06. Code TNA_ Final MKL1.pdf

www.nasa.gov

The cornerstone of NASA’s current and future missions

Computational Aerosciences Branch Code TNA

Cetin Kiris & Shishir Pandya Branch Chief| Aerospace Engineer

Code TNA Computational Aerosciences Branch

NASA Advanced Supercomputing Division (NAS) NASA Ames Research Center ü Perform large-scale simulations for

• Human Launch : Space Launch System (SLS), Orion, Commercial Crew - Databases and induced environments for ascent and abort, staging, debris, plume impingement, retro-propulsion, launch environment ü Advance computational aerosciences predictive capabilities to support NASA’s current and future missions/needs/requirements

• Aerodynamic

• Aeroacoustics

• Aerostructures ü Work with NASA High End Computing Capability (HECC) experts to adapt our toolsets for the next generation computing platforms

Role of Computational Aerosciences

Multiphysics Prediction Capability for Launch Environment

11/3/22 ü Goal of this project was to develop a capability for high-fidelity full-scale time accurate simulations of the launch environment acoustics including the water-based sound suppression system.

ü Novel and provably stable high-order method for multiphase flow developed and integrated into the LAVA Cartesian Adaptive Mesh Refinement (AMR) framework.

Photographs of a test of the NASA Kennedy Space Center (KSC) water deluge system which is used to suppress the immense sound pressure levels felt by the vehicle and surrounding structures during ignition.

Visualization of previous LAVA SLS launch environment simulation without water deluge system.

Particles track the plume path and are colored by temperature (white hot, dark cool).

Surfaces are colored by pressure (red high, blue low).

Validation with the 5% Scale Model Acoustic Test (SMAT)

11/3/22

Photograph of the Scale Model Acoustic Test (SMAT)

“Ignition Overpressure Computational Fluid Dynamics Simulation of Scale Model Acoustic Test PC123-FA-HF-01“ Peter A. Liever February 2017

Computational geometry showing point probe locations underside of SMAT mobile launcher

Comparisons of numerical (blue) pressure signal and recorded (gray) pressure signal as a moving average and with the shaded region showing RMS.

multiphase simulations with plume & water

11/3/22

0.0 0.2 0.4 0.6 0.8 Time (s)

P

SI

G

LAVA Dry

LAVA Wet

Multiphase Simulations for Upcoming Artemis Flight at Launch Complex LC-39B

0.0 0.2 0.4 0.6 0.8 Time (s)

P

SI

G

LAVA Dry

LAVA Wet Dry

Wet

Movies:

Comparison of ignition overpressure wave for dry and wet case at two sensor locations on the vehicle.

Maximum gauge pressure felt on the underside of Mobile Launcher during the ignition of the SLS engines.

Volume rendering of mass fractions. Liquid water is colored blue. Plumes are colored yellow to purple and water vapor is white. Solid Rocket Boosters (SRBs) starting indicate time T=0

Aerodynamic Databases

1. SLS Block 1 Crew

• ~4000 ascent cases

• ~40k booster separation cases

(two analysis cycles)

2. SLS Block 1 Cargo (discontinued)

• ~1000 ascent cases

3. SLS Block 1B Crew

• ~1000 ascent cases

• ~2500 booster sep cases

• Next up: 10k more bsep

4. SLS Block 1B Cargo

1. ~2000 ascent (2 payload fairings)

5. SLS Block 2 Crew

• Next up: booster sep

• Ascent coming soon

6. SLS Block 2 Cargo

• No foreseeable tasking

Space Launch Systems (SLS) – Stage Separation

SLS stage separation simulations showing numerical Schlieren

Predicting Orion LAS Acoustics: Wind Tunnel ü Can Computational Fluid Dynamics (CFD) provide difference in vibration strength due to vehicle geometry difference with QM-1 test article? Yes, provided spatially varying expected delta-dB.

ü Can CFD predict relationship between vibration strength and trajectory point obtained from wind tunnel test: altitude, velocity, angle of attack? Yes, validated predictions with wind tunnel data.

