CFD For HVAC by Fluent

Biomedical
A Digital Gas Mask
Glass & Fibers

Shattering Glass Tank Design Methodologies
Reacting Flow
Three-Way Catalytic Converter Aftertreatment
Worldwide Offices
CFD in Germany
VOL XII ISSUE 1 SPRING 2003
Fluent
Celebrates!
HVAC
Supplement
Inside!
Editors Note
his year it is with

great pride that we
commemorate the
20th anniversary of our
first sale, and the 15th
anniversary of the launch
of our company. Bart
Patel, the founder of Fluent Inc. and current CEO,
provides some perspective on the growth of our
company and CFD in general on the back page
of this newsletter. Inside, we have chosen to honor
these important milestones by devoting the opening section to reacting flow. Reactions have been
a part of Fluents software from the start, when
a three-component combustion model (fuel, oxygen, and products) was a key feature. Since then,
reacting flow capabilities have advanced with every
release. Many of these are highlighted in the articles in our Reacting Flow section that begins on
page 4. New combustion models, surface reaction models, and methods for solving multi-step
reactions are featured. The use of user-defined
functions to simulate a low-density polyethylene
(LDPE) reactor is also described. To top it off, we
even simulated the candles on the front cover
using one of the combustion models in FLUENT!
Immediately following the Reacting Flow section is an article on bubble column reactors where
the focus is not on the reactions, but on the multiphase models that can be used to describe the
complicated flow physics in these devices. Several
biomedical applications are also featured in this
issue. Three of these make use of scanning equipment for generating geometry files for parts of
the human body. This technique is becoming
more and more popular for applications of all
kinds, where CAD files for complex geometric
forms do not exist. Another technique that is
growing in popularity among CFD users is the
coupling of two (or more) software products
together. Examples highlighted in this newsletter include the determination of sail shape for
given wind conditions, the optimization of an
automobile shape for reduced drag, and the solution of glass fiber drawing under the action of
convective and radiative heat transfer.
On the Cover:
Celebrating our anniversary are six
Fluent employees with a total of
139 years of service to the company
(left to right): Michael Engelman,
Barbara Hutchings, Bart Patel,
Sharon Everts, John Murray, and
Ferit Boysan
On the Supplment Cover:

Pathlines, colored by velocity
magnitude, illustrate the flow
around the Swiss Re Headquarters
in London
Fluent News is published by
10 Cavendish Court
Lebanon, NH 03766 USA
1-800-445-4454
2003 Fluent Inc.
All rights reserved.
FLUENT, FIDAP, GAMBIT, POLYFLOW,
G/Turbo, MixSim, FlowLab, Icepak,
and Airpak are trademarks of Fluent
Inc. All other products or name
brands are trademarks of their
respective holders.
This issues supplement features applications

in the HVAC industry, with a cover story on the
airflow in and around the striking Swiss Re building that is nearing completion in London. The
article describes the innovative mixed ventilation plans for the interior office space. Indeed,
many of the stories are about optimizing interior ventilation for a variety of buildings. Two are
targeted at smoke management scenarios, and
one focuses on industrial hygiene. Together, these
stories emphasize how CFD is becoming a mainstay in the design phase of new construction all
over the world.
A new department is being introduced in this
newsletter in which we focus on the use of CFD
in and around one of our worldwide offices. In
the first installment, Fluent Deutschland is featured, along with the current state of CFD usage
in Germany. The Visions of the Future series continues with an interview of Herve Buisson and
Christelle de Traversay from Vivendi Water in
France, commenting on the use of CFD in the
environmental industry. The Support Corner focuses on a topic that generates many questions from
users: how to set up a network of Windows PCs
for parallel computing. The information provided
should help some of you take this important step
forward if, until now, you have been hesitating.
Finally, we would like to extend our congratulations to the Swiss team Alinghi for their
recent win of the Americas Cup, and to the Ecole
Polytechnique de Lausanne for providing
Alinghis design team with important guidelines
from numerical simulations, many of which involved
FLUENT. While your simulation efforts may not
result in the presentation of a coveted trophy,
we understand and appreciate the importance
they bring to your company and to your own
understanding of a particular process. Please consider contributing an article to Fluent News that
describes the work you have done with our
software.
Liz Marshall
fluentnews@fluent.com
feature stories
Contents
reacting flow
Reacting Flow Models in FLUENT
Three-Way Catalytic Converter
Aftertreatment
Optimizing Plant-Scale LDPE Reactors

The Berl Combustor Revisited
departments
Deposition: One Layer at a Time

Surface Reactions in Catalytic Tubes
applications
10
chemical
12
oil & gas
13
biomedical
18
22
Separating Water & Gas
31
worldwide offices
39
academic news
42
product news
45
support corner
47
partnerships
48
around Fluent
16
A Digital Gas Mask

AccuSpray on Demand
Locating the Nasal Valve with FIELDVIEW
The Heartbeat of Pulmonary Modeling
marine
Flying Sails on the Computer
First European Americas Cup Winners
automotive
Grid Morphing
The Sounds of the Road
customized applications
25
materials
26
glass & fibers
30
visions of the future
Bubbling Columns
24
28
20
Environmental CFD
CFD in Germany
Channeling Chaos
Students Take Home ANTEC Prize
Re-entry Vehicle Shocks
FLUENT Ported to Itanium 2/HP-UX Platform

Mixing Simulation Gets Easier
FIDAP 8.7.2 Released in April 2003
POLYFLOW 3.10 Coming in June 2003
The Draw of POLYFLOW-FLUENT Coupling
Parallel Computing on a Windows Cluster
LMS SYSNOISE Link to FLUENT

Spatial Provides CATIA V4 Translation
for GAMBIT
Happy Anniversary Fluent!

New! Online Training
Customizing Food Steamers
Keeping Printer Touch Temperatures Low
31
hvac industry supplement
Shattering Glass Tank Design

Methodologies
Chill Ripples in Glass
S2
ventilation
S6
industrial hygiene
S7
smoke management
electronics & semiconductors

Optimizing Transformer Designs
Putting the Spin on Semiconductors
polymer processing
Molds that Feel the Pressure
S3
Swiss Re Headquarters Inside & Out

Looking Out for Crew Comfort in Space
California Living
Air Flow Befitting the US Marines
Breathing Easier in the Workplace
Containing Smoke in Complex Atria

Fire Scenarios in the Budapest Sports Arena
reacting flow
GaAs growth on a rotating substrate

and parasitic deposition on reactor walls
Reacting Flow
Models in FLUENT
By Graham Goldin and Genong Li, Fluent Inc.
LUENT 6.1 is unique in its vast offerings

for simulating reacting flow. Many
models are available for gas, solid, and
liquid fuels, for both gas-phase and surface
reactions.
For gas-phase combustion modeling, rapid
solutions can be obtained using the fast chemistry assumptions in the eddy dissipation (or
Magnussen), equilibrium mixture fraction,
or premixed models. These models are the
work-horses of current combustion simulations and are widely employed. The eddy dissipation model assumes that reactions occur
infinitely fast and that the reaction rate is limited by the turbulent mixing rate. The equilibrium mixture fraction model tracks the
progress of a mixture fraction and its variance rather than multiple species, and makes
use of a PDF function for the turbulence-chemistry interaction. It can include intermediate
species and radicals in a reaction system as
long as the reactions are fast and those species
can be assumed to be in equilibrium. The
flamelet model extends the equilibrium mixture fraction model to include finite rate effects
due to high flow strain rate.
The recent gas-phase combustion models in FLUENT incorporate full finite-rate chemistry into flame simulations. FLUENT can import
detailed kinetic mechanisms in CHEMKIN format, and the use of ISAT (see the articles on
pages 6 and 7) now makes the solution of
multi-dimensional chemistry simulations
affordable. The eddy dissipation concept (EDC)
model and the PDF transport model, imple-
PDF transport particles colored by temperature for a

turbulent jet diffusion flame
Fluent NEWS spring 2003
mented through a collaboration with

Professor Stephen Pope of Cornell, are available to account for turbulence-chemistry interaction. Both models allow for arbitrary complex
reaction mechanisms.
Coal combustion begins with the
devolatilization of fuel from the coal particle, and much effort has been spent on accurately capturing this process, since it
ultimately governs the flame temperature and
hence, pollutant formation. The recently introduced chemical percolation devolatilization
(CPD) model uses coal structure data based
on nuclear magnetic resonance measurements
to characterize devolatilization, and predictions of off-gases have agreed well with data.
Liquid fuel combustion begins with an accurate description of the spray and its breakup
into droplets. Evaporation follows, and subsequently, combustion. Models for all of these
processes are available in FLUENT and compatible with the deforming mesh model, so
that in-cylinder combustion can be readily
simulated.
Surface reaction modeling, used by
engineers in the semiconductor industry and
for applications such as gas reformers and
catalytic converters, involves reactions
between species in the gas phase and on a
surface. For a deposition process, gaseous
species are adsorbed at the surface, where
reactions take place. These reactions leave
behind deposited surface species and cause
the release of other species back into the gas
phase. For etching processes, there may be
no deposited surface species, only reactions
that produce species that are released into
the gas. To correctly model both deposition
and etching, FLUENT allows for three types
of species, which can be either reactants or
products: gas species, site species (adsorbed
at the surface), and bulk species (left
behind on the surface following the reaction).
Many other reaction models are also available. Those for the prediction of pollutant
formation, such as NOx and soot, are widely used and customizable, if desired.
Chemical reactions in packed bed reactors
can be simulated using either the porous media
model or the fixed-bed Eulerian granular multiphase model, with the added option of specifying different reactions in different zones.
In addition to the many built-in capabilities,
reactions that depend on micromixing or population balance theory can be simulated
through user-defined functions.
In summary, reaction modeling is a mature
capability in FLUENT that covers a wide range
of applications, and the articles on the next
several pages present just a sampling of these.
More examples can be found on our website, www.fluent.com, or by calling your local
Fluent office or distributor.
reacting flow
Three-Way Catalytic
Converter
Aftertreatment
By Yong Yi, Fluent Inc.
he majority of gasoline-fueled automobiles today have an emission control system that uses a three-way catalytic converter. The purpose of the three-way catalyst is to convert carbon
monoxide (CO), nitrogen oxides (NOx), and unburned hydrocarbons
to carbon dioxide, water, and nitrogen so that emissions from gasoline engines can be rendered less harmful to the environment.
Catalytic converters are built from structures called monoliths.
The monolith forms the basic framework of the converter, and acts
as an inert substrate for the catalytic coating. A layer of washcoat
is first deposited on the substrate, and the catalysts (often precious metals such as platinum, palladium, and rhodium) are then
deposited on the washcoat. In order to reach the required conversion efficiencies for a practical converter, the surface area for
reactions must be very large, and this large area is provided by
the monolith geometry and the highly porous washcoat. To optimize the design of a catalytic converter, it is important to investigate not only the flow field, but the chemical reactions and heat
transfer in the system as well. The distributions of temperature and
species throughout the device play an important role in its performance.
FLUENT 6.1 is a powerful tool for reaction simulations. In addition to offering a number of modeling options for treating reacting flow, a new reaction model is available for reactions and heat
transfer inside porous regions, such as the monolith in the catalytic converter. With the parallel computing capability in FLUENT,
this model can easily include the effects of multiple species and
reactions. Interoperability with CHEMKIN is also available, allowing FLUENT to read complex gas or surface reaction mechanisms,
if needed.
For the catalytic converter, the reaction mechanism is taken from
Reference 1. Exhaust gas, consisting of O2, N2, C3H6, H2, H2O,
CO, and NO enters the converter from one runner with a uniform
flow rate of 0.01 kg/sec and temperature of 600K. The wall of the
converter is assumed to be adiabatic. The surface-to-volume ratio
of the porous media is assumed to be 3000 m-1. The exhaust species
diffuse to the surface of the washcoat, and are adsorbed by platinum and rhodium to become sites species. Surface reactions take
place, and product species are released from the reacting surface
by desorption. Sixty-one surface reactions were used to model the
conversion of this mixture.
The temperature distribution on a plane cutting through the
exhaust pipes shows a temperature rise due to the catalyst reaction taking place. This result is reflected in contour plots of other
species concentrations as well.
Geometry and mesh of the generic 3-way catalytic converter
Temperature distribution on a cutting plane through

the exhaust pipes with only the left runner open
reference
1
Mass fraction of CO on the cutting plane through

the exhaust pipes with only the left runner open
Chatterjee D., Deutschmann O., and Warnatz J., Detailed Surface

Reaction Mechanism in a Three-Way Catalyst, Faraday Discuss., Vol. 119,
pp. 371-384, 2001.
reacting flow
Optimizing
Plant-Scale
LDPE Reactors
By Nitin H. Kolhapure* and Rodney O. Fox, Department of Chemical Engineering, Iowa State University, Ames, IA
*Presently at DuPont Engineering Technology, Wilmington, DE
ith the ever-increasing availability of high-performance computing tools, CFD is becoming

a significant technology, though still not dominant, for reactor design in the chemical process industry. CFD is emerging as a design tool for the
development of new processes and optimization of existing ones at a fraction of the cost and time of traditional
experimental and pilot-plant approaches. At Iowa State,
the ability of CFD to simulate turbulent reacting flows
in processes involving fast, mixing-sensitive reactions has
been investigated. These flows are characterized by interactions between large and small chemical and mixing
time scales that play a significant role in determining
reactant consumption (yield), product quality (selectivity),
and reactor stability. Traditional reactor models based
on idealized flow assume perfect micromixing and fail
to account for such interactions.
To improve upon these models, a comprehensive CFD
algorithm that links FLUENT with a sub-grid-scale multienvironment micromixing (MEM) model and detailed
low-density polyethylene (LDPE) chemistry has been developed for plant-scale tubular reactors. In LDPE reactors,
a small amount of initiator is injected into a preheated
monomer flow to start a complex series of reactions that
produce polymers of varying length (molecular weight).
These reactors are extremely sensitive to local mixing
conditions due to stiff and highly exothermic kinetics

and hence, they serve as an excellent test case for commercial reactors where control of the reaction conditions and optimization of the reactor performance (i.e.,
reactor stability, initiator efficiency, polymer molecular
weight distribution) are desired.
An interactive interface was created for the project
using user-defined functions (UDFs) in FLUENT. C routines for the MEM model and FORTRAN routines for a
customized in-situ adaptive tabulation (ISAT)1 algorithm
for the LDPE chemistry were compiled and linked to
FLUENT. The continuity equation, the k- model, the MEM
model, and the chemistry were solved sequentially at
each grid point in a 2D axisymmetric computational domain.
An unsteady coupled implicit solver was chosen to limit
the effects of truncation errors on the solution. The UDF
interface updated the mixing and chemical source terms
at each time step as per the formulation in the MEM model
and the ISAT algorithm. The interface also provided an
ability to account for the inter-dependence of the kinetic, physical, and thermodynamic properties of the polymer reaction mixture. ISAT enabled the inclusion of a
total of 16 species and offered ten-fold computational
gains by replacing the conventional direct integration
with a less expensive multi-linear interpolation. It proved
to be a powerful technique to include chemistry calculations in CFD without restricting the degrees of freedom of the chemical composition vector. More details
of the CFD algorithm with the UDF interface and the
MEM model can be found elsewhere.2,3
The CFD results demonstrated the capabilities of the
algorithm to capture the strong coupling between micromixing and complex chemistry and predict the complete
reacting flow information, including species and temperature distributions close to physical reality. The flow
information at the micro-scale provided important insights
into the occurrence of small-scale temperature fluctuations (hot spots), deterioration of polymer quality, and
The mean mass fractions for initiator (top, 0 to 1.15x10 -3), monomer (middle, 0.95 to 1), and temperature
(bottom, 250 to 307C) inside a tubular reactor (d = 3.8 cm, L = 10 m)
The injection region (0 to 0.2 m) is zoomed

in to highlight the non-uniform initiator
distribution, which caused a loss of 64%
initiator compared to plug-flow conditions
reacting flow
The
Berl Combustor Revisited
By Graham M. Goldin, Fluent Inc.
jets inject natural gas. The mixture then enters

a quarl, which expands to a hexagonal combustion chamber. Because of its complex physics
and ample supply of experimental data, the
BERL combustor has served as a benchmark
test case for combustion models in FLUENT
for many years.
Since the fuel jets in the cross-flow air-stream
cannot be accurately modeled in 2D, a 3D sector representing 1/24th of the burner is modeled. The simulation makes use of the
realizable k- turbulence model, and the P1
radiation model. The chemistry is described
by a 9 species Augmented Reduced Mechanism
(ARM), which was derived from the detailed
natural gas mechanism by making steady-state
assumptions for certain species.2
Results for radial NO predictions are in good
agreement with experimental measurements
at 27mm and 432mm downstream of the quarl,
despite the many assumptions made in modeling the turbulence, chemistry, radiation, and
their interactions with each other. Radial profiles of temperature and other species concentrations follow the same trends. In addition,
ISAT provides a net speed-up of 65 for this case.
Without ISAT, a simulation that can be completed overnight would require a month of
run-time!
loss of initiator under extreme operating and

mixing conditions. The influence of feed temperature, initiator concentration, and degree
of premixing on steady-state reactor performance
was helpful in making wiser, more well-informed
operational decisions. By replacing pilot-plant
tests, CFD offered a low-cost alternative to explore

a variety of design options for optimizing initiator consumption while controlling the product quality and reactor safety. Though
validation of such a CFD approach against key
experimental data remains an integral and essen-
Contours of NO on a center plane near the quarl
120
80
Pope S.B., Computationally Efficient

