Sürdürülebilir Toplu Konut Yerleşmesi Tasarımı İçin Pareto Genetik Algoritmaya Dayalı Bir Model Önerisi : SSPM
Sürdürülebilir Toplu Konut Yerleşmesi Tasarımı İçin Pareto Genetik Algoritmaya Dayalı Bir Model Önerisi : SSPM
Dosyalar
Tarih
2016
Yazarlar
Aksoy, Yazgı
Süreli Yayın başlığı
Süreli Yayın ISSN
Cilt Başlığı
Yayınevi
Fen Bilimleri Enstitüsü
Institute of Science and Technology
Institute of Science and Technology
Özet
Sürdürülebilir yerleşme ve bina tasarımı geleneksel bina tasarımına göre daha karmaşık bir sürece sahiptir. Bunun nedeni birlikte alınması gereken fakat birbiriyle çelişen pek çok tasarım kararının erken tasarım evresinde alınması gerekliliğidir. Sürdürülebilir tasarımın erken evresinde tasarımcıya yardımcı bir karar destek sistemine ihtiyaç kaçınılmazdır; fakat tasarım endüstrisinde, bilgisayar tabanlı araçlar tasarım sürecinin önemli bir parçasını oluşturmalarına rağmen, erken tasarım evresi bilgisayar desteğinin en az kullanıldığı evre olmuştur. Bu çalışmada, pareto temelli bastırılmamış sınıflandırmalı genetik algoritma (NSGA-II) yöntemi kullanılarak pek çok tasarım kriterini göz önünde bulundurması gereken tasarımcıya karar verme sürecinde yardımcı olacak evrimsel bir model geliştirilmesi amaçlanmıştır. Sürdürülebilir Arazi Yerleşme Planlama Modeli (SSPM) olarak adlandırılan model, evrimsel algoritmaların sürdürülebilir arazi yerleşiminde kullanılmasıyla, erken tasarım evresinde toplu konutların araziye yerleşiminde mimarı yönlendirebilecek bir karar destek sistemi olarak önerilmiştir. Evrimsel algoritmalarda en iyi çözümler, evrimsel süreçlere maruz bırakıldıktan sonra, olası çözümler kümesinden elde edilir. Eğer eniyilenecek tek bir amaç varsa, problem "tek amaçlı eniyileme problemi"; birden fazla amaç olması durumunda ise "çok amaçlı eniyileme problemi" olarak tanımlanır. Çok amaçlı eniyileme problemlerinde, birbiriyle çelişen amaçlar nedeniyle problemin çözümleri birden fazladır. Bu çözümlere etkin çözümler denir. Pareto analizi değişik sayıdaki önemli etkenleri, daha az önemde olan etkenlerden ayırmak için kullanılan bir tekniktir. Pareto-optimal kavramının evrimsel algoritmalara uyarlanmasıyla, SSPM modelinde kullanılan pareto temelli çok amaçlı evrimsel eniyileme algoritmaları geliştirilmiştir. Buna göre, popülâsyondaki en iyi çözüm, amaçların herhangi biri içinde en kötü olmayan ve en azından bir amaç içinde diğerlerinden daha iyi olan çözüm anlamındadır. Tasarımın ilk aşamasında eldeki tasarım problemine uygun tüm sınırlandırmaları göz önünde bulundurarak toplu konut yerleşim önerileri sunan evrimsel modelin tanımlanması ve modelin alan çalışması yapılarak değerlendirilmesi bu çalışmanın konusunu oluşturacaktır. Sürdürülebilir tasarım sürecinde, sezgisel ve uygulama tecrübesine dayanan geleneksel tasarım yöntemi yerine, hesaplamalı yöntemlerle problem alanına daha geniş açıdan bakılarak tatminkâr çözüm alternatifleri elde edilebilir. Tasarımın ilk evresinde sayısal ortamda planlanmış ve tasarlanmış bir çevre, tasarımcının yaratıcılığının doğru yönde gelişmesini sağlar. Bu sayede üretilen oldukça fazla sayıda alternatif objektif olarak değerlendirilebilir. Tez kapsamında, mimari tasarımda öncelikle çok amaçlı eniyilme modelleriyle sürdürülebilir tasarım konuları incelenmiştir. Tezin 4. bölümünde bu anlamda geliştirilen iki örnek model tanıtılmıştır. İlk olarak Weimin Wang, Hugues Rivard ve Radu Zimeureanu tarafından geliştirilen, yeşil bina tasarımı için planlar üreten eniyileme modeli ele alınmıştır. Bu çalışmada bina biçimlerinin genetik algoritma kullanılarak eniyilenmesi amaçlanmıştır. Çok amaçlı genetik algoritma, üretilen plan alternatiflerini duvar