Vol. 06 / Issue.

01 / 2017

EDGE
Accelerate Innovation
with CFD & Thermal
Characterization

Team Velarde Bose Huawei

Gears up for Automotive Technologies
Red Bull Air Systems Oustandingly
Race Page 47 Accurate
Page 12 Models
Page 16

mentor.com/mechanical
Free Yourself
from the Shackles of
Traditional CFD
The Democratization of CFD
democratize/ dmkrtz/
verb: make (something) accessible to everyone.

NAFEMS World Congress 2017

11-14 June | Stockholm | Sweden
The Democratization of CFD- Is it Possible? - Keith Hanna
1D-3D CFD and 3D-1D CFD: Simulation-Based Characterization - Robin Bornoff
Subtractive Heat Sink Design of Optimization Using CFD - Robin Bornoff
Frontloading CFD - Mike Gruetzmacher, and Keith Hanna
Using Modern CAD-Embedded CFD Code for Numerical Simulation of Heat
Transfer in Vanes and Blades - Andrey Ivanov

2 mentor.com/mechanical
 Perspective
Vol. 06, Issue. 01

Greetings readers! I have been thinking recently about the

Democratization of CFD, not least because our Mechanical
Analysis Division mission statement over the last three years
since I became General Manager recognizes it:

To democratize the use of thermal

simulation & test solutions to multiply the
Mentor Graphics Corporation productivity of any engineer, anywhere in
Pury Hill Business Park,
The Maltings,
the world, at any time.
Towcester, NN12 7TB,
To democratize any engineering simulation software is not an easy thing; it requires
United Kingdom
clever engineering approximations, innovative numerical methodologies, lots
Tel: +44 (0)1327 306000
of automation, plus tremendous attention to user experience and ease of use.
email: [email protected]
Nevertheless, I believe we have been pivotal in our sector in achieving unique
Editor: workflows, integrated simulation & test solutions, and 1D-3D workflows that yield large
Keith Hanna user productivity gains through a fixation on democratization without compromising
accuracy and engineering usefulness. Read more on page 6.
Managing Editor:
Natasha Antunes
The customer stories in this edition of Engineering Edge lead with Team Velardes
Copy Editor: attempt at winning the Red Bull Air Race World Championship with an aerodynamic
Jane Wade study of their plane using FloEFD. Huawei in China use our thermal simulation &
test tools, FloTHERM and T3Ster, to optimize the thermal performance of their
Contributors:
communication devices a great cutting edge application. There are great interviews
Doug Kolak, Ivo Weinhold, Jane Wade, John Parry,
and articles from two French leaders in Automotive Lighting, Paul-Henri Matha at
John Wilson, Kate Boyd, Keith Hanna, Matt Milne,
Renault and Jean-Paul Savier at ELS. Bose Automotive Systems in America are also in
Mike Croegeart, Mike Gruetzmacher, Richard
the spotlight as they use FloTHERM and our unique subtractive design methodology
Wilton, Robin Bornoff, Roland Feldhinkel
to optimize their heatsink's size and weight for almost the same thermal performance:
With special thanks to: a significant cost saving. Several great applications of FloEFD are reported including
Bose Automotive Systems Irkut in Russia (high lift wing flaps), E-Cooling GmbH in Germany (room-level power
Dr. Schneider Group electronics thermal), Dr. Schneider in Germany (optimizing automotive HVAC ducts), and
E-Cooling GmbH race car brake cooling design by a student at the Halmstad University in Sweden.
Electronic Cooling Solutions
Graz University of Technology With regard to product news, FloTHERM XT 3.0 has recently hit the market with some
Groupe Renault cool fan modeling capabilities, plus a standalone MicReD DynTIM S Dynamic Thermal
Huawei Technologies Interface Material (TIM) Tester with an integrated T3Ster unit. FloMASTER recently
Infolytica Corporation celebrated 25 years since the commercial company was founded (Flowmaster Ltd),
Innolux Corporation and a technical feature article on its new capability, Waste Heat Recovery, is outlined
Irkut Corporation herein. Finally, Jurgen Sprengle in Australia reports using FloMASTER to simulate a cool
JS Pump & Fluid System Consultants municipal domestic water supply network probably one of the biggest ever in the
Koenigsegg Automotive AB codes history and a thing of beauty to look at.
Team Velarde
ZFW

2017 Mentor Graphics Corporation,

all rights reserved. This document contains
information that is proprietary to Mentor
Graphics Corporation and may be duplicated
in whole or in part by the original recipient
for internal business purposes only, provided
that this entire notice appears in all copies. In
accepting this document, the recipient agrees
to make every reasonable effort to prevent
unauthorized use of this information.
All trademarks mentioned in this publication are
the trademarks of their respective owners.
Roland Feldhinkel, General Manager,
Mechanical Analysis Division, Mentor Graphics

mentor.com/mechanical 3
12

News Engineering Edge 34 Graz University

of Technology
6 Democratization 12 Team Velarde Mixed Flow Diffuser Pump CFD
Investigation-methods and
of CFD Gears up for Red Bull Air Race
test rig evaluation
8 New Release: 16 Huawei Technologies
Oustandingly Accurate Models 36 Koenigsegg
DynTIM S & T3Ster Regera
Booster 18 E-Cooling Sports Car Brake Cooling Simulation
Cooling Power Electronics at Room Level
8 New Release: 22 Infolytica Corporation 42 Irkut Corporation
FloTHERM XT 3.0 Vehicle Thermal Management Simulation
Commercial Aircraft Wing
High Lift Modeling
9 Mentor HEP with MotorSolve and FloMASTER

24 Dr. Schneider Group 47 Bose Automotive

10 Mentor Learning Shortening Product Development Time Systems
Center Finding new ways to make
28 Groupe Renault
11 Masters Course Driving down Automotive Headlamp Costs
our Products Cool!

Highlighting ELS
11 Harvey Rosten Award
Winners Announced

4 mentor.com/mechanical
 Contents

24 36

56
52 ZFW
Successfull Electrothermal Simulation
Study with Powertester and FloEFD

56 Electronic Cooling
Solutions
Unique Design to Generate UAV
Electrical Power in Flight

60 Organic Rankine Cycle

Camparing FloMASTER to
Experimental Data

62 Innolux Corporation
Thermal Design Approach for
Automotive Display Integration

64 JS Pump & Fluid

System Consultants
Predicting Water Treatment Plant
Performance with FloMASTER

Regular Features
20 Ask the GSS Expert
27 Interview:
27
Paul-Henri Matha, Groupe Renault

32 How To...
Characterize an Automotive Thermostat

68 Geek Hub:
Spiralled Skyline

70 Brownian Motion

mentor.com/mechanical 5
The Democratization
of CFD a Personal
Journey
By Roland Feldhinkel, General Manager, Mechanical
Analysis Division, Mentor Graphics

I
have been thinking about a phrase
that has been dear to my heart for
many years now, but has only recently
De

become one of the latest buzzwords

to hit our Industry: the democratization
of CFD. When I think of my career in
mo

CFD, I would say that it has all been about

democratizing CFD. Ever since I set up
Nika GmbH [1] in 1999 with the goal of
Analysts
producing CAD-embedded CFD software,
cra

I have had the dream of moving CFD down

the pyramid of users to the designers and
engineers in the world who can benefit
tiz

from it. These are the men and women

who want to dip in and out of CFD rather
than the full-time analysts and CFD experts
ati

who have been the main exponents and

advocates of CFD over the years. I recall
on

talking about democratizing CFD back

in 2004 during an interview for a German
CAD/CAM engineering magazine [2]. Designers
I am therefore pleased to see that for the
upcoming NAFEMS World Congress in
Sweden this June that my colleagues, Keith
Hanna and Ivo Weinhold, have dug deeply
into this subject in writing their paper The Figure 1: CFD User Pyramid

6 mentor.com/mechanical
 News

Democratization of CFD is it Possible? [3] CAD-embedding; and along with more fundamental research. The
because I, like them, believe that there are 1D fluid system modeling / 2D CFD. biggest impact on democratization will come
many strands to solving the democratization from PLM-enabled CAD-embedded CFD and
challenge and no one single factor is The table in figure 2 summarizes our cloud deployment. Moreover, user experience
delivering a solution to this problem today. A conclusions on each of these democratization needs to be front and center. Some CFD
democratization deficit is plaguing the larger factors in terms of their impact on how many Apps in key industries will have a place in the
CAE and MCAD markets too. One could extra CFD users could be expected to come democratization thrust, but the very challenge
argue that the creation of Onshape a few from the democratization factor, and how much of CFD democratization itself will be an
years ago by Jon Hirschtick is an attempt to the productivity of existing and new CFD users agent for change in our industry. It is entirely
democratize MCAD in the cloud. With CFD will be improved significantly by the factor. conceivable that one day our grandchildren
we are inherently trying to solve very complex will laugh at us when we say we spent two-
second order non-linear differential equations I believe that most engineers live in a CAD- thirds of our time in creating geometry and
through empirical models that approximate centric, PLM-enabled universe, and CFD meshes when we did fluid flow simulation with
the physics and chemistry of a given industrial software implementations have to deal with CFD!
application. There is always therefore a risk for customer workflows to make engineers as
ill-posed CFD simulations of getting the wrong productive as possible with the least amount References:
answer as much as the right answer. Unless a of physics and CFD-specific knowledge [1] Engineering Edge Magazine, Mentor
code is robust, its meshing has to capture the possible. Users want realistic simulations Graphics, Vol 3, Edition 1, May 2014, p5
relevant phenomena, and the right boundary in real-time, or as close to it as possible.
conditions and material properties are known [2] Strmender Erfolg, CAD-CAM Magazine,
Democratization of CFD is indeed possible,
(or chosen), not to mention the underlying March 2004, pp18-20
but to get from 5-10% market penetration to
quality of the CFD solver. Even after 40 years 80-90% in the next 30-40 years will require [3] Hanna, R.K. and Weinhold, I.W., The
of commercial CFD, the market is still barely more thinking out of the current box and Democratization of CFD is it Possible?,
10% penetrated; that is, the number of actual probably radical rethinking of CFD code NAFEMS World Congress, June 2017,
users divided by the total possible number structures, automation and user experience, Stockholm, Sweden.
of users of CFD, is quite small. And that
number has only doubled in the last five years
despite significant improvements in computer
hardware speed and available CPU RAM.
There are pros and cons to several solutions
being offered by the CFD Industry today to
bridge this democratization gap:

Appification;
High Performance Computing (HPC) and
faster solvers;
Multivariable design optimization;
Cloud-deployed solutions;
Open Source CFD;
User Experience and ease-of-use
enhancements;

Figure 2: Impact of Democratization Factors on CFD Penetration in Future

mentor.com/mechanical 7
New Release:
DynTIM S & T3Ster Booster
(240A/11V)
E
arly in 2017 the MicReD team, includes mechanics for thickness control, TIM measurement capability as the combined
answering market demand, a diode for heating and sensing control systems with an identical User Interface.
released the DynTIM S, combining electronics, and heating and sensing circuits. The MicReD team also updated and released
the power of T3Ster with the The in-built technology from T3Ster enables the T3Ster Booster (240A/11V) which is
unique capabilities of the DynTIM into one measurement by heating up the thermal available to order now. It enables easier
standalone unit. This brings a more cost structure, switching off the power and then measurement of IGBT half-bridges, both the
effective TIM measurement system to the measuring the thermal transient (temp. gate drivers and the heating current sources
market since there is no requirement to vs. time). The unchanged measurement are floating, not grounded. The new 240A
purchase a separate T3Ster system. control and results processing tool allows booster can heat MOSFET on open channel
quick and simple test setup and a clear and measure on body diode. This new
This new release is used for measuring bulk and intuitive interface to review and process booster can be rack mounted to become part
thermal conductivity of thermal interface the data or prepare it for reporting. This of the measurement set-up easily.
materials by varying thickness. The DynTIMS standalone system offers exactly the same

New Release:
FloTHERM XT 3.0
A
t the beginning of 2017, the cooling solution. The FloTHERM XT product Enhanced Parametric Study
FloTHERM XT product line can be used at all stages of the electronic Functionality: The addition of a Design of
released V3, and the latest design process, from design concept through Experiments scenario table allows users
version offers new functionalities manufacturing. Its CAD-centric technology and to set up a number of studies to ensure the
developed from customer feedback. This robust mesher simulates complex geometries best coverage of the design field. These
leading Electronics Cooling Simulation with ease, speed, and accuracy for improved scenarios can also be sent to remote
tool simulates thermal effects of complex product quality and reliability. machines with more capacity. The addition
geometries. Key capabilities include : of this key enhancement was based on
Key Features in the FloTHERM XT customer feedback.
Simulation of spinning parts; Release: Improved User Interface: Enhancements
Explicit Models for Spinning Parts: The to the user interface makes it easier to set
Temperature-dependent power simulation;
new sliding mesh capability easily meshes model parameters and find mesh controls
Enhanced Design of Experiments for improved thermal design predictability.
the rotation region to accurately simulate the
parametric studies functionality to determine
effects of spinning parts such as fans and Design Tool Integration: Two approaches
best design coverage; and
rotors used in electronics devices such as are available for tool interoperability in this
Thermal Territory Simulation to enhance fansinks and drones. release: Files can be directly imported
accuracy with heat transfer from Temperature-Dependent Power through the FloEDA Bridge module with
components. Simulation: When attached to an ease-of-use features such as undo/redo,
object or planar source, the FloTHERM access to power displayed in tree, and the
These features are important for electronics XT software will use the objects own graphical representation of component
devices used in the transportation, aerospace, average temperature goal to adjust the types. The second approach uses a neutral
defense, and consumer electronics markets. applied power, enabling users to capture file format allowing the user to interact with
the leakage current effects in dies. This the setup of a model outside of the standard
feature was developed from customer interface.
Based on the electronics cooling DNA of
Mentor Graphics market-leading FloTHERM feedback. FloTHERM XT is available now. For more
technology and the FloEFD CAD-centric Thermal Territory Simulation: This new information, visit the product website: www.
philosophy, the award-winning FloTHERM feature is critical for detailed copper under mentor.com/products/mechanical/flotherm/
XT product is the industrys first integrated the component designs, used to draw heat flotherm-xt/
mechanical design automation (MDA) and away from that component - providing an
electronic design automation (EDA) electronics accurate simulation methodology.

8 mentor.com/mechanical
 News

Training the Next

Generation of
Engineers
By Dr. Ivo Weinhold, University and Technology
Partnerships Manager, Mechanical Analysis Division

M
Mentor Graphics universities, colleges, and schools with HEP Mechanical Analysis
Mechanical Analysis leading edge design tools for classroom Design Package
Division has launched instruction and academic research to help The HEP Mechanical Analysis design
a new University and ensure that engineering graduates enter package ensures that engineering students
Technology Partnerships Program, into industry, proficient with state-of-the-art have access to the latest Computational
offering the divisions leading software tools and techniques. Through HEP, Mentor
Fluid Dynamics, Electronics Thermal Design,
tools for Computational Fluid Dynamics, strives to develop long term relationships
and Fluid Systems Simulation software used
Electronics Thermal Design, and Fluid with engineering schools and universities
in industry. From high school programs
Systems Simulation for engineering around the world. To date, Mentor is proud
through to higher degree university research,
education and academic research. to have partnered with more than 1,600
Mentors physical simulation software
academic institutions worldwide.
provides both the ease-of-use and technical
Engineering trends like Industry 4.0,
depth needed for all educational needs. The
Internet of Things, integrated digital Program Benefits design package contains only full capability
product development processes, massive The program helps engineering students software versions without reduced feature
product customization, and the revolution by providing simulation software, technical sets or artificial limitations. Currently, the HEP
in manufacturing and factory automation, support, training, and resources during Mechanical Analysis Package contains the
create new career paths for the next school and post-graduation and is open following simulation tools:
generation of engineers. Research and to all educational institutions. The software
development departments of large industrial is donated and levies a nominal annual FloEFD;
companies, design teams of small and support charge based on the design
medium enterprises, and innovative start- package(s) used, irrespective of the number FloEFD for Siemens NX, FloEFD for Solid
up entrepreneurs alike are demanding of licenses. HEP members have access to: Edge;
comprehensive engineering education for FloMASTER;
their innovators of tomorrow. Simulation of As many copies of the software as needed; FloTHERM XT;
physical product functionality, seamlessly
The same customer support services as FloTHERM, FloTHERM PCB; and
integrated within virtual product development
corporate customers;
processes, will be an integral part of future FloVENT.
engineering career paths, requiring sound Mentors advertised public training
expertise in using the latest simulation classes on a space-available basis, For more information and a HEP
technology. Engineering schools across the and on-demand training (for nominated membership online application form visit
globe are embracing this demand, striving to faculty staff members); and www.mentor.com/company/higher_ed/
best prepare their students for a successful Online product demonstrations and or contact Ivo Weinhold: ivo_weinhold@
professional career. Mentor Graphics' tutorials. mentor.com
Mechanical Analysis Division is committed to
support engineering education at all stages
by providing easy access to its advanced Dr. Ivo Weinhold is University and Technology Partnerships Manager for the Mechanical
engineering tools for physical simulation Analysis Division of Mentor Graphics. In this role he promotes Mechanical Analysis'
and test, allowing students to gain hands- simulation software and thermal test solutions for engineering education, helping the next
on, real-world, industry-level experience for generation of engineers to become familiar with cutting-edge technology for physical
tomorrow's simulation-driven virtual product simulation and test.
design workflows.
Prior to Mentor Graphics, Ivo worked for Flomerics, NIKA, Fluent and Fluid Dynamics
Higher Education Program International, starting his career in the CFD industry in 1993. His engineering, sales
Mentor Graphics Corporation founded and management roles through the years have included technical support, sales and
the Higher Education Program (HEP) to marketing of electronics cooling CFD software, and over 13 years Product Management
further the development of skilled engineers for the FloEFD line of CFD products at NIKA, Flomerics, and Mentor Graphics.
within the industry. The program provides

mentor.com/mechanical 9
Mentor Learning
Center
Interview with Lane Giles, Director of Education
Services, Mentor Graphics

Q. What is a Mentor On-Demand 1. View Mentor experts demonstrate concepts classes, as soon as that class ends, they
Training library? and features in the live software; begin forgetting what they learned. The
A. Mentor On-Demand training libraries 2. Practice in fully licensed virtual lab industry estimates that students forget
provide customers with a full featured library environments without using a license of their around 70% of what they learn in a class,
of content that covers material delivered installed software; depending on how quickly that can apply
in Instructor led training. Students watch the learning. Mentor On-Demand training
live topical videos by Mentor experts, 3. Search for answers to technical questions at enables enterprises to send everyone to
practice in secure licensed online labs, and their moment of need; class on the same day at the same time.
assess their proficiency with knowledge 4. Discover new features as the library content For example, if you have 400 engineers
checks and assessments. Each learners is updated in real-time; scattered across divisions and locations,
activity is tracked and progress completion 5. Learn about new software release features access to the library is simultaneous.
certificates can be printed from their profile. without needing to purchase another
training course or subscription; Q. How old is the Mentor Learning
Q. How long can I access the videos and Center and how has it been accepted?
6. Demonstrate Mentor proficiency with
labs in the library? A. The MLC initially launched with the first
completion certificates; and
A. On-Demand training is provided library in late 2015. Thanks to Mentor's
to individual users on a 12 month 7. Use the library as a reference to keep their unique development methodology, and
subscription. Although a learner could go knowledge fresh and current. seasoned subject matter experts, the MLC
through the materials for their unique role in now has a very robust and active publishing
roughly a week, they can refresh knowledge Q. Why should an enterprize consider cycle. The content that is added during an
and practice design work in the virtual labs Mentor On-Demand training? annual subscription keeps users engaged in
during the annual subscription. A. Enterprizes with 10s or 100s of design the learning process and has helped 1000s
engineers rarely have the budget, time, of users across 55 countries become more
Q. What are the benefits of the library or capability to schedule these individuals efficient and productive with Mentor tools.
subscription? into traditional training classes in a timely People are saving hours of development
A. Learners benefit from the annual manner. Furthermore, if they were able time due to proper use of the tool and its
subscription in several ways. They can: to schedule and send engineers to functionality.

