19 vues

Transféré par Ana Mora

cfd

cfd

© All Rights Reserved

- David Tamayo Case Study
- Colebrook White.( 1939)
- FLUID LAB Friction
- Tese Important
- ChE 190 Summary of Journals
- turbulent flow.pdf
- Homework 3
- A Parallel Adaptive Mesh Refinement Algorithm for Predicting Turbulent Non-premixed Combusting Flows
- 7.Moin BFS
- 2018 KUHNEN Destabilizing Turbulence in Pipe Flow
- biotechnology lab report
- CC15 Laminar Flow
- -03-28-heatexchanger-fan.pdf
- Lab 7 Osborned
- Venturi
- renold #
- inestabilidad poisuielle turbul
- Jirka 1994 Turbulence in Open‐Channel Flows
- Fulgosi_2003.pdf
- Chap1_n3 chassaing

Vous êtes sur la page 1sur 14

NUMERICAL

SIMULATION OF

INDUSTRIAL FLOWS

The numerical simulation of industrial

flows is an increasingly important means

to solve a large variety of fluid flow

problems, such as internal flows, external

aerodynamics, spray cooling, film coating,

environmental and biological flows, and

power generation. Several general-

purpose codes are available at VKI and

are used to model a wide range of fluid

flow processes including turbulence, heat

transfer, multiphase flows, chemistry, etc.

The von Karman Institute is working in

collaboration with several companies to

examine the fundamental and applied

aspects of these processes and to provide

efficient tools to aid industry in decision-

making.

2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

NUMERICAL SIMULATION

OF INDUSTRIAL FLOWS

The numerical simulation of industrial flows is an in-

creasingly important means to solve a wide variety of

fluid flow problems such as internal flows, external

aerodynamics, spray cooling, film coating, environ-

mental and biological flows, and power generation.

Several general-purpose codes and a large number of

specific VKI codes are now available for industrial flow

applications at the Institute. They model a wide range

of fluid flow processes including turbulence, heat

transfer, multiple phases, compressible flows and

chemistry and can be applied to complex flows in-

cluding free-surfaces and porous media. When used

to provide fast, reliable and low cost solutions of en-

gineering flow problems, they become efficient tools

to aid industry in decision-making and the preparation

of sophisticated and costly experimental tests. A se-

lection of typical fluid dynamic problems to which nu-

merical approaches have been applied at the VKI is

presented here to illustrate the applicability of such a

methodology to the industrial and engineering fields.

Comparisons with experimental data are also made

for some of the applications.

The computation of heat transfer by an impinging air

jet is an important research activity with two major in-

dustrial projects that will each involve experimental

validation. An impinging jet is an interesting means

to increase heat and/or mass transfer rates.

The first project deals with the cooling of fast-moving

steel sheets by multiple jets. Numerical simulations

have been performed for the investigation of the ef-

fect of the jet Reynolds number, stand-off distance, jet

spacing and jet angle on the mean heat transfer coef-

ficient on the steel strip. Figure 1 shows a typical re-

sult of the influence of the strip velocity on the flow

structure between the jets.

The second project concerns the design of a thermal

anti-icing system for the leading edge of a commer-

cial aircraft and has been undertaken on the basis of

numerical simulations. The anti-icing system is com-

posed of impinging hot jets and also of hot air chan-

nels located below the surface of the wing (double

skin). The numerical results has been compared with

experimental measurements using infrared thermog-

raphy.

The ventilation of a filling tunnel used in

the pharmaceutical industry has also been investigat-

ed numerically. The main objective is to identify the

occurrence of flow recirculation in the tunnel and to

limit such a phenomenon by proposing modifications

of the design to improve the flow uniformity. A typi-

cal result showing velocity magnitude in different

planes is given in Figure 2.

The two-phase flow in an airlift reactor configuration

is analysed using the two-phase models of the com-

mercial code FLUENT. The airlift reactor consists of an

upcomer and a downcomer with equal cross-section.

The distance between the bottom and the middle sep-

arating wall is kept constant.

Numerical Simulation of Industrial Flows

96 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Figure 1: Influence of a moving strip

on the flow structure

Figure 2: Velocity

magnitude in the tunnel

Figure 3: Void fraction and velocity

magnitude in an airlift reactor

The distance between the upper end of the mid-

dle wall and the free surface is varied during the

study. Air is injected through holes in two hori-

zontal pipes at the lower end of the upcomer.

Results of void fraction and velocity magnitude

are shown in Figure 3. The highest values for the

void fraction are obtained at the injection and

also close to the free surface where air is escap-

ing. Velocity magnitude obtained by numerical

simulation shows the airlift produced by the

bubbles and will be compared with experimen-

tal data.

