URBAN TEXTURE ANALYSIS AND ITS RELATION TO BUILDING ENERGY CONSUMPTION MARCEL IGNATIUS (B. Eng. Tarumanagara University, M. Sc. (Building Science), NUS) A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY DEPARTMENT OF BUILDING NATIONAL UNIVERSITY OF SINGAPORE 2014 ACKNOWLEDGEMENT First and foremost, I would like to thank God, whose many blessings have made me who I am today. He has been giving me everything to accomplish this thesis: patience, health, wisdom, and blessing. There are also several people with whom I am indebted for their contribution in this research and study. I would like to express my sincere gratitude to my advisor Prof. Wong Nyuk Hien for the continuous support, patience, and motivation during my Ph.D study and research. His guidance, enthusiasm, encouragement, and vast knowledge have helped me in all the time of research and writing of this thesis. I am forever grateful to Prof Wong for being an extraordinary supervisor who showed me the road and helped me started on this post graduate path. My sincere thanks also goes to Dr. Steve Kardinal Jusuf, for his guidance and stimulating discussions, and not to mention providing me with many insights on my research topic. He was always available for my questions and share generously of his time and vast knowledge. Also for my colleagues and friends: Adrian Chong, Nedyomukti Imam Syafii, Lee See Quin, Erna Tan, Terrence Tan, Norish Ishak, Shan Shan, and Anseina Eliza, who have been helping and supporting me in and out with their research works, which without them, this study would not see the light of the day. I would also like to thank Alex Tan, for his effort and time on research and writing guidance, not to mention reading out my final thesis draft. I want to convey a great thank you for my colleagues from Center for Sustainable Asian Cities (CSAC) for those all those lunch and gathering sessions: Ivan, Mizah, Nazim, Zdravko, Sari, Quyen, Irina, Chelsea, Daniel, Zhang Ji, and Agnes. They have been such great friends and colleagues, whose encouraging words and company kept me going when coffee ad lost its stimulating effect. ii I would like to express thanks to my beloved parents, Joseph and Sevianti, who have been giving birth to me at the first place, and providing me with endless encouragement, motivation, and all of support that I need in whole of my life. I have been given too much love from you both, which is indispensable and invaluable. Lastly, and most importantly, I am sending a very special thanks to my amazing Suri for all her morale support and endless yet tireless care. Her support, encouragement, quiet patience, and unwavering love were undeniably the bedrock upon which the past few years. I am forever grateful for having her not just as a great wife, but also as an excellent mother who has been taking care of our son, Jayden. Thanks for all your consideration, it helps me getting through my post graduate years. iii TABLE OF CONTENTS DECLARATION i  ACKNOWLEDGEMENT ii  TABLE OF CONTENTS . iv  EXECUTIVE SUMMARY ix  LIST OF TABLES . xi  LIST OF FIGURES . xiii  NOMENCLATURE xviii  Chapter INTRODUCTION 20  1.1  Research Problem . 20  1.1.1  Urbanization and Megacities 20  1.1.2  Climate Change impact on building energy performance in tropics. . 21  1.2  Singapore context 22  1.2.1  Geographical overview . 22  1.2.2  Singapore Weather and Climate information . 23  1.2.3  Tropical climate 23  1.2.4  Singapore Urban Development 24  1.2.5  Building energy consumption in Singapore 25  1.3  Research question . 25  1.4  Scope and objectives of research 27  1.5  Organization of study . 29  Chapter LITERATURE REVIEW . 31  2.1  Introduction . 31  2.2  Climate and Built Environment . 34  2.2.1  Atmospheric layers and climate zones 34  2.2.2  Urban area in the tropics and solar radiation. 37  2.2.3  Urban Heat Island (UHI) . 38  2.2.4  Microclimate condition and urban texture 42  2.3  Urban Texture Variables 43  2.3.1  Early studies on urban form and density parameters 45  2.3.2  Floor Area Ratio . 48  2.3.3  Site Coverage . 50  2.3.4  “Spacematrix” variables . 51  2.3.5  Compactness/Compacity . 52  2.3.6  Sky View Factor . 34  2.3.7  Sky Exposure Factor 35  2.4  Urban Texture, Microclimate, and Energy Performance 36  2.5  Cooling Load and Heat gains (ASHRAE, 2009) 41  2.5.1  Conduction gain through exterior surfaces 42  2.5.2  Solar Gain through windows 44  2.5.3  Heat Gain from Fresh Air Intake 45  2.6  Analysis tools . 45  iv 2.7  Summary and Knowledge Gaps 48  Chapter HYPOTHESES and RESEARCH METHODOLOGY 54  3.1  Hypothesis 54  3.2  Methodology . 55  3.3  Parametric Study 58  3.4  Scenario Builder . 60  3.4.1  “Parent” Scenario . 61  3.4.2  “Children” Scenario 62  3.4.3  Scenario naming and examples . 67  3.4.4  Obtaining Sky View Factor ( 3.4.5  Obtaining Sky Exposure Factor ( 3.5  ) . 68  ) . 69  Local Weather Data . 70  3.5.1  Local Outdoor Air Temperature 71  3.5.2  Generating 24-hour temperature profile . 73  3.5.3  Altitude influence on outdoor air temperature 75  3.5.4  Solar Radiation Quantification for Weather Data . 77  3.6  Building Simulation . 77  3.6.1  Integrated Environmental Solutions Virtual Environment (IES-VE) . 78  3.6.2  Boundary condition settings . 80  3.7  Final Deliverables . 84  3.8  Importance and potential contribution of the research . 84  Chapter PRELIMINARY STUDIES 85  4.1  STEVE tool validation on predicting outdoor temperature in urban area. . 