7/28/2019 AA V7 I1 Flight Simulator

1/3 2013 ANSYS, INC. ANSYS ADVANTAGEVolume VII | Issue 1 | 2013 35

FlightSimulator

Aeroelastic Prediction Workshopassesses computational methods forpredicting unsteady ow felds.

By Thorsten Hansen , Customer Support, ANSYS, Inc.

ANALYSIS TOOLS

T he early stages of aerodynamicdesign typically are based on theassumption that the mechan-ical structures are rigid, a concept thatgreatly simplifies the design process.

As the process progresses, the designermust take into account the elasticityof the mechanical structures, becausethis often has a major impact on aero-dynamic performance and structuralintegrity. Aeroelasticity refers to the inter-action between the aerodynamic flow

eld and the mechanical structure. It isone of the most challenging aspects of designing aerospace structures becauseits di cult to model the tightly coupledmultiple physics involved in the phe-

nomenon. Traditionally, companies havedeveloped in-house methods that use cus-tom aeroelasticity codes. However, withthe relentless demand for higher- delity information to support improved aero-dynamic efficiency, coupled with the

high cost of maintaining these codes, theindustry is actively exploring the develop-ment of new techniques.

To address this challenge and assessstate-of-the-art methods and tools for the

prediction and assessment of aeroelasticphenomena, the American Institute of Aeronautics and Astronauts hosts theAeroelastic Prediction Workshop as a con-tinuing education event. It provides animpartial forum to evaluate the e ective -ness of current simulation tool codes andmodeling techniques as well as to deter-mine whether or not they are able toaccurately simulate challenging nonlin-ear aeroelastic problems involving vor-tices, shock waves and separated flow

that result from aeroelasticity. In prep-aration for the 2012 workshop, ANSYSused ANSYS Fluent and ANSYS CFX com-putational uid dynamics (CFD) softwareto solve the problems, then presentedthe results at the meeting. A select set of

The designermust take intoaccount theelasticity ofthe mechanicalstructures,because thisoften has amajor impacton aerodynamicperformanceand structuralintegrity.

7/28/2019 AA V7 I1 Flight Simulator

2/3 2013 ANSYS, INC. ANSYS ADVANTAGEVolume VII | Issue 1 | 2013 36

ANALYSIS TOOLS

results is illustrated in this article. The full data set is availableonline (see endnote).

The Aeroelastic Prediction Workshop challenges the simu-lation community to apply best practices to predict unsteadyaerodynamics characteristics for rigid or weakly coupled aero-elastic systems. The test con gurations were selected to increase

in complexity, from fully turbulent with attached ow and weakshocks to transient separation conditions with strong shocks andsigni cant interactions between ow features. For each con gu -ration, participants were asked to provide a convergence study,steady-state analysis, and time-accurate response to forced oscil-lations using best practice guidelines to quantify numerical andmodel errors. The simulation results were compared with windtunnel test data. Aeroelasticity problems have a high computa-tional intensity, so the ANSYS team used high-performance com-puting (HPC) platforms with up to 2,048 cores on a Cray XE6 tosolve the workshop challenges.

RECTANGULAR SUPERCRITICAL WING

The rst test con guration, called the rectangular supercriticalwing (RSW), is a structurally rigid rectangular planform wingthat oscillates at a designated pitch amplitude and frequency.The wing is mounted to a small splitter plate that is o set 6inches from the wind tunnel wall. The RSW was originally testedin the NASA Langley Transonic Dynamics Tunnel in 1983. Thestudy was performed at Mach 0.825 with a 4 million Reynoldsnumber. Steady-state data was obtained with the model heldat a xed angle of attack. Separate dynamic data was acquiredby oscillating the model in a pitching motion about the 46 per-cent chord line. Forced pitch oscillation frequencies of 10 Hz and20 Hz were used in the two dynamic test cases.

The RSW was meshed with ANSYS ICEM CFD using hexahedralelements and a scalable grid that provides consistent mesh qual-ity after grid re nement. This application appeared to be quitesimple at rst glance. However, the simulation conducted withboth Fluent and CFX revealed modeling complications becausethe splitter plate and wing model were enveloped in the bound-ary layer. Despite that fact, both the static and dynamic simula-tion results showed general agreement with experimental dataat higher span locations where the in uence of the wind tunnelboundary layer and splitter blade is low.

