Vous êtes sur la page 1sur 52

Fluent News


Reverse Flow
Catalytic Converter
Heats Up
A Computational
Cure for Tires
Rapid Flow
Modeling for PLM
Dr. Ice and his
Skeleton Crew





RBAN CANYONS. The phrase brings to mind

images of city streets lined with skyscrapers and either
raging winds funneling through narrow passageways
or excessive heat that cannot escape. Architects and
city planners pay close attention to these phenomena,
and try to assess the role that building geometries and
sitings play on the local climate. While the weather can
be used to augment the internal HVAC system of a
building, excessive wind can create hazards for
passersby and be responsible for spreading the exhaust
from automobiles and smokestacks to more distant
areas. In this issue, we focus on weather and its impact
on buildings and their surroundings in a series of articles. The series starts on page 5, where the combined
effects of wind and solar radiation on a college building are analyzed. A simulation of the wind patterns
near a Belgian landmark appears on page 7, and the
role of wind in dispersing pollutants from cooling
towers is reviewed on page 8. The measurement of
wind and airborne particulates by the French national
weather service is the focus of an article on page 10,
and the impact of wind on measurements of rainfall
appears on page 12.
With the 2006 Winter Olympics around the corner, we
are pleased to report on two cases where sports
equipment is being improved through the use of CFD
(p. 16-17). The Sports Engineering Research Group
(SERG) in the UK has used FLUENT to simulate the
flow around a skeleton rider, and the Institute for
Research and Development of Sporting Equipment
(FES) in Germany has used it to study the four-man
bobsled. Watch for these teams when the events take
place in Turin, Italy in February. A number of other

interesting applications of CFD also appear in the

newsletter, and include waste water treatment (p. 14),
a novel catalytic converter (p. 20), and the curing of a
rubber tire (p. 26).
We are pleased to introduce two new products in this
issue: FLUENT for CATIA (p. 27) and studentFLUENT
(p. 36). The benefits brought by an existing product,
FloWizard, to a companys product design cycle are
reported on page 28. The Support Corner takes
another look at drag laws in FLUENT (p. 32), following an introductory article on the same topic in the
summer issue of Fluent News.
In early November, the heavy capacity Ariane 5 ECA
launcher delivered two communications satellites into
orbit. The launch was celebrated by the European
Space Agency and the dozens of companies who have
contributed to this program. As a tribute to Ariane 5,
our entire supplement is devoted to stories from
European companies who have worked towards the
success of this program. The stories show how FLUENT
has been used to simulate cryogenic flows (p. s4, s8,
and s10), heat transfer (p. s4, s8, s12), free surface
phenomena (p. s4, s10), and instabilities (p. s10, s14).
As the seasons change, seek shelter from the weather
and enjoy this issue of Fluent News. We look forward,
as always, to hearing about your work. 


Fluent News is published by

10 Cavendish Court Lebanon, NH 03766 USA

1-603-643-2600 www.fluent.com
2005 Fluent Inc. All rights reserved.
Editor: Liz Marshall
Assistant Editor: Susan Wheeler
Contributing Editors: Erik Ferguson and Keith Hanna
Design: Lufkin Graphic Designs
FlowLab, Icepak, Airpak, and FloWizard are trademarks of
Fluent Inc. Icepak and Airpak are joint developments of
Fluent Inc. and ICEM-CFD Engineering. All other products
or name brands are trademarks of their respective holders.

Pathlines and surface temperatures on a building at
Michigan Technological University, computed using
RadTherm and Fluent; temperatures from a thermal
imaging camera are shown across the middle
Courtesy of ThermoAnalytics and Monte Consulting


Pressure oscillations in the combustion
products inside a solid rocket motor
Courtesy of Avio Spa
Service Optique CSG




Solar Loads in
FloWizard Conjures
up the Atomium









Cooling Tower


Research Activities
at Mto-France


Gauging Rainfall


Waste Water Treatment
Gets an Oxygen Boost


Shape Optimization of
a Defroster Duct


Berlins Olympic


Reverse Flow Catalytic

Converter Heats Up


Dr. Ice and his

Skeleton Crew


Incipient Cavitation in
a Steering Rotary Valve



Distilling Exergy


Lithium Jet Hydraulics


A Computational Cure
for Radial Tires

Fluent News Fall 2005


FLUENT for CATIA: Rapid Flow
Modeling for PLM


FloWizard at MMA


Quality & Reliability in
Engineering CFD Simulations


Drag Laws 102


CFD for Future Engineers


Airfoil Noise in a Turbulent Jet


studentFLUENT Goes to College











Ariane 5 Reaches
for the Skies


Space Engineering
Activities at


Cryogenic Flows
in Rocket Engines



Fluent News Fall 2005

The Path to


Ariane 5 Internal
Cavities Beat the


Pressure Oscillations
in Solid Rocket


Solar Loads
in Northern Climates
By Craig Makens and Amit Shah, ThermoAnalytics, Calumet, Michigan, USA
and Matthew Monte, Monte Consulting Company, Houghton, Michigan, USA


use of multiple analysis codes to account for all
desired phenomena. As an example, an approximate
representation of the Raymond L. Smith Mechanical
Engineering Building on the campus of Michigan
Technological University in Houghton, Michigan was
recently used for an environmental simulation in the
presence of transient solar radiation. The model
demonstrated the combined use of FLUENT and
RadTherm software for architectural analysis.

solar shadowing, solar loading through glass, and

infrared band heat transfer are rapidly computed in
RadTherm. Interior heat loads, infiltration and ventilation effects are also accounted for in the thermal
calculation. This combined approach allows architects and building designers to more accurately
improve the thermal performance of each zone of a
planned structure, to test the efficacy of energysaving devices like low-e windows, or to develop
passive thermal heating and cooling designs.

Traditional HVAC and energy design makes use

of empirical methods based on one- and twodimensional approaches. These methods, while very
powerful and established, often neglect wind effects
and many other relevant situational factors, such as
the surrounding terrain and neighboring structures.
By combining FLUENTs accurate prediction of wind
flow around a building with RadTherms comprehensive environmental analysis, transient diurnal
thermal results can be generated with minimal computation time. Wind flow patterns and eddy current
effects are thus captured in FLUENT, while accurate

The R. L. Smith Building geometry was created in

Rhinoceros and meshed with ANSA for thermal
analysis in RadTherm. The thermal model geometry
consisted of 32,000 surface quads, including interior
walls and floors. A preliminary thermal analysis was
performed using RadTherm's built-in wind model.
The building exterior wall temperatures were then
exported to FLUENT to be used as a boundary
condition profile in the next phase of the solution.
For the FLUENT calculation, a high resolution CFD
mesh of 1.5 million tetrahedral cells was generated in

A cutting plane through the fluid domain shows the wind velocity magnitude; the building geometries are
colored by surface temperature with exterior walls ghosted to show internal rooms
Postprocessed in EnSight

Fluent News Fall 2005


GAMBIT. A steady-state analysis was performed using a bulk air

temperature of 6.8C and wind speed of 5 m/s from the north,
representing a cold north wind a common occurrence for
autumn in this location. The flow analysis in FLUENT captured
advective effects as the wind moved around the structure.
After the flow analysis was completed, the standard FLUENT
menu export command was used and "RadTherm" chosen as
the file type. This command exported a Patran Neutral file
containing surface mesh geometry and convection data
(convection coefficients and fluid temperatures on an elementlevel basis). These data were then imported into RadTherm and
mapped onto the lower resolution geometry as a boundary
condition for the external surfaces of the buildings. Interior
building surfaces retained the 1D fluid nodes used for natural
and forced convection computation in RadTherm. A complete
multimode thermal analysis was then carried out, including
transient solar conditions with loading through the windows.
The solar model considers global position, time of day, cloud
conditions for predicting direct and diffuse solar loads, surface
characteristics and glass characteristics. Post processing was
carried out in EnSight, courtesy of CEI.
This model illustrates the comprehensive in-situ analysis that
FLUENT and RadTherm can provide to buildings and downtown areas where complex flow and radiation effects render
traditional empirical methods inadequate. The approach used
in this model provides architects and building designers with
more accurate heating and cooling requirements, which leads
to proper and efficient HVAC sizing for each zone. The
increased efficiency yields fast returns on the engineering
investment, especially during high cost energy markets. 


Particle trace of wind pathways colored by velocity magnitude. Note the high speed wind
canyon between the R.L. Smith Mechanical Engineering Building and the Chemical
Sciences Building to the east (right). In November 1994, several windows in the northwest
area of the Chemical Sciences building were cracked by high speed winds. The windows
had no mechanism to be opened, so it was not the result of them slamming shut, but
rather the intense flexing within their rigid frames. The Smith building was constructed
in the early 70s, and no wind analysis was done at the time. Wind canyons between these
buildings can be quite severe, as many students and professors can testify

Diurnal surface temperatures of buildings and terrain,

viewed from the southeast; varying environmental
parameters include diffuse and direct solar loads,
direct solar angle, cloud cover, and apparent sky

Postprocessed in Ensight

Postprocessed in EnSight

Fluent News Fall 2005


FloWizard Conjures up the

By Corine Chauvin, Fluent Benelux


of the flow around a proposed new
building is an important step prior to
construction. In particular, the intensity of the wind at human height (2m)
near the base of the building is a key
component to pedestrian comfort.
FloWizard now makes it easy to accomplish such simulations quickly. The
building geometry can be taken from
the architects CAD software, loaded
into FloWizard, and the flow region
around the building will be automatically created and meshed. By specifying
the wind speed and direction, the flow
conditions in various regions can be
readily determined.

As an example, the airflow around one

of the most famous European monuments, the Atomium in Brussels,
Belgium, has been computed. This
structure, with a total height of 100
meters, was built in 1958 for the
Brussels Worlds Fair. It is a representation of iron atoms in a unit cell (a
repeating structure in a solid). At the
time of its construction, experimental
studies were performed in a wind tunnel to test the wind loads for a variety of
conditions. What took months of physical testing back then can now be done
using FloWizard in only a few hours!
The challenge for the CFD simulation
was to capture the airflow around the

Mesh on the Atomium

building and contours of
pressure on the surface

arms of the monument, and behind

the spheres. At the inlet to the simulation domain, a large rectangular volume containing the monument, a
normal constant velocity of 10m/s
(22mph) was applied. A fine mesh
with a boundary layer near the walls
was used along with the realizable k-
turbulence model. After a few hours
of run time, a solution was generated.
It showed a low velocity region
behind the building and local recirculation zones under the building and
behind the spheres. As the millions of
visitors who have visited the monument over the years would agree, the
wind currents near the base were not
found to cause extreme discomfort. 

Velocity vectors on the

central cutting plane show
regions of recirculation

The Atomium monument in Brussels

Fluent News Fall 2005


Chalk Point Coal Fired Power Station (2640 MW), Maryland

Courtesy of Power Plant Research Program, Department of Natural Resources, Maryland

Cooling Tower Drift

By Robert N. Meroney, Emeritus Professor of Civil Engineering, Colorado State University, Fort Collins, Colorado, USA


mechanical and natural draft cooling tower installations can contain water treatment chemicals that
can be hazardous if they make contact with plants,
building surfaces, or human activity. Prediction of
drift accretion is generally provided by analytic
models as found in the US EPA-approved ISCST3 or
SACTI codes. However, these codes are not suitable
when cooling towers are located in the midst of
taller structures and buildings. A CFD calculation
including a Lagrangian prediction of the stochastic,
gravity-driven trajectory descent of droplets is a
better approach in this kind of environment. One
such calculation has been performed and compared
to data from the 1977 Chalk Point Dye Tracer
Experiment in preparation for using such methods
in more complex building configurations. The
numerical analysis predicts plume rise, surface concentrations, plume centerline concentrations, and

Fluent News Fall 2005

surface drift accretion within the bounds of field

experimental accuracy.
Estimation of the impact of cooling tower drift on
the downwind deposition of droplet-born toxins is
difficult. A few field studies performed between
1965 and 1984 examined cooling tower plume
rise, visibility, and downwind concentrations.
Unfortunately, only a couple of these actually measured deposition rates downwind. Despite limited
field data, concern about drift and deposition led to
the development of more than a dozen separate
analytic models to predict downwind ground-level
concentrations and accretion rates. Chen [1] compared ten drift deposition models using a set of
standard input conditions for a natural-draft cooling
tower, and found that most of the models agreed
within a factor of three. However, when all ten models were compared, the predicted maximum drift

Particle-laden exhaust flows in a typical urban setting where cooling towers emit 300 micron particles in an
8.5m/sec exhaust stream, using reference wind speeds of 5m/s at an angle of 240 from true North; pathlines
are shown at left, and particle tracks are shown at right


Particle tracks downwind of the modeled Chalk Point

cooling tower and deposition regions located at 500
and 1000 m downwind

Calculations for the Chalk Point Cooling Tower

simulation were performed using FLUENT on a
domain 2000m long, 1000m wide and 500m high,
using 165,000 tetrahedral cells. The simulated
hyperbolic cooling tower height was 124m, with a
diameter of 54.8m at the tower exit. The plume vertical exhaust speed was set to 4.5m/sec, and mean
wind speed profiles were set to field values of 5m/s
at a height of 100m. Rather than specify the actual
temperatures, virtual temperatures were used to
account for the water vapor content in the plume
mixed with the ambient humidity of the background atmosphere. The plume virtual ambient

The height of the centerline of the cooling tower

plume was determined based on the height of the
maximum in the water vapor and temperature profiles downwind of the cooling tower. The calculated
points agreed very well with the predictions of the
Briggs plume-rise formula calculated by Hanna [3] as
well as with the trend of the visual observations for
plume height recorded during the experiment.
Predictions of ground level and plume centerline
water vapor concentration were compared to values
predicted by the ISCST3 program, and the agreement was within 25%. The calculations were done in
terms of log K factors, where K is the dimensionless
water vapor concentration: CUref / Qsource, where C is
the actual concentration, Uref is the approaching
wind velocity at the cooling tower release height, and
Qsource is the water vapor content of the exhaust
emissions at the cooling tower exit. Particle tracks for
a typical Rosin-Rammler particle distribution release
with a mean diameter of 0.09mm and spread
parameter, n, of 0.65 were also examined. The
calculated deposition accretion magnitudes were
compared to observed and analytic values predicted
by ISCST-3 and Hanna [3]. The CFD grid face values
for the specified inlet profile and Rosin-Rammler
representation of the Chalk Point source droplet
distribution agreed within factors of 0.75 and 0.5 at
0.5 and 1.0km, respectively. 


Chen, N.C.J.: A Review of Cooling Tower Drift

Deposition Models; Oak Ridge National Laboratory,
ORNL/TM-5357, 1977.
Policastro, A.J.; Dunn, W.E.; Breig, M.; Ziebarth, J.:
Comparison of Ten Drift Deposition Models to Field
Data Acquired in the Chalk Point Dry Tracer
Experiment; Symposium on Environmental Effects of
Cooling Tower Plumes, U. of Maryland, May 2-4, 1978.
Hanna, S.R: A Simple Drift Deposition Model Applied to
the Chalk Point Dye Tracer Experiment; Symposium on
Environmental Effects of Cooling Tower Plumes, U. of
Maryland, PPSP CPCTP-22, WRRC Special Report No.
9, May 2-4, 1978.

Plume rise (m)

Once the overall flow and turbulence fields were

calculated, the Lagrangian discrete phase model
(DPM) used a sample of this data to predict the
downwind distribution of a phase distribution
equivalent to measured field cooling tower exit
values. Ground level accretion of the particles was
noted at the 0.5 and 1.0km distances downwind of
the cooling tower.


Briggs, 1975
Observed, 1977



X (m)



Cooling tower plume rise comparison

Plume concentration (m-2x106)

Results from the 1977 Chalk Point Dye Tracer

Experiment are described in papers and reports by
Hanna [3]. These experiments are considered to
have produced the best single source of cooling
tower deposition data available. Two natural draft
hyperbolic cooling towers are located on the Chalk
Point site in Maryland, on a peninsula that extends
into the local bay and wetlands. The two towers and
the turbine building are located along an east-west
line, and are separated from one another by about
500ft. The hyperbolic cooling towers are 400ft
(124m) tall, 374ft (114m) in diameter at the base,
and 180ft (54.8m) in diameter at the exit.
Instruments to measure drift deposition were placed
at 5 intervals on 35 arcs at distances of 0.5 and
1.0km north of the cooling towers. The average
deposition rate of the dye-tagged sodium droplets
on the 0.5 and 1.0km arcs was 1080 and
360kg/km2/month, respectively. Drift droplet sizes
at the measurement stations had a mass median
diameter of 340 and 260m on the 0.5 and 1.0km
arcs, respectively. Most of the drop sizes were
between 250 and 450m on the 0.5km arc and 200
and 400m on the 1.0km arc.

temperature was set to 295.3K, and the virtual

exhaust temperature was set to 315.3K. Buoyancy
was included in the calculation.


ISC-3 Uref = 8m/s





X (m)

Predicted plume centerline concentration

Deposition (Kg/m2-s)

deposition differed by two orders of magnitude,

and the downwind locations of the maximum differed by one order of magnitude. These comparisons occurred before improved sets of field data
from the Chalk Point Dye Tracer Experiments
became available (after 1977). Policastro et al. [2]
compared most of the same drift deposition models
to the new Chalk Point experimental data, and concluded that None of the existing models performed well. A number of researchers have used
CFD previously to calculate cooling tower plume
behavior, but none of the CFD calculations found in
the literature predicted deposition levels downwind
of cooling towers.


