Multidisciplinary Design Optimization in the Conceptual Design Phase

Abstract

Traditionally, the aircraft design process is divided into three phases: conceptual, preliminary and detailed design. In each subsequent phase, the fidelity of the analysis tools increases and more and more details of the design geometry are frozen. In each phase a number of design variants is generated, fully analyzing them with the tools available, and then doing trade studies between important design variables to finally choose the best variant. In the past, this approach has shown good results for 'Kansas city' type aircraft, which could be decomposed into different airframe parts with distinct functions, such as wings, tail, engines and fuselage. Each part needs to fullfill its own set of requirements and could be designed and optimized relatively independently from the others. For the new generation of large transport aircraft, such as the Blended Wing Body (BWB), the traditional design approach is less suited. The Blended Wing-Body - studied by Boeing and many others as a future long-haul transport aircraft concept - is characterized by an integrated airframe, in which the aforementioned parts can no longer be clearly distinguished. The Blended Wing-Body features many and strong interactions between the various design disciplines and airframe subparts. Using the traditional design doctrine, these interactions greatly increase the required time to design. Over the past years,Multidisciplinary Design Optimization (MDO) is being considered as an alternative. Nowadays, in industry the MDO approach is mainly used in the detail design phase and for isolated, well-defined design cases. The goal of this project is to create an MDO framework which can aid the designer in optimizing entire aircraft designs in the conceptual phase. This framework is shaped to the Bi-level Integrated System Synthesis (BLISS) strategy. This strategy splits the optimization into two levels: a disciplinary level, and a system one. Before optimization, BLISS performs a sensitivity analysis to obtain linearized global sensitivities of the design objective and constraints to each of the design variables. Validation is done using three cases: two sample problems from literature with known solutions, and the optimization of a simplified Boeing 747 wing for maximum aerodynamic efficiency using an aerodynamic and structural model. All three cases were optimized succesfully. Finally, as a proof-of-concept for MDO, the framework is required to find an conceptual design of the Blended Wing-Body with minimum structural weight and minimum drag across a given mission. Meanwhile, structural, aerodynamic and performance constraints had to be satisfied. The problem features 5 disciplines, 93 constraints, 110 states and in total 92 design variables. Again, BLISS could converge to a solution, requiring 4 hours per cycle. By tuning the design variables, BLISS managed to converge to a final design in 22 cycles. The final design satisfies all constraints, except for the large local Mach number on the outboard wing. Similar problems were identified in several other Blended Wing-Body studies. The results support BLISS as a viable candidate method for introducing MDO in the conceptual design practice.