Optimization of centrifugal fan design using genetic algorithm and cst method

Abstract

Centrifugal fans, integral components in constrained spaces such as civil aviation ovens, are designed to meet stringent performance requirements. Their unique operational mechanism involves drawing air axially, redirecting it by 90 degrees, and passing it through the blades in a spiral trajectory. This redirection, combined with the application of centrifugal forces, significantly influences airflow and contributes to their ability to generate high total pressure. Consequently, centrifugal fans are ideal for high-pressure applications where reliable and efficient airflow is paramount. Centrifugal fans possess different characteristics compared to axial fans. Initially, the airflow begins in an axial direction, but before reaching the blades, it undergoes a 90-degree change in direction. Within these fans, the blades guide the air in a spiral pattern, which combines with centrifugal force. This centrifugal force drives the airflow outward and determines its direction. For forward-curved blades, the change in airflow direction has a significant impact on performance. Centrifugal fans generate higher total pressure, and the lift force of the blade profile plays a limited role. Backward-curved blades do not produce traditional lift forces. Their convex side acts as the pressure side, which is an unconventional feature for blade profiles. Additionally, a blunt leading-edge design enhances efficiency and strengthens structural integrity. Centrifugal fans are notable for their high efficiency and relatively low noise levels. Thanks to their robust structural strength, they can withstand pressures up to 7,500 Pa while maintaining stable performance. Commonly used in clean air, gas, and ventilation systems, they are also highly effective in systems with variable resistance. The transitions between blades are gradually widened. In most cases, blades are manufactured as hollow steel profiles welded to the backplate and shroud. In some cases, the fan wheel may also be fabricated using aluminum casting or CNC machining methods. It is well known that the performance of centrifugal fans is largely dependent on blade shape. In this context, Bezier curves, B-spline curves, and NURBS curves are frequently employed for the parametric design of fan blades. However, with the increasing demand for high-speed or curved blade shapes, lower-order curves often fail to adequately define these designs. The Class-Shape Transformation (CST) method, proposed by Kulfan, is a highly accurate modeling approach capable of achieving precise shape designs with fewer variables. This method is widely applied across various aerospace fields, such as blade profile optimization, compressor blade design, and aircraft design, and is also beneficial for the parametric design of fan blades. The research leverages the CST method to parameterize the fan geometry. The CST method is chosen for its capability to accurately describe complex geometrical shapes using a reduced number of variables, thereby simplifying the optimization process without compromising the fidelity of the design. By integrating CST with Genetic Algorithms (GA), this study seeks to explore a broader design space efficiently and uncover optimal configurations that maximize fan performance. A critical component of the optimization process is the use of advanced numerical simulations. ANSYS Fluent, a robust computational fluid dynamics (CFD) tool, was employed to conduct fluid flow analyses and gather performance data for various design iterations. These simulations provided insights into the aerodynamic behavior of the fan, highlighting the impact of geometric changes on performance metrics such as total pressure, efficiency, and flow rate. To facilitate an efficient optimization workflow, a communication loop was established between MATLAB's Genetic Algorithm Toolbox and ANSYS Fluent. This loop enabled seamless data exchange and automated the iterative process of design evaluation and optimization. Key parameters, such as blade angle, chord length, and hub-to-tip ratio, were defined and systematically varied within predefined limits to explore their influence on fan performance. In this study, the blade design of a two-dimensional centrifugal fan was optimized through an integrated computational approach. Numerical simulations were performed using ANSYS software, with the necessary calculations to ensure robust analysis. Fan performance was evaluated based on Fluent CFD analysis, with the results systematically recorded and processed in Excel. The optimization process employed a GA implemented through MATLAB's built-in libraries. Parametric blade optimization was achieved using the CST method, the k-epsilon turbulence model, and the GA, forming an iterative computational loop between MATLAB and ANSYS. Following optimization, the blade designs were analyzed to identify the configuration with the most favorable parameters. The study's findings underscore the significance of geometric parameterization and optimization in enhancing the performance of centrifugal fans. By systematically exploring the design space, the research achieved improvements in performance metrics, meeting the stringent requirements of civil aviation applications. The integration of the CST method and GA proved to be a powerful combination, offering a structured and efficient approach to fan design optimization. In conclusion, this study provides a comprehensive framework for optimizing centrifugal fan designs, bridging the gap between traditional methodologies and modern computational tools. The results not only contribute to the field of fan design but also demonstrate the potential of advanced parameterization and optimization techniques in achieving superior performance in constrained engineering applications.