ISC3 Drift and depletion

Observed, 1977
Hanna, 1978



X (m)



Deposition observed and predicted

Fluent News Fall 2005


Research Activities at
By Philippe Nacass, Mto-France, Toulouse, France


Weather Service), the Centre National de Recherches
Mtorologiques (CNRM, or The National Center for
Meteorological Research) is the department responsible for conducting most of the organizations meteorological research activities. The center is primarily
oriented towards the needs of the public in the areas
of meteorology, weather forecasts, and the physics
and dynamics of the atmosphere. Its work also covers
related fields, such as atmospheric chemistry (acid
rain and ozone), surface oceanography, the physics
and dynamics of snow cover (avalanches), surface
hydrology (floods), and urban pollution.
At CNRM, the development of new-generation
atmospheric models is an ongoing effort. To carry
out this mission, CNRM hosts approximately 225
permanent staff (one-third being research scientists), and 45 students and visitors. At the national
level, the research is conducted in close cooperation
with many universities and atmospheric laboratories, such as the Centre National de Recherche
Scientifique (CNRS, the National Center for
Scientific Research). At the international level,
CNRM collaborates with the European Center for
Medium-Range Weather Forecasts (ECMWF) in the
United Kingdom and the National Center for
Atmospheric Research (NCAR) in the USA, among
others. The CNRM also participates in international
large-scale field experiments and multidisciplinary
research programs, such as the International
Geosphere-Biosphere Program and the World
Climate Research Program.
Some of the data collection at Mto-France is done
using atmospheric research ships and aircraft, which
deploy a variety of sensors for state parameters,
atmospheric chemistry, cloud physics, and remote
sensing. Instruments used for these measurements


Fluent News Fall 2005

have fundamental sources of uncertainty that can

often be quantified. One source of error is distortion
from the hull or fuselage of the ship or aircraft.
Movement of air around these bodies can impact
not only the flow speed and direction, but the concentrations of various constituents or particulate
matter being measured. In fact, airflow distortion
can generate errors that are larger than those inherent in the measurement sensor. To make the most
accurate atmospheric measurements, these errors
must be understood and minimized by a suitable
selection of sampling location.
To better understand the influence of objects on
measurement error, wind tunnel studies have traditionally been done, even though such tests are costly and time-consuming to perform. They are limited
by the wind tunnel speed and the physical size of
the model. Numerical modeling has also been
employed to simulate the airflow disturbance over
ships and around aircraft. From 1988 to 1994,
potential flow codes were used to understand
the flow characteristics in critical areas where sensors are placed. These codes provided reasonably
accurate estimates of flow and particle behavior at
locations outside the aircraft boundary layer.
In recent years, advances in CFD have permitted
faster, more accurate representations of airflow
around bodies. Airflow characteristics (speed and
direction, for example) and particle trajectories in
measurement regions can be predicted using CFD,
allowing better placement of research instrumentation and measurements of greater accuracy. Since
1995, Mto-France has used FLUENT for this
purpose, tasking some individuals with full-time
effort. CFD is currently being used to optimize
measurement equipment and its positioning on a
number of aircraft and ships.


The Aircraft Fleet at Mto -France
In cooperation with other French governmental organizations, Mto-France has
operated several research aircraft over the years, such as a Piper-Aztec, a MerlinIV, and a Fokker-27 [1]. For these aircraft, CFD has been used to correct in-flight
measurements made by sensors and instruments designed and mounted on the
fuselages prior to the introduction of CFD. In 2005, Mto-France, through
SAFIRE [2], began to operate two new instrumented aircraft, an ATR-42 (a biturboprop) and a Falcon-20 (a bi-turbojet). For these aircraft, CFD was used to
study the best position of the instruments, and to make sure that they, in turn,
have no harmful influence on the aircraft for all possible flight attitudes.

The research aircraft ATR-42, a bi-turboprop operated

as part of the SAFIRE project

The Design of New Airborne Instruments

Mto-France has also used CFD to develop new aircraft instruments, including
special sensors, and to optimize their shape. This effort has proved to be very
important for the design of sensor inlets that measure airspeed or aerosol
concentrations. For manufactured instruments, CFD is used to illustrate the
airflow disturbance in the volume of measurement. The aerodynamic forces and
moments on the outside body of these instruments are also calculated by
FLUENT for certification by the French Aviation Administration.

Airflow Studies Around Aircraft

CFD simulations for specific aircraft have proved to be valuable to the research
community, both for optimal sensor placement and for interpretation of measured data. In most uncertainty analyses, it is assumed that the air reaching the
sampling inlet or sensor is representative of the free stream atmospheric flow.
However, the movement of air around an aircraft fuselage can impact not only
the flow speed and direction, but the concentrations of various atmospheric constituents as well. The additional uncertainty caused by the fuselage is a function
of the location and type of measurement being made. For studies of aerosol
particles in clouds and in the atmosphere, for example, a wide range of particle
sizes (0.001 to 1000m in diameter) must be measured accurately. Their
distribution at various locations around the aircraft varies significantly from free
stream conditions, and CFD is useful for quantifying this discrepancy.

The airflow disturbance in the volume of

measurement for a manufactured sensor

Pressure is used to
compute the
aerodynamic forces
and moments on
the pylons mounted
under each wing of
the ATR-42; the
pylons are used to
carry additional

For studies of aerosol particles in clouds, the trajectories of particles ranging from 0.1m (left) to 100m (right) in diameter must be measured accurately
Fluent News Fall 2005



Airflow Studies over Research Ships

Measurements made from ship borne instruments are biased due to the effect of
the ship on the flow of air to the instruments as well as turbulence from the
air/water interface. The presence of the ship causes the air flow to a particular
instrument site to be either accelerated or decelerated, displaced vertically or, to
a lesser degree, in the horizontal direction. Although recognized for some time,
it is only recently that the problem has been addressed using 3D CFD models.
These simulate the flow over particular ships, quantify the effects of flow distortion, and hence correct the ship-based measurements.
Since 1998, Mto-France has used FLUENT to model the flow around one of
the research ships operated by a French Institute [3] for various relative wind
directions and wind speeds. It has used the resulting estimates of error in the
measured wind speeds to correct measurements of fluxes between the air and
the sea. Comparison of the data recorded by the ship with the CFD results has
suggested that the flow distortion on the measured wind speed is dependent on
the incident wind speed. Pathline traces of the predicted flow field, beginning far
upwind of the ship and passing through the instrument inlet, allows the vertical
displacement of the flow reaching the site to be estimated. These and other
CFD results have been useful to researchers who have obtained meteorological
measurements from aircraft or ships in the past or to those making comparisons
between ground, aircraft, ships, buoys and satellite data systems. 

The presence of a research ship can distort

the flow around it; contours of pressure
coefficient are shown

An illustration of how CFD has been

used to correct ship-based measurements

By Andrew J. Newman and Paul A. Kucera, Department of Atmospheric Sciences,


are required for many different applications such as
river/flash flood forecasting, water resource management, and agriculture. Precipitation measurements are retrieved through the use of a variety of
instruments located on the surface (e.g. rain
gauges) or from remote observations such as satellites or ground-based radars. Despite advances in
technology, rain gauges are still considered the
standard for surface rainfall measurements. Often,
these data are used for verification of remotely
sensed rainfall estimates. Therefore, it is important
to understand and quantify the errors associated
with the in situ precipitation estimates.

When using a rain gauge, a variety of errors can

occur due to calibration and sighting problems and
instrument failure. While these errors are mostly
preventable, environmental conditions can lead to
wind induced catchment errors that are not generally preventable. Catchment errors occur because a
rain gauge modifies the environment in which data
are collected. Field measurements show that these
errors can range from 1-2% to over 10%, depending on the rainfall intensity and free stream wind
velocity [1]. A few prior studies have used CFD to
simulate precipitation gauges with results that have
fairly good agreement with observations [2, 3].
Using FLUENT, similar instruments are now being
modeled for comparison with previous studies, and
for the development of a tool for evaluating the flow
characteristics around new precipitation monitoring
As an example, the flow around a Qualimetrics
tipping-bucket rain gauge mounted on a 1m pole
has been studied. The computational domain is
1.8m x 1.2m x 2m and comprised of roughly
200,000 tetrahedral cells, with local refinement
around the rain gauge to better resolve the detailed


Fokker-27 was co-funded by Mto-France, the National Center for Scientific Research
(CNRS), the French Space Center (CNES) and the National Geographical Institute (IGN).

Service des Avions Franais Instruments pour la Recherche en Environnement (SAFIRE,

Facility for the French Aircraft Instrumented for Environmental Research), created in
2005 with Mto-France, CNRS and CNES.

Institut Franais de Recherche pour lExploitation de la Mer (IFREMER, French Research

Institute for Exploitation of the Sea).


Fluent News Fall 2005


University of North Dakota, Grand Forks, North Dakota, USA

flow there. A steady wind velocity with a height

profile was specified along with the standard k-
turbulence model. The results were compared with
one of the earlier studies [2], even though a slightly
larger wind speed (4m/s versus 3m/s) was used. The
results illustrate the acceleration of the flow over the
top of the gauge along with the formation of a
vortex in the gauge catchment area. A flow profile
such as this can cause raindrops, especially small
ones, to miss the rain gauge, giving rise to an
underestimate of the actual rainfall.

450,000 tetrahedral cells were used, again with a

refined mesh in the vicinity of the instrumentation.
The worst-case scenario has the wind coming from
behind the camera housing and flowing toward the
sample volume. The CFD results demonstrate that
this condition creates a shadow effect in the sample
volume that may cause under catchment of the
hydrometeors. Wind flow conditions that are not
along the camera line-of-sight do not affect the
hydrometeor sampling as much. 

The flow around an instrument that is used to
image hydrometeors (raindrops or snowflakes, for
example) has also been studied. The instrument,
called a disdrometer or Rain Imaging System (RIS)
and developed at NASA/Wallops Flight Facility by
Dr. Larry Bliven, photographs hydrometeors that fall
through a sample volume located between the
camera and light source. The RIS is unique because
it can provide measurements with less wind interference than other instruments of similar capability [3].
In a FLUENT simulation of this device, roughly

Sevruk, B.: Wind Induced Measurement Error for HighIntensity Rains. Proc. International Workshop on
Precipitation Measurement, WMO Tech. Document
328, 199-204, 1989. [Available online at

Nespor, V.; Sevruk, B.: Estimation of Wind-Induced

Error of Rainfall Gauge Measurements Using a
Numerical Simulation. J. Atmos. Oceanic Tech., 16,
450-464, 1999.

Nespor, V.; Krajewski, W.F.; Kruger, A.: Wind-Induced

Error of Raindrop Size Distribution Measurement Using
a Two-Dimensional Video Disdrometer. J. Atmos.
Oceanic Tech., 17, 1483-1492, 2000.

Velocity magnitude for the rain gauge simulation

Streamlines colored by velocity magnitude with static

pressure contours on the rain gauge

Contours of velocity for the Rain Imaging System with the worst case wind scenario, in which the wind blows
from behind the camera (left) towards the sample volume (right)

Pressure contours on the camera and mount and

pathlines colored by velocity

Fluent News Fall 2005



WasteWater Treatment
By Alban Poirier, Vincent Perrin, and Jrme Cluzeau, AIR LIQUIDE, Gaz Industriels Services, DAP, Les Loges en Josas, France




Turbine aerator


waste water treatment. Most of the bacteria that are
responsible for the decay of organic material are
aerobic, so the dissolved oxygen in the waste water
must be replenished by an outside source. The
aeration process allows bacteria and sludge to be
put into contact. Efficient oxygenation is, in fact,
essential to the success of aerobic biological treatment. In the most difficult cases, particularly with
industrial effluents, pure oxygen boosting (rather
than air boosting) is a very efficient solution; it can
be applied to most basins, even those that were not
originally designed for oxygen. CFD can be used
to validate the technical choices for oxygenation
without industrial risks and to predict the performance of existing and future basins.

Water outlets
Waste water treatment basin

Pathlines illustrate the flow field for high (top), nominal (middle), and low (bottom)
oxygen flow rates for VENTOXAL


Fluent News Fall 2005

In a recent project, FLUENT was used to characterize the hydrodynamic behavior of an industrial rectangular waste water tank in the presence of four
floating turbine aerators and two types of oxygen
transfer device: a TURBOXAL (floating at the basin
free surface) and a VENTOXAL (immersed in the
waste water basin). The goal of the project was to
simulate the initial performance of the basin, and
then to improve its performance before the installation of the equipment by optimizing two parameters: the location of the new equipment and the
flow rate distribution. Using a Lagrangian (DPM)
calculation, a discrete phase of oxygen bubbles was
coupled to the continuous phase of water, taking
into account the hydrodynamic effect of the oxygen
plume. Of particular interest were the oxygenation
homogeneity, the mixing efficiency, and the interaction of the plumes with the different equipment.
The presence of low velocity zones, which represent
a significant risk for sludge deposits (and the
development of filamentous bacteria) and shortcircuiting (hydrodynamic flows with low residence
time) were sought as well.


Gets an Oxygen Boost

Validation of these two AIR LIQUIDE oxygen boosting devices was difficult because of the high reciprocal impact of the gaseous phase on the flow. In
particular, one of the fitting parameters is the length
of the bubble streams, measured experimentally
and compared with the modeling results using
a typical bubble size (measured and correlated).
The VENTOXAL device was first simulated on an
instrumented biologic water treatment plant, and
provided computed velocity and concentration
fields for comparison with data. The validation
allowed simplifying modeling assumptions to be
identified. The TURBOXAL was also the object of a
preliminary 3D CFD simulation, used for the development of this new device, that took into account
the impeller rotation. These results also allowed a
simplified 3D model to be developed for the waste
water basin simulation that used values of axial,
radial and tangential velocities. A simplified model
of the turbine aerator was developed as well, based
on technical data supplied by the manufacturer.
The CFD simulation allowed the influence of
various parameters to be studied. For example,
varying the location of the oxygen boosting
equipment or modifying the flow rate distributions
were considered in order to improve the mixing
performance of the basin. The oxygen residence
time could also be optimized by displaying the
upward velocity of the plumes.

Validation of a simplified VENTOXAL CFD model

of an industrial basin

Cross-section of the velocity field 0.57 m above the

bottom of the basin, showing the low velocity zone
along the right edge

Once the basin began operating, the experimental

data showed very good agreement with the CFD
modeling results. The plumes were located exactly
where the CFD model predicted they would be, and
a very good quality of floc was observed at the exit
station as a result of a good velocity distribution.
Overall, the CFD modeling effort saved time
in determining the best implementation of the
equipment, and it will contribute to new oxygen
boosting projects in the future. 

Iso-surface of 0.1 m/s velocity,

showing the low flow zones in the basin
Fluent News Fall 2005




By Ralf Gollmick, FES, Berlin, Germany and Mathias Jirka, Fluent Germany

The German bobsled rounds the bend


(FES) in Berlin has been described as a gold mine, as the equipment that has
been developed there has often proved to make a difference during competitions. Since 1962, engineers, mathematicians, physicists, and craftsmen at the
FES have been involved in the development of customized sports equipment for
use by athletes in training and competition. Along with sports such as rowing,
canoeing, sailing, and cycling, the recent emphasis has been to focus on the
upcoming 2006 Winter Olympics in Turin, Italy. Of particular interest are the
sleds used for the skeleton and bobsledding events, where it is as important
to minimize friction as it is to keep the air resistance of the athletes and their
equipment at a minimum.
To achieve enhanced results in this area, the FES engineers have been performing
flow simulations using FLUENT. In their view, CFD has become indispensable for
equipment development in competitive sports. Over the years, hardware gains
have enabled numerical simulation of three-dimensional flow patterns with
ample precision and effective comparison between variants. At FES, the goal is
not to achieve a comparable result in relation to the wind tunnel, but to create
a foundation that makes systematic comparison between variants more reliable.
In the bobsled research, a Linux cluster fitted with AMD Opteron 64-bit
CPUs was used for the CFD simulations. Using the cluster, the calculation of a
complete model with six million cells could be completed in about an hour. A
dual-processor Intel Xeon 3GHz computer with 4 GB of RAM also accomplished
comparable tasks in an acceptable time.

The 4-man bobsled in the wind tunnel

The realizable k- model with standard wall functions was relied upon for all of
the computations. Initial trials with the shear-stress transport (SST) k- model
were also successful. The realizable k- model had the advantage of being welladapted for simulating regions of stalled flow. In addition, the flow underneath
the sled is known to generate increased resistance due to the shear between the
floor (base layer) and the upper layers of the fluid that are moving faster. To
investigate the effect more precisely, a moving base layer was used in the simulations in conjunction with the oncoming air flow boundary condition. Such
detailed scenarios could not be properly reproduced in a wind tunnel.
The simulations of different sled variants at FES led to the development of design
modifications. One change gained prominence because it caused a significant
reduction in resistance. The modification involved the shape of the hood, and it
resulted in better formation of the wake area and a subsequent decrease in the
strongly turbulent area behind the sled.

Dynamic pressure contours on the sled, ice, and athletes


Fluent News Fall 2005

As is done for Formula One (F1) race cars, the bobsled prototypes were tested in
wind tunnels at TU Dresden and the BMW plant in Munich. The results in those
cases were found to be in good agreement with the CFD simulations. After
having complied with all of the technical prerequisites for a good performance
in the Olympics, Germanys bobsleds should again be in peak position to return
from Turin with even more medals. 


