of maintainability scoring ……………………………… 5.2.1 Mathematical principle …………………………………………… 5.2.2 Maintainability Handbook ………………………………………… Maintainability scoring for basement …………………………………… … Maintainability scoring for facade …………………………………………… Maintainability scoring for wet area ………………………………………… Maintainability scoring for roof ……………………………………………… Maintainability scoring for sanitary-plumbing system ……………………… Maintainability scoring for HVAC system.………………………………… Maintainability scoring for elevators ………………………………………… Maintainability scoring for electrical system ………………………………… Maintainability scoring for fire protection system …………………………… Overview of scoring of all subsystems ……………………………………… Summary …………………………………………………………………… 132 132 132 133 133 135 137 139 141 144 146 149 152 154 154 CHAPTER 6.1 6.2 COMPREHENSIVE MAINTAINABILITY SCORING SYSTEM Introduction …………………………………………………………………… Data analysis ………………………………………………………………… 6.2.1 Data processing in ExpertChoice (EC 11.5) ……………………… 6.2.2 Dealing with inconsistency ……………………………………… 6.2.3 Selection of threshold limit of IR and the best dataset …………… 155 155 155 155 157 158 vi 6.3 6.4 6.5 6.2.3.1 Preliminary test result …………………………… …… 6.2.3.2 Statistical test result …………………………………… 6.2.3.3 Rank reversal …………………………………………… 6.2.3.4 Selection between Dataset and …………………… 6.2.4 Derivation of GW from LW ……………………………………… Results and discussion ………………………………………………………… 6.3.1 Influence of criteria: location and height ………………………… 6.3.2 Influence of sub-criteria: location ………………………………… 6.3.3 Influence of sub-criteria: height …………………………………… 6.3.4 Relative importance of C&A and M&E systems………………… 6.3.4.1 Influence of location on C&A and M&E systems……… 6.3.4.2 Influence of height on C&A and M&E systems ……… 6.3.5 Relative importance of all nine subsystems ……………………… 6.3.5.1 Rank 1: HVAC system ………………………………… 6.3.5.2 Rank 2: Elevator system ……………………………… 6.3.5.3 Rank 3: Facade ………………………………………… Application of GW in COMASS …………………………………………… 6.4.1 Example of a calculation in COMASS …………………………… Summary …………………………………………………………………… CHAPTER 7.1 7.2 7.3 7.4 7.5 TESTING AND APPLICATION ……………………………… 173 Introduction …………………………………………………………………… Operational validity …………………………………………………………… 7.2.1 Details and specification of the prototype building ……………… 7.2.2 Scoring for basement ……………………………………………… 7.2.3 Scoring for facade ………………………………………………… 7.2.4 Scoring for wet area ……………………………………………… 7.2.5 Scoring for roof …………………………………………………… 7.2.6 Scoring for sanitary-plumbing system …………………………… 7.2.7 Scoring for HVAC system ………………………………………… 7.2.8 Scoring for elevators ……………………………………………… 7.2.9 Scoring for electrical system ……………………………………… 7.2.10 Scoring for fire protection system ………………………………… 7.2.11 Scoring for entire building ………………………………………… Sensitivity analysis via Monte Carlo simulation ……………………………… Web based application of COMASS ………………………………………… 7.4.1 Defect Library …………………………………………………… 7.4.2 Maintainability Scoring System …………………………………… Summary …………………………………………………………………… 173 173 173 174 176 178 180 182 185 187 189 192 195 196 197 197 198 200 CHAPTER 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 159 159 160 160 161 163 163 163 165 166 166 167 168 170 170 171 171 171 172 CONCLUSIONS ………………………………………………… 201 Introduction …………………………………………………………………… Research summary showing achievement of research goal…………………… Key findings ………………………………………………………………… Knowledge contribution ……………………………………………………… Industry contribution ………………………………………………………… Limitation of the study ……………………………………………………… Scope for future research …………………………………………………… Concluding remarks ………………………………………………………… 201 201 203 205 207 208 208 209 vii BIBLIOGRAPHY ………………………………………………………………… 210 APPENDIXES ………………………………………………………………… 240 Appendix A Appendix B Appendix C1 Appendix C2 Appendix C3 Appendix C4 Appendix C5 Appendix C6 Appendix C7 Appendix C8 Appendix C9 Appendix D Appendix E Survey questionnaires …………………………………………… Results of defect rating survey …………………………………… Maintainability guidelines for basement ………………………… Maintainability guidelines for facade ……………………………… Maintainability guidelines for wet area …………………………… Maintainability guidelines for roof ……………………………… Maintainability guidelines for sanitary-plumbing system ………… Maintainability guidelines for HVAC system …………………… Maintainability guidelines for elevators …………………………… Maintainability guidelines for electrical system ………………… Maintainability guidelines for fire-protection system …………… AHP survey results ………………………………………………… Results of sensitivity test by Monte Carlo simulation …………… 240 251 275 288 307 319 331 347 358 371 390 404 410 LIST OF PUBLICATIONS ………………………………………………………… 414 viii the factors those influence overall building maintainability A direct comparison of various subsystems is not scientifically viable For example water tightness