10 mentor.com/mechanical
 News

Unique Embedded Lighting

Systems Masters Course in Paris
T
he worlds first Advanced Masters
degree in Embedded Lighting
Systems (ELS) at the prestigious
Institute dOptique in Paris, at
ESTACA Ecole dIngnieurs and at the
State School of Design in Paris is now
entering its third year. With a select annual
intake of twenty engineers and scientists
from across the world, and led by ELS
Chair Jean-Paul Ravier, it seeks to cater
for the rapidly expanding market for car
lighting design and engineering which is
growing at 7.5% annually. Jean-Paul Ravier, ELS Chair

knowhow to his new academic role with and a science. We have designers who are
It is estimated that there are around 50,000 over forty years of experience in the lighting having to do engineering to see if their ideas
R&D engineers involved in automotive lighting sector, thirty of those at Valeo S.A. And having are feasible; and we have engineers who are
systems catering for over 1,000 car models secured sponsorship from Renault, Valeo, appreciating the aesthetics of good design to
around the world today. Historically, automotive PSA and Automotive Lighting GmbH, ELS apply their technical knowledge to it. You have
lighting evolved slowly over its first 100 years, selected Mentor Graphics and its FloEFD to be multidisciplinary to operate in this new
but in the last decade it has seen an explosion product line to help students understand how paradigm.
in interest with the advent of Light Emitting an advanced CAD-embedded 3D CFD tool
Diodes (LEDs) and more recently laser lighting. can be used to thermally optimize the design ELS aims to stay ahead of technology trends
Increasingly, function and styling are becoming of lighting assemblies. such as ever more complex lighting systems,
intertwined in automotive lighting design embedded sensors in head and taillights,
as head and tail lights become aesthetic Commenting on the automotive lighting lighting for autonomous cars, the advent of
signatures for both car brands and individual revolution, Mr Ravier notes that Modern OLEDs (Organic LEDS), and Adaptive Driving
models in ever more complex design situations. automotive lighting is developing dramatically. Beam (ADB) systems. For more information
Headlights are seen as the eyes of the car. on the ELS course, visit www.embedded
Mr Ravier brings tremendous practical Their design is somewhere between an art lighting.com

Mentor Graphics Team Receives the

Harvey Rosten Award for Thermal
Heatsink Optimization Methodology
S
EMI-THERM 2017, announced underperforming portions of a heatsink to
that Dr. Robin Bornoff, Dr. John save weight and cost without compromising
Parry and John Wilson, a team overall thermal performance. This method
from Mentor Graphics Mechanical using Mentor Graphics FloTHERM
Analysis Division, received the Harvey technology provides a variety of automated
Rosten Award for Excellence in thermal optimization approaches to gain deeper
modeling and analysis of electronics. The insights into thermal characterizations to
team received the award for their paper, determine the best thermal design solution.
Subtractive Design: A Novel Approach to
Heatsink Improvement, at the 33rd annual We are extremely proud of our team for their
IEEE Thermal Measurement, Modeling and commitment and dedication in discovering
Management Symposium (SEMI-THERM) ways in which heatsinks may be optimized,
in San Jose, California. stated Roland Feldhinkel, General Manager thermal experts is a tremendous honor,
of Mentor Graphics Mechanical Analysis and particularly personal since this award
The Mentor Graphics team created a Division. To be recognized by the selection is named after the co-founder of Flomerics,
unique methodology to sequentially remove committee consisting of highly esteemed which Mentor Graphics acquired in 2008.

mentor.com/mechanical 11
Aces High for
Team Velarde
FloEFD explores external aerodynamics for
Team Velarde in the Red Bull Air Race World
Championship
By Matt Milne, Application Engineer, Mentor Graphics

I
don't know about you but I'm competitive against the other more
partial to a bit of Iron Maiden experienced pilots.
now and then. And if there's a
more fitting, more intense, higher It is often said in racing that to stand
energy soundtrack than "Aces High" to still is to go backwards. With this in
accompany the Red Bull Air Races then mind, Team Velarde continued
I want to know about it! Established working hard through the
in 2003, the Red Bull Air Race World off-season to make
Championship sees pilots from around their goal a reality.
the world compete against the clock "Regarding the
as they navigate their aircraft through a race plane,
challenging obstacle course in a breath- we are
taking combination of high-speed low-
level flying.

2017 sees Team Velarde begin their third

year in the championship. Pilot Juan
Velarde is looking to build upon his
impressive 2016 record where
he took the top spot
in qualifying in one
round and proved
he could be

12 mentor.com/mechanical
Aerospace

mentor.com/mechanical 13
aiming to have a new set of winglets and
a few other aerodynamic improvements
ready by the second race of the season,"
explained Velarde. "We are working with
a team of engineers to optimize the
performance of the plane," he added.

Team Velarde approached Mentor Graphics

for help investigating the aerodynamics of the
race plane. Specifically they wanted to use
CFD to investigate the performance benefits
Figure 1. CL vs. (Full span model) Figure 2. CD vs. CL (Full span model)
of the new winglets. Working with CAEsoft
in Madrid, Team Velarde used 3D scanning
technology to create a 3D CAD model of the but it still had to be done manually. And A new simulation was set up using the CAD
aircraft which was then provided to Mentor then came meshing. For conceptual geometry supplied by Team Velarde. The
Graphics for CFD analysis. models with a relatively clean shape some time taken to prepare the model, generate
degree of automation was possible, but the mesh and begin the first flow calculation
While modern 3D scanning techniques are for more complex models meshing had to was less than one hour. Furthermore,
without question impressive, it turns out that be completed manually. I remember clearly our previous work on a wide range of
the technology does a rather better job than some cases where meshing alone took more well-known external aerodynamics test-
some might desire for CFD analysis. Close than a month. cases meant we could follow a set of well
inspection of the CAD model revealed a documented best practices for this class
number of excrescences and other lumps This is all in stark contrast to the situation of simulation and be confident of obtaining
and bumps. For traditional CFD, such today with FloEFD. Put simply, the consistent and reliable results.
features might be the cause of considerable SmartCell meshing technology employed
pain, requiring time and effort to modify the by FloEFD addresses meshing challenges Initial calculations were run at low-medium
CAD geometry to a state where it is suitable with minimal user effort. Regardless of angle of attack (AoA) at M=0.25, comparing
for meshing. However, with FloEFD these whether the geometry is a clean conceptual the original clipped wing and revised
features are simply not a problem. model, or dirty with lumps, bumps and "winglet" designs. These calculations were
excrescences, FloEFD can mesh it. With run using the complete aircraft span (i.e.
During my time in aerospace research, a less time spent on geometry preparation and port and starboard sides) but omitting the
lot of effort was spent working on the CAD meshing, the user has more time to think propeller blades. It was assumed that the
geometry to get it into a suitable state for about the results, what they mean and how propeller wash has only an incremental
meshing - the process was well developed to improve performance. effect on the overall wing performance and

14 mentor.com/mechanical
 Aerospace

that since the wing leading edge is unswept

any interaction between the propeller and
wingtip flows is negligible. Each calculation
used approximately 1.9 million cells and took
around two hours to run using 12 cores (Intel
Xeon E5-2643 v3).

It can be seen from Figure 1 that the

"Winglet" design yields a slight increase in
lift curve slope (dCL/d). Figure 2 shows
that there is also a reduction in drag. Figure 4. CL vs. (Half span model) Figure 5. CD vs. CL (Half span model)
These predictions are expected from basic
aerodynamic theory, since both designs dimensions used to non-dimensionalize optimization and further work in this regard is
effectively increase wing aspect ratio. the integrated forces and moments. Each recommended for the future. Based on these
calculation used approximately 5.6 million findings, the team are currently working on
Further simulations were run to quantify cells and took around seven hours to run modifications to the race plane.
performance of the "Winglet" design at using 12 cores (Intel Xeon E5-2643 v3).
medium-high AoA at M=0.25. It is noted that The first race of the 2017 Red Bull Air Race
this area of the flight envelope (i.e. accurate Predictions of integrated forces are shown World Championship took place in Abu
prediction of onset and development of flow in Figures 4 and 5. As expected the lift Dhabi on 10-11 February. After setting an
separation and associated effects) continues curve slope remains linear at low-medium impressive fifth fastest time in qualifying,
to be a challenge for all CFD methods today. AoA before becoming non-linear as flow Team Velarde took a sensational second
While validation against well-established separations develop. A thorough investigation place finish in the race, a best ever result
aerospace test-cases has shown FloEFD to of the predictions obtained using FloEFD is for the team. The second race takes place
perform acceptably in this regard, caution ongoing. This will give more insight into the in San Diego on 15-16 April. With luck,
is nevertheless advised when interpreting development of the flow topology, and hence FloEFD will prove to be the ace in Team
computed predictions in this part of the flight ways in which the "Winglet" design might be Velarde's hand and help them on their way
envelope - the ability to obtain plausible optimized to improve performance. to success in the 2017 Red Bull Air Race
looking solutions is no guarantee of accuracy. World Championship.
The improvement from the original clipped
To increase computational efficiency, only the wing to the "Winglet" design is expected and More information:
starboard half of the aircraft was simulated. indeed is predicted by basic aerodynamic www.juanvelarde26.es
The effects of asymmetry in the model were theory. Performance of the "Winglet" design www.redbullairrace.com
assumed to be negligible and the propeller has been quantified for a wider range of
blades were again omitted. Appropriate AoA. Further improvement of the "Winglet"
revisions were made to the aircraft reference design may be possible through the use of

mentor.com/mechanical 15
Huawei Delivers
Outstandingly
Accurate Models
By Yake Fang, Senior Engineer, Huawei Technologies Co., Ltd.

P
ackaging high-performance
multi-core IC devices used in
communication applications
is a key challenge for both
manufacturers and system integrators.
Traditionally a System-in-Package (SiP)
has been taken, with chips mounted side-
by-side, allowing differing semiconductor
Core 2
technologies to be mixed. More recently
stacking silicon die has become more Core 3 Core 1
commonly used, as volumes have
increased, making the System-on-Chip
(SoC) design approach, which has
much higher upfront non-recoverable
expenditure, the cheapest option overall.
These multi-heat source devices require
very careful thermal design frontloaded in Figure 1. Overview of SoC device construction
the design process, due to the very high
cost of changes during detailed design. suppliers rather than rely on their thermal transient simulation to get a baseline result for
data, which is often not sufficiently accurate this, as yet, uncalibrated model, which we built
The increased functionality and power or suitable for design. As it is based on the using the best available information during the
dissipation has made the thermal design of thermal transient measurement principle, design. The temperature vs. time graphs we
communication device applications crucial in giving the highest available fidelity, we turned got from FloTHERM and T3Ster seemed to
todays increasingly high computing power to Mentor Graphics T3Ster hardware to match quite well.
communication device industry. These perform a series of thermal tests so our thermal
products present a very challenging thermal engineers could better understand the thermal Using a model of the package in the same
environment for our SoC packages, with the impedance characteristics of these high- thermal environment in FloTHERM allows
need to design to very tight thermal tolerances performance products, and reduce time and the temperature vs. time response from the
to meet size, weight and form factor goals. resources needed to achieve a good design. FloTHERM simulation to also be converted
into a Structure Function, and then the two
Huawei always pays attention to thermal As a test vehicle for this new design workflow Structure Functions compared. This is a plot
challenges in product design and uses thermal we chose to investigate a 3-core SoC device of the cumulative thermal capacitance vs. the
analysis throughout the whole product R&D typical of those now used in communication cumulative thermal resistance that the heat
phase. We have the ambitious program to device applications and assembled in a experiences as it leaves the package. If the
develop an internal thermal design flow for Package-on-Package (PoP) format that allows model exactly matches reality for the package,
multi-core SoC and SiP devices that would package stacking. The FloTHERM model is then the two Structure Function curves should
enable us to use the highest possible quality shown in figure 1. match perfectly, as the heat flow paths from
thermal data and models to facilitate our own the die junction out to the environment should
design efforts. As soon as the first working packages were be identical.
produced, they were measured using Mentor
We have used FloTHERM for almost 15 Graphics T3Ster hardware, which records the We then post-processed this temperature vs.
years, so we have a rich depth of experience temperature response of the part to a power time data to create a Structure Function graph
and so are able to create detailed thermal step from just a few microseconds until steady- for the T3Ster and FloTHERM results, and
models of our SoC and SiP products. state is reached, with an accuracy of 0.01C. compared these. We were not expecting an
However, we needed a way to test that these exact match, as there are always uncertainties
packages achieved the as-designed thermal Having creating the detailed numerical model surrounding material thermal properties.
performance, and to test packages from other of the package in FloTHERM we performed a The thickness and uniformity of bond lines,

16 mentor.com/mechanical
 Electronics Cooling

"To design successful communication devices, Huawei first performs thermal

simulations in FloTHERM, and then calibrates the detailed thermal model based
on accurate thermal measurement data obtained from T3Ster. In parallel with this
Huawei optimizes the design through different material selection to decrease thermal
resistance, confirmed using T3Sters transient thermal measurement. This process
is much faster and requires fewer resources than we needed before, and provides
Huawei with a highly-accurate thermal model to use in our system-level modeling,
helping us get products to market faster than our competitors."
Fang Yake, Senior Engineer, Huawei Technologies Co. Ltd.

and contact resistances within the structure

where one material is not in perfect thermal
contact with its neighbor materials. However,
we were somewhat surprised at how big the
differences were when we compared Structure
Functions, as shown in figure 2. The Structure
Function comparison is a very powerful tool
for showing up differences between the model
and the experimental result, and where those
differences are in the heat flow path.

Figure 2 shows the Structure Functions

for the initial FloTHERM simulation (black)
vs. the T3Ster measurement data (blue),
shown here when Core 1 is powered. The
two Structure Functions are clearly quite
different. At the far right of the plot the thermal
capacitance becomes infinite, as heat reaches Figure 2. Initial simulated curve (black) vs.curve measured with T3Ster (blue)
the environment. For the FloTHERM model
this occurs at a lower thermal resistance
than found in practice, so the model is
underestimating the temperature rise. Some
of this difference may be due to deficiencies in
the way the cold plate is modeled. However,
the differences seen on the left of the graph
are within the package model itself, and close
to the heat source. Simulation vs. test results,
when cores two and three are powered, both
showed similar discrepancies.

To modify the model manually to bring it in

line with the experimental results would be
very time consuming and error prone as it
would be a guess-and-correct process that is
impractical to undertake a fast-paced design
environment like ours. However, part of our Figure 3. Final Calibration Result Showing Excellent Match
reason for choosing T3Ster is that its results
can be read into FloTHERMs Command within the package, ignoring the part of the The result, shown in figure 3, gives an
Center tool, which allows automatic calibration curve that represents heating of the board excellent match within the thermal resistance
of multiple user-selected parameters to and the cold plate, so truncating the transient range up to the junction-to-board thermal
adjust the model to match the measurement calculation to 0.15 s. The optimization requires resistance (RthJB). This has proved to us the
data. This calibration gives a very good fit, a list of parameters, the exact values of which effectiveness of FloTHERMs auto calibration
something that would be extremely difficult are uncertain, and a possible range for their technology coupled with T3Ster, and the
and time consuming to do manually, and so thermal conductivity values. The parameters calibrated model can be used for board-level
results in a model that has very high accuracy selected were the C4 bumps, with a and system-level thermal analysis with a very
for us to use in system design. conductivity ranging from 40 W/mK to 70 W/ high degree of confidence in the fidelity of the
mK; the underfill (0.2 W/mK to 0.7 W/mK; the results.
To do this automatic calibration, we selected package substrate (10 W/mK to 20 W/mK);
the time range corresponding to heating and the solder mask (0.5 W/mK to 3 W/mK).

mentor.com/mechanical 17
E-Cooling: Cooling Power
Electronics at Room Level
By John Wilson, Technical Marketing Engineer, Mentor Graphics

T
he conversion of electrical
power with power electronics
has become a key technology
that enables many of todays
innovations for electric vehicles as
well as electricity production. Though
the applications vary greatly there are
two features they have in common:
the electric components generate a
significant amount of heat, and are
trusted to operate reliably. A key
factor in the reliability of electronics
is the operating temperature. To
maintain the operating temperature
within specifications with a high heat
dissipation, a proper thermal design
must be developed.

Challenges
The market of inverter solutions for
photovoltaic is growing. The Middle East,
Africa and India are expected to massively
invert in photovoltaic. In these regions,
an average summer's day temperature
of 50C is usual. The customers require
reliable inverters integrated in enclosures
or containers without air conditioning. As Figure 2. The enclosure with its electric Figure 3. Streamlines starting from fan
the cooling air takes away the electrical components and fan colored with the velocity
losses, its temperature usually exceeds the
environment temperature. Dust or sand
entering the electric enclosures must also
be avoided. Therefore, filters are necessary,
despite large pressurelosses.

E-Cooling, an engineering consultancy

firm based in Germany, has expertise in
the thermal design of Power Electronics,
Transformers and Chokes at the component
level or at room level where the computational
domain enhances a complete enclosure,
container or room. At room level, it must
be ensured that the cooling path provides
adequate airflow to all critical components.

Solution
E-Cooling performs thermal and airflow
design with FloEFD from Mentor Graphics.
One of their recent designs involved the
overall cooling of a 2.2 MW frequency
converter enclosure. Such a frequency
converter could be used for a wind power
station to provide the grid with the correct
frequency or for an electrical drive in order Figure 4. Frequency Converter Fans

18 mentor.com/mechanical
 Power Electronics

to control the rotation speed. This power is

reached by two frequency converter boards,
each has three IGBT power modules and
two dedicated axial fans. Each board
dissipates 1,350 Watts.

