Show simple item record

dc.contributor.advisorCotsovos, Doctor Demitrios
dc.contributor.advisorVal, Professor Dimitry
dc.contributor.advisorLagaros, Doctor Nikos D.
dc.contributor.authorAhmad, Afaq
dc.date.accessioned2019-04-23T15:45:47Z
dc.date.available2019-04-23T15:45:47Z
dc.date.issued2018-07
dc.identifier.urihttp://hdl.handle.net/10399/3477
dc.description.abstractThis project aims to develop a radically new stable, robust and computationally efficient structural analysis procedure capable of realistically and objectively predicting the nonlinear response of reinforced concrete (RC) structures. This procedure will be suitable for both research and practical applications and will be capable of effectively solving design optimization and reliability problems which require extensive parametric studies. For this purpose, Artificial Neural Networks (ANNs) are employed which require significantly less computational resources compared to more traditional approaches of structural analysis based on the non-linear finite element analysis (NLFEA). The procedure is based on the simulation of the nonlinear behaviour of each RC element (ranging from which include typical beams and column) through the use of a model which consist of a finite element incorporating an ANN the latter predicting brittle modes of failure and the associated load-carrying capacity. For this purpose, databases consisting of test data obtained from experiments carried out on a range of simple (determinate) structural configurations (e.g.; Beam, Column, T-beam and Slab) are developed. Subsequently the published test data is used for training the ANN models. The predictions obtained from the trained ANN models are then compared to the predictions of the relevant design codes and alternative assessment methods concerning specific aspects of RC structural behaviour at the ultimate limit state (ULS). For validation of these ANN models, limited nonlinear finite-element analyses are also conducted. These models are then used to form ANN-FEA models to simulate more intricate RC structural configurations consisting of more than one structural elements. In the latter ANN-FE models, ANNs are essentially used as a failure criteria when conducting non-linear static push over analysis. The stability and robustness of the proposed structural analysis method, as well as the validity and objectivity of its predictions, is ensured through a comparative study of the predicted behaviour of RC frames under static loads with its experimentally and numerically established counterparts. The predictions obtained from ANN-FE models are compared to their counterparts obtained from professional and research analysis packages for the case of a number RC structures. The proposed procedure employs the ANNs as failure criteria defining the loadbearing capacity and mode of failure exhibited by the individual RC beams and columns during the pushover analysis. The results show that the ANN-FE model predicts the structural response of RC at ULS with more accurately as compared to industrial tools (i.e., SAP 2000) and in less amount of time without requiring the high computational resources as compare to research tools (i.e., ABAQUS).en_US
dc.language.isoenen_US
dc.publisherHeriot-Watt Universityen_US
dc.publisherEnergy, Geoscience, Infrastructure and Societyen_US
dc.rightsAll items in ROS are protected by the Creative Commons copyright license (http://creativecommons.org/licenses/by-nc-nd/2.5/scotland/), with some rights reserved.
dc.titleReinforced concrete (RC) structures analysis and assessment with artificial neural networks (ANNs)en_US
dc.typeThesisen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record