85  4.1.1  Background and objective 85  4.1.2  Methodology . 86  4.1.3  Analysis 88  4.1.4  Results and Analysis 88  4.1.5  Importance for the overall research study 99  4.2  Comparison of STEVE tool and ENVI-met as temperature prediction model. 99  4.2.1  Background and objective 99  4.2.2  ENVI-met and STEVE comparison 99  4.2.3  Methodology . 100  4.2.4  Result and discussion 102  4.2.5  Summary 104  4.2.6  Importance for the overall research study 105  Chapter RESULT ANALYSIS AND DISCUSSION . 106  5.1  Introduction . 106  5.2  Local Temperature Profile Result 106  5.2.1  Local and Background Temperature comparison 106  5.2.2  Energy Simulation Comparison using Local and Background ambient temperature data. . 116  5.3  Simulation Result and Analysis 120  v 5.3.1  Selected result outputs from IES-VE . 120  5.3.2  Overall simulation result from IES-VE 121  5.3.3  Correlation analysis with floor and surface Area 127  5.3.4  Data normalization . 128  5.3.5  Normalized data analysis with urban geometry variables 131  5.4  Summary 135  Chapter PREDICTION MODELS DEVELOPMENT 137  6.1  Prediction Model concept . 137  6.2  Non-linear Regression Model . 138  6.3  Correlation analysis 139  6.4  Thermal load unit regression models development . 140  6.4.1  Sensible cooling load unit regression model development 142  6.4.2  Envelope conduction gain unit regression model development . 144  6.4.3  Solar gain unit regression model development 145  6.4.4  Fresh air intake gain unit regression model development . 147  6.5  Summary and discussions . 149  6.6  Models strength and accuracy . 153  6.7  Conclusion 156  Chapter SENSITIVITY ANALYSIS 158  7.1  Introduction . 158  7.2  Establishing variables limit range . 158  7.3  Sensitivity analysis of prediction models 162  7.3.1  Envelope Conduction Gain Unit . 163  7.3.2  Solar Gain Unit . 169  7.3.3  Sensible Cooling Load Unit 175  7.3.4  Fresh Air Intake Gain Unit 181  7.4  Conclusion 183  Chapter MODEL APPLICATION ON CASE STUDIES . 185  8.1  Introduction . 185  8.2  Case Study 185  8.2.1  Locations 185  8.2.2  Acquiring urban texture variables . 187  8.2.3  Thermal load calculation 187  8.3  Case Study 190  8.3.1  Site Selection . 190  8.3.2  Methodology . 191  8.3.3  Design Iterations 193  8.3.4  Thermal load prediction 196  8.3.5  Additional analysis components . 200  8.3.5.1  Wind analysis using Velocity Ratio ( 8.3.5.2  Urban outdoor temperature condition and greenery implementation   203  8.3.5.3  Outdoor thermal comfort analysis   211  vi )   200  8.4  Conclusion 215  8.4.1  Thermal load study . 215  8.4.2  Ambient temperature, wind, surface modifications, and energy performance 216  8.4.3  Outdoor Thermal Comfort 218  8.4.4  Conclusions 220  Chapter CONCLUSION . 223  9.1  Important findings for each research objective 223  9.1.1  First objective . 225  9.1.2  Second objective 227  9.1.3  Third objective 228  9.2  Research contribution 229  9.3  Research limitation . 230  9.4  Suggestions for future work and research . 230  9.5  Design for the future – density and open space? 231  Chapter 10 PUBLICATION AND CONFERENCE LIST . 235  10.1  Conferences . 235  10.2  Publications 236  Chapter 11 REFERENCES . 237  Chapter 12 APPENDICES 242  12.1  Influence of urban density on air temperature within Singapore central business district. 242  12.1.1  Background and Objectives. 242  12.1.2  Case Studies 242  12.1.3  Methodology . 243  12.1.4  Results . 244  12.1.4.1  Temperature maximum ( 12.1.4.2  Temperature average ( 12.1.4.3  Temperature minimum ( ) analysis   244  ) analysis   246  ) analysis  . 247  12.1.1  Relationship between urban morphology and air temperature 247  12.1.2  Summary 249  12.1.3  Importance for the overall research study 250  12.2  Parametric study on urban planning model for high density city. 251  12.2.1  Background and objective 251  12.2.2  Methodology . 251  12.2.3  Results . 253  12.2.3.1  Temperature Maximum ( 12.2.3.2  Temperature Average ( 12.2.3.3  Temperature Minimum ( ) Map  . 253  ) Map  . 254  ) Map   255  12.2.4  Summary 256  12.2.5  Importance for the overall research study 258  12.3  Table of scenarios for parametric study (per site) 259  vii 12.4  Tabulation of all scenarios (per district scale). . 273  12.5  Compilation of monthly dry bulb temperature from Changi MET station (Singapore). 287  12.6  Buildings Envelope Data 288  12.7  ENVI-MET simulation result on different wind speed. . 291  12.8  Predicted Temperature Difference between STEVE and ENVI-met 292  12.9  Comparison of STEVE and ENVI-MET temperature. 293  12.10  Local ambient temperature results (STEVE tool results) . 294  12.11  Thermal output results, simulated by using Apache under IES-VE. 298  12.12  Matrix table for ECGU. 302  12.13  Matrix table for SGU 305  12.14  Energy consumption calculation 307  12.15  Reviewers’ Comment . 