Dr. Ice
and his Skeleton Crew

Pressure contours on a
simulated skeleton slider, with
pathlines colored by velocity magnitude
Postprocessed by Ensight

By David Curtis, Sports Engineering Research Group (SERG), Sheffield University, Sheffield, UK


its climate being conducive to winter sports.
However, with the Winter Olympics in Turin, Italy
coming up in February 2006, a new breed of athlete
has come along. Kristan Bromley, the top-ranked
skeleton bobsled competitor in the UK, and current
European Champion, is a leading British medal
prospect. He was also the first British male to win a
winter sports world cup series when he dominated
the 2003/2004 season. Bromley is perhaps unique
in the sport because he has a Ph.D. in ice sports
technology, and saw the value of CFD in his previous job with the defense company BAE. His passion
for skeleton grew while working on a BAEsponsored skeleton sled project, and it has led him
to dedicate himself to becoming the next Olympic
champion. Indeed his gold medal quest is tightly
coupled to his passion for engineering. He and his
brother Richard, who is also his coach, have set up
a pioneering technology company to support his ice
sports dream and coordinate his UK Sport-funded
performance program for the Olympics. Members
of the press in the UK have been fascinated by
this unique athlete, and have taken to calling him
Dr. Ice because of his sporting, technological, and
academic achievements.
In 2003, Bromley approached the Elite Sports CFD
Unit a part of the Sports Engineering Research
Group (SERG) in Sheffield, UK and asked them to
provide CFD flow simulation support to increase his
chances of success. SERG has had considerable success in the past with the British Olympic Cycling
team in Athens during 2004. Bromley maintains a
philosophy of using advanced technology to
enhance on-ice performance, We look to our
strengths in R&D to help bridge the gap we have
with the stronger winter sports nations, so that I can
compete on equal terms. His belief that elite-level

athletes also require elite-level support led him to

collaborate with SERG and make use of Fluents CFD
software, since both have a proven record of success
in elite sport.
Dr. John Hart and his colleagues in the Elite Sports
CFD Unit began by laser scanning a full-size flexible
mannequin on a skeleton sled, They then used specialist surfacing software to produce a high-quality
CAD surface on top of the scanned data. This prepared the model for import into GAMBIT to create
the required volume mesh. A computational mesh
suitable for CFD analysis was then constructed using
a combination of GAMBIT and TGrid. A typical
skeleton sled and mannequin mesh consisted of
approximately seven million tetrahedral and prismatic
elements. The prisms were required over the entire
surface of the modeled geometry to accurately
capture the surface boundary layers and flow separations. FLUENTs CFD solver captured turbulent
flow effects by using the realizable k- turbulence
model in conjunction with non-equilibrium wall
functions to accurately resolve the boundary layer
The initial area of CFD interest for Bromley has been
skin friction characteristics associated with the customized skin-suit in which he competes. He wanted
to assess small changes in surface texture in terms of
their impact on minimizing his overall aerodynamic
drag. The ultimate proof of the CFD work will come
in Turin when all of Bromleys hard work, his mental
and physical conditioning, and the technologies
behind his sled will be put to the test over the few
short minutes of the Olympic competition. 

Fluent News Fall 2005



Distilling Exergy Savings

By Ricardo Pulido, Leodegario Monroy, and Ricardo Rivero, Instituto Mexicano Del Petroleo-Exergy Group, Atepehuacan, Mexico
and Yi Dai, Fluent Inc.

T IS NO SECRET that oil makes the worlds economies go around. Because

oil processing can be expensive, energy intensive, and detrimental to our environment, efforts to make the oil refining process cleaner and more efficient are
ongoing at a number of research laboratories. At the Instituto Mexicano del
Petroleo-Exergy Group, engineers have been focusing on distillation. At
refineries, distillation is used to separate crude oil a mixture of hydrocarbon
compounds into a number of constituents. Of particular interest to this
group is diabatic distillation technology, whose benefit is measured in terms of
exergy savings. Exergy, simply stated, is usable energy that can do work, and
is a quantity that tends to decrease over time. It is emerging as an increasingly
useful measure of efficiency as process improvements are considered throughout the refining industry [1].

The diabatic distillation pilot plant

The geometry of one stage of the

diabatic distillation tower

Pathlines, colored by liquid naptha

volume fraction, illustrate the flow of
the liquid and gaseous components: gas
enters through the inlet at lower left,
passes around the plate at right, and
enters the upper region; liquid naptha
enters through a slot inlet on the top
left and drips down the left-hand plate
onto the horizontal condenser plate

In a classical adiabatic distillation tower, heat is supplied externally by means

of a heater. As the exhaust vapors rise, the temperature drops and condensation occurs on plates positioned in stages at different heights. The condensed
liquids are extracted separately, since condensation occurs at different
temperatures for the exhaust constituents. Any remaining exhaust vapors exit
after the uppermost condenser, carrying with them excess heat that is extracted externally by means of another condenser. In a diabatic distillation tower,
heat is supplied or extracted not only by an external source or sink, but from
process fluids inside the distillation process itself, which is why diabatic distillation is considered exergy efficient. There are a number of studies that support this technology for its future potential in the refinery industry worldwide
[2, 3, 4].
A numerical and experimental study is currently underway to gain a better
understanding of a diabatic distillation tower. In the first phase of the study,
the isothermal flow of liquid and gaseous naptha is considered in a typical
stage. The process simulation package ASPEN is used to generate gas and
liquid mole fractions and flow rates at various zones within the stage. These
are used as boundary conditions in a FLUENT calculation that makes use of the
volume of fluid (VOF) model to track the motion of the two fluids on a mesh
of 425,000 cells.
The early results have provided a clear view of where the liquid tends to
collect on the plates within the stage, and they are serving as a basis for the
first round of physical testing. Using the numerical results, proposals for
modifications to the tower will be made and tested in a quicker, less expensive
way than would be required if only experiments had been run. In the future,
heat and mass transfer (condensation) will be included, and different diabatic
designs will be compared numerically prior to fabrication, to save time
and money. 


Contours of liquid naptha volume

fraction show the liquid running down
both vertical plates and collecting on
the horizontal plate


Fluent News Fall 2005

United Nations Environment Programme. DTIE.


Kenney, W.F.: Energy Conservation in the Process Industries, ISBN 0-12-404220-1,

Academic Press, Inc. New Jersey 1984.

Linnhoff, B., Polley, G.T. and Sahdev, V.: General Process Improvements Through
Pinch Technology, Chemical Engineering Progress, June 1988.

Rivero, R.: LAnalyse dExergie: Application la Distillation Diabatique et aux

Pompes Chaleur Absorption, Thse de Doctorat, Institut National Polytechnique
de Lorraine, Nancy 1993.



Focus on CFD
For the Ariane 5 Launcher


Ariane 5 Reaches
for the Skies


Space Engineering Activities


Cryogenic Flows
in Rocket Engines


The Path to Passivation


Ariane 5 Internal Cavities
Beat the Heat


Pressure Oscillations
in Solid Rocket Motors



Ariane 5
Reaches for
the Skies
By Vincent Canu and Gilles Lebiez, Fluent France
and Keith Hanna, Fluent News

SPACE today
depends on
on the
the Ariane
Ariane family
family of
of heavy-duty
(610 ton)
ton) launchers
launchers and
and the
the future
future Soyouz
and Vega
Vega launchers
launchers for
for medium
medium and
and small
payloads (13
(13 tons).
tons). Since
Since their
their inception,
Ariane rockets
rockets have
have been
been the
the most
most successful
commercial satellite
satellite launcher
launcher systems
systems ever,
with Ariane
Ariane 44 alone
alone being
being responsible
responsible for
for the
launch of
of 182
182 satellites
satellites of
of all
all kinds
kinds in
in 14
14 years
before its
its retirement
retirement in
in 2003
2003 over
over 450
450 tons
in all.
The space
space industry
industry in
in Europe
Europe today
today employs
thousands of
of men
men and
and women
women in
in hundreds
of companies
companies across
across the
the continent.
continent. SpaceSpaceof
related systems
systems contribute
contribute aa tremendous
amount to
to our
our scientific
scientific knowledge
knowledge and
technological development
development in
in the
the modern
world. They
They are
are responsible
responsible for
for the
the hundreds
of satellites
satellites launched
launched since
since the
the 1970s
1970s that
have paved
paved the
the way
way for
for global
global positioning
systems in
in our
our cars,
cars, improved
improved weather
weather foreforesystems
casting and
and global
global climate
climate monitoring,
detailed mapping
mapping of
of our
our land
land masses
masses and
ocean depths,
depths, modern
modern high
high speed
speed telecomtelecomocean
munications, and
and increasingly
increasingly sophisticated
defense and
and security
security aids.

Ariane 5 (last heavy version) qualification

flight in February 2005. The launch was
performed during the day (it usually happens
during the night) in order to allow a more
detailed post flight analysis


Fluent News Fall 2005

Photo copyright ESA/CNES/ARIANESPACE Service Optique CSG



The Ariane 5 rocket was first put on the drawing board in 1987 and subsequently had its first successful launch in 1997 from the Guiana Space Center near Kourou, French Guiana. Its success is
the result of a unique collaboration between many European companies who all make different
parts of the launcher system. Over the years, engineers have faced immense technological challenges, including cryogenic propulsion systems, aerothermodynamics, fluid-structure interaction,
thermal protection, and aeroacoustics to name a few. A variety of in-house and commercial CFD
codes, including FLUENT, have been used to successfully model these diverse phenomena. In fact,
CFD has been used to model nearly every part of the launcher and launch process [1], such as:
Time (sec)


Prior to launch

Launcher roll-out and local climate modeling

0 to 7

Launcher main engine ignition

Solid booster ignition and take-off

30 to 50

Transonic flight


Flight at maximum dynamic pressure

540 to 550

Stage separation; ignition of upper stage engine and

upper stage engine plume impact on the main stage

By comparison with the Ariane 4 rocket, which relied little on CFD, Ariane 5 has made widespread
use of this technology for understanding, refining, and ultimately accelerating the space
transportation design process, within the constraints of safety and quality assurance demanded
of such a unique system. Of the many companies involved with Ariane 5, CNES, the French
governmental space agency, led its technological development in association with commercial and
academic partners such as EADS Space Transportation, Air Liquide, Snecma, Sener, Fiat Avio Spa,
and Cryospace, among others. CNES was formed in 1962 to assist the French government in shaping Frances space policy. It leads the programs funded by the French government and represents
France at the European Space Agency (ESA) and in international space activities and partnerships.
It has about 2,500 personnel at four sites across the world, and in addition, it promotes and
encourages space applications and industrial space related R&D. For 40 years it has been involved
in driving the design and development of European Launchers with respect to safety, production,
quality assurance and launch ground facilities.

Classical launch in geostationary transfer orbit (GTO)

(with reentry of the main stage into the sea), followed by
satellite maneuvers to reach geostationary orbit (GEO);
solar arrays are then unfolded to get the operative
configuration of the spacecraft
Courtesy of CNES

Real tests to validate or compare CFD results for Ariane 5 are expensive, hard
to do (especially for cryogenic liquids), and limited in generating useful data.
Actual Ariane flight data is best for CFD validation, but it, too, is very
expensive to generate. Thats why we need reliable CFD codes.
Currently, CNES uses some 90 different pieces of simulation software, most
of which need to be ISO 9001 certified. For future space systems, CNES is
interested in using validated software (including CFD codes) to evaluate
reusable launcher components; electrical, nuclear and solar propulsion systems in space; as well as miniaturization and nanotechnology, using lighter
materials with different compositions. Ultimately, performing virtual launches
with simulation software will be a challenge.
- Isabelle Rongier, CNES
The civil space industry has become highly competitive in the post cold war world and the Ariane
space program wants to keep its dominance in the satellite launching arena by constantly updating
its technology. However, as with NASAs recent space shuttle troubles, the modern Ariane 5-ECA rocket also had a failed mission. On December 11, 2002, the cooling tubes in the central Vulcain 2 liquid fuel rocket failed, ultimately leading to the rockets mission being aborted. This caused a major
reevaluation of the Ariane 5 launcher and CFD was used extensively for the next two years to
identify the source of the failure and fix the cooling problem. The changes, many of which were
simulated and later tested, led to a successful launch of a new rocket with enhancements to its third
stage in February, 2005. In honor of this achievement, all of the stories in this supplement focus on
Ariane 5 applications that have made use of FLUENT software. Cryogenics applications are featured
in stories by Air Liquide and Cryospace (page s4), Snecma (s8) and CNES (s10), and heat transfer in
pressure-controlled cavities is described in an article by EADS (s12). The complex flow characteristics
of gaseous combustion products in solid rocket motors are shown on the supplement cover, and
described in the article by Fiat Avio Spa on page s14. 

EADS carries out most of the construction of the Ariane 5 Launcher, and uses
FLUENT for aerothermal analysis and
aerodynamics. Each Ariane launcher
is unique because each payload is
different. CFD is used to give technical
confidence and reliability to the design
team. In their planning it helps to cut
costs and it also yields technical insights
that would not otherwise be available.
However, senior engineering experience
also plays a major role in the design
process in terms of checking the CFD
predictions rigorously.
- Loic Cheriaux and
Jean-Marc Carrat, EADS


Launch-Vehicle Modeling, European Space Agency (ESA) Bulletin 120, November, 2004.

Fluent News Fall 2005




Space Engineering

By Jerome Lacapere, Air Liquide, Sassenage, France, and Mathieu Gardette, Cryospace, Les Mureaux, France

NGINEERING ACTIVITIES relative to the cryogenic propellant tanks of

the European space launch vehicles were first developed at Air Liquide DTA
(Advanced Technology Division) for the Ariane 4 upper stage tank (also
referred to as H10). Subsequent development occurred at CRYOSPACE, a
joint venture between Air Liquide (55%) and EADS ST (45%), for the Ariane
5 main stage (EPC) and upper stage (ESC) liquid hydrogen (LH2) tanks.
Development of the Ariane 5 upper stage liquid oxygen (LOX) tank (ESC
LOX) and the helium sub-system (SSHEL) was later performed at Air Liquide.
Engineers at Air Liquide developed in-house software in the 60s and 70s to
compute and predict the thermo-hydraulic behavior of the propellants
(liquid hydrogen as fuel and liquid oxygen as oxidizer) in their respective
tanks. The software models have been periodically improved, and they
remain dedicated engineering tools. During the past five years, CRYOSPACE
and Air Liquide DTA have also been using and benchmarking FLUENT for
their specific cryogenic applications.


The main function of a stage tank is to thermally condition the propellants in

order to feed them to the engine in well-controlled temperature and pressure
ranges. Over the years, a specific expertise has been developed and maintained for modeling the thermodynamics inside these tanks. In particular, the
simulations must:

(upper stage)

compute the heat fluxes entering the tank volume when it is

subjected to a severe external environment (ground conditions on
the launch pad or flight conditions during the ascent phase, for



(main stage)


He sphere

Vulcain/Vulcain2 engine
Diagram of the Ariane 5 rocket, showing the cryogenic storage tanks
Courtesy EADS SPACE Transportation / SERGE WITTEMANN


Fluent News Fall 2005

accurately predict the heating rate of the liquid propellant and

especially the stratified temperature profile that develops at the
top of the tank due to natural convection along the tank walls
compute the gaseous mass flow rate needed to maintain the
ullage pressure (or that in the space above the liquid) in a
specified range throughout the flight (before and while the liquid
is draining)
All of these computations lead to a calculation of the thermal residuals,
which correspond to the mass of cryogenic liquid that is not compliant with
the engine specifications in terms of temperature and pressure. Other
calculations are performed to compute the diphasic residuals. These are the
small amount of cryogenic liquid remaining in the nearly empty tank at the
end of the engine feeding process. When this point is reached, external
perturbations and the very high flow rate of ingestion could cause bubble
ingestion, which must be avoided. Both types of residuals can be predicted
using complex 3D simulations.




Thermal Residuals

If the tank pressure is regulated at a constant value during the flight, the
NPSP depends mainly on the pressure loss in the engine feed lines between
the tank and the turbo-pump, and the temperature of the propellant that is
fed to the engine. The minimum acceptable NPSP value corresponds to a
maximum acceptable draining temperature. The mass of propellant with a
temperature exceeding this upper limit is unburnable. Since a launch vehicle
always needs to optimize the ratio of its used propellant mass to its loaded
mass, the residual mass should be minimized. It is therefore important to
limit the heating rate of the propellant by an adequate insulation design and,
if thermal stratification occurs, to be able to predict the draining temperature
and its evolution over the full propulsive phase.
One place where thermal stratification occurs is in the upper stage liquid
hydrogen (ESC LH2) tank. FLUENT was chosen to simulate this problem
because of its volume of fluid (VOF) model with heat transfer, and its unstructured mesh capability. The goals of the simulations included predictions
of thermal stratification and feed temperatures at the tank outlet with an
accuracy better than 0.1K. When deemed acceptable, the computations
were done with 2D axisymmetric geometries. However, the particular shape
and location of the feed line (tank outlet) sometimes required the use of a
180 3D problem domain, which made use of a symmetry plane. One of the
main concerns was the optimization of the mesh. In addition to capturing
the turbulent convective boundary layers and moving free surface, there
were local areas where the fluid velocity was known to be relatively high (in
the boundary layer and close to the tank outlet), and other areas that could
be considered dead zones with near zero velocity. Additional complexity was
due to the fact that almost all of the parameters were time-dependant
and many, such as heat fluxes on the walls, longitudinal accelerations, and
draining flow rates, were highly variable.

minimum NPSP
LH2 tank pressure (bar)

Thermal stratification
Before launch, but following the filling of the tanks, the cryogenic propellants
are thermally stabilized into a saturated state. Since the tank pressure values
are typically close to 1.1bar, the stabilized temperature values are 20K
(-253C) for LH2 and 90K (-183C) for LOX. A few minutes before launch,
the tanks are pressurized with helium and brought to the flight pressure
values. The pressurization process is aimed at stabilizing the tank structures,
which will have to withstand significant mechanical loads during the ascent
phase, and at providing a sufficient net positive suction pressure (NPSP, or
difference between the static pressure in the tank and the saturation pressure
in the pump) to prevent any cavitation in the turbo pumps during the draining (engine boost) phase.


