Newsletter
Simulation Based Engineering & Sciences
Year
Trapped-vortex approach for
syngas combustion in
gas turbines
Extreme Ensemble
under Uncertainty - Optimization
of a Formula 1 tire brake intake
Fluid Refrigerant Leak in a Cabin
Compartment: Risk Assessment
by CFD Approach
Research of the Maximum
Energy Efficiency for a Three Phase
Induction Motor
9
n°2 Spring 2012
International CAE Conference
21-22 October 2012
Reducing Fuel Consumption,
Noxious Emissions and Radiated
Noise by Selection of the Optimal
Control Strategy of a Diesel Engine
Newsletter EnginSoft Year 9 n°2 -
3
Flash
While the warm summer waves surround us, the teams of
EnginSoft and the company’s global Network are preparing
one of the major annual CAE summits: the International CAE
Conference which will take place from 22nd-23rd October
2012, in Pacengo, Lazise (Verona) – Italy.
With this Newsletter, we extend a warm invitation to our
readers to join us for a most productive and creative Gettogether in Northern Italy!
The Conference, which has become a brand in itself over the
past 28 years, will present the state-of-the-art of CAE in
diverse industries. Under a networking “umbrella”, the
program will feature the latest innovations in the use of
simulation and engineering analysis, as well as the various
initiatives of EnginSoft and its international partners.
Speakers from leading companies in product development
will demonstrate how the use of CAE enhances efficiency
and delivers ROI.
A small foretaste is
presented in this
Newsletter, which as
you may note, has a
slightly
different
appearance,
nine
years after its first
edition. To reflect our
international network,
our role and culture,
EnginSoft has also
renewed the layout of its website and brochures. We hope
that you like our new look, and we welcome your feedback!
This issue reports about ENEA’s trapped vortex approach for
syngas combustion. We look at how the optimal control
strategy of a diesel engine helps to reduce fuel
consumption, emissions and radiated noise. Fabrizio
Mattiello of Centro Ricerche Fiat outlines the Center’s risk
assessment with CFD. We hear how optimization can perfect
the design of a high-performance engine’s exhaust pipes
and a steel case hardening process. This issue also presents
Rossi Group’s optimized motor which fulfills the criteria for
the IE3 premium efficiency class. Standford University,
University of Naples, EnginSoft Americas illustrate their
work for a three-dimensional geometry using an extreme
ensemble. We learn about high pressure die casting, one of
EnginSoft’ s key expertise areas, and the importance of
structural CAE for the development of today’s appliances.
ANSYS and EnginSoft maintain excellent collaborations with
the food&beverage industries. For this issue, we interviewed
Massimo Nascimbeni from Sidel, one of the world’s leading
groups for packaging solutions for liquid foods.
Our software news highlight ANSYS 14 and the ANSYS
product family, HP and NVIDIA Maximus, and their benefits
for ANSYS Mechanical. We introduce EUCOORD, a web-based
collaborative solution for FP7 EU Project Management, and
the new Research Projects that EnginSoft supports in 2012.
We are delighted to announce the opening of EnginSoft
Turin, the launch of EnginSoft’s Project Management Office,
our membership with AMAFOND and our cooperation with
Flow Design Bureau (FDB) in Norway. The Japan Association
for Nonlinear CAE informs us about a new framework for CAE
researchers and engineers.
To hear more and to discuss with the experts of CAE and
Simulation, please join us on 22nd and 23rd October – We
look forward to welcoming you to Lazise!
Stefano Odorizzi
Editor in chief
Ing. Stefano Odorizzi
EnginSoft CEO and President
Flash
4 - Newsletter EnginSoft Year 9 n°2
Sommario - Contents
CASE STUDIES
6
10
17
22
26
32
35
38
Trapped-vortex approach for syngas combustion in gas turbines
Extreme Ensemble under Uncertainty
High Pressure Die Casting: Contradictions and Challenges
Reducing Fuel Consumption, Noxious Emissions and Radiated Noise by Selection of the Optimal Control Strategy of a Diesel Engine
Fluid Refrigerant Leak in a Cabin Compartment: Risk Assessment by CFD Approach
Numerical Optimization of the Exhaust Flow of a High-Performance Engine
Multi-Objective optimization of steel case hardening
Research of the Maximum Energy Efficiency for a Three Phase Induction Motor by means of Slots Geometrical Optimization
CAE WORLD
42
Applicazioni strutturali CAE nel settore Elettrodomestici
INTERVIEW
44
EnginSoft interviews Massimo Nascimbeni from Sidel
HARDWARE UPDATE
46
Productivity Benefits of HP Workstations with NVIDIA® Maximus™ Technology
SOFTWARE UPDATE
48
Uno sguardo alle principali novità riguardanti l’integrazione dei prodotti ANSOFT in bassa frequenza in piattaforma ANSYS
Workbench 14
RESEARCH & TECHNOLOGY TRANSFER
52
54
56
Updates on the EASIT2 project: educational base and competence framework for the analysis and simulation industry now online
EnginSoft in £3m EU Partnership to Optimise Remote Laser Welding
DIRECTION: Demonstration of Very Low Energy New Buildings
ENGINSOFT NETWORK
58
59
60
EnginSoft and Flow Design Bureau (FDB) launch collaboration.
Dinamica esplicita: nuovo competence center EnginSoft a Torino
Avvio del Project Management Office di EnginSoft
The EnginSoft Newsletter editions contain references to the following products which
are trademarks or registered trademarks of their respective owners:
Forge and Coldform are trademarks of Transvalor S.A.
(www.transvalor.com)
ANSYS, ANSYS Workbench, AUTODYN, CFX, FLUENT and any and all ANSYS, Inc. brand,
product, service and feature names, logos and slogans are registered trademarks or trademarks of ANSYS, Inc. or its subsidiaries in the United States or other countries. [ICEM CFD
is a trademark used by ANSYS, Inc. under license]. (www.ansys.com)
LS-DYNA is a trademark of Livermore Software Technology Corporation.
(www.lstc.com)
modeFRONTIER is a trademark of ESTECO srl (www.esteco.com)
Grapheur is a product of Reactive Search SrL, a partner of EnginSoft
(www.grapheur.com)
Flowmaster is a registered trademark of Menthor Graphics in the USA.
SCULPTOR is a trademark of Optimal Solutions Software, LLC
(www.optimalsolutions.us)
(www.flowmaster.com)
MAGMASOFT is a trademark of MAGMA GmbH. (www.magmasoft.de)
Contents
For more information, please contact the Editorial Team
Newsletter EnginSoft Year 9 n°2 -
JAPAN COLUMN
61
The Japan Association for Nonlinear CAE: a New Framework for CAE
Researchers and Engineers
EVENTS
64
EnginSoft diventa socio di AMAFOND: Associazione nazionale fornitori
macchine e materiali per fonderie
64
METEF Foundeq 2012: EnginSoft soddisfatta della partecipazione
all’esposizione
66
66
66
67
Constructive Approximation and Applications
Graz Symposium Virtual Vehicle
International modeFRONTIER Users’ Meeting 2012
Event Calendar
5
Newsletter EnginSoft
Year 9 n°2 - Summer 2012
To receive a free copy of the next EnginSoft
Newsletters, please contact our Marketing office at:
[email protected]
All pictures are protected by copyright. Any reproduction
of these pictures in any media and by any means is
forbidden unless written authorization by EnginSoft has
been obtained beforehand.
©Copyright EnginSoft Newsletter.
Advertisement
For advertising opportunities, please contact our
Marketing office at: [email protected]
EnginSoft S.p.A.
24126 BERGAMO c/o Parco Scientifico Tecnologico
Kilometro Rosso - Edificio A1, Via Stezzano 87
Tel. +39 035 368711 • Fax +39 0461 979215
50127 FIRENZE Via Panciatichi, 40
Tel. +39 055 4376113 • Fax +39 0461 979216
35129 PADOVA Via Giambellino, 7
Tel. +39 049 7705311 • Fax +39 0461 979217
72023 MESAGNE (BRINDISI) Via A. Murri, 2 - Z.I.
Tel. +39 0831 730194 • Fax +39 0461 979224
38123 TRENTO fraz. Mattarello - Via della Stazione, 27
Tel. +39 0461 915391 • Fax +39 0461 979201
10133 TORINO Corso Moncalieri, 223
Tel. +39 011 3473987 • Fax +39 011 3473987
www.enginsoft.it - www.enginsoft.com
e-mail: [email protected]
COMPANY INTERESTS
PAGE 26 FLUID REFRIGERANT LEAK IN A
CABIN COMPARTMENT: RISK ASSESSMENT
BY CFD APPROACH
EnginSoft GmbH - Germany
EnginSoft UK - United Kingdom
EnginSoft France - France
EnginSoft Nordic - Sweden
Aperio Tecnologia en Ingenieria - Spain
www.enginsoft.com
Cascade Technologies
www.cascadetechnologies.com
Reactive Search
www.reactive-search.com
SimNumerica
www.simnumerica.it
M3E Mathematical Methods and Models
for Engineering
www.m3eweb.it
ASSOCIATION INTERESTS
NAFEMS International
www.nafems.it • www.nafems.org
TechNet Alliance
www.technet-alliance.com
RESPONSIBLE DIRECTOR
Stefano Odorizzi - [email protected]
PAGE 42 CAE STRUCTURAL APPLICATIONS
IN THE INDUSTRIAL APPLIANCES SECTOR
PRINTING
Grafiche Dal Piaz - Trento
The EnginSoft NEWSLETTER is a quarterly
magazine published by EnginSoft SpA
Contents
Autorizzazione del Tribunale di Trento n° 1353 RS di data 2/4/2008
PAGE 10 EXTREME ENSEMBLE
UNDER UNCERTAINTY
ESTECO srl
www.esteco.com
CONSORZIO TCN
www.consorziotcn.it • www.improve.it
6 - Newsletter EnginSoft Year 9 n°2
Trapped-vortex approach for
syngas combustion in gas turbines
The so-called trapped vortex technology can potentially
offer several advantages for gas turbine burners. In the
systems experimented with so far, this technology is mainly
limited to the pilot part of the whole burner. The aim of the
work we present here, is to design a combustion chamber
completely based on the trapped vortex principle,
investigating the possibility to establish a diluted
combustion regime, in case of syngas as fuel. This article
presents some results obtained by a 3D CFD analysis, using
both RANS and LES approaches.
Introduction
A key issue in combustion research is the improvement of
combustion efficiency to reduce fossil fuel consumption and
carbon dioxide emission. Researchers are involved in the
development of a combustion technology able to
accomplish energy savings with low pollutant emissions.
The differences between syngas and natural gas combustion
are mainly two. For the same power, fuel mass flow should
be 4-8 times higher than natural gas, due to the lower
calorific value. Premixed combustion of natural gas and air
is one of the most commonly used methods for reducing NOx
emissions, by maintaining a sufficiently low flame
temperature. This technique poses some problems with
syngas because of a significant presence of hydrogen and
the consequent danger of flashback in the fuel injection
systems. In the case of a non-premixed diffusion flame,
diluents such as nitrogen, carbon dioxide and water, or
other techniques, such as MILD combustion, can be
employed to lower flame temperatures and hence NOx.
The systems developed so far use combustion in cavities as
pilot flames for premixed high speed flows. The goal is to
design a device operating entirely on the principle of
trapped vortices, which is able, based on its intrinsic nature
of improving mixing of hot combustion gases and fresh
Case History
mixture, that represents a prerequisite for a diluted
combustion and at the most a MILD combustion regime. The
trapped vortex technology offers several advantages for a
gas turbine burner:
1. It is possible to burn a variety of fuels with medium and
low calorific values.
2. NOx emissions reach extremely low levels without dilution
or post-combustion treatments.
3. It provides extended flammability limits and improves
flame stability.
MILD combustion is one of the promising techniques
proposed to control pollutant emissions from combustion
plants. It is characterized by high preheating of combustion
air and massive recycle of burned gases before reaction.
These factors lead to high combustion efficiency and good
control of thermal peaks and hot spots, lowering NOx
thermal emissions. Two important aspects are crucial for
MILD combustion. First of all, the reactants have to be
preheated above the self-ignition temperature. Secondly,
the reaction region has to be entrained by a sufficient
amount of combustion products. The first requirement
ensures high thermal efficiency, whereas the latter allows
flame dilution, reducing the final temperature well below
the adiabatic flame temperature. In this way, reactions take
place in a larger portion of the domain in absence of
ignition and extinction phenomena, due to the small
temperature difference between burnt and unburnt gases
and as a result a flame front is no longer identifiable; this
is why MILD combustion is often denoted as flameless
combustion.
Another advantage is represented by the fact that the
temperature homogeneity reduces materials deterioration.
Because of the limited temperature, NOx emissions are
greatly reduced and soot formation is also suppressed,
thanks to the lean conditions in the combustion chamber,
Newsletter EnginSoft Year 9 n°2 -
due to the large dilution level and the large CO2
concentration.
The introduction of MILD technology in gas turbines is of
great interest because it is potentially able to answer two
main requirements:
1. A very low level of emissions.
2. An intrinsic thermo acoustic stability (humming).
7
CFD models
The simulations, performed with the ANSYS-FLUENT code,
have been carried out according to steady RANS and LES
approaches. The models adopted for chemical reactions and
radiation are the EDC, in conjunction with a reduced
CO/H2/O2 mechanism made up of 32 reactions and 9
species, and the P1, respectively. NOx have been calculated
Air tangential
mass flow
[kg/s]
Air tangential
flow velocity
[m/s]
Air vertical mass
flow
[kg/s]
Air vertical flow
velocity
[m/s]
Fuel mass flow
Air/Fuel global
Air/Fuel
primary
[kg/s]
Fuel mass flow
velocity
[m/s]
0.01238
75
0.00592
62
0.00462
61
3.96
1.28
Table 1 - Boundary conditions for the reference case.
Prototype description
The TVC project concerns gas turbines which use annular
combustion chambers. The prototype to be realized, for
simplicity of design and measurement, consists in a
linearized sector of the annular chamber having a square
section of 190x190mm (fig. 1).
The power density is about 15 MW/m3 bar. The most obvious
technique to create a vortex in a combustion chamber
volume is to set one or more tangential flows. In this case
two flows promote the formation of the vortex, while other
streams of air and fuel, distributed among the tangential
ones, feed the "vortex heart". The air flows placed in the
middle provide primary oxidant to combustion reaction,
while the tangential ones provide the air excess, cool the
walls and the combustion products, in analogy to what
happens in the traditional combustion chambers, in which
this process occurs in the axial direction. The primary and
the global equivalence ratio (Fuel/Air/(Fuel/Air)stoic) were
Fig. 1 - Burner geometry.
equal to 1.2 and 0.4, respectively. Given the characteristics
of the available test rig, the prototype will be tested under
atmospheric pressure conditions, but the combustion air
will be at a temperature of 700 K, corresponding to a
compression ratio of about 20 bar, in order to better
simulate the real operating conditions. The syngas will have
the following composition: 19% H2 - 31% CO - 50% N2 LHV
6 MJ/kg. The reference case boundary conditions are
reported in table 1.
in post-processing. In order to save computational
resources, the simulations have been conducted only on one
sector (1/3) of the whole prototype reported in the figure
1, imposing a periodicity condition on side walls. A
structured hexahedral grid, with a total number of about 2
million cells has been generated.
Results
A big effort has been made to properly modulate flow rates,
velocities, momentum and minimum size of the combustion
chamber. It’s clear that inlets placement plays a key role in
the formation of a energetic vortex which can be able to
properly dilute the reactants and create sufficient residence
time that favorites complete combustion. The resulting
configuration establishes a perfect balance between the
action of the tangential flows, which tend to generate the
vortex and the action of the vertical flows that tend to
destroy it (fig. 2).
