Analysis of Reinforced Concrete Chimney Using Artificial Neural Network

Abstract

The present major project work deals with the utility value of Artificial Neural Network for the Reinforced Concrete Chimney. In the field of engineering, the present day scenario due to paucity of time needs one to arrive at solution with minimal input. To reach the desired solution with accuracy using conventional technique, some validation is needed by an expert or an experienced person. In such circumstances some artificial intelligence tool can accomplish this task in absence of an expert or an experienced being. A chimney as a structure is a tall slender stack like structure, behaves as a free cantilevered structure. It is not the complete solution for pollution control but it is to be provided for the purpose to reduce the concentration of flue particles that are harmful to mankind. The analysis of R.C.Chimney is very time-consuming process as far as dynamic and modal analysis is concerned. It requires a clumsy computation, as it requires to follow the standard guidelines. Herein, the value of A.N.N. has been utilized for the analysis purpose.

In present work, as far as analytical solution is concerned, the separate spreadsheets have been developed after having studied the proper and related literature from various sources. In spreadsheet calculation, the calculation of dynamic properties likes Mass at various sections, Moment of Inertia about an axis, Area of cross section, Eigen Value, Eigen Vector, Time Period etc… have been calculated as per the vital requirement of A.N.N. like as input vector and target vector. Based on that, as per codal provisions in IS 4998-1992(PART 1), due to vortex shedding, due to drag and gust factor force, the calculation of wind force has been carried out for along wind load as well as for across wind load of simplified approach as well as for Random Response approach. Aforesaid cases gives the amplitude which is tip deflection of chimney, Critical Wind Speed, Sectional Shear Forces and Sectional Bending Moments for Simplified as well as for Random Response approaches related to the aforesaid cases of Wind Load. As per IS 1893-1984 criteria for earthquake analysis, the modal analysis has been carried out with reference to concern literature and the modified acceleration

Abstract

have been found out. It depends upon damping factor, soil condition, modal values, time period, importance factor, and zone factor. Using the value of modified acceleration, reliably the sectional shear forces and bending moments have been calculated up to desired no. of modes. Finally, the Design shear force and Design Bending Moments have been carried out using the method of combination the modes is SRSS (Square Roots of Summation of Squares).

As far as utility value of Artificial Neural Network for the analysis of R.C.Chimney is concerned, reliable and sufficient amount of data needs to be provided for the training purpose. As A.N.N. operates like human brain, where learning takes place with real life experiences becoming input data, more the experiences better is learning and better response to the similar situation. Herein, for the study purpose, total no. of data for training as well as for testing purpose is 44 cases, with different configuration of Reinforced Concrete Chimney such as height of chimney, top diameter, bottom diameter, top thickness, bottom thickness, basic wind speed as per its location (Vb), earthquake zone etc… have been gathered from the actual practice. Out of 44 cases, for the training purpose, 30 cases have been chosen. And after stack the proper training, the remaining data have been tested and its robustness, sensitivity could be measured with analytical solution. There are no. of Neural Network model used for real complex structure but out of that, herein, the Feed Forward Back Propagation Neural Network Model has been fully utilized for the analysis of reinforced concrete chimney. The present study could show the value of A.N.N. in to relevant chapter in form of graphical representation.

The outline of relevant Chapters is given in to the specific chapter called as synopsis.