Phase Description:
initial saturated state
A to B: ground pressurization
B to C: waiting phase w/o draining
C to D: draining phase (engine on)
engine cut-off)


LH2 saturation curve

Thermodynamic evolution at tank outlet


max temperature = 23.5K











LH2 temperature (K)

LH2 saturation curve over a range of temperatures, and the pressure evolution
prior to and during launch and ascent

Thermal stratification predicted by a 2D

axisymmetric model of the ESC LH2 tank during
draining; warmed propellant starts to stratify at the
top of the liquid volume, and a hot spot appears at
the tank bottom in a zero velocity area

Fluent News Fall 2005




Liquid sub-cooling
Although the more frequent concern is the heating rate of propellant inside the tanks, there are
also reverse situations in which problems result
from excessive sub-cooling. For the EPC tanks in
the main stage, the LOX and LH2 tanks share a
common bulkhead, and a strong thermal coupling occurs at the wall interface. At the bottom
of the LOX tank, heat is conducted toward the
LH2 side, which serves as heat sink. Locally, the
liquid oxygen is driven below its saturation temperature into a sub-cooled state. The main stage
engine must operate within a given temperature
range at the pump inlet, however, and this leads

to a requirement for a minimum sub-cooled temperature. Part of the functional studies performed
on the propellant tanks included modeling the
thermodynamic evolution inside the tanks and
verifying that the feed temperatures do not drop
below the allowable range.
A 2D axisymmetric simulation was performed for
the bottom of the LOX tank in the region of the
common bulkhead. In addition to the liquid volume, the surrounding metallic tank structure and
tank insulation were meshed as well. This allowed
the external thermal environment specifications
to be imposed directly as boundary conditions.

2D axisymmetric model of the bottom of the EPC

LOX tank, showing a sub-cooled (below 91K) layer
of LOX growing in the tank bottom

Natural convection inside the liquid propellant

was simulated along with conductive heat transfer in the structure walls and insulation during the
transient simulation. The results were used to
predict the temperatures in the sub-cooled
region and to make sure that the temperatures at
the pump inlet were within the necessary range.

Temperature gradients in the tank structure

The tank structure is made primarily of aluminum alloys, and another set of simulations was performed for analyzing temperature gradients. Using fine meshes for the solid wall material, one objective of the simulations was to accurately compute the conductive heat fluxes resulting from thermal gradients, which can represent a fair amount of the heat budget entering the tank. The simulations were also
used to export temperature profiles for input to mechanical analysis models. For these calculations, the temperature boundary condition
was typically imposed on the cryogenic side of the structure at the wall/propellant interface.
As an example, an axisymmetric simulation was performed for the insulated lower skirt and ring of the upper stage LH2 tank. The tank
was completely full, and still on the ground. The steady-state simulation showed a 265K temperature gradient along the 700 mm long
skirt that generates a steady heat flux of 9 kW toward the tank inside.

2D axisymmetric
model of the
insulated lower
skirt and ring of
the ESC LH2 tank

Liquid sloshing in the tanks

The behavior of the cryogenic fluids inside the tanks is complex
for a number of reasons. First, external perturbations during the
launch cause sloshing and rolling of the free surface of the liquid.
As learned from many years of development and flight measurement processing, the sloshing amplitudes experienced by the propellant can become high enough to modify the heat and mass
transfer at the liquid/gas interface and to perturb the thermodynamic equilibrium in both phases. Typically these conditions
cause the tank pressure to drop and the liquid temperature to rise.
Second, in the gaseous dome above the liquids in the tanks, there
is some amount of non-condensable helium present (which is
used as a pressurizer). While it is not enough to violate the
assumption of pure vapor in the LH2 tanks, it is significant in the
LOX tanks, so must be taken into account in numerical modeling.
Third, in the near future, the upper stages of Ariane will be
required to operate in ballistic mode, and to have engine
reignition following this phase of operation. During ballistic flight,
the vehicle is in a micro-gravity environment. The walls of the
tanks are wetted because of the predominance of capillary forces
(with Bond number about 1) and because of the specific wetting
property of cryogenic fluids (with contact angles less than 5).
Because of the increased surface area, the cryogenic liquid undergoes increased heating by external heat fluxes and by heat and
mass transfer with the gas phase.
To address these needs, the VOF calculations need to take into
account heat and mass transfer at the liquid/gas interface. A
special model has been developed at Air Liquide for this purpose,
and is now undergoing a complete validation. As part of the
validation, characteristic tests have been performed with cryogenic liquids, including sloshing tests in a small cryostat (diameter
~ 20cm) with liquid nitrogen and liquid oxygen (performed in the
framework of the French-German program COMPERE). These


Fluent News Fall 2005

tests have shown the characteristic evolution of gaseous pressure

and propellant temperature that are directly linked to the heat
and mass transfer at the interface. In some cases, the pressure
evolution can be large, and this was shown to result in significant
condensation at the free surface. The numerical analysis of these
experiments was carried out using FLUENT and the special heat
and mass transfer model was incorporated through a user-defined
function (UDF). The simulation results were found to be in good
agreement with the experimental data.
After a complete validation of the model, it will be used to simulate thermal stratification and pressure evolution in a cryogenic
tank in a micro-gravity environment, so that temperature and
pressure can be computed precisely for the ballistic phase.

Temperature stratification
in the cryostat during
sloshing, with the range
limited to 85K

Volume fraction of helium in the

gaseous phase during sloshing,
showing the condensation of
nitrogen close to the free surface



Diphasic Residuals
As the tank draining nears the end, some free surface disturbances may
occur, particularly for the LOX in the upper stage tank, leading to sudden
gas inclusion in the outflowing liquid. Because feed pump operation is not
recommended in such conditions, the engine must be stopped before any
gas ingestion occurs, and a residual mass of liquid will remain in the tank. The
quantity of diphasic residual mass depends on the dynamic conditions
during the draining process.
In quiet conditions, perturbations in the liquid surface are minimal, so gas
ingestion is small. However, rolling can be imposed to the stage and transmitted to the drained propellants by the launcher attitude control system,
and sloshing can occur. A vortex can form with an associated residual mass
that increases with the intensity of these disturbances. To compensate for this,
anti-vortex devices have been added to LOX tanks since the development
of the Ariane 4 upper stage in order to delay the ingestion of bubbles in
the collector and reduce the mass of the diphasic residuals. The impact of
these devices is now being computed for a number of different flight configurations with different degrees of rolling associated with different lateral
perturbations. These computations are being performed with the complete
3D geometry, starting from the very beginning of the launch phase and
finishing at the end of the upper stage thrust phase, when the tank is draining.
The computations focus on two periods of time. The first is dedicated to the
thrust of the first stage, and lasts about 10 minutes. The tank is not draining,
but the fluids are subjected to external perturbations, particularly spinning.
The second is dedicated to the thrust of the second stage, when the tank is
draining and subjected to external perturbations. Strong vortex formation
appears at the end of this phase with bubble ingestion. This phase lasts about
15 minutes.
Preliminary studies have included a comparison of numerical and experimental
results, with tests performed on sub-scale tanks filled with water and at
conditions with a similar Froude number. Very good agreement was
achieved, from quantitative and qualitative points of view. For example,
when the first bubble was ingested, the predicted residual mass of water was
within 10% of the measured value. In addition, the qualitative behavior of
the free surface in the numerical computations matched the observed behavior, both in terms of the time when dips in the surface were seen to occur and
the location and size of the subsequent vortices. Following the validations,
further computations were carried out using an actual flight configuration.

Anti-vortex devices

Geometry of the LOX tank bottom with internal anti-vortex equipment

Stable vortex
and bubble

Three seconds after the onset of a smooth dip in the surface, a stable vortex is
created as the free surface approaches the anti-vortex device. Some bubbles
escape from the free surface, and this instant corresponds to the end of the
engine feeding process; the remaining liquid in the tank corresponds to the
diphasic residual

FLUENT is now used intensively to compute the pressure and temperature
evolution in cryogenic tanks during all flight phases from pressurization on
the launch pad to the last droplet ingestion in the turbo-pump. In the future,
the tool will be adopted for use in a microgravity environment. For this
purpose, the development of new local models is needed, and validation
of these models will be a difficult task. Non-dimensional numbers will be
heavily used to recreate experiments correctly on earth, since cryogenic
experiments in true microgravity conditions are difficult and expensive to
perform. Despite the fact that transient 3D computations are CPU intensive,
the final goal is to perform them with complex internal geometry coupled
with thermodynamic analysis and heat and mass transfer at the wall and at
the liquid/gas interface. After complete validation of the relevant models,
computations such as these could be performed within a decade. 

Massive gas ingestion

The last phenomenon to be observed is the massive gas ingestion in the

collector, which always occurs at the same angular location, due to the curvature
of the collector

All these activities have been performed with support from CNES.

Fluent News Fall 2005




Cryogenic Flows in
By S. Zurbach and L. Ballester, Snecma, SAFRAN Group, Vernon, France


to CNES by delegation of the European Space Agency, and Snecma has been
responsible for all of the Ariane cryogenic rocket engines. The Ariane 5
launcher and its powerful, expanded capacity version Ariane 5 ECA are
thrusted by the cryogenic rocket engines Vulcain or Vulcain 2 and HM7.
Vulcain 2 is an extension of the Vulcain cryogenic engine, operating in a gas
generator cycle with two separate turbopumps. The combustion chamber
is cooled by cold hydrogen flowing through regenerative circuits and the
nozzle extension is cooled by cold hydrogen that flows through helicoidal
tubes (a dump cooling system).
Rocket engines are exposed to severe mechanical and thermal loads, such as
high vibration, a wide range of temperatures (from 20K to 3600K), and a
wide range of pressures (from vacuum to 200 bar). Due to this extreme
environment and the performance quality and reliability required, an
understanding of the physical processes and complex technologies relevant
to these operating regimes is needed. As the prime contractor for the Ariane
cryogenic engines, Snecma has been modeling cryogenic flows for the
sub-systems of the rocket engines for several years. The specifics of cryogenic
flows depend largely on the thermodynamic behavior of the propellants. For
oxygen, hydrogen, and methane, a real gas approach is mandatory, since the
ideal gas equation of state is no longer valid. For both injection and cooling,
the propellants operate in subcritical and supercritical regimes.

Vulcain 2 at the test bench

Oxygen density as a function of temperature and pressure


Fluent News Fall 2005

Besides hot tests, validated simulation tools are being used more and more
to ensure the reliability of space propulsion equipment. CFD allows new technology to be evaluated without performing high cost tests. It is also a tool for
assessing development risks and production non-conformance. One important example of the potential of CFD simulation was illustrated during the
Flight Recovery Program (FRP) [1]. The FRP was set up after the maiden flight
of the Ariane 5 ECA in December 2002, during which the nozzle extension
lost its mechanical integrity. During the FRP, a reinforced concept for the

Thrust chamber with the helicoidal cooling system and a thermal map of the
helicoidal tube (supercritical hydrogen + solid)



Rocket Engines
nozzle extension was defined, and the new design was produced and
qualified within a very tight time schedule.
As the thermal load of the nozzle extension is the major contributor to its
mechanical integrity, the efficiency of the dump cooling system has been
extensively studied using FLUENT. It consists of welded rectangular tubes
carrying supercritical hydrogen as a coolant. In the simulations, heat exchange
in the regenerative circuit is modeled with a coupled approach involving the
hot gas side and the coolant side, since very high heat fluxes are generated (10
- 100MW/m2). On the coolant side, real gas effects are accounted for and a
modified real gas equation of state is used. A 3D approach is necessary to fully
catch the three-dimensional nature of the flow inside the tubes, generated by
the helicoidal shape of the tube walls. The thermal methodology has been
validated on a subscale nozzle experiment, and on full scale nozzles, with dedicated thin thermocouples implemented on the tube walls. Post test expertise
of the tube and nozzle hardware has been systematically used to anchor the
CFD predictions to a metallographic analysis. The test results are in good
agreement with the CFD predictions. The 3D effects predicted in the flow of
hydrogen in the helicoidal tubes was confirmed by the test measurements. The
predicted wall temperature and the thermal gradients within the solid part
were in line with the measured thermal map as well.
In addition to the nozzle cooling analysis, Snecma has developed specific combustion models to predict the combustion efficiency in a rocket chamber [2]. For
cryogenic engines, the atomization of the reacting fluids is often performed by
coaxial injectors. Liquid oxygen at 90K flows at low speed through a tube, which
is surrounded by an annular high speed flow of gaseous or liquid hydrogen. In
order to guarantee sufficient atomization of the liquid oxygen and efficient
mixing of the combustion products, optimization of the injection plate and injectors is necessary. For this purpose, a balanced analysis of both experimental tests
and FLUENT predictions has been performed. One challenging component of
these calculations is the prediction of the turbulent combustion at very high

Mean temperature field for a LOX/H2 rocket engine injector

pressures, in excess of 100 bar. Because the combustion chamber pressure

is higher than the critical pressure of the injected propellants, the turbulent
combustion regime is called transcritical. Despite these complexities, an accurate
prediction of the reacting cryogenic flows has been obtained by the development and validation of specific models created within the framework of R&D
programs at Snecma. These models, implemented in FLUENT through userdefined functions (UDFs), have been extensively compared to sub-scale measurements and applied to characterize full scale rocket gas generators and chambers.
For the forthcoming years, cryogenic flows will be computed for real gas
mixtures. In the combustion chamber, oxygen and hydrogen are injected at
very low temperatures so that the classical ideal gas equation of state is no
longer valid to describe properly all of the thermodynamics, such as density
and enthalpy. It is therefore mandatory to have a CFD solver with a real gas
formulation for a single fluid (for the cooling tubes) but also for mixtures to
predict the flow field in the main combustion chamber, where supercritical
turbulent combustion of hydrogen and dense oxygen occurs. Important
components for the success of this work include numerical solver robustness
and stability, thermodynamic integration of the necessary models, and a time
reduction for large eddy simulation (LES) calculations. 
For the evolution of the Vulcain Engine, a new nozzle extension was developed
by Volvo Aero in Sweden, under a contract from EADS-ST GmbH in Germany.
EADS-ST is responsible for the Vulcain 2 Thrust Chamber, under a contract with
Snecma in France, who is responsible for the Vulcain 2 Engine.


Ferrandon, O.; James, P.; Girard, P.; Terhardt, M.; Blasi, R.; Johnsson, R.; Damgaard,
T.: Vulcain 2 Nozzle Extension: Integrated European Team and Advanced
Computational Model to the Service of Nozzle Design; AIAA-2005-4535, July 10-13,

Vingert, L.; Zurbach, S.: LOX / Methane Studies for Fuel Rich Preburner; AIAA 20035063, July 20-23, 2003.

Characterization of a LOX / CH4 coaxial injector for two different operating

points; the predicted (left) and visualized (right) mean flame length are compared
Fluent News Fall 2005




LH2 tank

LOX tank

Diagram of the
Ariane 5 ECA
cryogenic upper
stage (ESC)

Surface grid for the LOX

tank (center, bottom) and
the LH2 tank (outer, above)

The Path to Passivation

By S. Petitot, S. Casalino and B. Vieille, French National Center for Space Studies (CNES), Evry, France, and
B. Lazaro and E. Gonzalez, SENER, Universidad Politecnica Madrid, Spain

Start of

X-axis force
X-axis momentum

Beginning of

Spin of

End of

X-axis force
Y & Z-axis momentum

X-axis force



Time (s)


Torques (Nm)




Torques (Nm)


500 1000 1500 2000 2500 3000 3500 4000

Time (s)


500 1000 1500 2000 2500 3000 3500 4000

Time (s)

The chronology of forces and moments applied to

the ESC before and during the passivation phase


Fluent News Fall 2005


after it delivers its payload, which is typically satellites into geostationary
transfer orbit, none of its parts will be reused. For these missions, the
cryogenic upper stage (ESC) is designed to achieve a stable status so that it
can remain in orbit, structurally intact, for 25 years. A sequence of operations
has been developed to make this outcome problem-free. An important step
in the sequence involves the removal of any residual liquid propellant in the
tanks. During this so-called passivation phase, the tanks are depressurized by
safety valve openings and the liquid is allowed to evaporate. The passivation
phase is very important to study, because the evaporation occurs at a time
when the ESC is subjected to linear and angular accelerations, and the
induced liquid motion can give rise to forces and moments that can strongly impact the dynamic behavior of the ESC module.
A study has been conducted using FLUENT to simulate the liquid motion in
the ESC in the period leading up to and during the passivation phase. When
the simulation begins, the ESC has forward axial motion, and a constant
torque is applied to make it start spinning about this axis. After 150 seconds,
it is spinning with a rotation speed of 45 degrees/s and the torque is stopped.
The axial speed decreases throughout this period, and after a total of 400
seconds, the valves open and the passivation process begins. After 300 more
seconds, the ESC is accelerated abruptly and torques about the other axes are
applied. In the final stage, which lasts for another 50 minutes or so, the ESC
undergoes a tumbling motion as it gradually decelerates.
The ESC has two cryogenic tanks, which contain liquid oxygen (LOX) and
liquid hydrogen (LH2). Both tanks are included in the FLUENT simulation,
using a total of 513,000 hexahedral cells. The fluid properties for the isothermal calculation are defined using CNES software. The coupled motion of the
ESC and contained liquids is captured using a six degrees-of-freedom (6DOF)
model, implemented through user-defined functions (UDFs). Viscous stresses
are included through the use of the RNG k- turbulence model. The volume
of fluid (VOF) model with the geo-reconstruct scheme is used to capture the
evolution of the free surface. Three fluids are defined: helium gas (the
primary phase), liquid hydrogen, and liquid oxygen, and surface tension is
taken into account. The external forces and torques described above are
prescribed in a UDF, and are modified by the forces and torques that result
from the fluid sloshing.