In this sense, it is worth
pointing out that it is
especially the tangential flow
further from the outlet which
is the most effective. The
negative effects on vortex
location and size resulting
from a reduction of its
strength, compared with the
other inlets, have been
evident. Even the distance
between the two vertical
hole rows has been properly
adjusted. In fact, the upper
tangential stream flows between the two vertical rows on
the opposed side and, if the available space is insufficient,
it doesn’t remain adherent to the wall and the vortex is
destroyed. In principle, a significant presence a very
reactive hydrogen, can produce elevated temperatures and
fast reaction, especially near fuel inlets. For this reason,
inlet velocities are sufficiently high to generate a fast
rotating vortex and then a rapid mixing. Further increase in
fuel injection velocity has a negative influence on vortex
Case History
8 - Newsletter EnginSoft Year 9 n°2
Fig. 2 - Temperature (K) field on central section plane. (left) RANS (right) instantaneous LES.
gas recirculation, which associated
to a good degree of mixing with
reactants, represents a necessary
condition for MILD combustion. In
order to quantify the degree of
mixing, the following variable has
been mapped:
MIX=|H2-H2mean|+|H2OH2Omean|+|CO2-CO2mean|+|O2O2mean|+|CO-COmean|+|N2-N2mean|
Fig. 3 - OH mass fraction on different planes.
shape and position, without slowing reaction and reducing
temperature peaks.
The aim of the LES simulation has been to analyze the
unsteadiness of the system. The vortex appears very stable
in the cavity, i.e. trapped. No vortex shedding has been
noted.
Compared to axial combustors, the minimum space required
for the vortex results in an increase in volume and a
reduction in power density. On the other hand, carbon
monoxide content, with its slow chemical kinetic rate
compared to natural gas, requires longer residence time and
a bigger volume.
It can be observed that high temperature zones (fig.
2) are concentrated in the vortex heart. If the
hydrogen content is rapidly consumed, the carbon
monoxide lasts longer and accumulates given that
the amount of primary air is less than the
stoichiometric value. The mean LES and the steady
RANS fields have provided very similar results.
It is interesting to evaluate the residence time inside
the chamber. In the central part of the chamber, the
residence time ranges from 0.02 to 0.04 sec.
The establishment of a MILD combustion regime
depends especially on a sufficient internal exhaust
Case History
If all the species involved were
perfectly mixed, MIX should be zero everywhere. In practice,
the more MIX tends to zero, the more reactants and
products are well mixed. If the zones immediately
downstream the inlets are neglected, MIX assumes very low
values in the chamber. This supports the fact that the
vortex is able to produce the expected results.
The recirculation factor, i.e. the ratio between exhaust
recirculated and fresh mixture introduced, is about 0.87,
while exhaust composition is: 0.19% CO2, 0.05% H2O,
0.005% CO, 0.039% O2, 0.713% N2.
Fig. 4 - Maximum temperature vs compressor ratio.
Newsletter EnginSoft Year 9 n°2 -
9
recirculation ratio increases only to 0.92. A 30% decrement
of the equivalence ratio (leaner combustion) causes a
reduction in temperature and a subsequent increment in
unburnt species, especially CO. NOx emissions are in general
unimportant for all the cases mentioned.
Fig. 5 - EICO (CO emission index) vs compressor ratio.
Fig. 6 - EINOx (NOx emission index) vs compressor ratio.
Fig. 7 - EICO and EINOx vs equivalence ratio, operating pressure 20 bar.
In order to determine where reactions are concentrated, it
is useful to analyze radical species distribution, such as OH.
The fact that radicals are not concentrated in a thin flame
front, but well distributed in the volume (fig. 3), represents
an evidence of a volumetric reaction regime.
A sensitivity analysis has been conducted varying the
boundary conditions around the references previousy
reported in table 1. The quality of the different cases was
judged in terms of pollutants emission indices (g
pollutant/kg fuel), in particular for CO and NOx. CO is an
indicator for incomplete combustion. Its presence in the
exhaust is favored by low temperature and lack of oxygen.
On the contrary, NOx are favored by high temperature and
oxygen abundance. A 30% increment of the tangential flow
velocity causes a faster rotation of the vortex, but the
If the operating pressure is augmented from 1 bar to 10-20
bar at constant geometry and inlet velocities, the mass flow
rates and then the burner power increases by about 10-20
times, even if the specific power density (MW/m3 bar)
remains constant. The temperature and pollutants trends for
those cases are reported in figures 4-6. The maximum and
average temperatures in the chamber increase almost
linearly, while EICO increases and EINOx decreases as
pressure increases. For each operating pressure, it is then
possible to identify an optimal equivalence ratio condition,
at the intersection of the two curves in figure 7, where the
major pollutants are kept down at the same time.
Scaling
One of the aims of this work has been to verify if when
varying the prototype dimensions, their behavior would
remain unchanged or not, in terms of fluid dynamics and
chemistry. The scaling method based on constant velocities
has been chosen for this purpose, because all the physics of
the system are based on a fluid dynamic equilibrium, if one
refers to the above discussion.
If one imagines to halve all sizes, the volume will be
reduced by a factor of 0.5x0.5x0.5=0.125, while the areas
will be reduced by a factor of 0.5x0.5=0.25. Then mass flowrates will be scaled by a factor of 0.25. Therefore, the power
density (power/volume) will increase by a factor of
0.25/0.125=2. The simulations show that the behavior of
the burner remains unchanged, in terms of velocity,
temperature, species, etc. fields. It can be concluded that,
if the overall size of the burner is reduced, the power
density increases accordingly to Power0.5.
Conclusions
A gas turbine combustion chamber prototype has been
designed based on the trapped vortex technique.
A preliminary optimization of the geometry has led to a
prototype in which the actions of the different flows are in
perfect balance, and a vortex filling the entire volume has
been established. The large exhaust recirculation and the
good mixing determine a diluted combustion regime, which
helps to keep down flame temperature. A sensitivity
analysis has allowed to determine the optimal operating
conditions for which the contemporary lowering of the
major pollutants species was achieved. In addition, a
scaling study has been conducted which demonstrated how
the prototype keeps its characteristics unaltered if halved
or doubled in dimensions.
A. Di Nardo, G. Calchetti, C. Mongiello
ENEA-Italian National Agency for New Technologies
Energy and Economic Development
Case History
10 - Newsletter EnginSoft Year 9 n°2
Extreme Ensemble under Uncertainty
The development of robust design strategies coupled with
detailed simulation models requires the introduction of
advanced algorithms and computing resource management
tools. On the algorithmic side, we explore the use of
simplex-based stochastic collocation methods to
characterize uncertainties, and multi-objective genetic
algorithms to optimize a large-scale, three-dimensional
geometry using a very large number (a so-called “extreme
ensemble”) of CFD simulations on HPC clusters.
In this article, we concern ourselves with the optimization
under uncertainty of a Formula 1 tire brake intake to
maximize cooling eciency and to minimize aerodynamic
resistance. The uncertainties are introduced in terms of tire
deformation and free stream conditions.
The simulations environment Leland has been developed to
dynamically schedule, monitor and stir the calculation
ensemble and to extract runtime information as well as
simulation results and statistics. Leland is equipped with an
auto-tuning strategy for optimal load balancing and fault
tolerance checks to avoid failures in the ensemble.
Introduction
In the last few years, clusters with 10,000 CPUs have
become available, and it is now feasible to design and
optimize
complex
engineering
systems
using
computationally intensive simulations. This development
highlights the need to create resource managers that deliver
cost-effective utilization with fault tolerance.
The BlueGene/L cluster with 65,536 nodes was designed to
have less than one failure every ten days. In fact, this
cluster and others like it, experience an average of one
processor failure every hour. In light of this, it is necessary
to study, develop, and to continually improve strategies for
the ecient completion of large simulations. Theoretical work
has been published in the literature that suggests that
advanced algorithms might be available although they have
only been demonstrated using test functions on a small
number of compute nodes.
Case History
The design process involves running an extreme number of
large computations or “extreme ensemble” (in the range of
thousands), in order to create a robust solution that will
remain optimal under conditions that cannot be controlled
(“uncertainties”). We call this process optimization under
uncertainty. The ensemble is a list of runs generated by the
optimization and uncertainty analysis algorithms that is
dynamic in nature and not deterministic. This means that
the number of additional simulations is dependent on the
results of the prior converged simulations.
In this article, we explore the computational design of a
Formula 1 tire and brake assembly using large-scale, threedimensional Reynolds-Averaged Navier-Stokes simulations
on a high performance computing cluster. The purpose of
designing the brake duct is to increase the amount of air
captured by the duct while minimizing the total drag of the
tire. This multi-objective optimization problem is tacked
using a genetic algorithm which produces a Pareto front of
best solutions.
An uncertainty analysis of 4 specific points on the Pareto
front (minimum drag, maximum cooling, best operating
point or trade-off, and baseline F1 tire geometry) is
summarized in the results section of this paper. Some of our
future work will include a study on how uncertainties can be
invasively incorporated in the optimization procedure,
producing a probabilistic Pareto front rather than analyzing
the sensitivity of the deterministic Pareto due to
uncertainties. For such a study, approximately 400
simulations have to be performed per optimization cycle
(i.e. generation). When the results of these 400 simulations
are analyzed, an additional list of 400 simulations, each
with a unique range of input parameters, will be created for
the next generation of the optimization process. The values
of these input parameters are not known a priori.
The optimization procedure needs to account for
uncertainties arising from variables in ow conditions as well
Newsletter EnginSoft Year 9 n°2 -
11
Fig. 1 - Leland flowchart
as from variability in the flexible tire geometry. This
complex baseline geometry consists of 30 million mesh
cells. In order to generate an optimal design under
uncertainty, the mesh is deformed locally, creating 5000
unique simulations to perform. Each simulation (or
realization) will be run on our in-house cluster using 2400
cores; the full design process should take approximately 2
weeks to complete.
The second part of this article is dedicated to the
development of a software platform able to reduce the total
time needed to carry out an engineering design process
such as the one described above. Leland, the simulations
environment we have developed, allows us to schedule the
resources, to monitor the calculation ensemble and to
extract runtime information as well as simulation results
and statistics on the y. Leland is equipped with an autotuning strategy for selecting an optimal processor count.
Moreover, an implemented fault tolerance strategy ensures
that a simulation or a processor stall is detected and does
not impact the overall ensemble finish time. The results of
this study document the actual computational time savings
achieved through the efficient use of resources with Leland,
as opposed to submitting individual jobs on the cluster, one
at a time, using traditional queue managers (e.g. Torque,
SLURM, etc.).
Robust Design Algorithm
The impact of uncertainties in the robust design process is
characterized using the Simplex Stochastic Collocation
(SSC) algorithm, which combines the effectiveness of
random sampling in higher dimensions (multiple
uncertainties) with the accuracy of polynomial
interpolation. This approach is characterized by a superlinear convergence behavior which outperforms classical
Monte Carlo sampling while retaining its robustness. In the
SSC methods, a discretization of the space spanned by the
uncertain parameters is employed, and the simplex
elements obtained from a Delaunay triangulation of
sampling points are constructed.
The robustness of the approximation is guaranteed by using
a limiter approach for the local polynomial degree based on
the extension of the Local Extremum Diminishing (LED)
concept to probability space. The discretization is
adaptively refined by calculating a refinement measure
based on a local error estimate in each of the simplex
elements. A new sampling point is then added randomly in
the simplex with the highest measure and the Delaunay
triangulation is updated. The implementation of advanced
algorithms to improve the scalability of Delaunay
triangulation in higher dimensions, in order to circumvent
the curse of dimensionality, has not been fully investigated
as part of this study. There are proofs in literature that show
that Delaunay triangulation can achieve linear scaling with
higher dimensions.
In this work, we analyze a nontrivial multi-objective
problem within which it is not possible to find a unique
solution that simultaneously optimizes each objective:
when attempting to improve an objective further, other
objectives suffer as a result. A tentative solution is called
non-dominated, Pareto optimal, or Pareto ecient if an
improvement in one objective requires a degradation of
another. We use the NSGA-II algorithm to obtain the nondominated solutions, hence we analyze the more interesting
solutions on the deterministic Pareto set in presence of
uncertainty. Our goal here is to prove the importance of
considering the variability of several input conditions in the
design process.
For all these solutions, the SSC is used to obtain a
reconstruction of the objective function statistical
moments, refining the simplexes until an accuracy threshold
is reached.
Dynamic Resource Manager - Leland
The structure of Leland is based on a workflow through I/O
sub-systems that represent the software applications (i.e.
Case History
12 - Newsletter EnginSoft Year 9 n°2
Sculptor, Fluent, Tecplot, Matlab etc.) involved in the
process. This environment is designed to run natively on
any high-performance computing (HPC) system, by
integrating with the job-submission/queuing system (e.g.
Torque). Moreover, it does not require continuous
management: once the analysis is initiated, multiple
simulations are submitted and monitored automatically. In
Leland, a job is an instance of the entire multi-physics
simulations, which might include grid generation, mesh
morphing, flow solution and post-processing. The main
objective of Leland is to set-up a candidate design as a job,
to manage it until it is completed and to gather relevant
results that are used to inform the optimization under the
uncertainty process. ROpt (robust optimum), shown in
Figure 1a, is the engine behind this design environment.
(1a)
(1b)
(2)
simulation using 1 processor divided by the total time
required to finish the simulation using p processors (see
Equation 2).
The speed-up curve in Figure 2 was generated by artificially
replicating an HPC simulation. The time required to
complete an HPC simulation is primarily a function of three
factors: i) portion of the code that is not parallelizable
(tserial in Equation 1), ii) portion of the code that is
parallelizable (t parallel in Equation 1) and iii) the
communication time between CPUs (tcomm in Equation 1).
The serial portion of the code in the example illustrated in
Figure 2 is constant (5000 seconds) and not a function of
the number of processors allocated to the job. The length of
time required to complete the parallel portion of the code
in the example shown in the same figure, is 5 million
seconds divided by the number of processors used. Finally,
there will always be some latency between CPUs and this is
characterized by the communication time between nodes.
The linear penalization we used in this example is 40
seconds per processor, but the latency slowdown could also
be a more complex function related to the specific
application.
(3)
Linear speed-up, also referred to as ideal speed-up, is
shown as the green dotted line in the middle plot of Figure
2. An algorithm has linear speed-up if the time required to
finish the simulation halves when the number of processors
is doubled. It is common for fluid dynamic simulations to
experience speed-down; this occurs when the total time
required to finish the simulation actually rises with an
increasing number of processors. Leland has the ability to
recognize the point at which speed-down occurs (at about
400 processors in Figure 2) and never uses more than this
number of processors. The rightmost plot in Figure 2 shows
the eciency (defined by Equation 3) curve for this artificial
HPC simulation. The eciency typically ranges between
values of 0 1 estimating how well utilized the processors are
compared to the effort wasted in synchronization and
communication. It is clear from this plot, that the highest
eciency occurs with the lowest number of processors.
Given the design and/or uncertain input variables, the ROpt
continuously generates new design proposals (samples)
based on the evolutionary strategy and/or analysis of the
uncertainty space, until a convergence criterion is met.
The Job Liaison shown in Figure 1b, defines the
characteristics of each single job and continuously monitors
the progress of the simulations until completion, in order to
communicate the objective evaluations back to the ROpt. It
is the job of this module, to continuously monitor for faults,
stalls, or errors, to ensure that the total runtime is not
detrimentally affected by processor/memory failure.
The Job Submission engine, shown in Figure 1c, ensures
that the correct number of jobs is always running on the
cluster. The variables (number of cores, number of jobs,
etc.) from the input file that are used to initialize the runs
are dynamic, which means that they can be edited on the y
and the system will respond accordingly.
Leland has the ability to dynamically
select the optimal number of processors
to run per realization. This is achieved
by auto-tuning. The user selects an
optimal window of cores to use per
realization prior to launching the full
ensemble. The auto-tuning algorithm
then samples the space by using a
unique number of cores per realization
in the ensemble. Once two or more
realizations are complete, the auto- (a) Total time required to complete simulation
tuning algorithm can start to construct as a function of the number of processors (left),
an application-specific speed-up curve Speed-up curve (middle), and Eciency curve
(Figure 2). Speed-up is defined as the (right)
total time required to finish the Fig. 2 - Sample HPC simulation diagnostics
Case History
(b) Number of simulations that would be
completed in a 24 hour window with 1000
available processors using exactly p processors for
each simulation
Newsletter EnginSoft Year 9 n°2 -
13
elements is considered for a fully detailed 3D
wheel model (Figure 4). The simulations that
require
geometrical
modification
(for
optimization for uncertainty) are created using
Sculptor, a commercial mesh deforming software
from Optimal Solutions.