and indoor air quality are the key concerns for maintainability of basement and HVAC system respectively These two have nothing in common except cost and cost is not a parameter to define criticality – the basis used for individual scoring When it comes to overall maintainability i.e ease of maintenance for overall improvement of a building’s performance, FMs essentially set priority in order to generate the best possible business within a limited maintenance budget (Shen et al.,1998) Apart from objective parameters, their decision is intuitively guided by subjective parameters such as political, financial, social, economic and legal importance of maintainability (Jones & Sharp, 2007; Spedding et al., 1995) Gilleard & Yat-lung (2004) further demean this process as ‘possibly arbitrary’ due to lack of structured decision analysis tools or over emphasis of certain KPIs Simulation models are unsuitable as they require large amount of input data (Boussabaine & Kirkham, 2004) It should be noted that information is exceptionally rare in the field of defect or fault studies as people are reluctant to share or document their experience from the fear of legal harassment or loss of reputation (Wardhana & Hadipriono, 2003) 3.4.1 Selection of AHP as suitable technique for integration Hence for integration of building subsystems into COMASS, it was essential to select a weighted additive system that can handle both objective and subjective parameters, tolerate natural inconsistency in judgement to certain extent and generate project specific and performance based information Such sophisticated tools inherently suffer from complexity, costly simulation or documentation, fuzzy nature of weighting, user dependence etc leading to a limited application (Crawley & Aho, 1999) As a result rudimentary techniques are still popular in spite of their lack of precision (Chew & Das, 2008) 53 Analytic Hierarchy Process (AHP) was critically selected to fulfil the requirements of this research AHP has been reported as the most systematic MCDA tool (Lee & Burnett, 2006) and most suitable for FM related studies (Wu, Lee, Tah & Aouad, 2007) because it: ● Can deal with complex, unstructured and multi-attribute decision (Partovi, 1994) ● Combines both qualitative and quantitative approaches into one empirical method (Cheng & Li, 2001) ● Can handle inconsistency to a certain extent which is a common limitation of human mind while making complex decisions ● Provides a clear rationale for the decision (Forman & Selly, 2001) ● Generates easily comprehensible results through simple software application Since its development by Prof T.L Saaty in 1980s, AHP has found extensive application in commercial decision making or empirical research (Easley, Valacich & Venkataramanan, 2000; Ho, 2007; Vaidya & Kumar, 2006) In building research, especially in subjective priority setting, use of AHP is prevalent, to name a few, sustainability grading for CASBEE (JSBC, 2004); SBTool-Taiwan (Chang, Chiang & Chaou, 2007); maintenance priority setting (Reddy et al., 1994; Shen & Lo, 1999; Shen, Lo & Wang, 1998); selection of intelligent building components (Wong & Li, 2008); accessibility of buildings (Wu et al., 2007); priority setting in construction quality (Tan & Lu, 1993) etc However AHP is not exempted from criticisms for consistency cut-off, left-right eigenvector inconsistency etc (Apostolou & Hassell, 1993; Johnson, Beine & Wang, 1979; Lin, Wang & Yu, 2008; Pelaez & Lamata, 2003) For more than two decades, AHP has counteracted numerous criticisms (Saaty, 2005) and have managed to retain its basics unmodified Unlike other tools seeking optimality, AHP suffers from artificial restrain on decision to certain extent, but yet it is the most structured and flexible method (Bana d Cotsa & Vansnick, 2008) The mathematical principle of AHP has been discussed in Section 2.7.3 With the goal of ‘higher building maintainability’, the following steps were followed for COMASS: 54 ● Construction of hierarchy using top-down approach - goal at top, criteria and sub-criteria at subsequent levels and building subsystems at base ● Questionnaire design to obtain group judgement ● Selection of respondent & sample size ● Data collection (objective pair-wise comparison and its subjective reasoning) ● Data analysis and calculation of global weights of subsystems 3.4.2 Construction of hierarchy In AHP, construction of hierarchy depends on how the given problem is decomposed into various units to suit the level of complexity of the problem (Saaty, 1997) In order to generate RW of nine subsystems contributing in building maintainability, the proposed AHP model was developed (Fig.3.3) Details of the components are presented in the following sections Higher Bldg Maintainability GOAL Bldg Location Criteria Residential zonel Subcriteria