UDF: compute source term on each projection of

momentum equation using angular and linear acceleration calculated at the previous time step

X Ang.Acc(rad/s2)






Resolving fluid governing equation:

Fluent Segregated Solver



Time (s)



Time (s)



UDF: Resolve the two new sets of equations using

fluids forces and torques computed by FLUENT:

return the new angular and linear acceleration

Y Ang.Acc(rad/s2)




Iterate time step


The algorithm describing one time step of the coupled calculation

At CNES, a turbulent approach was used, and the results were compared to
laminar calculations done by SENER, a Spanish consulting firm with scientific advisers from the University of Madrid. To achieve a comparable angular
speed after 150 seconds, the torque in the laminar case had to be stopped
10 seconds early, at 140 seconds. Turbulent effects make a full 150 second
spin-up period necessary, but the axial angular acceleration remains non-zero
following this time as it takes time to decay. The difference in behavior is due
to the increased viscous dissipation in the turbulent case. When a turbulent
approach is used, the liquid phase forms a nearly uniform layer that rotates
along the outer wall. The laminar calculation, on the other hand, predicts
that the fluid forms a rotating localized mass instead. The movement of this
mass back and forth inside the vessel gives rise to angular accelerations in the
non-axial directions, and these grow in amplitude, causing an instability
inside the vessel. For the turbulent case, small off-axis oscillations develop
only after 700 seconds, when additional thrusters of the attitude control
system are activated. This initiation time of oscillation development is very
close to the time seen during the first flight of the ESC-A stage in February
2005. The amplitudes of these turbulent oscillations remain considerably less
than when the laminar approach is used.



Angular accelerations of the fluid in response to the

torques and forces applied by the thrusters; the large
amplitude oscillations predicted by the laminar model (the
z-angular acceleration is similar to the y-component)
correspond to bulk rotation of the liquid

Liquid volume fraction in the LH2 (top) and LOX

(bottom) tanks at the end of the spin-up cycle, computed
using a turbulent (left) and laminar (right) approach

An examination of the liquid volume fraction in the tanks helps explain the
difference in the laminar and turbulent behavior. After the spin-up period has
ended, at 150 seconds, a side-view of the tanks shows that the turbulent
approach predicts a cohesive liquid domain, especially in the LH2 tank. When
laminar conditions are assumed, large pockets of liquid separate and flow
independently in the tank, reducing the off-axis fluid force components and
increasing the liquid moments. Thus turbulence acts to improve the kinetics
of fluid movement along the side walls under spin conditions. For both tanks,
the turbulent approach predicts a smoothly rotating mass of liquid at the
wall. For the laminar simulation, the perturbations early on develop into an
unstable rotation of a localized liquid mass inside the vessel. Essentially, the
laminar approach delays the time needed by the fluid to reach a smooth
rotation at the side walls under spin conditions, especially in the LH2 tank
where a huge fragmentation of the liquid occurs. By contrast, the effective
viscosity of the turbulent approach allows a well-behaved fluid mass to rotate
along the LH2 tank perimeter. 

Laminar (left) and turbulent (right) predictions of the liquid surface,

colored by liquid velocity, after 150 seconds
Fluent News Fall 2005




Ariane 5

Internal Cavities Beat the Heat

By Loc Cheriaux and Jean-Marc Carrat, Thermal Aerodynamics & Hydrodynamics Analysis Department-EADS-Space Transportation, Les Mureaux, France

ESC/VEB cavity

EPC/ESCA cavity


The geometry of the launcher cavities, with the LOX tanks in blue and the LH2
tank in yellow

Ariane 5 on the launch pad in French Guiana

Photo copyright ESA/CNES/ARIANESPACE Service Optique CSG

HE NEW 10-TON PAYLOAD ARIANE 5 (the so-called A5ECA) is composed

of two main cryogenic stages separated by internal ventilated cavities. These
cavities are flanked by very different temperature extremes on the sides, so the
development and validation of a thermal model for them, particularly including convective effects, is critical to the success of the project. Several studies
have been performed throughout the life of this program using FLUENT to
simulate convection and conjugate heat transfer. This work continues at EADS,
with the goal of completing a global validation of both cavities using full scale
test data and incorporating local weather conditions into the CFD model.

There are two main ventilated cavities, both with helium conditioning. The
ESC/VEB cavity is between the third stage of the rocket (ESC-A) and the
payload compartment, with a size of about 30 m3. The EPC/ESC-A cavity is
between the two cryogenic stages, and is about 150 m3 in size, with a height
of nearly 6 m. There are many technical challenges associated with this
thermal configuration. First, an acceptable thermal environment is needed
for different types of equipment, such as the electronics and the engine.
Ventilation of the cavities with helium gas can provide the proper environment, but it is expensive and there is an associated constraint on the tank
pressure. Thus, the flow of helium in the cavities needs to be optimized.
Second, a compromise must be found that provides efficient thermal
protection to the tanks, but which is also lightweight and inexpensive.
Consider the EPC/ESC-A cavity. This cavity has two very cold walls on its floor
and ceiling, which are the upper and lower storage tanks for cryogenic fuels.
The liquid hydrogen (LH2) is at a temperature of 20K, and the liquid oxygen
(LOX) is at a temperature of 90K. The vertical side walls, on the other hand,
are exposed to the hot temperatures of French Guyana, the site of the Ariane
5 launch pad. In this environment, temperatures range from 273K (0C) to
313K (60C). The walls of the cavity are made from different materials, such
as aluminum and composites, which have different thermal conductivities.
Using a helium gas flow rate through the cavity, the challenge is to maintain
thermal equilibrium for several hours on the launch pad with the tanks filled!
Modeling the flow and heat transfer in the cavity poses a number of
challenges as well. In addition to the complex geometry and physics, the
numerical model must take into account several critical issues:
The overall cavity is large (several meters across), yet the walls can
be only a few millimeters thick, so this poses a meshing challenge.
Very high thermal gradients are anticipated, so this also impacts
meshing decisions. In addition to regions of refined mesh, a
smooth, high quality mesh overall is desired for the combined
calculation of conduction and convection.
The conjugate heat transfer needs to account for boundary layers,
turbulence, laminar/turbulent transitions, separation, a variety of
thermal boundary conditions, and the effects of phase change of
ambient water vapor.


Fluent News Fall 2005



The mixed convection that will occur will include forced and
natural components, and due to the height of the cavity, the
latter cannot be neglected.
The flow is compressible close to most of the ventilation inlets
(where the injections are nearly sonic), and incompressible in
the remainder of the cavity volume, especially in the many low
Mach number recirculation zones. While this impacts the mesh
refinement in the inlet regions, the choice of numerical solver
(segregated vs. coupled) is still an issue in this situation.
While steady-state flows provide much important information,
transient analyses are also needed.
The forced convection in the cavity is the result of the helium gas flow into
the volume. Natural convection results from the warm walls and cold floor
and ceiling. It gives rise to a circulation loop, with upward flow at the hot
lateral walls and downward flow, accelerated by the cold ceiling, through the
center of cavity. For this cavity, both the Rayleigh number (Ra) and Reynolds
number (Re) have the same order of magnitude: 109. The flow is therefore
highly turbulent and buoyant. For these conditions, a full 3D Navier-Stokes
simulation is required to understand and characterize the coupled convective
and conductive heat transfer phenomena.

The geometry of the EPC/ESC-A

cavity includes many details, most of
which were included in the CFD
model. The two components above
combine to form the geometry used
for the simulation on the right

Construction of the 3D geometry for the EPC/ESC-A cavity required some

simplifications in order to focus on the salient flow features. The model was
not intended to reproduce the exact flow structure throughout the cavity,
but instead, to provide the true global budget of mass and energy throughout the domain. It was created by a surface-based CAD-system, and included the convective flow domain as well as the material structures (for the
conduction calculation). A conformal hybrid mesh of about 1.5 million cells
was built using GAMBIT. This mesh size was chosen to meet several (often
conflicting) requirements, including CPU and RAM resources, convergence
time, and the need for solution precision. Both size functions and the
automatic structured boundary layers option were used during the meshing
process. The realizable k- turbulence model with standard wall-functions
was used, since it is an industrial approach particularly well suited to this
problem. Parallel computations were used to reduce the CPU time.
To date, validations have been made through full scale tests on the ground.
The validation process involves

The validation work has proved to be very successful. Transient predictions of

the average gas temperature inside the cavity and at a point on the wall are
in very good agreement with experimental data.
Based on the use of CFD software (FLUENT) with pre-flight validations using
full scale tests, the thermal calculations and especially the inner convection
field have been identified as critically important to the A5ECA program. CFD
has demonstrated that it can help engineers to better understand complex
phenomena, investigate local optimization, and/or define corrective action.
In the future, there will be a shift from the current macro cavity model to a
more detailed global computational model, by taking advantage of steady
improvements in solver numerics, physical models, and computer capabilities. The end goal will be to reduce computation time yet still make progress
on modeling the thermal coupling between the gas and wall. 

Contours of the fluid velocity field on a slice

through the EPC/ESCA cavity

Temperature (C)

computing the external environment of the launcher using

meteorological data on the ground (solar and wind conditions)
applying this specific environment to the global thermal models
comparing the calculated and measured temperature, and
identifying discrepancies, understanding them, and adapting the
thermal models, as needed.

Wall temperature
Inter stage ambient temperature

13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00


In a comparison between a transient simulation and measurements on the

ground, a global discrepancy of less than 2C was achieved for the average gas
temperature (green) and wall temperature (blue)
Fluent News Fall 2005




Pressure Oscillations
in Solid Rocket Motors
By M. Telara, Avio Spa, Colleferro (Rome), Italy, and
F. Stella, F. Paglia, and M. Giangi, Dipartimento di Meccanica e Aeronautica Universit di Roma La Sapienza, Rome, Italy


of large solid rocket motors (SRMs). In this kind of motor,
pressure oscillations lead to thrust oscillations and to significant dynamic loads that often cause the need for dampers
on the payload, reducing the payload mass capacity and
launcher usability. The problem has attracted the attention
of several researchers during the last few decades and,
although pressure oscillations cannot yet be completely
controlled, the physics of the mechanism that causes them
in SRMs is now well understood. Pressure oscillations
originate in the combustion chamber of an SRM after most
of the fuel has been burnt and after vortices have developed
in the flow of the gaseous combustion products. A feedback
loop can cause these vortical perturbations to grow until a
self-sustained pressure oscillation becomes established.
A solid rocket motor consists of a few elements, including:
the solid propellant, called grain, a case, thermal protections, a nozzle, and an igniter. For large SRMs, which for
technological reasons are characterized by a segmented
grain geometry, vortices in the combustion chamber flow
can be produced in three different ways:
By obstacle vortex shedding behind the tip of
frontal thermal protection devices
By parietal vortex shedding from the propellant
By angle vortex shedding off angles in the grain

Ariane 5 geometry the PTFs

placed on the frontal surfaces of solid
propellant segments are circled in red


Fluent News Fall 2005

Obstacle vortex shedding

Parietal vortex shedding



Angle vortex shedding

The three mechanisms of vortex shedding on the full P230 geometry

The interaction between aerodynamics and acoustics that is

at the heart of self-sustained pressure oscillations in the SRM
combustion chamber is essentially a feedback loop. When the
vortex shedding frequency is close to the natural frequency of
the chamber, pressure oscillations grow and reach a maximum.
The numerical simulation of such complex phenomena is a
challenging task, since every one of the steps involved in the
feedback loop has to be properly accounted for. In the past,
codes devoted to the prediction of pressure oscillations have
been developed and extensively applied. Several authors have
presented numerical results for SRM pressure oscillations, but
in some cases the results are substantially different from
experimental data [1].
One of the goals of the present study is to evaluate FLUENT
for its ability to estimate the frequencies and amplitudes of
pressure oscillations in the Ariane 5 P230 booster. Before
attempting numerical simulations of the full scale P230 motor,
a number of other tests were conducted on simpler geometries,
ranging from cold flow tests to small scale SRM (called LP6)
simulations. For the sake of conciseness, in the present paper
only one cold flow test and the P230 results are presented.
Both analyses share the same idea in approaching the
problem, which is that the pressure oscillation is basically a fluid
dynamics phenomenon that can be adequately reproduced
using a pure CFD approach. For the simple cold flow case, the
fluid flow was assumed to be in the axial direction toward the
SRM nozzle. For the small scale and full scale SRM models, inlet
conditions were imposed normal to the side walls, simulating
solid propellant combustion. Axisymmetric models were used
for all cases studied. Each time-dependent simulation began
with a long initial transient during which time the feedback
mechanism became established. After this period, data acquisition began for the analysis of the chamber acoustics.

In all of the numerical simulations performed, an impressive

agreement between the results and experimental data was
observed both in terms of oscillation frequency and amplitude.
In the cold flow test case, errors lower than 2% on the frequency
and lower than 13% and 6% on the amplitudes of the first and
second modes were obtained. A previous work [2], on the
other hand, reported an error of one order of magnitude on the
The most significant results were obtained for the full scale
P230 SRM, and these details are described in Reference 3.
To conduct this analysis, the flow and geometrical conditions
corresponding to the second peak of the Ariane 5 MPS P230
motor was chosen. The second peak occurs at about 70% of
the total combustion time, so most of the fuel has been burned
at this point. During the numerical simulations, control points
were located at the motor head where the static pressure was
sampled during bench/flight tests. In order to test the grid
dependence of the results, three different meshes were adopted,
ranging from a coarse mesh (100,000 cells) up to a very fine
one (1,200,000 cells). All results from the three meshes studied
predicted the main peak at 20.7 Hz, which is in good
agreement with the values measured during the firing test
and/or obtained from flight data (around 21.2 Hz). It is worthy
to observe that this difference is inside the error bar of the
numerical simulation, or frequency resolution of the analysis,
which is derived from the time step and the number of samples
taken. These results demonstrate the ability of the method to
predict oscillation frequencies even with a coarse mesh.
On the contrary, the amplitude of oscillations was found to
change considerably with mesh density, going from 40 mbar

Amplitude (mbar)

Frontal thermal protection (PTF) devices are present on

the front surfaces of propellant segments in Ariane 5 P230
boosters. Made of different types of ethylene-propylenediene-monomer (EPDM) based rubbers, the PTFs are used to
prevent the combustion of the frontal face of the grain
segments. During flight, the inert material of the PTFs burns at
a slower rate than the propellant, leading to annular ring
protrusions in the combustion chamber that act as obstacles to
the flow. After much of the propellant has burned, the flow past
these rings produces regions of high shear and causes periodic
vortex shedding. The PTF rings are, in the authors opinion, the
most important source of vortex shedding in the Ariane 5 P230
booster rocket.





80 100 120
Frequency (Hz)





Pressure spectrum of the full Ariane 5 P230 geometry, computed

using the mid-sized mesh
Fluent News Fall 2005




5, 6

Presense of unstable
shear layer (1)

Vortex roll-up and

advection (2)

Vortex impingement
on a surface (3)

Coupling on an acoustic mode

of the combustion chamber (6)

Acoustic wave
production (4)
feedback (5)

Feedback loop leading to self sustained pressure oscillations

in SRMs

(coarser mesh) to 151 mbar (finer mesh). This

result showed that a mesh sensitivity analysis is
needed whenever the amplitude of oscillations is
of interest. The amplitudes of oscillations
obtained using the two finer meshes were very
close to each other (149 and 151 mbar) showing
that the numerical solutions had reached asymptotic behavior. The results for oscillation amplitude were compared with experimental data
obtained from bench/flight data of the P230
booster. The comparison was conducted on the
basis of energy in the frequency band 17-25Hz,
and showed a difference from averaged flight
data lower than 5%.

Details of obstacle vortex shedding

Details of parietal vortex shedding


Fluent News Fall 2005

FLUENT numerical simulation of the simplified cold flow test;

vortex shedding behind the PTF rings and changes in the
propellant grain geometry are shown

It is worth noting that all mechanisms of vorticity

production in SRMs have been observed and
properly reproduced using CFD. The results
suggest that vortices initially produced from the
shear layer on the PTF, are merged in the initial
phase of the process and then mixed with the vortices extracted from the side (propellant) walls. In
this manner, large, non-homogeneous vortex
structures are produced and transported along the
motor to the nozzle exit. In spite of the complexities of this flow pattern, numerical simulations
conducted have shown an impressive agreement
with bench and flight experimental data, showing
that the adopted methodology can properly
resolve such fine details of the flow. 