Due to international regulations concerning the

reduction of solvent-based painting, powder

coating becomes a very attractive process. In

such a process fine solid particles of paint are

transported by a gas flow over a moving strip.

The VKI has been involved in the numerical sim-

ulation of such a complex technique. In a first phase,

the study is focused on the influence of the speed of

the moving strip, the gas flow rate and the geometri-

cal parameters such as the width and the thickness of

the strip on the flow structure within the channel.

Numerical simulations are also used for the design of

complex three-dimensional flow systems as shown in

Figure 4. In this study, different geometrical configu-

rations have been simulated and compared. The ob-

jective is to have a uniform velocity profile at the out-

lets of the nine horizontal slits shown in the figure.

Porous material and deflectors have been added to the

geometry to obtain the required results.

One of the key objectives for building designers is to

maximise glare-free natural daylight in buildings while

minimizing solar heat gain. External solar shading re-

duces the amount of direct solar heat gain thereby re-

ducing the building's cooling requirements. To study

wind effects on window shades, a series of full-scale

tests were performed in the L-1 wind tunnel on a so-

lar protection system. Numerical simulations have

been performed to gain a better understanding of the

unsteady flow field around a single shade

and around a set of sun pro-

tection shades. Figure 5 shows

the velocity magnitude and lift

coefficient for two different

time steps corresponding to two ex-

tremes values of lift coefficient.

The residence time in a chemical

tank has been computed and

analysed. Particles have been

tracked inside the tank in order to

have a histogram of their transit

time. These values have been com-

pared with experimental data com-

ing from the industrial site. The

geometry of the tank is com-

plex and consists of different

side tubes for the inlet and out-

let of the various products. The

surface grid of the tank is repre-

sented in Figure 6.

Figure 6: Surface grid of the tank

Numerical Simulation of Industrial Flows

97 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Figure 4: Surface grid of a complex

three-dimensional geometry

Figure 5: Unsteady flow field around a sun

protection device

With the continuous increase in operating speeds of

high speed trains, several problems arise when trains

enter confined areas (tunnel, train station). It is well

know that a train passing through a tunnel generates

pressure waves that propagate back and forth to the

portals where they are reflected. The resulting tran-

sient pressure may affect passenger comfort and

cause damage on certain elements of the train.

Numerical simulations are therefore performed to de-

sign airshafts located inside the wall of the tunnel to

reduce the pressure wave. Three-dimensional tran-

sient simulations are performed using the geometry

of a typical high-speed train as shown in Figure 7.

In collaboration with the

International Polar Foundation,

the VKI is carrying out the aerody-

namic design of the future Belgian Antarctic Research

Station. The station will be built on a small granite

ridge at an elevation of less than 20m above the sur-

rounding snow surface. Wind tunnel tests are per-

formed to asses snow drift and snow erosion.

Numerical simulations are used to determine the wind

load on different parts of the building. Contours of the

pressure distribution, C

p

, are shown in Figure 8 for a

typical orientation of the buildling on the ridge.

Numerical Simulation of Industrial Flows

98 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Figure 7: surface

mesh of a typical

high-speed train

Figure 8: contours of C

p

on the polar station

99

TURBULENT FLOW

AND

HEAT TRANSFER

The VKI has considerable experience in

investigating heat transfer in various

complex flow systems. The latest applica-

tions are related to fast cooling of moving

steel sheets, the anti-icing of aircraft,

compact heat exchangers and powder

coating processes. In recent years, VKI has

maintained a continuing effort on the

development of an in-house LES code

(MIOMA) to study in depth the mechanisms

of turbulent transport. The application of

Multi-Domain techniques made possible

the simulation of the flow around 3D

bluff bodies. Advanced post-processing

algorithms were also developed to detect

coherent structures from CFD or PIV results

and extract statistical information from this

population.

2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Turbulent Flow and Heat Transfer

100

HEAT TRANSFER

IN COMPLEX FLOW SYSTEMS

Research in heat transfer in complex situations such

as impinging gas jets, flow over ribbed surfaces, and

compact heat exchangers has been pursued. The ap-

plications under concern are fast cooling of moving

steel sheets, the anti-icing of aircraft and the powder

coating process.

The study of design parameters allowing the optimi-

sation of a gas cooling system based on multiple im-

pinging jets has been carried out on a test facility sim-

ulating at scale 2:3 a real industrial fast cooling system

[MP17]. The local and mean convective heat transfer

coefficients in an arrangement of gas slot nozzles have

been determined by means of infrared thermography

combined with the thin heated foil technique [MP48].

