Aerodynamic Analysis of a DEP Surveillance Aircraft
Via Machine Learning and Quantum Approach
Aerodynamic Analysis of a DEP Surveillance Aircraft
Via Machine Learning and Quantum Approach
24.4
90.27 %
1.0 %
Cruise L/D ratio, RANS CFD. Maximum 24.85 on the clean polar.
1.0 %
90.27 %
1.0 %
ML surrogate against an independent CFD case
720×
90.27 %
90.27 %
Faster than solving the flow field from scratch, using a trained surrogate.
90.27 %
90.27 %
90.27 %
Agreement with NASA 2025 CFD reference
One platform, three computational frontiers
High-fidelity CFD
High-fidelity CFD
High-fidelity CFD
OpenFOAM + OpenVSP multidisciplinary framework — k-ω SST turbulence, actuator-disk rotor modelling, energy-thermal and flight-envelope analysis.
Machine learning
High-fidelity CFD
High-fidelity CFD
Random Forest surrogate over 486 configurations (R² = 0.972), SHAP explainability, and a live Streamlit performance predictor.
Quantum hybrid
High-fidelity CFD
Quantum hybrid
A PyTorch–Qiskit variational circuit trained on lattice-Boltzmann data achieved 1.000000 fidelity at 8 × 8. Gate counts grow polynomially for streaming and exponentially for collision, with break-even near 16 × 16.
What the study found
AERODYNAMICS
AERODYNAMICS
AERODYNAMICS
Over-the-wing rotors boost lift
Over-the-wing rotors raise CL by 12.89 % and CD by 12.01 % at the cruise point, taking L/D from 24.215 to 24.398. Cruise sits at 98.3 % of the clean-polar optimum of 24.85, leaving margin in hand when conditions move.
ENDURANCE
AERODYNAMICS
AERODYNAMICS
4.11 hours on station
4.11 h on station at a 100 km radius, from a 306.6 kg pack holding 122.6 kWh at 400 Wh/kg and 80 % depth of discharge. Raw endurance reaches 7.18 h at the same cell energy. Ground-level noise from 3 000 m is 9.1 dB, more than 20 dB below rural ambient, alongside a low thermal signature.
VALIDATION
VALIDATION
VALIDATION
From simulation to flight
A bench smoke tunnel and a 1:9.8, 120 cm span PLA and foam prototype put separation onset near 14°, in line with the blown CFD case at 14 to 15°. The prototype flies, confirming that the distributed layout can be built and balanced.
EFFICIENCY
VALIDATION
VALIDATION
Data-driven design
The surrogate screens 1 000 designs in milliseconds and agrees with an independent CFD case to 1.0 %. Trained on lattice-Boltzmann data, the hybrid model returns a section estimate about 720 times faster than solving the field.
Run the models yourself
Two live Streamlit tools let you predict aerodynamic performance and explore the quantum flow model in real time.
© 2026 Batuhan Türk · İTO Science and Art Center
Welcome to the AeroQubit
This is a brief interactive preview rather than the full project, giving you a quick look at its key features and concepts. Enjoy exploring!