Bldg systems Subsystems Commercial Industrial zone l zonel C&A - Civil & Architectural −− Basement −− Facade −− Wet area −− Roof Mutually exclusive for location option; Bldg Height Mixed zonel Low ≤ st.2 Medium High 8-20 st >20 st M&E - Mechanical & Electrical −−Sanitary-plumbing −− HVAC −− Elevators −− Electrical −− Fire protection Mutually exclusive for height option; st = storey Figure 3.3 The hierarchy model for building maintainability 3.4.2.1 Goal In the context of this research, w.r.t different criteria and sub-criteria, the relative impact of various building subsystems was to be determined in order to achieve higher maintainability for the entire building Hence it was set as the goal of the hierarchy 55 3.4.2.2 Criteria Expert respondents, who participated in the previous pilot survey, commented that though the objective parameters have higher influence on individual subsystems, but the overall maintainability involves subjective parameters In order to make the business grow, the building standard should match rather outsmart other buildings in the same area Building height often guides the complexity of building systems and in turn influences decision in maintenance management That means, a building of same floor area and facility will be treated differently if the building height or location changes This concept matched with the fourth maintainability parameters i.e external factors as suggested by Chew et al (2008) Building age was finally dropped from hierarchy model after long debates as the issue is addressed by pre-defined replacement schedule and there is a very limited scope to set priority For example, if all the sprinkler heads need replacement in a certain year, fund will be allocated for that job irrespective of the relative importance of fire protection system in general Secondly, compared to physical age, functional obsolescence plays a greater role in commercial building sector and major alteration to combat the same is a sporadic and unpredictable phenomenon Hence location and height were finalised as criteria 3.4.2.3 Sub-criteria under location Table 3.2 Various location options used in AHP model Zone Residential Commercial Industrial Mixed URA terminologies Residential, residential with commercial at 1st storey only, residential/ institution Commercial, hotel, business park, civic and community buildings Business 1, business Commercial & residential, white, business park – white, business 1– white, business - white Urban Redevelopment Authority or URA (2003) divides Singapore in 31 different zones For this research, only those zones were selected as sub-criteria where a commercial building is commonly located Building locations were regrouped under four main zones; namely, residential, commercial, industrial and mixed (Table 3.2) The areas with main use of 56 healthcare, education, places of worship, civic institutions, nature reserve, utility, port, prison etc were not relevant for this study 3.4.2.4 Sub-criteria under height There is no universal definition of tall buildings in terms of height or number of storey (Chew, 2001) Structurally, a building is tall when its structural analysis and design are affected by the lateral loads, particularly sway (Taranath, 1988) On the other hand fire engineers designate a building as tall when emergency evacuation is so difficult that the fire should be fought internally (NFPA 70) From maintainability point of view, the cut-off levels of low, medium and high were defined by the complexity of building subsystems A high rise was defined as a building where due to its height the building systems and components become complex and difficult to maintain The growing complexity with height for nine subsystems was reviewed from codes or standards A simple grading was assigned as: = not applicable / not required; = basic requirement; = moderate requirement and = complex requirement When there was an abrupt increase in the total grade with height, a threshold point was set to obtain the classification as: (1) low-rise: ≤ storeys (~24m); (2) medium-rise: 9-20 storeys (~24-60m) and (3) high-rise: > 20 storeys (Fig 3.4) Hence, the second set of sub-criteria under the criteria of height was classified as low, medium and high 3.4.2.5 Mutual exclusiveness of sub-criteria Different options for zone and height are mutually exclusive; i.e a building can not be simultaneously be low and mid-rise For each possible location-height combination, the RWs were renormalized by ignoring all other non-applicable options Such example was illustrated by Saaty himself in determining consumer preference (Saaty, 2001, Page 102 – 106) 3.4.2.6 Alternatives In order to derive relative impact of building systems on maintainability, the alternatives for this hierarchy were nine subsystems grouped under C&A & M&E systems (Section 