Kourta, K.: Computation of vortex shedding in solid

rocket motors using time dependent turbulence
model., J. of Prop. And Power, Vol. 15, No.3, pp.
390-400, 1999.

Anthoine, J.; Buchlin, J.M.; Guery, J.F.: Effect of

Nozzle Cavity on Resonance in Large SRM:
Numerical Simulations. J. of Prop. and Power, Vol.
19, No. 3, pp 374-384, 2003.

Stella, F.; Paglia, F.; Giangi, M.; Telara, M.: Numerical

Simulation of Pressure Oscillations in Solid Rocket
Motors. European Conference in Aerospace Sciences
(EUCASS) Moscow, 2005.


Shape Optimization
of a Defroster Duct
By Krithika Veluchamy, Detroit Engineered Products, Troy, Michigan, and Pepi Maksimovic, Fluent, Inc., USA
Courtesy of Hyundai-MOBIS

N AUTOMOTIVE DEFROSTER is designed to remove frost from the windshield while adhering to government regulations regarding the time to clear a
minimum specified area. Hot air is blown from the HVAC unit of the vehicle and
through a duct to the inner surface of the windshield, where the defrosters
performance is measured by the air temperature and velocity distribution.
Optimization of the defrosting pattern is commonly achieved by altering the
angle of the guiding grille at the windshield end of the duct to maximize the
velocity impinging on the glass. Such intuitive tuning of the grilles, however,
usually fails to produce flow patterns without some dead zones (areas of no air
flow). Furthermore, the current numerical methodology which involves making
design changes in the CAD system and remeshing prior to simulating the flow
field again and again is painstaking and time-consuming.
New techniques for design optimization have emerged that are based on
parameterization of the CFD model. They automate the analysis process and
greatly increase engineering through-put while reducing an engineers time and
effort. These techniques employ the integration of stand-alone software tools
and the automation of the simulation processes. In this approach, software such
as Meshworks/MORPHER from Detroit Engineered Products (DEP) is used for
parameterization of the CFD geometry and for morphing, or making rapid
changes to the shape of an object within an allowable range of geometric parameters. FLUENT computes the air flow pattern for each new geometry. iSIGHT from
Engineous Software enables the coupling (data exchange) between FLUENT and
Meshworks/MORPHER, and provides optimization tools. The process is set up to
run in a batch mode by executing a sequence of commands listed in script files.
The overall purpose is to optimize a so-called objective function. For the case
of a defroster duct from Hyundai-Mobis, the objective function is to optimize the
windshield air flow pattern so that uniform flow is achieved over the largest
possible windshield area. This highly automated procedure requires less time and
effort than that required to manually build new geometries and meshes for each
geometry change considered, and yields optimized and refined design in the
shortest development time.

Geometry of the automotive cabin with the defroster duct (red), windshield
(blue), and symmetry plane (purple).

Four shape design variables (DV) were defined for the defroster duct. These
geometric parameters governed the shape of the duct and the angle of three
of the grille vanes. A baseline FLUENT model was created, and subsequently
parameterized for the four shape design variables using MeshWorks/MORPHER.
A design of experiments (DOE) process was constructed and, to fully explore the
design space, a total of sixteen grille angle changes were studied. A parameter was
introduced to assess and compare the results. Called the Percentage Area,
it was defined as the percentage of total windshield area having velocity less
than 1 m/s. Thus the objective of the exercise was to minimize this parameter for
optimum performance. After the simulations were completed, an optimization
algorithm was used to determine which case resulted in the minimum Percentage
Area. In addition, a second optimization technique was performed in iSIGHT.
The study showed that the optimal duct design yielded a Percentage Area
value of 17%, which is significantly better than the 42% value recorded for the
baseline duct shape. 

The baseline (original) geometry of the defroster duct (top) and the optimum
design (bottom)


Fluent News Fall 2005



Reverse Flow Catalytic

By B. Liu and R.E. Hayes, Dept. of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta, Canada, and Y. Yi, Fluent Inc.

The compact reverse flow catalytic converter used for

automotive applications; a rotating valve at the top is
used to change flow direction


natural gas to fuel automobiles. New fuels, however,
may require redesign of the catalytic converters used
to reduce emissions of carbon monoxide, hydrocarbons and nitrogen oxides. The natural gas engine
poses a challenge for emissions control because,
under some operating conditions, the exhaust temperature is too low to achieve reaction in the converter. Under these conditions, the catalytic flow reversal
reactor (CFRR) offers potential. In the CFRR, the feed
is periodically switched between the two ends of the
reactor. Switching the feed allows energy to accumulate in the reactor, resulting in high temperatures that
can sustain a reaction. Although CFRRs have been
used in industrial applications, their use in automobiles presents difficulties. The large gas velocity
requires a switching time of about 5 to 15 seconds.
One recent design uses a single cylindrical monolith
core, separated into two halves by a flow divider. A
rotating valve on the top directs the flow into one
side or the other. At the University of Alberta, 3D simulations of this system have been performed using
FLUENT to help optimize the system design. The
model has been validated with experimental results
using the exhaust from a natural gas engine.
The 3D model of the full catalytic reactor contains
porous catalytic sections consisting of a monolith
support coated with a catalytic washcoat, and open
sections. The catalytic sections are modeled using
the porous media option, and the permeability is
adjusted to account for the channels. In the flow
direction, the permeability is based on experimental
pressure drop measurements. To discourage flow in
the transverse directions, it is set one to two orders
of magnitude higher. The flow through the catalyst
section is laminar, and in the open sections, the
standard k- turbulence model is used.

The surface mesh used; the second computation zone

for the solid temperature is shown outside the actual
catalytic converter in the middle


Fluent News Fall 2005

Two approaches have been used to model the reactor. In the homogeneous model (the single zone
model, or SZM), the solid and fluid phase properties
are not differentiated, whereas in the heterogeneous
model (the dual zone model, or DZM) separate equa-

tions are used for the fluid and solid phases. In the
SZM, the catalyst sections are modeled as a porous
medium, with the bulk thermal properties being used
along with an apparent thermal conductivity. The
DZM uses two computational zones. The first zone
covers the entire reactor geometry, including the
monolith and empty sections, the wall, the insulation, and the fluid phase. The fluid phase is treated
the same as in the SZM; that is, the monolith sections
are modeled using the porous media option with
laminar flow. The second computational zone
accounts for the solid temperature. This zone is built
by duplicating the meshed domain of the porous
sections into a new domain, which is identified as a
solid in FLUENT. The density and thermal conductivity
for this solid are effective properties. The bulk
density and an anisotropic effective thermal conductivity are used to account for the porosity. A mesh of
64,500 cells is used for the SZM, and one of 90,710
cells is used for the DZM.
The conservation equations are solved in each
computational cell, and the two zones are coupled
by source terms. A matching process is done by a
user-defined function (UDF) before starting the flow
solution. The link between each pair of fluid and
solid cells is established. During the solution, relevant information is abstracted from each fluid cell
and its corresponding solid cell. Exchange terms are
computed and used as source terms during the next
iteration. The source terms include heat and mass
transfer terms, reaction rates, and thermal energy
generation terms, which are calculated using UDFs.
The boundary conditions imposed are mass flowrate
and temperature at the inlet, and heat loss from the
reactor by convection to the surrounding air. The
interface between the catalytic monoliths and the
inside reactor wall is treated differently for the two
approaches. In the case of an SZM, there is one
continuous phase throughout the domain, and thus
no need to impose a thermal boundary condition
between the monolith surface and the wall. For the
DZM, continuity of flux at the wall is achieved


Converter Heats Up
through the fluid phase, and thus an adiabatic
boundary condition is imposed at the surface of the
catalyst sections.
The only curve-fit parameters in the model are the
kinetic rate constants, which were obtained from
ten steady-state uni-directional flow experiments
conducted in the University of Albertas engine lab.
The model was then tested against transient
uni-directional flow experiments, also performed
in the lab. In the latter experiments, the engine
operating conditions were gradually changed,
producing a transient response. The outlet temperatures and methane conversion predicted by the
two models were found to be in good agreement
with measurements, even though the DZM
temperature and conversion were both predicted
to be higher than the SZM values. Furthermore,
the outlet temperature was found to be higher
than in the experiment, while the methane conversion at the outlet was found to be slightly lower.
This result indicates a discrepancy between the
model and experiments in the calculation of heat
loss. The numerical results for the uni-directional
flow also show that there is not a large difference
in the predictions of the SZM and the DZM.
Considering the experimental error, the observed

difference is insignificant. Computationally, the

DZM uses more CPU time for a given simulation.
The number of computational nodes in the DZM is
higher, and because the fluid and solid calculations
are coupled, a smaller time step is needed to
provide an equivalent solution.
For low temperature feeds, reverse flow is required
to heat the reactor adequately so that conversion
can occur. In a second set of experiments, the
reactor started at steady-state in uni-directional
flow, and then the reverse flow mode was activated.
While in reverse flow mode, the average reactor
temperature was found to increase, as expected.
The numerical results show how the reversing flow
keeps a hot spot in the central portion of the
reactor, so that methane conversion is possible.
Overall, the reverse flow catalytic converter offers a
practical method for achieving methane conversion
in the emissions from a natural gas engine. The
DZM offers superior model prediction when there
are large differences in solid and fluid temperatures
resulting from sharp changes in the inlet conditions.
It is required for reverse flow in most conditions,
and perhaps for uni-directional flow with sudden
large changes in exhaust gas temperature. 

Temperature (K)

Dual zone model
Single zone model





Temperature (K)

Methane conversion (%)


Dual zone model

Single zone model


Dual zone model
Single zone model




300 400
Time (s)



Comparison of the reactor outlet temperature (top)

and methane conversion at the outlet (bottom) for the
experiments and the two models for the transient unidirectional flow case


Time (s)


Average reactor temperature during the reverse

flow process; the temperature rise reflects the
accumulation of thermal energy in the reactor

Temperature fluctuations in the catalytic converter

fluid (middle) and solid (outer) zones for the DZM
as the exhaust flow enters from alternating sides at
the top
Fluent News Fall 2005



Incipient Cavitation
By Jess Esarte, Centro Multidisciplinar de Innovacin y Tecnologa de Navarra CEMITEC, Noain-Navarra, SPAIN, and
Miguel Marcotegui, TRW Automotive, Pamplona-Navarra, SPAIN




A schematic of the TRW rotary valve

HE STANDARDS FOR COMFORT in the automotive industry have reached such a high level that any
noise detected by the components can cause them
to be rejected. In particular, manufacturers of
hydraulic steering systems, in their ongoing drive
towards technological specialization and perfection,
have faced noise problems that usually must be
solved through experimental means. Among the
different kinds of noises generated within a
hydraulic steering system, that coming from the
rotary valve has been of special interest. It is well
known that this noise appears as a result of the
cavitation that the oil suffers when flowing through
the narrow, convergent and divergent sections of
the valve. In a recent project, the Research and
Technological Center of Navarre, Spain (CEMITEC),
in collaboration with TRW Automotive, have
developed a 2D steering valve model in order to
better understand the cavitation phenomenon and
be able to predict if, where, and when cavitation is
going to take place.
A TRW rotary valve was used for the investigation.
The valve has twelve chambers machined on its
outer surface, six on the stator and six on the rotor,

with a 60 radial disposition. Hydraulic steering oil

enters the valve radially at high pressure through
three pressure inlets in the stator, and flows through
narrow passages (convergent and divergent) in the
system either to the left or the right side of
the hydraulic piston, depending on which direction
the steering wheel is twisted.
It is experimentally known [1] that cavitation occurs
in certain locations inside the rotary valve, where
the oil flows through a passage and there is an
accompanying drop in pressure. However, any geometrical change or modification to the operating
conditions makes it impossible to predict whether
or not cavitation will take place unless new bench
tests are carried out. Experimental procedures such
as this are both time-consuming and expensive. For
this reason, it is important to have a computational
model that is capable of predicting the appearance
and severity of cavitation.
Two-dimensional FLUENT models were created to
study cavitation for convergent and divergent
channels. The geometry of a single passage was
meshed with 100,000 cells, and the boundary


Cavitation predictions by the model (left and center) and by experiments (right) in the divergent section


Fluent News Fall 2005


in a Steering Rotary Valve

regime). To do this, the cavitation parameter, K, will
be used. This parameter is defined as the ratio of the
maximum pressure drop to the pressure recovery
that the fluid experiences on its way through the
convergent and divergent sections [2, 3].


conditions were set so that the oil flowed in one

direction or the other. The standard k- model was
used for turbulence. In the first round of simulations, the cavitation phenomenon resulting from
shear layer instability was predicted by FLUENT,
and the results were in very good agreement with
experimental data. The model was unable to
predict any cavitation produced as a consequence
of an extreme pressure reduction, as the experiments had shown. In order to improve the CFD
model, the oil viscosity was modified according to
the Ostwald law for non-Newtonian fluids. This
law states that the fluid viscosity falls drastically
once its shear rate exceeds a certain value. From
this point the viscosity keeps a constant but lower
value. When the viscosity drops, the pressure is
reduced, giving rise to the onset of cavitation.
By introducing this revised viscosity law through
user-defined functions (UDFs), the model predicted the cavitation generation in the divergent
section, exactly as the experiments had shown.

No cavitation





For a specific torsion angle of the rotary valve (2),

there is a particular operating condition or flow
regime beyond which cavitation begins. Beyond
this point of so-called incipient cavitation, or Ki, as
the velocity is increased, the cavitation intensity
rises while the cavitation parameter, K, decreases. 


Fluid velocity (m/s)


Cavitation parameter, K, vs. fluid velocity for the case

where the rotor is twisted by a torsion angle of 2 with
respect to the stator; cavitation begins only after the
point of incipient cavitation has been passed


Behrens, H.W.; Harpole, G.M.: Review of Hiss Noise

Research Results. TRW Steering Division, March 1992.

McCloy, D.: Cavitation and Aeration: The Effect on

Valves and Systems. Fluid Power Institute, Milwaukee
School of Engineering.

Rau, J.; Miller, L.: Reduction of Flow-Induced Noise Due

to Cavitation in an Integral Power Steering Gear Rotary
Valve. TRW Cross Gear Division, 1989.

Now that the model has been validated, the next

step will be to determine the cavitation intensity as
a function of the operating conditions (flow

Cavitation predictions by experiments (left) and the model (center and right) in the convergent section

Fluent News Fall 2005



Lithium Jet Hydraulics

By Valeriy Kolesnik, Alexander Mikheyev, and Nikolay Loginov, Institute for Physics and Power Engineering, Obninsk, Russia

Target assembly of the IFMIF: the deuteron

beam strikes liquid lithium and produces
energetic neutrons at a rate of 1017 neutrons/sec
[3, 4]

OR MORE THAN FIFTY YEARS, scientists have been

striving to achieve the controlled fusion of light elements (hydrogen isotopes) into heavier products
(mostly helium, with the additional production of
energetic neutrons) as a means of producing energy.
Several concepts have been under study at laboratories
around the world, and have met with varying degrees
of success. One key ingredient of a probable fusion
reactor of the future is a blanket of lithium lining the
interior walls. The lithium will absorb the energetic
neutrons, fueling a heat exchanger for the production
of electricity, and will also produce tritium, a hydrogen
isotope that can be used in subsequent fusion reactions. In addition to lithium, other materials will be
present inside the reactor, and their response to continued bombardment by neutrons is the subject of
ongoing research.
The International Fusion Materials Irradiation Facility
(IFMIF) is currently being planned by Japan, Russia, the
EU, and the United States to test materials for their

potential use in fusion reactors of the future. At the IFMIF,

a beam of energetic neutrons will strike samples of various materials to test their performance under likely reactor conditions. The neutrons will be produced in a target
assembly by an accelerated beam of deuterons (deuterium nuclei) that strikes a liquid lithium jet. Since the lithium jet will be exposed to unusual conditions, such as
vacuum and imposed curvature, numerical simulations
are being carried out to better understand its behavior.
The lithium jet will be emitted from a special nozzle
and will flow at high velocity along a concave surface
inside the target assembly area. While the outside surface of the jet will be bounded by a rigid wall, the
inside (free) surface will be exposed to a vacuum. At
this stage of the research, 2D isothermal simulations of
the jet flow have been carried out using FIDAP. The
studies have examined the flow inside the nozzle and
along the free surface of the jet, as well as the influence
of the nozzle edge angle on the shape and behavior at
the jet free surface.


Back wall



Fluent News Fall 2005

The path of the liquid

lithium through the
target assembly

Outline of the CFD domain, and detail

of the mesh in the nozzle region


The target assembly through which the lithium

flows consists of:
A flow straightener;
A two-stage convergent nozzle [1,2] ;
A curvilinear section
The initial calculation corresponds to an experimental
model where the nozzle edge is oriented at an
angle of 6230' from the test chamber wall. A uniform velocity profile is set at the inlet to the flow
straightener. Inside the flow passages, the walls are
assumed to be ideally smooth and completely
wetted by the lithium. At the jet free surface, the
vacuum condition is represented by a pressure of
0.001 Pa. The material in the vacuum region is
assumed to consist of non-condensable gases that
are chemically passive to lithium. The turbulent jet
flow through the assembly is captured using the
standard k- model. The results show that the liquid
is squeezed and accelerated where the channel in

the nozzle first narrows, and again at the second

stage of the nozzle. The mean jet velocity at the
nozzle outlet is 20m/s, and the pressure drop across
it is about 100 kPa. Close inspection of the liquid
velocity in the nozzle edge zone shows tiny drops of
liquid leaving the free surface, even though the
calculations did not estimate the size and frequency
of these drops.

geometry behind the jet leads to a radial acceleration

of the lithium and a subsequent centrifugal force
of the liquid on the wall. The liquid near the wall is
compressed quickly, but to conserve the jet volume,
the jet surface deforms. Following the change in
curvature, the thinned jet profile increases gradually
along the remainder of the wall. 