Implementation of Combustion Chemistry Using
In-Situ Adaptive Tabulation, Combustion Theory
and Modeling, 1, pp. 41-63, 1997.
http://www.et.byu.edu:8080/~tom/Papers/
Hemant-WSS96/WSS.html.
60
40
20
references
1
FLUENT 27 mm
FLUENT 432 mm
Experiment 27 mm
Experiment 432 mm
100
NO (ppm)
he focus of FLUENT 6.1 gas-phase combustion modeling is to provide affordable,

detailed, finite-rate chemistry. With the new
models, kinetically controlled processes such
as pollutant formation (NOx, CO, etc.) and
flame ignition/extinction can be simulated with
high fidelity.
The difficulty in including detailed kinetics
is the extreme non-linearity of the chemical
mechanism. Large computational times are
required to integrate the equation set, and special care is required to properly couple the chemistry with the turbulent flow. For these two
reasons, most commercially available chemistry codes are limited to physical dimensions
of zero or one.
To overcome the massive computational
demands of detailed chemistry simulation in
2D and 3D domains, FLUENT 6.1 incorporates
ISAT (In-Situ Adaptive Tabulation1), which can
accelerate chemistry calculations up to a thousand-fold. For a chemical mechanism with N
species, ISAT builds N-dimensional chemistry
tables during the simulation. The expensive
kinetic integrations are mitigated by retrieving the appropriate values from the table. ISAT
can be used with two turbulence-chemistry
interaction models in FLUENT 6.1: the Eddy
Dissipation Concept (EDC) model and the PDF
Transport model.
To demonstrate the power of ISAT, a
FLUENT 6.1 simulation of Sandias Burner
Engineering Research Laboratory (BERL) industrial combustor has been performed using the
EDC model. The BERL combustor consists of
an annulus with swirling air, into which 24 fuel
0
0.0
0.1
0.3
0.2
0.4
radial position (m)
0.5
0.6
Comparison of FLUENT predictions of NOx with experimental

data for radial scans 27mm and 432mm downstream of the
quarl
tial part of the design procedure, it opens greater opportunities for the development of safe and efficient chemical processes at reduced costs and time. The study has
brought turbulent reacting flow simulation for singlephase finite-rate chemistry closer to realistic chemical
process engineering applications.
references
1
Pope S.B., Computationally efficient implementation of combustion chemistry using in situ adaptive tabulation.
Combustion Theory and Modeling, 1:41-63, 1997.
Fox R.O., Computational methods for turbulent reacting

flows in the chemical process industry. Revue de lInstitut
Franais du Ptrole, 51:215-243, 1996.
Kolhapure N.H. and Fox R.O., CFD in polymer reaction engineering: Combining polymerization chemistry and detailed
flow models. DECHEMA Monogr., 137:247-271, 2001.
The effect of micromixing is shown through local temperature fluctuations in the reacting environment
(top, 250 to 329C) and higher polydispersity (molecular weight distribution) (bottom, 0 to 7.15)
reacting flow
Deposition:
One Layer
at a Time
s device sizes continue to shrink below 90nm,
the semiconductor roadmap suggests that
atomic layer deposition, or ALD processes will
be required for a variety of applications, such as the deposition of barriers for copper interconnects, the creation
of tungsten nucleation layers, and the production of highly conducting dielectrics. In the ALD process, two or more
precursor gases flow over a wafer surface in an alternating manner, so that the gases can react with the sites
or functional groups on the wafer surface. When all of
the available sites are saturated, the reaction stops and
an inert gas flow purges the excess precursor molecules
from the region. The process is repeated, as the next
precursor gas flows over the wafer surface. A cycle is
defined as one pulse of precursor 1, purge, one pulse
of precursor 2, and purge. This sequence is repeated
until the final thickness is reached. These sequential, selflimiting surface reactions result in one monolayer of deposited film per cycle.
ALD is a stable process over a wide range of temperatures,
and as a result there is a linear relationship between the
thickness of the layer deposited and the number of deposition cycles. Because of the periodic pulsing of reactants and purge gases in short intervals, the ALD process
is challenging to simulate using CFD. The modeling strategy for one cycle should include convective and diffusive transport of reactants to the surface, transient boundary
Ti
N
0.6
0.4
cumulative thickness (Ang)
growth rate (Ang)
site coverage
A typical ALD reactor along with pathlines

colored by velocity, and TiN growth rate on the wafer
1.0
0.8
0.8
0.6
0.4
0.2
0.2
0.0
0
Where (a) is an adsorbed site/surface species, (g) is a gas phase species,

and (s) is the deposited solid or bulk species.
The TiN growth in a typical ALD reactor was investigated in FLUENT 6.1
by looking at the number of Ti atoms in the TiN layer per unit area, as a
function of the number of deposition cycles. Transient simulations indicated
that the gaseous concentration of TiCl4 increases during the first pulse, as
does the adsorbed site species TiCl2. After the purge, an NH3 pulse is initiated and reacts with the site species TiCl2. No cross-contamination between
the pulse cycles was observed, suggesting that the purge time could be
reduced. The model also predicted that the growth rates are uniform over
successive ALD cycles.
The results shed light on different growth regimes: a transient regime
where film thickness for one deposition cycle increases towards a constant
value, and a saturated regime where film thickness for a given deposition
cycle is constant. The simulations also suggested that growth rate has a
linear relationship with the number of deposition cycles. The CFD model
provides a framework for investigating the influence of geometrical parameters and different precursor doses on film thickness. The results implied
that the NH3 dose significantly affects the TiN film growth. For low ammonia dosing, the growth rates are significantly lower than those associated
with high ammonia doses.
1.2
TiCl2
NH2
1.0
NH2(a) + TiCl4(g) TiCl2(a) + 2HCl(g) + N(s)

TiCl2(a) + 2NH3(g) 2NH2(a) + 2HCl(g) + Ti(s)
By Balaji Devulapalli, Fluent Inc.
1.2
conditions to account for pulse sequences, adsorption of the monolayer of

the first precursor on the wafer surface, and subsequent surface reactions
with the second precursor gas.
One important application for ALD is the deposition of a TiN (titanium nitride)
diffusion barrier using TiCl4 (titanium tetrachloride) and NH3 (ammonia) precursors. Since the properties of barrier films are related to their nucleation and
growth mechanisms, a detailed surface chemistry model is needed to characterize adsorption, desorption, and heterogeneous surface reactions on the
wafer surface. The self-limiting surface reactions used in the model are:
10
20
30
40 50
time (s)
60
70
80
0.0
0
10
20
30
40 50
time (s)
60
70
80
40
Ti
N
20
0
0
10
20
30
40 50
time (s)
60
70
80
Fractional coverages of TiCl2 and NH2 (left), growth rate for Ti and N as a function of pulse time (middle), and predicted film thickness at the wafer center over 10 cycles (right)
reacting flow
Mass fraction of CH4
n many industrial applications, such as wafer processing

by chemical vapor deposition (CVD), combustion of
solids, chemical etching, and catalytic combustion, reactions occur on wall surfaces. During the last few years,
detailed mechanisms for certain surface reactions, including rate constants, began appearing in the literature.
Although there is a strong need to simulate such problems numerically, especially in conjunction with flow field
simulations that include gas-phase reactions, the stiffness of the reaction systems and the complex
mass/energy interactions between the surface species
and gas phase species often make such problems difficult to solve.
The new surface reaction model in FLUENT 6.1 allows
for arbitrary, complex reaction mechanisms, involving
any number of gas phase and/or surface species, and
reactions between different gas species, gas and surface
species, and different surface species. Surface reactions
are fully coupled with the flow simulation, so that the
distribution (coverage) of different surface species and
the deposition rate of bulk species on a wall can be tracked.
In addition to dealing with reactions on actual walls,
the model can also account for surface reactions on unresolved walls in a porous medium. Moreover, there is full
flexibility in the problem setup different reaction mechanisms can be defined in different fluid or porous zones.
To illustrate the new capabilities in FLUENT, a tubular catalytic reactor, which represents a single pore of
an actual monolithic catalyst, was simulated. The tubular reactor is 10cm long and 2mm in diameter. The inner
tube surface is coated with platinum (Pt), which serves
as a catalyst to initiate and/or accelerate reactions. A
22-step reaction mechanism1 was used in the simulation.
The system involves seven gas phase species (CH4, O2,
H2, H2O, CO, CO2, and N2) and eleven site (surface)
species (H(s), O(s), OH(s), H2O(s), C(s), CO(s), CO2(s),
CH3(s), CH2(s), CH(s), and Pt(s)). Both gas phase species
and surface species can be depleted or created as a result
of surface reactions. This causes the concentrations of
gas phase species and the coverage of site species to
change along the pipe. The results show that methane
is oxidized quickly after it enters the pipe, and that CO
and CO2 are produced. Changes in site species coverage include increases in H2O(s), H(s), and OH(s), and
decreases in CO2(s) and C(s).) The coverage of some surface species, such as platinum and oxygen, remains relatively constant. Due to the effective numerical algorithm
in FLUENT for reaction simulations such as this, convergence
of this example was rapid and well-behaved.
Mass fraction of CO
Contours of mass fraction of the major gas species; diameter exaggerated for clarity
Surface
Reactions in
Catalytic Tubes
By Genong Li, Fluent Inc.
100
os
pts
10-1
10-2
ohs
surface coverage
10-3
10-4
10-5
h2os
hs
10-6
Pt catalyst, T = 1290K
CH4
Air
Mass fraction of CO2
cos
d = 2mm
-7
10
L = 10cm
Schematic of the problem, with the diameter greatly

exaggerated
10-8
reference
10-9
Deutschmann O., Maier L.I., Riedel U., Stroemman A.H., and

Dibble R.W., Hydrogen Assisted Catalytic Combustion of
Methane on Platinum, Catalysis Today 59, p. 141-150, 2000.
cs
0.025
0.050
Z (m)
0.075
0.1
Surface coverage profiles along the length of the tubular reactor
chemical
Bubbling Columns
By Vivek V. Buwa and Vivek V. Ranade, Industrial Flow Modeling Group, National Chemical Laboratory, Pune, India
Instantaneous gas volume fraction (left) and liquid velocity vectors

(right) for a superficial gas velocity of 0.14 cm/sec, simulated using a
single-group Eulerian multiphase model (H/W: 4.5)
10
A set of simulations was performed to study

the effects of superficial gas velocity, sparger configuration (including bubble diameter), and the height-to-width (H/W) ratio
of the column on the low frequency oscillations and time-averaged flow variables,
such as vertical liquid velocity and gas holdup. The results indicated that the dynamic characteristics are sensitive to bubble size,
as produced by different sparger configurations. See Reference 2 for additional details
of experimental measurements and CFD
simulations.
The single-group findings prompted the
launch of a project to develop a multi-group,
or multi-fluid model based on a discrete
population balance methodology. The population balance model, which accounts for
bubble coalescence and break-up, was developed and mapped onto FLUENT through
user-defined functions. While developing
population balance models, it is essential
to ensure the conservation of certain properties of the bubble population. During coalescence and break-up processes, 1) the
mass of bubbles should be conserved, 2)
the number of bubbles should be appropriately reduced or increased, and 3) the
interfacial area should be appropriately
reduced or increased. Since bubble population is represented by a finite number
of groups, it is difficult to satisfy all of these
three conditions simultaneously. Thus, in
this work, mass conservation and adjustments to the bubble number were incorporated in the population balance models,
but adjustments to the interfacial area were
0.020
time-averaged gas-hold-up
nsteady multiphase flows are frequently encountered in chemical

process equipment. Bubble column
reactors, even though simple in construction,
are characterized by a host of inherently
unsteady complex flow processes with widely varying scales of space and time. For example, recirculating reactor-scale flow processes
coexist with microscopic flow processes
around individual bubbles. The overall multiphase fluid dynamics controls the fluid
mixing and inter-phase transport processes, which in turn determine the reactor performance. Most of the early work in this
area was focused on predicting time-averaged flow properties with the help of a few
adjustable parameters. While time-averaged
characteristics can help provide general guidelines for reactor design, the effects of the
unsteady flow characteristics are lost.
Experimentally validated CFD models
need to be developed, therefore, for accurate prediction of the dynamics of gas-liquid flows in bubble columns.
At the National Chemical Laboratory, a
rectangular bubble column has been constructed for this purpose. Its geometric simplicity allows for systematic experiments and
numerical simulations with minimal computational demands. The specific geometry was chosen to complement earlier
experimental work1. A jet of air, injected at
the center of the base of the water-filled column, gives rise to a meandering plume of
bubbles. Wall pressure fluctuations have been
recorded to characterize low frequency oscillations that correspond to local recirculating flow. The effect of various design and
operating parameters on the plume oscillation period has been investigated. In addition to wall pressure fluctuation measurements,
single-tip voidage probes have been used
to record the local instantaneous void fraction.
CFD simulations of the bubble column
have been performed using several multiphase approaches. The Eulerian multiphase
model in FLUENT was initially used to simulate the 3D, unsteady gas-liquid flow. Each
of these so-called single-group simulations used water for the primary phase and
a single secondary phase of air bubbles represented by an average bubble diameter.
Exper.
17x25x7
32x47x11
61x92x19
0.016
0.012
0.008
0.004
0.000
0.00
0.04
0.08 0.12 0.16

column width (m)
0.20
Time-averaged gas holdup calculations show

the dependence on grid density (Superficial
gas velocity: 0.14 cm/s, H/W: 2.25)
chemical
14
(a)
0.20
instantaneous gas hold up over 0.1s
0.18
instantaneous gas hold up over 1.0s
0.16
gas hold-up
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
10
20
30
time (s)
40
50
(b)
0.14
recorded for each 0.01s

moving average of 1.0s
local gas hold-up
0.12
0.10
0.08
0.06
0.04
0.02
0.00
10
30
20
time (s)
40
50
10
20
30
time (s)
40
50
(c)
0.016
0.014
local gas hold-up
not. With this approximation, the maximum error in predicted interfacial area is about 10% for the smallest group, and it decreases for
larger groups. The model has been used to study the evolution of the
bubble size distribution in bubble column reactors, and has shown
reasonably good agreement with experimental measurements. The
results are encouraging and the model is being extended to various
other multiphase systems, such as stirred tank reactors. These models can be easily extended to simulate gas-liquid mass transfer.
In another set of simulations, the Lagrangian discrete phase model
(DPM) in FLUENT was used to follow the motion of individual bubbles. This approach provides information on bubble scale processes,
which is necessary for any rigorous modeling of reactions and heat
and mass transfer. The simulation results have been validated against
experimental measurements. For example, the plume oscillation period calculated from the numerically predicted voidage fluctuation time
series using DPM simulations agrees well with the experimental measurements and Eulerian simulations. The time-averaged vertical liquid
velocity (based on LDA measurements1) and gas hold-up measured
at different column heights are in reasonably good agreement with
both the Eulerian and DPM approaches. The power spectrum of bubble passage frequencies obtained by the transient DPM simulations
also shows good agreement with experimentally measured bubble passage frequencies.
Gas-liquid and gas-liquid-solids flows in cylindrical bubble columns
have also been studied. The gas-liquid flow was found to be highly
chaotic in comparison to the quasi-periodic flow observed in the rectangular bubble column. Single-group simulations using the Eulerian
multiphase model were carried out for the gas-liquid mixtures, and
a few three-phase simulations (with gas, liquid, and granular phases)
were carried out for the gas-liquid-solids mixtures to study the effect
of solids loading on key dynamic and time-averaged flow properties.
The results, which have been compared to measurements, are still preliminary but are encouraging. They will be used in the future to help
clarify the dynamics of complex multiphase flows in bubble columns.
In another set of simulations, CFD models were used to predict mixing time, an important parameter for reactor engineering. In these
simulations, the liquid phase mixing was simulated using transient and
time-averaged flow. The mixing time values obtained using time-averaged flow were found to be much larger than those obtained by fully
transient flow. For example, at a superficial gas velocity of 0.14 cm/s,
the mixing time obtained using time averaged flow was 26.2 s in comparison with 15.4 s obtained using unsteady flow. The latter agrees
well with the experimentally measured mixing time of 16.0 s. The effects
of H/W ratio, sparger configurations, and gas velocities on the liquid
phase mixing time have been investigated using CFD as well, and the
results have been validated using experimental measurements.
0.012
0.010
0.008
0.006
0.004
0.002
0.000
Instantaneous bubble/gas volume fraction distribution and corresponding voidage

fluctuation time series obtained from (a) experiments, (b) DPM, and (c) Eulerian
multiphase simulations (Superficial gas velocity: 0.14 cm/s, H/W: 2.25)
12
10
8
6
References
4
2
0
0.0
wall pressure fluctuation measurements

Eulerian simulations
DPM simulations
0.2
0.6
0.4
superficial gas velocity (cm/s)
0.8
1.0
Pfleger D., Gomes S., Gilbert N., and Wagner H.-G., Hydrodynamic simulations of
laboratory scale bubble columns fundamental studies of the Eulerian-Eulerian modeling approach, Chemical Engineering Science, 54, p. 5091-5099, 1999.
Buwa V.V. and Ranade V.V., Dynamics of gas-liquid flow in a rectangular bubble column: Experiments and single/multi-group CFD simulations, Chemical Engineering
Science, 57, p. 4715-4736, 2002.
The plume oscillation period predicted by the Eulerian multiphase and

discrete phase models, compared to experiment (H/W:2.25)
11
oil & gas
spillweir
baffles
inlet
region
water
outlet
Separating
Water & Gas
By Steve Turner, Zeta-pdm Ltd., Isle of Wight, UK
The interface between the water and gas close to the

vessel inlet
% of particles settled
100
80
60
40
20
0
inlet design 1
inlet design 2
increasing particle size
Settling efficiency for sand for a range of particle sizes and

two inlet designs
12
eta-pdm Ltd. is a world leader in the oil

& gas industry, focusing predominantly
on separation processes. In the recent
design of a separation vessel, Zeta-pdm engineers wanted to investigate the effect of different inlet designs and baffle arrangements
on separation efficiency. The separator will be
used to process a mixture of water, hydrocarbon
gas, sand, and oil. During the process, water
will separate from the gas and oil while sand
particles settle to the bottom of the container. CFD was used to simulate the multiphase
flow field for a number of prospective
designs, so that engineers could improve their
understanding of the separation process, and
assess the designs for the improvement each
would offer.
The separation vessel consists of an inlet
region, a series of specially designed baffles,
and separate outlets for the water and gas.
The flow pattern is a function of the inlet design,
baffle design and position, vessel dimensions,
inlet velocity, and mixture composition.
Material enters the vessel in a highly turbulent state, consisting of liquid, bubbles, and
particles. Since the quantity of oil in the mixture is very small, it is neglected for the purpose of the simulation. The material passes
through the baffles, which work to calm the
flow and enhance the separation process. The
flow conditions need to be controlled so that
the sand particles entering the vessel can settle, the gas bubbles can rise, and each constituent can be removed from the vessel through
the desired outlet.
The multiphase mixture was simulated in
FLUENT using the Eulerian multiphase model
in combination with the discrete phase

model (DPM). The Eulerian model, in which
separate sets of fluid equations are used for
each fluid, was used to track the bulk separation of the gas and water phases, the two
primary ingredients in the incoming mixture.
The sand particles were simulated using the
DPM, since it is the most efficient way to track
the motion of particles with a range of sizes.
Gas bubbles were also tracked using this method,
to assess the separation efficiency of the vessel as a function of bubble size. A hybrid mesh
of approximately 600,000 cells was used. The
solution was performed on a network of computers, using FLUENTs parallel processing capability.
Examination of iso-surfaces of the gas-water
interface close to the inlet is one way that engineers analyzed different inlet designs. The fluids are highly turbulent and well mixed in this
region, but the large volume of buoyant hydrocarbon causes the gas to rise rapidly and separate from the water. Different designs were
found to make this process more or less efficient. Examination of planar contours of volume fraction near the inlet and first few baffles
were used to assess the inlet design as well
as the calming effects of the baffles. Using the
DPM results, different inlet designs were evaluated for their ability to separate bubbles and
sand particles of a range of sizes. Results showed
that while two designs may perform comparably for small bubbles and particles, one worked
better for larger sizes. The results were used
to finalize the design for the new separator,
which is now in operation.
biomedical
as masks are prime protectors of emergency

responders in toxic environments. Researchers
at the Prins Maurits Laboratory of TNO build,
test, and improve gas masks for protection against
chemical and biological warfare agents. Gas masks
protect the wearer by purifying air in filter canisters. After filtration, the air is distributed inside the
gas mask to minimize the thermal load on the face
and condensation of water on the eyepieces. TNO
develops new designs of gas masks using CAD/CAM
methods, and its researchers use CFD to study the
masks without the need for building costly prototypes. The term digital gas mask has been given
to gas masks that are developed and studied in this
manner.
The first digital gas mask was based on an existing gas mask supplied by the Dutch Ministry of Defense,
who funded the project. Creating the computational
mesh was a challenge, since a digital CAD/CAM model
was unavailable. To generate the gas mask geometry, a different strategy was followed instead: a 3D
scan of the gas mask was made using tools from
the German company, Vitronic GmbH. The geometry could not be scanned directly because of the
complex internal shape of the gas mask while worn
on the face. The geometry was therefore built up
by scanning three parts: the face piece, which covers the entire face; the nose piece, a smaller mask
piece surrounding the nose and mouth that is located inside the face piece; and a mannequin head.
Special software was used to unify the parts in a stereolithography (STL) file format. Engineers from
Fluent Europe imported the STL file into GAMBIT
for meshing. The end result was an unstructured
mesh comprised of 290,000 cells. Using this mesh,
FLUENT simulations were initiated to study the flow
patterns inside the mask that are normally hidden
from view, and to study the residence time distribution (RTD).
The studies of flow inside the mask were focused
on the vicinity near the eyepieces during the breathing cycle. Because of moisture in the breath, there
is a likelihood that water will condense on these surfaces. CFD calculations were used to show that during inhalation, flow from the inlet rapidly introduces
a supply of fresh air to the region around the eyepieces in a swirling pattern. This periodic freshening of the eyepieces prevents water condensation
from developing.
The RTD calculations showed that during inhalation, 25% of the tracer particles (injected at the inlet)
leave the face mask at the outlet after the average
residence time is reached (computed as volume divided by volumetric flow rate). This simple exercise indicates that there are some dead zones or recirculation
regions inside the mask. Future efforts will focus on
locating and minimizing these regions using CFD.
Over the next year, plans are in place to improve
the CFD techniques for the digital gas mask. This
tool will help TNO optimize the design of gas masks
that are currently being developed using CAD/CAM
methods.
Digital
Gas Mask
By Coen van Gulijk, TNO Prins Maurits Laboratory, The Netherlands
3D scanning
Courtesy of Vitronic GmbH., Germany
real gas mask
digital gas mask
CFD model of a face mask after scanning;