2.2) These two groups pose very different technical properties and such grouping had a major advantage over direct comparison The numbers of elements to be compared at each level 57 being reasonably small - lesser than the optimum limit of seven to nine (Miller, 1956); the judgments were expected to be highly consistent Increase in compelxity in building systems with no of storeys Subsystem Components 2,3 9-15 16-20 Dry riser 21-25 Fire Fire hydrant Protection (SCDF 2007: Cl 6.2) Nil Fire sprinkler (CP 52: 2004, Cl 4.5) Nil Fire sprinkler (CP 52: 2004, Cl 5.6) Nil Fire alarm (SCDF 2007: Table 6.3A) Manual Fire exit (SCDF 2007: Cl 2.6) Nil Fire lift (SCDF 2007: Cl 6.6.3) Nil Nil Nil > 60 One way system Accessway (SCDF: 2004, Cl ) 41-60 Minimum lift Communication system (SCDF: 2007, Cl 8.2) 26-40 HVAC Pressurisation of exit (SCDF: 2007, Cl.7.2.1) Distribution (CP 13: 1999, Cl 6.4) Wet riser Continuous monitoring Divided in stages Automatic staircase > staircase Required Nil Required if not naturally ventilated Nil Supply air via evenly distributed vertical duct Sanitary- System type plumbing (PUB, 2004) Single stack Electrical Emergency power for lift (SCDF: 2007, Cl.6.6.2) Nil Fully ventilated fire lift Nil Mech Elevator transport (BS 5655-6:2002, Cl 9.2.2) Facade Access system (Chew, 2004) Total score Legend score score Geared traction fire lift + another lift Gearless traction nos nos nos 10 11 11 17 17 20 21 22 23 Significant change in score score score Low rise Medium rise High rise Figure 3.4 Building height vs complexity of building systems 3.4.3 Development of questionnaire The survey questionnaire was comprised of four sections (Appendix A2) After a precise introduction of research objective, the first section briefly defined the key terminologies used in this study, such as maintainability, its controlling parameters, building systems and subsystems The second section was to collect the respondents’ profile in order to explore if there is any correlation between the respondents’ profile and feedbacks In third section the concept of pair-wise comparison and the nine point scale (Table 3.3) were illustrated with a 58 simple worked-out example In the last part, the actual questionnaire for grading was presented Elements were compared pair-wise according to their relative influence on overall maintainability This last section contained total 24 matrixes (Table 3.4) The section containing matrix F1 to I3 was lengthy and repetitive Graphical symbols or ‘Visual cookies’ (Figure 3.5) were designed to clearly demarcate various locations and heights for a better comprehension and to retain respondent’s attention Table 3.3 The 9-point pair-wise comparison scale (Saaty, 1996) Value Intensity of importance of element over other Equal Equal to moderate Moderate Moderate to strong Strong Strong to very strong Very strong Very strong to extreme Extreme Table 3.4 Structure of AHP questionnaire Matrix A B C D1-D4 E1-E3 F1-F4 G1-G3 H1-H4 I1-I3 Nos 1 4 Matrix size (2×2) (4×4) (3×3) (2×2) (2×2) (4×4) (4×4) (5×5) (5×5) Description Location vs height Among zone options Among height options C&A vs M&E systems C&A vs M&E systems Among C&A subsystems Among C&A subsystems Among M&E subsystems Among M&E subsystems Preceding element(s) Overall maintainability Location Height different locations different heights C&A for different locations C&A for different heights M&E for different locations M&E for different heights Note: matrixes with same alphabetic codes (e.g D1-D4) are mutually exclusive $ Location: Residential Area Location: Commercial Area Height: Low Location: Mixed Area Location: Industrial Area Height: Medium Height: High Figure 3.5 Graphical symbols to denote various locations and height of the building 59 3.4.4 Selection of respondents and sample size Facility managers with more than years of experience who participated in the first survey were chosen to be the respondents or decision makers (DM) Experts involved in pilot survey suggested that minimum years experience is required for holistic understanding of the multifaceted problem of maintainability in terms of technical, financial and functional aspects Additionally the practising consultants and senior managers involved in strategy making of developer and FM companies were included in the survey This group of personnel in an average have an experience of more than 15 years and can be considered as industry experts Total 58 contacts were made comprising of 27 owner & developer’s representatives (46%), 19 FM company employee (33%) and 12 individuals (21%) In the first survey, the potential respondents who explicitly opted not to participate were not contacted 3.4.5 Data collection – survey and interview Face to face individual interviews were conducted with the DMs During the interviews, other than the actual