Multidisciplinary Design Optimization Techniques for Branching Trajectories

Abstract

Fully reusable two-stage-to-orbit vehicle designs that incorporate ‘branching’ trajectories during their ascent are of current interest in the advanced launch vehicle design community. Unlike expendable vehicle designs, the booster of a reusable system must fly to a designated landing site after staging. Therefore, both the booster return branch and the orbital upper stage branch along with the lower ascent trajectory are of interest after the staging point and must be simultaneously optimized in order to achieve an overall system objective. Current and notable designs in this class include the U. S. Air Force Space Operations Vehicle designs with their ‘pop-up’ trajectories, the Kelly Astroliner, the Kistler K-1, one of the preliminary designs for NASA’s Bantam-X study, and NASA’s proposed liquid flyback booster designs (Space Shuttle solid booster upgrade). The solution to this problem using an industrystandard trajectory optimization code (POST) typically requires at least two separate computer jobs — one for the orbital branch from the ground to orbit and one for the booster branch from the staging point to the landing site. In some cases, three computer jobs may be desired: one from launch to staging, one for the upper stage, and one for the booster flyback. However, these jobs are tightly coupled and their data requirements are interdependent. This paper expounds upon the research necessary to improve the accuracy, computational efficiency, and data consistency with which the branching trajectory problem can be solved. In particular, the proposed methods originate from the field of Multidisciplinary Design Optimization (MDO).