Folie 1 > 12. STAB-Workshop > A. Krumbein

1
Automatic Transition Prediction and Application
to 3D Wing ConfigurationsCurrent status of
development and validation
2
Outline
Outline

• Introduction
• Transition Prescription
• Transition Prediction
• Modeling of Transitional Flow
• Transition Prediction Strategy
• Preliminary Results ONERA M6 wing
• Outlook

3
Introduction
Introduction

• aerospace industry requirement
• RANS based CFD tool with transition handling ?
• different approaches
• RANS solver stability code eN method
• RANS solver boundary layer code
• stability code eN method
• RANS solver boundary layer code
• eN database method
• RANS solver transition closure model or
• transition/turbulence model

- prescription - prediction - transitional flow
modeling - automatic, autonomous
4
Introduction
Introduction

• aerospace industry requirement
• RANS based CFD tool with transition handling ?
• different approaches
• RANS solver stability code eN method
• RANS solver boundary layer code
• stability code eN method
• RANS solver boundary layer code
• eN database method
• RANS solver transition closure model or
• transition/turbulence model

- prescription - prediction - transitional flow
modeling - automatic, autonomous
5
Introduction

• Structured approach
• FLOWer laminar BL method for swept, tapered
  wings
• eN database methods for TS and CF
  instabilities
• FLOWer
• 3D RANS, compressible, steady/unsteady
• structured body-fitted multi-block meshes
• finite volume method, cell-vertex scheme
• explicit Runge-Kutta time integration
• multi-grid acceleration
• mainly eddy viscosity models, Boussinesq

6
Prescription
Transition Prescription
- automatic partitioning of flow field into
laminar and turbulent regions - individual
laminar zone for each element -
different numerical treatment of laminar
and turbulent grid points, e.g. mt 0
in laminar zones
7
Prescription

• transition line on ONERA M6 wing, 4 points on
  upper and lower side

PTupp(sec 2)
PTupp(sec 1)
PTupp(sec 3)
PTupp(sec 4)
8
Prediction
Transition Prediction
- RANS solver ? shall predict transition points
automatically! - stability database ? shall
yield accurate values of transition points! -
eN database method ? needs highly accurate BL
data! ? BL adaptation in NS grid ? very time
consuming, coupling with grid
generator NO! ? laminar BL method ? fast,
cheap, easy to couple YES! - restrictions
? linear stability theory ? parallel flow
assumption - independent of mesh topology, grid
structure, 2D or 3D - integration paths grid
lines of the structured grid
9
Modeling
Modeling of transitional flow
- algebraic models for the transition length
ltr ? Reltr 5.2 (Restr)3/4 downstream of
RANS laminar separation point ? Reltr 2.3
(Red(str))3/2 downstream of BL laminar
separation point ? Reltr 4.6 (Red(str))3/2
downstream of TS instability - intermittency
function ? g(s) 1 exp (- 0.412 3.36 (s -
str)/ltr2 ) s arc length
starting at the stagnation point
displacement thickness
10
Strategy
Transition prediction strategy
- coupling structure
11
Strategy
- algorithm
set stru and strl far downstream compute
flowfield check for RANS laminar separation ?
set separation points as new stru,l cl ? const.
in cycles ? call transition module ? use
outcome of eN-databases or BL laminar
separation point as new transition point set
new stru,l underrelaxed ? stru,l stru,l d,
1.0 lt d lt 1.5 convergence check ? Dstru,l lt e
12
Results
Preliminary Results
- ONERA M6 wing single-element semi-spa
n A 3.8 swept LLE 30 LTE
15.8 tapered l 0.562 - based on
ONERA D airfoil (symmetric),
perpendicular to 40-line - designed for
studies of three-dimensional flows from low to
transonic speeds at high Reynolds numbers

13
Results

• feasibility
• 1 block-grid, 384,000 points
• M? 0.84, Re? 2.0?106, a - 4.0
• turbulence model Baldwin-Lomax
• critical N-factors NcrTS 4.0, NcrCF 2.0,
  arbitrariliy set