ranking of building subsystems, the logic the DMs applied while comparing the elements were noted down in details so that this subjective information can be compared with the objective information in the form of the local weights of various elements at a particular level of the hierarchy Such comparison was planned to allow a better understanding of the integration of building systems and subsystems Apart from briefing the respondents about the research, verbal instructions were given to fill up the survey questionnaire It was found that most of the respondents were not familiar with pair-wise scale and was unable to understand why the questionnaire was so long to compare only few elements Hence these points were explained to them in details and the importance of criteria, sub-criteria and elements were highlighted to avoid pitfalls such as: ● Missing information: failure to provide judgement for a particular criterion ● Misinterpretation of elements: e.g locations (residential, commercial, industrial and mixed zones) can be misunderstood as the function of the buildings 60 ● Difficulty in using pair-wise scale: e.g marking on both sides of the scale ● Inconsistency in judgement: e.g loss of concentration after some time or intransitivity especially for larger matrix 3.4.6 Data analysis and global weight (GW) calculation In AHP method, data analysis includes data processing, dealing with inconsistency, aggregation of the results into group decision making (GDM) and finally derivation of local and global weights Inconsistency and GDM are highly debated topics After a thorough literature study on these two issues, the option most suitable for this research was selected 3.4.6.1 Inconsistency ratio Saaty’s recommended threshold limit of 10% inconsistency ratio (IR) has been challenged by the argument that in reality very few of the comparison matrices can satisfy this criterion (Karapetrovic & Rosenbloom, 1999; Murphy, 1993) Especially with increase in matrix size, inconsistency increases dramatically (Pelaez & Lamata, 2003) and hence much of the evaluation information is lost (Cho & Cho, 2008) Expensive reassessment does not guarantee a better output (Tam, Tong & Chiu, 2006) Probable solutions of this problem are: ● Redefine the threshold of consistency index (Aguaron, & Moreno-Jime’nez, 2003; Apostolou & Hassell, 1993; Crawford & Willams, 1985; Pelaez & Lamata, 2003) ● Improve inconsistency to comply with Saaty’s 10% rule (Tam et al., 2006; Lin, Wang & Yu, 2008; Monsour, 1997) Accountability of these methods is highly debatable as they lack real life application Hence Saaty’s original recommendation to determine inconstancy was not ruled out and effort was taken to improve any clerical error Apostolou and Hassell (1993) remarked that data with average IR of > 10% does not vary significantly from the data with ≤ 10% IR Hence data beyond 10% IR was explored and the results were compared using (1) preliminary visual observation of graphs; (2) non-parametric statistical test and (3) rank reversal For two and 61 more datasets (related samples with ordinal values) Wilcoxon signed rank test and Friedman test were applicable respectively (Tan, 2004) AHP being a subjective and focussed method, a small sample size is allowed, rather preferable (Lam & Zhao, 1998) In fact Cheng and Li (2001) argue that an AHP survey with a large sample may be impractical as a large number of ‘cold-called’ respondents may provide arbitrary answers, leading to very high inconsistency Alike others (Lam et al 1998; Wong & Li, 2008) they have used a sample size less than 10 However in this research, it was decided that if number of acceptable data sets are very less (≤ 5), reassessment should be done 3.4.6.2 Aggregation of results Weighted aggregation of individual judgement reflects the judgement of a cohort and is called group decision making (GDM) Synthesis can be done by (Forman & Peniwati, 1998): ● Aggregating individual judgments (AIJ): DMs acts as a unit or a new individual, e.g during a meeting ● Aggregating individual priorities (AIP): DMs acts as separate individuals, e.g in questionnaire survey When each DM acts in his or her own right, with different value systems, then concern is about alternative priorities of each DM ● Aggregating individual’s derived priorities in each node in the hierarchy This method is technically feasible within AHP framework but it is less meaningful and seldom used Hence in this research AIP was most suitable Next issue was to select among various mathematical principles of summing, namely, geometric mean and their weighted versions (Arbel, Saaty & Vargas, 1987), row geometric mean method (Crawford et al.,1985) and logarithmic least square (Kwiesielewicz, 1996) Among all, the geometric mean method (GMM) is more consistent with the meaning of both judgments