BENCHMARK SUPERCRITICAL WING

The benchmark supercritical wing (BSCW) is another structur-ally rigid rectangular planform wing that oscillates at a desig-nated pitch amplitude and frequency. The BSCW has a NASASC(2)-0414 airfoil. All data in this study was obtained at Mach0.85 with a dynamic pressure of 200 psi, xing the Reynoldsnumber at 4.49 million based on the wing chord. Dynamic datawas obtained for the BSCW by oscillating the model in a pitch-ing motion about the 30 percent chord. Steady-state informationwas calculated as the mean value of oscillatory time histories.This con guration was chosen because the experimental data

shows highly nonlinear unsteady behavior, particularly shock-separated transient ow.

Pressure coe cient results determined by the CFX solver showa signi cant in uence on the turbulence model for shock predic -tion on the upper surface for the static solution. The dynamic sim-ulation was solved with a converged steady-state solution as the

initial conditions and displacing the mesh to match the harmonicwing motion. Simulation showed this test con guration to havea very complex ow eld driven by the shock-induced separated

ow. The dynamic cases were harder to predict, and some varia -tions were seen between test results and simulation.

HIGH REYNOLDS NUMBER AEROSTRUCTURAL

DYNAMICS MODEL

The high Reynolds number aerostructural dynamics (HIRENASD)test con guration is a 34-degree aft-swept, tapered clean wingwith a BAC 3-11 supercritical airfoil pro le. The semi-span testarticle is mounted to the ceiling using a noncontacting fuselagefairing connected to a turntable, balance and excitation system.The HIRENASD wing is geometrically more complex and has asmall amount of structural exibility that is used to oscillate the

wing in its structural modes to acquire unsteady aerodynamicdata. The HIRENASD was excited at the second bending mode fre-quency at approximately 80 Hz forces applied using piezoelectricstacks in the mounting hardware.

The structural exibility in this test con guration increasedthe complexity of the simulation challenge. ANSYS Mechanicalwas used to simulate the structure and calculate mode shapes.Static aeroelastic equilibrium results predicted by the CFX simu-lation closely matched experimental values. The dynamic caseswere solved by using the converged steady-state solution as theinitial conditions and displacing the mesh to match the har-monic wing motion predicted by structural nite element anal -

ysis (FEA). Agreement was again good at the mid-span crosssections in the dynamic test cases.

ANSYS software performed consistently in all three testcases. The results with CFX and Fluent converged rapidly toa single solution, both as a function of grid density and timeinterval. Results were validated by running a grid and time

Comparison of ANSYS Fluent CFDresults to wind tunnelexperiment for RSWdynamic test case at10 Hz

Rectangularsupercritical wing inwind tunnel PHOTO COURTESY NASA.

7/28/2019 AA V7 I1 Flight Simulator

3/3 2013 ANSYS, INC. ANSYS ADVANTAGEVolume VII | Issue 1 | 2013 37

step sensitivity study. Despite the complexity of the ow eldsin these test con gurations, simulation results in most casesshowed good agreement with wind tunnel tests. As expectedwith problems of this di culty, some deviations were identi -

ed in the simulation results, and these are being investigated.ANSYS continues work on improving capabilities to solve

these and other aeroelastic simulations. Publications sum-marizing the data from the 2012 Aeroelastic PredictionWorkshop are being prepared for conferences in 2013. ANSYSis already planning to participate in the next Aeroelastic

Prediction Workshop and will use ANSYS Workbenchto seamlessly couple the CFD and FEA simulations tomore accurately simulate aeroelastic behavior.

Thanks to Cray for the high-performance computing resources used by ANSYS

for this project.

Simulation results in most cases

showed good agreement withwind tunnel tests.

Fluent calculationsfor BSCW show good

t with experimentalmeasurements.

Simulationpredictions of unsteady-statecoe cient of pressurefor HIRENASDcorrelate well to windtunnel experiments.

ANSYS CFX turbulence model error for BSCW at steady-state conditions Comparison of static aeroelastic equilibrium simulation predictions toexperimental measurements for HIRENASD

BSCW setup forwind tunnel testingPHOTO COURTESY NASA.

HIRENASD is 3-Daeroelastic wing withgeneric fuselagemodel. PHOTO COURTESYRWTH AACHEN.

Splitter Plate

BSCW

Fairing

ReferenceAeroelastic Prediction Workshop, https://c3.nasa.gov/dashlink/projects/47/