A calculation of that part of the jet volume occupied
by the liquid is also of great interest to researchers.
Described by the function F, F=1 corresponds to the
case when all the volume is filled with liquid, and F=0
to the case when the given volume is empty. The
results show that there is no vacant space in the bulk
of the jet volume or adjacent to the wall. At the jet
free surface, however, a significant drop in F occurs in
a 1-2 mm thick layer, which is associated with wave
generation. The perturbations at the jet surface in this
region develop as a result of the curvilinear path that
the jet is forced to take. The curvature in the wall

Ida, M. et al.: Thermal-Hydraulic Characteristics of

IFMIF Liquid Lithium Target. Fusion Eng. Des., 63-64,

Shima, A.: Theory of Direct and Inverse Methods to

Obtain Nozzle Shape. Mem. Inst. High Sp. Mech.,
Japan., Vol.17, No. 164, p.61-86, 1961/1962.

Ida, M; Nakamura, H; Ezato, K; Takeuchi, H.:

IFMIF_WalkThrough03_0317, JAERI, Japan, 2004.


Turbulent kinetic energy for a nozzle

edge angle of 6230'

Contours of velocity magnitude for

the lithium jet after 0.013 seconds
for a nozzle edge angle of 6230'

Deformation of the jet velocity field at

the vacuum interface, where the channel
curvature changes
Fluent News Fall 2005



A Computational Cure
for Radial Tires
By Christophe Waucquez and Antoine Dozolme, Fluent Benelux

1 min.

4 min.
Compound A
Compound B

The temperature on the mold, compounds, and bladder

at the start of the problem, along with the mesh

URING IS THE FINAL STEP in tire manufacturing whereby a green tire built from layers of rubber compounds is formed to the desired shape in a
press. Heat is transferred to the tire from the surroundings, which are maintained at a higher temperature. The added heat causes a curing reaction
(vulcanization) of the rubber compounds to begin,
converting the layers of compounds into a strong
elastic material and binding them with internal reinforcing cords.
The curing process is energy-consuming and has a
strong impact on the final tire properties. Given the
temperature history, the cure cycle may be optimized to minimize capital and energy expenditures.
The conventional method is to directly measure
profiles of the temperature as a function of time
using thermocouples inserted into various parts of
the tire, and then to use the measured profiles to
predict the quality of the final product. This method
is costly and time-consuming however, so computer simulations are now being performed by engineers in this industry as an effective alternative. The
ample data provided by numerical simulations can
be used to assess the structural characteristics of the
finished tire and to optimize the process.


Fluent News Fall 2005

In this example, POLYFLOW is used to simulate the

curing process using a simplified model of a tire. A
passenger car radial tire typically consists of 15 or
more layers of rubber compounds, but in this example, only two layers are used. The layers are assembled in a mold with sidewalls and a flexible bladder
on the inside surface. Steam and hot water are the
major sources of heat used for curing reactions.
Steam circulates in the mold and pressurized hot
water circulates through the bladder.
In addition to the transient boundary conditions,
temperature-dependent properties, and a geometrical description of the tire, bladder, and mold, a cure
calculation also requires an accurate kinetic model
for the reactions that take place inside the rubber
compounds. For the two component case considered here, a fairly general model is used that
accounts for both an induction period, before any
changes to the rubber properties are measurable,
and a curing stage, during which these changes
take place. A dimensionless variable is used to characterize the state of cure (SOC) during the process.
One popular way to compute the SOC is through
rheometry. The torque required to maintain a given
strain on a rubber specimen changes as the material properties change. The torque is constant and the
SOC is 0 during the induction period; both quantities change during the curing process; and when
the curing process is complete, the torque is again
constant but the SOC has a value of 1.
The histories of temperature and the state of cure are
among the most important predictions of such a calculation. The state of cure predictions, in particular,
can be used to ensure that the time required to complete the curing process is not over-estimated. 

6 min.

8 min.

12 min.

The temperature distribution (lower half) and state of

cure (upper half) for the tire and it surroundings from
1 to 12 minutes (top to bottom), at which time a
uniform temperature of 170C has been achieved


The PLM embedded solution

cycle of FLUENT for CATIA

Automated flow volume

extraction and meshing

Knowledge-based design

Simulation setup and automated solution steering


Rapid Flow Modeling for PLM

By Laurent Collonge, FLUENT for CATIA Product Manager, and Andr Bakker, FloWizard Product Manager

O YOU OFTEN WONDER how to better integrate CFD into your companys product development process? How to manage the simulation data?
How to integrate not just CAD and CFD, but also
stress analysis, optimization software, and more?
The answer may lie in the use of a Product Lifecycle
Management (PLM) solution. PLM software tools are
used to document and support the complete life
cycle of industrial products and to manage related
services, such as product maintenance. PLM products
are usually integrated with engineering software,
such as CAD, and other enterprise management
tools, such as project planning, resource scheduling,
cost estimation, and knowledge capture software.
Fluents large customer base includes many worldleading manufacturing companies who share the
need for tight integration of their engineering and
business processes. Most of these companies are
making significant investments in PLM to further
streamline their operations. To meet their growing
needs, Fluent and Dassault Systmes of Paris, France
formed a partnership to develop the first CFD software that resides completely inside the CATIA PLM
solution product. FLUENT for CATIA (FfC) is more
than CFD embedded into CAD: it is CFD absorbed

into the PLM bloodstream of the worlds leading

manufacturing companies. FfC ensures that these
companies make the most of their PLM investments.
In addition to knowledge and data management
capabilities, PLM companies need fully generative
relationships between the design of their manufacturing-ready geometric models and their mathematical simulation models, which include structural,
thermal, and CFD analysis. Simply having the CFD
model show up in the same graphics window as the
CAD model is not enough. The CFD data has to fully
reside in the PLM softwares database structures.
One important benefit of such a system is that the
geometry, CFD, and other analysis models work
together to allow for complete knowledge-based
optimization studies that involve all aspects of the
design. Engineers can rapidly turn their CAD models into flow models, and store the resulting optimized designs and supporting CFD data together,
using the same data management system. To
accomplish this, deep access to both the PLM and
CFD software codes is required, so Dassault
Systmes and Fluent worked closely together in
developing FfC. FfC uses original CATIA components wherever possible, including meshers and

optimizers. Some of these components were actually optimized conjointly by Dassault Systmes and
Fluent to enable a tighter relationship between the
two codes.
With FfC, the rapid flow modeling concept first
introduced by Fluent with FloWizard, is now fully
integrated in the PLM process. Rapid flow modeling
is an approach to CFD simulation aimed at reducing
overall time and increasing efficiency. It allows for
quick engineering design validation throughout the
product lifecycle. FfC embraces the high level of
automation that is key to successful rapid flow modeling. Tasks that often require a lot of manual intervention, such as meshing, solution steering, and
reporting, are fully automated. FfC allows PLM companies to reduce their development time, while continuing to produce competitive and innovative
FLUENT for CATIA has already been tested extensively at global industrial companies. The overwhelmingly positive response it receives has further
strengthened the partnership between Fluent and
Dassault Systmes. FLUENT for CATIA is scheduled
to be available for general distribution within the
next few months. 

Fluent News Fall 2005



MMAs balancing valve STV 25 is used to adjust and

balance the flow in heating and cooling systems.
Examples of usage areas include mains, paths, branch
lines, shunt groups, and cooling baffles. The valve is
equipped with self-sealing measuring sockets, placed at
a 45 angle in relation to the wheel center. The wheel
is equipped with a digital display, and is used to set
the valve at the desired flow (or valve sizing
coefficient) value. When the flow value of the valve is
set, it is locked. After locking, the valve can be closed
but cannot be opened to a higher sizing coefficient
value than the one set.
The geometry shown above was created in Pro/ENGINEER Wildfire

Balancing a STV valve


Fluent News Fall 2005


FloWizard at MMA

ERSION 2 OF FLOWIZARD rapid flow modeling software was released in August. It offers many new features,
including improved CAD connections and the ability to calculate remotely. While presenting FloWizard 2 at the
Scandinavian UGM in Gothenburg, Sweden, Andr Bakker, FloWizard Product Manager, spoke with Ivan Bublik,
Design Engineer and FloWizard user in the R&D Department at MMA.

The flow region as shown in FloWizard, with

the inlet in red and the outlet in purple

The total pressure distribution on the valve

body walls

The velocity magnitude distribution in the flow

AB: Can you tell me about MMA?

IB: MMA is a well-known trademark for thermostatic
radiator valves used for indoor climate control.
Together with Irish Myson Heating Controls,
MMA makes Business Unit Controls as part of
Finish Rettig Group. MMA is based in Markaryd
in Southern Sweden. Established in the 1950s,
the company has 120 employees who develop,
manufacture, and market a wide range of radiator
fittings, heating, cooling, and balancing valves.

AB: How does FloWizard fit into your workflow?

IB: We use FloWizard for new product development
and for applying existing products to new applications. My work includes the design and testing
of thermostatic sensors and mixing and cooling
valves, and performing technical computations.
To design our systems, we have worked with CAD
and experimental testing. FloWizard provides us
the opportunity to obtain the necessary flow and
pressure drop characteristics of our prototypes
much faster than we could before.

AB: What were your expectations of FloWizard?

IB: Our main expectations were to obtain reliable
results that would match experimental measurements from our testing laboratory. We expected
to reduce the number of prototypes and shorten
development time as well. FloWizard met all of
our expectations. We compared FloWizards
predictions with results we measured in our
experimental tests, and they compared very
well. In fact, an important reason to acquire
FloWizard was the robust solver, which contains
reliable physical models for turbulent flow and
heat transfer. FloWizards ability to work directly
from our existing CAD models and monitor
essential flow parameters for them helped us
reduce prototyping and development time.
AB: How often do you use FloWizard?
IB: My background is in CAD, mainly using Pro/E.
Neither MMA nor I were involved in CFD prior
to acquiring FloWizard. Getting started with
FloWizard was easy, and I use FloWizard daily right
now. I also make extensive use of FloWizards
scripting ability to perform repetitive and parametric calculations. Our FloWizard license began about
six months ago, and over that time, we have done
thousands of simulations.

We use FloWizard in combination with experimental testing. We perform pressure drop tests
both for prototypes and serial manufactured
valves. The testing liquid is usually water at room
temperature, which is easy. There is a growing
demand, however, for cooling systems that use a
variety of viscous media at very low temperatures.
We need information about how our valves
operate in such conditions, but experimental
testing is very difficult for that. FloWizard gives us
the ability to predict how the flow is altered when
we use low temperature viscous liquids. It allows
us to analyze real-life situations for which we
cannot perform tests.

Do you share FloWizard results with

your customers?


Yes indeed, we do. Our customers use databases

made from laboratory tests and the FloWizard
results for sizing the valves. We have several cases
where those combined results directly benefited us
to give necessary support to our customers. For
example, in a recent project we had to obtain the
flow through each individual valve in a series, for a
variety of cone positions, pressure drops, and cooling liquids. FloWizard allowed us to analyze this
complicated system quickly and efficiently. Using
FloWizard saved us both significant cost and time. 

Flow path lines colored by velocity magnitude

Fluent News Fall 2005



Quality & Reliability in

Keith Hanna from Fluent News recently interviewed Tim Morris, Chief Operating Officer of NAFEMS, and Althea
de Souza, Chair of NAFEMS CFD Technical Working Group, to find out more about this independent simulation
standards organization and what it is doing for its members in the CFD community worldwide.



What is NAFEMS and what is your

NAFEMS started in the
early 1980s as a nonprofit organization in the
UK focused on the finite
element (FE) structural
analysis techniques
being rapidly adopted
by industry, and in particular the safe and
Tim Morris
reliable use of the tools
by engineers. Increasingly, with so many
engineers beginning to rely on computer simulation methods as vital components within
their product development processes, a need
grew for an independent authority and an
association of users to resolve best practice use
of this technology. It was out of this environment that NAFEMS was born. Since then, both
structural analysis and CFD have experienced
rapid growth in usage with many companies
investing heavily in these technologies around
the globe. However, for both established and
new users of this software how can they ensure
the best returns on their investments? How do
they develop and enhance their simulation
capabilities? And crucially, how do they know
they are using the technology in the most
effective way possible? This is where NAFEMS
steps in as an impartial best practice guide and
network for users.

How big is NAFEMS and how is it


TM: Currently we have about 700 companies who

are subscribing members ranging from
Fortune 500 companies to small engineering
consulting firms in many countries around the
world. About 30% of our members are UK
based but we are seeing a rapid growth in
North America, with 10-15% of our members


Fluent News Fall 2005

based there, and the remainder mostly in

Europe. About 15% of our members are academics and although the largest proportion of
people are doing FE analysis, our CFD group is
expanding rapidly. We now have branches in
Germany, France, Italy, America, the Nordic
region and Iberia, as well as in the UK, and we
are seeing a cross-section of members in all
industry sectors. NAFEMS has a governing
council of twelve directors from six countries.
There are seven geographically-based steering
groups and five technical working groups. All
committee members in the groups participate
on a voluntary basis, and represent leading
industrial companies, software vendors, and
academics. Currently, our five working groups
are: Computational Structural Mechanics,
CAD/FE Integration and Optimization,
Computational Fluid Dynamics, Education &
Training, and Analysis Management. In many
ways I think the CFD industry today is where
the FE industry was 15 years ago.


What services does NAFEMS offer to

members from the CFD Community?

The CFD working group started in 1995 with the

objective of promoting
the safe and reliable use
of CFD and to provide
best practices, training
and education for CFD
users of all levels. The
work of the group is
aimed at encouraging
and supporting the
Althea de Souza
growing use of CFD in
industry and the expansion of applications and
situations to which it is applied. A key activity
of the CFD working group is to produce guidance documents on industrial applied CFD. To
date, a number of publications have been
produced which we hope will benefit novices


Engineering CFD Simulations

and experts alike. These are available at a nominal price to members, but non-members can
buy them as well. Many new and potential
users can find CFD intimidating because it can
be a very jargon-laden subject. We therefore
wanted to produce guides that unpacked the
language of CFD, dealt with issues in a clear
and simple way, and gave good practical
industrial CFD advice.


from all industry sectors and a range of countries. Eighty percent were using commercial
CFD packages and most used CFD for over
60% of their working hours. The majority had
over one year of experience with CFD. Many
of the respondents wanted more information
on validations of CFD simulation results, the
assessment of solution quality, the theory of
CFD, and basic meshing and CFD solver
guidelines. They also requested more
workbooks of CFD examples, specialist and
advanced CFD publications, and basic or
introductory CFD information. The survey also
identified the most useful format through
which the material should be made available,
that is, via booklets and online website

What do you perceive to be the main

general issues facing CFD members of

ADS: Today with the profusion of easy to use and

push-button CFD codes for designers, the
unwary can easily get wrong or unphysical
CFD predictions without being aware of it.
Many people want roadmaps through the
minefield of CFD usage. We find that meshing
is still a big issue especially compared to structural analysis, as is CAD embedding of CFD.
The CFD industry is awash with buzzwords like
multiphysics and multibody dynamics and
these terms need to be defined clearly in the
context of CFD. Many companies are interested in integrating their CFD tools into their
design processes but how do they do so
effectively? Ultimately, the bottom line is
the quality and reliability of industrial CFD
simulations and this is the most important
issue today for many NAFEMS members.


I understand that you commissioned the

first independent CFD industry user
survey of your members recently. What
were your findings?

ADS: This year the CFD Working Group decided to

carry out a worldwide survey on the availability of CFD information resources. The purpose
of the survey was to identify which topics are
insufficiently covered at present, but of interest
to those involved, either directly or indirectly,
in CFD analysis activities. Most of the 300
respondents to our survey were CFD analysts


Subject Area Interest responses in the


Advice Documents published by the NAFEMS CFD Working Group

Why do Computational Fluid Dynamics?

How to Get Started with Computational Fluid Dynamics
How to Plan a CFD Analysis
How to Understand Computational Fluid Dynamics Jargon
Introductions to Grid & Mesh Generation for CFD
1st NAFEMS Workbook of CFD Examples
CFD Analysis Guidance for Good Practice
2nd NAFEMS Workbook of CFD Examples
How to buy CFD Services
How to choose a CFD System

Fluent News Fall 2005



Drag Laws 102

By Liz Marshall and Sergio Vasquez, Fluent Inc., USA

N THE LAST ISSUE OF FLUENT NEWS, we presented an overview of the basic drag laws in FLUENT. These laws are appropriate for many
applications of the discrete phase model (DPM) and the standard or granular multiphase models, using either the Eulerian or mixture
formulation. Drag contributes to the momentum exchange between different phases in a system. The momentum exchange term, which
appears in each component of the momentum equation, is proportional to the velocity differential between phases and a drag function,
which is obtained empirically for a specific system and flow regime.
Most drag laws make the assumption that the particles are spherical and that drag is largely dependent on the particle diameter and
relative Reynolds number. There are many applications, however, where these simplifying assumptions cannot be made. In this article, we
present some of the alternative drag laws available in FLUENT and the special circumstances that require them.