310  viii EXECUTIVE SUMMARY Global urbanization has caused significant increase of urban dwellers. In year 2003, United Nations estimated that by year 2030, up to billion people will live in urban areas which will be 61% of the world's population. Urbanization brings major modification on natural landscape; buildings are erected, soil has been transformed into roads and pavement, greenery has been vastly reduced, etc. The deterioration of the urban environment through urbanization can be seen from a phenomenon known as urban heat island (UHI); where cities record higher temperatures in comparison to their non-urbanized surroundings. Overheating due to solar radiation and ambient temperature increase in big cities affect the thermal balance within the environment, where building occupants are adapting by utilizing air conditioning to achieve comfortable internal condition. However, on district level, quantification of climatic condition effect on buildings and vice versa needs further exploration and observation. A parametric study involves urban texture variation has been conducted to observe its effect on district energy performance. Thus, these parametric study scenarios, which focus on non-domestic/office function, are simulated using Integrated Environmental Solutions (IES) to study the identified urban texture related to the energy performance, specifically on district cooling load and external heat gains through building envelopes. This whole process implements weather files that accounts for the UHI effect to derive models which characterize certain urban texture along with its related energy performance. The findings serve to identify the relevant urban texture variables which characterize urban density and form, such as floor area ratio (FAR), open space ratio (OSR), story height (ST), gross site coverage, (GSC), and sky view factor (SVF). Thermal load calculation method is developed to illustrate how combination of several urban texture variables affects or influences the cooling load and heat gain ix 12.14 Energy consumption calculation PRECINCT DATA FAR/Plot Ratio : 10 Block Area (buildable) : 63,000 m2 Max gross floor area (GFA) : 630,000 m2 AIR CONDITIONING SYSTEM Air Conditioning type : Central chilled water plant Design System Efficiency (DSE) : 0.7 kW/RT (1 RT = 3.5168525 kW) Mechanical fan system : 0.67 W per m3/h (Variable Air Volume type) (Sensible cooling load was obtained using the thermal load models) OCCUPANCY Occupancy density (office) : 10 m2/person Operating hours : 55 hours/week People Ventilation Rate : 5.5 l/s or 19.8 m3/h (CMH) Number of people : Gross Floor Area/10 m2 = 63,000 pax POWER BUDGET Lighting Power density : 15 W/m2 Equipment Power density : 16 W/m2 307 Calculation Table AIR CONDITIONING  1  40%  2  3  4  50%  5  6  7  60%  8  9  1‐ mass  2‐ mass  4‐ mass  1‐ mass  2‐ mass  4‐ mass  1‐ mass  2‐ mass  4‐ mass  LIGHTING  (MWh/yr)  District Energy Performance  (kWh/m2/yr) Air  D  Total   (MWh/yr)  Water  C  EQUIPMENT  (MWh/yr)  B  ENERGY   CONSUMPTION  (MWh/yr)  A  ENERGY   CONSUMPTION  (MWh/yr)  ANNUAL  COOLING LOAD  (MWh/yr)  Massing Type Site Coverage T  118165  23519  23903  27027  28829  103277  163.93  119357  23756  23903  27027  28829  103515  164.31  116849  23257  23903  27027  28829  103015  163.52  102615  20424  23903  27027  28829  100182  159.02  104565  20812  23903  27027  28829  100570  159.64  99312  19766  23903  27027  28829  99525  157.98  91979  18307  23903  27027  28829  98065  155.66  94853  18879  23903  27027  28829  98637  156.57  93565  18623  23903  27027  28829  98381  156.16  Notes: T Annual cooling load (MWh), using thermal load model for sensible cooling load. Annual energy consumption for air conditioning (water side), using the T value and Design System Efficiency (DSE) of 0.7 kW/RT which the converted into kW value. A Hence: (T x DSE) / 3.517 Annual energy consumption for air conditioning (water side) Several steps and assumptions are required to obtain this output: People Ventilation Rate District occupancy = 630,000 m2/10 m2 = 63,000 people B Fresh air requirement = 63,000 x 19.8 CMH = 1,247,400 CMH Mechanical Fan System energy consumption Then, it is assumed that the fresh air requirement for occupants counts 10% from total air circulation. Which means: 308 Total air circulation = 11,247,400 CMH Energy consumption for air conditioning (air side) is using the mechanical fan system efficiency of 0.67 W/CMH (Variable Air Volume type): 1,247,400 CMH x 0.67 W/CMH = 8,357,580 W Hence, the annual energy consumption for mechanical fan system: 8,357,580 W x 55 hours x 52 weeks x 10-6 = 23,902.68 MWh Annual energy consumption for lighting, uses the lighting power density (15 W/m2) C C = Floor Area x 15 W/m2 x 55 hours x 52 weeks x 10-6 = 27,027 MWh Annual energy consumption for equipment, uses the equipment power density (16 D W/m2) D = Floor Area x 16 W/m2 x 55 hours x 52 weeks x 10-6 = 28,828 MWh 309 12.15 Reviewers’ Comment Reviewer #1 # Comments Responses Author listed out a few typology of urban Real world example of typical city grids morphology. In order to illustrate the of commercial district has been realistic nature of these choices, author displayed in Figure 3.3. should draw a parallel with any real world example, as much as possible. All equations should not be labelled as, Equation labels have been revised say, "equation 1.2", but the conventional format can be followed, that is, (1). Author used STEVE tool to deploy 10 • There was a mistake regarding the measurement points. How and why are number of measurement points. It is these chosen? supposed to be 39 points. 