Cost Estimation of Reinforced Concrete Buildings by Using Neural Network and Multi Regression Analysis

In this study, an Artificial Neural Network and Multi Regression Analysis have been used to evaluate the strengthening cost and total cost of reinforced concrete buildings. To obtain strengthening cost, 377 reinforced concrete buildings which have been designed according to the 1975, 1997 and 2007 Turkish Earthquake Codes have been checked and strengthened according to the new code (2018 Turkish Earthquake Code). After that, to obtain the total cost of the buildings according to the new code, 84 different reinforced concrete buildings have been designed according to the 2018 Turkish Earthquake Code. 4 different places at 4 different earthquake zones in İstanbul have been chosen to make the study. The professional program Sta4CAD has been used to model, analyze and strengthening those reinforced concrete buildings. The parameters which affect the cost of the buildings will represent the input and the strengthening cost and total cost of the buildings will represent the output. When the old buildings will be checked according to the new code, they may not satisfy the conditions of the code. Since the new code has more general rules. According to that, those old buildings will need strengthening. Section enlargement method, addition of shear wall and other methods described in chapter 4 will be used so that the old buildings will satisfy the new code. For strengthening cost of Reinforced Concrete buildings, 13 parameters have been chosen accordingly. These parameters are: Number of Storey (N), Concrete Class (C), Steel Class (S), Plan Area (A), Shear Wall Ratio (SWR), Column Ratio (CR), Earthquake Code (EQ), Stirrup Spacing, Soil Type (ST), Earthquake Zone (EZ) Torsional Irregularity, Weak Column-Strong Beam and Soft Storey. The output parameter for the study is the strengthening cost, which are in Turkish Lira according to the unit prices of materials in Turkey. For total cost according to TEC 2018 8 parameters have been used. Those parameters are: Plan Area (A), Number of Storey (N), Concrete Class (C), Steel Class (S), Shear Wall Ratio (SWR), and Column Ratio (CR). Finally the input parameters of the strengthening cost will be sorted accordingly to the importance. According to the study, the prediction accuracy of the Artificial Neural Network that has been trained, found 94% accuracy for the strengthening cost calculations of buildings. However for the Multi Regression Analysis Method, 71% accuracy has been found for strengthening cost. For total cost, Artificial Neural Network gave 97% accuracy and for Multi Regression Analysis Method 95% accuracy has been found.