The software is used to generate multiple CFD
mesh model variants, while keeping CAD and grid
generators out of the design process loop, thus
saving design time and costs substantially. The
(a) Outer view of tire
generated models are then used to compute the
(b) Inner view of tire
Fig. 3 - Front right tire of the Formula 1 race car used in this study showing green airfoil
air flow around the tire geometry by a parallel
strut used to secure tire to the experimental wind tunnel facility and the outer brake duct
CFD solver (Fluent v12.1.4). It is important to
(magenta) used to cool the brake assembly
closely monitor the skewness of elements in
Sculptor to ensure grid quality. If the deformation in
This speed-up curve will guide Leland's auto-tuning
Sculptor is too large, the CFD solver will diverge. The
algorithm in assigning the optimal number of cores per
boundary conditions, computational setup, and
realization (which may not be in the user’s original
experimental comparison for this case are outlined in
window). Since an ensemble of this size takes more than a
separate studies.
few weeks on a large cluster, multiple job submissions need
to be submitted to the local queuing system.
These jobs are typically limited to 24 hour runtimes (or a
Optimization Variables
A local mesh morphing software, Sculptor (v2.3.2), was
wall clock time of 24 hours). Thus, it is essential that the
used to deform the baseline Formula 1 brake duct (Figure
3). Specific control volumes were used to deform the brake
duct in three dimensions, namely i) width of opening
(Figure 5(a)), ii) height of opening (Figure 5(b)) and iii)
protrusion length (Figure 5(c)). Each design variable was
allowed to change by 1cm as depicted in Figure 5.
(a) Isometric view of ground plane
showing contact patch
(b) Streamwise cut plane showing
mesh inside rotor passages
(c) Spanwise cut plane showing full
brake assembly
(d) Top view of plane cutting through
the center of the tire
Fig. 4 - Four different views showing the Formula 1 tire mesh
auto-tuning algorithm recognizes how many hours remain
prior to the job terminating due to the wall clock time, and
tries to increase the number of cores to finish as many
realizations as possible within a specific time frame.
Application Description
Leland is used here to optimize the shape of a F1
tire brake duct (magenta color in Figure 3(b)),
taking into account the geometrical uncertainties
associated with the rotating rubber tire and
uncertain inflow conditions.
The objectives are to minimize the tire drag [N]
while maximizing the captured mass flow (kg/s)
needed to cool the brake assembly. A
computational mesh consisting of 30 million
Uncertain Variables
Multiple uncertain variables were tested to determine their
sensitivity to output quantities of interest using a DOE
(design of experiments) approach. Some of the uncertain
variables were based on the in flow conditions (i.e. yaw
angle, turbulent intensity, turbulent length scale) while
others were based on geometric characteristics of the tire
(i.e. contact patch details, tire bulge radius, camber angle).
Figure 6 shows 9 geometric modifications which were
performed. Each subfigure shows the minimum, baseline F1
tire geometry, and maximum deformation for each uncertain
variable.
From the results of a purely one-dimensional perturbation
analysis, the turbulence length scale (on the order of 0m
2m) results in less than a 0.1% difference in both, the mass
flow rate through the brake duct and the overall drag on the
(a) Brake duct width
(b) Brake duct height
(c) Brake duct length
Fig. 5 - Brake duct optimization variables
Case History
14 - Newsletter EnginSoft Year 9 n°2
tire. Conversely, both the mass flow rate and tire drag are
very sensitive to the turbulence intensity.
The mass flow rate decreased by 7.8% compared to the
baseline (less cooling) with 40% turbulence intensity, and
the tire drag increased by 7.2% with 40% turbulence
intensity. This analysis confirms that the car performance
decreases with “dirty” air compared to “clean” air. The
sensitivity of the output quantities of interest caused by
the tire yaw angle is reflected in the first row of Table 1.
The remaining rows in Table 1, show the sensitivity of mass
flow rate and drag force to geometric characteristics,
specifically contact patch, tire bulge radius, tire
compression, and brake duct dimensions.
In the end, the three most sensitive uncertain variables,
namely tire contact patch width, tire yaw angle, and
turbulence intensity, were selected for the optimization
under uncertainty study. The tire contact patch width was
able to expand and contract up to 1cm, the tire yaw angle
varied between 3, and the turbulence intensity varied
between 0% and 5%.
Results
Formula 1 engineers are interested in primarily three factors
related to tire aerodynamics: i) overall tire lift and drag ii),
Fig. 6 - Subset of uncertain variables tested for sensitivity in output
quantities of interest
cooling performance of the brakes, and iii) how the tire air
flow
affects
downstream
components
(wake
characteristics). All three factors are tightly coupled which
makes the design quite complicated, especially when
uncertainty in the flexible tire walls and upstream
conditions can negatively effect the car performance.
Figure 7 illustrates the wake sensitivity caused by flow
traveling through the tire hub and exiting from the
outboard side of the tire. If the flow of air is not allowed to
pass through the tire hub (see top left and bottom left
images in Figure 7), there is no mass eux from the outboard
side of the tire and the wake is quite symmetric about the
wheel centerline. The wake is dominated by a counterrotating vortex pair and both the inboard (left) and
outboard (right) vortex are of similar size. Alternatively, if
the flow of air is allowed to pass through the tire hub, the
inboard (left) vortex becomes larger than the outboard
(right) vortex causing wake asymmetry (see top right and
bottom right images in Figure 7).
Table 1 - Mass flow rate into the brake duct and drag force on the tire
sensitivity for 9 uncertain variables and 3 design variables
Case History
The results of the single parameter perturbations indicated
previously show that the mass flow rate through the brake
duct and tire drag force are more sensitive to the brake duct
width than the brake duct height or length (in the range of
deformation between 1cm). The physical explanation of this
result becomes evident when visualizing iso-contours of
turbulent kinetic energy around the tire. Figure 7 presents
the difference between a low width configuration (top) and
high width configuration (bottom). The larger width of the
brake duct causes a larger separation region immediately
behind the brake duct in addition to higher turbulence
levels in the shear layer immediately behind the inboard
back edge of the tire.
Newsletter EnginSoft Year 9 n°2 -
15
selection 6 (e.g. mating pool, parent
sorting) and reproduction (e.g.
crossover and mutation). Leland was
used to handle the job scheduling and
management and as a result, the time
required to complete the 450
simulations was 2 days compared to
about 4 days without using Leland,
which requires submitting jobs manually
to the job queuing system using a
constant number of processors.
Among the Pareto set (see Figure 9), the
design that achieves the highest mass
flow rate is shown in blue and the
design that achieves the lowest overall
drag on the tire is shown in magenta.
The green design is labeled as the tradeFig. 7 - Wake sensitivity (shown by streamwise x-velocity contours for a plane located 1.12 wheel
off
design, since this design tries to
diameters downstream from the center of the tire) for a simplified tire with wheel fairings (top
left), baseline F1 tire (top right), baseline F1 tire with blocked hub passages (bottom left), and
achieve the highest mass ow through
simplified tire with artificial mass eux from blue segment (bottom right)
the inlet of the brake duct while
minimizing the total drag on the tire.
The baseline geometry, reported in red, was shown not to
be on the Pareto front in the deterministic setting.
In the previous results, once the tire configuration and
other input conditions are specified, the solution is
uniquely determined without vagueness. On the other hand,
when uncertainties are present, the results have to be
expressed in a non-deterministic fashion either
probabilistically or as ranges of possible outcomes. The
approach we followed here using the SSC is strictly nonTable 2 - Multi-objective optimization strategy
intrusive, in the sense that the existing tools are used
The Pareto frontier showing the optimal brake duct designs
without modifications, but the solution - or more precisely,
under no uncertainty are illustrated in Figure 9. Ten
their probability distributions - are constructed performing
generations, which equates to 450 simulations, were needed
an ensemble of deterministic analyses. Further details about
to eventually construct the Pareto frontier. Further details
the uncertainty quantification strategy can be found in
about the optimization strategy can be found in Table 2.
Table 3.
This table reports the settings of the NSGA-II algorithm
The variability of the four geometries described above
adopted to drive the main phases of the genetic algorithm:
(namely trade-off, highest mass flow, lowest drag, and
baseline) as a result of the uncertainties
in the tire yaw angle, turbulence
intensity, and contact patch width, are
presented in Figure 10. The variability
of the minimum drag design is the
highest, as illustrated by the spread of
magenta dots, followed by the maximum
mass flow design shown by blue dots,
trade-off design presented by green
dots and baseline design reflected by
red dots. The colored dots in this figure
represent the mean probabilistic values
and the black lines represent 1 standard
deviation
of
the
probabilistic
distribution. It is evident in this figure
that the optimal designs, on average,
move away from the Pareto frontier,
decreasing the overall performance of
Fig. 8 - Turbulent kinetic energy contours for the minimum drag configuration (top) and maxithe
racing car.
mum cooling configuration (bottom)
Case History
16 - Newsletter EnginSoft Year 9 n°2
Fig. 9 - Deterministic Pareto front (left); the green, blue, magenta, and gray brake ducts in
the subfjgure on the right correspond to the trade-off, max cooling, minimum drag, and baseline
configurations respectively
Table 3 - Uncertainty quantification strategy
A similar conclusion can be drawn when we look at the
probability density of the drag force and the brake mass
flow (Figure 11). The former shows a large excursion of
both, the position of the peak and the support, while the
Conclusions
In this work, we introduced an ecient method to perform
massive ensemble calculations with application to a
complex Formula 1 tire assembly optimization case.
Special attention has been placed on the creation of an
effective resource manager to handle the large number of
computations that are required. Since the geometrical
uncertainties associated with rubber tires and inflow
uncertainties associated with upstream “dirty” air, have an
impact on the dominating solutions, their presence has to
be taken into account in the design process. The next step
of this study is to consider the presence of uncertainties
invasively
in
the
optimization
procedure, generating a probabilistic
Pareto front rather than analyzing the
sensitivity of the deterministic Pareto
due to uncertainties.
Fig. 10 - Pareto frontier for F1 wheel assembly showing the variability of the minimum drag
(magenta), baseline (red), trade-off (green), and maximum cooling (blue) designs to uncertainty
in the in flow conditions and exible tire geometry.
(a) Drag force on tire
(b) Mass flow through tire inlet brake duct
Fig. 11 - PDF's of the output quantities of interest used for this study
Case History
latter is only marginally affected. This
directional sensitivity under uncertainty
with respect to drag force might suggest
that only the drag minimization could
be treated as a probabilistic objective,
while the brake mass flow optimization
can be handled using conventional
(deterministic) optimization. Since the
solutions identified above move away
from the deterministic Pareto, the
optimization process cannot be
decoupled from the uncertainty
quantification process. We plan to
tackle the joint problem in a future
study.
Acknowledgments
The authors would like to acknowledge
first and foremost Sculptor Optimal
Solutions, specifically Taylor Newill and
John Jenkins, for their generous
support, training, and licenses. The
authors wish to thank Dr. J. Witteveen
for providing the initial version of the
Simplex
Stochastic
Collocation
algorithm and Steve Jones and Michael
Emory for helping with the resource
allocation manager. The authors also
thank Toyota Motor Corporation - F1
Motorsports Division for providing the
original geometry used in this study.
John Axerio-Cilies, Gianluca Iaccarino
Stanford University
Giovanni Petrone
University of Naples “Federico II”
Vijay Sellappan
EnginSoft Americas
Newsletter EnginSoft Year 9 n°2 -
17
High Pressure Die Casting:
Contradictions and Challenges
This article is aimed at offering an overview of the actual
status of the High Pressure Die Casting (HPDC) technology,
putting into evidence both the critical aspects and the
potential advantages. Specific attention is paid to quality
requirements from the end-users, production rate achievable,
process monitoring and control, European and worldwide
scenarios.
This overview leads to the individuation of the 6 most
relevant challenges for the HPDC industry: “zero-defect”
production, real time process control, understanding the role
of process variables, process optimisation, introduction of
R&D activities and disseminating the knowledge about the
HPDC technology.
Introduction
The High Pressure Die Casting (HPDC) process is particularly
suitable for high production rates and is applied in several
industrial fields; actually about half of the world production
of light metals castings is obtained by diecasting. In HPDC
of Aluminium alloys, “cold chamber machines” (Fig. 1) are
used: the alloy is molten and kept in a crucible, from which
a dosing system loads the injection chamber. The main steps
of the process are the filling of the cold chamber, the
injection into the die, and the extraction. HPDC is a complex
process, not only due to the phase transformation the metal
undergoes when solidifying in the die. In fact:
• high pressure die castings are produced by pouring liquid
metal into a metallic shot sleeve;
• a steel piston accelerates quickly and transports the
metal into a steel die, resulting in metal velocities
between 100 and 200 kilometres per hour;
• the subsequent extremely short fill time of 50 to 100
milliseconds guarantees perfect fill of small sections in
the die such as ribs before metal solidification;
• when the metal solidifies, the volume diminishes leaving
shrinkage in the casting;
• the process tries to overcome these physical phenomena
by pressing liquid metal into the die using a couple of
hundred atmospheres.
While scrap rates in other production lines, such as
machining, are measured in ppm (parts per million), in HPDC,
due to the complexity of the process, casting scrap rates lie
in the range of % (parts per hundred).
Fig. 1 - Schematics of HPDC process
HPDC is widely employed, for instance, in automotive
components (about 60% of light alloy castings in this field
are made by HPDC – a rough estimation could be 80-100 kg
of HPDC components in the “average EU car”) but the amount
of scraps is sometimes “embarrassing” (it is not so rare to
have 5 to 10% of scraps, due to different kinds of defects, in
almost all cases detected during or after the final operations
(e.g. machining or painting).
Case History
18 - Newsletter EnginSoft Year 9 n°2
The Contradictions of HPDC Process and the Related
Challenges
When the HPDC process is approached, at least six
contradictory features can be immediately individuated:
1. The final application fields of HPDC products are
increasingly requiring enhanced quality, reliability and
safety BUT Among the foundry & forming technologies,
HPDC is certainly the most critical one in terms of defect
generation and poor reliability of products
2. HPDC is a manufacturing route which becomes highly
advantageous when elevated production rates are
required BUT Actually, the quality control is performed at
the end of the production stage, with no real-time
correction of critical parameters.
3. A variety of HPDC process parameters is measured today
and used for defect detection BUT The quality relevant
process parameters are monitored by individual control
systems of equipment and no interfacing to the resulting
part quality takes place.
4. The HPDC process is highly automated and extended use
of process simulation techniques is done by the
companies BUT The in-field HPDC process setup and
optimisation is still based on specific experience and
skills of few people.
5. Specific studies have demonstrated that HPDC companies are
innovation-sensitive, with investments mainly directed at
increasing production capacity (27%), efficiency (22%),
product quality (13%) BUT Within SMEs, the internal R&D
potential of HPDC companies is relatively limited, and
innovation requires multi-disciplinary integration and
cooperation (which are quite unusual for SMEs)
6. HPDC products are typically addressed to Large Industries,
for assembling components, systems, machines to be
used in widely extended markets BUT 90% of EU light
alloys HPDC foundries are SMEs, with an average number
of employees going from 20 (IT, PT) to 40 (SE, FI, DK),
80 (DE, ES), up 140 (NO)
Fig. 2 - Defects classification and origin
It seems that each attractive issue of HPDC is
counterbalanced by technical or structural difficulties,
leading, in the opinion of the authors, to six inter-related
challenges which must be faced by HPDC-oriented
companies:
1. leading HPDC to “zero-defect environment”;
2. by introducing real-time tools for process control;
3. by monitoring and correlating all the main process
variables;
4. by making the process set up and optimisation a
knowledge-based issue;
5. thanks to multi-disciplinary R&D activities;
6. impacting on HPDC foundries scenario.
In the next paragraphs, each of the six challenges mentioned
above will be described, both in terms of actual technologies
and approaches and of future and innovative solutions.
