Artificial Neural Network Modeling of Wind Loads on R.C.C Chimneys with Interference Effects

Abstract

In the design of tall chimneys, estimation of exact wind loads is very difficult. Empirical as well as analytical approach proposed by different researchers for the estimation of along wind response of a chimney doesn’t give satisfactory results. Codal recommendations are only for the estimation of along wind response. No codal recommendations exist for estimation of across wind response of a chimney in isolated chimney case or in interference due to a chimney situated near the main chimney, which often dominates over wind response. Wind tunnel testing is the sole alternative resorted because simulation of atmospheric boundary layer and structural modeling is possible only in wind tunnels. Wind tunnel tests become a laborious and time-consuming affair for interference studies. Artificial Neural Networks (ANNs), because of their capability to map the relationship between the input and output parameters, have been used in the present study, which can predict the along wind and the across wind response of a tall chimneys of different heights with interference effect. ANN modeling has been used to predict the along wind and the across wind response of a tall chimneys of height 83m ,102.3m and 224.24m. In which 102.3 m high chimney is used as test chimney and other two are used for training the network. The developed ANN networks have been trained with 102.3 m chimney of the experimental data and satisfactory results have been obtained thus reducing the wind tunnel testing upto some extent. The number of patterns in training data set has been determined by plotting scatter diagrams between experimental and ANN predicted values of response in interference condition for different number of training patterns.