This paper will discuss the use of computational fluid dynamics (CFD) for the Space Launch System (SLS) program to model the ascent phase of flight. The ascent phase begins shortly after the vehicle clears the launch tower and extends to the first staging event. To model SLS’s ascent, over 1000 numerical solutions of the Navier–Stokes equations were solved, and this analysis has been repeated for five different SLS configurations. To manage this demanding ascent CFD task, the SLS program has developed the Computational Aerosciences Productivity & Execution software. The paper also discusses some of the ways that CFD and high-end computing have advanced in the last decade and offers some comparisons to CFD used in the Space Shuttle Program.

This paper discusses the validation of the ascent surface pressure database developed using Computational Fluid Dynamics (CFD) for the Space Launch System (SLS) Program with Artemis I flight data from the ascent phase of flight. The ascent phase begins shortly after the vehicle clears the launch tower and extends to the first staging event. This work compares sensor data from the successful Artemis I flight to point pressures extracted from the implemented surface pressure database for Artemis I and other available CFD data sets. The database is constructed such that the pressure at any location can be extracted a posteriori so that there is no need have the correct sensor locations while running the CFD. A final source of experimental validation is provided from steady pressure sensitive paint (PSP) data obtained during a force and moment test in the Unitary PlanWind Tunnel at NASA Ames Research Center. Most extracted surface pressures from CFD data sets match well with flight data and with the limited PSP data available. The differences that do exist are mostly caused by the way flight data was calibrated. The flight instrumentation had significant drift, so the flight data at each static pressure tap was calibrated to have a pressure coefficient of zero at liftoff and at another instance approximately two minutes into flight. However, many sensors, for example those on conical surfaces, are not expected to have a pressure coefficient of zero at high Mach numbers. Changing the calibration to the Newtonian aerodynamics pressure coefficient greatly improves the comparison to CFD. Several points remain where CFD and flight data still do not match, and these are in locations near the core stage engines, where the thermal environment and geometry were not modeled with high accuracy in this work. Overall, the Artemis I flight data strongly validates the continued use of this or similar CFD methods for launch vehicle surface pressure databases.

The Space Launch System (SLS) employs two Space Shuttle–derived solid rocket boosters, which separate from the SLS core while still experiencing appreciable aerodynamic loads. Creating an aerodynamic database for this phase of flight can be challenging due to the large number of independent variables needed to fully constrain the problem and the complex flow induced by exhaust plumes of the booster separation motors and core main engines impinging on other parts of the vehicle. This paper details recent efforts in generating aerodynamic data used to create databases for the SLS during the booster separation event using viscous computational fluid dynamics (CFD) simulations obtained using NASA’s FUN3D solver. Particular challenges faced when modeling a complex problem, such as booster separation, are presented. Reductions in interpolation error estimates were observed through the introduction of a physics-based covariance approach to building the CFD run matrix, eliminating infeasible booster location permutations that could potentially skew final response surfaces. Interpolation error control is shown using test cases outside of the main database. Code-to-code comparisons between the FUN3D and OVERFLOW solvers are also presented to further verify the results.

The successful launch of the Artemis-I mission in November 2022 was made possible, in part, by years of rigorous vehicle simulation and scaled testing. Correctly anticipating the complex physics of the booster separation event was one of many necessary challenges. The successful booster separation of Artemis-I yielded flight data with which the fidelity of these predictions could be assessed. In this paper, the flight data and flight simulations are reconciled to present a unified assessment of the booster separation event. With this assessment, predictive confidence can be reinforced in support of the crewed Artemis-II mission.

This paper will discuss the use of Computational Fluid Dynamics (CFD) for the Space Launch Sys- tem (SLS) Program to model the ascent phase of flight. The ascent phase begins shortly after the vehicle clears the launch tower and extends to the first staging event. To model SLS's ascent, over one thousand numerical solutions of the Navier-Stokes equations are solved, and this analysis has been repeated for five different SLS configurations. To manage this demanding ascent CFD task, the SLS program has developed the Computational Aerosciences Productivity & Execution (CAPE) software, which is introduced in the paper. The paper also discusses some of the ways that CFD and high-end computing have advanced in the last decade and offer some comparisons to CFD used in the Space Shuttle Program.

The Space Launch System Aerodynamics Task Team is responsible for delivering aerodynamic force and moment databases from liftoff through ascent until the rocket leaves the Earth’s atmosphere. The process for developing the ascent portion of this database is described in the current paper. The data used to develop the database were generated using a combination of wind tunnel testing and CFD simulations. The details of the wind tunnel testing performed at the NASA Ames Unitary Plan Wind Tunnel and CFD simulations performed using FUN3D at wind tunnel and flight conditions are discussed, and comparisons of these data sets are provided. The methods used for converting the source data into the final database response surfaces with corresponding uncertainty are also detailed.