Non-spherical particles
For non-spherical particles, a special drag law is available for the DPM. Called the
non-spherical drag law, it uses the diameter of a particle with the equivalent
volume to compute the relative Reynolds number [1], which is computed from
the relative velocity between the phases. The drag function makes use of this
spherical relative Reynolds number and a shape function, which is defined as the
ratio of the surface area of the sphere of equivalent volume to that of the actual
particle. In a study of high aspect ratio switchgrass in a co-fired coal burner [2],
the results showed that by modeling the cylindrical particles using these modifications, more accurate predictions of particle trajectories and residence times
were obtained, and predictions of other combustion-related aspects of the
process were improved as well.

A liquid fuel spray in an IC engine simulation

Sub-micron particles
Sub-micron particles are so tiny that they can respond to the random motions of
a fluid, in addition to the bulk flow. In FLUENT, a drag law is available for
sub-micron particles [3] that differs from most other drag laws in that it does not
depend on the relative Reynolds number. Instead, it is a modified form of Stokes
drag law for flow over a stationary sphere that depends on the fluid viscosity, the
particle diameter and density, and a Cunningham correction factor that depends
on the mean free path. In laminar flows with sub-micron particles, a Brownian
motion option is available for the DPM, and this so-called Stokes-Cunningham
drag law is the most appropriate choice for modeling momentum exchange.
Liquid sprays
Numerical models for the spray of liquid droplets into a gas involve complex
physics, such as droplet collisions and breakup as well as the formation of sprays
through atomization. These phenomena can be modeled only through careful
consideration of how a droplet or larger liquid region responds when it is
subjected to the external forces applied by the surrounding gas. An accurate
description of drag is also critical to the success of any spray model. In FLUENT,
the dynamic drag law is available for this purpose. It computes the drag for a
sphere and for a disk of equivalent volume. An equation is solved to assess the
degree of distortion of the drop, and based on the result, a linear interpolation is
done between the spherical and disk drag limits to obtain an appropriate drag
on the drop [4]. This calculation is done dynamically during the trajectory
calculation routine.

Harvesting switchgrass for use in biomass boilers

Courtesy of the US Department of Energy


Fluent News Fall 2005


CFD for
By Leon Liebenberg, University of Pretoria, Pretoria, South Africa,
and Danie de Kock, Qfinsoft, Pretoria, South Africa

The Magelli drag law outperforms the SN law for a

turbulent stirred tank containing large (675 micron)
particles; the Eulerian granular multiphase model was
used for the simulation [7]

Solid-liquid mixtures
Multiphase mixtures of granular (particulate) material and a turbulent
liquid can sometimes be modeled using the basic Schiller-Naumann (SN)
drag law in FLUENT. The success of this model depends on how much the
fluid turbulence influences the drag. The ratio of the particle diameter, dp,
to the Kolmogorov length scale of the flow, , an approximate measure of
the size of the smallest turbulent eddies, provides a good measure of the
importance of this effect. For dp/ < 0.2 (or thereabouts) the role of turbulence is weak and the SN drag law should suffice. For larger values of dp/,
a modified drag law should be used that takes the turbulence into account.
One law that was first proposed by Magelli [5], and later published by
Pinelli [6] has demonstrated very good results. Whereas the SN drag law is
derived from the settling velocity of a particle in a still fluid column, the
Magelli law computes the settling velocity of a particle in a turbulent fluid,
and uses this to compute a corrected drag function. While not available in
the standard release of FLUENT, this custom drag law can be obtained by
contacting your local Fluent office. 


Haider, A.; Levenspiel, O.: Drag Coefficient and Terminal Velocity of Spherical and
Nonspherical Particles. Powder Technology 58, p. 63-70, 1989.

Gera, D.; Mathur, M.P.; Freeman, M.C.; Robinson, A.: Effect of Large Aspect Ratio
of Biomass Particles on Carbon Burnout in a Utility Boiler, Energy & Fuels, 16(6),
p. 1523-32, 2002.

Ounis, H.; Ahmadi, G.; McLaughlin, J.B.: Brownian Diffusion of Submicrometer

Particles in the Viscous Sublayer. Journal of Colloid and Interface Science, 143(1),
p. 266-277, 1991.

Liu, A.B.; Mather, D.; Reitz, R.D.: Modeling the Effects of Drop Drag and Breakup
on Fuel Sprays. SAE Technical Paper 930072, SAE, 1993.

Magelli, F.; Fajner, D.; Nocentini, M.; Pasquali, G.: Solid Concentration Profiles in
Slurry Reactors Stirred with Multiple Impellers: Recent Results. Engineering
Foundation Conference Mixing XI, Henniker, NH 1987.

Pinelli, D.; Nocentini, M; Magelli, F.: Solids Distribution in Stirred Slurry Reactors:
Influence of Some Mixer Configurations and Limits to the Applicability of a Simple
Model for Predictions. Chem. Eng. Comm. 118, p. 91-107, 2001.

Montante, G.; Rondini, A.; Bakker, A.; Magelli, F.: CFD Predictions of Solid
Concentration Distributions in a Baffled Stirred Vessel Agitated with Multiple PBT
Impellers. CHISA 2002, Prague, August 25-29, 2002.

Danie de Kock, Fluent representative for South Africa, presented a two-hour lecture
to 20 top students on the rudiments of CFD; thereafter, each of the students
simulated the flow around an airfoil using FlowLab

Annually, during South Africas winter school holidays, the University of Pretoria
presents a space and aviation camp for a select group of students in grades 11
and 12. The camps main goal is to stimulate interest in careers in mechanical
and aeronautical engineering by using the excitement of aviation technology
and space travel.
To qualify for the 2005 four-day camp, students had to make a conceptual design
of an unmanned combat aerial vehicle. They also had to demonstrate distinctions
(+80% averages) in at least four subjects, including math and science.
Only twenty of the candidates were selected to participate in this unique and
prestigious event, which included the following activities: lectures on aeronautics, space travel and space medicine; practical sessions on designing, building
and testing model rockets and a device to measure rocket altitude; subsonic wind
tunnel testing of airfoils; lectures and tutorials on flow simulation using FlowLab;
flight in a military cargo plane during which in-flight refueling with two Cheetah
fighter jets was simulated; two aircraft factory visits; flying a simulator of a
locally manufactured unmanned aerial vehicle; spinning in a centrifuge to three
and a half Gs, and simulating high-altitude flight in a hypobaric chamber. 

Fluent News Fall 2005



Airfoil Noise in a
By Lilla Edit Koloszr, Patrick Rambaud, Jrme Anthoine, von Karman Institute for Fluid Dynamics (VKI), Rhode-Saint-Gense 1640, Belgium

Adimensional pressure fluctuations


FLUENT 6.1.22
FLUENT 6.2.3
Implemented UDF

0.58 0.60 0.62 0.64 0.66

Pressure fluctuations at a listener position of 3

wavelengths from the transversely oscillating sphere


fluid equations. Due to the non-linearity of these equations and the importance
of different scales, it is a difficult task to predict the sound produced by fluid
flows, especially for high Reynolds number, subsonic conditions. Since the sound
field associated with subsonic flows represents only a minute fraction of the
energy in the flow, the accuracy of numerical simulations must be very high to
capture the sound generation. This is particularly dramatic in free space and at
low subsonic speeds. However, the fact that the sound field is in some sense a
small perturbation of the flow can be used to obtain approximate solutions.
Aeroacoustic analogies, which allow sound generation mechanisms to be
separated from sound propagation, have been developed for this purpose.
Progress in computational aeroacoustics (CAA) nowadays allows a direct numerical simulation of the hydrodynamic and acoustic fields at once. The Reynolds
numbers that can be achieved (at reasonable computational costs) by these
accurate numerical approaches are still below the range that corresponds to
practical engineering applications, however, and acceptable accuracy on the
calculated sound field occurs only for Mach numbers in the high subsonic range.
At the opposite end of the spectrum is a low Mach number turbulent jet at a relatively large Reynolds number. Of particular interest is the noise produced when
an airfoil is placed into the oncoming flow from a jet under these conditions. A
numerical investigation of this aerodynamic noise source has been performed
using a multi-domain hybrid method, which consists of two steps:
1. Near the noise source, the flow field is obtained from LES computations performed using FLUENT.
2. In the far field, the Ffowcs-Williams Hawkings (FW-H) integral method
is used to predict the sound propagation. This aeroacoustic analogy
allows for predictions of the sound produced by solid surfaces
immersed in the unsteady flow field.
Even though FLUENT has the FW-H capability, an acoustic postprocessing module based on this method has been developed at VKI as part of an academic
research project during the nine-month Diploma Course postgraduate program.
The postprocessing module is implemented in FLUENT through user-defined
functions (UDFs). The primary advantage of this approach, compared to writing
a postprocessing module using separate software such as Matlab for example, is
that the variables from the solver can be accessed and, therefore, used directly in
the noise calculation.

Computational domain and mesh in the wing region


Fluent News Fall 2005

The new acoustic module takes into consideration only the noise induced by
vibrating body forces on a surface, namely the dipole source. It was first validated
through a basic acoustic test case: a transversely oscillating rigid sphere that serves
as a pure dipole source. The results were compared to the analytical solution of the
problem and to the acoustic module in FLUENT (6.1.22 and 6.2.3). The implemented
module showed good agreement with the analytical solution and with the solution
obtained with FLUENT in the far-field region where the method is valid. It revealed
however that the user should be careful to define the listener position to guarantee


Turbulent Jet
that it lies in the far field, a position that depends on the distance from, and the
frequency governing the sound source. Otherwise the predictions of pressure
fluctuations were found to be questionable. This is due to the fact that the FW-H
method assumes plane wave propagation while, in the mid field, the wave
propagation is between the ideal spherical and plane wave forms.

The unsteady aerodynamics of airfoils involves two phenomena. One is the flow
separation, either at the trailing edge or, if the wing is stalled, on the suction side.
The other is the interaction with oncoming vortical, periodic, or random
disturbances. Such effects influence the sound field. This basic investigation of
airfoil-vortex interaction can be used in two major areas of industrial applications.
First, such a phenomenon governs the noise propagation from a helicopter rotor
blade. The second major application of airfoil induced noise propagation is
related to turbomachinery. Here the blades are always submerged in a turbulent
flow, so sound production due to the fluctuating pressure is encountered.
For the jet flow over the wing, the noise is primarily due to the interaction of
large eddies with the body, so the pressure fluctuations on the body surface are
needed for the acoustic module. The flow calculation in the source region is
performed using the large eddy simulation (LES) approach in FLUENT. The mesh
is refined near the solid boundaries, in order to reach y+ values less than 5. A very
fine mesh is used at the leading edge where most of the vortex dynamics occurs,
to give enough resolution for noise prediction. The number of faces on the wing
surface is 12,000 and the total mesh size is about 913,000. Because of the large
number of cells, statistically fully converged data are not yet available, since only
five flow-through calculations have been performed to date.
The results at one instant in time show that the wing has a strong effect on the
flow field and that the leading edge plays a key role in the evolution of the
incoming coherent structures. These structures either have to deviate from their
original path in order to bend away from the wing or impact the leading edge,
which tears them apart. Such a vortical flow induces pressure fluctuations on the
wing surface. The fluctuating pressure level at the leading edge is around 100
times higher than that at the trailing edge, so the strongest acoustic source is
likely to be found near the airfoil leading edge.

Instantaneous flow field from the jet outlet to the leading edge of the wing;
coherent structures visualized with Q = 30000 colored by pressure
(Q is the second invariant of the velocity gradient tensor)

Adimensional pressure fluctuations

Once the implemented acoustic module was validated, it was applied to the case of
a wing in a turbulent jet flow of Mach number M = 0.1. A 2D NACA0012 wing is
placed into the jet flow at a distance of 6 times the jet diameter downstream of the
jet outlet. The chord of the wing profile is equal to the jet diameter, so the Reynolds
number based on either the jet diameter or the wing chord is the same (36,000).

Implemented UDF
FLUENT 6.2.3



Time (s)


Pressure fluctuations at a listener position of

4 wavelengths from the wing




angle (deg)



-60 -40 -20

Sound Pressure Level (dB)




The prediction of noise propagation is determined at a given far-field position, just

above the wing, using the implemented UDF and the FW-H module in FLUENT. The
same background LES calculation is used as the source, and the pressure fluctuations
at a listener position of 4 wavelengths from the wing (in the far field) are in
excellent agreement. The close results show that the assumption of considering only
the dipole source term is valid for noise prediction in the far field. The directivity of
the acoustic signal was also computed. The predicted null region in the vicinity
of 90 degrees is typical of acoustic scattering by a finite streamlined body. 



Directivity of sound pressure level at a listener

position of 4 wavelengths from the wing

Fluent News Fall 2005



Fluent Worldwide
Corporate Headquarters


Goes to College
By Shane Moeykens, studentFLUENT Product Manager

Fluent Inc.
10 Cavendish Court
Lebanon, NH 03766, USA
Tel: 603 643 2600
800 445 4454
Fax: 603 643 3967
Email: info@fluent.com

USA Regional Offices

Ann Arbor, MI 48108
Tel: 734 213 6821
Santa Clara, CA 95051
Tel: 408 522 8734
Morgantown, WV 26505
Tel: 304 598 3770

Austin, TX 78746
Tel: 512 306 9299
Evanston, IL 60201
Tel: 847 491 0200

European Regional Offices

Fluent Benelux
Wavre, Belgium
Tel: 32 1045 2861
Email: info@fluent.be
Fluent Deutschland GmbH
Darmstadt, Germany
Tel: 49 6151 36440
Email: info@fluent.de
Fluent Europe Ltd.
Sheffield, England
Tel: 44 114 281 8888
Email: info@fluent.co.uk
Fluent France SA
Montigny le Bretonneux, France
Tel: 33 1 3060 9897
Email: info@fluent.fr
Fluent Italia
Milano, Italy
Tel: 39 02 8901 3378
Email: info@fluent.it
Fluent Sweden AB
Goteborg, Sweden
Tel: 46 31 771 8780
Email: info@fluent.se

2D simulation of vortex shedding behind a cylinder

Asian Regional Offices


across many campuses in the US. Laptops and individual PCs have become synonymous with the hand
calculators students equipped themselves with in
the 70s and 80s. It is not uncommon for engineering students to be required to purchase their own
PC along with textbooks and other school supplies.
With this increase in personal computer access,
students are now expressing an interest in licensing
their own individual copies of software, from word
processors to CFD tools, such as FLUENT. As part
of this shift in software usage, some key universities
in the US are downsizing computer labs from
centralized campus-based facilities to individual
PCs. Recognizing this trend, Fluent released
studentFLUENT in September, 2005.
Initially available in North America only,
studentFLUENT offers all of the advanced
functionality and physical models provided
by FLUENT 6.2, using a maximum cell limitation
of 20,000 elements. Unlike FLUENT 6.2,
studentFLUENT can be purchased directly by
students for use on personal laptops. A student


Fluent News Fall 2005

can setup, converge, and postprocess CFD problems away from campus-based computer facilities
using this tool. All activities except geometry
modeling and meshing can be performed with
studentFLUENT. Students will access GAMBIT on
campus or be provided geometry and mesh files
by a professor for use in their class.
Virginia Polytechnic Institute and State University
(VT) decided to use studentFLUENT in the Fall 2005
Fluids-Heat Transfer Design course. This is a required
class for all undergraduate mechanical engineers,
and is typically offered to 250 senior-level students
each year. Dr. Karen Thole, Assistant Department
Head of the Mechanical Engineering Department at
VT, commented studentFLUENT provides students
with a useful tool for their design projects. In addition, employers of VT students have voiced the need
for BSME students to have some CFD experience.
The use of studentFLUENT in a required course is
vital in that it represents a modernization to the VT
ME curriculum; we believe this modernization will
help our students gain fundamental knowledge and
increased skills. 

Fluent Asia Pacific Co., Ltd.

Tokyo, Japan
Tel: 81 3 5324 7301
Email: info@fluent.co.jp
Osaka, Japan
Tel: 81 6 6359 7371
Fluent Software (Shanghai) Co., Ltd.
Shanghai, China
Tel: 86 21 53855180
Email: info_china@fluent.com
Fluent India Pvt. Ltd.
Pune, India
Tel: 91 20 2293770
Email: info@fluent.co.in

Anova Ltd. Turkey
ATES Korea
Beijing Hi-key Technology Corporation Ltd.
Cavendish Instruments de Mexico, S.A. de
C.V.(CIM) Mexico, Venezuela, Argentina,
Chile, Colombia
FEM++ Israel (POLYFLOW only)
Flowmen Technology Co., Ltd. Taiwan
Fluid Codes Ltd. UK (serving the Middle East)
Fluvius Pty. Ltd. Australia & New Zealand
INNOTECH Ltd. Hungary
J-ROM Ltd. Israel
PlasmaVenture Ltd. Russia & CIS
Process Flow Finland & Baltics
Qfinsoft South Africa
Regional Technologies Corp. Ukraine
SimTec Ltd. Greece (serving Southeastern Europe)
SMARTtech Services & Systems, Ltd. Brazil
SymKom Poland
Techsoft Engineering s.r.o. Czech Republic &
Slovak Republic
TENSOR srl Romania