The study aims at determining the effect on the aver-

age heat transfer coefficient of the slot Reynolds num-

ber up to a value of 100000, with the nozzle spacing

normalised by the slot hydraulic diameter from 6 to

18, the normalised nozzle emergence length, E/S, from

5 to 17, and the normalised nozzle-to-strip standoff dis-

tance Z/S from 3 to 10. The geometrical arrangements

tested include perpendicular (90) and tilted (60) noz-

zles. Typical IR thermogrammes shown (Figure 1) are

a faithful depiction of the impinging flow pattern ac-

cording to the system design.

At large W/S-values (Figure 1a), the low spanwise dis-

tortion of the isotherm contours denotes the two-di-

mensional character of the impinging flow. For a jet

configuration characterised by small W/S-values, as

featured by Figure 1b, a jet-merging phenomenon is

observed and the local distribution of the heat trans-

fer coefficient does not mimic the nozzle arrangement

[MP49].

The axial distribution of the Nusselt number, Nu,

based on the hydraulic slot diameter, S, and corre-

sponding to the jet arrangement of Figure 1a, is plot-

ted in Figure 2. Peaks of Nusselt number of 400 and

more can be reached at the impingement of the

peripheral perpendicular jets. However, the two

60-tilted jets, designed to create additional moving

strip stabilisation yield a lower cooling.

Three-dimensional numerical simulations have been

performed with the code Fluent. The meshing consists

of typically 500000 hexahedral cells. Simulations are

conducted on half of the domain taking advantage of

the symmetry plane. A k- turbulence model with wall

functions is used for the computations. The boundary

conditions are imposed based on the experimental

data (temperature, velocity, heat flux).

After validation from comparison with the laboratory

IR results, the numerical model has been applied to

industrial situations [AJ30]. In particular, it can illus-

trate the enhancement of the heat transfer produced

Figure 1: IR mapping of the temperature field

a)

b)

Figure 2: Streamwise Nusselt number distribution

Figure 3: Effect of H

2

/N

2

mixture on the heat

transfer coefficient obtained with Fluent

using k- model

2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Turbulent Flow and Heat Transfer

by using a H

2

/N

2

gas mixture instead of air. Such an

investigation would be difficult in the pilot facility for

safety considerations, but can easily be approached

with CFD. Figure 3 shows the axial distribution of the

heat transfer coefficient obtained for 5% and 60% H

2

,

respectively, as normalised by the heat transfer coef-

ficient for air only. The increase of hydrogen content

has a strong effect on the heat transfer rate, particu-

larly in the impingement region. At constant Reynolds

number, using 30% H

2

may readily double the cooling

efficiency.

Instead of a steady k- turbulence model, a more ad-

vanced simulation of the turbulence is obtained with

the Large Eddy approach which is providing a large

amount of unsteady results. This approach is now ma-

ture enough to be extensively used and proposed.

In Figure 4, an example of in-

stantaneous temperature con-

tours on a plate cooled by a

round and cold turbulent jet is

presented [MP136].

Such round jets are also used in recent designs of anti-

icing systems for the wing and tail surfaces of air-

planes. In this technique, hot round jets are issued

from a perforated supply duct, called a piccolo. The hot

air flowing on the inner surfaces of the skin prevents

ice formation on the leading edge and subsequently

on the wing upper surface through a double skin sec-

tion when severe atmospheric flight conditions are ex-

perienced. Another concept based on effusive heating

through a porous skin section is also considered. Such

thermal de-icing systems have been studied at the VKI

via experimental approaches involving infrared ther-

mography and numerical simulations.

The experimental investigation of a full-scale imping-

ing round jet system was performed on a leading edge

section of an aircraft wing. Figure 5 gives a view of the

actual model used. Different operating conditions and

geometries of the multiple jet system were tested to

determine heat transfer correlations and optimise the

de-icing performance.

Three-dimensional numerical simulations of the

aerothermal behaviour of the multiple jet system have

also been performed numerically to make a predesign

of the de-icing configuration. The standard steady vis-

cous equations are solved in combination with a k-

turbulence model. Figure 6 shows typical mappings

of the heat transfer coefficient on the leading edge

skin for two piccolo designs.

Research involving heat transfer in confined flow is

also pursued. A heated double skin section of the

wing upper surface consists in a kind of plate heat

exchanger made of two metallic plates sandwiching

small aluminium slats to form seven parallel slit-type

channels, 0.3m long and 0.02m wide. The channel height

may vary from 0.001 to 0.002m, respectively. The top

wall is modelled by a thin heated thermofoil while the

bottom wall is insulated. Air at laboratory room

temperature is blown uniformly through the double

skin at chan-

nel Reynolds

n u m b e r s

ranging from

2700 to 6700.