external features are omitted
Close-up of flow
patterns near
the eyepiece
13
biomedical
AccuSpray
on Demand
By Vince Sullivan, PhD and Anjana Bhuta-Wills, BD Technologies, Research Triangle Park, NC;
and Shoreh Hajiloo, ICEM CFD Engineering, Livonia, MI
The BD AccuSpray Delivery System is
designed to reach acceptable pressure
for actuation with minimal variability
Fluid is forced into the swirl chamber through a valve, resulting in rotational
flow prior to release
14
300
average particle size (m)
n accurate CFD model is a potentially powerful

development tool for evaluating and predicting
the performance of medical devices. In studies
conducted at BD Technologies, FIDAP was used to evaluate the operation of the BD AccuSpray Nasal
Delivery System. The model developed in this study accurately simulated fluid flow through the device at appropriate device actuation velocities.
The BD AccuSpray device is a nasal delivery system
based on BDs Hypak prefillable syringe technology. It
creates a spray by forcing liquid through a pressure swirl
atomizer when the user depresses the plunger on the
device. A thin intact sheet of liquid is formed in the shape
of a cone at the exit orifice, and breaks up into droplets
of an appropriate size for delivery of drugs to the nasal
mucosa. The AccuSpray device has been used to deliver a new live attenuated influenza vaccine, which has
recently completed Phase III clinical trials, and is currently under review at the FDA. Other intranasal vaccines are in the preclinical testing stage.
correlation 1 calc. SMD (m)

correlation 2 calc. SMD (m)
experimental SMD (m)
250
200
150
100
50
0
27
33
80
40
velocity (mm/sec)
Average particle
size (Sauter
mean diameter)
as a function of
plunger velocity
for two
correlations and
experimental
measurements;
agreement is best
when velocities in
excess of 80
m/sec are used
100
A model was developed using four design stages. First, internal single-phase
flow characteristics were modeled. Second, a 2D model of the two-phase atomization flow at the device nozzle was simulated. Third, empirical equations were
obtained from the literature, relating particle size to model output parameters
such as cone angle and intact sheet thickness at the nozzle; these were used
to validate the model. Fourth, a 3D , two-phase flow of the atomization process
was simulated using FIDAPs volume of fluid (VOF) model. These results were
then compared to average spray particle sizes produced by AccuSpray, as determined experimentally by a Sympatec laser diffractometer.
FIDAP was used to illustrate the path of the fluid as it moves past the valve,
into the swirl chamber and out the nozzle. In addition, the VOF models were
able to simulate the formation of a hollow cone, and the initial jet breakup of
the fluid as it leaves the AccuSpray nozzle.
Predictions of average particle size were obtained using semi-empirical correlations based on the fluids physical properties and the FIDAP results. There
was a close correlation between the model and experimental results, indicating that the model accurately simulates the AccuSpray operation. The experimental particle size data showed that reproducible particle size in the appropriate
range for nasal deposition occurs when plunger velocities exceed 80 mm/sec.
While the experimental results were closely matched by model results at higher velocities, the model underestimated particle size at lower velocities. The reason for the discrepancy at lower actuation rates is believed to be due to the
fact that the model assumes fully turbulent flow. This assumption appears to
be correct only at higher velocities.
Using CFD to model liquid atomization is complex because the random physical instabilities at the air-liquid interface that cause atomization can be difficult
for CFD to capture. Even so, the AccuSpray study has given BD confidence that
their new design does allow the user to reliably reach an actuation velocity high
enough to achieve full spray atomization. The results of this study indicate the
power of CFD modeling, when combined with rigorous analytical testing, to accurately predict device performance.
biomedical
Locating the
Nasal
Valve
with
Flow pathlines colored by velocity magnitude

for a simulation of steady-state inhaled
breathing with a plug flow inlet condition
FIELDVIEW
By R.A. Segal, J.M. Sheppard, J.S. Kimbell, CIIT Centers for Health
Research, Research Triangle Park, NC
t the CIIT Centers for Health Research (CIIT),

FIDAP is being used to model airflow in
the human nasal passageways. Computed
airflow patterns are used to simulate gas uptake
and particle transport to determine the potential toxicological risk of inhaled materials. The accuracy of the nasal airflow predictions are tested
by constructing a hollow plastic replica of the nasal
cavity using stereolithography, a rapid prototyping
process, and comparing simulated pathlines with
dye streaklines in the hollow mold.
Another way to test model accuracy is to compare simulated nasal pressure drop and resistance
with measurements made in people. Most of the
nasal resistance measured in people has been reported in the vicinity of the nasal valve, an area near
the front of the nose where the cross-sectional
area of the airways is smallest. However, the nasal
valve is difficult to locate because its exact location is different in each individual. An approximate location can often be found, however, using
acoustic rhinometry (AR), a noninvasive process
that calculates the nasal cross-sectional area and
volume as a function of distance into the nose

by acoustic reflections from a hand-held wand
placed against the nostril.
Last year, a summer intern at CIIT was assigned
the task of locating the nasal valve in a FIDAP
model using FIELDVIEW. To do this, planes at different tilts were swept through the nasal model,
and the resulting cross-sectional area on each plane
was calculated. The intern was able to organize
the search using FIELDVIEWs scripting language,
FVX, to compute cross-sectional areas throughout the nasal model. By using the iso-surface definitions and the integration tool, she was able to
sweep through the model with planes at various angles. To validate the procedure, the crosssectional areas were compared to those acquired
from AR measurements of the corresponding stereolithography mold.
Using AR, measurements were made on the
right and left side of the hollow plastic nose independently. The thresholding abilities of FIELDVIEW
were used to isolate the different sides in the model.
The sweep plane for calculating cross-sectional
area was allowed to tilt from left to right as well

as front to back because the positioning of the
AR wand may not be perpendicular to the airspace and therefore may select a minimum plane
whose orientation is not straight. FIELDVIEW was
useful because it allowed automation of this task
by looping through sweep planes with different
definitions. The data were continuously output
during the running of the script so that analysis
of the numbers could take place as the computations were progressing.
This process allowed the location of the nasal
valve to be found in a systematic and reproducible
way. In addition, the results compared fairly well
with the AR data calculated from the stereolithography model. This information provides credibility for the nasal models developed at CIIT, and
helps build the interface between measurements
made on people in clinical settings and simulations of biological systems.
The authors thank Dr. Matthew Godo from
Intelligent Light for his assistance with this project.
FIELDVIEW is a
registered trademark
of Intelligent Light.
90
60
45
15
Orientation of several sweep planes used for locating the nasal valve
Location of the nasal valve in the original human nasal model
15
biomedical
The
Heartbeat of
Pulmonary
Modeling
By Rob Woolhouse, Fluent Europe Ltd.
he use of CFD in new and diverse application areas

is becoming more widespread. In one exciting new
field, magnetic resonance imaging (MRI) technology is being combined with computer modeling to simulate the flow in complex physiological channels within
the human body. One advantage of using CFD for this
purpose is that it allows multiple tests and experiments
to be carried out to minimize clinical research. Another
is that virtual simulations on an individual prior to surgery can make the operation proceed more smoothly
and result in a more successful outcome.
As an example of this promising new capability, flow
through a pulmonary artery has recently been studied.
These results are being used to highlight areas that can
result in clotting sites or aneurysms. Clots are formed
in low and stagnant flow regions where low fluid shear
and high residence times are observed. Conditions such
as high surface pressure, shear stress, or strong gradients can result in an aneurysm, where the vessel wall
bulges outward, forming a pocket. The repair of aneurysms
The pulmonary artery,

imaged using the
FLUENT 6 mesh
16
biomedical
is normally done in a surgical procedure in which
a stent is inserted to stabilize the vessel. By using
CFD, the stent location and design can be modeled prior to the operation to determine the optimum size and orientation of the device, reducing
the risk of unintentional damage and the time required
for the procedure.
The actual process of converting patient data
into a suitable CFD geometry is not trivial. Many
steps are required, and for the pulmonary artery
project, several of these involved the collaboration
of Materialise (based in Leuven, Belgium) and their
proprietary software, Mimics and Magics. Mimics
converts MRI slices into a 3D solid model, and exports
in a variety of CAD compatible file formats, including stereolithography (STL). Magics is a dedicated STL editor with a comprehensive set of surface
repair tools.
For the pulmonary artery project, an MRI scan
of a chest cavity was obtained from the Sheffield
University MRI Unit. MRI scan slices are typically produced in a greyscale pixelated DICOM format, and
these were joined together to create a 3D solid model.
Vessels not connected to those of interest, as well
as bones and other tissue, were removed. The 3D
solid model was then exported as a 3D surface in
STL format, and further edited to remove all additional unwanted features, leaving the pulmonary
artery and its primary branches for the CFD model.
The resulting smooth geometry of the artery was
then read into GAMBIT, where a tetrahedral surface
mesh was created. Due to the complexity of the
model, further surface mesh adjustment and volume meshing were performed in TGrid.
The physics of blood flow through the body has
been the focus of a number of studies over the years,
and many of the findings were incorporated into
the current model. For example, fluid structure interaction can be neglected for the pulmonary artery,
since the thick vessel wall is designed to carry large
Mimics software from Materialise is used to create

a geometry of the chest cavity, and then bones,
lungs and unconnected vessels are removed
Pathlines inside the artery and
branches colored by velocity
quantities of blood under (relatively) high pressure, directly away from the heart. Initial checks
also confirmed that the flow regime was laminar. Because blood is a non-Newtonian fluid,
the shear effect on viscosity needed to be considered. The Carreau-Yasuda model was implemented through a user-defined function (UDF).
A velocity boundary was applied to the single
large inlet, with a transient, periodic profile that
reflects the flow supplied by the heart. Pressure
outlet boundary conditions (of equal pressure)
were used for the multiple outlets in the model,
and the flow split was determined by the vein
geometry.
Plots of velocity vectors indicate that there
are no recirculation regions or dead zones within the artery or its primary branches, which was
expected, since the scans were taken from a healthy
adult. Surface contours of wall shear stress show
an increase near some of the constrictions in the
vessels. However, it is unlikely that these sites
would result in the formation of an aneurysm,
since the flow in these regions is not directed
toward the surface.
Overall, this emerging technology shows promise for medical procedures in the future, since
it can provide important information specific to
an individual using non-invasive tools.
The author thanks the Sheffield University MRI Unit
for their assistance with this project.
Fluent Partnership
aterialise, based in Leuven,
Belgium, is a market leader in
the provision of rapid prototyping and software solutions for the medical device industry. It has a medical specialist
division that offers software for the production of medical models, which can be
used as masters for implants, or for the
planning of implant surgery. Under a newly
formed partnership, Materialise and Fluent

Europe will explore how FLUENT software
can interface with and extend the scope
of Materialises medical prototyping
solutions.
more.info@
www.materialise.com
Wall shear stress on the surface

of the artery and branches
17
marine
acht racing, in particular the Americas Cup, has always been
at the forefront of sailing research and development, with
the single goal of making boats go faster. The research, design,
and building of an International Americas Cup Class (IACC) yacht
is extremely expensive, so participants and suppliers are continually looking for inexpensive yet accurate ways to reduce R & D
costs and expedite results. In a collaboration between North Sails
Performance Resource Group and Dartmouth Colleges Thayer
School of Engineering, a Virtual Wind Tunnel (VWT) has been
developed to meet this need. The VWT enables North Sails engineers to do performance evaluations of full-scale sails and sail plans
on the computer and study the fluid-structure interaction between
the wind and sails. Initial emphasis has been put on downwind
sails that use relatively lightweight, stretchy materials compared
to modern, relatively rigid upwind sails.
The VWT is composed of three codes: MemBrain, North Sails
proprietary software for the structural analysis of sails, masts, and
rigging, GAMBIT, and FLUENT. These three components are linked
together in an iterative process, automated through the use of
GAMBIT templates. A VWT analysis begins with an assumed sail
shape and position (trim) and surface pressure distribution for a
given set of wind and boat velocities. MemBrain uses these initial conditions to compute deformations in the sail geometry by
balancing external aerodynamic pressure loads with internal stresses, which are governed by the characteristics of the sail material. Once a new sail shape has been determined, the new sail geometry
is transferred from MemBrain into an IGES file. GAMBIT automatically
reads the IGES file and generates a mesh. FLUENT is then started by template commands, and a journal file instructs the code
to read the mesh, set boundary conditions, and launch the calculation. The flow field and pressure distribution are computed
for the deformed geometry, and upon convergence, the new pressure distribution is exported into a file and used to update the
sail shape in MemBrain once again. An iterative process coupling
GAMBIT, FLUENT, and MemBrain ensues until the sail shape reaches static equilibrium, i.e. when the maximum displacement between
pressure updates is less than a preset value. GAMBIT templates
allow the entire process to be run with no intervention by the
user.
After static equilibrium has been reached, the sail forces and
moments are evaluated to see if re-trimming, or repositioning
the sails on the boat is needed in order to optimize the sails per-
Two Stars and Stripes IACC yachts under sail
Flying
Sails
on the
Computer
By H. J. Richter and K. C. Horrigan,
Thayer School of Engineering, Dartmouth College, Hanover, NH;
J.B. Braun, North Sails Performance Resource Group, Marblehead, MA;
and K. H. Kuehlert, Fluent Inc.
The initial and final sail shape and trim before and after the
optimization process
18
marine
formance, i.e. obtain the maximum driving
force or the maximum drive over heeling force
of the boat under the given wind conditions.
The VWT has many advantages over traditional wind-tunnel testing methods. First,
since VWT tests are performed on the computer, they are done at full scale, so the problems encountered when using scaling laws in
real wind tunnels are avoided. Second, the
computational flow domain around the sails
can be very large, and since no wind tunnel
walls are present, edge effects are non-exis-
tent. Finally, since the boundary layer above

the water is computed in the upstream computational domain, a more accurate description of the angle of attack at the site of the
vessel as a function of rig height is incorporated into the simulation.
In addition to optimizing sail performance,
this promising technology will be used in the
future for instrument calibration and predictions of bad air zones, where sailboats in the
wake of nearby sailboats experience greatly
diminished and changeable winds.
The fiber
layout in an
asymmetrical
spinnaker;
changes to the
fiber layout
cause changes
to the stretching
characteristics,
and thus the
sails response
to external
forces
Picture of an optimized sail showing pathlines

and the pressure distribution on the sails
First European
Americas Cup Winners!
T
he landlocked European nation of

Switzerland is not renowned for its seafaring traditions or its expertise in yachting. In Auckland Sound this past winter, however,
this stereotype was forever put to rest. The
Swiss-based Alinghi yachting team, skippered
by Russell Coutts, captured the Americas Cup
from two-time holders Team New Zealand in
a dramatic 5 0 series.
As was reported in the Fall 2002 issue of
Fluent News, researchers from the Ecole
Polytechnique Fdrale de Lausanne (EPFL)
the Official Scientific Advisor to the Alinghi
Challenge for the 2003 Americas Cup working in the group of Prof. Alfio Quarteroni, have
applied leading-edge numerical flow simulation techniques using FLUENT. These simu-
lations have provided valuable information to

the Alinghi Design Team, which, under the
direction of Grant Simmer, designed the winning high-performance Alinghi racing yacht.
The use of FLUENT along with in-house and
third party codes, has allowed the EPFL and
Alinghi to analyze hydrodynamic and aerodynamic flows, and even yacht/yacht interactions. CFD has allowed Alinghi to obtain a
competitive edge in an application area where
small performance improvements result in significant time gains.
According to Prof. Quarteroni, We are very
proud of Alinghis strong performance in the
Americas Cup, and we are pleased that our
CFD group was able to contribute to the
victory.
The Alinghi yacht during Race 1 in Auckland

Copyright Th. Martinez, Alinghi Team
19
Visions
Future:
of
the
Wastewater treatment plant of Nevers, France, certified ISO 14001

Courtesy of Gnrale des Eaux - Marcel Chevret
Environmental CFD
Keith Hanna from Fluent News recently interviewed Herve Buisson, Head of the Research Center
at Vivendi Water in Paris, France and Christelle de Traversay, CFD Program Manager at Vivendi
Water about the trends and challenges facing the environmental industrys use of CFD.
KH: Vivendi has become a world famous industrial conglomerate

over the last decade or so where does the Vivendi
Environnement (VE) group fit into the mix?
HB:
Vivendi, as a company, really started off in 1853 as Generale des Eaux,

a French municipal water company.
Over time, it extended its activities
to other environmental sectors, such
as waste management and transportation, and evolved to become the
worlds largest environmental company. In the 1990s, the company
diversified into print and broadcast
Herve Buisson
media, eventually buying out Universal
Studios in America, and in 1998, it changed its name to Vivendi.
In 2002, Vivendi Environnement separated from Vivendi Universal
to focus again on its original core competencies as the world
leader in environmental services.
KH: Can you describe the scope of your environmental company today?
HB: We turned over about $29 billion in 2001 with operations
in over 100 countries worldwide and some 295,000 employees. Vivendi Environnement is the only company in the world
that operates across the entire range of environmental services, with our four divisions covering water (Vivendi Water,
which accounts for nearly half of our total revenues), waste
management (Onyx), energy services (Dalkia), and public
transportation (Connex). US Filter is a leading supplier of
water industry equipment and services in the USA. Because
of our diversity we can develop integrated service packages
that offer a comprehensive, tailored response to the envi-
20
ronmental problems our customers face in both private and

public utilities. We have a strong customer focus, and are
proud of our long-term commitment to protecting the environment through ethical operating standards.
KH: You both work in Vivendi Environnements Anjou
Recherche R&D Center - what are the important technical issues for the water industry from your perspective?
HB: The industry today is driven by stronger and stronger regulations, and consumer demands, and our research is geared
as much to new or potential legislation as it is to process
and equipment improvements. Water is, for the most part,
a high volume, low end-value product, typically costing less
than $1/ton with typical city plants treating over 100,000
t/yr. In the past, our new product development cycles were
around 10 to 15 years in duration, but with frequent regulatory changes and a growing demand for cost efficiency
and more compact plants, Research & Development has come
very much to the forefront. Today, new process and equipment development cycles are typically three to five years in
duration. Through optimization of existing products and processes, our research group aims to supply Vivendi plant operators with solutions that are cost-effective (from a Life Cycle
Costs perspective) and tailored to specific water applications.
KH: Where does CFD fit into your rapidly shortening product development cycles?
CdT: CFD is a key component of all of our R&D work, in operations, engineering applications, and equipment design. Over
the years it has been validated for modeling many processes and pieces of equipment. In some instances, we have seen
cost savings of up to 30% over conventional pilot study approach-

es. CFD has given us a fundamental understanding of many
of our unit processes that have been viewed as black boxes
in the past. The flow visualization capabilities have been invaluable for demonstrating process behavior to our senior management and clients. Today, we see a greater need to optimize
our processes, which means that CFD usage has increased.
Our use of other software, such as AUTOCAD, FLOWMASTER,
MATLAB, and InfoWorks is also on the rise.
KH: What sort of CFD work do you do at Anjou Recherche?
CdT:
In addition to modeling unit processes, we tend to do standard fluid flow

simulations in ducts and pipework.
Increasingly, we are doing multiphase
simulations with two or three phases, because such systems are very common in the water industry. Once we
test and validate CFD for a given flow
application, we pass our know-how
on to plant engineers and to our partChristelle de Traversay
ners at Vivendi Water Systems and US
Filter. We have an internal CFD club where we share our experiences with CFD engineers throughout the company. CFD
offers us significant advantages over experimental methods
during pilot plant scale-up tests, where we want to investigate full-scale effects before a plant is constructed. Our newer
water treatment processes tend to be more compact than
older ones, because they are cheaper to build and operate.
Smaller equipment means reduced hydraulic residence times,
however, so we are faced with lower over-engineering margins for our equipment designs. To accurately predict behavior, CFD is critical to our design processes.
KH: Why did you choose FLUENT as your CFD code and what
benefits does it provide?
CdT: Fluent is the clear leader in CFD and has a strong presence
here in France. We like FLUENTs ease of use, range of turbulence models and customization capabilities. We were one
of the first users in France when we licensed FLUENT in 1991
and we have stayed with Fluent France ever since. Early on
when we decided that we needed CFD, we also decided not
to develop our own CFD codes in-house because it is just
not cost-effective. Instead, we wanted CFD software that allowed
us to attach our know-how to the code and FLUENT provides that.
Sludge volume fraction in a secondary

settling tank during dry weather conditions
(top) and after a storm event (bottom)
KH: How do you foresee CFD being used in the water & environmental industry in the long term?
HB: At Vivendi, we aim to give the best local access to our global development, and therefore want to deploy hydraulic modeling tools in an easy-to-use template format locally at every
major water plant, so that plant operators will be able to
get dynamic fluid flow information that allows them to make
informed decisions in real time. Ideally, we would want very
few buttons on the interface and background software that
is foolproof. The CFD code may be transparent and do multiple CFD simulations on a neural network for instance, or
perhaps pick out answers from a database of pre-calculated simulations to display to the operator as needed. In essence,
we are envisioning an advanced process control flow modeling tool that has CFD (and maybe other software tools)
integrated into it as necessary. In addition, we want faster,
better, and cheaper software!
We want to be able to model biological and chemical phenomena in conjunction with our CFD calculations. We work
with a range of world-class universities and colleges around
the world and would like to integrate the latest biological
and chemical research findings into our software modeling
tools. We hope to put more and more of these models into
FLUENT through user-defined functions (UDFs) or through
couplings with other software programs.
On the waste management and energy side of our company
we see a lot of potential for applying CFD, especially in incinerator modeling and solid waste management processes. Our
R&D colleagues dealing with these issues are using the same
CFD tools, to optimize intercompany transfers of know-how
and synergies. We anticipate doing more two- and threephase simulations with CFD, and anticipate such advanced
uses as multivariable analyses through Monte Carlo simulations using FLUENT output. While not underestimating the
inherent complexity of our raw material, water, with its complex chemical, biochemical, seasonal, and geographical variations, we strive to optimize our services, standardize our
equipment, and customize our engineering plants using software tools like FLUENT as powerful and smart templates
to embed our know-how.
Research Center hall at Maisons-Laffitte, France (Anjou

