AMERICAN INSTITUTE OF AERONAUTICS AND ASTRONAUTICS
1
INTEGRATED WING DESIGN WITH THREE DISCIPLINES
1
1
Copyright
©
2002 by NRC. Published by AIAA with permission. G. Shi, G. Renaud and X. Yang Structures, Materials and Propulsion Laboratory Institute for Aerospace Research National Research Council Canada Ottawa, Canada K1A 0R6 F. Zhang and S. Chen Aerodynamics Laboratory Institute for Aerospace Research National Research Council Canada Ottawa, Canada K1A 0R6
ABSTRACT
A wing design process, coupling the three disciplines of structures, aerodynamics and aeroelasticity, was successfully performed. This paper presents the multidisciplinary design procedure with emphasis on structural modeling and optimization. The linking and coupling techniques for the multidisciplinary integration is summarized and the methodology for automatic conversion of the aerodynamic wing shape and pressure distribution to the structural finite element model is presented. Furthermore, the automatic model generation used for structural optimization in the integrated design loop is described. The results obtained show that the developed techniques work well and that more complicated MDO operations can be undertaken.
INTRODUCTION
Multidisciplinary Design and Optimization (MDO) has gained wide acceptance in the aerospace industry. The increasing interest in this methodology is due to the complexity of aerospace systems, which requires efficient coordination of various disciplinary analysis capabilities and effective communication among potentially geographically separated teams. The design departments in aerospace industry are often strongly segregated by disciplines, such as structures, aerodynamics and aeroelasticity. Each department is only responsible for specific aspects of the engineering work required for designing an aircraft. In each discipline, specific discipline-driven design techniques are developed and used, for instance, FEM for structural analysis and CFD for aerodynamic analysis. In order to develop an integrated aircraft design approach, the coupling of multiple disciplines and the linkage of distributed design teams are two important issues that must be addressed in MDO. The Institute for Aerospace Research of the National Research Council of Canada launched a project for developing MDO strategies for aerospace systems integrating structures, aerodynamics and aeroelasticity. At this stage, the research activities focused on the development of linking and coupling techniques. This paper presents the initial results of a preliminary wing design by integration of the three disciplines. In order to provide more information on this research work, this paper focuses on the structural aspects and coupling techniques. Another paper provides more details on aerodynamic modeling and optimazition
1
.
MULTIDISCIPLINARY DESIGN CYCLE
The objective of the multidisciplinary design process presented in this paper was to find the lightest wing box design for certain flight conditions that satisfies predefined geometry, stress, flutter, and displacement requirements. The loads acting on the wing box were determined from CFD analyses and depend on the structural weight. Figure 1 illustrates the M6 swept wing geometry and flight parameters used in the design process. The integrated design process included three coupled disciplines. First, the aerodynamics discipline evaluated the pressure distribution on the surface of the wing. Second, the structures discipline calculated the wing box stresses and deformations resulting from the air pressure. Third, the aeroelasticity discipline
9th AIAA/ISSMO Symposium on Multidisciplinary Analysis and Optimization4-6 September 2002, Atlanta, Georgia
AIAA 2002-5405
Copyright © 2002 by the author(s). Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
AMERICAN INSTITUTE OF AERONAUTICS AND ASTRONAUTICS
2 predicted the flutter speed with data from the other two disciplines. Figure 1. Wing characteristics. Two formulations were implemented for the integrated design process. The first case involved structural optimization with aerodynamics and aeroelasticity analysis. The second case involved both structural and aerodynamics optimization, with aeroelasticity analysis. Both cases included an initial step for determination of the initial weight and pressure distribution corresponding to the initial design. The actual process of system optimization comprised three steps: 1) Minimization of the wing box weight, under stress, deformation, and flutter constraints. 2) Calculation of wing drag. If aerodynamics optimization is performed, the wing shape is modified to minimize the drag. 3) Evaluation of process convergence. These three steps were repeated until the process convergence met a certain tolerance criteria. The multidisciplinary design process is illustrated in Figure 2 and described in the next paragraphs. Figure 2. Multidisciplinary design loop.
Step 1: Structural and aeroelastic analysis and optimization
A finite element model of the wing box was created after each aerodynamics analysis to include the changes in pressure loads acting on the structural members. Furthermore, the wing shape changes were taken into account when aerodynamics optimization was present. The wing box structural displacements and stresses caused by the aerodynamics loads were calculated through a finite element procedure. This analysis was coupled to an optimization algorithm that interpreted the structural response and modified the wing box design to minimize the weight under predetermined constraints. This procedure was repeated until the lightest structure with acceptable stress and deformation under the aerodynamics loads was determined. Aeroelasticity analyses were performed in parallel to the static structural analysis. No optimization work was involved in this discipline and only flutter speed calculations were carried out. The flutter speed was used as a constraint in the structural optimization. The objective of this constraint was to ensure that the flight speed did not reach the critical flutter speed.