Local heat

transfer is in-

ferred from

the constant

heat flux and

the surface

temperature

is measured

by infrared

thermography.

Figure 4:

Large Eddy

Simulation of an

impinging jet

Black contours:

temperature in z = 0 plane

Colour flooded contours:

temperature on the plate

Figure 5: Aerothermal de-icing model

Figure 6: Typical CFD simulation

of aerothermal de-icing per-

formance obtained with Fluent,

using k- model

101 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Turbulent Flow and Heat Transfer

102

IR thermogrammes indicate that the heat flow is uni-

directional and that the data reduction can be per-

formed considering the streamwise evolution of the

temperature along a line located in the centre part of

the specimen.

Figure 7 shows the axial evolution of the local Nusselt

number based on the hydraulic diameter of the chan-

nel for the specimen of aspect ratio equal to 0.05

[MP50]. At a Reynolds number corresponding to the

laminar regime (Re = 2790) the decrease of the Nusselt

number with the streamwise distance is consistent

with the thermal boundary layer development [MP51].

The plateau value, which denotes fully developed

flow, agrees perfectly with the Nu value of 8.24 pre-

dicted for laminar flow between infinite parallel plates

and its streamwise location x/D

h

= 80 compares well

with the literature. The subsequent drop at the end of

the specimen denotes an effect of the exit design in-

volving holes drilled through the bottom plate before

the end of the channels.

An efficient way to control the thermal behaviour of a

body exposed to a gas stream, is to inject gas flow

through the wall, into the boundary layer developing

on the surface. The injection can be done through a

porous surface. When wall heating is performed with

such an injected hot gas, the technique, which is

known as effusive heating, could be applied to the de-

icing of flying bodies. An experimental investigation

of the effect of the blowing factor and the influence of

the porous wall design on the thermal efficiency of ef-

fusive heating has been conducted in the VKI CWT-1

facility as sketched in Figure 8.

Two porous walls have been tested. The first specimen

is a bronze porous matrix, 0.003m thick with porosity

of 34%. The second specimen is a thin laser-perforat-

ed titanium plate with a thickness of 0.0009m. The

drilled holes have a diameter of 50m and are regu-

larly ordered with a pitch of 10 diameters. The experi-

mental conventional effectiveness of the effusive heat-

ing, , defined as the relative temperature of the

heated wall with respect to free stream temperature

compared to the relative hot air temperature, is plot-

ted versus the blowing factor in Figure 9.

Figure 9 compares the present IR data to literature.

The agreement is found to be satisfactory for speci-

men 1 considering that the turbulence level and the

initial boundary layer state could be different in both

studies. However, for specimen 2 where higher blow-

ing factor values are reached, the findings show that

a saturation is reached far from the expected effec-

tiveness that should be close to unity. The explanation

lies in the porous pattern of specimen 2; indeed, the

excessive exit velocity of the hot air jets and the large

distance between holes (10d) as well, lead to entrain-

ment of cold air impinging on the porous structure as

sketched in Figure 9. Such a formation of mixing cells

characterised by high turbulence level lessens the ef-

fective thermal protection and the final result is op-

posite to what was expected.

An experimental study of convective heat transfer in a

channel implemented with perforated ribs immersed in

a turbulent boundary layer has been carried out [AJ9].

Infrared thermography associated with the steady

heated thin foil technique has been applied to obtain

the mapping of the heat transfer coefficient. Different

Figure 7: Streamwise distribution of Nu-number

in the plate heat exchanger at transitional regime

Figure 8: Effusive heating test in CWT-1

Figure 9: Effusive heating efficiency

versus blowing factor

Figure 10: Effect of the rib number on developed

flow: Chevron type turbulators Re = 32000,

= 0.533, P/H = 5

2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Turbulent Flow and Heat Transfer

types of perforations were compared in the case of a

single turbulator configuration. Conclusions were

drawn on the effect of the rib design on the heat trans-

fer enhancement. Next, the study focuses on the tur-

bulator with perforations of the chevron type. Then,

emphasis was given to the influence of the rib-to-rib

spacing, the open-area factor and the channel

Reynolds number on the thermal exchange.

Compared to a solid rib, a local thermal enhancement

factor of 3 can be expected just behind the perforated

turbulator. Fully developed flow is reached after the

5th to 6th rib as shown in Figure 10. The optimal de-

sign combines a rib pitch-to-height ratio of 5 with an

open area factor of 0.53 for channel Reynolds numbers

ranging from 30000 to 60000 as shown in Figure 11.

