Requirements for BVI Noise Study

mustardarchaeologistMécanique

22 févr. 2014 (il y a 3 années et 1 mois)

63 vue(s)

Requirements for BVI Noise Study

Sponsored by WHL

Romuald Morvant

Computational Fluid Dynamics group

University of Glasgow

Helicopter Study Weekend,

the Burn, April 2002

Summary

1
-

Previous progress

2
-

Vortex generation
-

Requirements

3
-

Review of CAA methods
-

Requirements

4
-

Conclusions

5
-

Further work


Review of the possible high
-
order schemes

Previous Work


Implementation of the UNFACtored method



Debugging of the 3D CHIMERA mode



=>RESULTS

7A Model Rotor

Results on a Coarse Grid & CHIMERA

Pressure and Normal force coefficients


CHIMERA Grids

HOVER test case:




Incorrect solutions on CHIMERA grid



Load distribution, Normal force coefficient


-

Reason? => vortex located in the overlapping


area (lower order of accuracy)


-

Solutions: Froude conditions?


Boundary conditions?


Numerical scheme?


Results

TEST Cases
SOLUTIONS
GAIN
LANN
Wing
0.25
ONERA
FORWARD non-lifting
0.95
2-bladed
7A
Transonic HOVER flight
4-bladed

Single grid
0.23
Chimera grid
-








Current Work


Aeroacoustic objective => BVI study




Generation of a vortex in EROS and PMB





Use of the ICEM package to generate the grid


Transformation of the grid







EROS
-
Vortex Generation

470


200


2 C
-
H grid, 8 chords length

EROS
-
Vortex
-
Airfoil Interaction

EROS
-
Decay in Vortex Strength

PMB
-
Vortex Generation


UNIFORM grid


Fine H
-
grid:


1000 x 135 x 2



-
> test for


the vortex convection



PMB
-
Vortex Convection

PMB
-
Decay in Vortex Strength

PMB
-

Pressure Evolution at the Vortex Core

CYCLE

------
>


Preserving Vortex Characteristics

HIGH
-
ORDER ACCURATE SCHEME



Convection of the vortex (less points/wavelength)



PB of dispersion for any scheme


FINE & SMOOTH GRID:



Capture the small scales



Memory requirements


-
> Necessity of parallelisation

-
> Use of PMB and test of the chosen implemented
scheme on PMB (cf: Unfactored method)

CFD Methods for CAA

-

Complicated Flow Field, strong sound directivity

-

Thickness and load noise are well predicted

-

Blade vortex interaction noise less well predicted

Near
-
Field Methods


CAA is currently moving towards CFD especially

for near
-
field predictions



-
> Computation of the fluid
-
flow mean motion


-
> Computation of the source of noise


-
> Capture of the flow non
-
linearities



Far
-
Field Methods

1
-

Kirchhoff method



Choice of the Kirchhoff surface



Calculation of the overall noise level


2
-

FW
-
H




Computation of the quadruple noise



Noise mechanisms are decoupled

Dispersion Relation

for
centered

schemes

V=d
w
/dk, dispersion velocity

V=d
w
/dk, dispersion velocity

Dispersion Relation

for upwind biased schemes

Discontinuous case


High Order Schemes & Discontinuities

Smooth case

One
-
Dimensional Case

CAA Requirements

High resolution between scales of the flow

-
> Low dissipation




High
-
order numerical scheme


with the appropriate grid's size



Representative flow
-
field data from CFD


Use of the Kirchhoff formulation


Faster calculations for the FW
-
H method

Summary of Current Efforts



Simulation of a vortex
-
airfoil interaction





Large dissipation of the vortex (larger near the airfoil)



Incorrect convection of the vortex



High
-
order accurate in space scheme


Grid refinement


Test: aeroacoustic calculations



Further Work

1) Implementation of an aeroacoustic module:


Surface extractor in PMB


Use of the Kirchhoff method


Test for simple test cases (ball & Gaussian source)


-
> Being tested


2) Choice of the high
-
order scheme and implementation





Further Work

Check of the capabilities of the solver


for the vortex generation



Use of the CFD data to get the frequency at which
the BVI occurs for a documented test case


Check the quality of the generated vortex


Check the capabilities of the schemes


Refine the domain so as to get good results


HIGH
-
ORDER SCHEMES

How to get higher
-
order of accuracy


over large period of time?




TIME
-

Explicit





Stability: Dx~=Dt
-
> non physical


-

Implicit



Smaller dispersion error, more effective when


coupled with high
-
order accurate in space




-
> Implicit 4
th

order exists (midpoint rule)


PB: the linearisation of the convective terms (2
nd

order)

How to get higher
-
order of accuracy


over large period of time?



SPACE : highly accurate spatial derivative of the flow (few
points/wave length)



-
> high
-
accuracy finite difference method:

-
Explicit scheme (large stencil)


-
Compact scheme (smaller stencil, use of the


derivative of neighbouring points)


High
-
order accurate scheme


1) Fourier series analysis



Spectral method



Dispersion Relation Preserving scheme (DRP)





Good quality wave resolution




Improper capture of wave discontinuities (central)




Not suitable when Mach increases


-
> Use smoothing techniques



High
-
order accurate scheme

2) Truncated Taylor Series Method



ENO, WENO, WCNS
-
> Explicit


-
> Under development (FV)



2
-
4 MacCormack
-
> Explicit & Implicit


-
> Central (unpublished papers)




Compact
-
>Explicit & Implicit


-
> More mature



damping scalar: solution less physical


filtering: high
-
order filter

Compact Schemes


CENTRAL:



they can be spectral
-
like




spurious oscillations (smoothing terms)


UPWIND:



dissipative behaviour appropriate at/near Bs



dissipative error (highest wave numbers)


Difficulties:
-

it requires smooth meshes


-

it must be adapted to the Bs


-

very sensitive to BCs on coarse grid

Compact Scheme

AEROACOUSTIC


Resolution in the wave number domain:


-
> Advantage of the filtering


-
> Computational cost: not much, truncation error


decreases and the number of required points near
the Bs diminish


Proposed Schemes

TIME:

-

implicit 2nd order accurate


SPACE:

-

3rd order upwind
-
biased in space

-

4th central
-
differences scheme