Abstract
In recent years, sustainable design methods have become a major concern within the building industry. There is also a growing awareness of the impact urban morphologies have on the overall energy and fuel consumption of a city. This paper investigates digital form-finding methods for generating an urban tissue to suit climatic conditions. In this research, a cascading series of genetic algorithms at multiple scales is coupled with environmental evaluation methods as fitness criteria. The methods devised in this paper integrate evaluation tools written with an object-oriented scripting language together with the Galapagos genetic solver in the Rhino/Grasshopper/Python platform. It is shown that the developed methods can be used to create large-scale urban layouts with improved street-level climate conditions as well as aggregations of buildings that function together to improve environmental and architectural parameters. The methodology developed in this paper is tested on a site with an area of approximately 1 km in Brooklyn, New York, chosen because its climate features a large yearly variation in temperature and wind regime. The existing surrounding urban fabric, along with the local climatic conditions, is taken as the initial input in order to develop algorithmic processes with sensitivity to the site context. Keywords: urban fabric; climatic optimization; multi-objective optimization; multiscalar; genetic algorithm 1. Introduction for generating an urban fabric that can utilise In practice, the development of an urban tissue environmental analysis as a driver for the creation typically occurs in separate phases at several different of forms, with the objective of improving the urban scales and is described in terms of separate projects, environment's microclimatic conditions. In this such as the master plan, detailed master plan and paper, genetic algorithms serve as the computational building design. In urban scenarios dealing with framework for a consideration of multiple evaluation multiple scales, decisions suited to one scale might criteria to determine the fitness of an urban morphology have an unfavourable effect on others with respect to at multiple scales. The research methodology microclimatic conditions. The climate-adapted urban represents an attempt to create an integrated digital process should therefore aim to integrate the different workflow on a single computational platform that uses scales—urban, block and building—in order to allow a genetic solver together with customised evaluation for control not only of the behaviour of the buildings tools for designing an urban fabric. themselves, but also of the nonbuilt spaces between In the following sections, environmental criteria such them. as solar access, wind flow and energy consumption are In contemporary architectural practice, many used as fitness criteria along with other architectural digital tools have already been devised to quantify parameters. These customised evaluation tools use the performance of a building in its environment. simplified calculation methods to account for the However, these tools are typically used in a posteriori runtime of the algorithms and are written for the Rhino/ analysis of the architectural design, that is, to evaluate Grasshopper/Python computational platform. The preconceived design alternatives, rather than as results of the algorithms produce street-grid layouts, integrated processes to generate design options. public open-space distribution, building masses This study focuses on form-finding methodologies and facade surfaces that are adapted to their local climate. The correlations of the design elements in the urban settings are considered in forming the overall *Contact Author: Soungmin Yu, Architect ARB/RIBA, algorithmic process, which comprises a cascading Kyung Gi Do, Yong In, Gi Hung Gu, Joong Dong series of genetic algorithms. The resulting morphology 962-1 Dong Yung Jae 1 Dong 201 Ho, South Korea of the algorithms presented in this paper is compared Tel: +82-31-681-445-553 to the existing urban fabrics of their respective sites. E-mail: soungminyu@gmail.com In this comparison, the result of the design experiment ( Received October 7, 2013 ; accepted March 10, 2014 ) Journal of Asian Architecture and Building Engineering/May 2014/324 317 showed improvements in the majority of the criteria, southeast/northwest orientation with a staggered grid to suggesting an overall improvement while maintaining maximise the self-shading capability of the buildings. In an architectural similarity to their surroundings. addition, the materials with which the buildings are clad help to minimise heat gain in the streets due to reflected 2. Theoretical Background radiation. It is suggested that this multiscalar approach, 2.1 Urban Scale combining the orientation of the urban grid, the massing According to Salat, unsuitable urban morphologies distribution of each block and the effect of the building can double the energy consumption of urban tissues. envelope on both its interior and the street, represents In support of his view, an analysis of multiple cities, a direction for further exploration of the correlations of such as