ABSTRACT

Experimental studies, which are essential methods in engineering, are powerful techniques in terms of designing structures, optimization, and improvement of design. However, the results illustrated by experimental studies evaluating the effect of multiple factors on measures of performance may not be interpreted as statistically significant. In addition, the limited resources (time, labor and financial resources, etc.) are among the most important limitations of the experimental studies. In this study, three main factors that determine the working conditions of the pneumatic system developed to reduce the sediment accumulation in the ballast tanks of the ships are considered and the effect of these factors on sediment reduction is examined statistically. With the classical statistical approaches applied within the study, the experimental data could not be interpreted in a meaningful way at the desired level. For this reason, instead of interpreting the relationships in the data set obtained from the experimental set through a statistical model, it was determined to reveal the relationship between the variables directly through the data. Therefore, although the number of data in the dataset is limited, the Artificial Neural Networks approach been applied (ANN). The number of input data in the ANN structure set greatly affects the accuracy of this approach. Although it is theoretically possible to increase the data set to an infinite number in the experimental study in question, this is not applicable due to resource limitations, particularly time and labor. Because of that, another application is created when forming the ANN model, considering the purpose of the experimental study. Some of the experimental data is used in the training phase of the ANN model, and the ANN model is provided to suggest an optimal working condition. Experimental results not included in the ANN training phase are used to experimentally compare the optimal working condition proposed by the model. In this study, ANN is used as a tool for optimization and the optimum operating condition suggested by the model provides the best result in terms of sediment accumulation reduction among all experimental data. In this study, the problems experienced in the statistical interpretation of the experimental study results for the optimization of the operating parameter of the system developed to reduce the sediment accumulation in the ballast tanks of the ships and the approach applied when using Artificial Neural Networks (ANN) are discussed.