LAUSR

Abstract A meta model was developed to evaluate the performance of electrothermal anti-icing systems for a rotorcraft engine air intake. A reduced-order model based on proper orthogonal decomposition and a general regression neural network was employed to build the meta model, which covered the entire envelope of icing conditions. A compressible Navier-Stokes-Fourier computational fluid dynamics code was used to simulate the three-dimensional airflow around the rotorcraft engine air intake. High-fidelity droplet trajectory and ice accretion codes were also used to calculate the collection efficiencies and ice accretions on the anti-icing surface of the engine air intake in the heat-on mode. By considering airfoil and rotorcraft engine air intake cases, a feasibility study was then conducted to assess the meta model, using the leave-one-out cross validation method. The module was updated using a self-organizing map that contained k-mean clustering for the new samples in the parametric space. Lastly, the performance of the electrothermal anti-icing system was evaluated to demonstrate the practical use of the meta model. The distributions of ice thickness and icing parameters on the anti-icing surfaces can be used as guidelines for determining the required level of heater power and the size of heat pads.