London, Berlin, Toulouse and Paris, has the different design elements considered at each scale. been carried out by research undertaken at MIT in 2.3 Genetic Algorithms collaboration with the University of Cambridge and An emerging discipline in architecture is the at the Centre Scientifique et Technique du Batiment use of computational genetic algorithms (GAs) to in collaboration with the Atelier Parisien d'Urbanism. solve design problems. GAs are modelled on the Both studies draw the conclusion that certain urban evolutionary process of natural selection to search for textures can reduce the energy consumption and carbon a fittest individual according to set criteria by repeating emissions of a city, confirming the importance of the the four phases of initiation, selection, crossover, and morphology of urban fabric to sustainability. These mutation (Kim, 2012). The advantages of GAs and findings indicate the importance of understanding the their ability to search through a vast design space for a potential of urban forms for climatic control. However, solution have been noted by numerous scholars. Adolphe (2000) has noted that it is difficult to evaluate Furthermore, GAs have an inherent ability to the effect of urban configuration on outdoor climatic integrate simulation tools as a part of the optimization conditions and the energy balance of a building due process, and when considering multiobjective to the extreme morphological heterogeneity of urban optimisation, they can demonstrate the results of the contexts. Adolphe proposes a set of morphological trade-off of different criteria (Caldas, 2003). These parameters including, among others, rugosity, advantages make optimisation routines using GAs porosity and sinuosity as three-dimensional geometric a desirable form-finding method for performance- abstractions to measure wind flow around and solar driven solutions that rely on simulation tools. GAs are access and energy balance of buildings. Moreover, now becoming a more common form-finding tool in research conducted by Kakon and Nobuo shows architecture, where practices such as SOM use GAs to the correlation between pedestrian comfort and the find the optimal solution for daylight quality, structural street canyon, where they employed street sections as efficiency, construction cost, view quality, carbon geometrical abstractions of urban forms. The present footprint, acoustic quality, programmatic compliance, study adopts these principles of geometrical abstraction etc. The availability of user-friendly evolutionary to integrate them into a process used to generate engines, such as Galapagos in the Rhino/Grasshopper urban forms. The research described above shows environment, encourages designers to utilise GAs that processing the amount of data needed to describe as a part of their design process. This tool vastly a large area of urban fabric is a challenge that needs simplifies the writing of a GA and enables the seamless to be addressed by using methods employing a set of integration of the generative process with the analysis geometric abstractions. process. Despite these advantages, GAs also have 2.2 Architectural Scale some known problems, one of which is the relatively Historically, the building envelope has had multiple long runtime of the algorithm, which results from the aesthetic and functional design implications, and one of time it takes to evaluate every single individual within these is the function of climate control. Today, software a population. Integrating the simulation tool with the helps designers simulate the heat and airflow around the genetic engine is greatly limited by computing power, building envelope, and they can use this information and in many cases the simulation tool needs to be to improve the performance of the design. The two simplified in order to provide real-time feedback. main concerns for designing a building envelope with passive design principles are the form and materiality 3. Methods of the building. The general massing, locations and This section discusses the overall algorithm and sizes of the openings on the building envelope have a focuses on the measures taken in developing the strong influence on the building's daylight quality, solar customised evaluation tools. In the experiment described heat gain and heat transfer, all of which are directly in this paper, the urban context is broken down into related to the building's energy consumption. urban, block and architectural scales, and each scale Masdar City in Abu Dhabi, designed by Foster + is addressed with a GA with multiple fitness criteria. Partners, is recognised as a precedent for a design At each stage, the border of the site is considered a approach that integrates urban form with individual part of the generative input for the simulation. Starting building scale. In Masdar, the street arrangement has a the process with the larger scale therefore allows the 318 JAABE vol.13 no.2 May 2014 Soungmin Yu information to trickle down, informing all the possible these tests were then factored with constants relating to results with data acquired in the earlier stages. diffuse solar radiation in the site according to the time The algorithms for producing