Modeling flow coefficient by using fuzzy SMGRT method

The quality of today's cities is determined by advanced infrastructure systems. A functional and well-designed infrastructure system provides facilities for more comfortable human life and contributes significantly to urbanization. Rainfall drainage systems are important components of infrastructure systems. Paralleling to the development in urbanization, in residential areas impermeable surfaces i.e. asphalt, concrete coatings, car parks, and similar impermeable surfaces increase both in number and area. For this reason, the rainwater falling to the surface especially in city centers passes to the stream in a shorter time with higher flow ratio. In order to prevent environmental disasters, it is necessary to remove the rainfall from the residential areas by means of drainage systems as soon as possible. Therefore, rainfall-flow relationship must be correctly determined. This is vitally required for designing hydraulics system. There are many different conventional or artificial intelligence techniques; deterministic or empiric; mathematical or physical models developed for the aim available in the current literature. It is almost impossible to deterministically model the hydrological events, because such events are very complex and have many uncertainties. In another word, just as many other natural events, hydrological events also imply indistinctness. So in modeling hydrological events, it can be preferred the nondeterministic methods instead of deterministic ones. However, when making such a choice, the modeler should avoid from black-box methods in order not to ignore the physical dimension of the event. In order to model flow coefficient in this study, the fuzzy logic, which keeps both the uncertainties and physic of the phenomenon, has been preferred to other modeling techniques. The fuzzy logic modeling technique depends on both the fuzzy sets theory and fuzzy rules base, which consists of linguistic and quantitative if-then statements. It is possible to note that the physics of the event can be reflected in both designing of fuzzy sets and in the fuzzy rule base. Therefore, the two most important issues in a fuzzy modeling are to precisely determine the fuzzy sets and the fuzzy rule base. Although there are many methods developed for this purpose, SMRGT (Simple Membership Functions and Fuzzy Rules Generation Technique), which is a new and capable to precisely determine both the fuzzy sets and fuzzy rules base together, has been chosen. On the other hand, SMRGT is not a purely data-based; it is not considered among black-box methods. It is understood from the current literature that the SMRGT keeps uncertainties and physics of the event, give quite realistic results and do not need trial and error process. The program designed by using the SMRGT method is operated in MATLAB software. The model developed to determine the flow coefficient consists of three sub-models. The first sub-model takes into account the meteorological features, the second and the last one takes into account the land use conditions and the soil characteristics, respectively. Then, a single flow coefficient was obtained by taking the average of the three flow coefficients obtained from the three models. The model's exists for Turkey. However, with a small change to be made by changing the boundary conditions of the models, the model can be validated for all hydrological / meteorological, land use and soil characteristics. In the thesis, an application of the model has been given and flow coefficient for the campus area of Sırnak University has been determined. Thus, the coefficient of flow has been tried to be calculated more precisely. In addition, the model can be applied to any part of the world by changing the existing boundary conditions. The thesis consists of six chapters. Following to the introduction chapter, the literature up to date is discussed. The third chapter contains the methodology of the thesis. The development of the models and their applications together with their results' discussion are presented in chapter four. The work is concluded in chapter five. The last chapter includes the list of references investigated for the thesis.