Leading HPDC to “Zero-Defect Environment”
Due to the extreme conditions to which the molten alloy
(injection speed up to 200 km/h, solidification in few
seconds) and the die (contact with a molten alloy at more
than 700°C and, after 30-40
Defect
% of
Predictable
Experimental
It can be monitored by
seconds, with a sprayed lubricant at
sub-class
occurrence by simulation? validation by
room temperature) are subjected,
X-Rays, light
temperature, pressure,
Shrinkage defects
20
only partially
microscopy
metal front sensors
the
difficult-to-keep-constant
process parameters and the lacking
X-Rays, light
Gas-related defects
15
no
air pressure, humidity
microscopy, blister test
interactions among process control
Optical inspection,
air pressure, metal front units, HPDC can be considered as
Filling related defects
35
yes
pressure tests
sensors, temperature
“defect-generating process”. Not
only an average of 5-10% scrap is
Undesired phases
5
no
light microscopy, SEM
shot chamber sensoring
consequently produced, also kind,
Thermal contraction
Optical inspection,
size and criticism of defects are
5
yes
temperature
defects
light microscopy
various. According to a recently
Metal-die interaction
temperature,
proposed defects classification for
5
only partially
Stereo microscopy, SEM
defects
ejection force
HPDC components, 9 sub-classes of
by advanced
defects (leading to more than 30
Out of tolerance
5
Visual inspection
geometry measures
simulation
specific defect types) can be
Lack of material
5
yes
Visual inspection
geometry measures
identified, as summarised in Table 1.
They present also their estimated
by advanced
Excess of Material, flash
5
Visual inspection
geometry measures
percentage of occurrences. Each
simulation
stage of the conventional HPDC
Table 1 - HPDC defects classification and possibility of prediction, validation and monitoring
Case History
Newsletter EnginSoft Year 9 n°2 -
19
of the manufacturing cycle, often only by visual inspection or
with a delay which still cannot not be still accepted, as it
strongly affects costs and delivery time.
Advanced sensors applied to HPDC process will allow first of
all the continuous control of the process itself, recording all
the variables evolution during each manufacturing cycle, in
order to individuate all the deviations from the optimal set
up. Obviously, completely new hardware & software solutions
are needed, to be incorporated into the die to achieve a
“reactive control” of the quality.
process can generate these defects, which at first-level
classification can fall into three categories (surface defects,
internal & surface defects, geometrical defects) as shown in
Fig. 2. All these defects are classified in further detail, their
morphology and origin are well-known and, in some cases, an
advanced engineering tool such as process simulation can be
used for their prediction (see Table 1). Such tools have been
experimentally validated, with reference to specific subclasses of defects. Some of the process parameters and the
experimental variables which possibly generate defects can
be monitored by dedicated sensors and devices. One of the
shortcomings of the HPDC process is that the overall
complexity of the process is not handled in one system:
machine controls measured machine parameters, furnace
temperatures are controlled separately, lubrication systems
only control lubricant pressures and application times and so
on. Within the die, where the solidification of the metal
takes place and the final part quality is determined nearly no
process parameter is measured, neither controlled nor related
to final part quality.
The challenge of leading HPDC to a “zero-defect
environment” requires advanced engineering tools capable of
managing the complexity of the process. Key-variables
identification, knowledge of variables-defects relationship,
implementation of real-time sensor device to monitor these
variables must be managed by these tools. They must also
have the ability to integrate all these information and to
carry out re-active strategies to instantaneously “balance”
the process in view of zero defect production.
Monitoring and Correlating all the main
Process Variables
Today’s state-of-the-art is that HPDC machines are equipped
with sensor systems which allow to measure basic process
data such as hydraulic pressures or piston velocities.
If the fill time varies only in the range of milliseconds, there
will be an impact on defect location. If however, the fill time
lies in the range of milliseconds, the piston speed should be
controlled one order of magnitude faster, which is not the
case in the state-of-the-art equipment. Until today, other
relevant issues are not taken into consideration at all,
although each impacts casting quality: thermal loads change
the geometry of the shot sleeve and lead to delayed fill and
liquid metal cools in the shot sleeve leading to presolidification. Taking into account the actual HPDC process
architecture (Fig.4),
• in Product and Process design, only mechanical
performances of the component are considered,
neglecting the real behaviour of the diecast alloy, which
results also from the presence and characteristics of
defects and imperfections;
• in the Overall Casting process, there is a very partial
control of the pre-casting stages (melting, degassing,
pouring, etc), focussed on the setup of the alloy
degassing, furnace level, cup movement to pour the shot
chamber. On the other hand, the HPDC machine setup is
actually the “best” controlled system, but it simply
“tries” to replicate the input setup, with the real shot
profile visualised after casting, allowing only “postmortem” evaluations;
• in the Die Management, the temperature and all the other
parameters of conditioning media (water, oil) are set up
and controlled, but correlations with the final quality of
castings are performed only when defect causes have to
Introducing Real-Time tools for Process Control
It is well-known that HPDC, whose typical cycle time ranges
from 40 to 60 seconds, is particularly suitable when high
production rates are required. The high costs associated to
HPDC tools/dies are recovered when at least 5.000-10.000
casting/per year are produced (Fig. 3). If this manufacturing
context is considered, it is clear that quality investigations
(carried out on a statistical basis or on the whole production)
are a critical issue, which must be as close as possible to the
moment at which defects are generated. In the actual HPDC
production, defect detection is usually carried out at the end
Fig. 4 - Actual HPDC process architecture
Fig. 3 - Competitiveness area for different casting processes of Al alloys
Case History
20 - Newsletter EnginSoft Year 9 n°2
be individuated, while no action to prevent defect
formation is carried out;
• in Post-Casting Operations (including machining, heat
and surface treatments), the main variables are certainly
recorded (e.g. in movement robot and cutting equipment;
the actual programming method allows to have positiontime control), but their setup is usually defined a-priori,
without considering the possible generation of defects.
From the described situation, the need for an effective
integration of the design and manufacturing chain with the
following HPDC components becomes evident:
• sophisticated and accurate results interpretation to
estimate the die and process output as well as the scale
and morphology of the defects;
• the use of advanced sensors to monitor the level of
hydrogen in the alloy, the temperature and volume in the
furnace and the temperature and volume in pouring cup;
• the use of real-time sensors to control position,
acceleration and velocities of the plunger and to correlate
them to product quality, on the basis of meta-models of
the machine behaviour, defining the robustness of the
process and die design and the range of quality
tolerances;
• specific sensors (temperature, pressure, humidity, airpressure) dedicated to real-time monitoring of the
thermal-mechanical behaviour of the die, including
special reaction devices to modify the process in view of
“zero defects” self-adaptation by active gate section
variation or venting valve modification;
• prediction of the durability of the die by simulating the
deterioration mechanisms;
• control of die surface lubrication by temperature, flux and
direction sensors as well as thermo-regulation by change
in temperature, flux and medium consistency;
• efficient thermo-regulation by temperature control of
heating/cooling
media
and
by
time
activation/deactivation, to optimise the heat balancing
of the die during production and/or in warm-up phase.
Making the Process Setup and Optimization a
Knowledge-Based Issue
The HPDC process only seems to be a “fully” automated
production process that promises repeatability, high
production rate and automation in any phase of the
production cell. Practically, the IT control of the process is
attributed to a single mechanism with active
synchronisation, but excluding any interaction with product
quality and equipment performance. The result is that the
process setup and its cost optimisation are essentially based
on the skills and competenceies of few persons. In other
words, it can be said that HPDC technology is more a
“personal” than the company’s ownership. This strongly
impacts production costs, whose quantification becomes
uncertain and variable. Furthermore, a traditional quality
management perspective is employed in HPDC, assuming that
the failure costs decrease as the money spent on appraisal
Case History
and prevention increases, and that there is an optimum
amount of quality effort to be applied in any situation, which
minimizes the overall investments. As quality effort is
increased, the costs of providing this effort – through extra
quality controllers’ inspection procedures and so on – get
higher than the benefits achieved. However, it is interesting
to note that the costs of errors and faulty products decrease
due to these investments. Furthermore, in this traditional
perspective, the inevitability of errors is accepted, failure
costs are generally underestimated (“reworking” defective
products, “re-serving” customers scrapping and materials,
loss of goodwill, warranty costs, management time wasted,
loss of confidence among operations processes), and
prevention costs are considered inevitably high. The
importance of quality to each individual operator is not
assessed, and preventing errors is not considered an integral
part of everyone’s work. Training, automatic checks, anything
which helps to prevent errors occurring in the first place are
often considered simply as costs. The breakthrough expected
for the HPDC process is the change from a simple mechanism
input setup to dynamic total quality management. This
implies that the set up process is accelerated and that a
continuous cost optimisation activity will autonomously run,
parallel to the process itself. The Total Quality Management
(TQM) method, which stresses the relative balance between
different types of quality cost, rather than looking for
“optimum” levels of quality effort, must be implemented in
HPDC foundries. TQM approach emphasizes prevention to stop
errors happening in the first place, rather than placing most
emphasis on appraisal. The more effort that is put into error
prevention, the more internal and external failure costs are
reduced. Then, once confidence has been firmly established,
appraisal costs can be reduced. Eventually, even prevention
costs can be decreased in absolute terms, though prevention
remains a significant cost in relative terms. Initially, total
quality costs may rise as investment in some aspects of
prevention as mainly training is increased. However, a
reduction in total costs will quickly follow.
Process data will be organised thanks to Statistical Process
Control (SPC) and Control Charts, to see whether the process
looks as though it is performing as it should, or alternatively
whether it is going “out of control”. An equally important
issue to take into account is if the variation in the process
performance is acceptable to external customers. This will
depend on the acceptable range (called specification range)
of performance that will be tolerated by the customers.
Multi-disciplinary R&D Activities
The fact that several disciplines and competenceies are
needed to run the whole HPDC design and manufacturing
chain is obvious. Process metallurgy, machine design,
automation, numerical simulation, heat transfer, furnace and
die design and construction, and many other are involved.
The actual problem is that the approach to HPDC is typically
mono-disciplinary: dies are designed and built for
productivity, but not interfaced with control systems,
lubrication is carried out for die safety and cycle time,
Newsletter EnginSoft Year 9 n°2 -
21
without taking into account metallurgical
quality, high productivity is always targeted
but a real cost analysis is never carried out.
The level of interaction among the disciplines
playing a role in HPDC field is still poor and
limited: each of them is certainly introducing
relevant innovations, which, however, are in
most cases not integrated or are fully
separated. The impact of a specific innovation Fig. 6 - Weight savings and market penetration for Al alloys in automotive applications
on productivity and costs is experimented directly in-field, with
Impacting on HPDC Foundries Scenario
high cost and no consistent background. Although, innovation
The relevant figures about the situation of EU HPDC foundries
is strongly needed by HPDC foundries (and they know this very
are the following:
well) they have no scientific, technological and organisational
• there are more than 2000 light alloys foundries in Europe
background to develop and apply it. These foundries are very
(Table 2);
limited through their SMEs structure, which intrinsically stops
• they are basically SMEs, with an average number of emplomulti-disciplinary innovation, which, on the other side, is the
yees around 50 (though most of them have less than 20
key for survival.
employees);
• the end users of diecast products are the transport industry
Methods and tools for the integration of different technologies,
(60%), mechanics (7%), electro-mechanics (9%), civil engidisciplines and competences are strongly needed, to lead the
neering (20%), with a growing trend in automotive and tranHPDC manufacturing sector towards a more knowledge-based
sport, (Fig. 5), supported by the reduction achievable in fuel
and interactive approach, leading to the optimal use of personal
consumption and emission (Fig. 6);
and company resources. Only an integrated use of each personal
• the production, due to the well-known effects of the crisis,
skill and expertise, in order to have a complete quality
has been strongly reduced in 2008 and 2009, with a partial
management of the HPDC process and products, will assure the
recovery in 2010 (see also Table 3);
development of a “new generation” of HPDC foundries.
• almost 30% of HPDC machines have been installed in foundries more than 25 years ago, thus they are very close to
obsolescence;
Italy
Germany
France
Poland
UK
• HPDC technology makes use, in almost all cases, of re-cycled
(the so-called “secondary”) alloys, with a relevant savings in
2008
960
346
335
245
236
terms of energy and natural resources.
2009
920
344
319
n.d.
220
Table 2 - Number of non-ferrous foundries in Europe (Al alloys foundries are
the 80% of them)
Country
2006
2007
2008
2009
2010*
Germany
773.000
882.000
802.000
560.000
759.000
Italy
897.000
912.000
820.000
560.000
727.000
Spain
129.000
125.000
110.000
81.000
94.000
Sweden
55.000
57.000
51.000
31.000
38.000
Table 3 - Production of Al alloys castings in European Countries (values in
tons): about 60% of the production is obtained by HPDC (*estimated)
For EU HPDC foundries and their survival of the crisis,
competitiveness, efficiency and innovation will be crucial. Of
key importance in the next years will be the implementation of
focussed dissemination activities for the personnel of the
companies involved in the HPDC design and manufacturing
chain. The concepts of multidisciplinary integration and
effective interaction for quality management need to be shared
and “absorbed” by anyone in the chain.
Performing these actions will allow HPDC foundries to achieve a
more mature and efficient approach to address large end-users,
and to exploit their relevant potential.
Acknowledgements
This article is the result of a survey carried out by the authors
with key people in the HPDC fields. In particular, we would like
to thank: Jeorg and Uwe Gauermann (Electronics GmbH), Lothar
Kallien (GTA, Aalen University), Marc Schnenider (MAGMA
GmbH), Lars Arnberg (NTNU, Trongheim University), Aitor Alzaga
(Tekniker), Luca Baraldi and Flavio Cecchetto (MOTUL).
Fig. 5 - Data on Aluminium alloys use in automotive
For more information:
Franco Bonollo, Giulio Timelli - Università di Padova - DTG,
Vicenza, Italy
Nicola Gramegna - EnginSoft, Padova, Italy
[email protected]
Case History
22 - Newsletter EnginSoft Year 9 n°2
Reducing Fuel Consumption, Noxious Emissions
and Radiated Noise by Selection of the Optimal
Control Strategy of a Diesel Engine
Despite recent efforts devoted to the development of
alternative technologies, it is likely that the internal
combustion engine will remain the dominant propulsion
system for the next 30 years and beyond. Due to more and
more stringent emission regulations, methods and
technologies able to enhance the performance of these
engines in terms of efficiency and environmental impact are
strongly required.
Our present work focuses on the development of a numerical
method for the optimization of the control strategy of a
diesel engine equipped with a high pressure injection
system, a variable geometry turbocharger and an Exhaust
Gas Recirculation (EGR) circuit. In this article, we present a
preliminary experimental analysis for the characterization of
the considered six-cylinder engine under various speeds,
loads and EGR ratios. The fuel injection system is separately
tested on a dedicated test bench, to determine the
instantaneous fuel injection rate for different injection
strategies. The collected data are employed for tuning
proper numerical models, able to reproduce the engine
behavior in terms of performances (in-cylinder pressure,
boost pressure, air-flow rate, fuel consumption), noxious
emissions (soot, NO) and radiated noise. In particular, a 1D
tool is developed with the aim of characterizing the flow in
the intake and exhaust systems and predicting the engineturbocharger.
Matching conditions, by including a short-route EGR circuit.
A 3D model (AVL Fire™) is assessed to reproduce into detail
the in-cylinder thermo-fluid-dynamic processes, including
mixture formation, combustion, and main pollutants
production. An in-house routine, also validated against
available data, is finally developed for the prediction of the
combustion noise, starting from in-cylinder pressure cycles.
Case History
Obviously, data exchange between the codes is previewed.
The overall numerical procedure is firstly checked with
reference to the experimentally analyzed operating points.
The 1D, 3D and combustion noise models are then coupled
to an external optimizer (modeFRONTIER™) in order to
select the optimal combination of the engine control
parameters to improve the engine performance and to
contemporary minimize noise, emissions and fuel
consumption. Under the hypothesis of a pilot-main
Table 1 - Engine data
Fig. 1 - Architectural layout of the engine
Newsletter EnginSoft Year 9 n°2 -
injection strategy, a multi-objective optimization problem
is solved through the employment of a genetic algorithm.
