TY  - GEN
AB  - This paper presents the results of an investigation conducted to assess the efficacy of Artificial Intelligence (AI) in optimising the hullform of a conceptual frigate. The frigate, with a length of 130 meters and a top speed of 24 knots, served as the subject for the experiment. A baseline hull was initially designed, and spatial constraints were defined for the study. Three different contractors were invited to optimise the bare hull resistance for various speeds and displacements. The optimisation process considered a weighted grading matrix based on the ship's operational profile.
One contractor employed traditional optimisation techniques, including empirical knowledge, regression series, and the modification of ship hydrostatic characteristics. A parametric hullform model was created to produce candidate designs. Final analysis was made using Computational Fluid Dynamics (CFD). Another contractor utilised machine learning with neural networks, with training data sourced from a parametric hullform model and CFD results. The final contractor employed a combination of a parametric hullform model, T-Search optimisation, potential flow, and CFD, with a specific focus on achieving maximum top speed with the lowest resistance.
Results indicated that traditional techniques improved bare hull resistance by an average of 8%. The use of neural networks significantly outperformed traditional methods, demonstrating a remarkable average improvement of 22% in bare hull resistance. However, a critical observation emerged regarding the optimisation focused on top speed only, where machine learning techniques demonstrated a 27% improvement on resistance. This improvement, however, came at the cost of a notable detriment to lower speeds resistance, resulting in an average overall increase in resistance of 16% across the speed range.
The findings of this study present a dilemma for naval architects and customers alike. While prioritising top-speed optimisation may lead to reduced capital expenditures (CAPEX) due to lower costs associated with machinery and auxiliary systems, it could also result in increased through-life costs due to high resistance at lower speeds, which constitute the majority of operational time at sea. This highlights the importance of a balanced approach and careful consideration when implementing AI-based optimisation techniques in naval vessel design. Ensuring a holistic view of performance and costs throughout the vessel's lifecycle is crucial for making informed decisions.
AD  - BAE Systems
AD  - BAE Systems
AU  - Lopes Gamboa, F
AU  - Paterson, N
DA  - 2024-11-06
DO  - 10.24868/11204
DO  - doi
ID  - 11204
JF  - Conference Proceedings of INEC
L1  - https://library.imarest.org/record/11204/files/.pdf
L2  - https://library.imarest.org/record/11204/files/.pdf
L4  - https://library.imarest.org/record/11204/files/.pdf
LK  - https://library.imarest.org/record/11204/files/.pdf
N2  - This paper presents the results of an investigation conducted to assess the efficacy of Artificial Intelligence (AI) in optimising the hullform of a conceptual frigate. The frigate, with a length of 130 meters and a top speed of 24 knots, served as the subject for the experiment. A baseline hull was initially designed, and spatial constraints were defined for the study. Three different contractors were invited to optimise the bare hull resistance for various speeds and displacements. The optimisation process considered a weighted grading matrix based on the ship's operational profile.
One contractor employed traditional optimisation techniques, including empirical knowledge, regression series, and the modification of ship hydrostatic characteristics. A parametric hullform model was created to produce candidate designs. Final analysis was made using Computational Fluid Dynamics (CFD). Another contractor utilised machine learning with neural networks, with training data sourced from a parametric hullform model and CFD results. The final contractor employed a combination of a parametric hullform model, T-Search optimisation, potential flow, and CFD, with a specific focus on achieving maximum top speed with the lowest resistance.
Results indicated that traditional techniques improved bare hull resistance by an average of 8%. The use of neural networks significantly outperformed traditional methods, demonstrating a remarkable average improvement of 22% in bare hull resistance. However, a critical observation emerged regarding the optimisation focused on top speed only, where machine learning techniques demonstrated a 27% improvement on resistance. This improvement, however, came at the cost of a notable detriment to lower speeds resistance, resulting in an average overall increase in resistance of 16% across the speed range.
The findings of this study present a dilemma for naval architects and customers alike. While prioritising top-speed optimisation may lead to reduced capital expenditures (CAPEX) due to lower costs associated with machinery and auxiliary systems, it could also result in increased through-life costs due to high resistance at lower speeds, which constitute the majority of operational time at sea. This highlights the importance of a balanced approach and careful consideration when implementing AI-based optimisation techniques in naval vessel design. Ensuring a holistic view of performance and costs throughout the vessel's lifecycle is crucial for making informed decisions.
PY  - 2024-11-06
T1  - Comparative Analysis of AI-Based Optimisation Techniques for a Conceptual Frigate Hullform Design
TI  - Comparative Analysis of AI-Based Optimisation Techniques for a Conceptual Frigate Hullform Design
UR  - https://library.imarest.org/record/11204/files/.pdf
VL  - INEC 2024
Y1  - 2024-11-06
ER  -