14
Results

• Validation, 1st test
• 1 block-grid, 800,000 points
• M? 0.84, Re? 11.72?106, a 3.06 ?
  classic CFD validation test case
• Tu? 0.2 ? N 6.485 using Macks
  relationship
• WT S2MA, Modane Center
• turbulence model Baldwin-Lomax, Spalart-All
  maras with Edwards mod. (SAE), Wilcox k-w
• critical N-factors NcrTS NcrCF 6.485
• transition prediction in 3 wing sections near h
  z/b 0.1, 0.5, 0.9

15
Results

• surface pressure and transition lines

• influence of TMs extremely low
• all transition points due to CF instabilities,
  except
• BL, h 0.1, lower side
• ? lam. sep.

16
Results

• cp-distributions at h 0.2, 0.44, 0.65, 0,9

• almost no difference to fully turbulent re-sults
• accuracy of results comparable to those of others
  (e.g. lite-rature, TAU code)

17
Results

• Validation, 2st test
• 1 block-grid, 800,000 points
• M? 0.262, Re? 3.5?106, a 0, 5, 10,
  15
• Tu? 0.2 ? N 6.485 using Macks
  relationship
• WT S2Ch, Chalais-Meudon
• transition detection in experiment sublimation
  of acenaphtene
• turbulence model SAE
• critical N-factors NcrTS NcrCF 6.485
• transition prediction in 4 wing sections
  near h 0.1, 0.44, 0.5, 0.9

upper side
lower side
18
Results

• transition locations from experiment at h 0.44

h 0.44
h 0.44
TS
TS
lower side
exp.
lower side
upper side
upper side
ls
ls
TS
ls
19
Results

• transition lines for a 5 and exp. transition
  locations at h 0.44

• Has acenaphtene triggered transition on the
  lower side?
• Is NcrCF correct?

TS
outcome of the database methods
h 0.44 on lower side
CF
20
Results

• max. N-factor curves for a 5 at h 0.44 on
  lower side from a linear stability code
  (from H.W. Stock using COAST (?) code)

NcrCF ? 3.2
In other cases, e.g. ONERA D infinite swept,
NcrCF ? 6.0 was found.
21
Results

• What is wrong?
• 1. Error in coding of the 3d coupling procedure?
• ? compute infinite swept wing flow for ONERA D
  airfoil using sweep angle at xTlow(h
  0.44)
• ? fails due to problems with BL code BL code
  does not converge
• ? another problem to be solved!
• 2. Is sweep angle correct?
• ? account for effective sweep angle Leff
  L DL arcsin (UT/U?)) due to
  influence of changing absolute wing
  thickness ratio UT velocity in the
  attachment line
• tested 1 ? DL ? 6
• ? cp around stagnation point must be
• reduced to prevent BL code crash
• ? are database results affected?
• ? another problem to be solved!

) G. Redecker, G. Wichmann, Forward Sweep A
Favorable Concept for a Laminar Flow Wing,
Journal of Aircraft, Vol. 28, No. 2, 1991, p.
97-103
22
Results

• What is wrong?
• 3. Is CF database method erroneous?
• ? ONERA D infinite swept successfully analyzed
  by ISM (TU-BS) with same program for M?
  0.23, Re? 2.4?106, an 4, L 60 using BL
  data from TAU code
• ? Results from CF database method are almost
  the same as those from linear stability code
  COAST.
• ? Is the functioning of the CF database method
  case dependent?
• 4. Are the grid lines of the structured grid a
  too bad approximation of the streamline?
• 5. Is the selected test case a reliable
  validation test case?

23
Outlook
Outlook

• clarification/solution of the problems
• convergence problems of BL code
• automatic determination and consideration of Leff
  in the iteration loop
• automatic reduction and adaption of cp around
  stagnation point
• guarantee that CF databse results are do not
  depend on manipulation of cp
• reproduction of the results of the ONERA D
  infinite swept case
• coupling with linear stability code LILO (G.
  Schrauf)
• empirical criteria for - attachment line
  transition - bypass transition -
  transition in laminar separation bubbles