malzemeleri ile birlikte değerlendirerek, maksimum doğal ışık ve minimum enerji tüketimine göre eleyerek, pareto optimal alternatifi kullanıcıya sunmaktadır. Bina biçimi yanında malzeme ve maliyeti göz önünde bulundurması açısından işlevsel bir model olarak tanımlanabilir, fakat erken tasarım evresinde birden fazla binadan oluşan kompleks sistemlerin araziye yerleşiminde yetersiz kalmaktadır. Zielinska, Church ve Jankowski tarafından geliştirilen SMOLA (Sustainable Multi- Objective Land Use Allocation Model) ise çok amaçlı mekânsal eniyilemeyi hedeflemektedir. SMOLA çelişen iki kriter olan maksimum açık alan ve maksimum kalkınma, komşu yerleşimlere uygunluk ve var olan yerleşim alanlarına yakınlık şartlarını göz önünde bulunduran kentsel ölçekli çok amaçlı mekânsal eniyileme modelidir. Model kentsel yerleşim problemleri için kısıtlayıcı olarak düşünülebilir, fakat mekânsal eniyileme modellerinin gelişen şehirlerin kentsel formunu yönlendirmede büyük rolü olduğu düşünülmektedir. Bu tez kapsamında önerilen sürdürülebilir bina modeli sadece enerji etkin bina modeli olarak değil, eskiz aşamasından başlayarak bina ölçeğinde sürdürülebilir arazi kullanımına önem veren bütünleşik bir modeldir. Bu konuda yapılmış diğer modellerden farklı olarak, Amerika kökenli LEED (Leadership in Energy and Environmental Design) ve İngiltere kökenli BREEAM (Building Research Establisment Environmental Assessment Method) arazi yerleşim kriterlerinin yanında yerel yönetmelikleri ve yerel iklim şartlarını da göz önünde bulunduran modelin, üretilen alternatifleri uygunluk puanlarına göre birbirleriyle kıyaslayarak pareto optimal sonuçlar üretmesi, tasarımcıyı sürdürülebilir arazi kullanım çözümlerine götürmesi hedeflenmiştir. Toplu konut yerleşmesinin tasarlanacağı arazi, matrisle tanım tekniği kullanılarak temsil edilmiş ve matrisler Excel'de oluşturulmuştur. Arazinin tanımlanmasında Excel programı, veri girişinin pratik ve kolay tanımlanabilir olması ve matrisle tanım tekniğine olanak sağlaması nedeniyle tercih edilmiştir. Matrisin satırlarını ve sütunlarını oluşturan her bir elemanı (hücresi), arazinin alanı 1 m² olan bir birimini temsil etmektedir. Arazi köşe noktalarının ve arazi üzerindeki mevcut elemanların x,y koordinatlarına göre matris oluşturulmaktadır. Arazinin topografik yapısını ve doğal oluşumları göz ardı etmeden, çözümün gerçekçi sonuçlar verebilmesi için, arazi verilerinin de gerçeğe yakın modellenmesi gerekmektedir. Bunu sağlamak için sayısal araziye eldeki eğim değerlerine göre üç boyutlu karakter kazandırılmıştır. Yükselti farklarının çözümünde, araziden gelen eğim değerleri kullanıcı isteğine göre farklı kotlardaki parçalara bölünerek modele tanıtılmaktadır. Bu sayede yüksek kot farkına sahip kademeli arazilerde teraslama yapılarak, topografya ile uyumlu arazi yerleşim çözümlerine gidilebilecektir. Sürdürülebilir bina tasarımını amaçlayarak geliştirilen pek çok evrimsel modelden farklı olarak, arazideki kot farklarını ve üretilen bina yüksekliklerini 3. boyutta kullanıcıya gösteren SSPM modeli, gerçek arazi verilerini sayısal olarak kullanarak, arazideki korunacak ağaçları, su öğelerini ve yeşil alanları üretimlerine dâhil etmektedir. Böylece güncel veri grubu üzerinden üretimlerini yaparak gerçekçi sonuçlar vermektedir. Modelin sınanması için Kağıthane'de yer alan 4500 m2' büyüklüğünde bir arazi seçilmiştir. Hipotetik bir arazi yerine, daha önceden bir bölümünün kullanılmış olması, sınırları içerisinde korunacak ağaç ve su öğelerini içermesi sebebiyle bu arazi tercih edilmiştir. Arazi İstanbul ilinde olması sebebiyle uygunluk fonksiyonlarının bir kısmını İstanbul İli İmar Yönetmeliği