Comparison of Higher-Order CFD Modelling with an unsteady Panel - - PowerPoint PPT Presentation

▶
comparison of higher order cfd modelling with an unsteady
SMART_READER_LITE
LIVE PREVIEW

Comparison of Higher-Order CFD Modelling with an unsteady Panel - - PowerPoint PPT Presentation

Oct 08, 2023 105 likes •318 views

www.DLR.de Chart 1 17th ODAS 2017 > Wilke Pade Scheme & UPM Comparison of Higher-Order CFD Modelling with an unsteady Panel Method Gunther Wilke DLR AS-HEL June 7 th 2017 17 th ODAS Aussios, France www.DLR.de Chart 2

slide-1
SLIDE 1

www.DLR.de • Chart 1 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Comparison of Higher-Order CFD Modelling with an unsteady Panel Method Gunther Wilke DLR AS-HEL June 7th 2017 17th ODAS Aussios, France

slide-2
SLIDE 2

www.DLR.de • Chart 2 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Overview

  • Motivation
  • Theory
  • Compact Pade Scheme (and JST Scheme)
  • UPM
  • Simulation Setup
  • Results
  • CFD Grid sensitivity study
  • UPM Grid sensitivity study
  • Comparison of CFD and UPM
  • Summary Conclusions
slide-3
SLIDE 3

www.DLR.de • Chart 3 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Motivation

  • Simulation of blade tip vortices and vortex structure
  • BVI noise in descent flight
  • interaction aerodynamics, for example tail shake
  • Difficulties with State of the Art Tools
  • CFD
  • 2nd order too dissipative
  • plenty of grid points and still not there
  • Simplified Methods
  • accuracy
  • generalization/justification
  • Challenges with higher order schemes
  • stability
  • efficiency
slide-4
SLIDE 4

www.DLR.de • Chart 4 17th ODAS 2017 > Wilke • Pade Scheme & UPM

  • Theory
slide-5
SLIDE 5

www.DLR.de • Chart 5 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Pade-Scheme

  • Higher order finite difference

scheme implemented by Stefan Enk in FLOWer → referred to as FLOWer4

  • 4th order spatial discretization

with 3rd order boundaries

  • 4th to 8th order filtering

with down to 2nd order boundaries

  • Line implicit
  • Grid transformation from arbitrary to

Cartesian grid

  • Not (yet) suitable for transonic flows

Howto get f '?

2h 4h 6h

1 1 2 2 3 3   

   

i + i i + i i + i ' i i

f f c + f f b + f f c = f a

Ansatz Solution through LU-decomposition (Thomas algorithm)

slide-6
SLIDE 6

www.DLR.de • Chart 6 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Jameson vs Pade-Scheme = G + S d F + dV W dt d      Jameson Finite Volume Pade Finite Differences ˆ = J G + ξ F + J W dt d

i i

          

 

 

G + S W F V Δt = RES

t t

                                 G + J t W + ξ F J Δt = RES

i i

  1 Flux Average Difference of Fluxes

slide-7
SLIDE 7

www.DLR.de • Chart 7 17th ODAS 2017 > Wilke • Pade Scheme & UPM

UPM

  • Unsteady Panel Method with

a free wake model

  • Ensures unsteady (pressure)

Kutta-condition

  • Tip vortex roll-up model
  • Viscous vortex core radius

model

Courtesy of Yin et al. 2015

slide-8
SLIDE 8

www.DLR.de • Chart 8 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Simulation Setup

slide-9
SLIDE 9

www.DLR.de • Chart 9 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Numerical CFD Setup

  • Dual-Time Stepping with 1,

1/4, 1/8 degrees timesteps

  • Residual Smoothing
  • 2V Multigrid on JST blocks
  • 6th order Pade Filter

with 4th order at the boundaries

slide-10
SLIDE 10

www.DLR.de • Chart 10 17th ODAS 2017 > Wilke • Pade Scheme & UPM

CFD Grid – Summary coarse medium fine blade 40k 323k 2.6 Mio fuselage 61k 490k 3.9 Mio background 1.4 Mio 11 Mio 88 Mio total 1.6 Mio 13 Mio 103 Mio fully coupled simulations with HOST

slide-11
SLIDE 11

www.DLR.de • Chart 11 17th ODAS 2017 > Wilke • Pade Scheme & UPM

UPM Grid - Summary coarse medium fine blade 480 1080 1920 fuselage 960 wake 10,800 28,800 86,400 total 13,680 34,080 95,040

  • ne-way coupled simulations

with HOST/METAR

slide-12
SLIDE 12

www.DLR.de • Chart 12 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Results

slide-13
SLIDE 13

www.DLR.de • Chart 13 17th ODAS 2017 > Wilke • Pade Scheme & UPM

JST Hybrid-Pade Grid Sensitivity Study (JST vs Pade)

slide-14
SLIDE 14

www.DLR.de • Chart 14 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Grid Sensitivity Study (JST vs Pade) JST Hybrid-Pade

slide-15
SLIDE 15

www.DLR.de • Chart 15 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Grid Sensitivity Study (JST vs Pade) JST Hybrid-Pade medium fine medium fine Cost increase about 50%

slide-16
SLIDE 16

www.DLR.de • Chart 16 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Grid Sensitivity Study (UPM) Vortex field Derivative of airloads

slide-17
SLIDE 17

www.DLR.de • Chart 17 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Grid Sensitivity Study (UPM) experiment coarse medium fine Cost range from working day to two weeks

slide-18
SLIDE 18

www.DLR.de • Chart 18 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Alternative Simulation Techniques Inviscid CFD Viscous CFD

Runtime factor UPM 0.3% Inviscid 63% Viscous 100%

UPM

slide-19
SLIDE 19

www.DLR.de • Chart 19 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Summary & Conclusions!

slide-20
SLIDE 20

www.DLR.de • Chart 20 17th ODAS 2017 > Wilke • Pade Scheme & UPM

Summary & Conclusion

  • Established a hybrid simulation environment for helicopter rotors within

FLOWer with the 4th order compact Pade scheme

  • Hybrid simulation with Pade scheme significantly improved vortex

conservation → better loads correlation → better acoustic correlation

  • Roughly 50% cost increase relative to the JST scheme
  • Investigated different grid sizes for UPM
  • reasonable results on all mesh sizes
  • time range from 7 hours to 2 weeks (reliable CFD ~ 4 Months)
  • Further investigations should examine different rotors to further validate

the applied methodologies