Roy J. Hartfield

Aerospace design optimization using a steady state real-coded genetic algorithm

Abstract This study demonstrates the advantages of using a real coded genetic algorithm (GA) for aerospace engineering design applications. The GA developed for this study runs steady state, meaning that after every function evaluation the worst performer is determined and that worst performer is then thrown out and replaced by a new member that has been evaluated. The new member is produced by mating two successful parents through a crossover routine, and then mutating that new member. For this study three different preliminary design studies were conducted using both a binary and a real coded GA including a single stage solid propellant missile systems design, a two stage solid propellant missile systems design and a single stage liquid propellant missile systems design.

Wind Turbine Airfoil Performance Optimization using the Vortex Lattice Method and a Genetic Algorithm

This paper examines the viability of using the combination of the vortex lattice method for aerodynamic performance prediction with a genetic algorithm for the optimization of the aerodynamic performance of horizontal axis wind turbine blades. The work described in this paper includes the adaptation of a vortex lattice code designed to predict propeller performance to wind turbine performance prediction and the optimization process including results for both single point and multipoint design optimization efforts. Background The economics of deploying large wind turbine farms as a substantial source of electrical power is driven in large part by the efficiency of power conversion from wind energy to rotational mechanical energy. In the 1920’s, Betz formulated the basic analysis for the limiting case for horizontal axis wind turbine efficiency and set up the guidelines for how windturbine efficiencies should be calculated. A modern explanation of the Betz analysis can be found in Reference 1. In the 1930’s Glauert applied classical aerodynamic methods to airplane propeller designs in an effort to optimize performance of the horizontal axis machine for propulsion. 2 In the 1970’s and 1980’s blade element and momentum theory models were developed and refined for modeling wind turbine performance 3-5 and the efficiency of these models made them ideally suited for the genetic algorithm optimization work performed in the 1990’s by Selig et al. 6 This method for optimizing horizontal axis wind turbines using genetic algorithms used an improved version of the momentum theory models and demonstrated a successful and efficient optimization strategy. Hampsey 1 improved upon the optimization efforts by using a B-spline approach for modeling the blades along with a panel method for performance prediction. In this important optimization using a relatively higher order method for aerodynamic prediction, the blade geometries were not modeled using traditional airfoil shape theory. In the development of performance prediction for sails, it has been shown that the prediction of aerodynamic loads using the vortex lattice method is often much more accurate than load predictions based on simpler momentum theory methods. As with the panel methods, computational efficiency necessary for optimization can be maintained with the vortex lattice method. 7 A vortex lattice method has been applied to ship propellers 8 and to airplane propellers and has been shown to accurately predict performance in appropriate applications. 9,10 The airplane propeller analysis reported in

Aerodynamic Loads over Arbitrary Bodies by Method of Integrated Circulation

A novel numerical method for the conversion of unstructured surface vorticity into directional integrated circulation has been presented. This method allows for the reduction of vorticity distributions on arbitrary three-dimensional bodies into their mathematically equivalent formulations in the Prandtl lifting-line theory. A discussion of its implementation into a practical numerical solver has been presented. This new solver was applied to simple wing shapes as well as complex aircraft geometries to demonstrate the applicability of such an approach to flow solvers. It was shown that the reduction to lifting-line theory was exact for simple shapes such as rectangular and elliptical wings. For complex aircraft geometries, such as leading-edge root extension and canard-equipped aircraft, it was shown that the method of integrated circulation allowed for accurate prediction of aircraft loads and the analysis of flow physics, in keeping with the Prandtl original concepts.

Optimization of hypersonic aircraft using genetic algorithms

Abstract A methodology has been developed to take advantage of hypersonic aircraft design characteristics to allow design optimizations using computational fluid dynamics and genetic algorithms. This methodology highlights the inherent advantages obtained from decoupling and then adding via explicit formulations, the viscous, thermal, species-transport and combustion physics into the implicitly solved Euler formulations of fluid flow. The resulting formulation is found to be robust and less sensitive to volume mesh refinement, thereby making it possible to use an automated mesh generation process. The mesh generator also takes advantage of the inherent two-dimensionality of the fluid flow around hypersonic aircraft and uses block-structured rhombohedrum meshes that are easy to control using an optimizer. A sample case study is presented in the form of the X-43 hypersonic aircraft which is then optimized for a Mach 6.5 flight using this new methodology.

Design, Testing and Optimization of a Constant Torque Propeller

This paper examines the viability of a passive variable pitch propeller for small aerial Remotely Piloted Vehicles (RPV’s) and Unmanned Aerial Systems (UAS’s) using a self adjusting mechanical pitch control mechanism. The work described in this paper includes the design and testing of a mechanical pitch change mechanism based on a constant torque spring for an off-the-shelf propeller. A propeller performance prediction code, based on the vortex lattice method, in combination with a genetic algorithm (GA) is used to optimize a propeller for given motor performance data. The optimization process includes results for both single point and multipoint design optimization efforts.