Eight degrees of freedom are defined, namely start of
injection, dwell time, energizing time of pilot and main
pulses, EGR valve opening, throttle valve opening, swirl
level, and turbine opening ratio. As this work shows, nonnegligible improvements can be gained, depending also on
the importance given to the various objectives.
EXPERIMENTAL ANALYSIS
The considered engine is an in-line six-cylinder
turbocharged Diesel engine, equipped with a common rail
fuel injection system (CR-FIS). The main engine
characteristics data, together with its architectural layout,
are reported in Table I and Figure 1, respectively.
A comprehensive characterization of the engine behavior in
terms of energy conversion performance, noxious emissions
and radiated noise is obtained under different operating
conditions.
A very important issue, in a 3D simulation, is the correct
specification of the fuel injection profile. In fact, the latter
has a great impact on the spray development, air mixing,
and fuel impingement/evaporation. A characterization of
the hydraulic behavior of the six-hole injector is made by
measuring the instantaneous mass flow rate on a dedicated
test bench under different injection strategies and rail
pressures. However, since the tested points cannot cover all
the engine working conditions, a procedure is developed to
gain parameterized injection mass flow rates, starting from
the available ones, and only using, as input variables, data
relevant to rail pressure, energizing current, and total
amount of injected fuel.
1D MODEL
A one-dimensional simulation code is employed to initially
predict the performance of the considered engine. The 1D
code solves the mass, momentum and energy equations in
the ducts constituting the intake and exhaust system, while
the gas inside the cylinder is
treated as a zero-dimensional
system.
3D MODEL
The multidimensional modeling
of the in-cylinder processes
characterizing the operation of
the considered engine, is
realized within the AVL Fire™
software environment. It is
conceived to obtain a rather
large database of results to be
processed by means of a multiobjective optimization tool.
Therefore, the engine model is
built by introducing simplifying
assumptions allowing cost-
23
effective solutions in terms of amount of time needed for
the computation of each engine operating condition. The
simplifications are properly chosen, depending on factors
including the required level of accuracy and the available
computing power, but also on the basis of the authors'
knowledge of similar engines and of high pressure injection
systems for Diesel fuel.
COMBUSTION NOISE ESTIMATION PROCEDURE
In order to understand the effects of the combustion
characteristics on the overall engine noise, an empirical
model is developed for the engine under test, whose results
are compared with those experimentally measured through
the employment of a noise meter instrument. The overall
noise is predicted starting from the in-cylinder pressure
data. The procedure is suitable to be implemented within
the optimization process.
OPTIMIZATION
The previously described 1D, 3D and combustion noise
models constitute the basis for the optimization of the
control strategy of the considered engine. A reference
operating condition, corresponding to the experimental
case measured at 1500 rpm, 4 bar brake mean effective
pressure (BMEP), is chosen (overall mass of injected fuel Mf
= 13.64 mg/cycle/cylinder).
The logical development of the optimization problem within
the modeFRONTIER™ environment is explained in Figure 2.
Basing on the values of 8 degrees of freedom (start of
injection, dwell time, energizing time of pilot and main
pulses, EGR valve opening, throttle valve opening, tumble
port valve opening, and turbine opening ratio), a Fortran
routine firstly computes the coordinates of the 8 points (xi,
yi, i=1,8) defining the overall injection profile, for the
assumed constant amount of total injected fuel, Mf. The
injection profile, together with the values of other control
parameters, are written in the input files of both 1D and 3D
codes, which are then sequentially executed. In particular,
Fig. 2 - Workflow of the optimization problem developed in modeFRONTIER environment
Case History
24 - Newsletter EnginSoft Year 9 n°2
the 1D code computes the engine-turbocharger matching
conditions and passes to the 3D code the cylinder-bycylinder averaged pressure, temperature and composition at
Inlet Valve Closing (IVC), together with the estimated swirl
level. At the end of each 3D run, a proper script routine
extracts the 3D computed pressure cycles and gives them in
input to the Matlab™ procedure, predicting the overall
combustion noise. Simultaneously, the NO and Soot levels at
the end of the 3D run are returned back to the optimizer.
during the optimization loop, the 3D code is executed
during the closed valve period, while the 1D code gives
information on the mass exchange phase. This allows
reconstructing the whole pressure cycle and the related
Indicated Mean Effective Pressure (IMEP). The BMEP is then
obtained thanks to a mechanical loss correlation. As a
consequence, the brake specific fuel consumption (BSFC),
can be estimated as:
where:
• Mf: mass of injected fuel
• BMEP: brake mean effective pressure
• V: engine displacement
(#112), low soot (#263) and low fuel consumption (#529).
The latter parameter is here considered as the most
important one. Each solution, in fact, provides a substantial
BSFC improvement with respect to the initial reference case
(#000). Table II also highlights the best and worst objective
levels, which are illustrated in blue and red colors.
Figure 4 compares the pressure cycles and noxious emissions
in the selected optimized solutions. A very advanced
injection process with a small pilot is specified in solution
#081; it realizes the best NO emission, which is even lower
than for the reference #000 case. A low BSFC is attained in
this case as well. The low NO emission is determined by the
occurrence of a premixed low-temperature combustion
(LTC), diluted by the presence of a high residuals
concentration. Soot emission is indeed comparable with the
reference case. A slight delay of the injection process,
coupled with a greater pilot injected fuel (#529) and a
reduced EGR amount, produces a strong NO increase, but
also provides the best result in terms of fuel consumption,
with a reduced soot emission. A further soot reduction is
obtained with a single-shot injection and a negligible EGR
rate in solution #263. In this case, however, both NO
emission and radiated noise reach very high values. Solution
#112, finally, presents a more delayed injection and a very
low EGR value. This is the way to obtain a reduced noise
As a next step, a multi-objective optimization is
defined to contemporary search the minimum
BSFC, the minimum Soot, the minimum NO and
the minimum overall noise. To solve the above
problem, the MOGA-II algorithm is utilized,
which belongs to the category of genetic
algorithms and employs a range adaptation
technique to overcome time-consuming
evaluations. As usual with multi-objective
optimization problems, a multiplicity of
solutions is expected, belonging to the so-called
Pareto frontiers.
Figure 3 displays a bubble chart of the
optimization process. It highlights the wellknown trade-off between NO and soot emissions.
Each bubble is colored proportionally to the BSFC
level, while the bubble size is proportional to
the externally radiated Sound Pressure Level
(SPL). This representation allows to easily locate
the best efficiency solutions (dark blue points).
These, however, also realize a considerable NO
Fig. 3 - Bubble chart of the optimization process
emission and a generally high radiated noise.
As expected, it is not possible to
identify a unique optimal solution; a
different choice must be made
depending on the importance
assigned to the various objectives.
As an example, Table II reports the
accomplishment of the solutions,
respectively, low NO emissions
(#081), acceptable radiated noise Table II - Values of the objective parameters in selected optimized solutions
Case History
Newsletter EnginSoft Year 9 n°2 -
25
utilization of a 3D CFD code. This
allows exploring the effects of the
injection strategy, EGR and swirl
levels on the in-cylinder pressure
cycle and the main pollutants
formation. The results of the fluiddynamic analyses also allow the
prediction of the radiated noise,
based
on
the
structural
characteristics of the engine under
consideration.
The coupling of these different
procedures with an optimization
Fig. 4 - Comparison of pressure cycles and noxious emissions in selected optimized solutions
code is realized to analyze very
different operating conditions. The developed procedure is
level, while maintaining fuel consumption very low. In this
able to select a proper combination of 8 control parameters,
case, unfortunately, the NO emission is much higher than in
identifying the best compromise solution between the
the reference case. The presented results highlight the
predefined objectives of low fuel consumption, low soot and
difficulties encountered in realizing an improvement among
NO emissions and low radiated noise. The methodology
all the considered objectives. The best compromise solution
presented has the main advantage of contemporary
is probably obtained in solution #081, which confirms the
considering many different aspects in the engine tuning
potential of the recently proposed innovative combustion
process, and highlights the difficulties in the selection of
modes. A non-negligible penalty on radiated noise,
the optimal solution, due to the presence of conflicting
however, has to be paid.
needs. Nevertheless, it is shown that it is possible to
improve the experimentally defined set point, mainly in
The above discussion demonstrates the difficulties in
terms of fuel consumption and NO emission, with a nonidentifying an optimal set of parameters, able to comply
negligible penalty however, on the radiated noise (solution
with so many conflicting needs, even when a modulated
#081).
injection process is considered and a high number of
degrees of freedom are available for the engine control. The
The optimal setting of the control parameters confirms the
best compromise is definitely accomplished through
trend of some current research in the field, going towards
increased EGR, boost and swirl levels and takes advantage
operating conditions characterized by increased EGR and
of an early injection strategy. Although the relationship
swirl levels, together with the arrangement of an early
between the overall engine behavior and the control
injection strategy. These settings, in fact, promote a typical
parameters is roughly expected, the proposed methodology
low temperature combustion, where NO emission can be
offers the advantage of quantifying the estimated
minimized without significant detrimental effects on soot.
improvements, taking into account mutual dependencies
and cross-correlation effects. In addition, the selection of
In the future, a further validation of the 3D code, together
the optimal solution can be carried out through a
with the inclusion of a kinetic scheme, will allow to more
standardized set of designer preferences, depending on the
directly analyze the recently proposed combustion modes
importance assigned to the single objectives. These may
and to extend the methodology to multiple operating
vary in the different operating conditions, say by focusing
conditions and objectives (CO and HC production). In this
the attention on the noxious emissions at low load, while
way, a variable combustion mode can be realized,
improving the performance and reducing the noise emission
depending on the working conditions and selected
at high load.
objectives.
CONCLUSION
This article presents integrated numerical techniques aimed
Daniela Siano
to characterize the behavior of a six-cylinder Diesel engine
Istituto Motori CNR - Italy
equipped with a CR-FIS, from both an energetic and an
Fabio Bozza
environmental point of view. In particular, a 1D code is
Universita' di Napoli Federico II Italy
employed to evaluate the overall engine performance by
Michela Costa
varying the control parameters that affect the engineIstituto Motori CNR – Italy
turbocharger matching conditions and the EGR level. The
classical limitations of this kind of numerical approach, i.e.
For more information:
the absence of a description of the mixture formation and
Francesco Franchini, EnginSoft
emission production processes, are overcome through the
[email protected]
Case History
26 - Newsletter EnginSoft Year 9 n°2
Fluid Refrigerant Leak in a Cabin Compartment:
Risk Assessment by CFD Approach
In the framework of the Greenhouse gas emissions reduction
initiatives the European Union (EU) requires that the airconditioned vehicles sold in EU countries use refrigerants
with Global Warming Potential (GWP) less than 150 starting
by January 1st, 2011 for new vehicles type and in all vehicles
by 2017 (regulation 2006/40/EC).
The air conditioning system of passenger cars, trucks and the
vehicles manufactured uses the hydro fluorocarbon R-134a as
refrigerant that has no effect on the atmospheric ozone layer
but has a GWP of 1430.
The Honeywell-DuPont fluid refrigerant named HFO-1234yf is,
at the moment, the first option to replace the R-134a.
Critical open issues of this fluid are:
• It is flammable (R12 classified), even if its flammability
is low (ASHRAE A2L);
• when burning, or a contact with hot surfaces, it produces
toxic gases (HF);
• high cost.
The present work has been focused to evaluate the risk of
burning associated to the new refrigerant fluid leak in the
cabin due to an evaporator fault in case of a vehicle crash
and in case of exchanger corrosion. A 3D model of a
B-segment passenger car cabin and HVAC module and ducts
has been analyzed. By means of CFD
approach the refrigerant concentration has
been identified for different working
conditions.
Activity objective
The purpose of the work is to predict the
refrigerant leakage distribution into the
passenger compartment both in case of a
vehicle crash and in case of evaporator
Case History
corrosion, in several working conditions, to identify the
potential risk of refrigerant burning in the cabin.
For this type of damages the leak diameters has a standard
size:
• evaporator damage due to the vehicle crash => Φ = 6.35
mm leak
• evaporator damage due to the corrosion => Φ = 0.5 mm
leak
With the CFD approach it is possible identify the refrigerant
flammable concentration within the passenger cabin.
Cabin model
The car cabin selected is a B-segment vehicle with 2 m3 of
internal air, typical size of a small-medium European vehicle.
This low air volume set the worst condition because the
refrigerant concentration will be greater than an higher
volume cabin with the same refrigerant leak rate.
The geometry will include the following objects:
• full cabin model with 1 and 4 passengers;
• HVAC (Heat Ventilation and Air Conditioning module with
condensate drain box);
• complete air distribution system: ducts, vent outlets with
vanes, floor ducts and outlets.
Fig. 1 - CAD of passengers, cabin, HVAC, ducts and outlets
Newsletter EnginSoft Year 9 n°2 -
27
Starting from the CAD evaporator surface
the effective one is defined: the real
opening area due to the difference between
the CAD surface and the closed surface of
the radiator pipes and fins. This area (green
colour in fig. 3) is very important to define
the correct air velocity from the evaporator
once the air mass flow data is known,
because the air velocity is the most
responsible of the refrigerant diffusion.
Fig. 2 - HVAC simplified, rear floor ducts and outlets
The condensate drain box is take into
account to consider the potential discharge of air and
refrigerant leak through this outlet. The 6.35 mm and 0.5 mm
leaks are placed in the middle of the effective evaporator
surface.
Cabin air exhaust
The cabin ventilation valves are modelled with three areas:
• the areas position is coherent wih the real air path
between the cabin and boot compartment;
• the areas size has been defined from experimental data to
reproduce the cabin fluid dynamic resistance value: 105
Pa @ 350 m3/h.
Fig. 3 - evaporator and leak geometry details
Mesh model
The mesh model has been realized in ANSYS Icem CFD
environment using tetrahedron cells mesh. The mesh
dimension realized is different for the several components,
with a high density into the HVAC to allows the code to
simulate the correct refrigerant transport phenomena by
means the air. Six different mesh model has been build-up for
the different geometry configurations: Vent, Floor, Bi-level
air distribution with 1 and 4 passengers. The mean mesh
dimension is equal to 7 million of cells.
Fig. 4 - cabin ventilation valves
Geometry
In the figure 1 are present the cabin components take into
account.
HVAC and floor ducts
The HVAC has been simplified respect the
original geometry because the air mass
flow is known, so it is not necessary to
introduce the blower and the evaporator
with their performance curve (fan pressure
and evaporator pressure drop). Therefore
the evaporator has been defined with its
surface toward the cabin only. Moreover
the air mixer valve is set in max cold for
the simulation, so the heater and its duct
portion into the HVAC is not considered.
Simulation conditions
The mesh model has been solved with ANSYS CFX-13
numerical code, using 8 parallel processors. The physical
conditions set are the following:
• transient simulation;
• thermal energy;
• turbulent model = K-ε;
• buoyant model;
• fluid model = variable composition mixture.
Fig. 5 - surface mesh of vent duct and front floor outlets
Case History
28 - Newsletter EnginSoft Year 9 n°2
Fig. 6 - surface mesh of HVAC with rear floor ducts and rear floor outlet
Where:
• G [kg/s] is the fluid mass flow
• ρf [kg/m3] is the density of the fluid
• Pc [Pa] is the critical pressure
• f(.) [-] is the correlation
• π1,…, πn are the non-dimensional
parameters
• a1,…, an are the coefficients of the correlation
For the two-phase flow the calculation requires the
evaluation of single phase flow at saturated upstream
conditions, then the basic equation for this condition is:
Fig. 7 - surface mesh with 4 passengers
The fluid components are defined by means of a variable
mass fractions (Variable Composition Mixture). In the
solution to a multicomponent simulation, a single velocity
field is calculated for each multicomponent fluid.
Individual components move at the velocity of the fluid of
which they are part, with a drift velocity superimposed,
arising from diffusion.
The transient model is used to take into account the real
amount of refrigerant fluid in the air conditioning system and
its discharge into the cabin, setting the correct discharge
time in the code. After the complete refrigerant delivering in
the cabin, an additional transient time of 30 minutes is
simulated to understand the diffusion of the refrigerant in
the cabin itself.
