Cross-sectional Design of Composite Rotor Blades Considering Manufacturing Constraints

摘要

In the design process, a cross-section of a composite rotor blade is required to possess some specified properties dictated by several often conflicting objectives that include performance, safety, and maintenance. These top-level objectives can be translated into specific requirements for sectional stiffnesses and moments of inertia, stress levels, and locations of shear and/or mass center, etc. Due to the complexity and limitations of computing capability, the designs are usually modeled as continuous-variable optimization problems without consideration of manufacturing constraints, such as values of discrete ply angles. As a result, the optimal designs obtained by an optimization approach often cannot be implemented in reality. In order to manufacture it, designers will naturally round the optimal solution to the closest one that satisfies the manufacturing constraints. However, this will risk violating the structural requirements. In this paper, manufacturing constraints are included in the optimization model. A two-phase method is developed to deal with this mixed continuous and discrete variable optimization problem. An efficient algorithm, sequential quadratic programming (SQP), is used to seek the optimal continuous solution from the relative large design space in the first stage. Manufacturing constraints are added in the second phase. The Genetic Algorithm (GA) is applied to search around the obtained optimal continuous solution for the optimal discrete solution. This two-step method is believed to be beneficial for designers, helping them to find an accepted cross-section layout faster than a one-step application of the GA. In addition to manufacturing constraints that can be explicitly specified, another important manufacturing consideration is related to the uncertainties associated with the manufacturing processes. To this end, the paper will present preliminary results regarding the effects of geometric imperfections on cross-sectional properties.