STUDI KOMPARASI OPTIMASI PARTICLE SWARM DAN ALGORITMA GENETIKA PADA PERANCANGAN FOIL

Abstract

Foil is a ship installed wings at bottom of hull support. If the speed’s ship increases, foil will produce a lift that causes the hull rised and exit from the water. Therefore, there will be a reduction in friction force that results in the ship's velocity increase. The choice of foil is important to improve the ship performance. The study aims to compare 2 optimization methods, namely Genetic Algorithm (GA) and Particle Swarm (PSO) produce maximum lift and minimum drag. The designed foil is Joukowsky type with angle of attack 2°, 4°, 6°, 8°, 10° and 150,000 Reynold numbers. The method is used to state initial geometry of Joukowsky foil, iterated JavaFoil, optimized by GA and PSO using GNU Octave software. They are compared foil produces the maximum optimum C_L/C_D value. Validation of lift coefficient (C_L) and drag (C_D) Airfoil S9000 is almost similiar with the Williamson experiment. Stall is at angle of attack 9.3^o, with C_L error of 3.66%, C_D error of 6.83%. It results in the allowed tolerances are below 9.5%. Validation with JavaFoil Solver shows accurate results, with error of 4.15%. GA optimization produces C_L/C_D averages of 41,777 so it grows 6,544%. Whereas the C_L/C_D average PSO is 36,197, improve of 4,793%. GA optimization has a fairly high convergence speed at the first, then followed by stagnation process, the resulted solution improves significantly. PSO is focused on local search so convergence is difficult to achieve. The recommended foil optimization is to use GA.

Keywords: Drag, GA, Hydrofoil, Lift, PSO