When integrating electronic systems into

an enclosure, it is important to determine
if there is enough airflow through all critical
components. This requires the accurate
representation of all the significant flow
obstructions as well as the fans. Here each
fan as well as the dust filter were modeled
with characteristic performance curves
(pressure vs. volume flow) provided by the
manufacturers. With the analysis, E-Cooling
is able to determine the flow distribution
within the enclosure and make any design
modifications to allow the neccesary airflow
for each component, including the use of
alternative fans.

In the lower portion of the enclosure the

model detail was not compromised. A
detailed representation of the choke and
auxiliary transformer was included as well
as the flow diaphragm flow upwards and
the exhaust grid flow downwards. A critical
aspect of the design is to ensure that a large
amount of the flow is going through the tiny
passages between the windings. This can
only be realized by obstructing the flow with
a diaphragm or innerwalls.

Benefits
Designing airflow and cooling solutions
at the room level for power electronics is
essential for providing reliable products to
the market. During a development project
the electrical requirements and losses Figure 5. Enclosure Fan
usually change and components are likely to
be replaced. With FloEFD embedded in the
MCAD program, the automatic Cartesian
meshing, and the possibility to start a
calculation with an initial field, E-Cooling
can update the thermal and flow model very
quickly to meet project datelines.

The most viable method of airflow design

prior to testing is with CFD simulation
based design software such as FloEFD.
FloEFD allows E-Cooling to quickly build
and assess the viability of an airflow and
thermal design. Karim Segond, Consulting
Engineer, E-Cooling.

E-Cooling in Berlin is an engineering

consultancy founded by Karim Segond.
Their expertize lies in providing 3D thermal
and flow analysis, enhancement and
development supporting electronics, electric
engines, and power electronics.

Figure 6. Lower portion of the enclosure

mentor.com/mechanical 19
Ask The GSS Expert

Excel export
function
By Kate Boyd, Application Engineer, Mentor Graphics

T
he ability to run FloMASTER via
a third party application, such as
Excel, has long been recognized by
users as an invaluable productivity
aid. It streamlines the post-processing effort
and opens up the power of the FloMASTER
solver engine to engineers without previous
experience of the software or who use Figure 1. Input Variable Parameters
FloMASTER infrequently. An experienced
user of FloMASTER can make and set up Figure 5. Excel GUI Automation spreadsheet
a model, then create an Excel spreadsheet
front end through which other users can
access the model and run simulations in a
familiarinterface.

In V8.0, accessing this facility has been

considerably streamlined: a simple export
button in conjunction with the use of
the existing variable parameters tool
automatically creates spreadsheets which
can be used to run and post-process
FloMASTERsimulations.

The export functionality is accessed from

the experiments tab of the Network View Figure 2. ECS System Figure 6. Excel Options
window. (Figure 1)

The result is a spreadsheet which will run

the variable parameter study outside of the
FloMASTER GUI. The spreadsheet can be
run directly with a FloMASTER solver licence, Figure 3. Output Variable Parameters
thus freeing up a GUI license for other users to
increase productivity.

For example, take a simple ECS system

containing a Compressor, Heat Exchanger
and Turbine as shown in figure 2. The aim
of the model is to examine a range of input
conditions and see how the established
system reacts, like a virtual test bench.

The system shows a simple arrangement of a

turbine and compressor powered by the same
shaft (modeled using controller components)
and a heat exchanger using RAM air as
acoolant.
Figure 4. Export to Excel from the Experiments tab Figure 7. Excel Add-ins
The input variables to be controlled are the
Hot Air Inlet Temperature, Shaft Speed, the Experiments tab or by simply typing the Curves and surfaces,
RAM Air Mass Flow Rate, and Hot Air Inlet value in square brackets into the input field of Scripts,
Pressure. The variables to be measured are the component.
Convergence and tolerance criteria,
the Outlet Temperature, the Power Output,
and the Heat Duty of the heat exchanger. Input parameters are not limited to just simple Materials, and
Creating the input parameters can be done on real values; they can also be set for: General Data.

20 mentor.com/mechanical
 FloMASTER Traditional Excel
Excel Export COM
Running simple experiments and post
processing results in excel

Steady state parameter investigations

Modelling What if Scenarios

Transient parameter effects

Large numbers of parameters and results

Table 1. Strengths of the FloMASTER Excel Export and Traditional Excel COM

Figure 8. A successful run of experiments

Creating output parameters is best done

by highlighting the result in question
Figure 9. Login details
and right-clicking to bring up the create
output parameter menu. However, Next the values for Input Parameters must
output parameters can be created on the be entered. The units of the inputs and
Experiments tab as well. outputs in the Excel sheet are the same as the
selected unit set for the Logon Project in the
After all of the input and output parameters FloMASTERGUI.
have been created and the model has been
validated, it can then be exported to Excel by Once the input parameters have been entered
clicking the button in the upper right corner of for all the desired experiments, the simulations
the Experiments tab. can run in batch mode by clicking the yellow
Run Experiments button.
An Excel window will appear with all of the
Figure 10. An unsuccessful run of experiments
input and output parameters already filled The results of the simulation(s) will be
in. By default, it is in a read-only state; it displayed in the lower half of the Excel
is recommended to save it on ones local spreadsheet in the results output window. Q: I received the error message: "No
machine at this point. project Path Provided on Experiment
It should be noted that the new Excel Sheet".
Since this is the first time running the export is not intended to replace the A: The field Logon Project value cannot be
automation, it may be best at this point to current functionality already in place in the blank. The Logon Project is the name of the
create a folder which automatically allows FloMASTER API. Both of the methods of folder in FloMASTER which will be logged in to
macros. More information on this can be linking to Excel have their own set of benefits. (typically called FloMASTER).
found in the user guide. This is not always
required; however it is invaluable for the The main difference between traditional Q: I have entered correct values for my
convenience of the users. COM and the Excel export is the flexibility of input parameters, but I can only see
traditional COM scripting. Any component 'Result IDs' and not 'Output Result' values.
Next the FloMASTER Add-in for Excel must input or result can be added with a few A: The simulation failed for the network. This
be activated. First, click the File tab, then go lines of code in a COM script, whereas in can be caused by a number of reasons.
to Options and click the 'Add-ins' category. In the export function, the user would have to Contact FloMASTER support for assistance.
the Manage drop-down, select Excel Add-ins create the parameter in the Experiments tab in
and click Go. FloMASTER and then export the model again. Q: Can I run the Experiment if some of the
input parameters values are not set?
The Add-Ins dialog will appear. There are a few common trouble shooting A: It is mandatory to add values for all of the
issues which can easily be resolved: Input Parameters. The current experiment will
Click Browse and browse to the location not generate results if any input parameters
of the FloMASTER application (product Q: I received the error message: "Create are missing.
of Mentor Graphics) directory. Select failed with exception: Could not load
FloMASTER.Addin-packed.xll and make sure file or assembly 'Flowmaster.Interfaces, Q: Can I run more than ten experiments?
that the checkbox is checked for the Add-In. Version=9.4.1.0, Culture=neutral, PublicK A: Yes. By default ten experiments are shown
Click OK. Once this is set up, the spreadsheet eyToken=a622c4c62ff70785' or one of its in Excel, however further experiments can
is ready to run. dependencies. The located assembly's be added by selecting the Input Parameters
manifest definition does not match the and Output Parameters columns, then
In order for the simulation to run, much like assembly reference. (Exception from dragging Excel's fill handle to create as many
in the GUI version of FloMASTER, login HRESULT: 0x80131040)". experiments as required. The number of
information must be given. Fill in the login A: The Excel.exe.config file is not at the right Input Parameters and Output Parameters
information into the appropriate boxes. Make version. The correct Excel config file will need columns should be equal for all simulations
sure the correct User Name, User Password, to be copied from the FloMASTER V8 folder tocomplete.
and Logon Project are entered. into the Office folder.

mentor.com/mechanical 21
Effective Vehicle
Thermal Management
Improving Vehicle Thermal Management
Simulation with MotorSolve and FloMASTER
By Doug Kolak, Technical Marketing Engineer, Mentor Graphics

W
hen looking at system
simulation in the automotive
industry there are many
areas where computer
aided engineering (CAE) can be deployed
effectively. One of the most valuable
systems on the vehicle to analyze is the
thermal management system. Traditionally
this importance has been derived from
the requirement to keep the internal
combustion engine at a reasonable
operating temperature. By doing so it
would ensure not only that the vehicle was
operating as efficiently as possible with
minimal pollution, but critically that it would
not overheat and experience significant
thermal damage.

Today, as automotive manufacturers continue

to shift towards additional electrification
of the vehicle, both in the powertrain with
hybrid and plug-in electric vehicles and in the
controls and displays, the need for thermal
management increases. FloMASTER is
often used for this type of analysis because
at a system level, it is able to account for all
of the different components in the various
systems such as the engine, thermostat,
pumps, heat exchangers, fans, batteries, and
power electronics. With all of these systems
interacting through the flow of under-hood Figure 1. 3D Motor Temperature Result
air, FloMASTER is able to simulate how the
different systems will potentially affect anothers calculations and a unique finite element to configure the cooling setup for the motor.
thermal performance. One of the newer analysis engine, allowing designers to quickly The 2010 Prius uses a partially submerged
components of these EVs and HEVs, that prototype motors and generators virtually. The cooling system that removes the heat through
now need to be included in the overall thermal purpose of this study was to incorporate the the outside enclosure. In MotorSolve, the
management of the system, are the electric motor and the thermal load that it generates cooling was set up as a submerged spray
motors. As with other components in a system into a full system level hybrid-electric vehicle type cooling with the number of jets calibrated
level model, it is the thermal performance of thermal management model. to match the data available. This resulted in 32
these components that can sometimes be nozzles each with a diameter of 2.5 mm and
challenging to find. Motor Model an oil velocity of 0.745 m/s. The engine oil that
For the study, the 2010 Toyota Prius was entering the cooling system was assumed to
One of the most popular ways of obtaining selected since it had full data available for be at 25C.
this type of information, especially early on which to build and calibrate the model. When
in the design process, is to characterize configuring the geometry of the model in For the characterization, two analyses need
performance virtually. In this case a finite MotorSolve, the rotor geometry was imported to be run, first of which being a 3D thermal
element tool called MotorSolve produced by using a dxf file and the stator lamination analysis. Since a motor is symmetrical, only
Infolytica Corporation. MotorSolve simulates was already part of the built-in library so the one section of the model needed to be
machine performance using equivalent circuit process was very quick. The next step was considered as shown in figure 1. The analysis

22 mentor.com/mechanical
 Automotive

Figure 2. Motor Efficiency Map Figure 3. Motor Cooling Figure 4. Full FloMASTER Vehicle Thermal Management Model
Loop in FloMASTER
is very fast, taking only a few minutes, but
providing several key factors for use in
FloMASTER. First it calculated the overall
thermal resistance which is used to calculate
the heat transfer coefficient. Second it
calculated the total thermal capacity of the
motor and housing. Because there is a
combination of parts in a motor made of
different types of metals, this total thermal
capacity is converted for use in FloMASTER
by calculating an equivalent mass of steel.

The second analysis run was to create

efficiency maps for the motor. There are
several factors that can affect the efficiency
of a motor but for this study three were
considered: motor temperature, speed,
and torque. To populate data for the
characterization, the temperature was Figure 5. FloMASTER Motor Temperature Result
limited to two data points, one at the cold
motor based on the efficiency maps from Configuration 1 has a separate cooling
ambient and one at the pseudo steady-state
MotorSolve and the motor speed and loop for the motor directly after the power
temperature from the thermal analysis. For
torque from the drive cycle, electronics,
speed, a range of values from 0-5000 rpm
was used, and for torque values from 0-250 A controller specifying the motor speed Configuration 2 combines the motor and
Nm. The analysis accounts for all losses of based on the drive cycle, power electronics into a single cooling loop,
the motor including friction, windage, Joule A controller specifying the motor torque and
heating, eddy current, and iron and stray. It based on the drive cycle, Configuration 3 places the motor cooling
produces a surface map of efficiency versus A gauge reading the temperature of the loop at the front of the under-hood air path.
speed and torque at each temperature. motor and feeding it back for calculation of
(Figure 2) the thermal load, and Figure 5 shows the results of the motor
temperature for each of the configurations.
A controller calculating the thermal load
System Model Overall the temperature is lowest for
based on the motor speed, torque, and
When looking at cooling the motor from the configuration 3 since it is the first system in
temperature.
system perspective, the first thing to consider the cooling pack and thus has the coldest
is how to represent the oil circuit that is air. Configurations 1 and 2 have a similar
This motor cooling sub-system can then be
dissipating the heat from the motor. One way rise at the beginning of the drive cycle but
incorporated into the larger vehicle thermal
to do this is by constructing a closed loop configuration 2 levels are 8-9C cooler due to
management model. With the setup previously
circuit comprised of a reservoir for the fluid, a the positioning, even though it is in the same
described the most likely place to connect this
heat exchanger, a pump, and a component loop as the power electronics. FloMASTERs
sub-system to the large model is through the
to represent the motor. For this study the versatility as a system level thermo-fluid
heat exchanger since it will likely share under-
motor is represented by a collection of solver, enables a wide range of designs to be
hood airflow with the other sub-systems in the
components (figure 3): considered in the vehicle thermal management
model. (Figure 4)
area. As these systems become more
Thermal bridge to calculate the amount complex, it is important to identify where to get
of heat transferred from the motor to the
Case Study
performance data from, such as MotorSolve to
Adding the motor cooling circuit means the
cooling oil, ensure accurate system simulation results.
model can now investigate how cooling the
Point mass to represent the thermal inertia motor will affect other parts of the vehicle
of the motor, and vice versa. As an example, three different
Heat source to apply a thermal load to the configurations were explored:

mentor.com/mechanical 23
FloEFD helps
Dr. Schneiders mission:
To make the Car the
Best Place in the World
By Enrico Lorenz, Research & Development, Dr. Schneider Unternehmensgruppe GmbH

D
r. Schneider, a successful bezels, covers and trims, grills and
family-run business, located windowframes.
in Kronach-Neuses in
Germany, supplies and As the automotive market expands
develops vehicle plastics components and grows, product development times
and systems for renowned automotive are dramatically reduced. For many
producers all over the world. The projects even the use of prototyping is
entire process chain is handled within no longer an option, making the need
the group, from research and design for reliable, fast and accurate simulation
to prototype construction. (Figure 2) tools essential to overcome these
Dr. Schneider has more than 3,900 challenges. Furthermore, frontloading
employees and 90 years of experience the Computational Fluid Dynamics (CFD)
in the plastics field, 60 of which are in simulation into the product development
the automotive industry. phase is crucial to the success at Dr.
Schneider. Mentor Graphics FloEFD
Dr. Schneiders primary products are visible tool is used mainly for the simulation
parts for the car interior: air vents (figure 1), of the air vents, one of Dr. Schneiders
stowage boxes, kinematic components, coreproducts.

24 mentor.com/mechanical
 Automotive

Figure 1. Extract of Dr. Schneiders products: air vents

"With FloEFD we can shorten product development time, quickly find

appropriate design variants, save costs and explain and present the flow
behavior of our products to our colleagues and customers."
Enrico Lorenz, Research & Development, Dr. Schneider Unternehmensgruppe GmbH

As FloEFD can be directly embedded

into Catia V5, CAD models were used
directly in the program with no need for
additional geometrical preparations. Airflow
distribution is investigated at a very early
stage and must adhere to strict customer
targets and specifications. For the example
shown in figure 4, the influence of a bezel
was investigated. The two variants of this
broadband vent, with and without bezel,
were simulated. The broadband vent
variant including the bezel shows a better
distribution of the airflow. The flow angle
into the lower area is enlarged. Additional
experiments in the lab using fog to
visualize the air flow confirmed the FloEFD Figure 2. Todays product development process
predictions. (Figure 5)

Enrico Lorenz, Simulation Engineer, explains:

We make up to 100 FloEFD simulations
per month per license. We need fast, robust
and accurate simulation results. In addition
we need an efficient way to create the result
reports and presentations. In combination
with the batch run functionality which allows
automatically runs of many variations,
the pictures for comparing the defined Figure 3. Driver, broadband and rear vent

mentor.com/mechanical 25
"FloEFD predicts what
we need, so it helps us
to make the car the best
place in the world."
Enrico Lorenz, Research & Development, Dr.
Schneider Unternehmensgruppe GmbH

results items are created automatically. The

influence of design changes on the fluid
flow behavior can immediately be identified.
(Figure 6)
Figure 4. Broadband vent and airflow

One of the major tasks is creating MS

PowerPoint presentations for internal
purposes and for customer presentations.
This might take more time than the
simulation itself. For this reason, a VBA
code is generated by the Dr. Schneider
engineers which automates this work in
MSPowerPoint.

With FloEFD embedded in Catia V5,

Dr.Schneiders engineers are able to meet
todays requirements of short development
times and fulfill the strict targets of
automotive manufacturers.