geometric models of the year (Robinson, 2011). at different scales and the tools used for determining Both methods were calibrated by testing a group of fitness criteria were written by the authors as 20 models against the Ecotect solar radiation tool, and customised evaluation tools. They were designed they show similar values and tendencies, validating to be suitable for the operation of a GA within the their use in the generative algorithm. limits of the computational framework of a laptop. The experiments investigated different evaluative strategies to be considered at different scales based on the geometric data to be processed. Different evaluation criteria were considered to determine fitness at different scales. A total of 10 evaluation criteria were considered in the experiment, each of which were translated into customised tools. The criteria were park influence, wind flow, winter solar exposure, summer solar exposure, porosity, passive zone, connectivity, heat transfer, sky-view factor and reflection. Of the customised evaluation tools, solar access, wind flow and energy consumption are further discussed in this section. Two main requirements were considered when writing these simulative tools: confirmation of results against existing software tools and a runtime efficiency Fig.2. Solar Radiation Result Comparison of GhPython Script enabling rapid numerous iterations. These methods and Ecotect took advantage of the use of different scales in the simulation, resulting in an analysis that was sufficient CFD analysis of urban fabrics is extremely time to describe the current levels of abstraction. consuming, taking up long periods of calculation in any Two methods were developed for use at different of the prevailing software packages (such as Winair and scales to calculate solar radiation. For the urban scale, ANSYS). Alternatively, different abstractions of the urban this study adopted an abstract street-canyon model from fabric were attempted by the authors to enable predictions Robinson (2011). This model relies on the angle of of wind flow without resorting to actual simulation. abstract facades or streets relative to the north and the In view of the fact that the geometry of the street height and distance of obstructions to quickly calculate canyon has a large effect on the flow and turbulence an estimate of their solar exposure over given time patterns of the wind (Nakamura and Oke, 1988), it periods. This method lacks accuracy in comparison to can be seen that typical street sections have typical other types of measurements, with results describing flow patterns (Erell, Pearlmutter & Williamson, 2011). only the percentage of time a surface is exposed to A hypothesis was formulated that the average rate of direct sun. However, it is extremely efficient, because wind flow on a given street has a good correlation it is based on abstract trigonometric calculations and with the relation between its width, height (on both can be performed on a database by describing solely windward and leeward sides) and length and its angle the street width and the angle and height of the two to the wind. In order to test this hypothesis, a series facades, without resorting to actual modelling. of CFD simulations was performed on models with different height parameters (which can appear on each side of the canyon), different widths and 15-degree variations in the wind direction. Fig.1. Ray Counting Method for Solar Radiation and Result of GhPython Script On a smaller scale, this study implemented the Fig.3. Examples of Wind Simulation Library Samples widely accepted ray counting method, testing the obstruction of solar vectors at different times as well The results were stored in a library and then used as the reflection of the same vectors. The results of to predict the wind behaviour of all the streets of a JAABE vol.13 no.2 May 2014 Soungmin Yu 319 neighbourhood-scale fabric. In this specific experiment, results of the evaluation tool to those of a validated the library took into consideration the morphological tool such as the Vasari energy calculation. A total characteristics of the existing site conditions of 20 individual forms of varied shapes and sizes preselected for the experiment in Brooklyn, New York, were evaluated with both the Vasari software and the before determining the range samples for street-aspect evaluation tool included in Grasshopper. It was clear ratio and wind direction. The library samples included from the graphs and the table that although the numeric a building-height range of 10-50 m and a street-width values of the results differed between the tools, the range of 15-35 m. For all the morphological variants, inclinations of the graphs showed the same trends in the possible wind angles were measured at every 15 both types of analysis. Although a precise prediction degrees, and the results were added to the library for of the energy consumption was not achieved, it was reference. The combination of the ranges of the width possible to use this method to compare different forms and height together with the varying angle of the wind and identify those that are preferable in terms of energy provided 189 variations, which were