ü Can CFD help reduce uncertainty in margins at high angle of attack? Yes

Volume rendering of pressure fluctuations p’ = p - <p>

Video Credit: Timothy Sandstrom

Wind tunnel model at transonic ascent abort with moderate angle of attack and side slip

Pressure on a cut plane through the nozzles for a transonic ascent abort scenario at high angle of attack (white is high, black is low)

Pad AbortQM-1 Ground Test

Supersonic Abort (Mach 1.2) Supersonic Abort (Mach 1.6)

Frequency (Hz)So un d Pr es su re Le ve l ( dB

-- Wind Tunnel Measurements

- LAVA Predictions

Flight Test Validation Pad Abort 1 flight test where Orion LAS accelerates from rest to 10x Earth’s gravity Video shows passive particles seeded at the nozzle colored by velocity magnitude:

white is fast, dark orange is slow

Video Credit: Timothy Sandstrom

So un d Pr es su re L ev el

-- Flight Test Measurements

- LAVA Predictions

- No Accel LAVA Predictions ü Can CFD identify difference in vibration strength due to vehicle acceleration and change in attitude? Yes, provided spatially varying expected delta-dB.

Frequency (Hz)

Shaded regions are approx. +/- 2 dB because of statistical uncertainty

Flight Test Validation

Ascent Abort 2 flight test where Orion LAS triggered at Mach 1.2

Video shows density on cut plane (red is low, blue is high), and axial velocity on the surface (black is low, white is high) ü Can CFD explain higher than expected vibration-levels recorded during AA-2 test? Yes, demonstrated that CFD predicts that vibrations do not reduce in strength as much as expected from empirical scaling laws.

ü LAVA team has made a big impact on the vehicle requirements for safety:

ü Prior to 2017, CFD predictions of acoustics were not used or trusted ü Now they help inform engineering decisions to ensure future astronaut’s safety

Supersonic Retro Propulsion (Mach 2.4)

Volume rendering of exhaust mass fraction where yellow is 100% and black is 1%.

Only the plumes on the left side of vehicle are rendered to elucidate the plume structure without obstruction

Supersonic Parachute/Fluid-Structure Interaction

Pseudocolor contours of Mach number shown on a cut plane through the center of the domain for the impermeable parachute in M = 2.2 flow and iso-contours of Q-criterion colored by Mach number ü Perform large-scale simulations for

• Sustained Atmospheric Flight : aerodynamic/aeroacoustic/aerostructural performance prediction, vehicle design, noise prediction, airframe/propulsion integration, safety analysis ü Advance computational aerosciences predictive capabilities to support NASA’s current and future missions/needs/requirements

• Aerodynamic

• Aeroacoustics

• Aerostructures ü Work with NASA High End Computing Capability (HECC) experts to adapt our toolsets for the next generation computing platforms

Role of Computational Aerosciences

• In-flight AirBOS images planned for Phase II flight test

• Capability developd with NASA Advanced Supercomputing (NAS) Viz team

• Mach number = 1.4, α = 2.05°

• Also direct comparison with tunnel model Schlierens

Computational Schlieren for X-59 Flight Test Support

X-59 Low Boom Flight Demonstrator Analysis ü LAVA Curvilinear is used to simulate numerous flight conditions and aircraft configurations for the X-59 Quiet Supersonic Aircraft program, generating databases of aerodynamic and noise predictions ü Refactorization of the LAVA Curvilinear flow solver has enabled near-field and far-field ground noise predictions to be generated with up to a 10X computational cost reduction ü A recently development space-marching routine coupled with the CFD solver has reduced the required CFD computational domain size, leading to simulation cost reduction

X-59 Low-Boom Flight Demonstrator aircraft Pressure disturbance contours shown, with wingtip vortices depicted in gray

Pressure disturbance contours on symmetry plane with sample nearfield boom signature

1818 18

Future and ongoing work NASA Commercial Supersonic Technology (CST) Jet Noise Prediction

Jet Noise Prediction With LAVA Wall Modeled Large Eddy Simulation (WMLES)

Isocontour of Q-criterion colored by vorticity magnitude and velocity on surface ü Only solver that currently supports NASA’s CST project with high-fidelity scale resolving simulations for jet noise prediction of complex geometries.

