Evolutionary design optimization of a torpedo-shaped autonomous underwater vehicle using a semi-empirical hydrodynamic model and design-specific CFD assessment


Özgan S., Bayat O., VARDAR D., Atilla D. Ç., Karagöz E. K.

Journal of Marine Science and Technology, 2026 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Basım Tarihi: 2026
  • Doi Numarası: 10.1007/s00773-026-01159-2
  • Dergi Adı: Journal of Marine Science and Technology
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Compendex, INSPEC, Academic Search Ultimate (EBSCO), Natural Science Collection (ProQuest), Earth, Atmospheric, & Aquatic Science Collection (ProQuest), Engineering Source (EBSCO), Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
  • İstanbul Üniversitesi Adresli: Evet

Özet

This study presents a computationally efficient design and assessment framework for a torpedo-shaped, twin-tube autonomous underwater vehicle (AUV). The evolutionary optimization stage is performed in MATLAB using a low-fidelity semi-empirical hydrodynamic model, whereas high-fidelity computational fluid dynamics (CFD) is applied subsequently and is not embedded within the optimization loop. The optimization study comprises both single-objective and bi-objective formulations under geometric, buoyancy, stability, and physical-feasibility constraints. Single-objective drag minimization is conducted using L-SHADE, IDEA, and CMA-ES, while the bi-objective problem of minimizing drag and improving longitudinal stability is solved using NSGA-II, MO-IDEA, and RVEA. The quality of the resulting Pareto fronts is assessed using the Hypervolume and Spacing metrics. L-SHADE provides the best convergence performance for the single-objective problem, whereas NSGA-II produces the highest-quality Pareto front in terms of convergence and solution distribution. The low-fidelity optimization model predicts drag forces of 1.88–2.00 N and drag coefficients of 0.18–0.20. Following completion of the optimization process, one selected geometry is modeled in SolidWorks and independently evaluated in ANSYS Fluent under steady, incompressible flow conditions using the k–ω SST turbulence model at an inlet velocity of 1.03 m/s. The detailed CFD analysis yields a drag force of approximately 2.10 N and a drag coefficient of approximately 0.20, corresponding to a value approximately 5–11% higher than the range of low-fidelity estimates. This difference is attributed primarily to the simplified geometric representation used in the semi-empirical model, whereas the CFD model includes the complete three-dimensional CAD geometry, including thruster tunnels and nonuniform hull transitions. The CFD result is therefore interpreted as a design-specific hydrodynamic assessment rather than validation of the entire optimization landscape. A physical prototype based on the selected geometry is subsequently manufactured and tested in a swimming-pool environment, demonstrating stable straight-line motion, depth-regulation capability, and practical feasibility. The proposed workflow combines extensive low-fidelity design-space exploration with independent design-specific CFD assessment and prototype-level experimental evaluation.