Where:
• C(.) is the single versus two phase flow
• tp1,…, tpn are the non-dimensional parameters
• b1,…, bn are the coefficients of the correlation
• sp and tp indicate the single and two phase flow,
respectively
For the present application, the hole on the evaporator has
been assumes equal to an orifice, and applying the Payne
model: the mass flow estimated are reported on the table 1.
Ambient @ 56°C
Ambient @ 15°C
AC on @
sat 4°C
AC off @
sat 56°C
AC on @
sat 4°C
AC off @
sat 15°C
hole 0.5 mass flow [g/s]
0.05
0.09
0.02
0.02
hole 6.35 mass flow [g/s]
7.61
13.95
2008
2008
Table 1 - fluid refrigerant mass flow
In particular, at 15 deg. of ambient temperature a leakage
condition of double phase refrigerant fluid has been defined
with a quality of 0.7 liquid and 0.3 vapor both for driving and
parking condition. Whereas at 56 deg. of ambient
temperature the refrigerant leakage will be in vapor phase
only.
Leak rate calculation – Payne model
The aim of the model is to generalize the correlation for
refrigerant mass flow through a short tube orifice used in
vapor compression cycle. Moreover, the model desired has to
correlate the mass flow rate of several different refrigerants
into a single closed-form equation, capable of predicting
mass flow rate over a wide range of conditions. The Payne
model adopted predicts both single phase and two-phase
mass flow rate.
Fluid models
About the gas-gas model, the fluid components are defined
by means of a variable mass fractions (Variable Composition
Mixture): additional transport equations are applied by the
code.
The Payne model is a semi-empirical one, based on nondimensional group build from geometrical parameter of the
orifice (as the hole diameter) and from the thermo-fluid
dynamics characteristic of the fluid (as the critical pressure
and temperature, density). The basic equation of the model
for a single phase flow is:
About the liquid-gas model, the liquid evaporation model is
a model for particles with heat and mass transfer.
The model uses two mass transfer correlations depending on
whether the droplet is above or below the boiling point.
The boiling point is determined through the Antoine
equation (vapor pressure equation that describes the relation
Case History
Newsletter EnginSoft Year 9 n°2 -
between vapor pressure and temperature for
pure components) that is given by:
Leak
Rate
where A is the Antoine Reference State
Constant, B is the Antoine Enthalpic
Coefficient and C is the Antoine Temperature
Coefficient. The particle is boiling if the vapor
pressure, psat, is greater than the ambient gas
pressure, pambient.
In the next picture the Antoine equation curve
extracted from the fluid refrigerant data.
To verify the difference between the gas and
the liquid leak, an additional test is done
using a reduced mesh model with the HVAC
only in front floor distribution. The boundary
conditions set are illustrated in Table 2.
Where ACH means air change per hour.
• 0.09 g/s (hole 0.5 mm
diameter) @ 56°C
ambient
• 13.95 g/s (hole 6.35
mm diameter) @ 56°C
ambient
• 0.02 g/s (hole 0.5 mm
diameter) @ 15°C
ambient
• 4.02 g/s (hole 6.35
mm diameter) @ 15°C
ambient
Sim.
Number
Parking
Thermodynamic
conditions
Air Exchange
rate
• 100% vapor @ 56°C
ambient
• 70% liquid, 30% vap
@ 15°C ambient
• 1.0 air changes
per hour @ T
ambient
• 2.5 air changes
per hour @ T
ambient
Fluid in saturation
conditions @ 56 and
15°C
29
Air Path
• dashboar
d outlets
• floor
outlets
Pass
• driver only
•4
passengers
32
Table 3 - number of simulations and boundary conditions in parking situation
Leak
Rate
• 0.05 g/s (hole 0.5 mm
diameter) @ 56°C
ambient
• 7.61 g/s (hole 6.35
mm diameter) @ 56°C
ambient
• 0.02 g/s (hole 0.5 mm
diameter) @ 15°C
ambient
• 3.48 g/s (hole 6.35
mm diameter) @ 15°C
ambient
Sim.
Number
Parking
Thermodynamic
conditions
Air Exchange
rate
• 100% vapor @ 56°C
ambient
• 70% liquid, 30% vap
@ 15°C ambient
• 2.5 air changes
per hour @ T
ambient
• 50 l/s @ 4°C
• 130 l/s @ 4°C
Fluid in saturation
conditions @ 56 and
15°C
Air Path
• dashboar
outlets
• floor
outlets
• bi-level
outlets
Pass
• driver only
•4
passengers
72
Figure 9 show the difference between the gas
and the liquid ejection:
• the gas leak is more diffused than the Table 4 - number of simulations and boundary conditions in driving situation
liquid near the leak hole yet;
• the liquid jet has a greater penetration towards the ducts
and therefore towards the cabin in spite of a lower mass
flow.
This means that it is possible to have a similar amount of
flammable refrigerant fluid in the passenger compartment
between vapor and liquid leak with different leak rate as long
as the air mass flow and the air distribution are the same.
Parking mode @ 56 ˚C
Parking mode @ 15 ˚C
Parking mode @ 15 ˚C
4.02 liquid-vapour leak
(quality 0.7 liq., 0.3 vap.)
Leak diameter [mm]
6.35
6.35
Air mass flow [ACH]
2.5 @ 56 ˚C
2.5 @ 15 ˚C
Air distribution
Floor
Floor
Simulation time [s]
32.26
32.26
Mass flow [g/s]
Table 2 - boundary conditions to compare/verify the vapour and liquidvapour models
Fig. 8 - Antoine equation curve for the fluid evaporation model
Fig. 9 - comparison between gas and liquid refrigerant leak in a simplified
model
Configuration of the simulations
The refrigerant charge considered is equal to 450 grams for
the vehicle selected (mean optimal charge for A and B
segment Fiat vehicles).
Simulation phases:
1. first phase (duration depending on leak rate) => full
refrigerant charge release
2. second phase (additional 30 minutes) => refrigerant
diffusion into the cabin
In the following table the simulations conditions analyzed.
Results analysis
The ANSYS CFD code allows to simulate the transport and the
diffusion of a gas into the air of which it is a part, impossible
to understand by experimental test because of the discrete
and poor number of sensors detector and their intrusive
geometry.
At the same time it is possible simulate the evaporation and
the transport with the diffusion of a liquid within the air.
Case History
30 - Newsletter EnginSoft Year 9 n°2
Design of experiments
The total simulations planned are 104. After some
simulations done it is clear that the configurations with 0.5
mm diameter of leak hole are not dangerous, it means that
there is no risk to reach the refrigerant flammable
concentration into the cabin compartment (both in driving
and in parking conditions, at ambient temperature of 15 and
56 °C).
This is the effect of the very low refrigerant mass flow,
therefore the flammable volume concentration remains very
closed to the evaporator (into the HVAC). Due to these
results some numerical experiment in these conditions have
not been done.
The configurations with 6.35 mm diameter of leak hole have
a potential dangerous, it depends on the HVAC blower speed
if it is switched on or off.
• blower on: there is not any risk, both with minimum (50
l/s) and maximum air mass flow produced;
• blower off: with the air mass flow moving in natural convection mode (1 ACH and 2.5 ACH) there is an amount of
refrigerant fluid with flammable concentration in the
cabin.
Results
In the next sections are presented the main results.
The simulations results are very similar between one and four
passengers, therefore only four passengers configurations are
shown and only the significant results.
About the simulations with 0.5 mm leak hole only one
configuration results are described because the other ones
give the same not dangerous results.
The results are defined in the refrigerant volume fraction
distribution between the lower and upper flammability level
of the reference refrigerant fluid.
Driving mode @ 56 ˚C
Boundary conditions:
• refrigerant mass flow: 7.61 g/s vapor leak
• leak diameter: 6.35 mm
• air mass flow: 2.5 ACH @ 56 ˚C, 50 l/s @ 4 ˚C, 130 l/s
@ 4 ˚C
• air distribution: bi-level, vent, floor
• 4 passengers
• time to discharge: 59.13 s
Case History
In any configuration with a leak hole of 0.5 mm (evaporator
corrosion) the refrigerant flammable concentration is very
closed to the evaporator, thus it is not dangerous for the
occupant, even with air in natural convection mode.
Driving mode @ 15 ˚C
Boundary conditions:
• refrigerant mass flow: 3.48 g/s, liquid-vapor leak (quality
0.7 liq., 0.3 vap.)
• leak diameter: 6.35 mm
• air mass flow: 2.5 ACH @ 15 ˚C, 50 l/s @ 4 ˚C, 130 l/s
@ 4 ˚C
• air distribution: bi-level, vent, floor
• 4 passengers
• time to discharge: 129.3 s
Newsletter EnginSoft Year 9 n°2 -
Parking mode @ 56 ˚C
Boundary conditions:
• refrigerant mass flow: 13.95 g/s vapor leak
• leak diameter: 6.35 mm
• air mass flow: 1.0 ACH @ 56 ˚C, 2.5 ACH @ 56 ˚C
• air distribution: vent, floor
• 4 passengers
• time to discharge: 32.26 s
31
Conclusions
The ANSYS CFD code allows to simulate the transport and the
diffusion of a gas within the air of which it is a part,
impossible to understand by experimental test because of the
discrete and poor number of sensors detector and their
intrusive geometry. With the code it is possible to identify
the volume location of the flammable fluid concentration and
propose the corrective actions.
The analysis, within the considered perimeter and test cases,
allows to affirm that if the flammable refrigerant fluid is
adopted as refrigerant in a conventional direct expansion
automotive air conditioning, in case of:
Leak due to corrosion
a) there is no formation in the passenger compartment of
zones where the refrigerant concentration is within the
flammability region
Leak due to an evaporator crash
b) there is no formation in the passenger compartment of
zones where the refrigerant concentration is within the
flammability region when the blower is active: air flow
higher or equal to 50 l/s
c) there is formation in the passenger compartment of small
volumes (front lower part) where the refrigerant
concentration is within the flammability region if the
blower is off and the air moves in natural convection
mode only (ACH lower or equal to 2.5)
Parking mode @ 15 ˚C
Boundary conditions:
• refrigerant mass flow: 4.02 g/s liquid-vapor leak (quality
0.7 liq., 0.3 vap.)
• leak diameter: 6.35 mm
• air mass flow: 1.0 ACH @ 15 ˚C, 2.5 ACH @ 15 ˚C
• air distribution: vent, floor
• 4 passengers
• time to discharge: 111.94 s
Further comment on case c):
• if the vehicle is parked with the A/C off it is not probable
that the evaporator has a dramatic fault like welding
problem
• if the evaporator crash is a consequence of a vehicle
accident likely, surely the engine compartment will be
damage seriously therefore the fluid will discharge
completely into the ambient due to the refrigerant pipes
collapse
• if the crash is a consequence of an accident likely, there
will be additional ventilation due to the cabin collapse
Therefore the case c), the only critical one, is not easy to get
in real application.
Fabrizio Mattiello
Centro Ricerche Fiat s.c.p.a - Thermofluid-dynamic and air
Conditioning Senior Specialist, Orbassano (Turin), Italy
Case History
32 - Newsletter EnginSoft Year 9 n°2
Numerical Optimization of the Exhaust Flow of a
High-Performance Engine
An automotive catalytic converter is a crucial yet sensitive
component of a vehicle’s exhaust system; its impact on the
overall design is considerable. In order to achieve high
efficiency and durability, good uniformity of the flow inside
the monolith and low pressure gradients are indispensable,
both depend heavily on the engineering design of the
exhaust pipes. Many of today’s high-performance engines rely
on exhaust systems with a minimum pressure drop. A joint
approach based on 3D CFD and
optimization tools can help engineers
to significantly improve the design of
the component and to meet different
objectives: high engine performance,
high emission reduction and long
durability of the catalytic converter.
This article describes how the
embedded CAD instruments of the CFD
code Star-CCM+ were linked with and
led by modeFRONTIER to modify the
geometry of an exhaust system. Using
this approach, the above mentioned
objectives could be achieved. The
numerical methodology was developed
and validated on a real exhaust system and delivered both:
increased catalytic converter efficiency and decreased
pressure drop.
engine. This approach is based on CFD simulations and
ensures the durability of the catalytic converter.
In this context, we should note that the optimization of the
exhaust flow requires a large number of simulations in order
to determine a good trade-off between the different project
targets. Engineers typically spend considerable time
modifying the geometry and setting up different
calculations. Therefore, the aim of this work was to develop
Fig. 1
Fig. 2
Aim of the work
The optimal design of an exhaust system is based on a tradeoff between the efficiency and the durability of the catalytic
converter and the efficiency of the exhaust system; the latter
is required in order to minimize the engine pumping loss.
Over the past few years, a correlation between numerical 3D
CFD results and experimental results has been developed
which allows to properly design the exhaust system of an
Fig. 3
Case History
Newsletter EnginSoft Year 9 n°2 -
33
• Pgrad, which is the radial pressure gradient. It was
recorded on plane C1, and it has to be lower than 20
mbar/mm in order to ensure the honeycomb structure
integrity.
Optimization methodology
Fig. 4 shows the modeFRONTIER layout. As we can see, the
displacement of the sketches and the control points
represents the input variables of the optimization process.
The objectives of the calculation are: to minimize the
pressure drop for each primary pipe, to minimize the mean
radial pressure gradient and to maximize the mean uniformity
index. Several constraints have been adopted to discard the
designs with a radial pressure gradient higher than 20
mbar/mm and with a Gamma factor lower than 0.85.
Fig. 4
In order to include computational costs starting from a
random DoE for about 20 designs, the response surfaces of
the objectives have been generated as functions of the input
variables using a RBF interpolation algorithm. This allowed
to perform a virtual optimization by using the MOGA-II
genetic algorithm. In a final step, the best designs were
an automated process in which
modeFRONTIER would drive all the
operations required for the geometry
modeling and the set up of the CFD.
The process finally allowed to
improve the design of the
component and to fulfill all the
objectives.
Geometry modeling
Exploiting the Star-CCM+ embedded
CAD features, the connecting pipe between the exhaust
manifold was directly modeled inside the CFD code, without
the use of any external CAD software, by lofting three
different sketches as illustrated in fig.1.
Fig. 5
It then became possible to modify the connecting pipe,
simply by adjusting the shape of the sketches and the
position of some control points (represented in yellow in
fig.2) or the position of the plane onto which the sketches
have been drawn.
Fig. 6
Outputs of the calculation
The work was performed for an 8-cylinder engine. The right
exhaust system comprised four primary pipes, hence four
different simulations were needed for each design.
In order to evaluate the accuracy of each design for each
simulation, three different parameters were recorded:
• DeltaP, which represents the pressure drop between the
inlet and the outlet section; it has to be as low as
possible in order to minimize the engine pumping loss;
• Gamma Factor, which is a uniformity index. It was
recorded on the plane C25, and it has to be higher than
0.85, in order to ensure a proper exploitation of the
catalyst inside the honeycomb structure;
Fig. 7
Case History
34 - Newsletter EnginSoft Year 9 n°2
Delta P [bar]
C1
C2
C3
C4
Baseline
design
1497
1293
1331
1625
Final design
1368
1103
1150
1268
-8.62%
-14.69%
-13.60%
-21.97%
Variance
Weight losses
Conclusion
Using modeFRONTIER it became possible for the authors to
develop a workflow into which all geometry modeling
operations and CFD simulations could be integrated and
automated. This approach also allowed to exploit the
response surface method in modeFRONTIER, to perform a
virtual optimization with the MOGA-II genetic algorithm. In
this way, a suitable design for the connecting pipe could be
defined by carrying out only a low number of real CFD
simulations. The developed methodology proved to be very
effective in reducing computational costs and operator time.
The best design delivered an increase of the mean gamma
factor of about +11.98%, a decrease of the mean radial
pressure gradient of circa -54.84% and a reduction of the
mean pressure drop of approximately -14.72%.
Prof. Ing. Gian Marco Bianchi,
Ing. Marco Costa, Ing. Ernesto Ravenna
University of Bologna, Italy
Current
geometry
Optimized
geometry
For more information:
Francesco Franchini, EnginSoft
[email protected]
Percentage
variation
validated with real CFD simulations. In fig. 5, the results of
the optimization process in terms of Gamma, Pgrad and
DeltaP are shown. Fig. 6 presents all the suitable real designs
out of which the solution with the lower mean pressure drop
was chosen.