stored in the use. library for the initial experiment. When these variations were compared to a Winair simulation of the same fabric, it was shown that a strong correlation in the tendency of the results could be expected, especially if the dimensions of the streets of the tested fabric were similar to those in the library. Because the CFD simulations contained in the library can be performed in advance, the library is theoretically unlimited in the amount of samples it can contain, and its accuracy will increase in direct relation to its size. Fig.6. Result Comparison between GhPython Script of Heat Transfer and Vasari Energy Simulation As further evaluative tools, additional parameters were encoded into the Rhino/Grasshopper/Python environment. Passive zone ratio, the first of these parameters, is an offset from the building envelope, usually double the net floor height in size, which is said to be the maximum depth at which natural lighting Fig.4. Result Comparison between GhPython Script (right) and and ventilation can be utilised. Additionally, sky view Winair (left) factor (SVF) was used to describe the cross-sectional For an energy evaluation tool, the widely accepted proportions of street geometries. It was employed calculations of building net heat transfer can be to describe the amount of sunlight gathered and the considered part of the equation to relate the form of amount of long-wave radiation released into the the building and the build-up of the building envelope atmosphere, both of which affect the urban heat island. (Jones 2008). This method includes elements such as At the building scale, it is an indication of the daylight wall construction, surface area, volume and glazing potentially available at each point in the building's percentage as well as the heat generated from building facade. In order to calculate this parameter, an upper usage and solar heat gain. Net heat transfer was used hemisphere of light rays is generated at each vertex of to approximate energy consumption in this project, the building's surface and the intersections of the rays estimating trends related to geometric form and with the surrounding environment are calculated. massing. 4. Case Study New York City was chosen as a test site because it has a relatively extreme climate—cold and windy in the winter, hot and balmy in the summer—and because it displays a differentiated wind regime from season to season, providing rich ground for climatic optimisation. In addition, the published documentations of New York City allowed for thorough research into existing local typologies and served as a basis for comparison to the resulting fabric. Fig.5. Heat Transfer Diagram The Fort Greene neighbourhood in Brooklyn was Because only a portion of the full energy calculation chosen as the urban tissue sample, because it has was employed, it was necessary to compare the medium-to-high density with both low- and high- 320 JAABE vol.13 no.2 May 2014 Soungmin Yu rise typologies and a mixed residential/commercial of typical sections comprising the street width and programme. The specific area chosen for this two height parameters at either side. The height and experiment was surrounded by different grid layouts width ranges were taken from measurements in the and could potentially bridge them in several ways. A existing urban fabric. The aforementioned ranges of 1 km area, although not empty, was taken as a site 15-35 m for the width of the streets and 10-50 m for to provide an experimental setting, and its existing the height of the buildings were used in the experiment tissue was used as a sample to determine programme, to generate the street sections. By grouping streets density, typical sections and block typology. Existing according to their angle relative to the north and connections with the neighbouring areas were assigning heights to each of these groups, it became considered starting points for the street layout. The possible to run a genetic simulation that distributes a results were always compared to this tissue in order set built volume onto an urban patch while optimising to accurately quantify improvements over the existing criteria such as facade solar exposure and wind flow. situation. It was thus possible to run a GA while maintaining In order to generate an optimised fabric using a key architectural attributes by placing constraints on GA, it was necessary to couple the analysis tools variables such as street width and facade height, which mentioned above with a genome set that can describe derived from the studied site. The benefit of using a GA urban geometry. This can be achieved by abstracting was clearly demonstrated by running a simulation with the geometric shapes that make up the fabric of a city multiple, conflicting evaluation criteria. It was shown by using the architectural concepts most relevant to the that in comparison to the existing fabric of Brooklyn, scale at hand. a 25% higher facade exposure to the sun in the winter months could be achieved, and average wind flow could be decreased in the winter by 35% and increased in the summer by 17%. Fig.8. Percentage of Daylight Hours Exposed to Direct Sunlight. Existing Fabric (left) and Urban