Preliminary Design Drag Calculation Using Advanced Paneling Schemes

An advanced, modular paneling scheme has been developed for atmospheric flight vehicles with an aim of predicting the drag during subsonic flight. A generic, three-dimensional grid generation algorithm has been developed for this effort that allows the construction of detailed aircraft geometries using the concept of geometric overlap in three-dimensional spaces. An automated grid refinement system has been developed that is fully autonomous in operation and converts the user defined grid into the refined grid by trimming the geometrical internal overlaps and generation of high fidelity surface grids over the exposed outer surfaces of the design airframe. A group of aerodynamic predictive code based on modified potential theory and semiempirical models has been created to determine the aerodynamic loads on the three dimensional airframe. The use of three-dimensional source distributions under the exposed surfaces of the fuselage structures and the use of vortex ring elements placed over quadrilateral panels deployed on the mean lifting surfaces provide the detailed distributions of pressure and velocity over the airframe surfaces. The use of relaxed wake methods has been used to investigate the development of fully defined wakes behind complex aircraft geometries and their effects on the fuselage is available for user investigation. Models have also been added for the determination of skin friction over the exposed surfaces using panel interaction schemes. The results for this preliminary design level drag prediction code have been found to be extremely encouraging in their compatibility to the experimental and flight test data.

Computational fluid imaging for iodine fluorescence in compressible flows

Computational Fluid Imaging (CFI) is a new field developing out of computational fluid dynamics (CFD). The effort described herein is the development and validation of a scheme for numerically imaging planar laser-induced iodine fluorescence in a compressible flow. The pressure and temperature required for the calculation of the fluorescence intensity are computed using a SPARK three-dimensional Reynolds averaged Navier-Stokes code developed at NASA Langley. The details of the fluorescence model are included herein and a comparison between a calculated image and an experimentally acquired image is shown.

Optimization of Combined Rocket and Ramjet/Scramjet Ballistic Missile Designs

A preliminary design optimization of a revolutionary concept involving integrated airbreathing propulsive assist is presented. The conceptual design for the proposed airbreathing thrust augmenter is an annular ramjet/scramjet integrated into a liquid-oxygen-rocket propellant rocket core. The airbreathing engine makes use of the same tankage and turbopump hardware as the rocket engine for supplying fuel. The aeropropulsive environment is simulated using an axisymmetric Euler solver with a decoupled semiempirical viscous model. The flight performance is simulated using a seventh- to eight-order Runge–Kutta integrator for the six-degree-of-freedom dynamics. The optimizer used for design iterations is a binary-encoded genetic algorithm with a proven track record for addressing complex flight vehicle design problems. Results show the marked improvement in performance achieved by the use of such thrust augmenters on a military-dominated niche of the overall spectrum of missile designs.

Missile trajectory optimization using a modified ant colony algorithm

A modified ant colony optimization (ACO) algorithm is applied to a single-stage solid missile design problem involving six degrees of freeedom (6-DOF) flight trajectory modeling. A local search procedure is also integrated with the algorithm, adding a search intensification ability that compliments the ability of ACO to thoroughly explore a solution space. The goal of this work is to evaluate the effectiveness of the ant colony optimization scheme by comparing its solution output quality with those of other, well-known optimization methods. Performance is based on solution “fitness”, or how closely each solution matches a specific set of performance objectives, as well as the number of calls to the objective function that are required in order to reach that solution. Additionally, an important performance criterion is to determine each algorithm0s capabilities of finding, not only a single quality solution to a design problem, but also a diverse set of additional, near-optimal solutions. The results of this study demonstrate that the modified ACO algorithm is a viable candidate as a high-performance optimization method, while the added local search method may or may not be a worthwile addition for the attempted optimization problem.

Aerospace design optimization using a compound repulsive particle swarm

This study shows the benefits of using a compound repulsive particle swarm optimizer (RPSO) for use in aerospace design applications. A new hybrid optimizer has been developed for this study involving a separate particle swarm imbedded within the standard RPSO. The engineering design problems for this study include the development of a desired solid rocket motor grain, a solid motor sounding rocket, and a single stage solid propellant missile system. The methods are used independently to match a prescribed set of parameters defining the propellant characteristics and/or the missiles geometry. The design study was conducted using both the standard RPSO and the newly developed compound RPSO. The algorithms are compared based on their speed and effectiveness in solving the design problem, with the figure of merit being a factor based on a set of desired performance goals.