3D wave transmission around permeable submerged breakwaters with the use of Artificial Neural Networks

3D wave transmission around permeable submerged breakwaters with the use of Artificial Neural Networks

Tavouktsoglou, N.S.

Stive, M.J.F. (mentor)
Simons, R.R. (mentor)
Visser, P.J. (mentor)
Verhagen, H.J. (mentor)

Civil Engineering and Geosciences

Hydraulic Engineering

Coastal Engineering

2013-10-07

Wave transmission is a very important design parameter for submerged porous breakwater design, as it defines the dimensions of the breakwater as well as the cost and construction process. The spatial distribution of the wave transmission parameter influences the current pattern on the lee of the breakwater and therefore the sediment transport process. For this reason it is important to create a design tool capable of predicting the variation in wave height around detached submerged breakwaters. This study has been conducted as an extension of the work completed by Amir Ahmadian for his PhD project the University College of London under the supervision of Professor Richard Simons. During his research a large number of experiments were conducted in order to create an extensive database on wave transmission around semi-infinite impermeable breakwaters. The results of these experiments where then used to create an ANN model capable of predicting the 3D wave transmission coefficients around submerged breakwaters. This thesis therefore aims to create an ANN model capable of predicting the 3D wave field around permeable submerged breakwaters, by using the algorithm architecture proposed by Ahmadian. To the author’s knowledge there are a limited number of experimental studies on the field of 3D wave transmission of permeable breakwaters and therefore creating an ANN model based on physical measurements is impossible. For this reason a large number of 3D experiments where performed using MIKE21 BW in order to create a database that will then could be used to train and test the ANN model. Important evidence of the significance of diffraction and breakwater permeability on the wave transmission phenomenon for submerged porous breakwaters where obtained. In addition the results of the simulations where then cross validated against the empirical formula provided by Vicinanza et al (2009). This analysis showed that the quality of the data was very good and could be used for training a Neural Network. During this process it was proposed that the empirical prediction formula of Vicinanza could be improved by introducing a correlation factor, as the numerical simulations showed strong evidence that the diffraction and wave transmission over and through the breakwater have a negative correlation. With regards to ANN modeling the algorithm showed that it has an excellent capability to predict the test dataset (obtained from MIKE21 BW simulation). The analysis of the ANN model revealed that the model predictions are in very good agreement with the prediction method of Vicinanza. Finally the sensitivity analysis of implemented showed that the permeability factor introduced to account for the effects of permeability has the most important contribution to the models performance. Concluding this thesis suggests that the proposed model has the potential to become a valuable design tool for engineering purposes in the field of submerged breakwater design.

submerged breakwaters
ANN
wave transmission

http://resolver.tudelft.nl/uuid:1f8c9197-287d-4dbb-be0d-8a5074e40adc

Student theses

master thesis

(c) 2013 Tavouktsoglou, N.S.