10 was the number of points that was implemented back at the initial stage, then it was changed to 39 points to obtain more accurate information from the whole precinct (see Figure 3.12). • The main criterion is that these points’ locations need to be fixed for all the scenarios. Also, it is important to avoid. Hence, these points are deployed mostly at street canyons, 16 at the perimeter, and additional 11 at the center. • The remaining 12 are located at the site boundary lines. There are no points deployed at the within the site as to avoid the possibility where there is a measurement point located under a building mass. • Throughout this thesis, STEVE tool Both STEVE tool description and its was mentioned and indeed this weaknesses have been added (Ch formed an important part of the 310 # Comments Responses research methodology. Author 3.5.1). should add a brief description on the concept behind STEVE tool. • Since STEVE tool is used and once again included throughout chapter 4, it is important to note the weaknesses of STEVE as well. This is important since author's own work depends heavily on the results of STEVE tool.  Equation 3.9 The purpose of using this model is to Mathematical expressions were given for obtain both diffuse and direct (Kd)t. It is useful for author to reflect on radiation for simulation purpose. This the physical implications, that is, write in is mainly because local full what these mean and their relevance meteorological stations measure to the study at hand. only the hourly or daily values of global radiation. Hawlader developed a method of estimating the diffuse component from the measured values of global radiation. Once the ratio of diffuse component or (Kd)h is known, the magnitude of the beam component can easily be obtained by subtracting the diffuse component from the global radiation values.  Hence, before running the simulation, the weather data template for each scenario has been equipped with proper diffuse and direct radiation component which refers on the recorded local weather data.  Figure 3.17 The graphs refer to the occupancy It is not well explained. Author should and schedule profile of lighting, use one of the graphs as example, and equipment, and infiltration. The X- explain in details what it means and its axis refers to the time of the day, 311 # Comments Responses overall implication to the study. while the Y-axis refers to the magnitude of the occupancy/usage.  These boundary conditions are kept fixed for all IES simulation scenarios, and in accordance to codes of practice in of office in Singapore Author mentioned that window-to-wall The 0.5 WWR was obtained based on ratio is fixed at 0.5 Although this is taken the survey data from 27 office buildings as a constant, it is important to make a around Singapore CBD. The average of note why this is reasonable assumption WWR from all these buildings is around and constant. 0.5. Hence, 0.5 WWR was considered as the assumption for the envelope condition of the scenario models. The differences between ENVI-met and A more detail observation on the STEVE tool results need to be differences between ENVI-met and elaborated further. STEVE tool has been published into journal paper (Wong et al., 2012)  Ch 5.2.2  criterions on selecting the scenarios: Author uses random selection to pick (1) The smallest site coverage and the random scenarios.  the most compact scenario, which is In some cases, complete random 20-20-01. (2) No 2-6 are the selected selection may not be the best option.  scenarios which have similar surface Author needs to explain whether area. (3) No 7-11 are the selected specific parameters are fixed and scenarios which have the same controlled randomization is done GFA. Buildings height has been kept instead, and justify doing so.  fixed each scenario. For example, as shown in figure 5.8, why is the height of all the buildings the same? It seems that some form of controlled randomization has in fact been done and so author needs to specifically explain this in this  These criterions on selecting the scenarios are based on the assumption that urban layout with similar GFA or envelope surface area may yield different impact on its outdoor temperature. chapter. 10 By looking at Table 5.2, there are The denominator is "total GFA" and "total No, there is no double counting of area. surface area". Is there any double- This is because total GFA refers to the floor area, and total surface refers to 312 # Comments Responses counting of area? summation of envelope area (walls and rooftop). 11  Figure 5.19 contains many graphs. These graphs are the scatter plots to Author needs to highlight the key points - see the relationship between especially any differences- in the link different urban texture variables (X- between the x and y variables. axis) and the dependent variables (SCL, ECG, SG, and FAIG). They show several variables which has strong correlation between them.  More importantly, these scatter plots highlight that there some who have strong curvilinear relationship, and some linear ones. 12   This line "do not have issues regarding correlation signs, . ", accepted in the statistic field as needs to be re-written so that it is common explanation to denote the more appropriate for an academic different correlation signs.  publication.  due to behavior of the models’ tables requires more description. variables (the Beta coefficients (B) Foremost, taking table 6.3 in and Pearson Correlation (r), showed particular, although the first model contradicting trends).  The correlation coefficient (r) that, to some extent, second model measures the strength of linear is just as good and even more relationship between two variables "powerful" than the first because and is bounded between -1 and 1, rand R2 are comparable with the first inclusive. Correlations close to zero model. represent no linear association Author highlighted the fact that "B" between the variables, whereas and "r" having different signs as a correlations close to -1 or +1 indicate reason to ignore the associated strong linear negative or positive model. Why is this a criterion? relationship, respectively.  