and priorities (Forman et al, 1998) Hence GMM was selected 62 Derivation of global weights from local weights 3.4.6.3 Local weights (LW) are weighs of elements within each level of the hierarchy and sum up to one LWs from each matrix for valid datasets can be combined into LWs of group decision by using AIP and GMM method Global weights (GW) are the weights of alternatives with respect to the goal and are derived from the multiplication of local weights obtained from every level This could be illustrated through a simple (1 goal C nos criteria A nos alternatives) hierarchy graded by N nos DMs as follows LW of m-th criteria is: (∏ j =1 aij )1/ N N LWC m = ∑ (∏ C m =i N a )1/ N j =1 ij for i, j = 1,2, , C (3.7) Where, aij = priority of factor over factor a j i.e aij = a ji (∏ j =1 aij )1/ N = Geometric mean of priorities set by N nos DMs, N LW of m-th alternative w.r.t p-th criterion is: (∏ j =1 aij )1 / N N LWAm , p = ∑ (∏ A m =i N a )1 / N j =1 ij for i, j = 1,2, , A GW of m-th alternative w.r.t goal: GWAm = ∑ C p =1 ( LWC p × LWAm, p ) (3.8) (3.9) For a deeper hierarchy the similar process is followed It should be noted that Equation (3.9) cannot be directly applied to mutually exclusive elements (Fig 3.3) and an additional renormalisation should be carried out 3.4.7 Development of COMASS First the calculated LWs and GWs were compared with the subjective data obtained from interviews of industry experts It was checked whether their logic behind the grading was in union of the numeric values Next the calculation of final maintainability score (MS) for the entire building in a simple format was carried out Once the MS and the GW of for all nine subsystems are determined for a building with a certain location and height range, the total maintainability score can be found by, MS Building = ∑i =1[(GWSystem )i × ( MS System )i ] (3.10) 63 3.5 Phase 4: Conclusion The COMASS framework was checked for its predictive accuracy and set ready for real life application through a simple online graphic user interface (GUI) 3.5.1 Checking the predictive accuracy COMASS has many factors where judgmental input may vary in a range of to Therefore, it is important that the output is not affected by such variation Predictive accuracy of a model, tool or decision analysis framework is reliable if it can pass the acid test of validation and / or sensitivity analysis Feasibility of such scrutinizing process for COMASS was reviewed and the most suitable method was selected However, for holistic grading systems where judgmental or conscience based values are used (Horvat & Fazio, 2005), these two aspects are not reported in literature 3.5.1.1 Validation Validation is the determination ‘that the code or model indeed reflects the behavior of the real world’ (DOE, 1986) or ‘a good representation of he actual processes occurring in the real system’ (IAEA, 1982) The main techniques are (Reddy et al.,1994): ● Convergent validity: if results derived by different methods are closely matched ● Internal consistency: if the method used by different individuals or by the same individual at different times yield results that agree with each other ● External validity: output of the proposed model matches with an externally obtained result, e.g real data The last one is the most straightforward route (Hills & Trucano, 1999) and widely embraced by maintenance related studies (Shen et al., 1998; Shohet et al.,2003; Spedding et al.,1994) These works are mainly post occupancy evaluation But COMASS can not be validated using this method The explanation is as follows 64 To validate COMASS externally, the defect profiles of few buildings with various design, construction and maintenance standards should be compared with respective maintainability scores and predicted defect lists That means all design, construction, maintenance details especially the flaws i.e any violation of COMASS guideline should be obtained Else an incomplete input data makes this direct comparison method fail (Oreskes, Shrader-Frechette & Belitz, 1994) In construction industry, incomplete data is a well known problem (Kaminetzky, 1991; Wardhana et al., 2003) There is a high probability that what is documented in original drawings or contract document might not be built at all (Clancy, 1995; Daoud, 1997) Hence a real building can never be used for validation of COMASS unless the whole process of its design, construction (including site modifications and sequence) and maintenance is fully documented It can be argued that at least the maintenance part can be assessed for an existing building to check whether violation of COMASS guidelines for maintenance yields a lower score and the predicted defect profile matches with existing defects This is feasible but scientifically unacceptable due to the facts: ● COMASS loses its relevance and