Recherche): bubble columns for ozone transfer
21
automotive
Grid Morphing
By Rajneesh Singh, General Motors Corporation, Detroit, MI
Design variables used in the parametric study
he aerodynamic design of an automotive vehicle

is an iterative process. It involves an interaction between
the designer, who proposes a shape for the vehicle, and the aerodynamics engineer, who evaluates the
shape for aerodynamic performance and provides enablers,
or guidelines for drag minimization to the designer. The
designer then incorporates these recommendations within the constraints of the design theme. After a number
of iterations, the process gives rise to a vehicle with an
improved drag profile. This process can take place using
clay models in a wind tunnel or using CFD. While quicker and less costly, the CFD approach still takes time because
of the need to create and mesh modified geometries
several times. At General Motors, a new automated process
has been developed using FLUENT and other software
tools that together help reduce the time required for
each geometric modification and therefore, each CFD
solution.
create CFD
model for the
baseline
geometry
start
end
yes
DOE
analysis
completed?
define design
variables and
setup model for
morphing
no
select set of
design variable
magnitudes
(iSIGHT)
save
design variables
and Cd for post
processing
perform
CFD analysis
and compute
aerodynamic
drag (Cd)
(FLUENT)
Flow chart of the automatic analysis process
22
generate CFD
mesh for
prescribed set
(Morpher)
The automated process for aerodynamic optimization makes use of FLUENT and a mesh morpher (Meshworks/Morpher, from Detroit Engineering
Products (DEP)). These codes are coupled via iSIGHT
(from Engineous Software Inc.) for an automatic
exchange of information and data. iSIGHT guides
the process by selecting the design variable magnitude, executing Meshworks/ Morpher to modify the mesh, and providing this mesh to FLUENT
for the next round of CFD computations. The process
can be set up to conduct traditional design optimization or Design of Experiment (DOE) studies.
It has a very quick turn-around time for multiple
calculations, and contributes to the vehicle design
in two ways. First, it can be used to develop the
enablers for the most efficient design. Second, the
process can be used to complement wind tunnel
testing. The design space can be explored to identify the aerodynamically critical regions of the vehicle, and the wind-tunnel test engineer can use this
information to reduce the number of wind-tunnel
tests.
A remote, high performance, parallel computing machine is used to run the coupled calculations
in a batch process mode. When FLUENT performs
the aerodynamic simulation, it executes a sequence
of commands listed in a journal file. The computation for each new design is started from the converged solution for the previous design, to reduce
the computational effort required. Since the CFD
computations are performed using a journal file, it
is also possible to produce a set of images to visualize the flow for each design. When combined with
averaged or integrated quantities (such as drag coefficient), these images help illustrate the flow characteristics of each case studied.
The automated analysis process for performing
a DOE study was recently illustrated using a generic automobile shape in a rectangular tunnel. A base
case and sixteen modified designs were created and
analyzed in the study. The mesh for the baseline
model was created with 1.3 million elements, with
prism layers on all of the car surfaces for an accurate resolution of the boundary layer. The designs
differed in the length and height of the rear deck
and roof edge. It took less than 2 days of time to
evaluate the 16 designs. Had more computational resources been available, more designs could have
been tested and an exhaustive DOE study conducted
to find the true optimal design. A conventional analysis procedure, in which the vehicle geometry is modified and a new mesh is constructed for each design,
would require at least an order of magnitude more
time to complete.
Pressure contours on the rear part of the

vehicle for various designs; the aerodynamic
drag is shown for each image
automotive
The
Sounds
of the
Road
By Sandeep Sovani, Fluent Inc. and Bipin Lokhande, Fluent India
The mesh in the vicinity of the noise-producing object has a strong impact on the
quality of sound that can be simulated using CFD
rom a fluid dynamics point of view, an automobiles side view mirror

(SVM) is a bluff body exposed to a high speed flow. The flow structure in the wake of an SVM is highly transient and subjects the vehicle surfaces in its vicinity, such as doors and windows, to significant unsteady
pressure fluctuations. This unsteady pressure variation ultimately propagates
inside and around the vehicle as noise.
Sound generated in and propagated through a fluid domain can be simulated using two methods:
Computational aeroacoustics (CAA), defined as a direct
simulation of acoustic fields generated by flow, and the
interaction of acoustic fields with flow. Direct implies that
computation is only based on fundamental physical principles
without reliance on empirical results.
Aeroacoustics models for propagation of sound from the source to
the receiver.
FLUENT can conduct aeroacoustic simulations using both of these approaches. CAA is handled by FLUENT through its well-established and extensively tested transient flow capability. In addition, two aeroacoustic models have
been implemented and tested. FLUENT 6.0s Lighthill-Curle acoustic module is capable of propagating sound generated by pressure fluctuations on
wall boundaries to far-field observation points. FLUENT 6.1 has a built-in
80
Experiment1
CFD Direct
CFD FWH
SPL (dB)
60
40
20
200
600
400
frequency (Hz)
800
The CAA and

aeroacoustics
model are both in
good agreement
with experiment
for a receiving
point not far
from the mirror2
acoustic module based on the Ffowcs-Williams-Hawkins theory that can calculate sound radiated by boundary and interior surfaces towards observation points inside or outside the computational domain. In addition, FLUENT
6.1 results can now be imported to SYSNOISE, an acoustics modeling tool
from LMS International.
For the SVM, the sound generated by the turbulent flow field in the wake
of the mirror has been simulated using CAA and the Ffowcs-Williams-Hawkins
formulation in FLUENT. The generic mirror shape consists of a half cylinder topped with a quarter sphere of the same diameter.
The CAA approach is executed by conducting a transient simulation of
the flow around the mirror with the LES turbulence model. Monitor points
are put at locations where microphones were placed in experiments reported in the literature1 and the transient static pressure signal is recorded at
these points. The Fast Fourier Transform (FFT) tool newly introduced in
FLUENT 6.1 is used to convert the transient pressure signals into frequency spectra.
The analysis based on the Ffowcs-Williams-Hawkins model starts with a
transient simulation of the flow field around the mirror. At the beginning
of the simulation, source surfaces for the sound and receiver (microphone)
locations are input. For the SVM, the mirror body and flat plate on which
it is mounted are selected as source surfaces. During the calculation,
FLUENT creates plots or files of sound pressure vs. time.
Sound pressure spectra show that both methods are in good agreement,
qualitatively and quantitatively, with experiment. The accuracy of aeroacoustic simulations is heavily dependent on that of the underlying transient
flow simulation. Time-step, grid resolution, and grid quality not only determine the accuracy of the predicted sound pressure level, but also the frequency band over which the simulation results are meaningful.
reference
1
Siegert R., Schwarz V., and Reichenberger J., AIAA

Paper no. 99-1895, 5th AIAA/CEAS Aeroacoustics
Conference, Seattle WA, May 10-12, 1999.
Lokhande B.S., Sovani S.D., and Xu J.,

Computational Aeroacoustic Analysis of a Generic
Side View Mirror, Paper no. 2003-01-1698, SAE
Noise and Vibration Conference, Traverse City, MI,
May 6-8, 2003.
Contours of velocity illustrate

the transient nature of the
flow around the mirror
1000
23
customized applications
Instrumented cooking tray, showing

thermocouples at the surface of the
peas and water left in the tray after
the cooking was completed
Customizing
Food Steamers
By Michael Engelman, Enductive Solutions, a sister company of Fluent
120
100
temperature (C)
80
60
40
20
0
-20
00:00 02:53 05:46 08:38 11:31 14:24 17:17 20:10 23:02 25:55
time (mm:ss)
Experimental profiles of temperature vs. time as measured by
the thermocouples positioned in the tray
Steam pathlines, colored by velocity magnitude, illustrate

the improved flow field in the steamer
more.info@
www.enductive.com
24
espite the fact that computer aided engineering (CAE) use is widening across
dozens of industries, the investment in
personnel, hardware, and/or software is beyond
the reach of many companies. Even if a focused,
turn-key product were available, the know-how
to develop such a product often is not. At
Enductive Solutions, CAE solutions are created for a specific application, through the use
of one or more software products that are integrated into a single, easy-to-use package.
In one recent project, a manufacturer of commercial food steamers wanted to increase the
capacity and efficacy of a steamer for a customers special requirements. The first step was
to investigate the functionality of the existing
device. It was determined that CFD would be
needed, so Enductive engineers worked with
the manufacturer on physical testing to
develop boundary conditions and a set of data
that would later be used for validation studies. The steamer was instrumented with
appropriate velocity probes and thermocouples and tested under four different operating
conditions. The tests showed that uneven heat
distribution within the steamer would make
it impossible to improve its performance without significant design changes.
Enductive engineers then obtained a
copy of the computer aided design (CAD) file
for the steamer and used it to create the geometry and mesh for a CFD simulation. Boundary
conditions and physical properties were
derived from the test results and information
provided by the manufacturer. The results of
the CFD simulation helped to explain the uneven
temperature distributions shown in the physical tests. In particular, they showed that the
location, speed, and direction of the steam
jets were far more important than the gas flow
rate entering the unit or distribution of food
inside the unit in governing the temperature

distribution in the steamer.
The results suggested that the operation of
the steamer could be greatly improved by any
change that results in a more uniform flow distribution, such as relocation of the jets, or introduction of manifolds, steam exhaust valves, or
recirculating fans. They also made it clear that
future CFD simulations of the steamer could
be done without heat transfer calculations, since
the flow pattern alone determines the temperature
distribution throughout the device.
The next step was to encapsulate the CFD
analysis within an interface that prompts the
user to enter only a few critical design parameters, such as the vessel size, tray locations, and
steamer type. The tool then performs the appropriate CFD calculation and returns the results
necessary to evaluate the performance of the
design.
Application-specific solutions such as this
can be done on-site or can be deployed to a
CAE specific Applications Service Provider (ASP)
such as the Fluent Remote Simulation Facility.
The customer can enter the parameters
needed to perform the analysis in a matter of
minutes and does not need to spend the time
that would otherwise be required to become
an analysis expert. The use of an ASP to perform the analysis eliminates the need to purchase and maintain expensive software and
hardware as well. The analysis returns the precise information that the user needs to determine the performance of the proposed design
and, just as importantly, helps provide an understanding of why the design does or does not
work so that it can quickly be improved. The
service puts advanced CAE tools in the hands
of the people who need them most front line
engineers to reduce engineering expenses
and bring products to market faster.
For The HVAC Industry
Newsletter Supplement
S2
ventilation
Swiss Re Headquarters
Inside & Out
Looking Out for Crew Comfort
in Space
California Living
Air Flow Befitting the US Marines
S6
industrial hygiene
S7
smoke management
Breathing Easier in the Workplace
Containing Smoke in
Complex Atria
Fire Scenarios in the Budapest
Sports Arena
HVAC industry
FOCUS on CFD
HVAC industry
ventilation
Swiss Re
Headquarters
Inside & Out

By Matthew Kitson, Hilson Moran Partnership Ltd., London, UK
For a given set of wind conditions

(illustrated by pathlines) FLUENT predicts
surface pressures (illustrated by contours)
on the building exterior and external
aerodynamic air flow characteristics
ondon is home to a striking new building 30 St. Mary Axe the

UK headquarters of Swiss Re, the international reinsurance company.
Designed by Foster and Partners, the building reflects the companys commitment to environmental sustainability. As part of the construction project, Hilson Moran Partnership Ltd., a consulting firm based
in England, was contracted to supply mechanical and electrical engineering
design services for the building.
The structure has a circular footprint, and each of the 33 circular stories of office space varies in depth from faade to core by 6m to 15m.
The ground floor comprises an entrance lobby with three banks of elevators and a separate retail space. The top of the building has private
dining rooms, a restaurant, and a multifunction area. Each floor plate
has 6 triangular shaped light wells (atria) placed symmetrically around
the perimeter of the faade, which rotate by 5 degrees every floor, giving the building a helical look when viewed from the outside.
Hilson Moran started on the design of the mechanical and electrical
engineering services for the building in 1997. Planning consent was obtained
in 2000 and construction on site started in 2001. The project is due for
completion and occupation by Swiss Re in the first quarter of 2004.
CFD was used in conjunction with Dynamic Thermal Modelling (DTM)
initially to calculate the thermal performance of the building. The results
were used as a basis to undertake further studies and a more detailed
analysis of other aspects of the design, including the ventilated faade,
the light wells, natural ventilation of the offices, entrance hall, and top
of the building dome. CFD has also been used to measure the external
conditions prevailing at the top of the building and interpolate these results
over the rest of the building to assess natural ventilation potential.
Based on a number of early CFD results, a team decision was made to
adopt a mixed mode ventilation design, which is not common for a high
rise building of this type. Natural ventilation may be used for up to 40%
of the year, and when conditions become unacceptable, the building will
be sealed to the outside and go into either a cooling or heating mode of
operation. The goal with this approach is to maximize the period of natural ventilation, and thereby minimize energy use and carbon emissions.
The natural ventilation is provided by perimeter window openings in
each light well, which serve as the lungs of the buildings, breathing when
S2
required. Fresh outside air reaches the office occupants without strong
breezes through balcony openings to the light wells. An active faade
with solar control blinds was adopted for the office areas and high performance solar glass in the light well areas. CFD was used to examine
design issues such as optimum ventilation rates, position of the blinds,
and entry and exit points for the ventilation air.
CFD was also used to examine the air flow on the office floors, and
ventilation schemes for the lobby, taking into account the high ceilings,
the heat generated by various lighting schemes, and drafts from the large
revolving doors. The top of the building, with its large glass dome surrounding a restaurant and multifunction mezzanine level, produced a number of design challenges as well. Hilson Moran was involved in several
key strategy design decisions for this region, ranging from the type of
glass used to the ventilation strategy. In the final design, the ventilation
is provided by a displacement system coupled with a chilled floor. CFD
analysis was used to examine the thermal performance of the dome in
both peak summer and winter periods, and the comfort of the diners
and visitors in these areas was evaluated.
Air speeds at head level, with windows

open on the windward side ventilation
t Boeing, an investigation has been

performed to predict the ventilation
and temperature characteristics of
the International Space Station Cupola, and
to ensure the adequacy of crew comfort
and safety during certain usage scenarios.
The Cupola is the Space Stations pressurized observation and work area. It is used
to support the Remote Manipulator System
(RMS), which is the stations robotic arm.
It also houses a Robotic Workstation
(RWS), which is used to control the RMS.
The Cupola permits the astronaut to view
the Earth, celestial objects, and visiting vehicles as well.
Fluents CFD software has been used to
evaluate the Cupolas ventilation with
and without the RWS operating. The entire
geometry and mesh were created using
GAMBIT. There is a requirement that an
effective air velocity in the Cupola habitable volume must be maintained within the
range of 15 to 40 feet per minute. The first
round of FLUENT results indicated that the
Cupola ventilation meets this specification
when the RWS is not powered. However,
when the RWS is operating, the air velocity is too high due to the elevated flow rate
created by the RWS fans. Furthermore, a
low velocity region is present where the
crewmember is usually positioned, at the
center of a vortex created by the flow
pattern.
In addition to examining Cupola ventilation, crew comfort has been evaluated by
investigating the air temperature around the
crewmember. When the RWS is not operating, it was found that, due to good air
mixing and low heat loads, the temperature around the crewmember stays at a comfortable level. When the RWS is activated,
however, there is reduced air mixing and
higher heat loads that cause uncomfortable
temperatures to develop.
After analyzing these results, it was found
that a laterally located vent on the RWS is
the culprit for the vortex air pattern during RWS operation. FLUENT was used to
find a resolution to this problem. A deflector was placed over the vent to hopefully
bring the ventilation closer to specification,
improve the air mixture, and lower the temperature around the crewmember. The results
of these simulations show that the deflector works by diverting the air from the laterally located vent directly out of the Cupola
hatch, thereby hindering the formation of
the vortex. This, in turn, results in lower
air velocities, an improved air mixture,
and more comfortable temperatures for the
astronaut.
HVAC industry
ventilation
Pathlines, colored by
velocity magnitude,
illustrate the flow
field in the cupola
when the RWS is not
operating
Looking Out for
Crew Comfort
in Space
By Jorge L. Zapata and Chang H. Son, The Boeing Company, Houston, TX
When the RWS is turned on, a vortex forms in

the cupola, causing discomfort for the
crewmember
By adding a deflector near a laterally mounted vent,

pathlines show that the flow is now acceptable
S3
California
Living
By Viralkumar Gandhi and Kishor Khankari, Fluent Inc.;
Kevin Blackwell and Rob Hammon, ConSol, Stockton, CA
onSol and the California Energy