Figure 5. Comparison of the broadband vent without and with bezel

Figure 6. Cut plots showing the velocity for different variations

Figure 7. Airflow distribution in the car interior

26 mentor.com/mechanical
 Interview

Paul-Henri Matha, Lighting Expert,

Groupe Renault Figure 1. Renault Laguna Headlight assembly thermal analysis (left) in FloEFD and airflow pathline details (right)

Q. Tell our readers a bit about Groupe the size of the headlight assembly package, Q. There are a lot of new technologies
Renault? such that it can be several centimeters shorter coming to automotive lighting what do
A. Groupe Renault is a long established than before, for instance. It is essential to you see as the most significant?
45bn/yr French multinational automobile manage costs early in the design cycle with A. If I look to the horizon we all see electric
manufacturer that makes a wide range of preferably multiple loops to optimize designs. vehicles and autonomous driving coming to the
cars, and vans (no trucks). It is a top ten OEM Questions we might want to ask include If you industry plus laser lighting and OLEDs (Organic
supplier and an international alliance with change the temperature of a bulb will it fail?, LEDs). All of these will pose different lighting
Nissan in Japan that makes us number three Can we design for a smaller heatsink with less challenges I think. Laser lighting is being used in
in the world, with subsidiaries that include weight? etc. Designing our headlamp LED high-end automobiles, but we are not sure how
Dacia and Renault Samsung Motors. We arrays so we can go down from five to three much it will impact the market. At the moment
are strongest in B and C category segments LEDs yet maintain lighting quality and thermal it is only used for high beam headlights and
cars like the Renault Clio, Megane, Scenic, levels can save several Euros per headlight there are questions over the efficiency of
Espace and Kadjar. We have been a pioneer in in costs. That soon mounts up over large light emission in terms of lumens/watt. Mood
automotive styling over the years and our C production runs. At Renault we have reduced lighting is being used today throughout all car
shaped headlight assemblies based on Light our headlamp costs by 50% between 2014 ranges and in terms of thermal impact, it is
Emitting Diodes (LEDs) are a signature part of and 2016 and we are aiming for a further 50% not that significant due to it being low power.
our marque these days. reduction in costs for our Generation 3 LED However, if they get hot the driver needs to
headlamps by 2018-2020. turn them off and driving in tunnels may require
Q. What is your background and your them on automatically for instance. In terms of
current role as a Lighting Expert in Renault? Q. Why did you select FloEFD as your OLEDs, there is talk of V2X (vehicle to exterior)
A. I actually trained as a Mining Engineer headlight thermal analysis tool? applications, but there are questions over
before I joined Renault in 2000 to become part A. Well, first we correlated FloEFD with OLED thermal reliability and the quality of the
of their, then, small lighting group where I have experimental measurements and it was very light source degrading over time, plus they
risen to become the resident Lighting Expert good. The big advantage with FloEFD we are relatively expensive. In terms of electric
in 2015. My role includes technologies and found is that you dont have to do meshing vehicles they are easier to design in terms of
processes associated with our global lighting and you can save a lot of time at this stage lighting compared to conventional internal
design activities with a primary focus on of CFD. Because it is embedded and highly combustion engines as there is no hot engine
optimizing our simulation and test workflows integrated into CATIA, our preferred MCAD cavity nearby plus condensation tends to be
thereby reducing costs in manufacturing. tool, one thermal engineer can be over twice less of a problem. Finally, autonomous vehicles
My work has a real impact on the financial as productive. It also moves the analysis tool will pose a whole new set of regulatory lighting
bottomline at Renault. I also constantly try to further up the design sequence. Invariably, challenges. With no driver in such a vehicle,
think out of the box and seek technology we can do many more design loops of how do you light up a vehicle so that it is
innovations that helps keep us ahead of the increasingly more complicated designs with apparent to other drivers that it is autonomous?
competition in a very competitive sector. FloEFD. We shoot for hundreds of simulations Do you have external red and green lights for
versus two or three in the past in our design instance? How do you determine standards
Q. Tell us about the Automotive Lighting optimization. With FloEFDs vapor absorption, between countries? All these need to be
Design Group in Renault-Nissan? deicing and condensation capabilities it is determined internationally still.
A. We have a large dispersed group around very well suited to our head and tail lighting
the world with most based in Paris but other applications in future. Q. Where do you see Thermal Analysis going
significant groups in Korea, Romania, Brazil and in future at Renault Automotive Lighting?
India. We are closely linked to the Electronics Q. How does thermal modeling of A. We are in the business of managing
Design Group inside Renault. We cover both headlights fit into the other physics headlight thermal and optical reliability. A good
simulation and test and deal with suppliers simulations Renault does? example of this is a Matrix Beam headlight
which Renault holds to very high specifications A. Of course we have to do EMC with ten LEDs for instance. What if some LEDs
for incoming components and assemblies. (electromagnetic confirmation), structural are on and some are off during a normal drive
analysis, vibration and optical modeling (using cycle? At the moment we deal with steady
Q. How does your role influence the Lucidshape) associated with each headlight state CFD simulations but in reality transient
bottom-line at Renault? design. And, naturally, a headlight is close behavior will become more and more important
A. A big part of my role is to understand the to the engine underhood cavity which can in determining headlight lifetimes of up to 10
supply chain for lighting inside and outside produce surface temperatures of 70 80 C -15 years. Ultimately my dream is to push
Renault to manage and reduce costs. To do which we have to take into account. Hence, CFD down to the design engineer such that
this properly, especially with suppliers, you have we have to devise simulation workflows that they would have a specification for a headlight
to have pre-development technical knowledge accommodate all our different tools to make and they could figure out for a given junction
of all components and things like their thermal sure the boundary conditions for each tool temperature at the LEDs, what size of heatsink,
performance. With the advent of LED lights is correct. Managing our software and the the overall size of assembly, and the number
we can save on CO2 emissions versus the old transition between them is becoming ever of LEDs that will be necessary to perform the
halogen lamps we used to employ. But also in more important. specification well.

mentor.com/mechanical 27
Driving down
Automotive Headlamp
costs at Renault
By Paul-Henri Matha, Lighting Expert, Groupe Renault

A
utomotive headlight design
is an important part of the
Renault brand these days and
our C shape headlights are a
signature part of the appeal of our cars.
About 30% of the cost of automotive
headlight assemblies can be found in the
mechanics and 70% in the electronics.
Hence, any savings that can be made on
the electronic side will have a profound
influence on the overall cost of these
units. This article will outline how we
have broken down our headlamp costs
and used thermal analysis tools to
incrementally optimize headlight designs
to achieve a 50% cost reduction in the
two years from 2014 2016 and how we
intend to halve it again in the next few
years.

In Generation 1 of our full LED headlight

assemblies we looked at six of our C and
D-segment vehicles from the Espace to
the Koleos. (Figure 1) We first standardized
all platforms to one common height
sensor, one common static leveller, one
common DRL/low beam/high beam driver,
one common central connector, and
common low and high beam modules facelift and stylistically we wanted to move
with two suppliers for each. We did this it to our LED based C-shape DRL lighting.
in one year by examining the partition of (Figure 3) There were four pillars to our
costs and standardized about 60% of the Generation 2 strategy:
components of our headlights. (Figure 2)
1. Be the first generalist automotive
It can be seen that the plastics we used
OEM with full LED headlights in this
only made up about 30% of the overall
B-Segment car,
assembly price. The volume effect is the
main cost factor for a headlamps part price 2. To reduce by a factor of two the
as is its supply entry ticket. However, by headlight part price between
moving from Halogen headlights in 2012 Generations 1 and 2,
(see figure 3) to LED based headlights in 3. To have a better LED lighting
2014 the overall costs went up four-fold. performance than the Clio Initiale
This in itself gave us the impetus to see if (which had Xenon 25W lights), and
we could cut costs in Generation 2 of our 4. To reduce overall headlight assembly
headlight evolution. depth by 50mm.

We focused our Generation 2 headlamp We standardized the Clio on a common

effort on our popular Segment B car, the LED Electronic Control Unit (ECU), a Figure 1. C and D-segment vehicle
Renault Clio, which was going through a common height sensor, and a common headlighting in this study

28 mentor.com/mechanical
 Automotive

Figure 2. Headlight components standardized in Generation 1 versus overall cost partition of a Headlight Assembly

leveller. We then focused on the LED Low

Beam light and reduced its price to 30%
by reducing the number of LEDs, the size
of the heatsink by 30%, and improved the
optical system. By doing all these things
(see Table 1) we improved the LED light
flux by 33%, and reduced the assembly
from eight to five LEDs. We also increased
optical efficiency by 25% and produced an
overall assembly size reduction of 50mm.
With the thermal improvements to the LEDs
we were able to increase the LED current,
increase the maximum junction temperature
usage and the flux derating at a lower
ambient temperature (Table 1). Similarly,
with the associated heatsink design, we
were able to get better junction temperature
management and better derating
management through our detailed thermal
simulations. (Figure 4)
Figure 3. Renault Clio Headlamp changes and cost reductions from 2012 2016: Generations 0 to 2
With respect to the overall headlamp
package size, figure 5 shows the 50mm Current Number of Tj @ 23C Start of
depth saving we were able to achieve Solution flux / LED
(mA) LEDs T ambiant derating
between Generation 1 with a halogen
headlamp and Generation 2 with a LED LED Low T ambiant
800 mA 200lm 8 120C
headlamp due to a better designed beam Gen1 60C
assembly. Figure 6 shows typical CFD LED Low T ambiant
simulations for a halogen headlamp in 1A 270lm 5 130C
beam Gen2 50C
Mentor Graphics CAD-embedded software,
FloEFD, illustrating the complex air flows LED High T ambiant
and thermal effects to be found on the 800 mA 200lm 4 120C
beam Gen1 60C
surfaces in the assembly.
LED High T ambiant
1A 270lm 4 130C
Looking deeper at our general CFD-based beam Gen2 50C
thermal analysis approach to headlamp Table 1: LED solution evolution from Generation 1 to 2 for the Renault Clio Headlamp

mentor.com/mechanical 29
design that typically is used to optimize
designs, we would normally be interested
in predicting lighting performance at 23C
outside the headlamp in ambient air and
up to a maximum of 70C for the outside
temperature for the outer boundary of LED
reliability. To validate our simulations, we
therefore carried out some experimental
tests where we fixed the ambient
temperature outside the headlamp at 23C
and installed eight thermocouples outside
the assembly (see figure 7) for a car with its
engine on and off.

Figure 8 shows thermocouple time traces

for both the engine on and car stationary for Figure 4: Renault Clio Headlamp heatsink weight evolution Generations 1 to 2
3hr 30mins; then lights switched on for 1hr
30mins with the engine on and stationary,
and then the lights on and engine driving
for 1hr 30 minutes. It is clear that the
temperatures can reach over 50C inside
the headlamp when the engine is idling and
the lighting is on for a prolonged period. In
addition, headlamp surface temperatures
can rise to 65C in certain idling conditions.
With other tests we were able to show that
with just a low beam on for an hour, the
temperature inside the headlight went to
20C and with both low and high beam on
for an hour an extra 5C in temperature was
measured.

We also carried out a series of tests where

we evaluated the effects of engine idling
and lights either on or off by looking at Rjth
of an Altilon LED with 3K/W and 3 chips
@ 1A and for a Delta (Tjunction Tcase)
of 20C. We were able to show that for
ambient temperatures of 70C and with
both low beam and high beam lights
on, together with the engine on that the
junction temperature of the LEDs comes Figure 5: 50mm Headlamp Package saving between Generation 1 (halogen) and Generation 2 (LED)
very close to the worst case scenario
of 150C. We concluded that it was not
possible to design a LED system if we
were to take into account all the use cases.
The OEM must therefore define the best
compromise. For example, at 23C, after
one hour engine idling, lighting performance
was shown to be at 100%, but if the
ambient temperature rose to 50C for the
same situation the lighting performance
would go down to 80%. In order to respect
this specification we concluded that a
thermal sensor had to be added to the PCB
so that the current could be reduced if the
temperature at the LED was greater than a
threshold we would define. We could then
do a thermal derating and a flux derating of
the full LED headlamp.

Going forward, we are putting in place an

action plan to deal with simulation and
testing of lighting in transient drive cycle Figure 6: FloEFD Thermal Predictions for a Renault Halogen Headlamp Assembly

30 mentor.com/mechanical
 Automotive

Figure 7: Location of the eight thermocouples for the Ambient Temperature Headlamp Engine Tests

modes. (Figure 9) As an OEM we want to

be able to simulate the impact of car speed
on its lightings thermal performance, and
in particular the thermal variation due to
speed of each of our engines. This will
make thermal CFD software key for lighting
engineers in future.

We also need to be able to model the

nearby engine bays thermal behavior in
parallel with the headlight simulation as
they affect each other. And we also see
the need for thermal management inside
the headlamp when thermal inductors will
be present. In short, Renault believes that
the OEM should be responsible for the
complete thermal system associated with
headlamp design. To do so will make the
OEM a market system leader.

Our immediate goal in the Renault Lighting

Team with Generation 3 of our project is
Figure 8: Thermocouple Temperatures outside the Figure 9: Headlamp Temperature Measurements for
to increase the LED flux of headlamps Headlamp Assembly for Various Engine on and light Engine Idling and during a Car Drive Cycle
from 270lm to 320lm by 2018 2020 and on/off scenarios
introduce new LED Drivers to deal with
higher power, plus we are aiming for the
introduction of Turn Indicators and AFS
functionalities, as well as overall assembly
size reduction. Finally, we are aiming for
a Height Sensor regulation evolution. Our
eventual target is to get our full headlamp
assembly to reduce in price by another
50% bringing it to the levels we saw
with halogen headlamps five years ago.
(Figure10)

Figure 10: Renault Headlamp Cost Reduction Roadmap to Generation 3 Headlighting

mentor.com/mechanical 31
How To...
Characterize an Automotive Thermostat for Thermo-fluid System Simulation
By Doug Kolak, Technical Marketing Engineer, Mentor Graphics

Figure 1. Thermostat Model in FloEFD

Automotive thermostats are one of

the most important components in a
vehicle thermal management system.
It is responsible for maintaining an
acceptable working temperature in the
engine both to prevent catastrophic
failure and keep it operating as efficiently
as possible. As vehicles have become
more complex with their cooling
systems, so too have thermostats. These
parts are no longer as simple as a single
flow path in and then either flow to
bypass or the radiator. It is easy for these
components to now have at least three
or four inlets and multiple outlets.

This increased complexity creates more

complex 3D flow patterns that are not
accounted for in any standard FloMASTER
component model. Engineers could try to
replicate the number of flow arms using Figure 2. Response Surface Models for Pressure Drops
discrete losses, but this does not allow for
the interaction of the effect on pressure error is eradicated by considering the effect goal is to characterize this component for
drop that the differing flow rates in all the of changes in branch pressure drops with use at maximum flow rate to the radiator,
arms have in reality, e.g. Venturi effects. changing flow rates in the other branches. so an assumption can be made that the
This is where applying simulation based Using FloEFD to perform full spectrum 3D radiator bypass is fully closed. This simplifies
characterization can increase the accuracy CFD characterization of the housing and the problem to a four-arm component with
of automotive cooling circuit simulations. integrating inter-dependent flow channel three inlets and one outlet.
In fact this is something Sudhi Uppuluri of interactions into a FloMASTER N-Arm
Computational Sciences Experts Group component improves simulation accuracy, The next step is to determine what the
hasexpressed, confidence and trust reducing physical bounds of the characterization should be.
testing times and ultimate time to market. At this point in the design the expected
"When modeling multi-arm junctions such coolant flow rates should be known, and
as automotive thermostats and pump FloEFD Model Setup if they are not, thermal balance for each of
housings, characterizing individual flow As an example, figure 1 shows a five arm the cooling loops could be done to estimate
paths as parallel and independent losses thermostat with four inlets (from the radiator, these values. With this information it is
can give rise to 15 to 20% error in predicting gear oil cooler, engine oil cooler, and radiator important to give a range of values centered
mass flow rates through an outlet. This bypass) and one outlet to the engine. The on the expected flow rate. The range can

32 mentor.com/mechanical
vary based on how well the system and hydraulic limits are available for review. At management model and connected to the
component are known but for this example this point the 3D visualization is also available appropriate parts of the model as seen in
a range of +/- 20% is applied and shown to identify each arm and also to create a figure 4. An initial run results in a system
in Table 1. In addition to the flow rates, a component symbol as seen in figure 3. The that is operating well with-in the bounds of
reference arm must be specified which in this 3D representation can be rotated and zoomed the characterization. This process allows
case will be the thermostat outlet. This arm to get the best angle, and the camera icon at for increased accuracy in the pressure drop
has an environmental pressure of 101325 the top can be used to take the image. Once through the thermostat when compared to
Pa which will remain fixed through all of the this is complete, the component will be saved previous ways of modeling in FloMASTER
simulations. into the FloMASTER database for use by and thus a more accurate representation of
anyone with access to it. the system behavior.
Design of Experiment Setup
These values can then be entered into With the component in the FloMASTER Download the White Paper:
FloEFD using the Characterization option database, it can be used in any model with Simulation Based Characterization. A 3D
for the type of Parametric Study. Once the a simple drag and drop similar to standard to 1D System Level Thermo-Fluid CFD
values are entered, a number of experiments FloMASTER components. In this example, Workflow: http://bit.ly/2nwjEmV
is required, in this case 120 simulations it is added to an automotive vehicle thermal
were run. FloEFD then creates a Design
of Experiments (DoE) that randomly and Parameter Min Value (L/min) Max Value (L/min)
equally distributes the combinations of flow
rates throughout the characterization space. Gear Oil Cooler Flow Rate 9.6 14.4
The DoE can then be run with the results
presented as dimensionless flow rate and
Motor Coolant Flow Rate 48.4 72.6
dimensionless pressure drop. These values
have been non-dimensionalizes based on the Motor Oil Cooler Flow
fluid properties to remove any dependency 39.4 59.4
Rate
on density and viscosity. This means when
Table 1. Oil Flow Rate Values
using this characterization in FloMASTER it is
not critical to use a fluid with the exact same
fluid properties.

Once all of the experiments are finished

successfully, the Response Surface Models
(RSM) can be created. These response
surfaces are N-dimensional mathematical
surfaces that relate an output to multiple
inputs. The first thing to check with the
RSMs is the error in their mathematical fitting.
The amount of error that is acceptable is a
subjective decision, but for this scenario <
5% is considered acceptable which all three
RSMs and under. If further refinements need
to be done, additional experiments could
have been added to the original DoE with the
new points fitting into the original distribution.
There is also the option to visually inspect the
RSMs as shown in figure 2.

Response Surface
Model Export and Import
Assuming everything is acceptable the RSMs Figure 3. FloMASTER N-Arm Creation
can then be exported to FloMASTER as a
.flonarm file. This file is actually an FMU
adhering to the FMI standard, but augmented
to also include additional information that
is used by FloMASTER such as the area of
each arm and the 3D representation of the
geometry for visualization in FloMASTER.

To create the thermostat component

in FloMASTER, New FloEFD N-Arm
Component can be selected from the
Launchpad. Once the file is loaded and the
import is successful, the branch data and Figure 4. FloMASTER Cooling System Model with FloEFD Thermostat

mentor.com/mechanical 33
System Level Evaluation
of 4-Quadrant Mixed
Flow Diffuser Pump
Behavior Using CFD
By Doug Kolak, Technical Marketing Engineer, Mentor Graphics

T
his work, System Level
Comparison of 4-Quadrant Mixed
Flow Diffuser Pump Behavior
Using CFD, is a derivative
of Investigation of the 4-Quadrant
behavior of a mixed flow diffuser
pump with CFD-methods and test rig
evaluation, by S. Hller, H. Benigni, and
J. Jaberg, IOP Conf. Series: Earth and
Environmental Science 49(2016) 302018,
DOI: 10.1088/1755-1315/49/3/032018,
used under Creative Commons 3.0.

When looking to analyze a thermo-fluid

system one of the most important aspects
to consider is: what will cause fluid flow
in the system. This may be as simple as
imposing a flow rate or using a difference in
pressure due to elevation, however one of
the most common real-world methods of
generating fluid flow is through the use of
pumps. Generally the behavior of centrifugal Figure 1. CFD Model of a Centrigual Pump
pumps is well known under normal
operation of forward flow with a forward The simulation was carried out using a
rotational direction of the pump. This is the full 360 model with traditional 3D CFD
most common usage for a pump and is software. The model consisted of the
usually well defined in terms of pump head suction bell domain including an inlet
and efficiency from the manufacturer. extension, the impeller domain, and the
stator domain including an outlet extension
However, there are scenarios where the as shown in figure 1. The inlet boundary
pump will need to run outside of this condition was a specified mass flow rate
normal operating mode, such as when and the outlet boundary condition was
the direction of flow or rotation change. defined as a static pressure. An entry and
An example of this is a pump trip, where exit length of five times the respective
to reliably calculate system transients like diameters was used to help redevelop the
maximum or minimum internal pressures flow profile. Figure 2. Head Curves for Different Blade Positions
it is necessary to obtain performance data
for all quadrants of operation. While it is The experimental measurements were made scale was chosen. To accurately compare
possible to determine this information at the Institute of Hydraulic Fluid Machinery the results of the CFD analysis and the
through physical testing, it is expensive, so at Graz University of Technology. Since a physical test, the test results were scaled
the ability to generate accurate 4-quadrant full size prototype pump with a mechanical up using IEC60193. As shown in figure 2,
pump characteristics using Computational power of over 1500 kW would exceed the the results of the simulation and the physical
Fluid Dynamics (CFD) can be valuable. capabilities of the test facility, a 1:3.6 model test showed good correlation.