The effects induced by classical turbulators are also

deeply investigated in a collaboration between the

Environmental and Applied Fluid Dynamics and the

Turbomachinery and Propulsion departments involv-

ing a Large Eddy Simulation study of multiple ribs

placed in a square duct. The innovative aspect of this

effort is found in the consideration of fully coupled

heat transfer; i.e. a time-resolved solution is proposed

both in the solid and in the air flow. For the first time

at VKI, rotational effects on turbulence and on heat

transfer are numerically investigated. A 2D cut of the

three dimensional mesh is presented in Figure 12

(top). The rotational effect on the coherent structures

contents is seen in Figure 12 (bottom), where half of

the duct is including the rotational effect (in red and

yellow). This image has been obtained by merging two

data files for a direct comparison purpose, the second

half of the duct (structures in green and blue) belongs

to the non-rotational case. It is worth mentioning that

the coherent structures are visualized thanks to an iso-

surface of the Qcriteria (energy based on a conditional

vorticity dominated by rotation).

Finally, due to international regulations concerning the

reduction of solvent-based painting, powder coating

becomes a very attractive process. In such a process

fine solid particles of paint are transported by a gas

flow over a moving strip. The particle depo-

sition is controlled mainly by an applied electrostatic

field and the turbulence of the channel flow. The VKI is

now involved in the numerical simulation of such a

complex coating technique. In a first phase, the study

is focused upon the influence of the speed of the mov-

ing strip, the gas flow rate and the geometrical pa-

rameters such as the width and the thickness of the

strip on the coating uniformity. Models have been spe-

cially developed to take into account the energy sink

due to the paint melting on the hot strip and radiative

transfer between walls and conduction in the strip.

An overall view of the surface meshing used to mod-

el the strip moving in a fixed channel is shown in

Figure 13 (left). Figure 13 (right) displays a typical gas

temperature mapping and a projection of the velocity

vector field in a channel slice.

Figure 11: Effect of the channel

Reynolds number on the mean Nusselt number

Figure 12b: Coherent struc-

tures; in a hot and steady duct

(blue-green); in a hot and ro-

tating duct (red-yellow). LES

obtained with Fluent

Figure 12a: 2D cut of the 3D LES grid

for the flow and the solid

Figure 13: Surface meshing (left) and top

mapping of the gas temperature and velocity

vector field in a channel section (right).

Obtained with Fluent, using k- model

103 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Turbulent Flow and Heat Transfer

104

SIMULATION OF TURBULENT SHEAR

LAYERS - DNS AND LES APPROACHES

Turbulence can be thought as a cascade of

flow structures of decreasing scale. Two

extreme options are open for its nu-

merical simulation: the first is to

resolve all the relevant scales oc-

curring within the simulated flow,

which is the aim of he Direct

Numerical Simulation (DNS); the

second one is to model all the tur-

bulent scales through an ad hoc

model, which is done in Reynolds-

Averaged Navier-Stokes (RANS) formulation. A bridge

between the two previous approaches is to adopt a

model only for the smallest scales, as done in Large

Eddy Simulation (LES). During recent years, the VKI

has focused on the development and the application

of the latter approach.

The LES approach to the modelling of the turbulence

is based upon the idea to decompose the velocity (and

any other state or transport variable) into two parts:

one representing the interaction between the turbu-

lence and the mean flow (i.e. the action due to the

biggest turbulent structures or large eddies) and the

other representing the contribution of the small scales

of turbulence. The turbulent structures belonging to

this range do not interact directly with the mean flow

but only with other turbulent structures.

The advantage of this concept lies in

the fact that the turbulent motions at

the level of small scales could be as-

sumed, at least in a first approximation,

to be isotropic and to exhibit an uni-

versal behaviour. Therefore, the influ-

ence of these small scales on the larg-

er structures could be modelled using

simple turbulent (or subgrid) models,

which should, in principle, apply to any

turbulent flow. The use of subgrid

models makes it possible to close the

mass and momentum conservation

equations for the large scales. In prin-

ciple, all the classes of flow which are

currently studied with the RANS ap-

proach could also be studied, with high-

er accuracy, with LES, albeit at the cost

of a CPU effort at least one order high-

er. Examples of possible applications of

LES could be: injection and combustion

processes; industrial processes; vehicle

aerodynamics; atmospheric flows and

pollutant dispersal; aeroacoustics.

The capability to accurately predict these flow prob-

lems would have a significant practical impact and the

final objective of ongoing research on LES approach

is, therefore, its application to flows of indus-

trial interest.