Scale Result from Algorithm (right) Fig.7. Overall Algorithmic Process in Three Scales Considering the orthogonal street layout of Brooklyn, a street could be defined simply as a line between two source points. By parameterising the source points of the main streets of an urban patch, Fig.9. Urban Scale Algorithm Result a large variety of main street configurations could be described. The source points were considered the At the block scale, it was possible to quantify the genome for a GA. Further subdivisions were made geometry of a typical Brooklyn block in a similarly based on the configuration of the primary streets by abstract fashion. Deriving the border of the block and recursively subdividing the initial grid with a set ratio the heights of its facades from the urban-scale results, of potential block width and length. The proportion it was possible to describe different typologies by of the blocks was again determined based on the local dividing the block into segments and controlling the urban morphology, which showed a length-to-width - offset distance of the edges of each of the segments ratio of 1:2 and typical block sizes around 200 m . For from their respective borders. For example, a courtyard the primary and secondary street network, the actual typology could be described as an offset of all the street width was determined according to the centrality centrally facing edges towards the circumference of the of the street as defined by a space syntax analysis, with block. streets increasing in width as their centrality increased. The percentage of open space of the existing urban The geometry of the streets was described by a series block could be defined as a used input. The analysis JAABE vol.13 no.2 May 2014 Soungmin Yu 321 of the existing urban fabric of the chosen site showed the initial volume. For this experiment, each vertex an average of 30% open space within each block, describing the block geometry was able to move in x which was later used in the simulation. The number of and y directions freely, with a maximum range of 4 m. subdivisions of building plots was also measured from The glazing percentages of the building facades were the existing site, and an average of eight building plots used as another genome set in order to further optimise per block were assumed in the experiment. Similarly, the energy consumption of the collection of buildings based on the existing height range in the site, the range within the block. The glazing percentage ranged from of building heights used was 10-50 m. 20 to 100%. The facades were divided according to Seven fitness criteria were used to generate the different elevations of the massing, and each elevation block morphologies. At this scale, a more precise was further subdivided every 10 m, allowing the solar calculation (ray counting method) was used to algorithm to assign different glazing percentages for determine the winter and summer solar exposure and different building heights as well as for elevation. was combined with other environmental parameters including energy use, SVF and passive zone ratio as well as architectural parameters such as the connectivity and privacy of the outdoor spaces. The weighting the different fitness criteria became a critical factor in the genetic simulations, with different ratio of the weightage of the fitness criteria leading to varied morphological results. The differences in the weightage of the fitness criteria were influenced by the programme. The blocks were divided into two main programmes, residential and commercial. In order Fig.10. Existing Fabric (left) and Urban-Scale Result from to determine the correct ratio, existing blocks were Algorithm (right) measured with respect to the analysis criteria, and it was observed that different building functions exhibited what can be described as typical weighting patterns. For example, in the Brooklyn area, residential blocks show a strong emphasis on passive zone ratio and open space privacy, whereas commercial blocks have a lowered net heat transfer and an increased SVF. Thus, it was possible to calibrate the GA to achieve blocks with properties similar to a typical commercial, public or residential block from that area, while achieving better scores in most of the measured criteria. An improvement of up to 30% could usually be achieved in the three or four parameters given the most weight in the fitness function, without significantly decreasing Fig.11. Urban-Scale Algorithm Result the score of the other criteria. An important aspect of this type of simulation The objectives for optimisation at the building scale is that several adjacent blocks can be optimised varied according to the assessment of the result of the simultaneously to achieve a result that is both better in previous block scale. Therefore, the conditions used terms of environmental criteria and shows emergent as fitness criteria varied accordingly for each block. patterns of public networks between the blocks. In Although specific algorithms can be shown to improve this way, the fabric can be worked on in patches; the properties such as energy use of buildings, they can different patches can be stitched together by rerunning also