Page 16 - IJOCTA-15-2

Artificial intelligence-assisted station keeping for improved drillship operations
states. This initial step simulates the vessel’s hy- or acoustic-based position references. The algo-
drodynamic behavior in diverse wave and envi- rithm reduces operational costs and improves re-
ronmental conditions. The data obtained from liability by eliminating dependence on external
these simulations form a comprehensive database technologies. Furthermore, the AI algorithm’s
that captures the vessel’s responses across differ- ability to dynamically calculate and apply correc-
ent scenarios. This database is critical as it serves tive thrust ensures precise station-keeping perfor-
as the primary reference for the AI algorithm dur- mance, even under varying sea states and envi-
ing station-keeping operations. Once the data- ronmental forces.
base is compiled, it provides a structured reposi-
tory of hydrodynamic response data correspond-
ing to various sea states and wave scenarios. This
4.5. Algorithm evaluation
database enables the AI algorithm to reference
specific responses and make informed decisions Figure 19 compares two surge response scenarios
about corrective measures. over a 300 s duration. The green line represents

The algorithm receives input from the vessel’s the uncontrolled surge response of the structure,
CoG, which is continuously monitored. When the which shows a significant drift from the initial po-
CoG location is input into the AI system, the on- sition of around 105 m to approximately 160 m by
board INS detects any shifts in its position caused the end of the simulation. In contrast, the pink
by environmental forces such as waves and cur- line (y = 105 m @LCG) represents the controlled
rents. The AI registers these deviations as the position, maintaining a constant surge at 105 m
vessel’s “response” to external forces. For numer- at the longitudinal center of gravity (LCG). It
ical simulations, an assumed value of this response demonstrates that the control system effectively
is entered into the AI system, representing the keeps the desired position against environmen-
positional or orientational shift due to external tal forces that would otherwise cause substantial
disturbances. drift.
The AI’s primary objective is to determine the
magnitude of the thrust force required to counter-
act the detected response and restore the vessel
to its original position. Using the magnitude of
the detected response, the AI searches the pre-
compiled database to identify the corresponding
corrective thrust force. This force is carefully
calculated to ensure alignment with the vessel’s
hydrodynamic characteristics as recorded in the
database.
Once the thrust force is determined, the AI Figure 19. Surge response without any control
calculates how this force should be distributed Abbreviation: LCG: Longitudinal center of gravity
among the vessel’s four thrusters. The total force
is divided and allocated to the thrusters to ensure
Figure 20 illustrates the structural force the
a balanced and effective counteraction of the de-
control system requires to maintain this position,
tected response. The allocation accounts for the ranging from approximately −1.5×10 N to 1.5×
5
orientation and position of the thrusters to maxi- 10 N. The highly oscillatory nature of the
5
mize efficiency and precision in correcting the ves- force profile indicates rapid and continuous ad-
sel’s deviation.
justments needed to counteract external distur-
When the calculated thrust is applied through bances. These force variations occur at seem-
the thrusters, the vessel’s response is mitigated, ingly irregular intervals with varying amplitudes,
allowing it to return to its desired position suggesting complex interaction of environmental
and maintain station-keeping effectively. This loads. The magnitude of these forces provides
methodology ensures the ship remains stable and crucial information for designing the actuation
aligned with its operational requirements even in system and structural components that would
challenging environmental conditions. This AI- need to withstand these repeated loading cycles.
driven approach to station-keeping is highly effi- This force profile excludes wave-induced loads,
cient and cost-effective, as it relies solely on real- meaning the actual required control forces in real-
time data from the INS system rather than exter- world conditions would likely be even more signif-
nal systems like global positioning system (GPS) icant when accounting for wave actions.
211