34 mentor.com/mechanical
 Manufacturing

Figure 3. FloMASTER Pump Trip Model

To further demonstrate the effectiveness

of creating this performance data with
3D CFD a pump trip simulation was
conducted using FloMASTER. The goal
being to evaluate the effect of discrepancies
between the experimental and the
numerically calculated pumps. The system
level thermo-fluid model shown in figure
3, contains two reservoirs connected by
the pump and piping with a static head
difference of 22 meters. The mixed flow
pump is connected directly to the outlet of
the lower reservoir and will operate at 507
rpm. The best efficiency point for the pump
is at 5m3/s which produces a head rise of
27m. The outlet of the pump is connect to
the downstream reservoir by a 2000m pipe,
with a diameter of 1.5m and an absolute Figure 4. 0 Degree Pitch Results
roughness of 0.025mm. To account for the
transient nature of the scenario, the method
of characteristics is used for the pipe
with a combined wave speed of 1000m/s
specified. Finally the inertia of the pump
including the drive is set to 1000kgm2.

The analysis is conducted for 120 seconds

with the pump tripping after five seconds.
Two different pitch angles were analyzed
in this way, 0 pitch angle and -9 pitch
angle, the results are shown in figure 4
and 5. The graphs compare the system
transients for flow rate (Q), pump rotational
speed (n), and pump outlet pressure (p) for
both the experimental curve and the CFD
generatedcurve.

As expected the decreasing rotational

speed once the pump has tripped results
Figure 5. -9 Degree Pitch Results
in an immediate pressure drop at the outlet
of the pump. The flow rate decreases at
a lower rate than the pressure due to the to occur. At this point a slight difference in behavior to the experimental results
inertia of the fluid mass inside of the pipe. the rotational speed and flow rate becomes when used in FloMASTER for a pump trip
During this initial phase of the simulation the evident. This difference is likely an effect analysis, thus providing a reliable option for
graphs show there is virtually no difference of the inaccurate prediction of the pump obtaining pump performance outside of the
in any of the three parameters between behavior in the CFD calculations near the normal operating range.
the experimental and CFD pump models. turbine runaway point.
However, once the pump changes the
direction of rotation and the direction of flow Ultimately the results from the CFD pump
reversed, a turbine runaway condition starts characterization provide comparable

mentor.com/mechanical 35
"With FloEFD, the configuration of the simulation is made from within the CAD
software, which eliminates the need for exporting/importing geometry. The software
also has features such as automatic meshing, a case configuration wizard and
integrated post processing, all from within the CAD software."
Arne Lindgren, Halmstad University, Sweden

Image Credit: Koenigsegg Automotive AB

36 mentor.com/mechanical
 Automotive

Sports Car Brake

Cooling Simulation with
CAD-embedded CFD
By Arne Lindgren, Halmstad University, Sweden

B
rake cooling is a crucial area
in motorsport and sports car
engineering. A recent thesis
project by Arne Lindgren
of Halmstad University in Sweden
considered different cooling solutions
for the extreme conditions of super cars.
The project, conducted for super car
manufacturer Koenigsegg Automotive AB,
had the objective to design an efficient
brake cooling solution for their latest
model, the Regera. (Figure 1)

Regera, which means "to reign" in Swedish,

is the first Koenigsegg car with hybrid-
technology, the combined power of its
internal combustion engine and its electric
motors exceeds 1,500 horsepower. Such
a powerful car needs effective and reliable
brakes. During a braking event from 300 to 0
km/h, the average brake power is over 1 MW
in the Regera.

Koenigsegg Automotive AB is a Swedish

company founded in 1994 by Christian
von Koenigsegg. The first prototype was
completed in 1996 and the series production
of their CC8S model started in 2002 [2].
The Koenigsegg CCR became the fastest
production car in 2005 and the CCX model
took the Top Gear lap record in 2006 with a
time that wasnt beaten for over two years.
In 2014 the One:1 model was introduced,
the world's first production car with a hp-to-
kg curb weight ratio of 1:1. The company
employs approximately 120 staff including
an engineering department which consists of
about 25 engineers.

Lindgren evaluated several brake cooling

designs for the Regera in his thesis. For

mentor.com/mechanical 37
his CFD simulations he used FloEFD
embedded in CATIA V5 from Mentor
Graphics experimental testing being
deemed too expensive and lacking in
acquisition of certain flow data [1].

While dedicated brake cooling is not

necessary for ordinary passenger vehicles,
it is a huge challenge for sports cars that
must tolerate racing conditions. The cooling
effect of the ambient air is normally sufficient
for ordinary car brakes during normal
driving conditions. Modern road vehicles are
equipped with internally vented brake discs
(at least at the front axle, and usually also at
the rear). The internal vanes help to pump air
through the disc, internal channels are where
the biggest heat dissipation takes place [1].

Over-heating of brakes can lead to a number Figure 1. Koenigsegg Regera

of problems:

Friction material degradation;

Thermal stress in the brake disc, which
can lead to distortion and stress cracks;
and
Brake fluid vaporization in the brake caliper.

These failures can lead to partial or complete

loss of braking, which is very serious.

With regard to racing cars and cars for track

driving, this issue becomes more complex.
Special cooling ducts which direct ambient
airflow to the brakes are used to ensure
sufficient cooling performance taking into
account the more frequent braking intervals
that occur during track driving.

Brakes are mainly cooled through the heat Figure 2. Koenigsegg Regera front brake.
transfer method called convection, where
a fluid absorbs heat and transports it away
from the hot object. Lindgren focused
his work on improving the convective
air cooling of the brakes, and limited his
studies to the front axle brakes as this is
where heat generation is at its greatest.
The brake cooling solution that was used
on the previous generation of Koenigsegg
cars was used as a baseline for the
coolingsimulations.

The baseline design consists of inlet ducts

in the front bumper of the car (that captures
ambient airflow) and flexible hoses that
channel the air to ducts (or nozzles), which
are mounted on the wheel bearing carriers
and direct the cooling air towards the
center of the brake discs. (Figure 3) For the
simulations, CAD models of the relevant
geometry were provided by Koenigsegg
which were used for the embedded
CFDsimulations. Figure 3. Brake cooling duct mounted on a Koenigsegg Regera upright (the baseline design). (Klingelhoefer, 2013, Ref 3)

38 mentor.com/mechanical
 Automotive

The main geometries used in the simulations

were:

The brake discs made of Carbon fibre-

reinforced silicon carbide (C/SiC), diameter
396 mm and thickness 38 mm, equipped
with radial ventilation channels;
The brake pads;
The brake caliper;
The upright (or wheel bearing carrier);
The hollow wheel axle;
The 19" wheel rim with tire; and
The wheelhouse geometry.
Figure 4. The geometrical wheelarch model (blue arrow indicates ambient airflow direction)
After simulating the baseline configuration,
different brake duct concepts were generated
and their cooling effect was simulated. The
same computational model was used for
both baseline and concept simulations,
only the brake duct geometry and, naturally,
the position of the duct inlet boundary
conditions were changed. The simulations
were conducted directly within the 3D
geometry models in the CATIA V5 embedded
FloEFD CFD software. This allowed for
efficient simulation of complex geometries.
An efficient and productive workflow was
found as a result of the automatic meshing
function, the case configuration wizard,
the post processing features, and because Figure 5. Wheel and ground boundary conditions. Rotating regions highlighted in turquoise
geometry could be modified directly within
the CADenvironment [1].

As to the objective of the project was to

investigate many concepts, a reasonable
calculation time was required. Therefore,
a complete simulation of the entire vehicle
was not expedient. A partial car body with
wheelhouse was used as well as wheel and
the brake as assemblies (figure 4) with similar
dimensions and ground clearance as the
Regera.

The ambient velocity was defined as 150

km/h, which is a typical average speed on
a race track. An additional airflow from the
radiator was applied on the inboard side of
the wheelhouse (figure 4, red arrows). The Figure 6. Disc and pad with FloEFD wall boundary conditions

Figure 7. FloEFD Computational mesh of the Wheel, Brake Disc and Wheel Housing

mentor.com/mechanical 39
airflow through the flexible hose from the inlet focused on the cooling of the brake discs Another early discovery was that the design
ducts in the front bumper was modeled as as most of the braking energy goes into of the cooling channels in the brake discs
a pressure to get a realistic flow rate for all the discs. The approach was to increase could be improved. But the brake disc
possible brake duct designs. The rotation the convective cooling by breaking up design was outside the scope of this project
of rotationally symmetric geometries, such the temperature boundary layer, which so it was not studied further. The break-
as the tires and brake disc friction surfaces, can be done by using high air velocity or through was the idea of putting the brake
were defined with wall conditions. The by introducing turbulence. In addition, duct inlet in the center of a wheel axle that
FloEFD sliding mesh approach was applied other design criteria were considered to has radial channels. This resulted in higher
to the non-rotationally symmetric parts. A ensure that the solution withstands forces, hose flow rate because the radial channels
3D body (rotating region) is used to define vibrations, and temperatures etc., that occur in the axle and the brake disc work together
which geometries should rotate, in this case in drivingconditions. as a centrifugal fan. Finally, this concept was
the rim spokes and brake disc channels. A supplemented by a passive cooling design
translation velocity of 150 km/h was applied The investigations showed that a local (not relying on airflow from the hose), realized
to the ground (figure 5) to include ground improvement of the airflow often led to worse by two ring-shaped plates perforated with
effects. Only convection was considered for heat dissipation in other areas simultaneously. slots. (Figure 10) The simulations showed
the CFD simulations as this is the easiest The airflow rate was simply not sufficient to that these plates improved the cooling of the
heat transfer process to influence and has improve the cooling over larger surfaces. discs friction faces, but Lindgren expresses
a proportion of about 60 90 % of the total
heat dissipation. The surface temperatures
applied on the surfaces of the parts
(figure 6) were based on values recorded
by Koenigsegg during track testing. The
simulation was conducted with approximately
3.5 million cells (figure 7) using the FloEFD
two-scale wall function technique, which
enables the use of a coarser mesh than
would be otherwise necessary in traditional
CFD codes.

The concepts were based on Lindgrens

own ideas, observed brake cooling designs,
and observations in other applications while
taking into account only concepts that
are possible to manufacture. 12 different
concepts were investigated using the
described FloEFD boundary conditions. With
the given settings, the simulation time was
approximately 24 hours on a computer with
a six-core Intel Xeon E5 CPU at 3.5 GHz and
32 GB RAM.

The baseline simulation results are shown

infigure 8.

Each concept was compared with the

baseline concept. The investigations Figure 8. Baseline brake duct highlighted in blue and table with the numerical results from FloEFD

Figure 9. Overall view and FloEFD results for concept No. 8 Figure 10. Overall view and FloEFD results for concept No.12

40 mentor.com/mechanical
 Automotive

skepticism to these results as the simulations

didn't include radiation (in reality the plates
would reflect heat back to the disc).
Concepts 8 and 12 were the most promising,
although concept 12 needs further analysis
or testing to see the effect of radiation. The
designs of the concepts and the results given
as the difference in percent from the baseline
values are shown in figures 9 and 10.
Concept 8 has the advantage that only a few
relatively simple additional parts are needed.

From the almost infinite number of possible

cooling solutions, 12 concepts were
analyzed (figures 11 - 13) and compared
with the baseline design from Koenigsegg. Figure 11. Baseline Case (a) temperature cut-plot on a horizontal plane through the
The two most promising solutions lead to wheel, and, (b) vector velocity on a vertical plane through the disc channels.
an overall thermal improvement of 14%
and 25%. Concept 8 was proposed as an
enhancement, which had the hose inlet in
the center of the wheel axle thus directing
the cooling air through radial channels to the
brake disc. FloEFD simulations indicated that
the proposed design should result in 14%
higher heat transfer rate compared to the
previously used cooling solution. In addition
to these cooling ducts, some passive cooling
devices could also be considered in future.

In this study FloEFD was able to provide

trend predictions within engineering
timescales for a wide range of concepts
although it was recommended that finer
meshes and radiation effects be examined Figure 12. Concept 8 velocity cut-plot horizontal plane (a) through the
more in the future which require more wheel, and, cut-plot vertical plane through disc channels (b).
computational resources. However, the
CFD results described here indicate that
when compared to brake cooling with the
previous Koenigsegg ducts analyzed as a
baseline, new concepts could be created,
analyzed and developed in an easy iterative
process. The simulations with the CATIA V5
embedded version of FloEFD made these
investigations possible because creating
a real prototype of each concept would
lead to high cost and time requirements.
The most promising solutions can now be
investigated deeper in terms of structural
analysis, manufacturing processes, and Concept 8, velocity 1 mm inside disc Concept 12, velocity 1 mm inside disc
finally produced as a prototype.

References
[1] Development of Brake Cooling, Arne
Lindgren, Bachelor Thesis in Mechanical
Engineering, 15 credits, Halmstad 2016-05-
20:
http://www.diva-portal.se/smash/get/
diva2:938489/FULLTEXT01.pdf
[2] Koenigsegg website: http://koenigsegg.
com/

Concept 8, velocity 1 mm outside disc Concept 12, velocity 1 mm outside disc

Figure 13. Comparison of Concept 8 and 12

mentor.com/mechanical 41
Winging It!
FloEFD provides Accurate and Fast Flight Load
Data for Aircraft Wing High Lift Devices at Irkut
By Andrey Chuban, Lead Design Engineer, Irkut Corporation

Figure 1. Typical tube experiment for a scale model aircraft

E
ngineers at Irkut Corporation
were looking for a way to get
better external aerodynamics
data than what they were
currently able to get from tube
experiments. Tube experiments consist
of building scale models of the aircraft
and measuring values in a wind tunnel.
This approach is expensive and has
some inherent disadvantages. These Figure 2. 3D CAD model of aircraft
include low Reynolds numbers and
higher airflow turbulence intensity for an alternative approach. As part of this to those of the tube experiment,
scale models vs. full size aircraft. the CFD tools needed to be validated Provide loads for the flight modes and
There are also inaccuracies due to the for this use, the tools tested included configurations, that were not tested
scaled geometry such as radii and Mentor Graphics' FloEFD and Ansys' during experiments,
points where different structures such CFX.
as the wings and fuselage meet. Compare solutions with or without
Finally, tests run in different wind The criteria for validating the tools were: ground effect simulation screen, and
tunnels can produce varied results. To Compare the FloEFD results with the
solve these issues Irkut turned to CFD as Obtain the computational results close load data obtained with CFX.

42 mentor.com/mechanical
 Aerospace

Figure 3. Final Mesh after a single adaption

= 7 deg. = 12 deg.
Figure 4. Computation Results: P on Wing Surface

The FloEFD software was installed on a speed gradients. The simulation was then run for two angles
double Xenon powered computer with of attack, 7 degrees and 12 degrees
48 Gb RAM. The mesh size was limited After construction of volumetric zones and several computational results were
to 4 million cells (about 1.4 million before around the nose of fuselage, high lift examined. The first result studied was the
mesh adaptation process). The density of devices, wing with engine and whole aircraft P on the wing surface. This showed an
the mesh was organized with the help of a basic mesh was refined automatically. The increase of pitch angle leading to the growth
volumetric zones to meet the basic rule: picture in figure 3 presents the final mesh of negative relative pressure along the wing
very dense mesh in areas with large airflow after a single adaptation. leading edge surface and upper surface of

mentor.com/mechanical 43
Figure 5. Computation Results: Mach Number Distribution along Wing Airfoil

Figure 6. Results Comparison: Slat Drag Coefficient Figure 7. Results Comparison: Slat Lift Coefficient

Figure 8. Results Comparison: Flap Drag Coefficient Figure 9. Results Comparison: Flap Lift Coefficient

the slats. So slat loads are the function of an

angle of attack.

The relative pressure on a bottom wing

surface doesnt really depend on the angle
of attack; flap loads are almost independent
of the angle of attack.

Next they examined the Mach number

distribution along the wing airfoil. In case
of a full flap extension, a high pitch angle Figure 10. Engine Thrust Effect: Flow Stagnation on the Flap Bottom Surface
may provoke flow separation on the flap's
upper surface. The relative pressure
growth on the upper surface led to flap ground effect simulation screen, a plate screenpresence.
loadsdecrease. was added to the FloEFD model to capture
the screen effect. This resulted in an The drag and lift coefficient plots illustrate
Since the experiment was run with the almost doubling of drag load level with that qualitatively identical airflow around

44 mentor.com/mechanical
 Aerospace

the slat in the half-way and fully extended The tube experiment and FloEFD calculations Conclusions
positions. show large increase (four times for FloEFD) The FloEFD results satisfactorily
of flap drag load for half-way extraction correlate with the experimental data
The addition of the screen also greatly configuration and about 20% increase of lift which are very close to the results of
raised the slat loading, but for flap, the load. For the outward flap tube experiment, ANSYS CFX calculations. The size of
opposite is true. results differs from CFD calculations due to the CFX mesh used for calculations
slightly different flap release angles of a tube was about 18 M cells, so these flow
The flap lift coefficient is especially affected model and real aircraft. computations took too much time
by screen, they observed a 15% load and required a very high-capacity
decrease at high angles of attack. The final criteria studied was the Reynolds equipment. Each FloEFD calculation
number effect. The tube model's small took about six to eight hours on a
The next stage of the validation was to scale leads to a difference in Reynolds standard workstation.
consider the engine thrust effect. The tube
number of a flow around the model and The ground effect simulation screen
experiment was also done with the engine
the aircraft. To determine the effect of the greatly affects the drag coefficient of the
simulator which demonstrated a 10%
Reynolds number, the scale model was slat and must be accounted for during
increase of flap lift at high angles of attack.
calculated in FloEFD. The results of the calculation. Outward flap loading is
To investigate the engine thrust, effect the
calculation of the aircraft and a scale model highly dependent on spoilers release.
boundary conditions at the inlet and outlet
of the engine were added to the model. This were almost identical. The inner flap load growth due to the
led to flow stagnation on the flap bottom
surface.

FloEFD also shows a flap load increase

during computation with the engine thrust
taken into account, but the gap between
the results with and without thrust tends
to decrease with the angle of attack
growth.