Acomparisonofsurfacepressurecoefficientmeasurementsobtainedusingpressure-sensitive paint (PSP) measurements with predictions from the computational fluid dynamics (CFD) code FUN3D is presented for the NASA SLS Block 1B crew vehicle. Overall, the flow features over the SLS configuration were captured by both the PSP data and CFD data at freestream Mach numbers ( M ∞ ) of 0.8 and 1.3. Overall, the flow features over the SLS are captured by the PSP data but the intensities of large pressure gradients are less intense than what was predicted by the CFD data. Several examples of this observation are given including the flow interaction at the booster nose cone edge, core body, and forward booster attachment hardware at M ∞ = 0.8.

This paper presents a method for adjusting sectional loads to match target values for integrated force and moment coefficients. In a typical application, the sectional load profile for one flight condition is calculated from Computational Fluid Dynamics (CFD) while the integrated forces and moments are measured in a wind tunnel experiment. These two methods do not generally result in identical predictions, and this leads to an inherent inconsistency between different data products. This paper aims to provide a procedure to remove that inconsistency. A sectional load profile for a launch vehicle splits the rocket into slices along its length and calculates the aerodynamic loading on each slice, which leads to a one-dimensional aerodynamic load profile that is used for structural analysis. Adjusting sectional loads, also known as line loads, is a nontrivial matter due to several consistency constraints. For example, the adjusted sectional normal force profile must be consistent with both the integrated normal force and pitching moment. To avoid such inconsistency issues, this paper presents a method using a Proper Orthogonal Decomposition (POD) to generate basis functions to adjust the sectional load profiles. As a corollary, this correction method enables the creation of an uncertainty quantification for sectional loads that is consistent with the dispersed integrated force and moment database and its uncertainty quantification. Several extensions to this technique, such as applying the method to the surface pressures, are considered.

NASA's new Space Launch System (SLS) will be the first rocket since the Saturn V (1967-1973) to carry astronauts beyond low earth orbit-and will carry 10% more payload than Saturn V and three times the payload of the space shuttle. The SLS configuration consists of a center core and two solid rocket boosters that separate from the core as their fuel is exhausted two minutes after lift-off. During these first two minutes of flight, the vehicle powers its way through strong shock waves as it accelerates past the speed of sound, then pushes beyond strong aerodynamic loads at the maximum dynamic pressure, and is ultimately enveloped by gaseous plumes from the booster-separation motors. The SLS program relies on computational fluid dynamic (CFD) simulations to provide much of the data needed to build aerodynamic databases describing the structural load distribution, surface pressures, and aerodynamic forces on the vehicle.

The development of the aerodynamic database for the Space Launch System (SLS) booster separation environment has presented many challenges because of the complex physics of the ow around three independent bodies due to proximity e ects and jet inter- actions from the booster separation motors and the core stage engines. This aerodynamic environment is dicult to simulate in a wind tunnel experiment and also dicult to simu- late with computational uid dynamics. The database is further complicated by the high dimensionality of the independent variable space, which includes the orientation of the core stage, the relative positions and orientations of the solid rocket boosters, and the thrust lev- els of the various engines. Moreover, the clearance between the core stage and the boosters during the separation event is sensitive to the aerodynamic uncertainties of the database. This paper will present the development process for Version 3 of the SLS booster separa- tion aerodynamic database and the statistics-based uncertainty quanti cation process for the database.

This paper presents details of Computational Fluid Dynamic (CFD) simulations of the Space Launch System during solid-rocket booster separation using the Cart3D inviscid and Overflow viscous CFD codes. The discussion addresses the use of multiple data sources of computational aerodynamics, experimental aerodynamics, and trajectory simulations for this critical phase of flight. Comparisons are shown between Cart3D simulations and a wind tunnel test performed at NASA Langley Research Center's Unitary Plan Wind Tunnel, and further comparisons are shown between Cart3D and viscous Overflow solutions for the flight vehicle. The Space Launch System (SLS) is a new exploration-class launch vehicle currently in development that includes two Solid Rocket Boosters (SRBs) modified from Space Shuttle hardware. These SRBs must separate from the SLS core during a phase of flight where aerodynamic loads are nontrivial. The main challenges for creating a separation aerodynamic database are the large number of independent variables (including orientation of the core, relative position and orientation of the boosters, and rocket thrust levels) and the complex flow caused by exhaust plumes of the booster separation motors (BSMs), which are small rockets designed to push the boosters away from the core by firing partially in the direction opposite to the motion of the vehicle.