be tailored to improve microclimatic conditions of the simulation on the patch borders, thus achieving the surrounding public spaces. For example, they can a well-knit fabric with different networks distributed be tailored to improve the quality of shared open space across the different scales. by maximising solar access for winter, minimising heat At the building scale, it was possible to rework the gain during summer, improving SVF and reducing the geometry of the resulting blocks, using another round reflected glare of a glass facade on the street level. of GA simulations to refine the buildings in terms Combined, these experiments form a holistic of plans, sections and elevations. In this round of approach to urban design, passing data sets from simulations, only blocks that exhibited low values in one scale to the other and insuring a continuous some of the parameters were chosen as starting points. improvement of the generated fabric. Because they The corner points of the building massing were can be performed on a single platform, they enable taken as the genome to modify the massing of the designers to make informed decisions at each of the block without deviating more than a set tolerance from different scales and to see the effect of the scales upon 322 JAABE vol.13 no.2 May 2014 Soungmin Yu evaluation criteria were selected in order to study their relationship with designed forms at the different scales. To further enrich the solution, other architectural criteria were introduced as part of the evaluation for the different scales. In some places, a designer's decision will surely be required when determining the priority of different criteria as well as in assessing the outcome of the algorithm and choosing between the different results. In order to be used for a real project, the evaluation Fig.12. Output of the Block-Scale (left) and Urban-Scale Results tools described in these experiments need to be refined from Algorithm (right) to improve their accuracy. The wind evaluation tool, for instance, has inherent limitations as it estimates the outcome by looking at the proportions of geometry in localised areas and predicting the combined global results. However, the accuracy of this estimation could be greatly improved by expanding the sample library with further case studies, including varying border conditions. In order to improve the accuracy of the net heat transfer tool, it would be feasible to add a library of different wall constructions and the heat generated from different usages of the building, as well as other criteria taken into account in a full energy calculation. In the architectural paradigm, there are several ways Fig.13. Building-Scale Algorithm Result to approach the different design issues tackled in this each other. The GA together with the framework of the study. The reflection and the albedo level addressed aforementioned experiment provides several solutions at the building scale could also be dealt with using at each scale level, allowing designers to intervene the materiality of the finishes instead of adjusting the and consider other design criteria to choose from form as suggested in the experiments. The generative alternatives which provide design options that consider methods for all the scales tackled here are based on climatic aspects. simple geometric rules such as subdivisions of lines, offsets and displacements of points and curves. These 5. Conclusion methods of generating geometries are sufficient for In this design study, the applicability of GAs to this particular set of experiments. However, these improving the climatic performance of multiscalar geometric rule sets are limited in what they can urban environments was tested and shown to have produce. For instance, these methods would not be potentially useful outcomes. As a design tool, the appropriate when applied in a context with curvilinear methods explored in this paper are thought to be most street layouts or staggered grids. Another feature appropriate as a part of the feasibility stage, where that could be implemented at the building scale is a the procedures described can be used to quickly test staggering of the facades into porches and terraces. different design conditions and ideas on a given site Similar simulations could be run in other places by varying density, programmes and materiality. in the world. In order to apply the design methods The result of algorithms can be used to assist the described in this paper in other locations, two main decision-making process, embedding into them the alterations will need to be made: expansion of the environmental consequences of the design. morphologies contained in the library for wind At the block and building scales, the method simulation and establishment of a geometric logic to developed shows potential for integrating various generate the forms. The tools that rely on equations climatic principles relating both to energy use or specific calculation methods, such as solar access, and pedestrian thermal comfort, which is rarely energy and SVF, could be used to reproduce the experimented with. Not only can individual blocks experiments in other contexts. However, further be optimised, but