focus as design is the main concern of maintainability Such ‘non-relevance’ yields false conclusion (Beck, Ravetz, Mulkey & Barnwell, 1997) ● Partial testing may create conflict with main focus and overlook any dangerous interaction (Saltelli, Tarantola, Campolongo & Ratto, 2004) Sargent (1998) has presented a comprehensive guide of verification and validation techniques, out of which operational validity is a widely accepted method that determines that ‘the model’s output behavior has the accuracy required for the model’s intended purpose over the domain of its intended applicability’ Any deficiencies found through the validation process can be checked for inadequate conceptualization of numerical inaccuracy The behavior of various specific elements in the proposed framework can be traced (followed) through the model to find if the logic is correct and if the necessary accuracy is obtained 65 3.5.1.2 Sensitivity analysis Quality check by sensitivity analysis is required when there are many uncertain inputs In this research objective data from real buildings were not available Such situation allows a posteriori method where a completed model is reviewed by a competent third party But this method can not truly reflect the SWOT analysis of a model (Saltelli et al., 2004) Especially in COMASS the factors under direct comparison is more than - the optimum limit beyond which human mind fails to provide consistent judgment (Miller, 1956) Strangled with similar problem of large input factors and little objective data, Reddy et al (1994) tried various methods and remarked the best one was assessment of a hypothetical facility by computing an overall score under certain criteria Outcome of Monte Carlo simulation of an input score uniformly spread over a given range was found to follow normal distribution As the mean value was in agreement with the result produced by their proposed RENMOD grading tool, they concluded the model was robust 3.5.1.3 The proposed testing method for COMASS In case of COMASS both suggestions were followed First a hypothetical multi-tenant office tower in CBD was taken for evaluation with common components of a commercial building To replicate a real life scenario, few mistakes were purposefully induced and checked whether COMASS can indicate them or not If the result is satisfactory, the scoring system was considered robust else a second consultation with experts were planned Secondly, for each subsystem Monte Carlo simulation was run with uniformly distributed inputs to check whether the results follow normal distribution It ensured that the model is not sensitive to small perturbation in inputs The elicitation process and analysis were meticulously checked to avoid any computational error 3.5.2 Practical application Practical application was planned through a simple online graphic user interface (GUI) Any proposed scheme can be judged against the ideal score ≈ 100% In GUI window, the user can enter scores against all applicable checklist factors and can evaluate any proposed scheme 66 The drawbacks and the long term effect (both critical and non-critical) of the proposal can also be viewed With trial and error a poor score can be improved The issue of budget allocation has been entirely left on the user 3.5.3 Conclusion and recommendation The key findings, both academic and practical contributions, SWOT analysis of COMASS model and scope of future research were summarised 3.6 Summary This chapter has provided a phase-wise overview of the entire research framework starting from the conceptualization Using FMECA concepts common defects pertaining to nine major building subsystems were graded statistically using data obtained from field investigation, and questionnaire survey Maintainability scoring principles were derived Next, the relative importance of these nine subsystems in maintainability of the entire building was determined using AHP method Both objective and subjective parameters were addressed Once the final scoring was formulated, the predictive accuracy of COMASS was checked through validation and sensitivity analysis 67 ... 10 4 10 8 11 0 11 4 11 8 12 2 12 5 12 8 13 1 13 4 13 6 13 8 14 0 14 2 14 4 14 7 14 9 15 2 15 4 15 7 15 8 16 0 16 1 16 2 16 2 16 9 17 2 ix Table Description 7 .1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 7 .10 C .1. 1 C .1. 2 C .1. 3 C .1. 4... C .1. 4 C .1. 5 C .1. 6 C .1. 7 C .1. 8 C .1. 9 C .1. 10 C .1. 11 C .1. 12 C .1. 13 C .1. 14 C .1. 15 C .1. 16 C .1. 17 C.2 .1 C.2.2 C.2.3 C.2.4 C.2.5 C.2.6 C.2.7 C.2.8 C.2.9 C.2 .10 C.2 .11 C.2 .12 C.2 .13 C.2 .14 C.2 .15 C.2 .16 C.2 .17 ... types of wall system ……………………………………….……… 15 23 24 24 25 25 26 33 34 38 43 49 55 58 59 69 72 76 88 90 96 10 3 10 9 11 6 11 7 12 4 13 0 14 9 15 5 15 6 15 8 15 9 16 4 16 4 16 5 16 6 16 7 16 8 16 8 19 6 19 8 19 8 19 8 19 9