The spoilers, release effect at a small angle,

(within 10 degrees) results in a substantial
increase of outward flap loading. Wherein
large spoiler release angles greatly disrupt
the flow around the flap, causing a
reduction of flap loading. The main problem
for the flap release kinematics designer is
a huge tangent load increase after spoiler
release.

10 degree spoiler release causes flow

separation on the flap aft edge and flow
acceleration on the leading edge.
Figure 11. Engine Thrust Effect: Flap Drag Coefficient

Figure 12. Spoilers Release Effect: Negative relative pressure growth on flap leading edge after spoilers release

mentor.com/mechanical 45
 engine thrust increase must also be question about the load data source at geometry between the tube model and
considered. the aircraft design stage. On the one the real aircraft.
The difference between the side there is a traditional distrust in CFD
experimentally defined and FloEFD and results and on the other side there is
ANSYS CFX computed loads raises the a difference in Reynolds numbers and

Figure 13. Spoilers Release Effect: Flow separation on the flap aft edge and flow acceleration on the leading edge

Scale True

model model

Slat drag
-0,11 -0,08 37%
coefficient

Slat lift
0,91 0,96 5%
coefficient

Flap drag
0,02 0,02 0%
coefficient

Flap lift
0,89 0,96 7%
coefficient

Figure 14. Spoilers Release Effect: Flap Drag Coefficient Figure 15. Reynolds Number Effect: Comparison of
scale model to True Model

46 mentor.com/mechanical
 Automotive

Figure 2. BOSE Automotive Audio Amplifiers

Bose Automotive
Systems Finding new ways to make our Products Cool!
By Brad Subat, Mechanical Engineer, Bose Automotive Systems

L
ike many industries these days, it
is vital to maintain class leading
technology and deliver products
to market in less time. For Bose,
one instrumental tool has been Mentor
Graphics thermal analysis software, first
FloTHERM and more recently FloEFD.
We switched to FloEFD since it handles
our more complex geometry and has
the benefit of being embedded in our
CAD tool (NX). While these tools provide
great insight into the thermal design of
products, we enter a new project with
a bias to use previous designs which
brings us to this article.

More than a year ago, I was intrigued

by the first Mentor Graphics blog (of
what became a several part series) on
Organically Grown 3D Printable Heatsinks.
So, while the example was nothing like
our actual products, I could see that this
Figure 1. Power Dissipation vs. Amplifier Output Power technology (once matured) could help us

mentor.com/mechanical 47
Figure 3. Power Amplifier Heat Source Locations Figure 4. Starting Rod Locations

meet our goals in future years. At Bose, we

believe we have Better products through
research, so while a tool might be effective
today, we are drawn to developing better
tools for the future. I then reached out to
Mentor Graphics to propose we start an
industrial research collaboration where
this automated growth approach could
be applied to something closer to our
products. They would focus their algorithm
development on our design space, while
Bose would provide feedback and testing at
critical progress points.

Designing products for the automotive

audio market has many constraints
and competing goals. In respect to our
amplifiers, we design them to work within a
wide temperature range, with high product
reliability and nearly perfect initial quality.
In the thermal realm, there are a few main
components that dominate our designs;
power amplifiers, voltage regulators and Figure 5. Reduction in Average Power Amplifier Temperature During Growth Process
digital signal processors. Traditionally, the
largest power dissipating (Pd) devices variables and conditions should be used to canvas of this growth study was a minimum
are power amplifiers. There are two main produce this data, so, focus on the relative nominal wall shell around a typical size PCB.
technologies that are used in the automotive differences. We included a few thick sections for heat
industry, linear (Class AB) and switching spreading and added three starting rod
(Class D), figure 1 Linear power amplifiers For the purpose of this initial investigation, locations (figure 4) from which the algorithm
are less efficient but historically have been the simplified model only had linear power builds.
more cost effective. In recent years, the amplifiers as components (figure 3) and a
industry has focused on improving the cost uniform heat source was spread over the In automotive, part re-use is important, so
effectiveness of switching amplifiers and the simplified PCB. car companies typically choose to have us
cost gap has narrowed significantly. The design a housing that can work in several
graph below shows the power dissipation Boses current design methodology for orientations. This allows them to have
differences versus power out for these our amplifiers is to use die cast housings design freedom from vehicle to vehicle
types of power amplifiers. On both axes, (both aluminum and magnesium) for five, of in regards to the amplifier location and
there can be endless discussions on what the six, sides of the enclosure. The blank orientation. During a standard development

48 mentor.com/mechanical
 Automotive

many orientations are considered. To Growth Progress

simplify this initial study, only the fins up axis
orientation was considered for growth
since it usually is the hottest orientation.

Figure 5 shows how the heat source

temperature drops, as the heatsink grows.
The sharper the initial drop, the greater the 0%
effectiveness of the added material. As the
growth continues, the rate of temperature
reduction slows. If the slope becomes very
small, this indicates that you are having to
add a lot of material for a small temperature
reduction. An ideal thermal design would still
have a large slope once the thermal goals
have been reached.

Figure 6 shows a few steps of the growth.

With the simulations in and showing 25%

favorable results, it was time to do actual
thermal testing to see if this innovative
simulation technique would correlate well.
Before machining our test housings, the
grown design was adapted in our CAD
tool to account for high level die casting
designrules:

Fin spacing: Tool steel is the inverse

of your part, so between fins there
needs to be enough tool steel to
provide cooling and strength during
50%
the die-casting process. If pushed too
far, the standing steel can break during
production requiring major repair or early
decommissioning of that cavity.
Draft: Like plastic injection parts,
die-casting needs a taper so it can be
pressed out of the tool.
One area not addressed, was mold
fill. In high volume, automotive die
casting process use an extremely short
high pressure process to inject molten
metal into the tool that can cool or trap 75%
air in a complex geometry.

As seen in figure 7, these results are very

encouraging for this additive technology,
as the unaided design used less material
and has a similar thermal result. For
thoroughness, we also tested the machined
housings to confirm our selected orientation
was indeed the hottest.

As with any research, it leads too many

other areas to explore. So as Mentor 100%
Graphics was working on growing a
heatsink, they decided to work on an
optimization algorithm in parallel, which was
subtractive by removing material from the
design. This algorithm was applied to the
grown heatsink tested above. The graph
figure 8 to the right shows the increase in Figure 6. Fin Growth and Temperature Distribution During the Growth Process

mentor.com/mechanical 49
"What was seen as an interesting idea over a year ago by Mentor
Graphics and Bose for a autonomously grown heatsink, has now
proven to be possible technology for the future."
Brad Subat, Mechanical Engineer, Bose Automotive Systems

Predictions and results:

Heatsink Fin only

volume surface area PA1 (C) PA2 (C) PA3 (C)
(mm^3) (mm^2)

Existing
design 268,475 100,904 119 124 117
(Prediction)

Grown 236,833 76,921 120 122 114

(Prediction)

Delta 12% less 24% less 1 Hotter 2 Cooler 3 Cooler

Figure 7. Existing Fin Design vs. Grown Fin Comparisons

Figure 8. Increase in Thermal Resistance vs. Heatsink Mass Removed

50 mentor.com/mechanical
 Automotive

the housings thermal resistance versus the

amount on heatsink fin material removed. Mass Removed
In this specific design the initial removal of
heatsink material is predicted to actually
lower the thermal resistance (cooler). Figure
9 shows the optimization of the fins and
predicted temperatures at severalstages. 0%
We decided to test at 22% since there
was a small leveling in the rise of thermal
resistance. (Figure 10)

Again, the results are very encouraging

for this subtractive algorithm, as heatsink
material can be removed for minimal
temperature rise. So if you have a product
11%
with thermal margin and are looking
to optimize it, this technique might be
immediately interesting for you.

A special thank you to the Mentor Graphic

team this past year for their hard work,
vision and passion for this collaborative
project. What was seen as an interesting
idea over a year ago by Mentor Graphics
and Bose for a autonomously grown 22%
heatsink, has now proven to be possible
technology for the future. I look forward to
the next steps in refinements from Mentor
Graphics to see if we can establish the
confidence needed to add to our product
development tools in thefuture.

46%

Figure 9. Fin Geometry and Temperature Distribution (Hardly Changing) During the Fin Mass Reduction Process

Heatsink volume Fin only volume PA1 (C) PA2 (C) PA3 (C)

Existing
design 268,475 100,904 119 124 117
(Prediction)
Optimized 219,114 62,268 122 123 117
(Prediction)

Delta 18% less 38 % less 3 Hotter 1 Cooler No Change

Figure 10. Grown Fin Design vs. Mass Reduced Fin Design Comparisons

The growth methodology involves a series of sequential FloTHERM simulations. The pin with the largest thermal Bottleneck number
is identified and another pin added to its coolest side. The process is then repeated. If the addition of a pin causes an increase in the
thermal resistance then that pin is removed and a pin addition to the next largest thermal Bottleneck location is attempted instead.

The subtraction methodology starts with a tessellated discretized definition of the heatsink. An initial simulation identifies the portion
with the lowest thermal Bottleneck, that portion is removed, the model re-simulated and the change in thermal resistance noted. The
process is repeated, building up a graph relating the change in thermal resistance to the decrease in heatsink mass.

Both methods are implemented using an Excel front end, VBA and FloTHERMs FloXML and FloSCRIPT technologies. The processes
are fully automated, not requiring any manual intervention due to the stability and robustness of FloTHERMs meshing and solver.

mentor.com/mechanical 51
Electrothermal
Simulation Study
by ZFW
Successfully carried out with MicReD
Power Tester and FloEFD
By Christian Rommelfanger, ZFW

Z
FW, Zentrum fr
Wrmemangement (Center
for heat management) based
in Stuttgart, specializes in
comprehensive services in heat
management and lifetime testing of
components and systems. No matter the
industry, our clients all basically want the
same thing: a fast and simple solution
for a specific thermal problem. We rely
on a fast and easy to use CFD system
that is capable of handling different
applications as our consulting projects
are spread through almost every industry
sector, from automotive to production
systems, and usually our clients give
us a call when a project is time critical. Figure 1. Forecast of the power electronics market
Thats why we use congruent CFD
software to give our customers a fast
and reliable answer to their questions.
Besides simulation we offer our clients a
wide range of measurement techniques
and test benches. We think, in a modern
engineering department, simulation and
measurement needs to fit closely together
to work effectively.

In power electronics, it is critical to

have a tight fit between simulation and
measurement. Even small differences in
the simulation can have a big difference
in predicting the lifetime of a component.
Within an industrial project, ZFW conducted
a detailed electrothermal simulation study
of a bridge rectifier as an example for the
use of coupled simulations to get more Figure 2. Level of Detail Model building Result quality

52 mentor.com/mechanical
 Power Electronics

Types of power electronics

Input Output Name Application Example

AC DC Rectifier
E-mobility, renewable energy systems or industrial applications.
Exchange between AC and DC electricity (socket <-> battery)
DC AC Inverter

In mobile phones to maintain the voltage at a fixed value independ-

DC DC Converter
ent from the state of charge

Change of frequency or voltage, for example in power distribution

AC AC Converter
network for change utility frequency 50 Hz to 60 Hz power grids

Table 1.

accurate results for reliability prediction of

powerelectronics.

The forecast of the power electronics market

predicts a rise of 200% in the next ten years.
(Figure 1) There are several reasons for the
increased demand in power electronics,
such as the rapidly increasing market for
e-mobility, the strong demand for renewable
energy, and the uprising market for personal
devices just to mention a few. In most
of these applications clients have high
requirements on the sustainability of the
device.

Keeping in mind that the rule of thumb of a

10 K change in temperature leads to almost
50% change in lifetime, it becomes clear
that even with typical errors from thermal
simulations, the error in the lifetime prediction
can be worse. A 5% error in power losses
within the device can have a serious impact
in lifetime prediction. Usually our customers
have +/- 10% error in their power loss
prediction.

Thats why it is very important for engineers

to have accurate models for reliability
prediction in early stages of development.
Furthermore, it is crucial to have good and
accurate boundary conditions that match the
application. Figure 3. MicReD Power Tester

In a nutshell, power electronics means the Most companies use the basic empirical law is an Arrhenius approach (Equation 2)
application of solid-state electronics to based Coffin-Manson model (Equation 1) and to implement the influence of the average
control electric power. In modern industry, add specific influences in the equation that junction temperature to the lifetime
there are plenty of applications where they observed in experiments. prediction.
power electronics helps to control power.
AC/DC rectifiers are used to change the Nf=a* T-n Ea
Nf =a* Tj-n*e
alternating current of the electricity grid K*Tm
network to direct current for loading the Nf: Cycles to Failure
battery of an e-car. In personal devices a,n: Empirical Parameter Ea: Activation Energy
like mobile phones, DC/DC converters are T: Temperature K: Empirical Parameter
used to maintain the voltage at a fixed value Equation 1 Tm: Average Temperature
independent from the state of charge of Tj: Junction Temperature
thebattery. A typical addition to the Coffin-Manson Equation 2

mentor.com/mechanical 53
The MicReD Power Tester was used to and so on. Which tactic is the right one CoffinManson: Nf = a* T-n
determine the empirical coefficients in the always depends on the specific application.
Coffin-Manson equation. As well as for Assuming the following constants for a
many additional parameters for various given application, that fit the basic Coffin- a= 10000000000
types of customer based lifetime laws. Manson law, we arrive at the results that n = 2,7
There are many different strategies to a change of the Junction Temperature of
determine the constants in the lifetime around 10 K leads to about 44% change in Nf (80 c) = 72720
laws for components, such as constant lifetime (Equation 3). Nf (70 c) = 104288
temperature change or constant current Equation 3

Figure 4. Electro-thermal simulations in FloEFD

Figure 5. Detailed Power loss distribution increases design accuracy

54 mentor.com/mechanical
 Power Electronics

"With these findings from the study, using the combination of the MicReD Power
Tester and FloEFD, reliable simulation models can be created and used in the future."
Christian Rommelfanger, ZFW

As mentioned before, a 10 K change in In the Si-chip itself power loss is 76,1 W. it is important to have accurate temperature
temperature leads to almost 50% change A stand-alone thermal model that would field prediction in the Si-Chip. Ordinary thermal
in lifetime. Because of this relationship use the overall power loss of 81,6 W, as a models give a good insight in the temperature
between temperature and lifetime, it volume source in the silicon would lead to distribution in an electronic system, but when
is critical to have accurate models for an error of 7%. This leads to a temperature it comes to component reliability detailed
electronics. The results of thermal models error of 8,4 K, assuming a junction electric/thermal studies need to be done due
can be improved by coupling the thermal temperature of 120C. Calculating the to their sensitivity to temperature. To simulate
simulation with an electric one. lifetime with the error of 8,4 K would result in the thermal chip behavior, it is very important
a 50% error in lifetime prediction. to have reliable material data. A parameter
Figure 4 shows an overview of how study that compares a measured voltage
electro-thermal simulations works with The temperature dependent electric drop to a simulated voltage drop can help to
FloEFD. The geometry is based on a resistance of the silicon chip is calibrated characterize the material value. The error with
single diode of a bridge rectifier. The electric through a parameter study of the measured a calibrated electro/thermal model can be less
boundaries are three power steps in 20 RSDon in the MicReD Power Tester for a given than 1K if the baseplate temperature is fixed
seconds with a length for each power step current. In the post process of the electric (coolplate) and the heat losses occurs due
of two seconds. The thermal boundary is simulation the voltage drop in the Si-chip Joule heating. The methodology used in this
a fixed temperature at the bottom of the can clearly be seen. example isnt just for diodes it can transferred
device. to MOSFETs and similar devices as well.
In the thermal simulation, the distribution
The 3D electric simulation can predict of the temperature through the diode is With these findings from the study, using the
the actual Joule heating in every part of shown. Notice that the bond wire works combination of the MicReD Power Tester and
the system. Two-way coupling between in this transient load chase, due their FloEFD, reliable simulation models can be
electric and thermal simulation works in thermal capacity as a heatsink. Comparing created and used in the future.
two directions. It enables the direct transfer this simulation with measurement results
from the power losses due to Joule heating in the MicReD Power Tester the error in References
into the thermal simulation (figure 5) and the temperature at the junction is less than 1K. https://www.zfw-stuttgart.de/
temperature for predicting the temperature
dependent electric resistance into the Conclusion
electric simulation. In reliability investigations of power electronics,

Figure 6. Voltage drop

Figure 7. Temperature distribution

mentor.com/mechanical 55
A Unique Design
to Generate UAV
Electrical Power
in Flight
By Guy Wagner, Electronic Cooling Solutions and Travis Mikjaniec, Mentor Graphics

A
s smaller UAVs are designed
with more sensors and
communications technology
for longer missions, the
additional electrical power to run them
drives the need to generate onboard
electric power. One way to create
onboard electrical power would be to
harness the remaining 80% waste
energy produced by the two-stroke
engine.

A team of engineers from Electronic

Cooling Solutions and Ambient Micro
designed, built, and tested an exhaust-
heat thermoelectric generator (EHTEG) that
can be incorporated into a UAV design to
harvest and convert this waste energy into
electrical power in flight [2] .

There are several sources of energy loss in

Figure 1. UAV with the EHTEG attached on top
a small engine all in the form of heat. The
main sources are the heat rejection from
the cylinder and cylinder head, loss from
Designing the Interior airflow on the outside (cool air) and the hot
and Exterior TEGs exhaust inside to estimate the temperature
friction, and the heat of the exhaust stream.
Heat exchangers on the inside of the difference, which enabled them to optimize
Though the heat from the exhaust system is
muffler absorb heat from the exhaust as the internal and external fins of the heat
the only truly usable heat source.
it flows through. The heat passes through exchanger and the number and location of
exchangers to 2inch. square TEGs the TEGs.
The EHTEG had several requirements:
mounted on the outside of the UAV and
finally passes through another row of heat They built engineering models of several
Mechanically robust and integrate into
exchangers to the open air. As the TEGs likely EHTEG configurations and ran them
the aircraft without compromising flight
safety, are exposed to the temperature difference on a test stand using the same engine and
between the hot inside exhaust air and propeller that is used in a MLB Company
Extract the required heat without Bat4 UAV. The models were validated for a
the cool outside air, they generate electric
impairing engine performance, range of operating parameters that simulate
current. The greater the temperature
Provide the largest possible temperature difference, the more current is generated. flight conditions
differential across the thermoelectric
modules while operating within the The team modeled the thermal design The engineers used FloTHERM software
maximum temperature limits, and of the system using Mentor Graphics' that models conduction, convection, and
Designed with minimal weight and FloTHERM Computational Fluid Dynamics radiation as well as the fluid flow of both
aerodynamic drag. (CFD) modeling software [1]. They simulated the external cooling air and the hot internal

56 mentor.com/mechanical
 Aerospace

exhaust gas. They created virtual models of

the exhaust system and performed thermal
analysis and test design modifications
quickly and easily before building any
physical prototypes. The results of the CFD
models correlated well with those obtained
on the engineering test bed.