kuralları oluşturmuştur. Araziye ait TAKS, KAKS, hmax, manzara, rüzgar ve iklim verilerine göre üretilmiş ilk popülasyon, sonrasında NSGA-II algoritması döngüsü içerisine sokularak pareto optimal arazi yerleşim bireyleri elde edilmiştir. Her nesilde üretilen birey sayısı 100 ile sınırlandırılmıştır, kullanılan bilgisayarın kapasitesi nedeniyle daha fazla birey üretimi modelin çalışma hızını oldukça düşürmektedir. Yerleşim, manzara ve rüzgâr açıları kuzey-güney-doğu ve batı olmak üzere doğrusaldır, açılı yerleşimler modelde üretim dışında tutulmuştur. Arazi, bünyesindeki kot farklarına göre kullanıcı tarafından teraslanarak modele tanıtılmıştır. Hem yerleşilen arazinin eğim verilerini okuyabilen, hem de yerleştirilen blokların yüksekliğini göz önünde bulundurarak yaklaşma mesafelerini hesaplayan model konut bloklarının yerleşimi için alternatif sonuçlar üretmiştir. Çalışmada farklı nesillerde, oluşan pareto eğrileri kıyaslanarak modelin üretimleri değerlendirilmiştir. İlk popülasyon, sonrasında NSGA-II algoritması döngüsü içerisine sokularak pareto optimal arazi yerleşim bireyleri elde edilmiştir. Uygunluk puanları verilen bireylerin toplam skorları 0 ile 1 arasında normalize edilmiştir; 1 en başarılı bireylerdir, 0 en başarısız bireylerdir; sonrasında turnuva seçilimi uygulanır. Turnuva seçiliminde skoru yüksek olan bireylerin üreme şansı daha fazladır. Bireyler kendi skorlarına göre oluşturulan matristen seçilirler. İlk üretilen bireylere bakıldığında arazideki korunması gerekli ağaçlar dikkate alınarak blok yerleşimleri yapılmış, Kağıthane Deresi ve 30'ar metre çevresi de koruma altına alınarak bloklar bu bölgeden uzağa yerleştirilmiştir. Bloklar manzaraya yönelmiştir. Birbirinin manzarasını kesmeyen fazla puanlı alternatiflerin üretim şansı da fazladır. Elde edilen ilk neslin pareto eğrisi eniyileme tamamlanmadığı için doğrusal olarak elde edilmiştir. Nesil ilerledikçe pareto eğrisi sıcak iklime sahip olan Kağıthane arazisi için hem manzara hem de rüzgar fonksiyonlarını maksimize etmek için konkavlaşmıştır. İleriki nesillerin pareto grafiğinde yeşille gösterilen pareto optimal bireyler hem manzara hem de rüzgarı maksimize etmiş ve üretilmiş diğer hiç bir birey tarafından bastırılmamış en yüksek puanlı bireyler olarak karşımıza çıkmıştır. Elde edilen pareto optimal arazi yerleşim bireyleri, kütle yerleştirme problemlerinde genetik algoritmanın iyi ve tutarlı sonuçlar verdiğini göstermektedir. Özellikle yoğun yapılaşmaya izin veren araziler üzerinde birbiriyle çakışan arazi yerleşim kriterlerini göz önünde bulundurarak blok yerleşimlerini yapmak geleneksel yöntemle oldukça zaman almaktadır. Modelin ürettiği pareto optimal arazi bireylerinin çok amaçlı genetik algoritmanın yardımıyla birkaç dakika içerisinde elde edildiği düşünülürse, bu modelin erken tasarım evresindeki önemli bir kullanım potansiyeline sahip olduğu söylenebilir. Sayısal ortamda planlanmış ve tasarlanmış bir çevre, tasarımcının yaratıcılığını doğru yönde ilerletmesine olanak sağlayarak oldukça fazla sayıda alternatifi değerlendirmede yardımcı olacaktır. İki farklı disiplin olan evrimsel algoritmalar ve sürdürülebilir tasarım ara kesitinde yer alan bu çalışma ile gerçekleştirilecek sürdürülebilir yerleşme tasarımı modelinde amaç, uzman bir tasarımcının yerini almak değil; tasarımcıya tasarım kararlarının alınması sürecinde yardımcı olmasını sağlamaktır. Sürdürülebilir tasarım sürecinde, sezgisel ve uygulama tecrübesine dayanan geleneksel tasarım yöntem ve teknikleri yanında; hesaplamalı yöntemlerle problem alanına daha geniş açıdan bakarak çözüm alternatifleri üretmek, tatminkâr çözümlerin elde edilmesini sağlayacaktır.
