Exchange Inlet Optimization by Genetic Algorithm


A. Rocket Flowpath Variables

A. Rocket Flowpath Variables

B. Air Flowpath Variables

B. Air Flowpath Variables

C. Four Branch Design

C. Four Branch Design

D. Five Branch Design

D. Five Branch Design

E. Six Branch Design

E. Six Branch Design

F. Individuals Evaluated (Most Fit Circled)

F. Individuals Evaluated (Most Fit Circled)

G. Exchange Inlet Components

G. Exchange Inlet Components

G. Optimum Exchange Inlet

G. Optimum Exchange Inlet

 

 

 

 

 

 

 

The design of the Exchange Inlet is structured that one can set a significant number of variables to tailor the resulting shape to meet a variety of design objectives.  These variables, shown in Figs. A and B are independently set but their influence on each other is not easily quantifiable.  Therefore, the optimization problem lends itself to a method like a genetic algorithm.  Shown in Figs. C, D, and E are various individuals from a population that is analyzed to find the most fit individual (where the fitness of a particular design, or individual, is defined by a combination of entrained airflow, shear layer area, and contour).  As the genetic algorithm progresses (Fig. F) through generations (as indicated by the value of k at the bottom of each graph), some combinations of the variables create aphysical designs (hollow dots in Fig. F) but as shown by the pattern of solid dots in the graphs, the genetic algorithm spends most of its time evaluating designs which carry the two variables on each axis that have values equal to the eventual optimum (i.e., the solid dots form lines along values for 0.3 along the y axis and 0.9 along the x axis).