Explanation is needed.  The second model was not chosen Assessment of table 6.3 and similar was chosen but one may also argue  That particular line has been The Beta coefficient or regression Bottom of the page, author said that slope (B) measures the "steepness" "logC appears to have different of the linear relationship between signs" and it is ignored in the two variables and can take any value subsequent regression. Why is this from −∞ to +∞. Slopes near zero 313 # Comments Responses the case? What is the physical mean that the response (Y) variable implication of omitting "C"? changes slowly as the predictor (X) variable changes. Slopes that are further from zero (either in the negative or positive direction) mean the response changes more rapidly as the predictor changes.  So, with these descriptions, the B and r should have the same sign (+ or -), but will almost never have the same value. Reviewer #2 # Comments Responses Ch 2.6 Apart from basic building geometries, I not see how the reservation building energy simulation (which usually expressed in the last paragraph is is done at the latter stage of the design material when one only needs to process) requires other data such as estimate the boundary conditions of the envelope constructions, interior building to be simulated. This only materials, air conditioning system, requires the basic building geometries. lighting type, or even the occupancy Am I missing something? schedule. These boundary conditions are required to obtain an optimum accuracy of the result.   On "knowledge gaps", I found the Knowledge gaps have been revised and statements over simplistic. rearrange by focusing into relevant Without demonstrating how the priorities. student appreciates the latest advances in coupled simulation techniques, statement and are a bit blatant.  In addition, these days, statement is also not too accurate. In the field of urban climatology, there are known methods, and there are 314 # Comments Responses already many studies.  Statement and may be the keys to the study. Ch 3.5.2 Although the method is not perfect, but The mathematical working of 3.5.2 is, to currently it is the only way to transform me, inappropriate. The method was not the Tmax, Tmin, and Tavg from STEVE developed and should never be used to into a 24-hour profile, with referring to estimate the temperature variations the recorded temperature from MET within the street canyon. But on the other station. Furthermore, this method at least hand, for a tall building urban context in acknowledges the temperature variation the tropics, the data is not available. in the urban canyons due to building geometry and the surroundings. Figure 4.3 – 4.4 The days are selected based on the How are the days "selected"? How about weather condition. STEVE was other days of the measurement period? developed by using clear and calm days’ data. Hence, only days which meet these criterions were selected for comparison. The graphs displayed in Fig. 4.3 – 4.4 are the examples of a clear and calm day weather condition. Meanwhile Fig. 4.5 – 4.6 use all filtered clear and calm days. I wonder why the STEVE and Envimet This comparison is a mini research study comparison cannot be done using a few which was done at the early stage of the of the scenarios that were previously PhD, before the parametric approach on defined. Instead it uses a model of developing the scenarios was buildings. Please explain. established.    It seems that only the temperature of There are two fields that were the weather file is modified in the modified. First one was temperature, simulation, correct? How about the second one was dew point. The wind? If only temperature is temperature was obtained from considered, then the observations in STEVE tool, while the dew point was p.127 are obvious. obtained by using both temperature and RH, by using this equation: And if this "obvious" observation is Td = T - ((100 - RH)/5.) translated into design guidelines,  then it is limited. An urban 315 Where Td is dew point temperature # Comments Responses environment optimized for air (in degrees Celsius), T is observed temperature may not be conducive temperature (in degrees Celsius), for living in the Tropics when and RH is relative humidity (in ventilation needs to be taken into percent). Apparently this relationship account. is fairly accurate for relative humidity values above 50%. It seems from Table 5.1 that the urban The small variation of predicted Tmin is predicted Tmin is only slightly higher due to the lack of solar radiation; hence than that of the MET station, and the the ambient temperature difference is difference is smaller than that of the most likely due to release of long wave Tmax. But based on literature, night time radiation from the urban surfaces. While UHI should be higher than daytime UHI, during day time, the high variation of any explanation? ambient temperature is because of both solar radiation exposure and overshadowing from buildings. Also, ref to Fig 5.2, I cannot understand The Tmax refer to the outdoor how Tmax can be higher when SVF is temperature condition at the hottest higher? It seems to mean that as one period of the day in Singapore (2-3PM). builds more, Tmax will lower! Hence, it is logical when the SVF is higher means the area is more exposed to the solar radiation, hence brings up the temperature. See also Fig. 5.3. The first paragraph of The observation made at for Fig 5.3 p.109 is most difficult to digest. refers to hottest period during day time especially for designing in the Tropics. It (Tmax), hence the logical explanation is actually demonstrates the key weakness that when an urban layout becomes of the study using shading and air more open, it increases the exposure to temperature as its main criteria solar radiation, hence during peak hour (2-3PM), the Tmax at open spaces is higher than at narrow canyons due to shading. 