This paper presents details of Computational Fluid Dynamic modeling of the Space Launch System during ascent. The primary focus of the paper is the flow simulation of the vehicle during ascent using the Overflow Navier-Stokes code. Computations of 739 first-stage flight conditions covering a range of Mach numbers, angles of attack, and roll angles were computed. The overset grid system contained 375 million grid points, and over 28 mil- lion CPU hours were used in the simulations. The simulations were run on the Pleiades supercomputer at the NASA Advanced Supercomputer Center at Ames Research Center. The data products from this work include integrated line-loads, surface pressure coefficients, venting pressures, and protuberance air-loads. Detailed comparisons were made of the aerodynamic performance predicted by Overflow and the wind-tunnel derived aero- dynamic database. A small number of the cases were run with two different turbulence models and with two differencing schemes. These results were used to quantify the sensitivity to the choice of the turbulence model and to the differencing scheme. The paper also introduces an effort to use the inviscid, unstructured Cartesian solver Cart3D to compute the aerodynamics during booster separation. Adaptive mesh refinement is being used to enable accurate simulations of sixteen booster-separation-motor plumes. The use of this tool is explored in preparation for building a booster-separation aerodynamic database.

This paper presents details of Computational Fluid Dynamic (CFD) simulations of the Space Launch System during solid-rocket booster separation using the Cart3D inviscid code with comparisons to Overflow viscous CFD results and a wind tunnel test performed at NASA Langley Research Center's Unitary PlanWind Tunnel. The Space Launch System (SLS) launch vehicle includes two solid-rocket boosters that burn out before the primary core stage and thus must be discarded during the ascent trajectory. The main challenges for creating an aerodynamic database for this separation event are the large number of basis variables (including orientation of the core, relative position and orientation of the boosters, and rocket thrust levels) and the complex flow caused by the booster separation motors. The solid-rocket boosters are modified from their form when used with the Space Shuttle Launch Vehicle, which has a rich flight history. However, the differences between the SLS core and the Space Shuttle External Tank result in the boosters separating with much narrower clearances, and so reducing aerodynamic uncertainty is necessary to clear the integrated system for flight. This paper discusses an approach that has been developed to analyze about 6000 wind tunnel simulations and 5000 flight vehicle simulations using Cart3D in adaptive-meshing mode. In addition, a discussion is presented of Overflow viscous CFD runs used for uncertainty quantification. Finally, the article presents lessons learned and improvements that will be implemented in future separation databases.

Recent developments on the automation of the X-rays approach to hole-cutting in over- set grids is further improved. A fast method to compute an auxiliary wall-distance function used in providing a rst estimate of the hole boundary location is introduced. Subsequent iterations lead to automatically-created hole boundaries with a spatially-variable o set from the minimum hole. For each hole boundary location, an averaged cell attribute measure over all fringe points is used to quantify the compatibility between the fringe points and their respective donor cells. The sensitivity of aerodynamic loads to di erent hole boundary locations and cell attribute compatibilities is investigated using four test cases: an isolated re-entry capsule, a two-rocket con guration, the AIAA 4th Drag Prediction Workshop Common Research Model (CRM), and the D8 \Double Bubble subsonic aircraft. When best practices in hole boundary treatment are followed, only small variations in integrated loads and convergence rates are observed for different hole boundary locations.

Characterization of the launch abort system of the Multi-purpose Crew Vehicle (MPCV) for control design and accurate simulation has provided a significant challenge to aerodynamicists and design engineers. The design space of the launch abort vehicle (LAV) includes operational altitudes from ground level to approximately 300,000 feet, Mach numbers from 0-9, and peak dynamic pressure near 1300psf during transonic flight. Further complicating the characterization of the aerodynamics and the resultant vehicle controllability is the interaction of the vehicle flowfield with the plumes of the two solid propellant motors that provide attitude control and the main propulsive impulse for the LAV. These interactions are a function of flight parameters such as Mach number, altitude, dynamic pressure, vehicle attitude, as well as parameters relating to the operation of the motors themselves - either as a function of time for the AM, or as a result of the flight control system requests for control torque from the ACM. This paper discusses the computational aerodynamic modeling of the aerodynamic interaction caused by main abort motor and the attitude control motor of the MPCV LAV, showing the effects of these interactions on vehicle controllability.

The NASA Advanced Supercomputing (NAS) facility at Ames Research Center has enabled remarkable breakthroughs in the space agency's science and engineering missions. For 30 years, NAS experts have influenced the state of the art in high-performance computing and related technologies.