The VKI is active both in the development

of in-house LES solvers [TH23] for fundamental

research and in the application of commercial codes (

Fluent; CFD ACE+) to the investigation of complex

flows for possible practical applications. In what con-

cerns fundamental developments, an important ad-

vancement, for the Environmental and Applied Fluid

Dynamics department, has been the completion

of the development of a state-of-the-art LES solver

for incompressible flow [MP196,MP197]. The code

(http://www.vki.ac.be/~giamma/mioma/) makes use of

the Finite Volume discretization, structured grids and

the Multidomain frame to describe full 3D geometries

(Figure 1).

The code is written in an advanced C language, makes

heavy use of abstractions, data structures and state-

of-the-art programming tools (CVS, GSL, PETSc,

Doxygen, Maxima) and has been designed to be the

Figure 1: Multidomain

discretization of the

computational field

around a wall-mounted

cube. Pressure field

[TH23].

Figure 2: Coherent structures in the wake

behind a wall mounted cube, visualized

applying Q criterion [TH23]. Reynolds num-

ber Re = 3000. Flow-field simulated by LES

2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Turbulent Flow and Heat Transfer

105 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

platform for fundamental research on the LES ap-

proach itself (discretizations, subgrid models, bound-

ary conditions) and the study of the mechanics of tur-

bulence. At present time, research is focused on the

study of the flow around bluff bodies and the under-

standing of the physics and the behaviour of the co-

herent structures which control the dynamic behav-

iour of this class of flow, an example of which is

shown in Figure 2 [TH23], for the case of flow around

a wall-bounded cube.

Commercial LES solvers have been applied to differ-

ent cases of complex flow. Details of these applica-

tions can be seen in the relevant chapters (multiphase

flows and aeroacoustics), but for the relevance and the

novelty of the research one can highlight two cases.

The first one concerns the simulation of the motion of

droplets of liquid alumina created by the combustion

process in the solid fuel rocket engines of the boost-

ers of Ariane 5 (Figure 3), performed by applying CFD

ACE+ [MP198]. The second one concerns the simula-

tion of the impact of a jet on a wing (Figure 4) to cre-

ate a realistic sound source for the study of the noise

created by propellers and helicopter blades, per-

formed by applying Fluent.

POST-PROCESSING OF NUMERICAL AND

EXPERIMENTAL DATA

In modern experimental and numerical techniques

such as PIV and LES, researchers are often producing

a huge amount of data. A relevant exploitation of

these data cannot be restricted anymore to classical

statistical techniques such as maps of ensemble aver-

age or of standard deviation. In front of this new need,

an important effort is dedicated at VKI to the develop-

ment and use of alternative post-processing techniques.

All of them share a key point; the definition of coher-

ent structures. These structures are present in an in-

stantaneous picture of the flow and they are thought

to be responsible for the concept of energy cascade

from large scales to small scales.If a clear definition of

these structures is not yet fully admitted, several cri-

teria exist allowing their detection. These criteria

(namely Q,

2

) are based on quantities derived from

the tensor gradient of velocity [TH1,TH23].

Qis the second invariant of the characteristic equation

associated with the tensor gradient of velocity.

When this variable is positive, the asso-

ciated flow is locally dominated

by pure rotation compared

to pure shear.

Figure 3: Coherent structures in a geometry

representative of a solid fuel booster,

visualized applying Q criterion [MP198].

Flow-field simulated by LES

Figure 4: Coherent structures in the flow of a round jet impacting a wing,

visualized applying Q criterion. Flow-field simulated by LES.

Figure 1:

Q visualization around

a circular cylinder (sign of the

streamwise vorticity). LES obtained with

VKI EA code MIOMA

Turbulent Flow and Heat Transfer

106 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

In Figure 1, this variable is used to visualize the shape

of a von Karman street produced by the flow around

a circular cylinder. The vortical rollers produced by the

shedding of the shear layers are linked by streamwise

vortical filaments illustrated in blue and red, depend-

ing on the sign of the streamwise vorticity. This Q cri-

teria has proved to be very efficient to illustrate the

body of a coherent structure in 3D.

2

is a variable representing the second eigenvalue of

the tensor where is the symmetric part of the ten-

sor gradient of velocity and is the asymmet-

ric part. When this variable is negative,

it has been proved that locally the

flow is dominated by rotation and

not by shear. At VKI, this criteria is

mainly used in PIV vector fields to

recognize the region where the

patches of vorticity are associated

to rotation but also to filter out the

patches of vorticity due to attached

shear layers. In Figure 2, a vortici-

ty field obtained from PIV is presented.

Figure 3 displays the associated

2

field

where only the area of vorticity dominated by

pure rotation is conserved.

