Hybrid of non-dominated sorting genetic algorithm II and invasive weed optimization for fan duct surface heat exchanger configuration optimization design

In this research, configuration optimization was conducted for FDSHX (fan duct surface heat exchanger) that is a new type of heat exchanger adopted in aero-engine heat management in recent years. Firstly, a method of Taguchi-ANFIS (Taguchi-Adaptive Neuro-Fuzzy Inference System), which can reduce training data volume utilizing orthogonal experimental matrix, was proposed to construct a rapid response mathematical model between three configuration parameters, including fin pitch, fin thickness, and oil passages number, and two design indicators, including heat transfer capacity and operating weight, based on the data prepared by an experimental validated FDSHX heat transfer capacity calculation method using heat transfer unit simulation. Secondly, a hybrid method of NSGA II-IWO (Non-dominated Sorting Genetic Algorithm II-Invasive Weed Optimization), which can simultaneously obtain the strong abilities of exploration and escaping from trap into local optima, was proposed to drive the constructed response mathematical model to enhance heat transfer through adjusting the three configuration parameters. Optimization comparison was performed between NSGA II-IWO and four classic optimization algorithms including GA (Genetic Algorithm), IWO (Invasive Weed Optimization), CA (Cultural Algorithm), and HS (Harmony Search). This research provides an integrated solution to help mechanical engineers improve the applicability and reasonableness of FDSHX design.