316 Reviewer #3 # Comments Responses Climate change part is rarely discussed This section has been removed. or assessed, and could therefore be omitted. Research limitation mentions “Future This aspect has been addressed in the Predictions” are not able to be assessed. future works (Ch 9.4) This aspect may be valuable in the context of climate change prediction. Ch 1.1.3 This section will be amended and moved Why Urban tools are introduced here? to the literature review chapter (Ch 2.6)  Literature review is poorly structured, Literature review has been revised and it feels rushed, picking on random reordered properply based on the range of topics loosely related to comments. thesis.  Not comprehensive and critical that helps to focus on topic  Need a more succinct, focused and relevant review would be adequate  Work back from findings to help determine what this review should focus on Figure 2.2 is no helpful or clear: the The intention here is not to divide the division between UHI, outdoor comfort components, it is to illustrate that and daylighting seems illogical (there are microclimate study comprises many 'aspects' that overlap all three topics). aspects which are not feasible to handle Although the thesis may not tackle everything into a single thesis. daylighting (why not?) it is nevertheless Furthermore, the author focuses on the important to recognise that urban building geometry, variation of geometry influences daylight availability temperature, shading, and their impact which in turn impacts on the cooling load on energy usage due to external heat and total energy consumption. Perhaps gain variation. light is even the most sensitive parameter with respect to built form.   It is a shame that courtyard form is 317 At the initial stage, the scope of the # Comments   Responses excluded from the list. study focuses on a CBD area which It should be noted that there is a mostly comprises offices or non- fundamental problem with taking domestic buildings. From local regular arrays (as those early observation, rectangular and slab examples of Martin and March) to type are the most common shape. assess the urban microclimate and While courtyard is not a common building energy use (which is why building type in Singapore, even for Ratti et al used real complex urban housing (since some literatures forms). mention that courtyard represents A grid by definition means that there typical housing form in hot and arid are prevailing directional climate). characteristics which will mean that Although, it is also true that since certain wind and solar orientations this study deals with many variations may be emphasised. Although real of building form in different density, cities, including Singapore, have the courtyard probably should have grid-like patterns, there are not been excluded.  variations in building height and   The purpose of using parametric dimensions, openings between approach is to control and test towers or raised podium or piloti various scenarios, and the configurations, etc., that will create possibilities are endless. Hence, increased randomness. author tries to deal with as many Research work by others has shown scenarios as possible which can fit that this randomness of the real into the time frame. urban texture can be very influential  The advantage of using real case with respect to the microclimate. This study is that it reflects the real presents a challenge to the adoption condition. However the downside, of simplified regular urban arrays for the findings and analysis are limited analysis, and this challenge must be contextually to that site only. addressed, discussed or at least noted.  Figure 5.1  Explain why the temperature range is small and above ambient for is due to the lack of solar radiation; hence the ambient temperature difference is most likely due to Tmin?  The small variation of predicted Tmin release of long wave radiation from Should the range not be larger, as the urban surfaces. While during day defined by the UHI effect being time, the high variation of ambient maximum at night? temperature is because of both solar 318 # Comments  Responses Why is the range larger for Tmax? radiation exposure and What is the data referred to in Table overshadowing from buildings.  5.1- from STEVE or measured?  The data referred in Table 5.1 are the results of STEVE calculation Why not include both? from all 462 hypothetical scenarios, hence measurement data were not available.  Explain why "urban layout . has a Oke refers that heat island is greater impact on daytime temperatures typically rather weak or absent in the than on night-time", when Oke refer to daytime but grows rapidly following the maximum UHI occurring at night sunset, reaching its maximum time? intensity (which can be translated as difference between rural background and highest urban temperatures 3-5 h later).  