Commission (CEC) are working on a
joint project to improve the energy efficiency and thermal comfort of residential homes.
These qualities can enhance the profitability of
homebuilders through reduced warranty and
callbacks. ConSol approached Fluent Inc. to develop a CFD model that can be used as a predictive tool to test the effect of several design and
operating parameters of the HVAC system of a
typical single-story, three bedroom house in
California. The main goal of this ongoing project is to test the effect of ventilation airflow rates,
and the location, size, and type of ventilation
registers on the distribution of air and temperature
within the home during the winter and summer months.
Fluent used Airpak software to develop a 3D
CFD model to simulate the winter situation in
which the home was heated by hot air supplied
through a number of ventilation registers placed
in the ceiling. The velocity and throw of each
register was obtained from the manufacturer of
the registers, and was modeled through the builtin functions for ventilation registers in Airpak.
Convective and radiative heat losses from the walls,
windows, and partitions of the home were modeled by considering the effective U values (representative of thermal conductance) for each
component. The feedback control from a thermostat was also taken into consideration in the
transient model by monitoring the air temperature at the thermostat location in the home, and
accordingly setting the ventilation fan to ON or
OFF positions to maintain the temperature within a 68F to 72F range.
The computational model accurately predicted
the distribution of airflow and temperature during the winter season. The thermostat cycle predicts that the fan turns ON about every 14 minutes
and remains ON for about 4 minutes. These predicted durations of the fan cycles helped quantify the heating duty on the home. The analysis
also showed that in certain rooms, the high airflow rates and throws of the registers, which
are specifically designed for the summer cooling conditions in California, cause localized hot
zones near the ceiling. The prolonged OFF cycle
of the fan helps reduce the thermal stratification and give rise to more uniform air temperatures,
which translates into improved thermal comfort for the occupants.
Distribution of air temperature at the end of the heating cycle

showing thermal stratification in the house
S4
Geometry of the three bedroom house studied
ConSol and CEC are planning to share these

computational results with builders and prepare
new design guidelines for residential ventilation
systems. These new guidelines, based on the
computational analysis, will help builders
reduce risks and costs by improving the quality, comfort, and energy efficiency of residential construction. ConSol and CEC find this
predictive tool a more cost-effective and efficient means of achieving this goal than
through physical testing. Indeed, measuring velocity and temperature at several thousand locations in a house, comparable to the output of
a CFD analysis, would be not only expensive
and labor intensive, but virtually impossible to
do!
74
temp (F)
HVAC industry
ventilation
FAN ON #1
FAN OFF #1
FAN ON #2
FAN OFF #2
72
70
68
66
12
16
time (m)
20
24
Thermostat cycles showing the cyclic duration of

heating and cooling cycles
Pathlines colored by air temperature showing the airflow entering from

the ventilation registers
28
he National Museum for the Marine Corps will

soon be under construction in Quantico, Virginia.
A project 10 years in the making, the Museum
is dedicated to sharing the heritage of the Marines with
the American public. Visitors will enter the museum
through an entry courtyard and lobby into a dramatic,
glass-enclosed atrium called the Central Gallery.
Symbolizing the Marines World War II victory at Iwo
Jima, the Central Gallery features a steel mast rising
180 feet from the floor to the top of a conical glass
skylight. The base of the gallery is underground and
measures 150 feet in diameter and 45 feet high. The
glass skylight structure rises an additional 110 feet above
the gallery walls. Several full-size airplanes will be suspended from the skylight structure, while kiosks and
displays will be located throughout the Central Gallery
and adjacent exhibit halls.
Given the large skylight glazing area, concerns arose
about the influence of solar heat gain in the summer
and condensation on the glass during the winter. Fentress
Bradburn Architects, the museum designers, asked
Architectural Energy Corporation, the projects energy, daylighting and sustainable design consultants, to
explore these issues using Airpak. Architectural Energy
Corporation has been performing sophisticated energy and daylighting modeling for many years, and has
added CFD to their portfolio of modeling tools. Another
purpose of the CFD analysis was to overcome the limitations of traditional HVAC design tools, which are constrained by a fully mixed (i.e. uniform temperature)
zone model, and tend to overpredict the anticipated
cooling load. In this tall glass structure, it was obvious that thermal stratification would occur during the
summer, thus requiring only the lower occupied zone
to be conditioned.
Using the CAD import tool in Airpak, a simplified
model was created with 700,000 cells. Conditioned
air was supplied by nozzle type diffusers located around
the perimeter of the Central Gallery drum at a height
of 18 feet. The nozzles were modeled using an effective area calculated by the Airpak diffuser macro to
achieve the correct throw specified in the manufacturers performance data. Air was returned or exhausted in three locations: a fan in the top of the skylight,
for venting to the outside, and return grilles in the entrance
ceiling and along the floor, for returning the remaining air to the air handling unit.
Several indoor and outdoor design conditions were
simulated to determine the extent of thermal stratification that would occur, and what supply air flow rate
was necessary to maintain comfort in the occupied
zone. A 90,000 CFM supply scenario was suggested
by a mixed-zone model to satisfy the cooling load. Airpak
showed that this scenario significantly overcooled the
occupied zone during summer design conditions.
Architects rendering of the building exterior
Air Flow
Befitting the
US Marines
By Galen Burrell and Michael J. Holtz, Architectural Energy Corporation, Boulder, CO
Exhaust fan
The geometry of the museum shows
the cylindrical underground base and
conical glass dome
Second floor balcony
Lobby return air
Supply diffusers
Floor-level return air grilles
continued on next page
S5
HVAC industry
ventilation
HVAC industry
industrial hygiene
Breathing Easier
in the
By Clyde J. Porter, Wyman Gordon Company, N. Grafton, MA
A worker in the process grinding area
The workstation layout before the analysis

was performed, showing ventilation air
velocity contours
In the optimized workstation layout

ventilation air velocity contours show
increased velocity at the point of grinding;
the absence of recirculation zones in this
design minimizes the amount of dust
entering the breathing zone of the worker,
who stands on the side of the bench
Temperature contours on a slice

through the gallery
S6
Workplace
C
ontrolling worker exposure to dusts and fumes

is a major concern at Wyman Gordon, a division of Precision Castparts Corporation, and
a leading manufacturer of metal forgings for the aerospace and industrial gas turbine industries. In the Process
Grinding Department at one plant, for example, sidedraft exhaust benches are used to limit exposure to
metal grinding dust. Airpak was recently used to compare the capture efficiency of these benches with other
types of exhaust systems in order to determine the
best approach to minimize the amount of dust in the
workers breathing zone.
Four different hood configurations were modeled
during the project, along with various combinations
of exhaust and supply airflow rates. These included
the existing side-draft hood, and modifications to it,
as well as three booth-type hoods, one with an open
roof, the others with a partial and a full roof.
The final selected design was a modification of the
existing hood that demonstrated the best combination of performance improvement and feasibility. Making
use of existing fans and filters, the new design will cut
worker dust exposure by up to 50%. Because the existing equipment could be utilized, construction cost savings of $250,000 for 10 workstations could be realized.
Without Airpak, it would have cost at least $25,000
and up to a year of additional work to develop and
test prototype hoods. As an added benefit, plant management has found Airpak model output very helpful in understanding why various project options were
selected.
Reducing the supply air flow to 60,000 CFM achieved

near ideal comfort conditions in the occupied zone
(75F), while allowing the skylight to stratify between
80F and 115F. A 40,000 CFM scenario was also analyzed, and showed overheating occurring in the occupied zone during peak summer design conditions.
Using these results, Architectural Energy Corporation
recommended lowering the cooling supply air flow
from 90,000 CFM to 60,000 CFM, allowing the design
team to downsize their HVAC equipment (and budget). All three scenarios required that at least 15,000
CFM be exhausted out of the top of the skylight to
keep temperature extremes below 120F.
Nighttime banquet conditions during winter were
also modeled to determine the potential of condensation forming on the glazing and framing system of
the skylight. Using Airpak and THERM, a glazing system heat transfer program developed by Lawrence
Berkeley National Laboratory, it was found that with
500 people eating hot meals on a cold winter night,
condensation would not occur. Taken together, the
results allowed the architects and clients to feel confident in the proposed Central Gallery HVAC and
skylight design.
trium spaces are a popular means of creating a

sense of openness and comfort for building occupants. The architecture of these spaces is
becoming increasingly complicated as designers work
to balance energy efficiency, aesthetics, and visual impact.
One of the challenges in designing such spaces, however, is engineering a smoke management system that
can maintain tenable conditions in the space so that there
is sufficient time for the occupants to escape in the event
of a fire. The difficulties are a result of the interactions
between the smoke, the architecture, and the airflows.
These interactions lead to disturbances in the rising smoke
plume, which in turn cause excess mixing of the smoke
with clean air, resulting in a larger volume of smoke to
be exhausted. For example, overlapping levels or bridges
across open spaces can lead to multiple balcony spill plumes,
and architectural features can narrow the available flow
area and cause local flow accelerations.
At RWDI, an internationally recognized engineering
firm, FLUENT has been used to better understand the
workings of complex atria in the presence of a fire. In
one recent study, an atrium was studied that consisted
of multiple levels and connected spaces. The space was
outfitted with a smoke management system developed
by following the local code, and the CFD results showed
that with this system in place, smoke would penetrate
into many of the occupied areas of the building.
In RWDIs experience, providing a code specified quantity of exhaust at the top of the atrium is not always sufficient for a safe atrium design. Other design strategies
are necessary to help keep smoke out of the occupied
zones, and RWDI uses a number of these to improve
smoke management systems. For example, an atrium
can be segregated into smaller and simpler atria when
a fire erupts. Segregation in this particular atrium led
to measurable reductions in undesired smoke propagation,
used less than half the exhaust air, and saved the owner
both capital and operating costs.
Proper use of CFD models for the design of smoke
management systems also requires an understanding
of smoke plume dynamics (including ceiling jets and thermal stratification), sprinklers, tenability (including visibility, toxicity, and thermal exposure) and external wind
effects. To ensure that the correct quantities of smoke
are being produced, it is important to use properly calibrated CFD methodologies. Otherwise, underprediction
of smoke transport may lead to an inaccurate assessment of the required exhaust rate. In addition, the smoke
generated by a fire in an enclosed space should not be
simulated using the same methods that are suitable for
a large open space. The proper use of CFD modeling
to assess smoke transport has allowed RWDI to demonstrate successful ventilation designs, leading to safer, cost
effective smoke management systems.
HVAC industry
smoke management
Smoke, represented by a gray iso-surface, penetrates into

many occupied areas of a building with a complex atrium
Containing
Smoke
in
Complex Atria
By Duncan Phillips, PhD, PEng, Senior Specialist/Associate and

Ray Sinclair, PhD, Principal, Rowan Williams Davies & Irwin Inc. (RWDI), Guelph, Canada
The same fire scenario as above, but with a smoke

management system that segregates the atrium
S7
HVAC industry
smoke management
Fire Scenarios
in the
Budapest Sports Arena

By Dr. Gergely Kristf, Mt Lohsz, Tams Rgert, and Pter Bodor, Budapest University of Technology and Economics, Budapest, Hungary
ecember 15, 1999 will forever remain in the memory of many Hungarians.
On that date, the Budapest Sports Hall, the largest indoor arena in
Hungary, burned down. The arena symbolized a source of great pride
in Hungary. The Hall was where crowds could experience sporting successes and failures, World and European Cups, and many other events. The
whole country was touched by the disaster. Donations were offered by artists,
entrepreneurs, public figures, and thousands of private individuals.
Construction has just been completed on a new Budapest Sports Arena.
Bouygues-Hungria Construction (builders of Parc des Princes and the Olympic
Facilities in Sydney) was selected by the Hungarian Government as the builder
for the project. The futuristic building will boast a seating capacity of 7,000
to 12,500 and an ordinary field area of 8,400m2. Its modular design is expected to attract events such as concerts and high-profile sport competitions,
including indoor motocross, horse shows, and even windsurfing.
A strong emphasis was put on safety during the design phase. In case
of fire, air vents will be closed, automatic doors will open, and the smoke
exhaust system will start to operate. During the early stages of the fire, the
developing combustion products will rise upwards and accumulate under
the ceiling of the hall, forming a continuous layer. Depending on the quantity of smoke generated and the flow rate of the smoke exhaust fans, the
lower edge of the smoke layer might gradually drop down to the top rows
of the grandstands. For this reason, the shape of the hall and the pace of
the fire will determine the time for a safe escape, in which everyone can
exit the hall without smoke poisoning or reduced visibility. An additional
consideration is the distribution of temperature in the cloud of smoke, which
can reduce the integrity of the building and cause radiation discomfort or
injury for the spectators.
To better understand these scenarios, Bouygues-Hungria charged the
Department of Fluid Mechanics of Budapest University of Technology and
Economics to perform numerical simulations and wind tunnel experiments,
with the goal of the former to study smoke propagation. Using FLUENT,
a simplified fire model was created. The fire was initiated in the vicinity of
the stage, and modeled as a simultaneous source of heat and carbon dioxide. Practical experience suggests that after ignition, the boundary of the
burning area spreads at a steady velocity, and the intensity of combustion
falling on a unit area is constant in space as well as in time. These phenomena were modeled with quadratically increasing sources of heat and
CO2. Since all possible fire scenarios could not be analyzed, smoke dispersion
was simulated for fire cases at three different locations, by using the same
power vs. time function. Thus while the simulation results provided valuable data for designers, they cannot be regarded as complete as far as all
possible fire scenarios are concerned.
The FLUENT results indicate that ten minutes after the fire ignites, the
smoke remains above the highest grandstand seats, so the people will have
ample time to vacate the building safely. The temperature in the smoke
cloud is about 100C, suggesting that the structure will not be in danger
at this time and the people will not experience too many smoke-related
side-effects.
The opening of the Budapest Arena was celebrated with a large
concert evening in March.
The new Budapest Sports Arena
Engineers perform a fire test inside the arena
Contours of temperature on a surface of 1% CO2 concentration, 10 minutes

after ignition, shows that the smoke has not yet reached the upper seats of
the grandstands, and that the temperature everywhere on this surface is
below 100C
S8
materials
Keeping
Printer
Touch Temperatures Low

By Francisco Zirilli, PhD, Xerox Corporation, Rochester, NY
ngineers at Xerox Corporation have used FLUENT to reduce the

temperatures in a new color printer in places that could be touched
by users when clearing a paper jam. The new printer, the DocuColor
iGen3, provides unparalleled speed, productivity, image quality, and
paper-handling capabilities. Many of its replaceable units are designed
for reuse or recycling. One of its internal components, the fuser roll,
is used to fix the toner image to the paper, and the roll must be quite
hot to do so. This creates an engineering challenge, since free convection can cause the temperature of customer accessible surfaces in
the printer to rise to a level that could cause discomfort.
Fusing, the process in which the transferred toner image is fixed
to the paper, is the last step in the xerographic process. Fusing the
toner to the paper is usually done by passing the paper through a set
of rolls that are forced to make contact with each other. The fuser roll
is in direct contact with the toner. It consists of a hollow aluminum
core coated with a rubber compound. This roll is heated internally using
a radiant lamp. The opposing roll is called the pressure roll, and is made
from solid steel. Successful fixing of the toner depends on maintaining the right fuser roll surface temperature and toner thermal properties. In the event of a paper jam, the customer may need to access
certain areas in the fuser region of the printer. By design, any surface
that the customer needs to come in contact with during jam clearance may not exceed specified touch temperatures.
Xerox engineers began the analysis process by importing IGES files
of the printer into GAMBIT, where the geometry was simplified and
an unstructured mesh of approximately 1.74 million cells was built.
Their model considered both conduction and free convection heat transfer. Temperatures were specified for the surface of the roll in order to
avoid the additional computational time needed to model the conduction inside the roll. The model was solved in two different ways.
A laminar flow solution was performed that was based on the Boussinesq
buoyancy approximation. A turbulent solution was also performed that
included the effects of buoyancy. When the simulations were compared
with physical experiments, both methods provided accurate results
to within 10% of experimental measurements, so the more costeffective laminar approach was used for subsequent analyses.
The results gave engineers all the information they needed to either
redesign the customer accessible components or shield them from the
high temperature sources. Natural convection was found to be the primary driver for carrying the heat from the fuser roll to the ends of the
printer compartment. Portions of the pins and brackets that are touched
by customers when clearing a paper jam reached unacceptably high
temperatures in an early design, so the engineers repositioned these
parts to move them out of the air stream. After simulating several design
iterations, they found a configuration that maintained the temperatures at acceptable levels. Because the thermal conductivity of a material has a significant impact on the perceived touch temperature, they
also used the results of the analysis to specify materials for user accessible components.
Temperature distribution of the fuser system, viewed from the

paper exit, with some components removed for clarity
Temperature distribution of the fuser system, viewed from the

paper entrance, with some components removed for clarity
25
glass & fibers
Glass
Tank Design
Shattering
Methodologies
By W.S. Kuhn, Stein Heurtey, Ris-Orangis, France
Simulation of a 120t/day electric cold top glass

tank designed and constructed by Stein Heurtey
(with courtesy of Akzo-PQ Silica); the horizontally
fed batch (left to right), illustrated by a surface
showing the batch-melt interface, extends nearly
to the back wall of the tank but does not touch
the electrodes; the half-circles indicate
measurement points
26
lass melting tanks are continuously operating reactors producing up to 1000 tons/day
of glass at temperatures up to 1600C. They
have relatively simple geometry but complex heat
transfer and glass melt convection. The transformation
of the raw materials to the final glass melt passes
through complicated thermo-chemical processes involving solid, liquid and gaseous phases. At Stein Heurtey,
a worldwide supplier of thermal installations for the
steel and glass industry, several detailed analyses of
glass melting tanks have been performed using
FLUENT, resulting in an improved understanding
of many of the complex processes at work.
Typically, simulations of thermal installations are
performed using several steps. First, the critical process
parameters need to be identified. One of these parameters for glass tanks is the residence time on the
critical trajectory through the melt volume. The
critical trajectory, which has a crucial impact on
the glass quality,1 is defined either in terms of transit time, temperature, or shear flow history along
its path. Second, the experimental and mathematical
means for assessing the key process parameters
have to be identified. CFD is often used to study
the critical trajectories because glass tanks are very
difficult to access for measurements. Third, the main
factors that influence the key process parameters
must be identified. For the critical trajectories, one
such factor is the intensity of recirculation patterns
in the melt. The batch blanket, which consists of
the raw materials floating on the surface of the

glass melt, acts as a heat sink during the process,
and the heat flow just under the blanket has a strong
influence on the recirculation of the melt.2
To simulate the critical trajectories, a large number of tiny particles are released in the batch blanket using the DPM model and the escape times
at the glass tank throat are compared. The particle path with the shortest residence time represents the critical trajectory in terms of time. Flow
field precision, the step size of the tracking calculations, and the fraction of recovered to injected particles are key numerical factors in determining
the accuracy of the trajectory predictions.3 A precise simulation of the heat sink under the batch
blanket is a prerequisite, because this heat sink determines the recirculation intensity of the glass melt.
To address the modeling of the heat sink and
other problems associated with the batch melting process, a collaboration between Fluent and
Stein Heurtey was initiated. The target was to identify the requirements for batch blanket modeling
and to develop the appropriate tools in
FLUENT. Today, 3D batch models that include reaction kinetics, phase change, widely varying material properties, and free surface prediction can be
solved using these tools. One Stein Heurtey glass
furnace for which numerous measurements are available has been simulated in FLUENT. In this furnace,
sodium silicate glass is melted by the dissipation
of electric current fed through 24 electrodes in the
glass melt. Having no combustion above the batch
and melt, this type of furnace is called a cold top
glass melter. The batch heat penetration and the
kinetics of the batch reactions are taken into account
in the simulation.4 Predictions for the batch blanket thickness are particularly useful for assessing
the remaining melt depth available for the electrodes. Electrodes that are too near to the batch
corrode faster and may even be deformed by contact with the rigid batch. The 3D batch modeling tools now allow for optimization of the melt
space and electrode arrangement under the batch
blanket. Many other glass melting process issues
can be studied using CFD as well. At Stein Heurtey,
FLUENT has proved to be a very useful tool for
problem solving and design improvement of its
thermal installations.
References:
1
Kuhn W.S., Moukarzel C., and Clodic D., Some

aspects to the minimum residence time in glass tanks
and its mathematical modeling, Proceedings of ESG
Conference Montpellier, 2002.
Kuhn W.S., Mathematical modeling of batch melting

in glass tanks, Chapter 2.2, pp. 73-125: in
Mathematical simulation in glass technology, Eds.
Krause D. and Loch H., Springer Berlin, 2002.
Moukarzel C., Kuhn W.S., and Clodic D., Numerical

precision of minimum residence time calculations for
glass tanks: The TC21-RRT1 case, Glass Sci. Technol.
Accepted for publication.
Kuhn W.S., Marmonier F., Bessette D., Muralidharan,

and Dutta A., First principles batch modeling and
validation on a large cold top sodium silicate melter,
ICG Conference Extended Abstracts, Edinburgh, 2001.
glass & fibers
hill ripples, which are also

known as press ripples or
flow ripples, sometimes
occur on the glass surface under certain conditions in many hot-forming processes, such as pressing and
casting. These ripples can be
observed on many low-quality hollow ware products, such as pressed
wine glasses, in the form of concentric
waves on the surface of the foot. The
name chill ripples reflects the fact
that this phenomenon occurs if the
temperature of the tool used in the
process is too low. Indeed, any low
temperature wall that comes into
contact with the glass can cause ripples to form. The obvious countermeasure of using a higher tool
temperature is very limited, since the
glass tends to stick to the tools if their
temperature is too high. The window between the critical temperature for the onset of ripples and the
sticking temperature can be very
small. Hence, for high quality products either the process conditions
must be controlled very carefully, or
expensive post-processing must be
performed.
Using POLYFLOW, two case
studies were performed to study the
formation of chill ripples. A model
of a casting process predicted the
formation of ripples on the glass surface if a low initial mold temperature was assumed. In agreement with
experiments, no ripples appeared for
higher initial mold temperatures. A

