19. yüzyıl İstanbul dizi konutlarının morfolojik andizine dayalı bilgi tabanlı tasarım modeli

Abstract

ÖZET Son yıllarda, bilgisayar destekli mimari tasarım alanında yapılan kuramsal çalışmalar, mimann sahip olması gereken uzmanlık bilgilerini içeren bir bilgi dağarcığından yararlanarak, tasanm sürecinin benzetimini yapan Bilgi- Tabanlı Sistemler ve Uzman Sistemlerin gelişimine imkan vermiştir. Bu tez kapsamında, mimari ürün olarak ortak bir dile sahip, 19. Yüzyıl istanbul Dizi Konutları'nın morfolojik analizine dayalı bir bilgi tabanlı tasarım modeli geliştirilmiştir. Öncelikle gün geçtikçe yokolan bu tarihi ve mimari öneme sahip bina tipolojisine ait kapsamlı bir veri tabanı oluşturulmuş ve bu veri tabanında yer alan mimari ürünün topolojisine, morfolojisine, geometrisine ve boyutlarına ilişkin bilgiler, dizi konut tasarımı yapacak uzman sistemin bilgi dağarcığını oluşturmak üzere kullanılmıştır. ilk bölümde, tezin amacı, kapsamı, bu kapsamda geliştirilen bilgi-tabanlı sistemlerle, uzman sistemlerin esasları, bu sistemlere baz oluşturan 19. Yüzyıl İstanbul Dizi Konutlarının ele alınmasının nedenleri ve morfolojik analizle elde edilen bilgilere dayanarak, bu tür konut tasarımı yapan uzman sistem algoritmasında kullanılacak yaklaşım ortaya konulmuştur. İkinci bölümde ise bilgi-tabanlı mimari tasarım modelleri ve bu modellerdeki farklı yaklaşımlar, literatürden örneklerle anlatılmıştır. Üçüncü bölümde morfolojik analiz yöntemleri ortaya konulmuş ve morfolojik analiz uygulamalarına literatürden örnekler verilmiştir. Dördüncü bölüm, 19. Yüzyıl Dizi Konutlarının ortaya çıkışları, özellikleri ve bu konutlara uygulanan analizlerle, analiz sonuçlarına yer vermektedir. Bu analiz sonuçlarından elde edilen bilgi, belli bir sistematik kurgu içerisinde, bir hipermedya ortamında kullanıcıya sunulmaktadır. Beşinci bölüm, morfolojik analizle elde edilen bilginin, tasarım ürününü oluşturmak üzere sözdizimsel kurallara dönüşümünü ve bunlara bağlı olarak geliştirilen, dizi konut tasanmı yapan bilgi-tabanlı tasarım modelini içermektedir. Altıncı bölümde ise geliştirilen modelin sunduğu katkılarla, uygulama olanakları sunulmakta ve tezin sonuçlan yorumlanmaktadır. XVİ

SUMMARY A KNOWLEDGE-BASED DESIGN MODEL BASED ON MORPHOLOGICAL ANALYSIS OF 19th CENTURY İSTANBUL ROW-HOUSES In the last few years the theoretical studies in the field of computer-aided architectural design, gave rise to Knowledge-Based Design Systems and Expert Systems which make simulation of design process by using a knowledge-base comprising the expertial knowledge of an architect. The aim of this thesis was firstly to analyze and interpret a chosen design product of common architectural language from different aspects and to restore the knowledge obtained, by means of graphical, numerical, verbal formats and various rules to form a knowledge-base; secondly to use this knowledge in the production of design and education process. Thus it was thought it would be useful for the 19th Century Row Houses of a typological category and of historical and architectural value to be analyzed in the context of this project in terms of preservation and restoration. Design process includes interpretation, design production and learning stages. This thesis proposed here, includes the 'interpretation process' and an expert system algorithm for design, based on the knowledge obtained from this interpretation. In the context of this thesis, a knowledge-based design model based on the morphological analysis of the 19th century Row-Houses has been developed. 19th Century Row-Houses were analyzed with all aspects, and a knowledge-based model of that typology has been developed by using the knowledge obtained from the analysis. Morphological, topological and dimensional analysis of the row-houses in Istanbul were made, and the syntactical rules has been formed to produce a design by using the vocabulary elements which belong to this house type. Thus a knowledge base, which includes all the information about these degrading houses which are of historical and architectural importance has been established. With the help of the rules obtained, the syntactical structure of the row-houses has been defined. Finally an algorithm of an expert system which will produce the row-house designs by using the knowledge base has been developed. In the first chapter the aim and the scope of the thesis were introduced in the context of the principles of knowledge-based design systems and expert xvlisystems. Also the reasons for choosing Istanbul Row-Houses are suggested, stating the importance of a knowledge base which includes all the information about these degrading houses of historical and architectural value. Row-houses in Istanbul, all of which were built in the second half of the 19th Century under Western influence, display a typological category of Western origin, which had no prototype in the traditional Ottoman City. Like other western cultural forms, these houses also, evolved within the pluralistic media of Istanbul, underwent mutation and joined the city structure. The row-houses display a line of development which is significant for the history and sociology of their period; thus from their location and distribution in the city, the social