whole areas can be developed assessments are required to determine the geometric simultaneously, with the built masses in different logic that could be applied in different city contexts. blocks functioning together to achieve a result that The use of source points, offsets and vertexes as could not have been achieved individually. genomes could still be valid if the intended results of The evaluation criteria considered in this paper form the morphologies are noncurvilinear. The ranges of the only a portion of all the considerations, criteria and height and width of the streets and of building height requirements that need to be met by an architectural need to be determined by users prior to running the or urban project. Only a handful of the most relevant algorithm. These height and width ranges need to be JAABE vol.13 no.2 May 2014 Soungmin Yu 323 17) Ratti, C., Baker, N. and Steemers, K. (2006) Energy consumption reflected in the library of the wind simulation. It is also and urban texture. Energy and Buildings, 37(7), pp.762-776. required that users have a prior understanding of how 18) Richens, P. (1997) Image processing for urban scale environmental different weights for multiple fitness criteria can be modelling. Paper presented at the Fifth International IBPSA distributed to suit the climate and required function. Conference: Building Simulation, Prague. Given that these alterations and modifications are 19) Robinson, D. (2011) Computer modelling for sustainable urban design. London: Earthscan. considered, it will be possible to employ methodologies 20) Robinson, D. and Stone, A. (2004) Solar radiation modelling in described in this paper and these types of experiments the urban context. Solar Energy, 77(3), pp.295-309. in other locations with different climatic conditions. 21) Salat, S. (2009) Energy loads, CO2 emissions and building stocks: Although the issues raised here represent challenges morphologies, typologies, energy systems and behaviour. In Lorch, and many improvements can be made to the system, R. (ed.), Building research information (pp.598-609). Routledge. the results of this series of experiments are promising. As actual improvements were made at many scales and in many criteria, the generative-evaluative loop of the GA seems to be the appropriate framework for this discussion. As the computational horizon expands, more and more variables can be integrated into the equation, and detailed architectural solutions may be achieved. References 1) Adolphe, L. (2000) A simplified model of urban morphology: application to an analysis of the environmental performance of cities. Environment and Planning B: Planning and Design, 28, pp.183-200. 2) Besserud, K. and Ingram, J. (2010) "Architectural Genomics", Skidmore, Owings, & Merrill LLP (BlackBox Studio). 3) Caldas, L.G. and Caldas, L.K. (2003) Genetic algorithms for optimization of building envelopes and the design and control of HVAC systems. Journal of Solar Energy Engineering, p.125. 4) Fazia, A. (1995) Dependence of outdoor thermal comfort on street design in hot and dry climate. PhD thesis. Reports of the Meteorological Institute of the University of Freiburg, Vol. 15. 5) Gonçalves, L. and Gazzola H. (2001) A comprehensive approach of the sky view factor and building mass in an urban area of the city of Belo Horizonte, Brazil, University of Minas Gerais. 6) Givoni, B. (1998) Climatic considerations. In Building and urban design. New York: John Wiley & Sons, Inc. 7) Hegger, Fuchs, Stark, Zeumer (2008) Energy manual: sustainable architecture. Basel, Boston, Berlin: Detail. 8) Hensel, M., Menges, A. and Weinstock, M. (2010) Emergent technologies and design. New York: Routledge. 9) Jabareen, Y.R. (2006) Sustainable urban forms: their typologies, models and concepts. Journal of Planning Education and Research, Sept, pp.38-52. 10) Jones, P. (2008) Thermal environment. In Metric handbook planning and design data. Oxford: Elsevier. 11) Kakon, N. and Mishina, K. (2009) An evaluation of increasing building height in respect of thermal climate in a high density city in South Asia using numerical modeling. Journal of Asian Architecture and Building Engineering, 8(2). 12) Kim, S. (2013) Interval estimation of construction cost using case-based reasoning and genetic algorithms. Journal of Asian Architecture and Building Engineering, 11(2). 13) Nakamura, Y. and Oke T. (1988) Wind, temperature and stability conditions in an east-west oriented urban canyon. Atmos. Envir., 22, pp.2691-2700. 14) Mardalevic, J. (1995) Daylight simulation, validation, sky models, and daylight coefficients. PhD thesis. DeMontfort University, Leicester. 15) Mitchell, M. (1998) An introduction to genetic algorithms. MIT Press. 16) Ranjbar, E., Motalaei, N., Reza, M. and Akbar, A. (2011) Climatic design initiatives according to wind. Northern side Persian Gulf. Case study: Old Boushehr. In PLEA 2011: 27th Conference on Passive and Low Energy Architecture (pp.291-296). 324 JAABE vol.13 no.2 May 2014 Soungmin Yu
Journal
Journal of Asian Architecture and Building Engineering
– Taylor & Francis
Published: May 1, 2014
Keywords: urban fabric; climatic optimization; multi-objective optimization; multiscalar; genetic algorithm