ü Understand and document strength and shortcoming's of methods used

Oblique shock

Prediction of Jet Noise with LAVA Wall-Modeled LES (WMLES)

• WMLES successfully applied to the prediction of jet noise within NASA CST

– Significant improvements in turnaround time (63x) compared to Hybrid

Reynolds Averaged Navier Stokes (RANS)/Large Eddy Simulation (LES)

– Improvements have enabled simulations previously not feasible

• A first of it’s kind database was generated utilizing WMLES within LAVA for NASA CST’s Prediction Uncertainty Reduction (PUR) technical challenge.

– A total of 20 high-fidelity scale-resolving simulations performed

– Each simulation takes around 24hr and generates 100 terabytes (TB) of data

– Excellent agreement with experiments within 1 decibels (dB)

(comparable to experimental rig-to-rig uncertainty) video credit: Timothy Sandstrom

Improved resolution range for far-field noise multi-stream heated jet jet surface interaction noise ü Increase predictive use of computational aerosciences capabilities for next generation vehicles

• The next frontier is to use scale resolving simulations to predict:

Sustained Atmospheric Flight

Unsteady loads and fatigue

Buffet and shock Boundary Layer (BL) interaction

Fan, jet, and airframe noise

Active flow control

Transonic Truss Braced Wing Simulations x/c

C P

0 0.2 0.4 0.6 0.8 1

LAVA SA-RC-QCR

LAVA SST-RC-QCR

Experiment Run 378

55% Span α ,deg.

-2 -1 0 1 2 3 4-0.2

0.2

0.4

0.6

0.8

1.2

Using SST, lowered difference in CL relative to experiment from 5.8% to 1.1% near the cruise condition.

Visualizing inflow / outflow of particle paths in and out of porous wall Boundary Conditions (BC) for Wind tunnel simulations.

• Ran various studies to develop best practices and gain an understanding of the sensitivities of the model.

– Grid adaption, grid refinement, component buildup, wall spacing, cavity / base correction, turbulence model & corrections, and time-step ramping, convective scheme discretization, starting conditions, and more.

• Refactorization of LAVA allowed for better and faster convergence and more accurate results

– 12 to 27 times Standard Billing Unit (SBU) reduction.

– Verified consistent results against original LAVA

• Presented paper (AIAA 2021-1531) at SciTech 2021 and received AIAA Applied Aerodynamics Best Paper Award.

Transonic Truss Braced Wing (TTBW) Simulations

• Refined meshing best practices for HRLES on the TTBW as well as performing Cartesian immersed boundary wall-modeled LES (WMLES) to investigate buffet.

• Will use these tools as well as transition model and more to work on scale resolving simulation to investigate high lift devices, deep stall, and aeroelasticity

Visualization of vorticity magnitude colored by u-velocity for a scale-resolving Hybrid RANS-LES (HRLES) simulation at 6 deg AoA

Investigated Buffet onset with Unsteady

RANS (URANS).

Simulation oscillations of shock caused by spanwise phenomena across several convective time units

WMLES on Cartesian grid to simulate buffet on TTBW

X

C p

0.4 0.6 0.8 1 1.2

-1.5

-1

-0.5

0.5

1.5 initial optimized

Design iteration

O p ti m a li ty

F e a s ib il it y

M e ri t fu n c ti o n

0 50 100 150 200 250 300

-8

-6

-4

-2

.0080

.0085

.0090

.0095

.0100

.0105

.0110

.0115

.0120 Optimality Feasibility Merit function

NASA TTBW: Conceptual Design and Aerodynamic Shape Optimization

A Transonic Truss-Braced Wing (TTBW) single-aisle aircraft configuration has been designed and optimized, which serves as a platform for evaluating tools and technologies for next-generation aircraft concepts:

1. Cruise-slotted (CS) airfoil TTBW (TTBW-CS)

2. Natural-laminar-flow (NLF) TTBW (TTBW-NLF)

3. Tail-cone thruster (TCT) TTBW (TTBW-TCT)

Optimized fully-turbulent TTBW configuration

Drag-minimization of a cruise-slotted infinite swept wing to determine best practices for TTBW-CS optimizationsBaseline TTBW-CS configuration for high-fidelity shape optimization

Scale resolving simulations: Certification and Qualification by Analysis

LARGE EDDY SIMULATIONS

(WMLES)

DELAYED DETACHED EDDY

SIMULATIONS (DDES)

REYNOLDS AVERAGED NAVIER

STOKES (RANS)