tendencies in the 19th Century capital can be discerned. Due to the fact that the row-houses built in Istanbul in the second half of the 19th Century, gives us insights into the social structure and morphological patterns of housing, it is thought to be instructive to study them in terms of their morphological and topological characteristics. In chapter 2, knowledge-based design systems are explained in detail. During the past decade, due to the introduction of artificial intelligence technologies, there has been a potential breakthrough in computer-aided design. As a result of this breakthrough, knowledge-based design systems has emerged with a great applicability in the domain of design. Knowledge- based systems employ facts as constituent elements like the data in databases. They can be defined as a collection of facts (data) and knowledge (rules) about the facts. One of the main properties of knowledge bases is the reasoning capability which can potentially support the inference of the new knowledge at any stage of design and separation of knowledge base from its reasoning mechanism. Inferencing mechanism is a tool which make derivations from both facts and knowledge, (Oxman, 1 992). The main components of a knowledge-based expert system are: an interface facility, facts base, a knowledge base, an inference engine, an explanation facility and a knowledge-acquisition facility, (Coyne, 1990). There are two heuristic approaches used in knowledge-based architectural design:. Top-Down,. Bottom-Up. The combinatorial use of the both approaches is usually prudent since the strengths and weaknesses of each method complete each other and give more effective results. The bidirectional propagation of constraints and multilevel analysis is illustrated with an application from literature which is related to the production of small house plans in top-down approach. Also two knowledge-based design models, one in top-down (GPRS) and the other, bottom-up (PREDIKT) are introduced. xvlliThe third and fourth chapters are related with the method used to analyze the row-houses in istanbul and the results obtained. Morphological Analysis method which is convenient for the classification and comparison of the different building typologies in the studies of architectural history is used to analyze these houses. In the context of this analysis, the geometrical and topological properties of houses are examined and houses are defined in terms of some numerical values, graphics and mathematical procedures. Although conventional architectural drawings (such as plans) describe this morphological structure, it is difficult to treat them objectively and quantitatively for the purpose of analytical discussion (Shoul, 1 993). Some graphic devices are used to represent the morphological patterns and the topological characteristics of dwellings. The methods used for the morphological and topological analyses in this research are Hillier and Hanson's morphological method, and Bloch's method of analysis in the properties and classification of rectangular arrangements.. Hillier and Hanson Analysis The morphological characteristics of a plan which represent the spatial relationships within the house and between the interior and exterior of the house are best revealed by the 'justified access graph' developed by Hillier and Hanson, in a justified access graph all rooms equally deep from a chosen reference space (in this analysis, the outside is chosen as the root space) are drawn on the same horizontal dashed lines. These lines or depth indicators are numbered from zero, to show the minimum number of doorways to be crossed in moving from the reference (root) space, always placed at zero line, to any chosen destination. The number of any particular depth line (0,1,2,..) is also the depth value of any space placed on that line. The depth values given to the spaces, form the basis for determining numerical measures of the spatial system of a given network (these measures are explained below), (Shoul, 1993). The morphological characteristics can be identified by syntactic terms; 'symmetry/asymmetry', 'distributedness/nondistributedness', which are related with the depth and permeability of the network system. Symmetry/asymmetry is about the integrating/segregating effects of a space in relation to the plan. Distributedness/ nondistributedness is about the possibility of getting into and around in the building in more than one way or only one way. These syntactic characteristics are described by RA (relative asymmetry), RRA (real relative asymmetry) and RD (relative distributedness). The Hillier Analysis used in this thesis constitutes the following numeric measures:. number of spaces,. depth of deepest space,. mean depth of all spaces (MD), xlxrelative asymmetry (relative mean depth) (RA), reel relative asymmetry (RRA), maximum vertex valency and the space it belongs, mean vertex valency, cyclomatic number, relative distributedness (RD),. Bloch Analysis Bloch produced a comprehensive catalogue of rectangular dissections which are the dimensionless grating forms of representation of architectural plans. Dissections are organized by means of an hierarchical classification. Dissections are first classified by their 