The engineers used the internal volume

and length for the muffler recommended by
the manufacturer to make the system act
as an efficient expansion chamber exhaust
system. These were used to develop the
first half-symmetry CFD models that would
determine the number of TEGs needed to
optimize the electrical output with minimal
weight. The model is symmetrical, so
building a half-symmetry model reduced the
number of elements down to 1.04 million
cells with no loss in accuracy.

The fin parameters of the internal and

external heat exchangers were kept
relatively constant while varying the location Figure 2. The positions of the external heat exchangers that resulted in the highest average power generation for the system
of the heat exchangers and the placement
of the TEGs. The goal was to maximize the
temperature differential across each TEG
to extract the most heat energy from the
455C exhaust gas.

13 configurations of heat exchangers and

TEGs were modeled in the first optimization
study. They tabulated the power output
from each configuration and chose the best
configuration.

After optimizing the placement of the

TEGs, it was found that the central fins
on the interior heat exchangers disrupted
the exhaust flow and the hot exhaust
gas wasnt reaching the front end of the
muffler. But if they removed the center fins,
the exhaust gas would not channel down
the center and the exhaust pulse would
not reach the front end of the muffler. So
instead of removing the central fins, they
placed them in a semicircular pattern.
(Figure 3) This configuration kept the
exhaust pulse moving through the
center of the muffler and it was able to
curl back symmetrically as the hot gas
flowed back along the outsides and
through the fins.

The interior heat exchangers decreased in

temperature as the hot exhaust gas flowed
from the front through the fins to the rear
of the muffler and then out of the vertical
exhaustpipe.

The FloTHERM simulation was able to show

the exhaust gas flow pattern for the heat
exchanger with the center fins removed.
(Figure 4) The flow pattern is disrupted Figure 3. The interior heat exchanger with the fins in a semicircular shape in the center

mentor.com/mechanical 57
before the exhaust stream reaches the front
end of the muffler.

The simulations showed that when the

semicircular-shaped fins are present, the
flow pattern retains its shape all the way
to the front of the muffler. (Figure 5) This
allowed them to optimize fin placement.

Typically, the outside heat exchangers on

TEGs are placed in a line. This causes an
issue because the units toward the rear of
the external airflow receive more preheated
air than the heat exchangers that are
upstream. This reduces the delta-T across
the heat exchanger and the power output.

Therefore, nine configurations were

analyzed to determine the optimum
fin parameters for the external heat
exchangers. The results were plotted
Figure 4. Thermal simulation shows the exhaust gas flow pattern without fins in the center
against the total power generated by the
TEGs. All of the custom heatsink designs
analyzed out-performed the stock heatsinks
for total power generation, but due to time
and budget constraints, stock heatsinks
were used for the first flight test model. This
resulted in a reduced power output 11W
compared to the custom heatsinks.

The outside TEGs were arranged in four

columns by two rows on each side of the
muffler. The TEGs maintained a cool side
temperature below 58.3C with external
air at 18 C and 22.3 m/s velocity, while
keeping the hot side temperature below the
maximum allowable temperature of 225C.
(Figure 6) The outside heat exchanger
temperature ranges from 26.1C on the
leading edge of the front fins to 58.1C on
the base plate next to the hottest TEG.

Some of the first simulations demonstrated

that the heat loss through all other surfaces
of the muffler had to be minimized to
Figure 5. Thermal simulation shows the exhaust gas flow pattern with fins placed in a semicircle in the center
maximize heat flow through the TEGs for
maximum power generation. Mineral wool
insulation was used on all the exposed
surfaces to minimize the heat loss. Mechanical Design temperature thermal interface material
The mechanical design of the EHTEG had at each interface in the thermal path to
two main requirements: maximize heat transfer from the inner
The power output of the EHTEG system
heatsinks through the thermoelectric
was modeled by summing the power
Able to support the thermal components generator modules and the outer heatsinks.
contribution of each pair of TEG modules.
and allow adequate flow of cold air Figure 8 shows an exploded view of
By first calculating the hot side and cold
side temperatures of each TEG pair, these around the external heatsinks, and theEHTEG.
values could then be used to compute the Be lightweight with minimal frontal area
open circuit voltage. The harvesting and to reduce aerodynamic drag. Converting to Usable Electricity
power conditioning circuitry matches the The thermal environment of each
equivalent series resistance for maximum The heatsinks were fastened from the outer thermoelectric module was slightly different
power transfer; thus, the voltage of the (cold side) to the inner (hot side) forgings. because of its location on the EHTEG, so
load resistance is exactly half the open The TEG modules were sandwiched each modules output was also different
circuit voltage. This data defines the power between the outer heatsinks and the from its neighbors. The input interface
harvested per TEG pair. EHTEG sidewalls. They used a high- modules received the output from a pair of

58 mentor.com/mechanical
 Aerospace

modules and converted the input voltage to

12 V. The input module also automatically
adjusted its input impedance to match the
source impedance, thus operating at the
maximum power transfer point.

The outputs of the input modules were

combined and fed to a single 12-V bus
regulator that provided a regulated 12-V
output to external loads. An electronic load
was included in the power conditioning
electronics for testing purposes. The
electronic load automatically adjusts its
resistance to extract the maximum power
available from the EHTEG system. The Figure 6. Cool-side temperatures of the outside TEGs.
data I/O board provided voltage levels
proportional to selected voltage and current
levels for input to the onboard data-logging
system.

In Summary
The team used FloTHERM to create virtual
models of the exhaust system, analyzing
various design configurations quickly before
building any physical prototypes. And the
results of the CFD models correlated well
with those obtained on the engineering
test bed. Since the TEGs are actually
thermoelectric coolers run in reverse, their
efficiency is only around 5%; that is, 5% of
the heat energy flowing through is turned
into electricity. If this efficiency rate can be
doubled, the technology could be used in
many practical and profitable applications.
New commercial opportunities are spurring
interest in thermoelectric power generation.
The design techniques described here
could be used to develop much higher
power output thermal energy harvesting
power systems.

References
[1] FloTHERM, http://www.mentor.com/ Figure 7. Single-side power output (W) for the flight configuration
products/mechanical/products/flotherm
[2] J. Langley, M. Taylor, G. Wagner, and S.
Morris, Thermoelectric Energy Harvesting
from Small Aircraft Engines, SAE
International, 2009.
[3] Marco Nuti, Emissions from Two-Stroke
Engines, Society of Automotive Engineers,
Inc., Chapters 7 and 9.
[4] Combustion Products Applet: http://
www.engr.colostate.edu/~allan/thermo/
page9/Combustion/combustion.html, Allan
T. Kirkpatrick, Colorado State University,
Fort Collins.

Figure 8. The inside and outside of the EHTEG

mentor.com/mechanical 59
Comparing
FloMASTER ORC Results
to Experimental Data
By Doug Kolak, Technical Marketing Engineer, Mentor Graphics

T
here is a strong current in the
automotive industry today, pulling
the future of vehicles more and
more towards electrification. While
this is a certainly a good direction, not only
because of the need for lower emission
rates and the reduced reliance on fossil
fuels; there are still several obstacles
that need to be overcome before the
roads are dominated with these vehicles.
As engineers continue to tackle these
challenges, it is important to remember
there are still opportunities to improve
vehicles that use the internal combustion
engine. Even with all of the optimization
that has gone into these components in Figure 1. Brunel ORC Test Rig
the past, approximately 60% of the fuel
energy is lost through heat, friction, and and employed. For this reason, it is critical that
exhaust loses [1]. a tool can quickly and accurately model these
complex systems for evaluating designs and
There are a number of techniques currently rapidly prototyping new concepts.
being investigated that deal with capturing
the energy of the exhaust gas. There are With the release of V8, FloMASTER has
several options that fall under this umbrella of introduced functionality to fully model ORC
Waste Heat Recovery (WHR), these include: systems so that engineers can consider key
Exhaust Gas Heat Recovery, Thermo-electric design parameters such as available superheat,
Generators, and Organic Rankine Cycle (ORC). high and low side system pressures, ORC
Each of these options does pose a number fluid selection, pump type and performance,
of challenges to system engineers, and this and heat source and expander selection.
is particularly true of ORC systems since they The tool was developed in conjunction with
are effected by both component and system Brunel University in the UK, who ran extensive
layout design. However, when compared tests to evaluate accuracy and robustness.
Figure 2. FloMASTER ORC System Model
to other exhaust gas energy capture Installation of rig based pressure, temperature,
technologies, ORC has been shown to achieve and flow instrumentation has demonstrated
heat to the fluid. The overall design goal of
efficiencies as high as 16% in automotive that FloMASTER is capable of predicting
the system is that it must be able to extract
applications [2]. overall system thermal efficiencies within 2%
5 15 kW of energy from the turbine without
of expected values. In addition to the work
having negative effects on the EGR or exhaust
ORC systems are typically designed to with Brunel, FloMASTER can be compared
systems. As shown in figure 2 the system is
recover waste heat from a process and to experimental work carried out by AVL
laid out as follows:
subsequently used to heat a working fluid until Powertrain Engineering Inc. and published by
it is a superheated vapor. The vapor is then SAE [3].
The pump is a positive displacement type
sent through an expansion device such as a
pumping, delivering the required flow rate
turbine, so that useful work can be extracted. The system represented by the model shown
based on the pressure drop of the system,
To effectively design an ORC system, it in figure 2 is a typical ORC system that would
requires that individual components are sized be found on a heavy duty truck. Due to its The fluid exits the pump and enters the
accordingly, that an appropriate working fluid common availability, ethanol is being used recuperator where it is heated by hot
is selected, and that system interactions, as the working fluid, and both the exhaust ethanol coming out of the turbine,
including a control strategy, are understood gas and the EGR gas are being used to The warmer fluid then enters the evaporator

60 mentor.com/mechanical
 Technical Article

Figure 3. ESC Test Points

where it is heated by the EGR gas until it

changes phase to a vapor,
The vapor then enters the superheater
Figure 4. B50 Test Point Temperatures
where it is heated further by the hot exhaust
gas,
The superheated vapor then enters
the turbine where it is expanded. This
expansion turns pressure energy into
rotational energy which can then be
translated into electrical energy by an
attached generator,
The still superheated vapor then enters the
other side of the recuperator, warming the
incoming liquid phase ethanol, and
The ethanol then enters the condenser
where it changes phase back to a liquid Figure 5. All Test Points Heat Inputs
to begin the thermodynamic cycle all over
again. simulation, is never greater than 1.7%, Figure 4
shows the temperatures around the model for
The FloMASTER model was constructed using the B50 test point. The heat input across the
all out-of-the-box vapor cycle components simulations also shows very close correlation
with no calibration required. For the with the greatest deviation from the B50 test
recuperator, a plate frame heat exchanger was point at 3.34% with all cases shown in figure 5.
chosen, while a shell and tube component was
selected for the other three heat exchangers. The other critical factors for the system is
the power generated and the overall system
As part of the study a number of simulation efficiency. Table 1 shows the values for
conditions were run based on the European the B100 test case. Overall the study and
Stationary Cycle (ESC). The European simulations match extremely well except for Table 1. Power Generated and System Efficiency
Stationary Cycle (ESC) is currently used across the pump power, however this value is so
Europe and is a part of the Euro III emissions small in the overall system it has minimal effect. and Shimokawa, K., (2014). A Study of the
regulations. The European Transient Cycle This design for the B100 test point is able to Rankine Cycle Generating System for Heavy
(ETC) and the European Load Response (ELR) deliver 13.54 kW assuming a 75% efficiency Duty HV Trucks. SAE Tech Paper, 2014-01-
test cycles are also part of the 1999/96/EC on the turbine, well within the desired range 0678, doi:10.4271/2014-01-0678.
directive. The ESC consists of 13 test points of the study. This leads to an overall system
efficiency of 11.2%, both quantities well below [2] Park, T. Teng, H. Hunter, G. Velde, B.
which are all run in steady-state. Figure 3 Klaver, J. (2011) A Rankine Cycle System
shows the test points recorded onto a graph 2% variation from experiment.
for Recovering Waste Heat from HD
of load vs. engine speed with the focus of the Diesel Engines - Experimental Results,
study on points B50, B75, B100, A100, and Developing ORC systems is both time
consuming and expensive, but being able to SAE Int. Publication 2011-01-1337, doi:
C100. These test points determined both the 10.4271/2011-01-1337.
temperature and flow rate of the exhaust gas frontload as much design work as possible
and EGR gas which directly determines the through the use of 1D system simulation, such [3] Teng, H., Klaver, J., Park, T., Hunter, G.,
amount of heat energy available to the system. as with FloMASTER, can give companies the van der Velde, B. A Rankine Cycle System
advantage of fewer physical prototypes and for Recovering Waste Heat from HD Diesel
Overall the agreement between the simulation faster time to market. Engines - WHR System Development,
results and the published experimental results SAE int. Publication 2011-01-0311, doi:
is very good. The variation in temperature References 10.4271/2011-01-0311.
versus published results across all of the [1 ]Furukawa, T., Nakamura, M., Machida, K.,

mentor.com/mechanical 61
Figure 1. Center Information Display Module

Thermal Design
Approach for Automotive
Display Integration
By Clemens J.M. Lasance, Philips Research Emeritus, Consultant SomelikeitCool

I
nnolux Corporation is a world leading does not solve the Navier-Stokes equations. conditions, coupling of elements to housing,
manufacturer of TFT-LCD displays, Furthermore, its radiation treatment was not decoupling LEDs from light guide, properties of
and supplies customers that include well-documented and caused further concern. the display stack, etc. These models were also
many of the world's leading information Hence, as a starting point for the numerical to compare with dedicated tests to determine
and consumer electronics manufacturers. analysis the CFD code FloTHERM from Mentor several unknowns, specifically the real-life
Innolux Corporation employs more than Graphics would be used. boundary conditions and thermal interface
68,000 employees worldwide and is a resistances.
leader in the global optoelectronics industry. The main objective was to provide preliminary
Innolux actively recruit and train R&D talent models based on the available (but not To get an idea of the CFD output, figure 2
to consolidate its prominent status in the sufficiently accurate) input data to study the depicts three screenshots for the three cross
industry. Innolux is striving to realize the feasibility of the proposed complete system sections through the hottest point. The module
ultimate in visual infotainment. Their unique as a sound starting point for subsequent is mounted vertically, showing the internal
intelligent management platform has not discussions regarding design decisions with airflow, for a total dissipation of 18.5W.
only enhanced the company's management the customer. Tests on an available prototype
capabilities, but also provides customers were performed early on to provide a first- Results were made available to the customer
with prompt, accurate delivery information. order comparison with the numerical models. for several system test conditions, heat
In a later stage, experimental calibration by transfer coefficients on the external surfaces
Engineers at Innolux Technology Europe B.V. means of dedicated tests, such as thermal to represent natural convection, with and
were confronted with the challenge of how resistances and thermal conductivities and without radiation, various choices of gap pad,
to solve expected thermal problems with the component thermal data, were required due PCB data, thermal conductivity changes,
integration of a display-centered system into to the inaccurate thermal data delivered. One etc. including component temperatures,
a high-end automobile. The thermal design key challenge was that the system was being which should be interpreted as local solder
philosophy behind the activities to be taken is designed to operate in an environment where temperatures. An additional temperature rise
the subject of this article. the ambient could be as high as 70C, while caused by the internal thermal resistances of
the testing had to be carried out at room the components should be added, but this
A thermal model was already available based temperature. is only appropriate when reliable component
on the code CST Microwave which had thermal data is available.
an embedded thermal solver, with detailed The numerical models were developed as
MCAD data as input. However, there were a tool for optimization: sensitivity analysis A comparison was made between the existing
some doubts concerning its accuracy for of gap pads, wall thicknesses and thermal model created using CST Microwave and
this application due to the fact that the code conductivities, PCB thermal data, boundary Mentor Graphics FloTHERM. After a series

62 mentor.com/mechanical
 Automotive

of trials, a reasonable match was obtained

between the two codes, for identical input
data and boundary conditions. However, there
were a number of reasons for not continuing to
proceed with CST:

The CST model does not solve the Navier-

Stokes equations, hence one never knows
the consequences except by comparison to
a code that does. Better to continue using
the CFD code. Knowing the errors arising
from the assumption, it was possible to
continue in conduction-only mode to speed
up the process of sensitivity analysis; Figure 2. Typical results with flow, total dissipation 18.5W

How CST solves the internal radiation

transfer is unclear, and subject to doubt; and system varies throughout the system so there Recommendations
On the issue of importing CAD files: the is no one temperature a component sees as How to retrieve the right thermal data for all
author is not in favor of using brute force its own ambient. Another challenge is that the critical components, including the maximum
when trying to get insight in thermal customer cannot control the design of the junction temperature, how this is to be
problems. While modern tools make it environment, in this case a car dashboard, determined, and exactly now useful thermal
easy to import CAD files, the not-so-easy into which this product fits, so defining the data such as Rjc and Rjb are generated,
next step is to get rid of all mechanical and boundary conditions for Innoluxs system is the requires investigation and discussion. It is
electrical details that are useless for thermal responsibility of their customer, not Innolux. highly recommended to ask the vendors of all
modeling. While the speed of modern critical components for more accurate models.
computers may allow millions of cells, the It would be much better for the supplier For many components even basic but useful
consequences are twofold: the layout to define a unique point at the outside of thermal resistances such as Rjb and/or Rjc are
becomes very complex, hindering insight their component or module where a given not available. As a customer, and especially if
in what is going on, and the CPU times go temperature should not be exceeded. Such a you build high-end or high reliability systems,
through the roof, hindering fast optimization. spec is independent of the application and so you are entitled to get the right data, and
avoids this issue. especially when you will be held responsible for
Comparison with Experiments system failures!
A number of experiments were available for an Tests with a Dummy System
existing prototype. While the author does not Tests with a dummy system should ideally Conclusions
recommend to use general results for checking focus on just a few issues, the most important The main conclusion is that, provided some
the accuracy of the numerical model, in this being the real-life external boundary conditions measures are taken, the estimated temperature
case, because the results are already available, for the complex display. Because of the specs (as far as they are known), could be met,
a comparison gave some idea of trends, complexity of the layout it does not make sense given the current dissipations and assumed
and could be used to extract some average to model every detail, because it makes the boundary conditions. However, it could be
boundary conditions in the lab. After calibration fitting considerably more difficult. Unfortunately, foreseen that under certain circumstances
of the boundary conditions, the model and the this principle is often forgotten. Determining these conditions would be worse, to such
experiments matched to within 2C. the boundary conditions that govern real-life an extent that natural convection only cannot
applications is always a challenge, because prevent overheating of the display at maximum
Under normal operating conditions the to get sufficient accuracy one should simulate ambient temperatures, and hence a forced
SOT solder temperature was above the a very big part of the environment wherein convection solution should be explored.
recommended 100C, the rest of the the device is operational, including the However, with the available numerical models
components appeared to be acceptable. The radiators, the windows, the people etc. noted of the system, it should be easy to check the
importance of radiation was tested, and found above. To avoid these problems, it is highly consequences of such design changes.
to be making a significant contribution to the recommended to impose a heat transfer
cooling. Fixing a strip of gap pad material coefficient, including radiation, to be calibrated In the opinion of the author, there is no other
between the LEDs, bracket, housing and with real-life measurements. way to reach these conclusions within a
carrier had limited effect, however introducing a realistic timeframe except by using a CFD-code
gap pad below between the bottom and main Gap Pads combined with dedicated tests to confirm
PCBs reduced the temperature by up to 6C It is clear from the analyses that using gap uncertain parameters required for the modeling.
and hence is considered effective. The results, pads or gap fillers improves the thermal The biggest hurdle to improving the speed and
particularly the SOT temperature, was found performance. Be aware that the thermal data accuracy of the design process is the lack of
to be sensitive to the assumed PCB thermal supplied by the manufacturers should be accurate data for the boundary conditions for
conductivity. Due to uncertainties in the model treated with caution, and gap pad data are not the system, coupled with the widely-spread but
data, a gap pad was introduced. an exception unfortunately. The test method incorrect habit of prescribing maximum ambient
usually applies far too high pressure, resulting temperatures for the components.
As in many other cases encountered, we were in too optimistic values. So, if critical, perform a
faced with a specified maximum ambient for dedicated test, ideally using Mentor Graphics
a module or component. This is impossible DynTIM system within which the pressure can
to design to, as the temperature within the be accurately controlled.

mentor.com/mechanical 63
Predicting Water
Treatment Plant
Performance with
FloMASTER
By Jurgen Sprengel, Principal, JS Pump and Fluid System Consultants

W
ater treatment systems
are designed to deal with
feed water quality. They
comprise mechanical
sub-systems made up of pumps, filters
and tanks which are connected by
valves and pipework generally installed
in a series arrangement. Chemical
treatment forms an essential part of
a water treatment process. However,
the fundamental driver is the hydraulic
process in producing the required
water conditions to enable successful
chemical treatment.