Best design: Connecting pipe geometry
The geometry of the connecting pipe between the exhaust
manifold and the monolith was modified considerably by the
optimization. The resulting geometry is clearly visible in fig.
7, it looks as if it was created by a human operator.
Gamma
factor
C1
C2
C3
C4
Baseline
design
0.80
0.75
0.77
0.79
Final design
0.87
0.87
0.88
0.86
Variance
+8.75%
+16.00% +14.29%
Best design: Engine efficiency
The re-shaping of the connecting pipe assured a significant
decrease of the pressure drop for each primary pipe flow. In
fact, compared to the original layout, the mean pressure drop
of the final design was about -14.72% lower which should
lead to a decrease of the engine pumping loss and enhanced
engine performance.
Best design: Catalytic converter efficiency and durability
The best design also improved greatly the efficiency and the
durability of the monolith, both are important factors for a
high-performance engine.
In particular, each primary pipe flow proved to be capable of
satisfying both the gamma factor constraint and the radial
pressure gradient constraint. Both are mandatory to obtain
the catalytic converter manufacturer's approval. The recorded
decrease of the mean radial pressure gradient was about 54.84% while the mean gamma factor increased by about
11.98%.
Case History
+8.86%
Fig. 8
Pgrad
[mbar/mm]
C1
C2
C3
C4
Baseline design
38.45
32.07
22.35
26.63
Final design
11.93
17.38
12.00
11.11
Variance
-8.62% -45.81% -46.31% 58.28%
Newsletter EnginSoft Year 9 n°2 -
35
Multi-Objective optimization
of steel case hardening
Steel case hardening is a thermo-chemical process largely
employed in the production of machine components to solve
mainly wear and fatigue damage in materials. The process is
strongly influenced by many different variables, such as
steel composition, carbon and nitrogen potential,
temperature, time and quenching media. In the present
study, the influence of these parameters on the carburizing
and nitriding quality and efficiency was evaluated. The aim
was to streamline the process by numerical-experimental
analysis to define optimal conditions for the product
development work. The optimization software used was
modeFRONTIER, with which a set of input parameters was
defined and evaluated on the basis of an optimization
algorithm that was carefully chosen for the multi-objective
analysis.
Introduction
For the deep analysis of industrial processes which depend
on different parameters, the use of computational multiobjective
optimization
tools
is
indispensable.
modeFRONTIER is a multidisciplinary and multi-objective
software written to allow easy coupling to any computer
aided engineering (CAE) tool. modeFRONTIER refers to the
so-called “Pareto Frontier”, the ideal limit beyond which
every further implementation compromises the system, in
other words, the Pareto Frontier represents the set of best
possible solutions.
The complex algorithms in modeFRONTIER are able to spot
the optimal results, even if they are conflicting with each
other or belonging to different fields. Optimization achieves
this goal by integrating multiple calculation tools and by
providing effective post-processing tools. The more accurate
the analysis, the more complex the later design process can
be. The modeFRONTIER platform allows the organization of a
wide range of software and an easy management of the
entire product development process.
modeFRONTIER’s optimization algorithms identify the
solutions which lie on the trade-off curve, known as Pareto
Frontier: none of them can be improved without prejudicing
another. In other words, the best possible solutions are the
optimal solutions.
Generally speaking, optimization can be either singleobjective or multi-objective. An attempt to optimize a
design or system where there is only one objective usually
entails the use of gradient methods where the algorithms
search for either the minimum or maximum of an objective
function, depending on the goal. One way of handling multiobjective optimization is to incorporate all the objectives
(suitably weighted) into a single function, thereby reducing
the problem to one single- objective optimization again.
However, the disadvantage of this approach is that the
weights must be provided a priori which can influence the
solution to a large degree. Moreover, if the goals are very
different in substance (for example: cost and efficiency) it
can be difficult, or even meaningless, to try to produce a
single all-inclusive objective function.
True multi-objective optimization techniques overcome
these problems by keeping the objectives separate during
the optimization process.
We should keep in mind
that in cases with
opposing objectives (e.g.
when we try to minimize
a beam's weight and its
deformation under load)
often there will be no “Intervento cofinanziato dall'U.E. – F.E.S.R.
sul P.O. Regione Puglia 2007-2013, Asse Isingle optimum because
Linea 1.1 - Azione1.1.2. Aiuti agli
any solution will be just
Investimenti in Ricerca per le PMI”
Case History
36 - Newsletter EnginSoft Year 9 n°2
Fig. 1 - workflow description of carburizing and nitriding
statistical analysis and data visualization. Design of
Experiments (DOE) is a methodology that maximizes the
knowledge gained from experimental data. It provides a
strong tool to design and analyze experiments, it eliminates
redundant observations and reduces the time and resources
to make experiments. Hence, DOE techniques allow the user
to try to extract as much information as possible from a
limited number of test runs. DOE is generally used in two
ways. Firstly, it is extremely important in experimental
settings to identify which input variables most affect the
experiment being run. Since it is often not feasible in a
multi-variable problem to test all combinations of input
parameters, DOE techniques allow the user to try to extract
as much information as possible from a limited number of
test runs. However, if the engineer's aim is to optimize his
design, he/she will need to provide the optimization
algorithm with an initial population of designs from which
the algorithm can "learn". In such a setting, the DOE is used
to provide the initial data points.
Secondly, exploration DOEs are useful for getting
information about the problem and about the design space.
They can serve as the starting point for a subsequent
optimization process, or as a database for response surface
training, or for verifying the response sensitivity of a
candidate solution.
Such an analysis has been performed in order to define the
behavior of different industrial steels after nitriding and
carburizing. These treatments are fundamental in industrial
applications. In many cases, they require a thorough control
of the processing parameters in order to achieve the best
mechanical and microstructural properties of the
components.
a compromise. The role of the optimization algorithm here is
to identify the solutions which lie on the trade-off curve,
known as the Pareto Frontier. All these solutions have the
characteristic that none of the objectives can be improved
without prejudicing another.
High performance computing nowadays provides us with
accurate and reliable virtual environments to explore several
possible configurations. In real applications, however, it is
not always possible to reduce the complexity of the problem
and to obtain a model that can be quickly solved. Usually,
every single simulation can take hours or even days. The
amount of time needed to run a single analysis, does not
allow us to run more than a few simulations, hence some
other smart approaches are needed. These factors led to a
Design of Experiment (DOE) technique to perform a reduced
number of calculations. Subsequently, these well-distributed
results can be used to create an interpolating surface. The
surface represents a meta-model of the original problem and
Experimental and numerical procedure
can be used to perform the optimization without computing
From a database of experimental results, computational
any further analyses.
models (so-called virtual n-dimensional surfaces) have been
The use of mathematical and statistical tools to
developed. The models are able to reproduce the actual
approximate, analyze and simulate complex real-world
process in the best possible way. The analysis has made it
systems is widely applied in many scientific domains. These
possible to optimize the output variables.
kinds of interpolation and regression methodologies are now
The applied method is called RSM (Response Surface
becoming common, even in engineering, where they are also
Methodology). RSM has been used for the creation of the
known as Response Surface Methods (RSMs). Engineers are
meta-models to simulate the actual process by using
very interested in RSMs for their computational work
physical laws with appropriate coefficients to be calibrated.
because they offer a surrogated model with a second
The RSM method consists of creating n-dimensional surfaces
generation of improvements in
speed and accuracy in computer
aided engineering.
Once the data has been
obtained, either from an
optimization or DOE, or from
data import, the user can turn
to the extensive post-processing
features in modeFRONTIER to
analyze the results.
The software offers a wideranging toolbox that allows the Fig. 2 - carbon concentration on the samples surface as a function of carbon potential and carburizing temperature;
user to perform sophisticated microhardness as a function of nitriding temperature.
Case History
Newsletter EnginSoft Year 9 n°2 -
37
Fig. 3 - input-output matrixes for the carburizing and nitriding process.
that are "trained" based on the actual input and output. The
surfaces are predicated on large experimental data sets and
are able to provide the output numbers that reflect the real
process of carburizing and nitriding. In the validation phase,
they have been included in the RSM, “trained” only
according to the remaining input conditions. The
numerically calculated output has been compared with the
experimental output, measuring the Δ error. The phase of
training and validation represents the Design of Experiment
(DOE). The carburizing and nitriding processes that have
been evaluated by the analysis with modeFRONTIER are
summarized in the workflow shown in Fig. 1.
The workflow is divided into data flow (solid lines) and logic
flow (dashed lines). Their common node is the computer
node in which the physical and mathematical functions
representing the carburizing and nitriding processes have
been introduced. In the data flow, all input parameters are
included and optimized in the numerical simulations and in
the corresponding output.
Results and discussion
When we analyze the carburizing output, we can observe
that the microhardness and carbon concentration in steels
are strongly dependent on the carbon potential. At the same
time, they depend differently on the carburizing temperature
as a function of the steel composition and on the
carburizing time (Figure 2a). For the nitriding process, we
can observe that the microhardness in steels is strongly
dependent on the nitriding temperature; however, it
dependents differently on the nitrogen potential as a
function of the steel composition and of the nitriding time
(Figure 2b). The input-output matrix is reported in Fig. 3. It
shows the weight of each single input parameter on the
output results. It is clear how the carbon concentration on
the surface (C[0]) is strongly dependent on the carbon
potential, then on the carburizing time and finally on the
carburizing temperature. The same behavior can be observed
for the carbon concentration at 0.5 mm from the surface
(C[3]), with a stronger dependence on the carburizing time.
The surface hardness (H[0]) is strongly and directly
dependent on the carbon potential while it is inversely
proportional to the carburizing temperature. The hardness at
0.5 mm (H[3]) of the surface is also directly related to the
carburizing time. Surface residual stresses (szz[0]) are
strongly dependent on quenching temperature, while
residual stresses at 0.5 mm (szz[3]) are also dependent on
carburizing time and temperature. As far as the hardening
of the nitriding steel/alloy is concerned, the nitrogen
concentration on the surface (N[0]) is strongly dependent
on the nitrogen potential and time. The nitrogen
concentration at 0.2 mm from the surface (N[5]) is directly
dependent on the heat treating temperature before nitriding
and inversely proportional to nitriding temperature and
nitriding potential. The surface hardness (H[0]) is strongly
dependent on the nitriding temperature, then on nitriding
time, then on nitriding potential and heat treating
temperature. The microhardness at 0.2 mm from the surface
(N[5]) is dependent on the same parameters with almost the
same weight. The residual stresses on the surface (szz[0])
are dependent on the nitrogen potential, while the residual
stresses at 0.2 mm from the surface (szz[5]) are dependent
on the nitriding temperature and heat treating temperature.
Conclusions
The case hardening of different steels has been studied
during thorough experimental investigations. We have
evaluated the different input parameters that have an effect
on the treatments, and we have analyzed the correspondent
mechanical and microstructural results. The data we
obtained have been employed to build a database which we
later analyzed with modeFRONTIER. In this way, it has
become possible to identify the optimal processing windows
for different steels and to evaluate the weight of the
different input processing parameters on the corresponding
mechanical and microstructural properties.
For more information:
Vito Primavera, EnginSoft
[email protected]
Pasquale Cavaliere, Università del Salento
[email protected]
Case History
38 - Newsletter EnginSoft Year 9 n°2
Research of the Maximum Energy Efficiency for a
Three Phase Induction Motor by means of Slots
Geometrical Optimization
Rossi Group from Modena is one
of Europe’s largest industrial
groups for the production and
sales of gear reducers, gear
motors,
electronic
speed
variations and electrical brake motors.
A few years ago, Rossi Group decided to rely on numerical
simulation in order to increase the value of its electrical
motors. Their simulation work focused on achieving
maximum energy efficiency for a three-phase induction
motor by geometrical optimization of rotor and stator.
RMxprt, one of the former Ansoft software tools. which
has been recently incorporated into the ANSYS Workbench
interface, was used for the optimization work at Rossi.
Fig. 1 - Efficiency classes for 50 Hz 4-pole motors (IEC 60034-30:2008)
Background
The topic of energy efficiency is becoming more and more
crucial for the design of electrical engines, in particular
when we consider that two third of the total energy
consumed by industry drives electrical motors. Until
recently, higher efficiency has been mainly achieved by
either increasing the motor’s global dimensions or by
changing the materials.
Owing to some standard manufacturing processes applied
by lamination makers, so far not much time or efforts have
been invested into optimizing the geometry of the slots.
Premium Efficiency
IE3
High Efficiency
IE2
Standard Efficiency
IE1
Recently, Rossi Group has decided to place more emphasis
on the geometrical optimization of the slots with the aim
Case History
Fig. 2 - CAD geometry of the motor (left) and its representation in RMxprt
(right).
Newsletter EnginSoft Year 9 n°2 -
39
Fig 3 - Rotor and stator slots, parameters are depicted.
to enhance the efficiency of its motors without raising
their global dimensions nor changing the materials used.
Efficiency classes
Electric motor technologies have continuously advanced
over the last two decades. Still today, AC induction motors
are the mainstream products sold in most industrial
markets.
For the purpose of energy efficiency, only recently a global
standardization has been established. This new IEC
classification harmonizes regional and national standards
that have been in use so far.
The recently introduced standard IEC 60034-30 defines
energy efficiency classes for single-speed, three-phase
cage-induction motors, 2, 4, 6 poles, 0,75÷375kW,
<1000V, 50 and 60Hz. (Table 1).
Figure 1 reports the value of efficiency versus rated power
for each of the three efficiency classes with regard to the
three-phase induction motor types used in the
optimization analysis described in this article.
RMxprt characteristics
RMxprt offers a machine-specific,
interface that allows users
to easily enter design
parameters and calculate
critical performance data,
such as torque versus
speed, power loss, flux in
the air gap, and efficiency.
The interface also allows to
determine lamination and
winding schemes.
Performance
data
is
calculated
using
a
combination of classical
electrical machine theory
and a magnetic circuit
template-based
approach. The software uses improved Schwarz-Christoffel
transformation to compute field distribution in both
uniform and non-uniform air gaps and applies Gaussian
Quadrature to treat the equivalent surface current of
permanent magnets. Leakage field and corresponding
inductance are derived based on analytical field
computation.
Using a model based on distributed parameters, RMxprt
takes skin effects into account and is able to calculate 3D
end effects. Due to a very efficient relaxation-based
iterative technique, the calculation of the saturation
coefficients in the equivalent circuit is fast and accurate.
This allows saturation effects to be taken into account
efficiently for all supported types of electrical machines.
Its ability to build completely parametric models and the
presence of an embedded optimizer allows RMxprt to
speed up the design and optimization process of rotating
electrical machines.
Motor data
The motor data are inserted into the pre-processing of
RMxprt by means of dedicated sheets. The model is
completely described in the RMxprt interface: geometrical
data, windings arrangement, rating plate and performance
Fig. 4 - Rotor and stator geometry of the original model (left) and the optimized model (right).
Case History
40 - Newsletter EnginSoft Year 9 n°2
Motor data
sheet
RMxprt
model
Optimized
model
Stator Ohmic Loss
232.3[W]
222.5[W]
167.4[W]
Rotor Ohmic Loss
179.4[W]
187.7[W]
137.9[W]
Iron Core Loss
95.0[W]
89.7[W]
117.9[W]
Frictional loss
18.0[W]
18.0[W]
18.4[W]
Stray Loss
59.8 [W]
60[W]
60[W]
Total loss
584.5[W]
577.9[W]
501.6
Input Power
4584.5[W]
4577.9[W]
4501.6[W]
Power Factor
0.82
0.85
0.83
26,62[Nm]
26,65[Nm]
26,34[Nm]
87.2%
87.4%
88.85%
Torque
Efficiency
data along with iron core material characteristics (B-H and
B-P curves).
Figure 2 illustrates the CAD geometry of the motor and its
representation in RMxprt.
The performances evaluated by RMxprt match the test
bench results. In particular, RMxprt estimates an
efficiency equal to 87.4% while at the test bench, the
motor reaches an efficiency equal to 87.2%. The latter
value does not allow the Rossi Group designers to register
the model in the premium efficiency class, and
consequently the motor belongs to the IE2 class.
