Neural Networks And Fuzzy Control for Bulk Terminal Operating Management
From Proceeding of Summer Computer Simulation Conference,SCSC95, Ottawa, July 24-26 1994
Pietro Giribone, Agostino G.Bruzzone

---

THE SYSTEM ANALYSIS

The simulation model utilises a simulator that was specially developed by the authors for plant design analysis. It was built according to object-oriented logic using the C++ programming language while personal computers were used as the hardware platforms.
In the specific case, planning includes booking ships in terms of time to guarantee the availability of different types of coal quality. The management problem arises due to the restricted number of slots available for berthing the ships to the plant pier.
In any case, this means that a certain number of booked ships will set sail after waiting at roads, due to faults or malfunctions.
On one hand this phenomenon leads to a major increase in costs in terms of lay-days and demurrage, while on the other it makes the terminal less competitive in the marketplace (in fact, the plant in question is privately owned and will serve power plants belonging to large national electric utility companies in the Northwest Mediterranean basin).
Correct planning will inconveniences, thus guaranteeing efficient purchasing for users. To do this it is not sufficient to adjust the flow of large coasting ships, but instead it is necessary to organise the arrival of smaller ships which will serve power plants and direct transhipment operations.
This scenario must also consider the co-existence of a power station at the site that uses part of the plant to handle purchasing directly, for which the arrival of some ships cannot be controlled, since it is the responsibility of another management process. On the other hand, some purchasing operations refer to annual contracts which still involve the acquisition of clearly defined loads, for which even these input flows are external to operating management.
However, the simulation performed with the model built for design analysis and technical-economic evaluation can be used to build a large data base relative to the behaviour of the plant, despite the fact that it is still in the design stage.
Therefore, with this knowledge, it is possible to train a neural network to estimate the availability at the coal yard for each quality required in order to plan purchases and to organise the current transhipment operations.