close inspection of the temperature
and flow field corroborates the conjecture that chill ripples are a consequence of the strong dependence
of glass viscosity on temperature, and
that thermal expansion effects are
not the dominant cause. More precisely, the model calculations show
that the ripples are caused by an
arching flow; once the glass
comes into contact with the mold,
the vicinity of the contact point is
cooled rapidly and thus immobilized.
Less viscous glass flows around the
immobilized region and eventually touches the mold in a new contact point.
In an examination of a pressing
process, the quantitative agreement
between the model predictions and
published experimental data1 for the
critical tool temperature is good if
the initial glass temperature is
high, but not as good if the initial
glass temperature is very low. More
importantly, these simulations
revealed many interesting details of
the evolution mechanism of chill ripples. An attempt to observe this evolution in an experiment would be
very expensive, because of the high
temperatures and the optical inaccessibility of the process.
Chill Ripples
in
Glass
By Ulrich Lange, Schott Glas, Mainz, Germany
Reference:
1
Optical fiber preforms without (left) and with (right)

chill ripples
Kluge W.D., PhD Thesis, University of

Freiberg, 1988.
The evolution of a
chill ripple by an
arching flow in
a casting process;
temperature
contours on the
glass and mold
are shown
t = 15.00 s
t = 21.67 s
The formation of
chill ripples in a
pressing process
with temperature
contours on the
glass and tool
(y-direction
stretched in
pictures on the
right hand side)
27
electronics
chneider Electric, a world leader in electrical distribution and industrial automation, develops products that are compact, yet have a
high level of electrical performance. Their products must meet strict
international standards and satisfy customer requirements such as reliability, safety, low cost maintenance, and dependability.
Electrical transformers are a good example of products that pose such
challenges to produce. To last an expected lifetime of 30 years, transformers need to be protected from high temperature peaks, which are
a major contributor to material ageing. One test applied to transformers under development involves a comparison of a transformer operating inside a kiosk (a concrete or sheet metal enclosure) to the same transformer
operating as a free-standing unit. In both cases, the transformers are assumed
to be loaded at the same nominal electrical power. The hottest surface
temperatures are measured and compared, and the difference must not
exceed 10C.
If the temperatures in the kiosk are too high for the contained transformer case, the kiosk design must be modified. Since fan cooling adds
an additional maintenance cost and reliability issues, only natural con-
Natural convection airflow inside a kiosk, resulting from transformer heating; the
transformer surface is colored by temperature, with red the maximum; pathlines
from the right louver are in blue and those from left are in red; flows are
homogeneously mixed through the kiosk as shown by traces leaving the top
Optimizing
Transformer Designs
By Dr. Laurent Tarbouriech, Schneider Electric, Grenoble, France
Surface temperature on the transformer and the flow circulation on a vertical

plane slicing diagonally through the transformer
Inside the transformer, oil is released through gaps between the coils and
magnetic circuits; the flow of oil is shown using pathlines, colored by
temperature; the oil jets generate toroidal structures on top of the coils
28
vection is considered for kiosk cooling schemes. To model the air flow
around the transformer and test different kiosk architectures, CFD simulations using Icepak have been performed. Results show that the size
and location of louvers on the sides of the kiosk can be optimized to satisfy the international standards and minimize manufacturing costs at the
same time.
The transformer itself is made of steel sheets and coils immersed in
an oil-filled tank. The oil is heated by the transformer losses, and the heat
is dispersed throughout the oil by natural convection currents. The tank
dissipates the heat to the outside by natural convection in the surrounding
air and by radiation. Conduction through the solid structures is negligible by comparison. In Icepak simulations of the transformer, natural convection was included in both the air and the oil. The thermal inertia of
the two fluids was accounted for through temperature-dependent properties for each of the two fluids. The surface power densities of the heatproducing elements (the coils and magnetic circuit) was specified as well.
A fine mesh in the vicinity of the transformer was used. To minimize
the global size of the mesh, a non-conformal interface was positioned
around the transformer, so that a coarser mesh could be used in the outer
regions of the model. The coils and magnetic circuit were defined as hollow blocks, and the resulting mesh consisted of 865,000 hexahedral cells.
Without the meshing tools available in Icepak, the mesh would have been
roughly twice as big.
Transformer simulations were performed to study the temperatures
on the transformer surface and the flow of oil on the inside. Among the
many interesting flow features indicated in the results were toroidal shaped
structures generated by the jets of oil above the coils. Studies such as
these make the internal transformer flows and cooling behavior easier
to understand. At Schneider Electric, they contribute to the efficient development of products in which the number of prototypes is reduced, and
the development costs are kept under budget.
semiconductors
media supply
process
chamber
process
levels
1-4
exhaust
wafer
chuck
Putting
the Spin on
Semiconductors
media drainage
Diagram of the spin process equipment
he wet chemical etching of single wafers is

one of the most important processes in the
semiconductor industry. It is often performed
using spin-process technology, developed by companies such as semiconductor equipment manufacturer SEZ AG. Spin-process equipment makes
use of a rotating chuck inside a process chamber. The wafer is placed on the chuck with the
side to be processed facing upward. Liquid etchant
is supplied through a port from above. The jet
hits the spinning wafer and the fluid is forced radially outward by the centrifugal force. The
process chamber has several vertically aligned annular plates, between which the unused etchant is
drained from the processor. By adjusting the wafer
height, several consecutive etching processes can
be performed during a single process run. Each
new wafer position allows the etchant to drain
through a separate set of plates, avoiding contamination of one etchant by another. Compared
to wet bench processes, in which a stack of wafers
is submerged into the etchant, spin processing
requires significantly less time to complete. In addition, this particular design from SEZ AG produces
By Bertram Schott and Jaroslaw Kaczynski, Carinthian Tech Research (CTR) AG,
Villach, Austria; and Andreas Baldy, SEZ AG, Villach, Austria
high uniformity and unmatched repeatability in

the etching process. CFD simulations carried out
at CTR AG using FLUENT have provided unprecedented insight into the fluid flow characteristics
involved in spin-process technology, and results
are being used to optimize the equipment design.
The performance of spin-process equipment
can only be improved when the fluid flow inside
the equipment is well known. Although the central feature is a fluid jet impinging on a rotating
wafer in a gaseous atmosphere, models that focus
on the gas flow alone can be very helpful in assessing the performance of the equipment. The gas
flow serves the purpose of exhausting toxic gases
released by the etching process, and trapping tiny
acidic droplets at the walls that result from the
spraying of the etchant. To meet these goals, it
is important for the gas flow to be homogeneous
and have high speeds just above the wafer and
plate surfaces. Optimizing the flow inside the spinprocess chamber by means of experiments would
be much too expensive. In addition, the complex geometry precludes the use of simpler models (such as boundary layer theory) to illustrate
the flow with any precision.

Instead of experimental methods or analytical models, CFD has been used to help visualize the gas flow structures in the spin-process
equipment. The weak interaction between the
liquid and gas justifies the decision to exclude
the etchant in the CFD models. The results have
allowed the geometry of the device to be optimized to improve the gas flow conditions. The
optimized design has alternating narrow and wide
exhaust channels. A mixture of gas and etchant
passes through the wide channels, where the slower speeds allow the etchant droplets to separate
from the gas flow and fall onto the plate surfaces.
By contrast, only pure gas passes through the narrow channels at relatively higher speeds. The combination of alternating exhaust channels has a
relatively low flow resistance, allowing for
increased gas flow rates inside the process chamber, and improved exhaust of toxic gases. Because
of successes such as these, CFD simulations are
now becoming more and more popular in industrial applications that depend on well controlled
fluid flow.
Gas flow in a 3D
model of the spin
process equipment
2D simulation
shows the gas
flow near the
wafer surface
and between the
drainage plates
29
polymer processing
Molds
Pressure
that Feel the
By Masahiro Matsuno, Keeper Company Ltd., Kanagawa, Japan
eeper is a manufacturer of rubber products for a

wide range of industrial and domestic applications.
Its oil seals are made using a compression molding process. The molds are filled with a heated polymer, and one or more walls are moved to compress and
shape the final product. Non-uniformities that develop during this process need to be avoided, since they
can have a negative impact on the shape and integrity of the product. At Keeper, early attempts to understand the filling process using a structural analysis code
failed, so last year, FIDAP was introduced to the company, and simulations using CFD were initiated instead.
Using silicone rubber as the working material, a typical oil seal mold was modeled in a 2D axisymmetric
simulation using the volume of fluid (VOF) model. At
the start of the transient simulation, a rectangular slab
of material was positioned in the corner of the mold.
As time progressed, the left mold wall was gradually moved
to the right, squeezing the rubber and forcing it to seep
out and fill the remainder of the complex mold space.
After about 5 seconds, when a preset position was reached,
the motion of the mold wall was stopped.
FIDAP predictions for the shape of the rubber as a
function of time were in good agreement with experimental results, with the FIDAP predictions for the free
surface location lagging the actual free surface measurements by about 0.5 seconds. Another discrepancy
observed was that the order in which certain corners
were filled in the experiment was not always the same
as the order in which they were filled in the simulation.
This difference was attributed to behavior observed in
the experiments, but not included in the model, such
as the apparent expansion or swelling at the surface of
the rubber as it rounds sharp corners and seeps into small
crevices in the mold. Because it is not fully plasticized,
this swelling occurs during the release of the stress that
acts on the rubber at the entrance of the cavity. This
phenomenon requires further investigation.
In future work, this and other effects will be incorporated into 3D models with more of the actual geometric features of the compression mold, in hopes of
improving the ability of CFD to capture more of the flow
details during processing. A transparent die is also being
developed so that the entire process can be monitored
visually.
up
kin
cavi_3
rubber
cavi_1
cavi_2
low
Comparison of the FIDAP predictions for volume fraction of rubber (right) with two sets of
experimental images as the mold closes (left and center)
30
The geometry of the axisymmetric model showing the initial

position of the rubber
worldwide offices
CFD in GERMANY
By Keith Hanna, Fluent News
FLUENT CFD simulation

of airflow around the
Brandenburg Gate, Berlin
Simulation done by Fluent Deutschland
using FLUENT. Mesh created by
HARPOON, courtesy of Sharc Ltd., UK,
and CEI Inc., USA
31
worldwide offices
The
Rapid
ince unification in 1989, Germany has
become the largest country in Europe
with the biggest economy in the
European Union, and the third largest in the
world after America and Japan. Germanys
industrial strength is due in considerable
part to the quality of the engineered products it produces. The automotive sector is
the largest, with some world-class marques
like Audi, BMW, Daimler Chrysler, Porsche,
and Volkswagen. In addition, Germany has
two of the largest chemical companies in the
world, BASF and Bayer, along with several
large companies in the power generation,
industrial machinery, domestic products,
materials processing, pharmaceutical, and
construction industries.
Germany exports nearly 500 billion
Euros worth of its manufactured products,
and with the recent global economic
downturn, the demand for German exports
has suffered a setback as well. This has resulted in difficult times for the German economy, which has recently experienced a slowing
growth and increased unemployment rate.
To become more competitive globally,
German industry has recognized the need
for efficiency improvements in its process-
GERMANY
Capital
Berlin
Population
83,251,851
GDP
2,063 billion Euros (2001)
GDP - per capita
26,200 Euros
GDP - composition
agriculture: 1%
industry: 28%
services: 71%
Total Area
357,021 sq km
Exports
machinery, vehicles, chemicals, textiles,

metals, manufactured goods, foodstuffs
Flow and pressure field around the ECO Speedster

Courtesy of Opel AG
32
worldwide offices
Growth
of
CFD in
Germany
es, along with the need to produce higher quality products. This has led to a widespread embracing of new technologies,
including computer-aided engineering,
across all industry sectors, with the automotive
industry leading the way.
German companies spend far more on
research and development in Europe than
other nations, demonstrating their commitment to technology innovation. Despite
this, they were, for many years, relatively
cautious about adopting commercial CFD
software, compared with the US, Japan, and
the UK. No major CFD companies origi-
nated in Germany in the 1980s, as they did

in the US and England, and it was only in
the early 1990s that CFD usage became widely accepted throughout German industry and
its leading universities.
A strong role in evaluating and adopting CFD technology has been taken by the
R&D groups at leading German companies
and through early adopters at key universities like Aachen, Darmstadt, Munich
Karslruhe, and Stuttgart. By the mid 1990s,
the CFD market in Germany started to become
very competitive, eventually taking off with
the advent of unstructured CFD codes and
faster, more powerful, and less expensive

hardware. Gradually, CFD software once
the preserve of analysts doing research projects began to be deployed by engineering and design departments in leading German
companies, with the automotive industry
again taking the lead. Cost savings, process
efficiency gains, and improvements in
product quality were major driving factors
for the rapid expansion of CFD use in Germany
during this time. The high accuracy and robust
solvers offered by the most established commercial codes appealed to CFD users in
Germany as well.
A sensor is used to measure mixing behavior

in a stirred tank with a dip tube
Courtesy of Merck KGaA; Photo courtesy of Chemineer, Inc.
Flow inside cylinders in a diesel engine

Courtesy of Deutz AG
33
worldwide offices
The offices of Fluent Deutschland
Fluent Deutschland
Expands its Services

luent Deutschland opened its doors in 1991,
with a central location near Frankfurt. In
addition to Germany, it serves Austria and
the German speaking parts of Switzerland.
Today, Fluent Deutschland is the largest CFD
company in Germany with a staff of more than
50 people. Its strength and rapid growth during the last ten years are due to a number of
factors. First, it is the only major CFD company
to provide a full range of CFD services in the
German language by locally-based engineers
who are focused on industry specific CFD applications. Second, it offers the widest range
of CFD codes in the German marketplace:
FLUENT, FIDAP, POLYFLOW, Icepak, MixSim,
and Airpak, making it the most industrially diversified CFD supplier in the country. Third, Fluent
Deutschland supplies a full range of CFD services including application consulting, training,
a web-based remote simulation facility, and
funded development.
According to Dipl.-Ing. Udo Weinmann, the
General Manager of Fluent Deutschland, Our
growth has come about because of our ongoing commitment to provide the best products
and services in Germany, especially with Fluents
easy-to-use, fully unstructured software tools.
In addition, our dedicated, industrially-focused
technical support teams work to make sure that
our users succeed with CFD. Our company has
a highly motivated workforce, and we have
F
At Boysen, we have been using FLUENT
for years for the numerical simulation of
unsteady and highly turbulent flow
patterns. These computations build the
basis for evaluation and optimization of
components regarding durability and
emissions. We value Fluent Deutschland
as a competent partner in support and
as a reliable provider of solutions to
complex problems."
Florian Lderitz
Friedrich Boysen GmbH & Co KG
34
had a healthy financial position throughout our

existence. Most importantly, we are striving
to provide better software and services, tailored to the particular needs of the German
CFD community.
During the last three years Fluent Deutschland
has made major inroads into the German automotive industry because of the versatility, robustness, accuracy, and reliability of FLUENT. Fluent
Deutschland leads the CFD market in Germany
due to the fact that unstructured CFD simulations are now the benchmark for commercial codes. Today, all of the major German
companies use CFD in one way or another.
The degree of use depends on the benefit that
can be derived from CFD within the company or industry, and the level of awareness of
CFD and its potential. German engineers want
accurate, easy-to-use CFD software from
geometry creation to grid generation to CFD
solution to postprocessing. Many German customers have seen significant cost savings, process
improvements, and product enhancements
through the use of CFD. In some German industries, suppliers to major companies are now
being asked to include CFD simulations with
their product offerings, and in others, such as
the built environment and power generation
industries, governmental legislation is making
CFD simulations compulsory.
worldwide offices
Technical
Support is
Key to CFD
Success
luent Deutschlands technical support group,
with a staff of over 20 engineers, provides highquality support in the local language, while drawing upon the expertise resident in the worldwide Fluent
organization, as needed. Three industry-focused teams
address a spectrum of CFD applications that range from
hypersonic flows to non-Newtonian plastics. According
to Dr.-Ing. Henning Rexroth, Technical Services Manager
at Fluent Deutschland, The support engineers are familiar with the problems and pitfalls of the real engineering world and how CFD can be successfully applied to
solve flow problems. They are realistic in setting the expectations of what CFD can and cannot do. We build longterm personal relationships with our clients, partnering
with them to make them successful with CFD. German
engineers and companies are demanding of their software products but are very loyal once they are happy
with a products capability and a companys services.
Many German companies have stayed with Fluent
Deutschland since the company was formed, and annual user surveys have shown a consistently high level of
satisfaction. Clients know that they can call the office
during German working hours and get expert support
to guide them on their CFD use and its deployment within their company. Fluent Deutschland is the only CFD
supplier in Germany to have an annual Users Group
Meeting in the German language. At these well-attended meetings, users meet with the Fluent Deutschland
staff directly and present their activities to peers in technical sessions. A strong corporate university program ensures
unrivaled support for academic users at all levels.
In addition to the support services offered by Fluent
Deutschland, other departments offer specialized services as well. For example, a local development team works
on key parts of the FLUENT code. A number of Fluent
global market managers are located in the Darmstadt
office. A local consulting services group offers timely solutions to clients custom needs. Training courses and seminars on industry-focused or product-specific topics are
also made available to new and experienced users.
Dipl.-Ing. Udo Weinmann, the General Manager of Fluent

Deutschland, has been with the company since its formation in 1991, having worked at Fluent Europe prior
to starting the Darmstadt office.
We are striving to provide better

software and services tailored to
the particular needs of the German
CFD community.
Dr.-Ing. Albrecht Gill joined Fluent Deutschland in 1994.

As Technical Services Manager for several years he was
responsible for support, consulting, training, and
development. Since October 2002, Dr. Gill has been leading the marketing & sales activities at the company.
We will continue to grow within

existing CFD sectors and in new
application areas because of our
reliable software products with
their ever-increasing functionality.
Growth among design engineers
will result from our tailored, easyto-use software products and
solutions.
Dr.-Ing. Carl-Henning Rexroth is the Technical Services
Manager at Fluent Deutschland. He joined the company
in 1997, and previously worked as the Power,
Environmental, and HVAC team leader.
We build long-term personal

relationships with our clients,
partnering with them to make
them successful with CFD.
Dr.-Ing. Markus Braun has been with Fluent since 1995.

