Friday, November 20, 2015

Building the Space Launch System Booster Separation Database

In this visualization, a shock wave (colored by pressure) is clearly shown at the front of the vehicle; farther back, booster separation-motor plumes are colored by Mach number. The simulation was run on the Pleiades supercomputer at the NASA Advanced Supercomputing facility at Ames Research Center.  The visualization was part of a NASA showcase of nearly 40 of the agency’s exciting computational achievements at SC15, the international supercomputing conference, Nov. 15-20, 2015 in Austin. Image Credit: Stuart Rogers, NASA/Ames; Ryan Rocha, University of California, Davis

NASA's new heavy-lift launch vehicle, the Space Launch System (SLS), will carry 15% more payload than Saturn V and three times the payload of the space shuttle, requiring innovative rocket design. The SLS configuration consists of a center core stage with four RS-25 engines and two solid rocket boosters (SRBs), which separate from the core as fuel is exhausted soon after liftoff. To help SLS design engineers understand how aerodynamic forces will affect the path of the SRBs away from the core stage during separation, researchers at NASA's Ames Research Center and the University of California, Davis, are running high-fidelity simulations of thousands of possible separation scenarios.

This project used CFD simulations with the inviscid Cart3D flow solver, developed at NASA's Ames Research Center, to compute the aerodynamic forces on the two solid-rocket boosters and the core. Additional terms for the database were built using the results from a wind-tunnel test and with the high-fidelity OVERFLOW CFD solver. The CFD calculations include simulating the aerodynamic effects of 22 different plumes firing during the separation, including 16 BSM plumes, the plumes from the four core-stage main engines, and two plumes from the boosters.

The aerodynamic data is a function of eight independent variables. These include three translation variables and two rotation variables to define the booster position and orientation relative to the core, the free-stream angle of attack and angle of side-slip of the core, and the relative thrust of the booster-separation motors. In addition, separate sets of computations were run to simulate the possible effects caused by one of the core-stage engines failing.

Over 22,000 Cart3D simulations were used to fill the parametric space describing the booster separation event. In addition, 390 OVERFLOW simulations were run to provide higher-fidelity data. Comparisons among the Cart3D results, the wind-tunnel data, and the OVERFLOW data showed that the Cart3D code could accurately predict the booster and core aerodynamics to within an acceptable margin.

The new database will be used by the Guidance, Navigation, and Control group at NASA's Marshall Space Flight Center to simulate the SLS booster-separation event. These simulations will be run to ensure that the boosters can separate successfully without recontacting the core under all possible flight conditions, in order to verify the design of the SLS booster-separation system.

All of the CFD simulations were run on the Pleiades supercomputer, primarily on the 24-core Intel Xeon “Haswell” nodes. Each of the Cart3D simulations required approximately 200 core hours, and the OVERFLOW runs required at least 20,000 core hours. A total of 12.7 million core hours were used to complete the computations. Most of these simulations were run over a two-month period, during which the project used 400 dedicated Haswell nodes (9,600 cores) and approximately 50 terabytes of short-term disk storage.


No comments:

Post a Comment