1980s - Present 1990s - Present 2020 - Present

• PHILOSOPHY: Solve for averaged solution by modelling all fluctuations

• Highly matured over the past few decades

• Extensively used in industry

• PHILOSOPHY: Model all fluctuations except for those in separated regions

• Incremental changes requires to existing RANS tools

• High maturity level – very popular in industry

• PHILOSOPHY: Resolve more than 95% of fluctuation energy and model the rest

• Ground up changes required over RANS frameworks and algorithms

• Limited to academic problems until 2020

FLOW SOLUTION:

Single snapshot (time averaged) 50mm-70mm spatial resolution

FLOW SOLUTION:

1E-4s of Temporal resolution 90mm of spatial resolution

FLOW SOLUTION:

2E-6s of Temporal resolution 5mm of spatial resolution

1 steady state RANS simulation: 45k core hours

1 RANS Simulation = 1.4s of DDES

(NEED about 6s)

1 RANS Simulation = 1.2s of WMLES

(NEED about 6s)

High-Lift Common Research Model: 𝐶𝐿max prediction

1) “High-Lift Common Research Model: RANS, HRLES and WMLES perspectives for CLmax prediction using LAVA” by Kiris et. al. AIAA Scitech Forum 2022 DOI: 10.2514/6.2022-1554

2) “A Reynolds-Averaged Navier-Stokes Perspective for the High Lift-Common Research Model Using the LAVA” by Duensing et. al. AIAA Aviation Forum 2022. DOI: 10.2514/6.2022-3742 and AMS Seminar https://www.nas.nasa.gov/pubs/ams/2022/05-26-22.html

3) “A Hybrid RANS-LES Perspective for the High Lift-Common Research Model Using the LAVA” by Browne et. al.

AIAA Aviation 2022 Forum . DOI: 10.2514/6.2022-3523 and AMS Seminar https://www.nas.nasa.gov/pubs/ams/2022/06-02-22.html

4) “A Wall Modeled LES Perspective for the High Lift-Common Research Model Using the LAVA” by Ghate et. al.

AIAA Aviation 2022 Forum . DOI: 10.2514/6.2022-3434 and AMS Seminar https://www.nas.nasa.gov/pubs/ams/2022/06-09-22.html https://arc.aiaa.org/doi/10.2514/6.2022-1554 https://arc.aiaa.org/doi/10.2514/6.2022-3742 https://www.nas.nasa.gov/pubs/ams/2022/05-26-22.html http://dx.doi.org/10.2514/6.2022-3523 https://www.nas.nasa.gov/pubs/ams/2022/06-02-22.html http://dx.doi.org/10.2514/6.2022-3434 https://www.nas.nasa.gov/pubs/ams/2022/06-09-22.html

High Lift Common Research Model (CRM-HL) at the stalled state

Particle traces colored by Mach number Curvilinear WMLES at the post CLmax Stalled State 1100M point grid (W-D) Timestep Size: 3.4𝜇𝑠

Largest Time-scale Motion: At least 1.5s Largest Relevant Length Scale: At least 7m (size of inboard separated region)

Fastest Time-scale Motion: Approx. 600 𝜇𝑠 (shedding time-scale for slat brackets)

Peak Mach (mean-flow) number:

Approaching 𝑀 = 1 (outboard slats) Peak suction: 𝑐! ≈ −15 (outboard slats)

Smallest Geometric Length Scale:

Approx. 2.5mm (blunt trailing edges) ü Systematic assessment of RANS vs DDES vs WMLES performed for the 4th High-Lift Prediction Workshop HLPW4 ü Unlike cruise conditions, in high-lift configurations, substantial shortcomings in RANS simulation ü In-board, corner flow at the fuselage-wing juncture primary challenge for accurate prediction of aerodynamics ü LAVA scale resolving simulations appear to very cost competitive to legacy RANS models and show improvements over RANS predictions

RANS from AIAA HLPW3 LAVA WMLES from HLPW4

Scale resolving simulations for Certification and Qualification by Analysis

High-Lift Configuration

LAVA is the only NASA in-house capability to contribute in the HLPW4 with RANS, HRLES, and WMLES simulations

Iso-surfaces of Q-criterion 𝜶 = 𝟏𝟗. 𝟗𝟖∘, Cartesian WMLES 3mm, 2010M point mesh

Open-Rotor Acoustics

UAS: Unmanned Aerial Systems

HALE: High Altitude Long Endurance

Urban Air Mobility – Noise Prediction

Isosurfaces of Q-criterion colored by vertical velocity and cut plane colored by logarithm of pressure gradient magnitude

WMLES to Predict Iced Aircraft Aerodynamics

Iso-surfaces of Q-criterion grid refinement around leading edge

Pressure coefficient at 𝛼 = 4.2∘ Lift coefficient

• The LAVA team initiated a development activity to predict flow around an iced aircraft.