With this criterion, it is now possible to do condition-

al averaging which allows the display of the part of the

classical ensemble average that is due only to a co-

herent structure belonging to a certain class [MP55].

In Figure 4, such an average

around a typical coherent structure

present in a channel flow is presented. The

streak-like coherent structure is formed by the en-

semble average of similar structures found close to

the wall. The length is displayed in wall units. The in-

clination towards the center of the channel should be

noticed.

Such conditional averaging may also be proposed

when only data associated with a range of Q values

are allowed to enter in the ensemble average

[MP215,MP194]. This new statistics is thought to be the

best tool to discriminate the impact of the coherent

structures on the mean flow. Figure 5 proposes the

Figure 2: Vorticity above a backward

facing step obtained from a PIV vector field

Figure 3:

2

field obtained from the PIV vector

field. The vorticity dominated by rotation is

displayed and associated to coherent structures

Figure 5: Streamlines in the middle plane of a

periodic rib duct flow. Top: classical ensemble

average, Bottom: Ensemble average of the data

associated with the higher values of Q.

LES obtained with Fluent

Figure 4: Conditional averaging of the flow

around a typical coherent structure in a plane

channel. LES obtained with

VKI EA code MIOMA

Turbulent Flow and Heat Transfer

107 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

comparison between a classical ensemble average

(top) and an ensemble average of the data associated

with the higher values of Q (bottom).

These criteria are now routinely used at VKI to recog-

nize a coherent structure from a 2D or 3D flow.

Moreover, an automatic algorithm to collect coherent

structures has been developed at VKI [AJ58,AJ37]. This

algorithm uses the wavelet properties to automatical-

ly recognize and extract isotropic coherent structures

that are also associated with high negative values

of

2

. Figure 6 shows that, for the flow over

a cavity at three different Reynolds

number, the wavelet

based algorithm has

proved to be very effi-

cient in automatically

analysing thousands of PIV

images.

As a reminder, a wavelet

may be presented as a lo-

cal filter. This filter is char-

acterised by a zero mean

and a support on which

the wavelet function is

not zero. There are many

wavelet functions but one of

them has been successfully used at VKI

to extract isotropic coherent structures; i.e. the

Maars Mother wavelet. This wavelet resembles a

Mexican hat (Figure 7) and is stretched or dilated to

compose a full wavelet family having a common

shape and displaying a full range of supports or scales

values.

The vorticity field obtained from PIV images is pro-

jected on the wavelet space. The position and the scale

of the wavelet associated with both the higher coeffi-

cients and to a clearly negative value of

2

are kept.

This information is translated in terms of position and

size of vortices as may be seen on Figure 6.

At this point, a full range of statistics based on the lo-

cal population of vortices may be proposed (Figure 8).

As an example of its application, this figure proposes

a local concentration of vortices extracted from the

data of Figure 6.

Besides this algorithm, re-

search is pursued in advanced

post-processing providing a

novel way to handle the ac-

quired information. The Proper

Orthogonal Decomposition be-

longs to this category [MP186]. This tech-

nique provides a mathematical base on which

the projection of a data set is energetically optimal, i.e.

the information is reduced to the minimum. These

properties have been used at VKI to extract from a data

set of PIV images uncorrelated in time the more prob-

able coherent structures (Figure 9).

Figure 8: Local statistics of the population of

vortices extracted with the wavelet algorithm

over a cavity at three different Reynolds number

Figure 6: Automatic extraction of positive (blue)

and negative (red) vortices done with the VKI

wavelet-based algorithm

Figure 7: Maars

Mother wavelet used

to extract isotropic

coherent structures

Although this method is attractive, our research

demonstrated the danger of a blind analysis of the

POD modes, forgetting that their mathematical nature

may be more important than their physical one. This

last drawback is balanced by the tremendous ability

of the POD technique to compress the information or

to synthesize the information in order to feed a LES

simulation with a synthetic reconstruction of the flow

with the help of some POD modes.

