Summary:
One of the main data used as a reference to schedule and manage port operations are Estimated Time of Arrival (ETA) and Estimated Time of Departure (ETD). The shipping agent needs to inform and update about such information. But generally, after the port call request occurs, estimated time data tends to remain static and is not updated as would be desirable, despite changes, delays and incidents arising indeed. This results into more challenging day-to-day operations due to the need of dealing with unexpected delays. In the specific case of ferry/RoPax passenger ships in the Strait of Gibraltar, operational management is even more complex since delays in rotations may affect the following port calls scheduled for the entire day. These scenarios can be further complicated during OPE, when 14 ships from 5 different shipping companies rely on 5 berthing spaces at Dársena de la Galera Terminal.
To handle these cases, a pilot project was developed with the aim to obtain (1) updated data in real time about the location of ships, (2) updated predictions or estimated times of arrival and departure, and (3) an optimisation model that takes into account possibly global delays within each schedule, in order to readjust operations (e.g. times, docking spaces, etc.) and ensure an excellent quality of services for end users.
This is why the prototype of an expert and predictive system was developed, which would assist the decision-making process as regards the berth assignment for ferry/RoPax ships. This expert system has two components: a predictive model to estimate times of arrival in port, and a mathematical optimisation model to manage docking operations and waiting times for ships arriving in port. The predictive model aims to estimate times of arrival in port of the ferry/RoPax ships with scheduled port calls in the next few hours at the Port of Algeciras. To do so, the following factors have been taken into account: current location through AIS data, delays occurred in previous rotations, historical behaviour patterns of ships and/or shipping companies, hour/day of the week, departure/return operations, weather conditions (wind and waves) and even the effect of the holiday calendar and pilotage exemptions.
Moreover, the mathematical optimisation model (linear programming) makes it possible to readjust berth allocations throughout the day as port calls take place. Said optimisation model is based on an objective function characterised, on the one hand, by minimising the difference between the time of arrival proposed by this model, and the one offered
by the prediction model. On the other hand, docking preferences, based on the virtual costs involved when ships are allocated docking spaces other than their preferred/favourite.In other words, if at the predicted time of arrival of a ship there is no preferred docking space available, the possibility of delaying its arrival may be considered, so that it can be allocated said docking space, or else a different one may be allocated, maintaining the predicted time of
arrival. The optimisation model will opt for one or the other option depending on the result of their decisionmaking
process.
As a result of this optimisation model, a berth allocation proposal and a required time of arrival (RTA) are made available for every upcoming port call. Both options will be adjusted according to the optimisation criteria chosen in each case, so that docking and time of arrival changes can be set. The latter should be done in such a way that the ship can reduce its cruising speed accordingly for a Just-in-Time arrival.
In addition, the model proposes a berthing schedule for the rest of the day’s calls, indicating necessary berth changes or arrival time requests.
The conclusions drawn from the proof of concept are:
- It has been shown that predictive models, since they deal independently with multiple factors regarding estimated times of arrival of ships in port, offer a better assessment of possible delays in comparison with average delays of the fleet as a whole.
- The predictive model, by using AIS data, makes it possible to obtain accurate information in the short term, and so, to make the appropriate operational decisions. In the medium term, it helps us focus on certain port calls with specific estimations of delay, which require special attention.
- Optimisation models allow, in a very simple and agile way, and on the basis of predefined criteria and objectives, to set precise docking times and reallocate docking spaces, taking into account their possible results in the future.
- The optimisation model realises port call schedules daily, on the basis of predefined criteria and with a minimised risk of future allocation changes, and the reason for this is a more regular distribution of said port calls.
- Lastly, having such an optimisation model makes it possible to devise a guidance fleet plan (blueprint) to avoid, as far as possible, future delays and changes.
The project’s innovation:
- Use of historical data and real-time data to assist and improve berth management decision making.
- Development of predictive and optimization models to improve berth management decision making.
The project’s product:
- Software prototype of an expert system to determine the optimal berthing and required arrival time of a Ferry/Ropax vessel scheduled to call in the next hours.
Applicability:
- The future product or decision support tool (based on the validated prototype) would be very useful for the operational management of berthing points dedicated to traffic in the Strait of Gibraltar, especially during the OPE period.
Leave a Reply