Nowadays as the aim to reduce the environmental impact of buildings becomes more apparent, a new architectural design approach is gaining momentum called sustainable architectural design. Sustainable architectural design process includes some regulations itself, which requires calculations, comparisons and consists of several possible conflicting objectives that need to be considered together. In design industry, advanced computer aided design tools have an important impact on design process, but still early design stage and sustainable design are problematic issues, and need to be solved. Sustainable building design refers to a process that begins with selecting the site and optimizing economic and environmental performance throughout a building's life cycle. In order to achieve a successful sustainable building, particular attention needs to be paid to the conceptual design stage when the most important decisions are taken; nevertheless it is the stage with least computer support. In architectural design process, generally optimization algorithms have been used to automate the generation of design layouts. Hence there is a need for design tools that can help designers better manage collaborative design development. It is becoming difficult to improve the performance of building design based only on improvements in individual disciplines. For this reason, better, system-orientated, holistic, multidisciplinary approaches to building design are needed. A successful green building design can be performed by the creation of alternative designs generated according to all the sustainability parameters and local regulations. Green building rating systems and certification programs are accepted as one of sustainability parameters. In this thesis, LEED (Leadership in Energy and Environmental Design) and BREEAM (Building Research Establisment Environmental Assessment Method) certification systems are going to be considered, as being the most representative building environment assessment schemes that are in use. Although these certification systems are used all over the world, the parameters are prepared according to America's and Britain's geographical, economic and cultural conditions, though other countries are experiencing difficulties during sustainable design process. As a result of this, green buildings should be designed also according to the climate of the region and local building construction regulations. To summarize, today some tools developed to assist designers during sustainable building design process, address more detailed design stages when important design decisions already have been taken in conceptual stage. As there are conflicting criteria's according green building rating systems sustainable site parameters, local regulations and local climate conditions, an efficient decision support system can be developed by the help of Pareto based non-dominated genetic algorithm (NSGA-II) which is used for several possibly conflicting objectives that need to be considered together. Genetic algorithm is a population-based search technique inspired from the biological principles of natural selection and genetic recombination. Genetic algorithm is a suitable method for multi-objective optimization problems because it can generate multiple Pareto optimal solutions in a single simulation run. NSGA is a very effective algorithm but it has computational complexity, lack of elitism. NSGAII was developed as a modification, which has a better sorting algorithm, incorporates elitism and no sharing parameter needs to be chosen a priori. The NSGA-II procedure has three features to find multiple Pareto-optimal solutions in a multi-objective optimization problem: It uses an elitist principle and an explicit diversity preserving mechanism, and also it emphasizes non-dominated solutions NSGA-II-based optimization process is used to develop the Sustainable Site Planning Model (SSPM). There are simultaneous optimization of several possibly conflicting objectives in multi-objective optimization problems that result in a set of non-dominated solutions. There does not exist a single solution that simultaneously optimizes each objective. In that case, there exist an infinite number of non-dominated solutions which are also known as Pareto optimal solutions. In sustainable design, maximum energy conservation and utilization of natural light can be given as an example of two conflicting objectives, so during sustainable design process Pareto genetic algorithm will be successful to generate design alternatives according to conflicting criteria. Genetic Algorithms (GAs) are adaptive heuristic search algorithm based on the evolutionary ideas of natural selection and genetics. As such they represent an intelligent exploitation of a random search used to solve optimization problems. GAs simulate the survival of the fittest among individuals over consecutive generation for solving a problem. Each individual represents a point in a search space and a possible solution. The individuals in the population are then made to go through a process of evolution. Although randomised, GAs are by no means random, instead they exploit historical information to direct the search into the region of better performance within the search space. The presence of multiple objectives in a problem, inprinciple, gives rise to a set of optimal solutions known as Pareto-optimal solutions, instead of a single optimal solution. In the absence of any further information, one of these Pareto-optimal solutions cannot be said to be better than the other. This demands a user to find as many Pareto-optimal solutions as possible. Classical optimization methods suggest converting