• Immersed boundary method with wall-stress boundary conditions from equilibrium wall model

• EG1164 case: full scale 72-inch-chord NACA 23012 with horn ice [Broeren, Andy P., et al. 2010]

• Reynolds number: 15.9 million; Mach number: 0.12

• Volume mesh: 143.7 million Cartesian cells with finest grid spacing ~0.9 mm

• Simulation can be completed using 200 Skylake nodes within 4 hours of wall-clock time for each angle of attack

WMLES to Predict Landing Gear Acoustics

• BANCIII benchmark LAGOON landing gear problem [Manoha & Caruelle 2015]

• 2-wheel landing gear (LG) which is a 1:2.5 nose LG for an Airbus A320 aircraft

• Reynolds number: 1.55 million; Mach number: 0.23

• The irregular shape of the landing gear with the struts and axle is challenging for body-fitted meshing, especially for a fully-dressed landing gear

• Laminar to transition tripping stripes imposed on wheels, axle and struts to model the actual flow past a real nose landing gear

• Detailed experimental results from F2 and C19zwind tunnels

27 Kulites on the geometry

Kulite 8 on the bottom of wheel Far-field flyover microphone # 9

(angular position 120∘downstream)

Kulite 10 on the side of wheel The LAGOON model tripping stripes are colored

34Surface pressure and streamwise velocity shown on planar slice for various database cruise power conditions.

Skin friction and pitching moment as a function of pitch angle for an unsteady forced pitch oscillation simulation.

X-57 Maxwell Aerodynamic Database Generation ü The LAVA Curvilinear CFD flow solver has enabled the efficient generation of large-scale aerodynamic databases, quantifying the distributed propulsion effects on NASA’s all-electric X-57 aircraft

• Over 2800 steady simulations for powered and unpowered flight conditions (left image)

• Over 40 time-accurate forced oscillation simulations to quantify dynamic stability characteristics (right image) ü Results are essential to accurately construct the piloted flight simulator, necessary to train X-57 pilots for safe flight ü Refactorization of the LAVA Curvilinear flow solver has resulted in a 9-12X reduction in simulation cost

• Large-scale database completion time has been reduced from years to months

Subsonic Single Aft-Engine (SUSAN) Aircraft Design and Support

• The SUSAN Electrofan hybrid-electric regional transport aircraft is being developed in a collaborative effort between ARC, GRC, LaRC, and AFRC.

• The LAVA group is contributing to the aircraft design from multiple fronts

1. Conceptual design and sizing

2. High-fidelity steady RANS CFD to investigate Propulsion Airframe Integration

(PAI)

3. Aerodynamic shape optimization to maximize benefits of Boundary Layer

Ingestion (BLI)

• LAVA contributions have helped to quantify overall aerodynamic performance, design the aircraft propulsion systems, and perform design trade studies to develop a feasible and efficient aircraft

Evolution of SUSAN Electrofan aircraft concept.

RANS simulations using the LAVA Curvilinear CFD solver inform design choices and quantify performance of the current Tail-Cone Thruster (TCT, left image) and proposed Distributed Electric Propulsion (DEP, right image) system designs.

Artistic rendering of current proposed SUSAN aircraft configuration during takeoff.

Initial concept Current concept

Aircraft Cabin Air Flow Modeling & Simulations

Source: Lin et al 2014 [Ref 4]

Explore adoption of a Cabin Air Common Research Model (CA-CRM)

• To serve as a test bed for simulation strategy development, verification and validation

• Commonly accessible to and contributed by NASA/DLR/NRC, the aviation industry, and the academia

• Essential to gain confidence in the modeling and simulation of both:

1. Cabin air flow environment

2. Cough/Sneeze etc. events (and the interaction with background flow / other passengers)

• Will provide essential insight into the appropriate physical/computational model fidelity

• Will aid cabin air simulations to be better integrated into the aviation industry design workflow

• Will increase confidence level of current practices using commercial software such as ANSYS, STAR-CCM+, etc.