'order1, (n), (n: the number of rectangles in the dissections or the number of spaces in the plan layouts) and then by their 'grating dimensions'. Each rectangle in the dissection occupies one cell in the grating. The next basis for classification is fronts', (f). A front is a rectangle which is in contact with the perimeter of the dissection. The fourth dimension of Bloch's classification relates to the valencies of vertices in the adjacency graphs of dissections. Bloch refers to the distribution of valencies among vertices as the 'graph partition'. This is given as a string of digits. The first digit gives the number of vertices with valency 1, the next gives the number of vertices with valency 2, and so on. As a final classification, Bloch's catalogue gives the 'symmetry1 properties of each dissection. Those dissections having no symmetry are labeled as 'identity1. Dissections with all the symmetries of the rectangle are labeled 'K4'. Dissections on nonsquare gratings with bilateral symmetry Dt are distinguished as to whether the axis of symmetry is vertical (D1V) or horizontal (D1h). Dissections on square gratings with D1 symmetry are always illustrated with the axis of symmetry up and down the page. There is the further possibility that square dissections may possess D^ symmetry about a diagonal axis (Di,dB). The cyclic symmetries C2, C4 are marked as in Steadman (Steadman, 1983). Also 'density1 measure of each rectangular dissection is evaluated which is the ratio of the order, n, to the total number of the grid cells, Im, (I and m are dimensions on x and y axis respectively), contained within the complete grating. In addition to the analyses of Bloch and Hillier, measures related to interior access and adjacency properties of the houses were introduced. These are:. the number of internal access relations, (İG),. the number of internal adjacency relations, (IK),. the ratio of the total number of access relations to the total number of adjacency relations, (İG/İK),. mean valency of interior access, (OrlGD),. mean valency of interior adjacency, (OrlKD). The results of these analysis are presented as morphological groups and a knowledge-base comprising the graphical, visual and verbal knowledge about this house typology. The knowledge base produced here is presented in the form of a catalogue comprising data base cards for each row-house group. The database system developed, is a simple kind of hypermedia which is an effective tool for storing and recalling information in computer systems. The hypermedia-based systems provide a data structure, graphic capabilities and a user interface. It contains textual information, images and graphics. The user interface of this hypermedia is formed with the use of Fox-Pro Software. The fifth chapter includes the transformation of the knowledge obtained from morphological analysis into the syntactical rules and the model which will produce design by using the vocabulary elements of this house type. The design generation process of the houses, is succeeded by the use of abstraction hierarchy in the top-down conception. The generative knowledge of a row house plan type is represented by rules that indicate the order in which decisions are taken in a typical routine or innovative design generation process. The model is mainly composed of two sessions; one is the generation of dimensionless house plans as rectangular dissections and the other is the function assignment to spaces and dimensioning. The rules are represented as stages as follows: 1. the first decision is about the zoning type of the plan being either 'exterior- exterior', or 'exterior-interior-exterior'; 2. the next level is related to the number of spaces at the street facade, whereas; 3. the third to the number of spaces at the back and; 4. the forth stage to the number of spaces in the interior zone of exterior- interior-exterior dissections; 5. the next stage includes the rule about the location of stairs. Stair can be located in three different ways within the zone (front exterior, back exterior and internal zone) it is placed in; it can either lie through the length of the zone, or can be withdrawn back or front situating at a corner of the zone; 6. the sixth level is about the assignment of functions and dimensions to the schematic house plans produced so far. First the stair types and then thespace groups indicated for each zone are assigned in different combinations. Finally, the dimensioning is realized due to the predetermined dimensions of the stair types and the fundamental dissection types (exterior-exterior, exterior-interior-exterior). It is expected that in the future studies, the algorithm introduced in the context of the study can be encoded in a programming language and used in design. The hypermedia-based database system developed, is thought to serve as an effective tool for the students to reach the necessary information they are looking for, with the use of current technologies. Thus the database could assist in architectural education by providing the necessary knowledge accumulation. It is also expected that the model developed using the results of this database, can assist the architecture students to comprehend the design process better.