the multiobjective optimization problem to a single-objective optimization problem by emphasizing one particular Pareto-optimal solution at a time. When such a method is to be used for finding multiple solutions, it has to be applied many times, hopefully finding a different solution at each simulation run. Over the past decade, a number of multiobjective evolutionary algorithm. The primary reason for this is their ability to find multiple Pareto-optimal solutions in one single simulation run. Since evolutionary algorithms (EAs) work with a population of solutions, a simple EA can be extended to maintain a diverse set of solutions. With an emphasis for moving toward the true Pareto-optimal region, an EA can be used to find multiple Pareto-optimal solutions in one single simulation run. The nondominated sorting genetic algorithm (NSGA) is one of the first such EAs. It is a popular non-domination based genetic algorithm for multi-objective optimization. Especially in green building design process many efforts have been made to integrate multi objective optimization models and sustainable design. Wang, Zmeureanu and Rivard optimized the building envelope using multi-objective genetic algorithm. They concentrate on building envelope because of its importance in environmental and economic performance of buildings. As a case study the design of a single-story office building located in Montreal, Canada is determined, only heating and cooling energy consumption are considered. The multi-objective optimization model they used for building envelope design, can be used to locates optimum or near optimum green building designs for given conditions (Wang, Zmeureanu and Rivard, 2005). In sustainable design land use is another subject that designers concentrate on. Zelinska, Church and Jankowski (2008) present a new multi objective spatial optimization model (SMOLA) which minimizes the conflicting objectives of open space development, infill and redevelopment, land use neighborhood compatibility and cost distance to already urbanized areas. These developed models are chosen for using evolutionary algorithms to solve different scaled design problems. The plan optimization model for green building design developed by Wang, Rivard and Zimeureanu is functional for considering material and cost information with building form, but it is insufficient for early stages of design. Also it is mostly concentrated on energy conservation; site planning for building complexes is disregarded. SMOLA model developed by Zelinska, Church and Jankowski generates land use patterns according to the building functions. For city scale the model is useful and has visual readability, but it is not convenient for small-scale problems. The SSPM model presented in this thesis is an integrated model that concentrates on wide sustainable design criteria; such as green building certification systems and local building codes for. The model presented in this thesis will generate site-planning alternatives for social housing according to sustainable design objectives. Multi-objective genetic algorithm is going to be used as an computational approach to generate design alternatives. The Sustainable Site Planning Model (SSPM) is written in Processing which is a Java based programming language. The first step of the model is to define of the site to the computer. Matrix definition technique in Excel is used to define the site with numbers. The defined digital site is used as a base to generate site planning alternatives by SSPM. Excel program is chosen for providing quick and easy data input for matrix definition. The rows and columns of the matrix represent 1 unit =1 m2 of the site. The x, y values of the site corner points and the center or corner points of existing elements are used to define the matrix. Each of the Excel cell is considered as a point in a coordinate system. R1C1 reference style in Excel is used to represent x and y axes to make data input apparent. The rectangular area which uses the values of maximum x and y coordinates of site corner points as its dimensions, is the boundary of the matrix. The cells inside the defined boundary are coloured according to their functions so that visual presentation is provided. In addition, user chooses climate type, the direction of sun, wind and view. SSPM generates site-planning alternatives on defined cells. According to fitness functions and crossover and mutation operators are applied to the population to find the pareto optimal site planning solution. The selected site must be digitalized closed to the real terrain data without disregarding the topography and the natural formations, so that the site is presented in 3D grid with its slope data. User will divide the site into different zones according to its slope data. As a result of this, terracing high-leveled sites will be possible and environmentally sensitive solutions will be able to generate. The SSPM model will generated social housing cells according to this terraced site-zones. The total fitness scores of the site individuals are normalized between 0 and 1; 1 is for most successful individuals, 0 is for unsuccessful individuals. At this stage tournament selection is done. Tournament selection works by selecting a number of individuals from the population at random, a tournament, and then selecting only the best of those individuals. The "tournament" isn't much of a tournament at all, it just involves generating a random value between zero and one and comparing it to a pre-determined selection probability. If the random value is less than or equal to the selection probability, the fitter candidate is selected, otherwise the weaker candidate is