Turbulent Flow and Heat Transfer

108 2006, von Karman Institute for Fluid Dynamics, Rhode-St-Gense, Belgium

Figure 9: POD mode of the decomposition of

a data set of PIV images obtained over a cavity

- David Tamayo Case StudyTransféré parDavid Tamayo
- Colebrook White.( 1939)Transféré parcarloscartasine
- FLUID LAB FrictionTransféré parKazankapov Nurlan
- Tese ImportantTransféré parGustavo Ronceros Rivas
- ChE 190 Summary of JournalsTransféré pardatUPstudentdoe
- turbulent flow.pdfTransféré parSudeesh Patel
- Homework 3Transféré parRaúl Spinoza
- A Parallel Adaptive Mesh Refinement Algorithm for Predicting Turbulent Non-premixed Combusting FlowsTransféré parMichael McDonald
- 7.Moin BFSTransféré parpratikmitra30
- 2018 KUHNEN Destabilizing Turbulence in Pipe FlowTransféré parHeric Silva
- biotechnology lab reportTransféré parEpíck Picqúe
- CC15 Laminar FlowTransféré par999impact
- -03-28-heatexchanger-fan.pdfTransféré parPrathmeshBhokari
- Lab 7 OsbornedTransféré parAbdur Rashid
- VenturiTransféré parAnurag Kumar
- renold #Transféré parOsama Hazaymeh
- inestabilidad poisuielle turbulTransféré parfrankkqqzz
- Jirka 1994 Turbulence in Open‐Channel FlowsTransféré parJoão Paulo Mendes Ferreira
- Fulgosi_2003.pdfTransféré parbrian790
- Chap1_n3 chassaingTransféré parCARLOS
- USACE - River HydraulicsTransféré parHarlysson Maia
- karman_dpTransféré parvijay10484
- Wrap Thesis Jouvray 2003Transféré parJacob Chezz
- Classical Turbulence ModelingTransféré parValkyrie Guild
- Conclusion ThermofluidTransféré parKhairul Ikhwan
- 2. IAHR (2005)Transféré parHarris Koftis
- UnitOpI(12-2) 060116 (1)Transféré parHairi Hang
- iahr-ke2Transféré parAhmad Sana
- Fluid Flow Regime INTROTransféré parKyle Saylon
- Experiment 1.docxTransféré parJoenel Fumera

- Flow in Turbo MachinesTransféré parosoreneg
- 1205763481Batchelor. Introduction to Fluid DynamicsTransféré pargioio
- The Mathematical Foundations of Mixing(Rob Sturman, Et Al)Transféré parAna Mora
- Cfd MethodsTransféré parAna Mora
- Conventional MeasurementTransféré parAna Mora
- Multiphase FlowsTransféré parAna Mora
- Biological FlowsTransféré parAna Mora
- Optical MeasurementTransféré parAna Mora

- XPLANE CS LawsonSoftwareTransféré parPatricia English
- Clarence Gravlee et al. - Heredity, environment and cranial form: a reanalysis of Boas'Transféré parNahts Stainik Wakrs
- Choosing a Mixed Methods Design (PPT07)Transféré parcelin_p
- Eamco Ele 004Transféré parMalik Owais
- cellandmollecularbiocatalog.pdfTransféré parphyaravi
- Favour Able Conditions for IntercomprehensionTransféré parErik Hemming
- CRYPTOGRAPHIC HASH KEY ALGORITHM TO MITIGATE WORMHOLE ATTACKS AND LURE CATCH ALGORITHM TO BLOCK THE ATTACKERSTransféré parIAEME Publication
- An Academic Cloud Framework for Adapting.pdfTransféré parannaicsdepartment
- Navi 1 Navi Xt User Guide 1.06Transféré parAshleigh White
- New vbTransféré parsyedwaliullah
- Spong_Robot Dynamics and Control - SECOND EDITIONTransféré parLuon Chan Dong
- The Mole and Amount of Substance in Chemistry and Education (1)Transféré parMontoya Duque
- EE6411 2009 Unit00 Introduction c ChapterTransféré parlila001
- SAC Fabian Johnson v. CCSF (2)Transféré parJoe Eskenazi
- Ecoc 2014 Final ProgrammeTransféré parBenedikt von Lindewiese
- Unix File SystemTransféré parsankalpsardana
- The Political Unconscious[1].Doc2Transféré parMohammed Jhilila
- ant_colony.pdfTransféré parvarunsingh214761
- European Food Law Handbook PDFTransféré parPhilip
- GXT2_datasheetTransféré parClayton Martinez
- MK 0071 ReleaseNotesTransféré parzacve
- sample-analytic-rubricTransféré parapi-287380814
- Doble - Martin and Moore - Boston 2013Transféré parAmmar Khalid Mayo
- Super Minds Level 1 Teacher's Resource Book With Audio CDTransféré parAeliana Ahmad
- Perdev - Lp Week 3Transféré parJenny-Vi de Dios
- Ole Wæver 10 books - politik versionTransféré parOleWaever
- Module 17 Asteroids.pdfTransféré parJohn Van Dave Taturo
- Organization CultureTransféré parSaba Amir Gondal
- cbse sample papers for class 7 science sa2.pdfTransféré parnavneet
- Passages for Note MakingTransféré parDeepak Dhariwal