The finding from this section does not imply the UHI impact, but merely the diurnal ambient temperature condition, which is highly influenced by the urban geometry due to its SVF or openness condition. Ch. 5.2.2 By looking at Table 5.2, there are You state that "eleven scenarios have criterions on selecting the scenarios: (1) been randomly selected"- this does not The smallest site coverage and the most come across as very rigorous or compact scenario, which is 20-20-01. (2) scientific as you may have randomly No 2-6 are the selected scenarios which picked unrepresentative scenarios. have similar surface area. (3) No 7-11 Justify the selection process. are the selected scenarios which have the same GFA. These criterions on selecting the scenarios are based on the assumption that urban layout with similar GFA or envelope surface area may yield different impact on its outdoor temperature. Table 5.2 refers to "Total Mass"- what is Total mass refers the number of this? Why not also include SVF and buildings within each site. SVF and 319 # Comments Responses surface to volume ratios for the 11 forms, surface-volume ratios has been added. to be more complete.   You should clarify if the differences  in energy performance between scenarios is an accumulation of both forms are more directly significant form and local temperature aspect. that the indirect effect of the change But from the 11 scenarios, it of temperature profile. indicates that variation in energy performances is significantly due to Is the variation within the form. 'background' climate data on energy  greater than the variation in ‘local’ scenarios would have been For example, form 20-02-01 uses simulated under the same weather 190 kWh/sqm, whereas form 70-20- condition (there would be no 01 uses 160 kWh/sqm- so about 30 localized temperature). Hence, the kWh/sqm is due to form. Whereas energy simulation result would have for 70-20-01 the difference between not been able to capture the energy ‘background’ and ‘local' is about 10 variation due to temperature profile. kWh/sqm. Therefore, form affects  energy significantly more than temperature profile and using the If so, what are the statistical energy background climate data instead, we use differences for all the forms and only obtain the energy variation due temperatures?  to form. Now, with generating the Are there scenarios where the form temperature profile to be embedded and temperature forces can cancel into the weather data, we can get a each other out (i.e. the negative more accurate energy variation, temperature conditions are because the localized temperature counteracted by more energy profile is the direct result of the urban efficient form choices)?   From the example of 20-20-01 and 70-20-01, if we exclude the local microclimate. Is this true?  By using the “background climate” for the weather data, it means all 462 microclimate?  The energy variation from different form. It should be noted that the outcomes  The objective of using parametric of conduction gain will be very approach is to control the variables, different for more complex forms- i.e. since the variability can be limitless. a building with a facetted facade will Hence, this study tries to include as have a much larger surface area many variations as possible within than its simplified form suggests. the feasible time frame.  The simplified forms thus give 320 Although the shape is using the # Comments Responses simplistic outcomes which in simple rectangular form, this study themselves are not very significant tries to vary the layout arrangement and therefore will be very sensitive to and openness through compactness other design aspects. control.  However the concerns will be addressed, and for future works, it is intended to test out more complex or radical forms.   The relatively small differences in performance that have been case study can be explained by the presented are a cause for concern boundary condition which has been because it raises questions over the set by the FAR control. Means, there robustness of the assumptions, are limitations of the range of forms uncertainties and errors, and the which can be tested due to meet the strength with which conclusions can maximum floor area.  be stated.   Regarding the wind speed For example, the author notes that difference, it is true that a more the maximum wind speed difference complex geometric form should have is only 0.2m/s- should a wider and been tested. But due to time more complex range of forms have limitation, this cannot be done under been tested? current time frame, but it has been taken into consideration for future Compactness only seems to make a study. difference of about kWh/sqm out of a total of about 152 kWh/sqm (Figure  A more radical difference will surely 8.17). The absolute maximum be encountered when urban forms difference in energy performance is with different FAR are tested, since only 13 kWh/sqm out of an average the floor area density will be of about 157 kWh/sqm- this seems significantly different.  