chosen. Chosen individuals are copied in direct proportion with their fitness scores, then they divided into two groups as parents. Next, crossover operator is applied between parents. Crossover selects genes from parent chromosomes and creates a new offspring. The SSPM model was tested on a site in Kağıthane in İstanbul. The reason for choosing Kağıthane was twofold. First, urban regeneration in the residential areas of Kağıthane has recently been a central issue of consideration in Turkey. Second, the selected study area is a realistic example that has reserved water supplies, green areas and polluted areas to test sustainable housing units. The focus of the case study has been the adaptation of the building blocks to local conditions, as well as the sustainable site usage parameters. The site layout alternatives that were generated by the SSPM model show a good adaptation to the site specific constraints and parameters, and good variation, within the limits that are imposed by the view, wind, floor area ratio (FAR) and floor space ratio (FSR) values. When we examine the first produced individuals, existing trees are protected and the building units are placed 30 meters far from Kağıthane River. Each housing units are directed to the view. The pareto curve of the first produced inividuals is linear because the optimization of site plan is not finished. As the generation progresses, the pareto curve for Kağıthane where has hot climate, becomes concave to maximize view and wind functions. For the next generations green individuals are pareto optimal individuals which have maximum view and wind scores and are not dominated by any other individuals. Pareto analysis and genetic algorithms are two built-in evaluation techniques guiding the creative output of optioneering tools. It can be argued that these evaluation techniques increase the creativity capacity of designer generating a large number of informed guesses at desirable designs. Pareto analysis focuses the set of design alternatives to the most optimal solutions, while stochastic methods like genetic algorithms introduce randomness to expand the space. As the environmental impact of buildings becomes more apparent, new architectural design approaches are arising. In this thesis, it is proposed to combine cellular structures with a multi-objective genetic algorithm for using its search ability to find Pareto-optimal sustainable site planning solutions for social housing complexes. This approach would introduce an effective computational design tool for early design stage of sustainable design, which does not currently achieved by current technologies.
Nowadays as the aim to reduce the environmental impact of buildings becomes more apparent, a new architectural design approach is gaining momentum called sustainable architectural design. Sustainable architectural design process includes some regulations itself, which requires calculations, comparisons and consists of several possible conflicting objectives that need to be considered together. In design industry, advanced computer aided design tools have an important impact on design process, but still early design stage and sustainable design are problematic issues, and need to be solved. Sustainable building design refers to a process that begins with selecting the site and optimizing economic and environmental performance throughout a building's life cycle. In order to achieve a successful sustainable building, particular attention needs to be paid to the conceptual design stage when the most important decisions are taken; nevertheless it is the stage with least computer support. In architectural design process, generally optimization algorithms have been used to automate the generation of design layouts. Hence there is a need for design tools that can help designers better manage collaborative design development. It is becoming difficult to improve the performance of building design based only on improvements in individual disciplines. For this reason, better, system-orientated, holistic, multidisciplinary approaches to building design are needed. A successful green building design can be performed by the creation of alternative designs generated according to all the sustainability parameters and local regulations. Green building rating systems and certification programs are accepted as one of sustainability parameters. In this thesis, LEED (Leadership in Energy and Environmental Design) and BREEAM (Building Research Establisment Environmental Assessment Method) certification systems are going to be considered, as being the most representative building environment assessment schemes that are in use. Although these certification systems are used all over the world, the parameters are prepared according to America's and Britain's geographical, economic and cultural conditions, though other countries are experiencing difficulties during sustainable design process. As a result of this, green buildings should be designed also according to the climate of the region and local building construction regulations. To summarize, today some tools developed to assist designers during sustainable building design process, address more detailed design stages when important design decisions already have been taken in conceptual stage. As there are conflicting criteria's according green building rating systems sustainable site parameters, local regulations and local climate conditions, an efficient decision support system can be developed by the help of Pareto based non-dominated genetic algorithm (NSGA-II) which is used for several possibly conflicting objectives that need to be considered together. Genetic algorithm is a population-based