1999 Fiscal Year Final Research Report Summary
A STUDY ON STOCHASTIC WEATHER ROUTING BASED ON INFORMATION OF THE ACCURACY OF NUMERICAL WAVE FORECAST
| Project/Area Number |
10650901
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
船舶工学
|
|---|
| Research Institution | Tokyo University of Mercantile Marine |
|---|
Principal Investigator |
HAGIWARA HidekI TOKYO UNIVERSITY OF MERCANTILE MARINE, Prof., 商船学部・情報システム設計工学講座, 教授 (30126338)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
SHOJI Ruri TOKYO UNIVERSITY OF MERCANTILE MARINE, Lecturer, 商船学部・情報システム設計工学講座, 講師 (50272729)
KUWASHIMA Susumu TOKYO UNIVERSITY OF MERCANTILE MARINE, Prof., 商船学部・情報システム設計工学講座, 教授 (30016943)
|
|---|
| Project Period (FY) |
1998 – 1999
|
| Keywords | NUMERICAL FORECAST / STOCHASTIC ROUTING / SIMULATION / ERROR ESTIMATION |
|---|
| Research Abstract |
In carrying out ship weather routing, since the forecasted environmental data inevitably involve forecast errors, the predicted passage time and fuel consumption on a certain route contain prediction errors. It is very useful if the standard deviations or 95% (99%) confidence intervals of passage time and fuel consumption on a certain route can be estimated at the departure time or during the voyage. By means of these standard deviations, the degree of uncertainties of the arrival time and fuel consumption can be shown quantitatively, which makes it easy to plan the operation schedule of a ship and to determine a margin for fuel oil to be loaded. Furthermore, it is possible to incorporate these standard deviations into the objective function in determining an optimum route from a stochastic point of view.
In this study, the accuracy of forecasted winds, waves and ocean currents is represented as the covariance matrices of the forecast errors and the correlations between successive forec
… More
ast errors. Based on these covariance matrices and correlations, a method of estimating the standard deviations of predicted passage time and predicted fuel consumption on a particular route is described. By incorporating these standard deviations into the objective function, a method to perform stochastic minimum time/fuel routing is proposed.
To verify the accuracy of estimated standard deviations of passage time and fuel consumption, the accuracy of a 10-day wave forecast operated at a certain weather routing company is analyzed first. This wave forecast model receives 10-day surface pressure forecasts from a forecast center and converts those data into wind fields. These wind fields are input into the spectral wave forecast model to produce 10-day wave forecasts. By comparing forecasted values with actual (analyzed) values over 4 months (December 1996 - March 1997), the accuracy of 10-day wind/wave forecasts in the North Pacific is investigated.
Then using 10-day wind/wave forecast data, a number of stochastic minimum time routing simulations of a container ship crossing the North Pacific are performed. In each simulation, the container ship is also navigated along the various fixed routes. On the minimum time route and the fixed routes, the passage times and fuel consumptions are predicted and their standard deviations are estimated. The container ship is navigated again along the minimum time route and the fixed routes using the actual wind/wave data, and the actual passage time and fuel consumption on each route are computed. By statistically comparing the errors of passage time and fuel consumption with their standard deviations, the accuracy of estimated standard deviations is quantitatively investigated and the effectiveness of stochastic routing is fully demonstrated. Less
|