Potential starting points may already exist

• DLR agreed to share Do728 data and geometry - Collaborating with: Daniel Schmeling, Christian Bauer, Andrei

Shishkin, Mikhail Konstantynov

• Experimental and numerical studies could be made available through workshop problems with participation from academia and the industry

Aircraft Cabin Air Flow Modeling & Simulations – Boundary Conditions

37J. Bosbach et al. ALTERNATIVE VENTILATION CONCEPTS FOR AIRCRAFT CABINS

• Need coordination with the experimental team at DLR to identify a target test case to simulate

• Boundary conditions need to be well defined:

➢ Inlets: Mass or volumetric flow rate, air temperature

➢ Outlets: Back pressure

➢ Cabin surfaces: can likely be assumed adiabatic due to ground level test

➢ Mannequins: Surface temperature

Lower Inlets

Dado panels

Upper Inlets Do 728 cabin section with 7 seat rows

Grid Paradigms in CFD

AMS Seminar: Predicting Orion Launch Abort Acoustics 38 ü High quality body fitted grids ü Low computational cost ü Reliable higher order methods ✘ Grid generation largely manual and time consuming üFully-automated grid generation üHighly efficient Adaptive Mesh

Refinement (AMR) üLow computational cost üReliable higher order methods ✘Non-body fitted à Resolution of boundary layers challenging ü Partially automated grid generation ü Body fitted grids ✘ Grid quality can be challenging ✘ High computational cost ✘ Higher order methods yet to fully mature

Structured Cartesian AMR

Unstructured Arbitrary Polyhedral

Structured Curvilinear

CFD 2030 Technology Development Roadmap

Visualization

Unsteady, complex geometry, separated flow at flight Reynolds number (e.g., high lift)

2030202520202015

HPC

CFD on Massively Parallel Systems

CFD on Revolutionary Systems (Quantum, Bio, etc.)

TRL LOW

MEDIUM

HIGH

PETASCALE

Demonstrate implementation of CFD algorithms for extreme parallelism in

NASA CFD codes (e.g., FUN3D)

EXASCALE

Technology Milestone

Demonstrate efficiently scaled CFD simulation capability on an exascale system

30 exaFLOPS, unsteady, maneuvering flight, full engine simulation (with combustion)

Physical Modeling

RANS

Hybrid RANS/LES

LES

Improved RST models in CFD codes

Technology Demonstration

Algorithms Convergence/Robustness

Uncertainty Quantification (UQ)

Production scalable entropy-stable solvers

Characterization of UQ in aerospace

Highly accurate RST models for flow separation

Large scale stochastic capabilities in CFD

Knowledge Extraction On demand analysis/visualization of a 10B point unsteady CFD simulation

MDAO

Define standard for coupling to other disciplines

High fidelity coupling techniques/frameworks

Incorporation of UQ for MDAO

UQ-Enabled MDAO

Integrated transition prediction

Decision Gate

YES

NO

NO

Scalable optimal solvers

YES

NODemonstrate solution of a representative model problem

Robust CFD for complex MDAs

Automated robust solvers

Reliable error estimates in CFD codes

MDAO simulation of an entire aircraft (e.g., aero-acoustics)

On demand analysis/visualization of a 100B point unsteady CFD simulation

Creation of real-time multi-fidelity database: 1000 unsteady CFD simulations plus test data with complete UQ of all data sources

WMLES/WRLES for complex 3D flows at appropriate Re

Integrated Databases Simplified data representation

Geometry and Grid Generation

Fixed Grid

Adaptive Grid

Tighter CAD coupling Large scale parallel mesh generation Automated in-situ mesh with adaptive control

Production AMR in CFD codes

Uncertainty propagation capabilities in CFD

Grid convergence for a complete configuration

Multi-regime turbulence-chemistry interaction model

Chemical kinetics in LES

Chemical kinetics calculation speedupCombustion

Unsteady, 3D geometry, separated flow (e.g., rotating turbomachinery with reactions)

CFD Vision 2030 Study report published in 2014