Adaptive Control Strategy for the Dynamic Positioning of a Shuttle Tanker During Offloading Operations

In deep water oil production, Dynamic positioning systems (DPS) strategy has shown to be an effective alternative to tugboats, in order to control the position of the shuttle tanker during offloading operations from a FPSO (floating production, storage, and offloading system). DPS reduces time, cost, and risks. Commercial DPS systems are usually based on control algorithms which associate Kalman filtering techniques with proportional-derivative (PD) or optimal linear quadratic (LQ) controllers. Since those algorithms are, in general, based on constant gain controllers, performance degradation may be encountered in some situations, as those related to mass variation during the loading operation of the shuttle tanker. The positioning performance of the shuttle changes significantly, as the displacement of the vessel increases by a factor of three. The control parameters are adjusted for one specific draught, making the controller performance to vary. In order to avoid such variability, a human-based periodic adjustment procedure might be cogitated. Instead and much safer, the present work addresses the problem of designing an invariant-performance control algorithm through the use of a robust model-reference adaptive scheme, cascaded with a Kalman filter. Such a strategy has the advantage of preserving the simple structure of the usual PD and LQ controllers, the adaptive algorithm itself being responsible for the on-line correction of the controller gains, thus insuring a steady performance during the whole operation. As the standard formulation of adaptive controllers does not guarantee robustness regarding modeling errors, an extra term was included in the controller to cope with strong environmental disturbances that could affect the overall performance. The controller was developed and tested in a complete mathematical simulator, considering a shuttle tanker operating in Brazilian waters subjected to waves, wind and current. The proposed strategy is shown to be rather practical and effective, compared with the performance of constant gain controllers.