very modest given the radical  The limitations encountered in this Overall, this practice has difference in simplified urban objecives. First to show how to use planning configurations tested (i.e. it the thermal load models in a case would not convince a planning study. Secondly, how to integrate authority to change its urban design other microclimatic analyses to policy which has many more complement the thermal load aspect. objectives than only to reduce Although it is not perfect at the energy). current stage, this attempt shows The change in comfort vote is no how different microclimatic aspects provides a more complete 321 # Comments Responses more than about 0.2 of one vote. information on the studied area. The average temperature difference   Hence, the small difference on the is less than degrees. result can be translated that other On page 235 the author claims that variant of shapes can be further the impact of form on energy is tested. "significant" but without qualifying this (e.g. compared to changing the facade design or the cooling system) it is difficult to defend the use of this word. 322 [...]... correlation, significance, and collineartity statistics tests under SPSS using stepwhise method 148  Table 6.8 Interpretation of SCLU model in relation to its predictors 149  Table 6.9 Interpretation of ECGU model in relation to its predictors 150  Table 6.10 Interpretation of SGU model in relation to its predictors 150  Table 6.11 Interpretation of FAIGU model in relation to. .. impact on building energy performance, since the usage of heating and cooling are highly related to both temperature and weather variations Building sector is accountable for more than 40% of global energy consumption and 30% of global greenhouse emissions, which comes from both commercial and residential usage Among the factors that contribute to the buildings’ emissions are building design, building. .. conditions (Wong and Chen, 2009a) 1.2.5 Building energy consumption in Singapore The hot and humid climate of Singapore also results in the excessive use of air conditioning in the buildings during daytime Generally, in hot and humid climate countries, the cooling demand in a building varies from 50-60% of the total building energy consumption (Kunchornrat et al., 2009) Research done by Building and Construction... the energy consumption by buildings alone accounts to about 57% of the whole electricity consumption in Singapore (Bing et al., 2005) Energy required in air conditioning is very much related to the cooling energy demand and the equipment’s operating efficiencies Cooling energy demand of a building is further divided into three groups; heat load from human activities, heat load from receptacle load and. .. methodology to study energy performance based on urban forms and geometry in order to answer the question Many simulation tools available on the market are able to perform a 27 rigorous and robust simulation and calculation of a certain building design However, it has become a reasonably difficult and an exhausting job to conduct the similar study on the urban scale Before buildings are built from empty lands,... form and microclimatic condition 2 To develop a prediction model that characterizes district/precinct energy performance due to its physical parameters and micro-climatic condition 3 To develop a matrix of urban texture derived from the geometry variables in order to determine the energy performance Understanding how urban texture could play a role in determining its urban microclimatic condition should... incorporates several form descriptors and heat gains (Ratti et al., 2005) 39  Figure 2.22 Factors that affect energy consumption in buildings; according to Baker and Steemers, building design accounts for a 2.5X variation, system design and occupants behaviour for a 2X variation each; the contribution of the urban context is not quantified (Baker and Steemers, 1992; Baker and Steemers, 2000; Ratti... about 52% of energy consumed in a typical commercial building in Singapore is utilized for air conditioning (Lee et al., 2004) Therefore, it is important to explore how far the urban texture and micro-climatic condition are affecting each other, and the possibility to transform this relationship into a practical analysis approach for urban performance evaluation It is deemed necessary to embark on... various urban designs to select the most compatible design program x LIST OF TABLES Table 2.1 The table shows the relationship between FSI, GSI, OSR, and L (Pont and Haupt, 2004) 51  Table 2.2 Urban Texture Variables and their relationship with microclimate studies 48  Table 2.3 Studies which emphasize the impact of urban texture variables impact on both microclimate and energy ... between urban texture, microclimate, and building energy usage 42  Figure 2.11 Two archetypal urban patterns, based on pavilions and courts (black represents buildings) with the same site coverage, building height and total floor space (Martin and March, 1972) 46  Figure 2.12 Fresnel’s diagram: all concentric squared annuluses have the same surface area, which is also equal to . model in relation to its predictors 150 Table 6.10. Interpretation of SG U model in relation to its predictors 150 Table 6.11. Interpretation of FAIGU model in relation to its predictors 150 Table. URBAN TEXTURE ANALYSIS AND ITS RELATION TO BUILDING ENERGY CONSUMPTION MARCEL IGNATIUS (B. Eng. Tarumanagara University, M. Sc. (Building Science), NUS). for the UHI effect to derive models which characterize certain urban texture along with its related energy performance. The findings serve to identify the relevant urban texture variables which