This particular Water Treatment Plant

(WTP) design is one that treats saline water Figure 1. WTP Block Diagram
produced from the coal seam methane gas
extraction process. It receives raw water
from the gas fields water gathering system
which typically contains between 4000 -
8000 mg of salt per litre. The raw water is
collected in a storage dam adjacent to the
WTP. The treated water and brine that result
from the treatment process are temporarily
stored in local storage dams. The treated
water is then dispatched for beneficial use
on an as-need basis, while the brine can be
further treated in a brine concentrator.

The WTP is a 20 ML per day WTP

comprising four duty Reverse Osmosis (RO)
Trains and one standby RO Train, designed
to achieve a treated water recovery rate of
85%. The hydraulic envelope of this study
comprises the following sub-systems: pre-
Figure 2. RO Train Hydraulic Sub-Model
treatment, filtration, hardness removal (ion
exchange), reverse osmosis, post-treatment developed. The membrane feed pressure installed at the RO Stage 3 brine outlet to
and storage. required to achieve the target recovery rate enable realistic modeling of a RO Train start/
for the selected raw water composition was stop event, it being controlled by a PID type
The RO Train Hydraulic Sub-Model obtained from the manufacturer. Variable controller set at the target reject flow rate.
Initially, a single RO Train Hydraulic speed controlled radial flow pumps were
Sub-Model (figure 2) including pumps, selected to produce the required pressures. The RO Train Hydraulic Sub-Model was then
membranes and reject control valve was A modulating Reject Control Valve was tested and optimized in isolation to ensure

64 mentor.com/mechanical
 Process

its functionality would meet all operational

requirements, both in steady state and
transient operation.

The Submerged Membrane

Filters Hydraulic Sub-Model
Submerged Membrane Filters are located
inside flooded cells with cell water level
and filtration flow rate being controlled
independently.

Raw water supplied by the upstream Inlet

Pumps passes through the Disk Filters and
enters the Submerged Membrane Filter
Cells. The cell inflow rate is controlled by
the Cell Inlet Valve via feedback controller in
order to maintain individual cell water level
set point. The downstream Filtrate Pump
extracts water through the internals of the
Submerged Membranes, its shaft speed
controlled by the water level set point in
the IX Feed Tanks. The water volume in
the cells allows for some buffer to even out
Figure 3. Submerged Membrane Filters Hydraulic Sub-Model
any short-term upstream and downstream
pump flow imbalances.

Backwashing of Submerged Membranes

is achieved by stopping the Filtrate Pump,
closing Membrane Outlet Valve and Cell
Inlet Valve and then starting the Backwash
Pump. As the Backwash Pump ramps up,
drawing water from the IX Feed Tanks,
the Membrane Backwash Inlet Valve and
Backwash Outlet Valve start opening, both
valves controlled by Backwash Pump
speed. Backwash waste water from the
submerged Membranes is directed to the
Site Pump Station for further processing, its Figure 4. Pump Set Ramp Speeds
clean water content being discharged to the
Raw Water Storage Dam.

The WTP Integrated

Hydraulic Model
The WTP Integrated Hydraulic Model
shown in figure 6 is essentially made up of
a multiple of sub-models. Auxiliary sub-
systems and control loops were added to
create a complete hydraulic model, now
consisting of almost 370 hydraulically active
components and including a total of 51 Figure 5. Sub-system flow rates Figure 6. Tank water levels
centrifugal pumps.

Pumps and valves were modeled based on

their actual performance characteristics.
Where not available from manufacturers,
performance information was utilized as
detailed in Fluid Flow Systems Second
Edition by D.S. Miller, BHRA 1990.

This model captures all essential elements

comprising the complete water treatment
system. FloMASTER allows the designer/ Figure 7. RO Pump Discharge Pressure Figure 8. Reject Control Valve: Reject Flow Rate, Valve
operator to identify any control deficiencies Position and Upstream Pressure

mentor.com/mechanical 65
or potential bottle necks in the overall sequence was reversed after 3,600s of followed by a Disk Filter backwash was
process. What-if scenarios can be quickly operation, resulting in the shutdown of simulated with four RO Trains in service.
analyzed and optimized such as the RO Train 4 and ongoing system dynamic This resulted in a temporary system flow
system response to certain equipment settling, eventually achieving stable system imbalance as the Filtration System capacity
failurescenarios. operation with three RO Trains in service. was reduced from 100% to about 75%.
Figure 5 shows the upstream sub-systems Normally automated plant operation of the
gradually adjust to rapid downstream flow Submerged Membrane Filtration System
The Operational Simulations
rate changes. Tank water levels (figure 6) would initiate a rotating backwash cycle
Two common transient operating scenarios
vary by a maximum of about 10% during for all active filtration units. However, in
have been selected for further analysis: RO
the RO Train Start/Stop cycle. The system this case the backwash cycle for one
Train start/stop and Submerged Membrane
dynamic settling period is about one hour filtration unit was manually initiated by
Filter and Disk Filter backwash cycle.
on RO Trainstart-up. setting Filtrate and Backwash Pumps start-
up and shutdown times as wells as the
RO Trains are started and stopped in
While pressure ramp rates for the RO HP opening and closing times for the Disk Filter
response to the volume of raw water being
Pump and RO Stage 2 Booster Pump BackwashValves.
available at the time. This is considered to
are acceptable, the ramp rate for RO
be a day-by-day operating scenario in the
Stage 3 Booster Pump is in excess of the The main purpose of the Submerged
life of a modern WTP.
acceptable membrane pressure loading rate Membrane and Disk Filters backwash
and hence does require further adjustment. simulation was to:
Regular Submerged Membrane Filter
backwash cycles are an ongoing Establish acceptable water level
The Reject flow rate, valve position and
requirement to keep the membranes at a variations in membrane cells,
upstream pressure in figure 8 all indicate a
sufficient level of cleanliness not to impact
smooth Reject Control Valve response to an Understand the impact on storage
on overall plant production capacity.
increasing upstream pressure. volume variation and tank capacity,
Determine the system recovery time and
The RO Train Start/Stop
The Submerged Membrane and maximum backwash cycle length, and
Simulation
The RO Train start/stop sequence was
Disk Filters Backwash Simulation Identify potential control system
A Submerged Membrane Filter backwash improvements.
manually initiated by setting fixed pump
start-up and shutdown time lines.

The main purpose of this RO Train Start/

Stop simulation was to:

Identify any excessive RO membrane

ramp pressure issues,
Understand the impact on storage
volume variation and tank capacity,
Identify any pressure surge issues,
Establish the overall dynamic system
settling period, and
Identify potential control system
improvements.

For this simulation of a total duration

of 6,000s, it was assumed that all
components in the hydraulic model would
be in service except for stand-by units
such as RO Train 5. A time step of 0.5s
was selected, sufficient to capture transient Figure 9. Cell Inlet Valve Position vs. Cell Water Level
events such as pump speed ramp and
Reject Control Valve position response to
PID controller input

The start-up simulation was initiated with

RO Trains 1 to 3 in service, each train
operating at 5MLD of raw water inflow.
RO Train 4 LP Pump was operating
at a shaft speed of 800rpm to provide
membrane flushing water with the Reject
Control Valve fully open. RO pumps were
then ramped up as shown in figure 4,
reaching their duty speed after 150s. This Figure 10. Submerged membrane cell water levels Figure 11. Tank water levels

66 mentor.com/mechanical
 Process

For this simulation a total duration of backwash cycle. However, the variance is especially in the event of frequent RO Train
5,000s was selected at a time step of 1s, 5% during disk filter backwash, both within start-ups. All other RO pump ramp rates
sufficient to capture transient events such acceptablelimits. have been confirmed acceptable.
as pump speed ramps and valve position
adjustments. Tank water levels vary by a maximum of Figure 5, 6 and 11 track the sub-system
about 25% during the backwash cycle. A flow rates and tank water levels throughout
Cell Inlet Valve position control is shown in period of about 2,000s is required for the the modeled transient events. Because of
figure 9. IX Feed Tank level to recover while the RO a linear relationship applied between water
Feed Tank level is almost unaffected. tank level and upstream pump speed,
Before the Filtration System backwash a relatively slow dynamic settling period
sequence was started, all four RO Train Conclusions hasresulted.
were operating continuously at rated Simulations showed that the modeled
capacity of a total raw water inflow rate of system design backwash cycle does have a In order to manage complexity, the hydraulic
20ML per day. After 300s of steady state greater impact on tank level variations than model incorporated only relatively basic
operation a Filtrate Pump shutdown was a RO Train start-up. Also confirmed was the system controls. A control system looking
initiated followed by a Backwash Pump conservative selection of tank sizes as the further upstream and forward feeding
start 60s later. This backwash operating maximum water level variation was less than flow and level signals would result in more
was reversed after 1,140s resulting in a 25%. However, further tank size reduction responsive sub-systems within the water
submerged membrane backwash cycle would be feasible and cost effective from a treatment plant.
duration of about 15 minutes. At 1,800s hydraulic perspective.
a Disk Filter backwash was initiated and
completed at 3,000s, a total duration of Figure 7 indicated the pressure ramp rate
about 20 minutes, this action concluding the provided by RO Stage 3 Booster Pump
backwash simulation. being in excess of the maximum acceptable
membrane pressure ramp rate of 0.7Bar
Cell water levels vary by less than 2% per second. If left uncorrected this situation
duringthe submerged membrane may lead to premature membrane wear,

Figure 12. WTP Integrated Hydraulic Model

mentor.com/mechanical 67
Geek Hub

Figure 1. Industrial example of two stacks

Spiralled Skyline
E
ver wondered why some For low Re numbers the vortices downstream
industrial stacks have spiral the cylinder are steady (figure 2), for higher Re
structures snaking skyward? numbers >100, the vortices begin to oscillate
By Mike Gruetzmacher, Well, since these are industrial (figure 3). The Strouhal number describes the
Technical Marketing plants, designer or artistic reasons can be frequency of the vortex shedding:
Engineer, Mentor Graphics eliminated. So it must be scientific.
St= fd
The spirally arranged extensions are called U
Scruton strakes, named after the British f=frequency of vortex shedding
engineer, Christopher Scruton. Working d=hydraulic diameter
together with D. E. J. Walshe at the National U=flow velocity
Physics Laboratory in Great Britain, he
invented the helical strake and first published These alternating shedding of vortices produce
the results in 1957[1]. These helical strakes periodic forces across to the flow direction.
are used to prevent vibrations caused by Each elastic component, including industrial
alternating vortices caused by the wind. stacks, has a natural frequency. If the excitation
frequency generated by the vortex shedding
Vortex-induced vibrations (VIV), are caused approaches the natural frequency of the stack,
by vortex shedding and can occur when resonance and vibrations can occur. This can,
long slender structures such as chimneys at worst, cause damage to the structure.
or car antennas are exposed to a flowing
fluid. The repeated pattern of wind causes According to Scruton, another dimensionless
alternating eddies or vortices, also known quantity was named, the Scruton number:
as a Krmn vortex street. But what do
those helical strakes actually do and why are 2Sme
Sc=
theyneeded? pb2ref
S =structural damping by the logarithmic
For the purposes of this investigation, I decrement
used Mentor Graphics FloEFD 3D CFD me =effective mass per unit length
Analysis Software, to better understand p =density of the fluid
thesestructures. bref =characteristic width of the structure

68 mentor.com/mechanical
The Scruton number is an indicator for the
design engineer to assess whether further
measures must be taken to avoid the effects
of vibration damage. Constructive measures
may include, for example, damping or other
structural engineering measures. Massive,
stable stacks seem to be more resistant
against these effects, as seen in figure 1. The
stack with the larger diameter on the left side is Figure 2. Steady flow at low Re number Figure 3. Oscillating at Re>100
not equipped with the Scruton spirals.

Focusing on the fluid dynamics, the aim

of the helical strakes is to interrupt the
alternating shedding of the vortices and
thus a suppression of the oscillating lateral
forces. I was able to quickly set up a simple
FloEFD model with the following boundary
conditions: Stack diameter: 1.5m, ambient
air velocity: 2 m/s (wind velocity 7.2 km/h), as
seen in figure 4.

Figure 5 shows a cut plot for velocity and at the

top left a curve of the lateral forces depending
on the time. We can see the oscillating force in
Figure 4. FloEFD simulation for one slice
y-direction caused by the alternating vortices.

In comparison, a further variant, which is

equipped with exemplary helical strakes. This
is demonstrated in figure 6. As we can see in
the velocity plot, the alternating vortices dont
occur anymore and the curve does not show
the oscillating course.

The example in figure 6, has three helices, each

120 degrees apart around the cylinder with
a pitch of 7m and 1.2 revolutions each. The
helices are not constructed all the way to the
Figure 5. Stack without strakes Figure 6. Stack equipped with helical strakes
top, since in the upper region the vortices are
already interrupted by the flow above the stack.

Comparison of the lateral force, time

dependent can be seen in figure 7.

As we can see in the graph in figure 7, the

stack that is equipped with the helical strakes
shows much lower lateral forces and does
not oscillate around the x-axis. These are,
of course, exemplary results for a slice of
the stack, but the qualitative difference is
clearlyvisible.

Thus, the Scruton strakes seem to present

an economical solution as opposed to a Figure 7. Time dependent lateral forces
structural reinforcement of the stack, while
at the same time minimizing additional load. technology to avoid the damage of protection (Great Britain), 1957.
Increasing mass often provides remedies, but tubes, in oil rigs, car antennas and electrical
https://www.youtube.com/
clever solutions are often the most efficient transmission towers.
watch?v=rlpUhgfEZPU
and here we could explain and present the
clever solution using FloEFD with its SmartCells References http://www.helicalstrakes.com/
technology. [1] C. Scruton and D.E.J. Walshe, A Means https://en.wikipedia.org/wiki/Scruton_number
for Avoiding Wind-excited Oscillations of
Additional applications of the described Structures with Circular Or Nearly Circular https://en.wikipedia.org/wiki/Strouhal_number
Scruton strakes can be found in measurement Cross-section, National Physics Laboratory

mentor.com/mechanical 69
Brownian Motion...
The random musings of a
Fluid Dynamicist
Brownian Motion or Pedesis (from Greek: Pess
'leaping') is the presumably random moving of particles suspended
in a fluid (a liquid or a gas) resulting from their bombardment by
the fast-moving atoms or molecules in the gas or liquid. The term
'Brownian Motion' can also refer to the mathematical model used
to describe such random movements, which is often called a
particle theory.

Do you Require a Receipt?

R
ecently I managed a very brave
thing, I left the gas / petrol
station without a single piece of
paper. I expected to walk across
the forecourt back to my car and have
someone demand that I prove I had paid
for my fuel. And of course without that
piece of paper I wouldnt be able to

Is this a generational thing or is the paper

comfort blanket important for any age? In this
digital age with pocket scanners and storage
on every connected mobile device, paper
surely isnt necessary? Most of us travel for
work and have endless receipts for hotels,
meals, taxis and other subsistence items
and I am willing to bet we all have electronic
expense systems. We return to the office and
scan all those pieces of paper back into digital
systems not long after they were output onto
paper from digital systems. Yet we still stand
at a hotel reception have them print out our
details to sign and check out and wait for the
often malfunctioning printer to deliver a receipt
to prove we stayed and paid. Or diligently take
that little piece of paper at every checkout
kudos to the gas station that offered me
the ability to walk away without that piece of
paper.

I Googled paper receipts, so I accept the

source is questionable, but in the US along
with 250 million gallons of oil, 10 million trees
and one billion gallons of water are consumed
just to make those thermal receipt rolls. That is
without the receipts printed on regular paper.
Staggering, so based on this alone perhaps
the days of paper are numbered, or maybe
we will always feel like we stole it if we dont
tuck a piece of paper into our wallets.

70 mentor.com/mechanical
Higher Education
Program Training the
Next Generation
of Engineers
The Mentor Higher Education Program helps engineering students by providing
simulation software, technical support, training and resources during school and
post-graduation. Open to all educational institutions, software is donated* and
gives students full access to latest Computational Fluid Dynamics, Electronics
Thermal Design, and Fluid Systems Simulation software used in industry.
HEP members have access to:
As many copies of the software as needed,
The same customer support services as corporate customers.
Mentors advertised public training classes on a space-available basis, and
on-demand training (for nominated faculty staff members), and
Online product demonstrations and tutorials
For more information and a HEP membership online application form visit
www.mentor.com/company/higher_ed/

* Software is donated and levies a nominal annual support charge based

on the design package(s) used, irrespective of the number of licenses.

mentor.com/mechanical 71
 tion 3
a
ul

D
Sim

CF
D
FD

Sim
sC

ulat
em

ion
1D Syst

Design
Optimization

Te n
s ti n g t io
Th er m r i z a
al Ch ara cte

Integrated Solutions
from Mentor
www.mentor.com/mechanical

72 mentor.com/mechanical