search technique inspired from the biological principles of natural selection and genetic recombination. Genetic algorithm is a suitable method for multi-objective optimization problems because it can generate multiple Pareto optimal solutions in a single simulation run. NSGA is a very effective algorithm but it has computational complexity, lack of elitism. NSGAII was developed as a modification, which has a better sorting algorithm, incorporates elitism and no sharing parameter needs to be chosen a priori. The NSGA-II procedure has three features to find multiple Pareto-optimal solutions in a multi-objective optimization problem: It uses an elitist principle and an explicit diversity preserving mechanism, and also it emphasizes non-dominated solutions NSGA-II-based optimization process is used to develop the Sustainable Site Planning Model (SSPM). There are simultaneous optimization of several possibly conflicting objectives in multi-objective optimization problems that result in a set of non-dominated solutions. There does not exist a single solution that simultaneously optimizes each objective. In that case, there exist an infinite number of non-dominated solutions which are also known as Pareto optimal solutions. In sustainable design, maximum energy conservation and utilization of natural light can be given as an example of two conflicting objectives, so during sustainable design process Pareto genetic algorithm will be successful to generate design alternatives according to conflicting criteria. Genetic Algorithms (GAs) are adaptive heuristic search algorithm based on the evolutionary ideas of natural selection and genetics. As such they represent an intelligent exploitation of a random search used to solve optimization problems. GAs simulate the survival of the fittest among individuals over consecutive generation for solving a problem. Each individual represents a point in a search space and a possible solution. The individuals in the population are then made to go through a process of evolution. Although randomised, GAs are by no means random, instead they exploit historical information to direct the search into the region of better performance within the search space. The presence of multiple objectives in a problem, inprinciple, gives rise to a set of optimal solutions known as Pareto-optimal solutions, instead of a single optimal solution. In the absence of any further information, one of these Pareto-optimal solutions cannot be said to be better than the other. This demands a user to find as many Pareto-optimal solutions as possible. Classical optimization methods suggest converting the multiobjective optimization problem to a single-objective optimization problem by emphasizing one particular Pareto-optimal solution at a time. When such a method is to be used for finding multiple solutions, it has to be applied many times, hopefully finding a different solution at each simulation run. Over the past decade, a number of multiobjective evolutionary algorithm. The primary reason for this is their ability to find multiple Pareto-optimal solutions in one single simulation run. Since evolutionary algorithms (EAs) work with a population of solutions, a simple EA can be extended to maintain a diverse set of solutions. With an emphasis for moving toward the true Pareto-optimal region, an EA can be used to find multiple Pareto-optimal solutions in one single simulation run. The nondominated sorting genetic algorithm (NSGA) is one of the first such EAs. It is a popular non-domination based genetic algorithm for multi-objective optimization. Especially in green building design process many efforts have been made to integrate multi objective optimization models and sustainable design. Wang, Zmeureanu and Rivard optimized the building envelope using multi-objective genetic algorithm. They concentrate on building envelope because of its importance in environmental and economic performance of buildings. As a case study the design of a single-story office building located in Montreal, Canada is determined, only heating and cooling energy consumption are considered. The multi-objective optimization model they used for building envelope design, can be used to locates optimum or near optimum green building designs for given conditions (Wang, Zmeureanu and Rivard, 2005). In sustainable design land use is another subject that designers concentrate on. Zelinska, Church and Jankowski (2008) present a new multi objective spatial optimization model (SMOLA) which minimizes the conflicting objectives of open space development, infill and redevelopment, land use neighborhood compatibility and cost distance to already urbanized areas. These developed models are chosen for using evolutionary algorithms to solve different scaled design problems. The plan optimization model for green building design developed by Wang, Rivard and Zimeureanu is functional for considering material and cost information with building form, but it is insufficient for early stages of design. Also it is mostly concentrated on energy conservation; site planning for building complexes is disregarded. SMOLA model developed by Zelinska, Church and Jankowski generates land use patterns according to the building functions. For city scale the model is useful and has visual readability, but it is not convenient for small-scale problems. The SSPM model presented in this thesis is an integrated model that concentrates on wide sustainable design criteria; such as green building certification systems and local building codes for. The model presented in this thesis will generate site-planning alternatives for social housing according to sustainable design