MODELLING OF CAPACITY INCREASE OF PORTS

In this study, an artificial intelligence based Port Capacity Prediction Model (CAPREM) was devoleped to predict handling capacity which forms the basis of revision plans that have considerable importance for ports. Time delayed multiple nonlinear regression, neurofuzzy and time series analysis sub-models were used to predict handling capacity of ports, error ratios and correlation values were computed to compare the results. This new model in port capacity planning was suggested to consider inherent uncertainties in the handling capacity by the application of deterministic, probabilistic and artificial intelligence sub-models. The artificial intellegence (neurofuzzy) sub-model is a time lagged total handling prediction model and it was the first model that gave a better prediction in port handling capacity as compared to neural network models in the literature.
Loading and unloading data of Mersin Port obtained from Undersecreteriat of Maritime Affairs were used for the application of the model. The existing capacity of Mersin Port and the time required to reach handling capacity were predicted by the developed model. Revision scenarios to increase the handling capacity of the port, such as purchasing various port equipment and improvement of port configuration were suggested. The newly developed model was successfully applied to Mersin Port.