Ph.D

Mr Seyyed Hosein Mousavi  will present his P.H.D thesis on " The Appraisal Model of Architectural Design of Natural Stone Cladding in order to Predict its Service Life  "Sunday, January 28, 2018, 15 pm The Amphitheatre of School of Architecture and Environmental Design . 
Supervisors: Dr. Ahmad Ekhlassi Dr. Seyed Bagher Hosseini 
Advisors: Prof. Jorge de Brito Dr. Rahman Farnoosh 
Jury: Dr. Ali Reza Falahi Dr. Katayoon Taghizadeh Dr. Farhang Mozafar Dr. Naser Koleini

Abstract:

In recent decades, several studies and international standards have been proposed to provide methods for assessing the durability of buildings, in addition to predicting their service life, for the reason that the service life prediction of buildings and their components has paramount importance for the concept of a sustainable environment, enabling a more rational use of resources.

On the other hand, the external cladding can be seen as the most exterior layer of the building, and therefore more exposed to agents causing degradation, it is also more prone to defects. Moreover, it can be drown from the previous researches, carried out on natural stone claddings, that inappropriate design and specification, in comparison with the other probable cause defects, are often directly or indirectly responsible for a large number of failure situations observed in stone claddings. Therefore, the current study proposed the evaluation model of the architectural design of natural stone claddings which can be considered as a supporting tool that help the designers to implement the optimal façade design with maximum service life.

Accordingly, this study was conducted in four stage. The first step was allocated to literature review in order to draw the definitions and principals required for the research process. The second stage was allotted to data collection using visual survey. In the third step, in order to identify the inputs of the prediction model of service life, the numerical index of the severity of degradation (Sw) was calculated for all samples. Then the statistical tests were used to determine the independent variables related to the dependent one. Finally, the service life prediction model of natural stone cladding will be established based on the selected variables in the last step. At the end, the validation of the predictive capability of the proposed model is performed based on statistical indicators and the sensitivity, the specificity and the accuracy of the model are evaluated.

In this study, data obtained from visual inspection of 507 samples in Tehran (Iran) and Lisbon (Portugal) are arranged into two categories: 1) the characteristics of the building and its façade, 2) the type and the extent of observed defects. After the calculation of the numerical index of the severity of degradation (Sw), in order to analyze various properties and understand their influence on the reference service life gain or loss, the ratio between the characteristic- related and global average service lives was established. Thus, designers are capable to neutralize the negative impact of some factors by changing the others in order to obtain the highest durability of stone claddings using this ratio.

The data which obtained from the inspection of 365 samples with direct fastening system in Tehran (162 samples) and Lisbon (203 samples) comprising 8 characteristics: the city, type of stone, color, size of stone plate (area), type of finishing, location of cladding, orientation and age was selected by statistical tests as well as the colinearity diagnosis was evaluated.

In the final step, the fuzzy model based on the adaptive neuro-fuzzy inference system (ANFIS) was used to predict the service life of natural stone claddings with direct fastening system. In order to optimize the proposed model, the evolutionary algorithm named Particle Swarm Optimization (PSO) was used.

The validation of the proposed model reveals the significant decrease in errors in addition to strong correlation between samples and the model which verifies the performance improvement in comparison with the models presented in similar studies. Moreover, considering the mean absolute percentage error (MAPE), the proposed model provides potentially very good predictions and according to the sensitivity, specificity and accuracy, it shows good classification capability and the outputs are closer to reality.

With respect the fact that all 8 variables used in proposed model are accessible in the design stage and ascertained by designers except orientation, age and the city, the proposed model can calculate the severity of degradation of natural stone cladding with considering the contractions of all variables. Therefore, this model can be consider as an aided design tool which makes it possible for designers to predict the service life of natural stone cladding in design stage and evaluate the influence of each characteristic on the reference service life gain or loss regardless of their knowledge related to service life concept.

 

Keywords: Natural stone cladding, Service life prediction, Degradation state, Fuzzy logic.