He is the Manager of the Development Group at Fluent
Deutschland, part of Fluents worldwide development
team located in 5 countries on 3 continents. His group
specializes in the discrete phase modeling tools in
FLUENT in the presence of moving and deforming walls.
They also carry out projects in the fields of plasma modeling, fuel cell simulation, and fiber modeling (for the
glass and textile industries). In addition, the team is working on biomass reaction modeling and multiphase flows.
35
worldwide offices
Experienced
Automotive
CFD Knowledge Base

erman automotive companies are always looking
for new ways to shorten
their design cycles. Toward this end,
virtual prototyping by way of computer-aided engineering is becoming the standard for a wide range
of automotive processes today. Fast
turnaround and accurate CFD predictions are critical, especially in light
of the rapid globalization that has
evolved throughout the automotive
industry. An important goal in this
environment is to be able to reduce
testing and troubleshoot virtual
prototypes early in the design cycle.
To meet these needs, German
automotive CFD users need robust
software that offers the latest physical models. Unstructured meshing
and parallelized solvers are also key
features, since automotive simulations
with several million cells are now common. Because CFD has gained
almost complete acceptance within this industry, its extensive capabilities have been transferred to a
widening range of applications.
Many non-traditional automotive simulations are now being performed,
Windscreen deicing
Courtesy of Visteon Deutschland GmbH
Pathlines illustrate the flow through a manifold
such as gas sloshing in a tank or cavitation in a fuel injector, both of which

make use of multiphase models. Use
of liquid spray combustion models
is also on the rise for in-cylinder simulations using the deforming mesh
model in FLUENT.
In addition, there has been an
increased demand for interoperability
with other virtual prototyping software. Two frequently used CAE tools,
CATIA for computer-aided design
and ANSA for surface meshing, are
frequently used in Germany, and tools
are available in GAMBIT to import
geometries and meshes from these
packages. Indeed, GAMBIT is well
equipped to import data from most
of the CAE software packages in use
today. This compatibility is an essential ingredient in the competitive
automotive manufacturing world.
The Fluent Deutschland automotive team tackles the full gamut
of automotive applications, including underhood flow and heat transfer, climate control, powertrain
analysis, external aerodynamics,
and aftertreatment simulations.
Courtesy of Filterwerk Mann + Hummel GmbH
Fluents Global Automotive Market

Manager, Dipl.-Ing. Werner Seibert,
is located in the Darmstadt office.
His role is to coordinate Fluents automotive teams worldwide, and to work
with key clients such as Ford, GM,
Daimler-Chrysler, Toyota, and PSA.
He serves as a conduit for automotive clients to feed their requests into
Fluents development schedule. Mr.
Seibert joined Fluent Deutschland in
1994, and previously worked as the
Automotive Team Leader. He has been
extensively involved with Fluents
benchmarking activities over the years,
particularly for external aerodynamics flows.
Members of the 2003 Automotive and Aerospace industrial team, from left to right:
Dr. Rolf Reinelt, Dr. Ingo Futterer, Marco Lanfrit, Frank Kaufmann (Team Leader),
Michael Ehlen, and Marco Oswald
36
worldwide offices
ermany has some of the biggest chemical
and materials processing companies in the
world. Some of the most complicated processing operations known to engineers, involving
mixing, polymers with complex rheology, multiphase
mixtures of liquids, solids, and gases, and chemically reacting flow are commonly found in these two
sectors of German industry. Difficult chemical process
operations such as these constitute some of the most
challenging applications that CFD can solve. To meet
the vast needs of these applications, many companies
make use of more than one CFD product from Fluent
Deutschland.
For the chemical processing industry, CFD has
yielded detailed information about unit operations
previously considered to be black boxes. Time and
again, it has helped improve process efficiencies, saving large sums in operating costs. Mixing simulations, for example, can be used to optimize impeller
locations and shaft speed so that existing equipment
can be used for new processes. CFD is unarguably
recognized as the most reliable way to reduce scaleup risks and troubleshoot equipment problems at
minimal cost. In addition, it is increasingly being coupled with 1D process simulation packages, such as
flow sheet software, to better analyze chemical process
plants and equipment.
In the materials processing sectors of plastics, glass,
food, metals, textiles, and consumer products, CFD
has become an enabling technology. It has been used
to evaluate innovative new ideas, leading to shorter product development times. For example, measurements are difficult or impossible to make in processes
involving molten glass, polymers, or metal, and CFD
predictions provide the best way for plant engineers
and managers to visualize the fluid behavior and how
it will change as modifications to the equipment are
considered.
Dr. rer. nat. Jochen Schtze of the Darmstadt office is the Product
Manager for MixSim, Fluents design tool for mixing applications.
MixSim 2, due for release this year, combines an easy-to-use interface with the GAMBIT preprocessor and the FLUENT 6 solver. Any
shape of impeller can be created automatically in a fully unstructured meshing environment. This flexibility allows for multiple impellers
and shafts and a variety of other tank internals, such as baffles, heat
exchangers, and dip tubes. Dr. Schtze joined Fluent Deutschland
in 1999 after completing a PhD in Biotechnology, simulating multiphase flows with mass transfer using CFD. He brings to his role a
wealth of user-defined functions (UDF) programming skills and mass
transfer related technical experience.
The Right Mix

Chemicals
and Materials
Processing
for
Cooling air flow during glass

pressing process of a cathode
ray tube (CRT)
Courtesy of Schott Glas
Temperature contours on the

mid-plane of a steam reformer
Courtesy of HuR ChemPharm
The 2003 Chemicals and Materials Team, from left to right: Dr. Mourad Lotfey, Dr.
Mark Pelzer, Annelie Groten, Ralf Lffler (Team Leader), Dr. Stefan Martens (Business
Development Manager), Dr. Christian Budde, and Dr. Jochen Schtze
37
worldwide offices
CFD software from Fluent has become an
integral part of our design services, due to
the increased planning security that
simulations make possible.
Efficient
Dr. Peter Vogel,

Gebude-Technik-Dresden GmbH
Power, Environmental, and HVAC
Solutions
he power generation CFD market is one of the most
mature in Germany, covering everything from boiler, burner, and turbomachinery manufacturing to
power plants of all types. In addition, Germany has a
very sophisticated level of public environmental awareness that has resulted in some very stringent pollution
abatement legislation, and a governmental drive for sustainable energy sources like wind, biomass, and fuel cells
in the near future.
Fluent is the leading CFD supplier to the Power Industry
in the world, with state-of-the-art models for combustion and heat transfer, and an experienced technical support team in Darmstadt. German power industry users
want to increase process and equipment efficiencies to
reduce costs, and to address pollution restrictions for
NOx, CO, soot, and heavy metal levels in off-gases and
power plant wastes. Fuel cell development is at the leading edge of power generation R&D, and CFD is helping speed up this technological development. In the field
of renewable energy, wind power generators are using
CFD extensively for blade design and wind farm mapping studies. German companies are increasingly
working to develop biomass furnaces and incinerators
as well.
In the building industry, fire and safety hazard evaluations are now customary, and CFD simulations are
frequently used during the design stage to maximize
the safety of occupants during a fire and minimize the
possibility of expensive litigation down the line. In addition to fire analyses, CFD is used to design the airflow
in many newly constructed German buildings to ensure
occupant comfort for a range of exterior weather conditions. Equipment designers in the HVAC and electronics
industries also use CFD to improve designs for
enhanced performance in operation. Fluent products
and the support staff at Fluent Deutschland are wellequipped to meet client needs in all of these important
application areas.
Pressure contours on the buildings of

Terminal 1 at the Frankfurt Airport in
northeast wind conditions; simulation
done by Fluent Deutschland
Contours of predicted velocity

magnitude at selected axial
and radial planes
Courtesy of KSB Aktiengesellschaft
Members of the 2003 Power, Environmental, and HVAC team are, from left to right:
Dr. Wolfgang Timm, Elmar Schneeloch, Stefan Braun, Michael Adler, Ingo Cremer
(Team Leader), and Dr. Ulrich Schmidt
38
Fluent Deutschland GmbH

Birkenweg 14a
D - 64295 Darmstadt
Tel.: +49 (0)6151 / 36440
info@fluent.de
academic news
Species contour map computed using CFD is in good agreement with experiment1
Channeling
Chaos
By Patrick Bennett, Manhasset High School, Manhasset, NY; Chris Wiggins, Columbia University, New York, NY; and Marc Horner, Fluent Inc.
icrochannels can be used to transport, mix, and process fluids

such as DNA. By using these small channels, hand held devices
can be created that perform the same task as entire laboratories
in a space one-thousand times smaller and with an accuracy one-million
times greater. Due to the small dimensions of microchannels (diameters
on the order of 200 microns), their Reynolds numbers are always low
often approaching zero resulting in Stokes flow and poor mixing.
A particular geometry that has sparked interest for its potential for chaotic and improved mixing1 utilizes an array of asymmetric chevron grooves
etched onto the floor of a microchannel to promote transverse components
in the fluid flow. CFD models for a number of microchannel designs of this
type have been created in GAMBIT and solved in FLUENT. Using periodic
boundary conditions and pressure-driven flow, the steady-state velocity field
was computed for each, and a series of species advection calculations were
run using a user-defined scalar (UDS) to simulate the transport and mixing of two distinct fluids. Contour plots of this scalar are in good qualitative agreement with experimental findings.1
Based on the UDS results, the degree of mixing was calculated through
a standard deviation function that associates complete mixing with a value
of 0 and complete segregation with a value of 0.5. These values were plotted against downstream length or converted into a percent mixed function for comparison with other geometries.
A nice benefit of using CFD for such a study is the flexibility that it yields
in terms of visualization. Traditional empirical methods typically make use of
Students Take Home

ANTEC Prize
ongratulations to Matthew J. Day and James T. Haring from
the Behrend College of Penn State University for winning the
Best Student Paper Presentation at ANTEC 2002, Blow Molding
Division, with their paper Simulation Study of Polymer Flow Through
an Extrusion Blow Molding Head. Their work included simulations
done with POLYFLOW, and appears in the Fall 2002 issue of the Society
of Plastics Engineers, Blow Molding Division Proceedings.
Species contours on the boundaries

of the mixing device illustrate ditch
mixing or mixing inside the grooves
confocal microscopy, which is difficult to do and only planar in nature.

FLUENT allows for 3D continuum images to be rendered, displaying such
things as the development of a transverse component to the flow, the creation of counter swirls, and the effectiveness of the ditch mixing process
in the mixer (mixing inside the grooves). FLUENT has allowed for a much
more robust exploration of the 3D, chaotic flow patterns in the system.
One geometric parameter of interest was the groove depth. Through
the optimization trials, where different groove depths were analyzed and
compared, it was shown that increasing the depth of the grooves both
decreases pressure drop and increases effective mixing. This leads to the
conclusion that the added volume of the grooves acts as a buffer to the
no-slip condition on the walls of the channel and grooves and allows for
stronger transverse components to be added to the flow, promoting the
stretching and folding actions that are required for mixing.
Reference:
1
Stroock A.D., Dertinger S.K.W., Ajdari A., Mezic I., Stone H.A., and Whitesides G.M.,
Chaotic Mixer for Microchannels. Science Magazine, 295, January 2002.
Editors Note: Pat Bennett presented this work at the American Physical Society,
Division of Fluid Dynamics Meeting last November, and entered the 2002 Intel
Science Talent Search and Siemens Westinghouse competitions. He was recognized as a semi-finalist at both events. A FLUENT user for three years, he
will enter Stanford University in September.
39
academic news
Re-entry Vehicle
Shocks
By Tracie J. Barber, University of New South Wales, Australia
he simulation of high Mach number flows is difficult experimentally, and actual flight tests are not
feasible. CFD is a convenient method to use to
study this type of flow and predict flight performance.
Other advantages include the ability to predict flow
properties that are difficult to capture experimentally, such as detailed pressure and temperature distributions.
At the University of New South Wales (UNSW), experimental, computational, and theoretical results were
recently compared for two cases that exhibit quite different shock behavior a cone and an Apollo module. This preliminary work will form the first stage of
ongoing research into re-entry vehicle and rocket flight
analysis. The two cases chosen represent two fairly simple bodies, exhibiting quite different shock behavior.
The Apollo module model is well-known to have a curved,
non-attached shock wave before it in supersonic flow.
The cone is a well-studied body in supersonics and exhibits
an attached shock. The variation in the shock waves
produced by the two bodies proves a useful test of the
CFD modeling capability, while also allowing the flow
in the base region to be studied.
The first body studied is a 15 cone, traveling at
zero angle of attack, and outfitted with two pressure
taps. The second is a 1/30th scale model of the Apollo
re-entry vehicle, also traveling at zero angle of attack,
and outfitted with four pressure taps. Visualization of
the shock waves on the actual bodies was performed
using the Schlieren method at the UNSW supersonic
wind tunnel facility at a Mach number of 3.05. At this
Mach number, the static pressure is found to be 11.32kPa
40
academic news
Comparison of CFD
and experimental
shockwave locations
for the Cone model
in the test section. Temperature was calculated to be 102.5K, local density 0.3848 kg/m3, and the local speed of sound found to be 202.9 m/s. Reynolds
numbers for the two cases, based on model characteristic length, were found
to be approximately 8.52x105.
The Schlieren method makes use of the high density gradients present
in flows exhibiting shock characteristics to enable visualization of the shock
waves. Simulations of the two vehicles were performed using FLUENT 6.
The Spalart-Allmaras turbulence model was used, and converged results
were obtained using adapted meshes and second order upwind differencing.
For computational efficiency, both cases were run as axisymmetric models. Although the vehicles are indeed axisymmetric, the wind tunnel test
section in which they are studied is not. Therefore any shock waves reflecting off the simulated tunnel walls, or their subsequent effects, are not correctly captured by the 2D models.
For the 15 cone, the photographic Schlieren result and the CFD result
both show an attached shock wave that can be seen as the dark straight
lines coming off the very front of the cone. The shock wave angle measured from the experimental image is 25.5 and from the CFD image is 25.1.
Pressure coefficient values found from the experimental pressure taps and
from corresponding CFD locations match well.
Tap Number
Cp (Exp.)
Cp (CFD)
1 base of cone
-0.1305
-0.126
2 side of cone
0.1856
0.184
Pressure coefficient values predicted by FLUENT for the 15

cone are in good agreement with experiment
For the Apollo module scale model, the unattached shock wave in the
photographic Schlieren result appears as a gradient in the image, off the
front of the body. The CFD predictions for the size and location of the shock
wave are in good agreement with experiment. Pressure coefficient values

are also calculated for the locations of the four pressure taps, and with the
exception of one site, good agreement is obtained. The site where the agreement is poorest is located at the base of the model. As this measurement
location is found just after a sharp corner, where the flow is subsonic, it is
likely that the turbulence model used is not accurate enough to capture
the recirculating flow in this region. Further work is planned to investigate
other turbulence models as the prediction of the effects on the afterbody
are also of interest.
Tap Number
Cp (Exp.)
Cp (CFD)
1 front corner of the module

2 behind the module
3 side of the module
4 front of the module, on axis
1.211
-0.1237
-0.1359
1.585
1.24
0.048
-0.135
1.69
Pressure coefficient values predicted by FLUENT for the Apollo scale model
are in good agreement with experiment
The theoretical downstream properties expected for the flow, based on

the equations for normal shockwave relations, can also be computed from
the FLUENT results and compared to a theoretical value. For the Apollo model,
the ratios of downstream to upstream (relative to the shock wave) values
of pressure, temperature, and density were found at the front central location of the model. For the most part, these ratios were found to be in good
agreement with the values calculated theoretically (from standard shockwave relationships). In particular, the good comparison for the temperature ratio across the shock (1.1%) is a useful indication of the validity
of the CFD model since no experimental data for the temperature was
available.
Comparison of CFD
and experimental
shockwave locations
for the Apollo model
41
product news
FLUENT Ported to
Itanium 2/HP-UX Platform
By Lee Fisher, Hewlett-Packard, Burlington, MA and Stewart Featherstone, Fluent Inc.
The Itanium 2/HP-UX system installed in

Visteons data center in December 2002
The Intel Itanium 2 processor, introduced in mid-2002, is designed for floating-point intensive 64-bit applications running
on servers, clusters, and workstations. Codeveloped with HP, this architecture has
made headlines for its parallel processing
design and future roadmap. A fully tested version of FLUENT (FLUENT 6.1.18) for
the Itanium 2/HP-UX 11i platform is now
available for download from the User Services
Center. Benchmarks have demonstrated
significant performance gains compared
to earlier HP-UX PA-RISC workstations.
Visteon Corporation has been an early
adopter of the new Intel/HP architecture,
and is using it for production runs. This
leading Tier I automotive supplier uses
FLUENT for critical analysis of automotive
climate control subsystems. By selecting
an Itanium/HP-UX server solution, Visteon

has been able to reduce cost, meet the growing performance requirements of its CAE
users, and bring all numerically intensive
computation into one centralized facility.
Itanium has delivered a flexible solution
for the wide variety of structural analysis,
CFD, and crash codes that Visteon uses to
support their automotive OEM customers.
Fluent and HP continue to enhance
FLUENT performance on this platform.
Recent optimization work has resulted in
gains that average 47% on benchmark test
cases. A prototype of this further optimized
version is now available by special request,
and a fully tested maintenance release is
planned for the summer. Fluent and HP
are also working to develop a port to the
Itanium/Linux platform.
Mixing Simulation
Gets Easier
By Jochen Schtze, MixSim Product Manager
ith the release of MixSim 2.0, the

numerical simulation of fluid flow
in stirred tanks will enter a new era.
The revised user interface allows for quick
setup of mixing equipment using a parts tree,
where objects can easily be added, modified,
and deleted.
Several brand name impellers and a variety of generic impeller types are available.
The flexible library file format allows for the
addition of any kind of impeller, and other
tank internals as well. CAD files of impellers
can also be read by MixSim and incorporated
into a vessel with other MixSim-specified internals.
After the mixing vessel which can include
multiple arbitrarily positioned shafts, baffles,
and various top and bottom shapes is completely specified, MixSim automatically creates the geometry and mesh in GAMBIT, sets
42
up a FLUENT 6 case file, and performs the

calculation, automatically displaying appropriate convergence monitors. New, intelligent postprocessing functions provide
informative process parameters and easy-tounderstand graphical presentation of the results.
As an added bonus, the MixSim-generated files can be read into the stand-alone products GAMBIT and FLUENT, or they can be read
back into MixSim for further modifications.
During a MixSim session, the MixSim console can be converted to the full FLUENT console with a simple mouse click. This allows
adjustments or extensions to be made to the
model definition. Finally, repetitive tasks can
be coded using FLUENT journal files, which
can be executed through the graphical user
interface. In short, the new MixSim goes a
long way towards making the simulation of
complex mixing tanks much easier.
A mixing tank, created by MixSim 2, contains three impellers

and four baffles; in addition to velocity vectors, velocity
contours are shown on the impellers and on surfaces near the
middle and lower impellers; pressure contours are shown on
the baffles; and iso-surfaces of helicity are shown behind the
top impeller blades
product news
POLYFLOW 3.10
Coming in June 2003

By Thierry Marchal, POLYFLOW Product Market Manager
OLYFLOW 3.10 is scheduled for release

in June 2003. The latest version includes
numerous features that will simplify tasks
for existing users and further extend the scope
of current capabilities.
A fully coupled fluid structure interaction
(FSI) model allows for the calculation of thermo-mechanical stress within a solid die or part(s)
of it. Deformations induced by the flow and
the possible impact of these on particle behavior is taken into account. Two new types of
optimization are available: 1) for extrusion applications, optimization of the die geometry using
the third party software VisualDoc to balance
the flow profile across the die lip, and 2) for
blow molding simulations, optimization of the
thickness map of the initial parison. User-defined
templates (UDTs) will allow users to define their
own templates by flagging parameters as easily as defining an evolution or a PMAT dependence. An inexperienced user will be able to use
a UDT to run similar simulations on similar geometries. The adaptive meshing technique has been
extended to refine the mesh when large variations in certain variables are calculated
across a given element. For example, this feature makes simulation of a thin thermal layer
caused by viscous heating less dependent upon
the initial mesh. Similarly, a sizing function for
adaptive meshing has been very helpful for simulations involving large deformations, such as
glass forming applications.
Extension of 64-bit compatibility to several UNIX platforms, a LINUX version, the PomPom viscoelastic model1, and slippage along
a rotating part are some of the other new features to be found in POLYFLOW 3.10.
reference:
1
McLeish T.C.B. and Larson R.G. Molecular constitutive equations for a class of branched polymers:
the pom-pom polymer, J. Rheol. 42(1),
p. 81-110, 1998.
here was a great deal of positive feedback from users

following the release of FIDAP 8.7 in December 2002.
The ability to run volume of fluid (VOF) simulations
in parallel was reported to be especially welcome.
Simplification of the setup of the free surface and evaporation models, the new partially coupled solver, and
the bundling with FIELDVIEW, FIDAPs new postprocessor,
are the features that generated the most enthusiasm.
Some users reported that the time between releases is sometimes too long, especially when they are waiting for the fix of a defect critical to them. To respond
to this request, releases will become available on the
Users Services Center more regularly. The first of these
is FIDAP 8.7.2, which is now available for download from
the USC. This maintenance release addresses many issues
that users have recently reported, with a special focus
on applications involving VOF and simulations being run
in parallel. For example, in FIDAP 8.7.2 VOF demonstrates robust performance when run in parallel on more
than 2 CPUs. Inconveniences that occurred with userdefined subroutines in parallel VOF simulations have disappeared. The accuracy of the results (as indicated by
mass balances) when tet meshes are used has been
improved. The performance of other models run in parallel has also been carefully addressed. Some former difficulties such as parallel runs on Windows 2000 with SLIP
entities, simulations involving FSI, or the use of the FDSTOP
file, have vanished in the new version.
Optimization of a PVC die