objectives. Multi-objective genetic algorithm is going to be used as an computational approach to generate design alternatives. The Sustainable Site Planning Model (SSPM) is written in Processing which is a Java based programming language. The first step of the model is to define of the site to the computer. Matrix definition technique in Excel is used to define the site with numbers. The defined digital site is used as a base to generate site planning alternatives by SSPM. Excel program is chosen for providing quick and easy data input for matrix definition. The rows and columns of the matrix represent 1 unit =1 m2 of the site. The x, y values of the site corner points and the center or corner points of existing elements are used to define the matrix. Each of the Excel cell is considered as a point in a coordinate system. R1C1 reference style in Excel is used to represent x and y axes to make data input apparent. The rectangular area which uses the values of maximum x and y coordinates of site corner points as its dimensions, is the boundary of the matrix. The cells inside the defined boundary are coloured according to their functions so that visual presentation is provided. In addition, user chooses climate type, the direction of sun, wind and view. SSPM generates site-planning alternatives on defined cells. According to fitness functions and crossover and mutation operators are applied to the population to find the pareto optimal site planning solution. The selected site must be digitalized closed to the real terrain data without disregarding the topography and the natural formations, so that the site is presented in 3D grid with its slope data. User will divide the site into different zones according to its slope data. As a result of this, terracing high-leveled sites will be possible and environmentally sensitive solutions will be able to generate. The SSPM model will generated social housing cells according to this terraced site-zones. The total fitness scores of the site individuals are normalized between 0 and 1; 1 is for most successful individuals, 0 is for unsuccessful individuals. At this stage tournament selection is done. Tournament selection works by selecting a number of individuals from the population at random, a tournament, and then selecting only the best of those individuals. The "tournament" isn't much of a tournament at all, it just involves generating a random value between zero and one and comparing it to a pre-determined selection probability. If the random value is less than or equal to the selection probability, the fitter candidate is selected, otherwise the weaker candidate is chosen. Chosen individuals are copied in direct proportion with their fitness scores, then they divided into two groups as parents. Next, crossover operator is applied between parents. Crossover selects genes from parent chromosomes and creates a new offspring. The SSPM model was tested on a site in Kağıthane in İstanbul. The reason for choosing Kağıthane was twofold. First, urban regeneration in the residential areas of Kağıthane has recently been a central issue of consideration in Turkey. Second, the selected study area is a realistic example that has reserved water supplies, green areas and polluted areas to test sustainable housing units. The focus of the case study has been the adaptation of the building blocks to local conditions, as well as the sustainable site usage parameters. The site layout alternatives that were generated by the SSPM model show a good adaptation to the site specific constraints and parameters, and good variation, within the limits that are imposed by the view, wind, floor area ratio (FAR) and floor space ratio (FSR) values. When we examine the first produced individuals, existing trees are protected and the building units are placed 30 meters far from Kağıthane River. Each housing units are directed to the view. The pareto curve of the first produced inividuals is linear because the optimization of site plan is not finished. As the generation progresses, the pareto curve for Kağıthane where has hot climate, becomes concave to maximize view and wind functions. For the next generations green individuals are pareto optimal individuals which have maximum view and wind scores and are not dominated by any other individuals. Pareto analysis and genetic algorithms are two built-in evaluation techniques guiding the creative output of optioneering tools. It can be argued that these evaluation techniques increase the creativity capacity of designer generating a large number of informed guesses at desirable designs. Pareto analysis focuses the set of design alternatives to the most optimal solutions, while stochastic methods like genetic algorithms introduce randomness to expand the space. As the environmental impact of buildings becomes more apparent, new architectural design approaches are arising. In this thesis, it is proposed to combine cellular structures with a multi-objective genetic algorithm for using its search ability to find Pareto-optimal sustainable site planning solutions for social housing complexes. This approach would introduce an effective computational design tool for early design stage of sustainable design, which does not currently achieved by current technologies.
Açıklama
Tez (Doktora) -- İstanbul Teknik Üniversitesi, Fen Bilimleri Enstitüsü, 2016
Thesis (Ph.D.) -- İstanbul Technical University, Institute of Science and Technology, 2016
Thesis (Ph.D.) -- İstanbul Technical University, Institute of Science and Technology, 2016
Anahtar kelimeler
mimarlık,
genetik algoritmalar,
konut,
tasarım ve yapım,
architecture,
genetic algorithms,
housing,
design and construction,
mass customization