Step 2: Aerodynamics analysis/optimization
Assuming that the wing shape was fixed (Case 1), the aerodynamics analysis was limited to determining the air pressure needed to maintain flight. To accomplish this, the angle of attack was adjusted at each aerodynamic step to ensure that the total lift was equal to the aircraft weight. In the present case, the aerodynamic analysis was conducted using the CFD solver KTRAN
2
. In case 2, a drag minimization was carried out for the ONERA M6 isolated wing, represented by seven sections in the spanwise direction. For simplicity, the original M6 wing planform was not altered and the optimization procedure only modified the section shapes. The free stream Mach number was set to 0.84 and the lift was kept equal to the weight of the airplane. The angle of attack was allowed to vary during the course of the optimization process. To obtain a realistic wing, geometry constraints forced the maximum wing thickness to be greater than or equal to 8% of its corresponding chord and the trailing edge angle to lie between 5 and 20 degrees, at each section. More details on aerodynamic analysis and optimization can be found in the reference
1
.
Optimization
Converged?Yes No
Structural &AeroelasticityAnalysis(Figure 5)Structural OptimizationAerodynamicsAnalysis/Optimization
Converged?Yes NoPressureGeomerty
Weight
3 1 7 . 3 i n 1 3 3 . 3 i n
471.0 inWing geo.: M6 swept wing Max thick. at root: 31.5 inSweep angle: = 26.7
Λ
25%O
1 7 8 . 7 i n
Mach number: M = 0.84Taper ratio: = 0.56
λ
Aspect ratio: A = 3.8
AMERICAN INSTITUTE OF AERONAUTICS AND ASTRONAUTICS
3
Step 3: Convergence evaluation.
The total weight of the aircraft was used for convergence purposes. This weight value was calculated using the optimized wing box obtained in step 1. For the results presented in this paper, a relative weight change smaller than 0.01% was required between two subsequent iterations as the criterion of termination.
WING BOX MODEL
A medium-complexity wing box model was assumed, composed of two skins, three spars, eleven ribs, and four stringers. A schematic of the wing box, without its upper skin, is shown in Figure 3. Figure 3. Wing box model (upper skin not shown). The structural analysis model consisted of 309 elements. Different element types and materials were used for the various structural components. The characteristics of the finite elements and materials composing the wing box are listed in Tables 1 and 2.
Component Element type Material Number of elements
Upper skin Shell Aluminum 40 Lower skin Shell Aluminum 40 Rib webs Shell Aluminum 44 Spar webs Shell Aluminum 30 Rib posts Beam Steel 55 Spar beams, Stringers Beam Steel 100 Table 1. FE properties of the wing box model
.
Material E
(Msi)
ν νν ν
σσσσ
y,t
(Ksi)
σσσσ
y,c
(Ksi)
σσσσ
y,s
(Ksi)
Aluminum 10.5 0.3 43 39 23 Steel 30.0 0.3 75 60 - Table 2. Material properties. The five columns of Table 2 correspond to the Young’s modulus, the Poisson’s ratio and to the tensile, compressive, and shear yield stresses, respectively.
AUTOMATIC MODEL GENERATION (COUPLING TECHNIQUES)
One of the most challenging aspects of multidisciplinary wing design and optimization is the sharing of disciplinary responses between the various analysis codes
3
. In the wing box case, all structural loads came from aerodynamic analysis. The quality of data transfer therefore directly affected the quality of the system analysis, and hence the final optimal design. The major difficulty came from the fact that the FE and CFD grids were different. The exchange of information from aerodynamics to structures was done during the finite element model generation. The data transmission was performed using ASCII files containing the linking information. Two types of coupling were considered in the present study. The first one concerned the outer wing geometry. It was assumed that the wing shape was controlled by aerodynamic performance and that it could be modified at any time outside the structures discipline operations. The structural geometric and finite element models therefore had to be updateable at the initialization phase of the structural optimization step. The second type of coupling concerned the aerodynamics pressure loads calculated by CFD that had to be converted to finite element nodal forces. Conceptually, the aerodynamics-to-structure coupling involved the first and fourth steps of the five-step automatic model generation procedure illustrated in Figure 4. The first three steps were optional if no aerodynamics optimization was performed. In that case, since the geometry was assumed fixed, the geometry and finite element model could be saved and re-loaded when needed. However, since the applied loads changed after each CFD analysis, step 4) needed to be performed before the start of each structural optimization. It should be noted that all finite element pre-processing was accomplished using PATRAN
4
. 1)
Outer wing geometry import
The aerodynamics-discipline wing shape was stored in a predefined data file as a collection of points located in three dimensions. For the problem presented in this paper, 11 equally spaced sections defined by 204 points were used. A function written in PATRAN Command Language (PCL) read this file and imported the points into the PATRAN database as
Ihre Neugier belohnen
Alles, was Sie lesen wollen.
Jederzeit. Überall. Auf jedem Gerät.
Keine Verpflichtung. Jederzeit kündbar.