Courtesy of Plastinnov, St Avold, France
FIDAP 8.7.2
Released in
April 2003
By Thierry Marchal, FIDAP Product Market Manager
Injection molding of a PET preform for a soda bottle
43
product news
The
Draw of
POLYFLOW-FLUENT Coupling
By Geraldine Deflandre, Fluent Benelux and Andy Young, Fluent Europe
ptical fiber drawing is a process

that is in wide use today as a result
of the growing number of applications that rely on optical fiber technology. It is a challenging process to model
accurately because of the complex governing physics inside and outside the fiber
material, and because of the changing fiber
profile. The fiber begins as molten glass
with a steep exponential temperature
dependence for the viscosity. During the
drawing process, the cross-section can
reduce by a factor of more than 10,000,
and the draw ratio, or ratio between the
velocity through the outlet and inlet sections, can exceed 100,000. The material
is subjected to mixed environmental conditions as well: radiative heating and convective cooling at the same time.
In a recent project, POLYFLOW and
FLUENT were coupled to address fiber drawing simulations in a rigorous manner. The
motivation for the work was to develop
an industrial-strength methodology that
could handle non-axisymmetric and hollow fiber profiles with high draw ratios,
and offer a comprehensive heat transfer
capability. The iterative calculation uses
POLYFLOW to compute the profile shape
and interior velocity field, and FLUENT to
compute the temperature and velocity fields
outside the fiber and the temperature field
inside the fiber.
Each code has particular strengths to
meet the specific challenges of this

process. POLYFLOW uses the Streamwise
remeshing algorithm that relocates the internal nodes as the profile is adjusted after
each iteration so that the best element quality is maintained, especially when high draw
ratios are encountered. FLUENT offers the
discrete ordinates (DO) radiation model that
allows for the absorption, transmission,
reflection, and refraction of radiation in the
presence of the semi-transparent fiber material. The model is capable of correctly capturing the radiative heat transfer to the glass
despite the rapid change in glass thickness
as the fiber is drawn. Accurate predictions
from a simulation such as this can be used
by engineers to minimize the heat requirements for the process.
The solution process begins in POLYFLOW,
where an initial fiber shape is computed
from the draw velocity. The fiber shape is
passed to FLUENT, where heat transfer to
(and inside) the fiber is calculated. The new
temperature field is then used by
POLYFLOW to calculate a revised fiber shape
and internal velocity field. The process repeats
until the fiber shape and temperature range
are within 1% of the previous values. In
2D and 3D test simulations, convergence
was achieved after 3 and 4 global iterations, respectively, and predictions of the
final fiber shape and temperature profile
were in very good agreement with
expectations.
Initialization
POLYFLOW
Filter
FLUENT
Filter
POLYFLOW
Convergence
no
yes
Solution
The coupled calculation between POLYFLOW

and FLUENT; filters are used to export data
between the two codes
Temperature contours and fiber cross-section for an axisymmetric simulation

showing the fiber after the initial POLYFLOW calculation without radiation (top)
and after the iterative loop (in which radiation is taken into account ) has reached
convergence (bottom)
Temperature contours on the fiber surface and
pathlines, colored by temperature, in the gas
44
support corner
he primary goal of running FLUENT in parallel is
to reduce calculation turnaround times by using
multiple processors (CPUs). There are two ways to
run FLUENT in parallel in the Windows environment.
In one model, multiple processors on the same
machine are used, and in the other model, multiple
machines in a cluster are used. The way in which memory is accessed is different for these two models:
Shared Memory
CPU
Distributed Memory
CPU
CPU
CPU
Memory
Memory
CPU
CPU
Memory
Memory
Memory
Parallel
Computing on a
Windows Cluster
CPU
CPU
By Diana L. Collier, Fluent Inc.
Memory (CPU and/or cache

memory) is shared between
the processors on a single
machine. Communication
between the processors is
handled locally on the
machine.
Each processor has its own

(private) memory associated with it. Communication
between the processors
occurs through a messagepassing interface (MPI).
The MPI is an interface for
a set of library functions that
processors in a distributedmemory multiprocessor can
use to communicate with
each other. Message passing software is required for
this task.
Last year, Fluent News featured a section on parallel

processing, and the support corner offered tips on how
to initiate a parallel processing run in FLUENT. Since that
time, many calls have come in with questions about how
to set up the hardware single multi-processor
machines or a cluster of machines (remote processors)
on a network especially when PCs running Windows
are being used. The following questions are typical of
those that have been asked.
Q.
I have a dual processor computer. How do I start

FLUENT using shared memory?
A.
Once FLUENT 6.1 is installed, open up a Command

Prompt and browse to your working directory.
Type: FLUENT 2d (or 3d, 2ddp, 3ddp) t2 (The
qualifier t2 starts up the two-process version of
FLUENT. The qualifiers 2ddp and 3ddp make use
of the double precision solvers.)
Q.
I have access to more than one computer, and

Id like to make use of these for parallel computing. How do the processors communicate
with each other?
A.
In this type of parallel implementation each processor has its own memory. The processors communicate
with each other through a socket communicator,
or MPI. Message-passing software is loaded on every
computer in the cluster and a Windows process
is started. Through the interface, computers coor-
dinate their tasks, such as sending and receiving arrays, synchronizing, and performing global operations (such as summations over all
cells), by sending and receiving messages to and from one another.
Q.
Which socket communicators and message-passing applications does

FLUENT support?
A.
FLUENT supports three of these tools: RSHD, MPICH, and MPI/Pro.
RSHD gets copied to the \FLUENT.INC\ntbin\ntx86 directory

when FLUENT 6.1 is installed.
MPICH is freely available from Argonne National Labs and can
be downloaded from their web site:
www-unix.mcs.anl.gov/mpi/mpich
MPI/Pro can be purchased directly from MPI Software
Technology, Inc. For more information, visit their web site:
www.mpi-softtech.com
Q.
What are the advantages and disadvantages of the three communicators that FLUENT supports?
A.
The advantages and disadvantages are summarized in the following

table:
application
RSHD
advantages
Free
A hosts file is not required
(can spawn nodes in the
FLUENT GUI)
Performs slightly faster
than MPICH
MPICH
MPI/Pro
disadvantages
Security issues (must use
an rhosts file for security)
Manual configuration is
necessary
Free
Argonne is continuously updating
Secure (password encryption)
Easy setup (no configuration)
Must use a hosts file

(cannot spawn nodes
in the FLUENT GUI)
Performance slightly better than

MPICH or RSHD
Secure (password encryption)
Easy setup (no configuration)
Cost
Must use a hosts file
(cannot spawn nodes
in the FLUENT GUI)
45
support corner
Q.
What is the difference between a hosts file and an rhosts file?
A.
A hosts file is a text document that contains the names of all the computers (and processors) on the network cluster on which you want to
run FLUENT. An rhosts file is the same, but is only required when you
are using the RSHD message passing software and you have concerns
about security. Note that there are restrictions regarding the naming,
storage, and configuration of the rhosts file. These are described in the
Security section of the RSHD installation guide, which can be found at:
www.FLUENT.com/support/installation/winfaq/rshd.htm
Q.
What does it mean to spawn nodes?
A.
A node is another name for a processor. To spawn means to initiate

a process. A parallel-processing system divides a complex problem into
smaller component tasks. The tasks are then assigned to the available
nodes. Spawning divides the computational workload into many tasks,
and assigns the tasks to the individual nodes. When the RSHD message-passing software is used, compute nodes on the network can be
selected by opening the Configure panel from the Parallel/Network
menu item in the FLUENT GUI.
Note: When the directory is shared you will notice a hand

under the directory.
4. Install the message passing software, making use of the comprehensive instruction guides available on the Fluent web site:
Code
Instruction address
RSHD
MPICH
MPI/Pro
www.FLUENT.com/support/installation/winfaq/rshd.htm
www.FLUENT.com/support/installation/winfaq/mpich.htm
www.FLUENT.com/support/installation/winfaq/mpi.htm
Q.
How do I configure FLUENT to run across a network using remote

processors?
5. Create a host file (required if you are using MPICH or MPI/Pro).
A.
The following steps outline the procedure for configuring FLUENT network parallel:
To create a host file, use a plain text editor, such as Notepad.

The file can have any name you wish as long as it does not
include any spaces. Save the host file to a working directory.
1. Install FLUENT 6.1. (It is only necessary to install FLUENT on

one of the computers on the network.)
Below is an example of a host file that uses two computers in

the cluster. computer1 has two processors, and must be listed
twice if both processors are to be used, and computer2 has
one processor.
2. Set the FLUENT.INC environment variables.

Click on the Start Menu, Programs, Fluent Inc Products,
Fluent 6.1, Set Environment. This will open a panel with
default environment settings. Click on Yes to modify the
settings to the values shown in the panel.
In order for all of the computers on the cluster to see
these necessary environment variables, you have to change
the local path to a network path. To do this, click on Start,
Settings, Control Panel, System. Choose the Advanced
Tab, and then choose Environment Variables. Edit the
FLUENT_INC and Path variables using the network path
format as shown in the panel below (where computer1
represents the name of the computer where FLUENT is
installed).
Q.
Once I have set up FLUENT and the necessary communication software, how do I start FLUENT to run in network parallel mode?
A.
Open up a command prompt and CD to your working directory.

Type:
FLUENT version version tnprocs pcomm path\\machine\share -cnf=host.txt
Where:
version specifies the version of FLUENT you want to run (2d, 3d,
2ddp, or 3ddp);
tnprocs specifies the number of processors you want to use (-t2
indicates that you want to use two processors);
3. Share the FLUENT.INC directory.
Right-click on the FLUENT.INC directory and choose
Sharing from the menu.
Click Share this folder
Click OK.
46
pcomm specifies the network communicator you are using (see

the table below for the correct pcomm command based on your
installed software);
partnerships
LMS SYSNOISE
Link to FLUENT
Spatial Provides
CATIA V4 Translation
for GAMBIT
AMBIT 2.1, due for general
used to produce the CFD domain.
release in Spring 2003, includes
Spatial is dedicated to constantly improvnative CATIA V4 translation proing our interop translators, said Linda Lokay,
vided by Spatial Corporation. Spatial, a
executive vice president of marketing and
Dassault Systemes Company, is the sole
development at Spatial Corp. Our native
provider of true, native CATIA translators.
CATIA technology makes our translators
Engineers at Valeo Engine Cooling, one
superior to any comparable products on
of many GAMBIT 2.1 beta test sites, were
the market. The results experienced by Fluent
very impressed with the new robust path
and their end users are indicative of our
from the CATIA V4 model file to automatic
continuous commitment to developing
meshing. GAMBIT has made great
products that contribute to our partners
progress in CAD import, cleanup, and autosuccess.
matic meshing, says Nicolas Franois,
research and development engineer at Valeo
more.info@
Engine Cooling, a major manufacturer of
www.spatial.com
compact heat exchangers and cooling systems for the automotive industry.
GAMBIT 2.1 is a clear
breakthrough in the path
from CATIA to high-quality meshes, says Erling
Eklund, GAMBIT product
marketing manager at
Fluent Inc. In a typical CAD
import sequence, a solid or
surface model created by
CATIA V4 is directly imported into GAMBIT 2.1. The
fluid volume is extracted
using simple Boolean operations, and straight-forAerospace valve model imported from CATIA V4 into
GAMBIT 2.1
ward cleanup tools are
SYSNOISE prediction of sound pressure level on an automotive

side-view mirror, based on flow-induced sources predicted in
FLUENT; the link to SYSNOISE complements the native capability
for noise prediction in FLUENT 6.1 (see article on page 23)
omputational aeroacoustics is of growing interest to analysts and designers, who would like to control or reduce
flow-induced noise. Fluent customers who use the popular acoustics package SYSNOISE from LMS International will
now be able to integrate their flow predictions from FLUENT with
acoustic solutions computed in SYSNOISE. The integration allows
FLUENT to be used for high-accuracy prediction of the flow-generated noise sources, with a coupling to SYSNOISE for prediction of how these sources are propagated and how they interact
with other acoustic phenomena.
The integration allows the time-varying surface pressure,
predicted in FLUENT, to be imported to SYSNOISE. Within
SYSNOISE, the flow solution data is processed into the frequency domain and becomes part of the acoustic prediction
performed in the SYSNOISE model. Mechanical excitations
and propagation of sound waves can also be included. Typical
applications might include noise induced by flow separation,
flow impingement, or from rotating fans or blowers.
The interoperability with FLUENT appears in SYSNOISE
Rev 5.6, and will be available from LMS International.
more.info@
www.lmsintl.com
Code
pcomm command
RSHD
MPICH
MPI/Pro
-pnet
-pnmpi
-pvmpi
\\machine specifies the name of the computer on which FLUENT is installed, and
\share is the name assigned when the directory FLUENT.INC was shared (for
example, in part 2 of the previous question, machine is computer1 and share is
fluent.inc);
hosts.txt specifies the name of the hosts file listing the computers on which you
plan to run FLUENT parallel. If you are not in the directory where the hosts file is
located you must specify the complete path to this file.
Sound confusing? It isnt, once you go through the process. All of these steps (and more
details) are available in the following comprehensive instruction guide to setting up FLUENT
network parallel: www.FLUENT.com/support/installation/winfaq/FLUENTpar61.htm
If you have questions or encounter any difficulties with the setup, please dont hesitate to
contact Installation Support at installsup@fluent.com for assistance.
47
Fluent Worldwide
around Fluent
Corporate headquarters
Happy Anniversary Fluent!

he year 2003 marks two important anniversaries
for our company. Twenty years ago, we celebrated
the sale of our first FLUENT license, and fifteen
years ago, we became incorporated. Many changes
have taken place since then, but much has stayed the
same. Many of the people who worked so hard to
bring the product and company together are still Fluent
employees today, and many of those who first licensed
our software have continued to be clients throughout the years.
In 1983, a small group of us at Creare, an engineering consulting firm in neighboring Etna, NH, introduced the first commercial version of FLUENT, which
had been developed by Prof. James Swithenbank and
his team, including Dr. Ferit Boysan, at Sheffield University
in the UK. The first version allowed for 2D or 3D structured grids using Cartesian or polar coordinates, steadystate flow, laminar or turbulent conditions, heat transfer,
three-component combustion, a dispersed phase, and
natural convection, with an easy-to-use, interactive frontend. We held our first sales seminar at Creare in October,
and the first sale of FLUENT was closed in December
to Fuel Systems Textron. The engineer who was first
to put his confidence in us is still a regular attendee
at our annual Users Group Meetings in the US.
In 1984, we sold six more licenses, and our business has grown steadily each year since then.
Meanwhile, Dr. Boysan launched our sister company, Flow Simulations Ltd. in Sheffield, which later became
Fluent Europe, the first of many European offices to
come. Because our early roots were as consulting engineers, a unique corporate ethos has defined us until
this day. Our mission has never been limited to software sales, but to make sure that all of our customers
get solutions to their engineering problems, whatever
they may be.
The first set of capabilities in FLUENT seems modest by todays standards, but they were state-of-theart back then. Never did we dream that in twenty years
we would reach the level of sophistication that we now
see in the pages of our newsletter: problems with millions of cells, moving geometry, LES turbulence modeling, and solution optimization!
When we began, we worked hard to educate industry managers about the existence and value of CFD.
Our early users were experts in their fields who were
not afraid to be the risk takers of the day. Once sold
on the benefits of CFD, they acted as evangelists to
convert their colleagues. Today, we see a top-level commitment at many organizations to make CAE an integral part of the production cycle, from beginning to
end. Our challenge now is to meet their needs by providing reliable software with easy access, customizability, and interoperability with other products.
Our steady growth, from the very first sale onwards,
has been mirrored by the ever-increasing capabilities
of our products as well as the number of users worldwide. The directions we have chosen to take have
always been the result of customer input. Thus, it is
you, our customers, with your continued support over
the years, who helped us become the company that
we are today, and I thank you. I also thank our capable and dedicated employees who are the heart and
soul of Fluent!
Bart Patel
CEO, Fluent Inc.
Take advantage of our new, online training courses! These topical training materials for FLUENT users are available at a reduced introductory rate of just $100 per
course. The courses are organized into manageable 30-60 minute modules and include
a number of relevant case studies and tutorials to enhance your practical knowledge and skills. You can stop and start as you please, and have access to the materials for up to 30 days.
Available courses include:
Turbulence Primer
Parallel Processing with FLUENT 6
Using User-Defined Functions with FLUENT 6
Solving Multiphase Flow Problems with FLUENT 6
Solving Combustion Problems with FLUENT 6
Solving Rotating Machinery Problems with FLUENT 6
Sign up today and take advantage of this opportunity to learn from anywhere,
at any time, and at your own pace. Introductory pricing will expire in 60 days. www.learningcfd.com
USA regional offices

Ann Arbor, MI 48104
Tel: 734 213 6821
Evanston, IL 60201
Tel: 847 491 0200
Santa Clara, CA 95051
Tel: 408 522 8726
Morgantown, WV 26505
Tel: 304 598 3770
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
Asian regional offices
NEW! Online Training
48
Fluent Inc.
10 Cavendish Court
Lebanon, NH 03766, USA
Tel: 603 643 2600
Fax: 603 643 3967
Email: info@fluent.com
Fluent Asia Pacific Co., Ltd.

Tokyo, Japan
Tel: 81 3 5324 7301
Email: info@fluent.co.jp
Osaka, Japan
Tel: 81 6 6445 5690
Fluent India Pvt. Ltd.
Pune, India
Tel: 91 20 6056381
Email: info@fluent.co.in
Distributors
ATES Korea
Beijing Hi-key Technology
Corporation Ltd. China & Hong Kong
Cavendish Instruments de Mexico, S.A.
de C.V. (CIM) Mexico, Venezuela,
Argentina, Chile, Colombia
FEM++ Israel (POLYFLOW only)
FIGES Ltd. Turkey
Fluid Codes Ltd. UK (serving Middle East)
INNOTECH Ltd. Hungary
J-ROM Ltd. Israel
LEAP Australia Pty., Ltd. Australia &
New Zealand
Process Flow Finland & Baltics
Regional Technologies Corp. Ukraine
Simcon International (Pvt.) Ltd. Pakistan
SimTec Ltd. Southeastern Europe
SMARTtech Services & Systems, Ltd.
Brazil
SymKom Poland
Taiwan Auto-Design Company (TADC)
Taiwan
Techsoft Engineering s.r.o
Czech Republic & Slovak Republic
TENSOR SRL Romania
Thermal Technologies/QFINSOFT
South Africa

CFD For HVAC by Fluent

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

CFD For HVAC by Fluent

Diunggah oleh

Hak Cipta:

Format Tersedia

Biomedical

A Digital Gas Mask

Glass & Fibers

his year it is with

On the Supplment Cover:

Fluent News is published by

This issues supplement features applications

Optimizing Plant-Scale LDPE Reactors

Deposition: One Layer at a Time

oil & gas

Separating Water & Gas

A Digital Gas Mask

glass & fibers

visions of the future

FLUENT Ported to Itanium 2/HP-UX Platform

Parallel Computing on a Windows Cluster

LMS SYSNOISE Link to FLUENT

Happy Anniversary Fluent!

Customizing Food Steamers

Keeping Printer Touch Temperatures Low

Shattering Glass Tank Design

electronics & semiconductors

Swiss Re Headquarters Inside & Out

Breathing Easier in the Workplace

Containing Smoke in Complex Atria

GaAs growth on a rotating substrate

LUENT 6.1 is unique in its vast offerings

PDF transport particles colored by temperature for a

Fluent NEWS spring 2003

mented through a collaboration with

Geometry and mesh of the generic 3-way catalytic converter

Temperature distribution on a cutting plane through

Mass fraction of CO on the cutting plane through

Chatterjee D., Deutschmann O., and Warnatz J., Detailed Surface

Fluent NEWS spring 2003

ith the ever-increasing availability of high-performance computing tools, CFD is becoming

conditions due to stiff and highly exothermic kinetics

The injection region (0 to 0.2 m) is zoomed

Fluent NEWS spring 2003

Berl Combustor Revisited

By Graham M. Goldin, Fluent Inc.

jets inject natural gas. The mixture then enters

loss of initiator under extreme operating and

tests, CFD offered a low-cost alternative to explore

Contours of NO on a center plane near the quarl

Pope S.B., Computationally Efficient

he focus of FLUENT 6.1 gas-phase combustion modeling is to provide affordable,

Comparison of FLUENT predictions of NOx with experimental

Fox R.O., Computational methods for turbulent reacting

Fluent NEWS spring 2003

cumulative thickness (Ang)

growth rate (Ang)

A typical ALD reactor along with pathlines

Where (a) is an adsorbed site/surface species, (g) is a gas phase species,

NH2(a) + TiCl4(g) TiCl2(a) + 2HCl(g) + N(s)

By Balaji Devulapalli, Fluent Inc.

conditions to account for pulse sequences, adsorption of the monolayer of

Fluent NEWS spring 2003

n many industrial applications, such as wafer processing

Mass fraction of CO2

Schematic of the problem, with the diameter greatly

Deutschmann O., Maier L.I., Riedel U., Stroemman A.H., and

Surface coverage profiles along the length of the tubular reactor