The RMxprt validated model is used to run an optimization
analysis in order to increase the efficiency of the motor.
The optimization analysis is operable since the model built
in RMxprt is totally parametric and will be performed by
Optimetric, an optimization tool embedded in RMxprt.
Figure 3 shows the rotor and stator geometry. The
geometry parameters managed by the optimization
analysis are reported as well.
Optimization analysis
Optimetric steers the optimization process by choosing
the parameter values with the aim to match the output
objectives, by means of a genetic algorithm.
OUTPUT
1. Efficiency parameter
2. Slot Fill Factor parameter
CONSTRAINT
1. Parallel stator and rotor tooth
OBJECTIVE
1. Maximization of the efficiency
2. Conservation of the SlotFillFactor parameter value.
The overall used parameters are:
• 11 slot geometric dimension parameters (these parameters are dependent in order to achieve parallel tooth
configuration in rotor and stator)
• winding arrangement parameters: Conductors for slot
and wire strand diameter.
To efficiently optimize the system, the defined objectives
are grouped by means of weights in a single objective cost
equation, that is subsequently being minimized.
Optimization results
More than 2000 designs have been evaluated. The cost
function is minimized at Id 1627. Figure 4 shows a
comparison of the original and the optimized geometry. In
table 2, the pertinent results of the original and optimized
model are compared with the data sheet characteristics.
The geometrical optimization results in a better
exploitation of the lamination and in a larger stator/rotor
slot able to accept more copper/aluminum, thus
significantly reducing the losses of the active parts of the
motor.
The optimized model reaches an energy efficiency of
88.85% which allows to register the motor in the IE3
premium efficiency class.
Conclusions
• RMxprt provided the necessary tools to build a model
that completely describes the performance of the present motor, subject of the optimization, with reasonable and minor errors.
• The results, evaluated by the optimization analysis,
achieved the goal to improve the energy efficiency
class of the motor from IE2 to IE3.
• Rossi Group, the manufacturer of the motor, is now
preparing the prototype with the geometric values of
rotor and stator slots as evaluated by RMxprt.
The optimization process is structured in the following
manner:
The prototype is currently under construction.
INPUT
1. Geometrical parameters of stator and rotor slots.
2. Wire diameter
3. Conductors per slot
For more information:
Emiliano D’Alessandro, EnginSoft
[email protected]
Case History
YOU GET
THE IDEA
WE PROVIDE
THE
TECHNOLOGY
www.e4company.com
42 - Newsletter EnginSoft Year 9 n°2
Applicazioni strutturali CAE
nel settore Elettrodomestici
Il settore elettrodomestici si può considerare, particolarmente
negli ultimi anni, un settore fortemente orientato
all’innovazione, a causa della sempre maggiore competitività,
all’importanza degli aspetti economici uniti alla crescente
attenzione per le questioni legate al risparmio energetico e
all’impatto ambientale.
Il cliente finale medio diventa sempre più consapevole ed
esigente, prestando maggiore attenzione ad aspetti quali il
prezzo, la facilità d’uso, i consumi energetici, l’estetica. Anche
la situazione dell'economia attuale ha inciso sulle abitudini di
spesa dei consumatori, inducendoli a ponderare con maggiore
attenzione le spese per beni durevoli, di cui gli elettrodomestici
fanno parte.
Per quanto riguarda gli aspetti energetici e ambientali
assumono sempre maggiore importanza anche i vincoli
normativi, in particolare le nuove Normative Europee tendono a
porre sempre più vincoli riguardanti l’energetica e il design a
basso impatto ambientale (eco-design).
In quest’ambito le analisi CAE, sia in ambito strutturale che
fluidodinamico e/o di ottimizzazione, contribuiscono in modo
Fig. 1 - Previsione della radiazione acustica di una lavabiancheria
CAE World
determinante al raggiungimento degli obiettivi che i produttori e il
Fig. 2 – Simulazione multibody di una
mercato richiedono.
Questo attraverso analisi prova di omologazione al trasporto di una
lavatrice
mirate che tendono a
verificare in modo “virtuale” vari aspetti del processo di
produzione e del funzionamento dell’elettrodomestico, dalla
resistenza strutturale nei confronti dei carichi di progetto alla
contemporanea esigenza di riduzione di peso e di materiale
impiegato, allo studio CFD dell’influenza di vari parametri sulle
prestazioni dei sistemi.
EnginSoft ha contribuito e contribuisce in modo determinante
alla diffusione delle analisi CAE nel settore, essendo key partner
delle maggiori aziende produttrici, italiane ed internazionali
(tra cui Merloni, Indesit, Candy, Electrolux..).
Le attività svolte da EnginSoft spaziano dal settore del freddo
(frigoriferi, congelatori) a quello del lavaggio (lavatrici,
asciugatrici, lavastoviglie) e della cottura (piani cottura, forni
a incasso). Considerando ad esempio la progettazione di
lavatrici e lavasciuga, le applicazioni CAE in ambito strutturale
vanno dall’esecuzione di analisi cinematiche e dinamiche
dell’intera struttura (analisi multi-body) al fine di ottimizzare i
supporti e determinare le forze trasmesse tra i componenti, alle
verifiche di resistenza dei cabinet e dei cesti. Per quanto
riguarda la resistenza della struttura, una fase particolarmente
importante è quella di trasporto dell’oggetto. Durante il
trasporto gli elettrodomestici pesanti (lavatrici ma anche
frigoriferi) possono subire urti di una certa entità. Per questo
motivo i nuovi progetti devono superare dei test standardizzati
prima di ottenere l’approvazione. La simulazione d’urto,
supportata preferibilmente da misure sperimentali, consente di
verificare in fase di progetto il corretto dimensionamento della
struttura e dell’imballaggio, nonché di valutare velocemente
eventuali alternative.
Newsletter EnginSoft Year 9 n°2 -
Fig. 3 - Analisi del processo di stampaggio a iniezione di un componente
per lavatrice
I cesti delle lavatrici possono poi essere analizzati nei confronti
di carichi dinamici e termici, considerando eventuali non
linearità dei materiali (es. polimeri). Per le componenti
polimeriche (ad esempio le vasche per lavatrici) è determinante
ai fini delle prestazioni finali anche la fase di processo
(stampaggio a iniezione). Simulazioni di processo sono
abitualmente svolte su questi componenti al fine di studiare il
corretto riempimento e raffreddamento dello stampo, la
presenza di eventuali punti di debolezza strutturale quali linee
di giunzione dei flussi, deformazioni post-stampaggio (che
possono pregiudicare il corretto montaggio dei componenti),
stress residui. Un altro campo di interesse è la previsione della
vita a fatica dei componenti strutturalmente più importanti (ad
esempio le crociere).
Per quanto riguarda il settore freddo (frigoriferi), oltre ai test di
trasporto a cui accennato in precedenza, l’utilità delle attività
Fig. 4 - Analisi termo-strutturale di un
cabinet di frigorifero
software ANSYS con programmi di ottimizzazione multidisciplinare e multi-obiettivo come modeFRONTIER.
EnginSoft ha esperienza di simulazione anche nel campo del
processo di produzione di celle frigo, ottenute per
termoformatura di polimeri. Le analisi consentono di cogliere le
distribuzioni di spessore sulla cella frigo, i particolare per
quanto riguarda i dettagli critici, ed eventualmente di eseguire
analisi strutturali sulla cella tenendo in conto gli spessori
derivanti dall’analisi di processo.
Anche nel settore del caldo (forni, piani cottura) le analisi
termo-meccaniche sono di fondamentale importanza ai fini di
indagare le temperature, gli stati deformativi e gli stati
tensionali raggiunti dai vari componenti. Temperature eccessive
possono provocare degrado di alcuni materiali. Per i piani
cottura, il funzionamento ad alte temperature unito alla
necessità di ottimizzare gli spessori di materiale impiegato,
possono portare ad eccessive deformazioni di origine termica
del piano stesso. Le analisi CAE (eventualmente non lineari per
considerare l’intervento di grandi deformazioni) consentono di
analizzare velocemente varie configurazioni e di apportare i
necessari correttivi.
Un altro settore di analisi a tutt’oggi meno diffuso ma non per
questo meno importante (anche a causa delle crescenti richieste
normative) è quello delle simulazioni di tipo acustico, come può
essere la previsione della radiazione acustica di una lavatrice in
condizioni di esercizio. Partendo dalla valutazione della risposta
vibrazionale (analisi dinamiche) è possibile arrivare alla
previsione dei livelli acustici e della potenza emessa.
Riassumendo, si comprende come in un mercato sempre più
esigente, in cui diventa basilare raggiungere standard
qualitativi sempre più elevati accanto ad altre esigenze quali la
riduzione del time-to-market, il risparmio energetico e di
materiali, l’apporto della prototipazione virtuale assuma un
importanza fondamentale. In questo ambito EnginSoft già da
Fig. 5 - Ottimizzazione del sistema di
raffreddamento di un frigorifero
di simulazione si manifesta nello studio delle deformazioni in
esercizio, della resistenza strutturale degli accoppiamenti e
nelle analisi termo-strutturali. In particolare l’analisi termica
agli elementi finiti può essere rivolta all’ottimizzazione degli
spessori di isolante, con lo scopo di ottenere il massimo spazio
possibile a disposizione dell’utente, mantenendo al contempo
inalterata la classe energetica cui il frigorifero appartiene. Tali
analisi possono essere svolte sfruttando la combinazione dei
43
Fig. 6 - Analisi dinamica di un forno
decenni si propone come partner d’elezione delle aziende per
l’innovazione del processo progettuale tramite le tecnologie CAE
e si pone all’avanguardia nell’affrontare insieme ai propri clienti
le sfide future.
Per maggiori informazioni:
Sergio Sarti, Maurizio Facchinetti, EnginSoft
[email protected] - [email protected]
CAE World
44 - Newsletter EnginSoft Year 9 n°2
EnginSoft interviews Massimo
Nascimbeni from Sidel
Sidel is one of the world’s leaders in solutions for packaging
liquid foods including water, soft drinks, milk, beer and many
other beverages and is one of Tetra Laval’s three industry
divisions along with Tetra Pak and DeLaval.
The Sidel Group, consisting of more than 5,000 employees in
32 sites over five continents, designs, manufactures,
assembles, supplies and sells complete packaging lines for
liquid foods packaged in three main categories: glass bottles,
plastic and drink cans.
Following a detailed assessment of customer needs, Sidel is
able to design complete solutions covering all the steps of
liquid food packaging, like processing, bottle design and
molding, blowing, filling and rinsing, washing and
pasteurizing up to wrapping, palletizing and logistics
management.
Research and development is fundamental for all these
aspects and is the key to improve the quality of the final
product as well as to increase lines productivity, that
involves challenges like increasing the production rate,
sustainability and safety or reducing costs and maintenance.
Mr. Massimo Nascimbeni is based in Sidel’s Parma site as
Simulation Engineer, focused on developing filling
technologies.
What is the role of virtual prototyping in Sidel?
The first step to improve our products and our production
lines is to gain knowledge on the physics involved.
We have based this process on our experiences in the field
and on our testing facilities for years, but if you want to go
deeper to really understand certain phenomena, involved for
example in the filling process, and be able to control and
improve the process you need simulation tools. Modeling
your process and product gives you the ability to design and
Interview
manufacture in reduced time and at lower costs by speeding
up the “experience process”: this is a requirement for any
company to stay competitive in the near future.
Why have you decided to rely on EnginSoft support?
EnginSoft has been a reliable partner since the start-up
phase, when you have to assess the cost and benefit of
introducing simulation in your company and in your design
and development process. In this
critical phase you have to
compare and make a decision
between different simulation
tools, you have to build a
new methodology and you
are asked to bring concrete
results to give evidence
of the advantages of
simulation.
Newsletter EnginSoft Year 9 n°2 -
EnginSoft ‘s support has been significant thanks to the long
and persistent experiencse EnginSoft has gained in
supporting companies in several industries and not just by
distributing software tools.
Me and the EnginSoft engineers have spent a lot of time
arguing and working together on specific issues that were
relevant to the company and were also good test benches to
build the simulation approach to our products.
Added to this solid base, ANSYS and Flowmaster provide
mature and robust software which have allowed us to meet
our design objectives with tools that are integrated in our
product development process.
What are the objectives of simulation in Sidel
Simulation activities can be performed at different stages
and with different degrees of detail.
We apply simulation from the early phases of the projects,
when you have to approach general aspects, you do not need
to go into the details, but you have to make important
decisions about the right direction to take.
Moreover simulation is fundamental to study in details and
to meet functional requirements of single components that
are the core of filling machines and that affect the reliability
and efficiency of the entire system.
A hierarchical approach has been implemented coupling 3D
(ANSYS) and 1D (Flowmaster) tools.
ANSYS CFD is usually applied to characterize and optimize
the behavior of parts and components of the filling machine.
This detailed information is then transferred to global models
developed in Flowmaster using a 1D approach. This for
example gives fast answers about the behavior of the
machine under different operating conditions or when you
change one component. In this way we are able to study and
improve our products and processes at different levels with
considerable reduction of physical prototypes and tests.
Sidel, with EnginSoft’s support has been working on
implementing an effective, efficient and robust design
process based on simulation. In Sidel we think that
improving the performance and the reliability of our filling
systems is essential to stay one step ahead of competitors.
For more information:
www.sidel.com
45
EnginSoft and ANSYS in the food
& beverage industry
EnginSoft works with the world’s leading players in the
food&beverage industry.
For these companies, food safety, process robustness and
productivity are the most important factors of their
research and development.
Food safety for the consumer is the first priority. Bottles,
packaging and machines have to be washed and in some
cases also sterilized. Moreover, aseptic or non-oxidant
conditions have to be assured for perishable food. All
these issues can be addressed and food safety can be
guaranteed with the support of simulation. Computational
fluid dynamics can give an insight in all these processes,
to study for example the interaction of chemical species
with packaging, machines and food. Thermal management
is another topic that is relevant for many applications, it
involves several disciplines from electromagnetism to
thermo fluid dynamics.
Also, process robustness is a key factor which leading
companies in these sectors that use industrial lines to fill
bottles and packaging with their products, have to
guarantee to their customers. Each component, the whole
line and the process itself, have to be reliable and robust.
Machines have to function in completely different
operating conditions (from polar to equatorial regions),
for diverse products (carbonated drinks, highly viscous
liquids sometimes containing solid parts) supervised by
different staff. Here, simulation can help to foresee the
machines' behavior in normal or critical scenarios, thus
reducing failures and maintenance time.
Last but not least, productivity has to reach the limit.
Hence, every step has to be optimized in terms of
efficiency, which means reducing time and keeping the
process quality at the maximum level. Performances
involve mechanical, thermal, electromagnetic and fluid
dynamics aspects.
The ANSYS technologies and EnginSoft's experiences in
engineering simulation can build synergy, to cover multidisciplinary applications and to give simulation the right
role, which is to support and drive the design process in
advance with respect to physical prototyping thus reducing development time and costs.
For more information:
Massimo Galbiati, EnginSoft
[email protected]
Interview
46 - Newsletter EnginSoft Year 9 n°2
Productivity Benefits of HP
Workstations with NVIDIA®
Maximus™ Technology
In today’s competitive manufacturing environment, getting to
market faster provides a huge financial advantage. Technology
that can cut long wait times that might otherwise force
engineering to put critical decisions on hold can be an
important investment in any successful design project.
A typical example is time spent waiting on results from a
conventional workstation used for engineering computations to
find out whether a design can withstand expected structural
loads and heat – a common bottleneck of simulation-driven
product design.
Industry leaders HP and NVIDIA have joined to create a new
class of workstations that deliver the highest levels of parallel
productivity to let design and engineering applications such as
CAD modeling, photorealistic rendering, and CAE simulation jobs
— all run at the same time. Built for high-end visualization and
computing, the HP Z820 workstation equipped with NVIDIA
Maximus technology is powered by a combination of NVIDIA
graphics processing units (GPUs), the Quadro GPU for visual
interactivity, and the Tesla GPU for massively parallel
computations.
Today’s conventional
design or engineering
workstation contains a
multicore CPU and a
professional GPU (like NVIDIA’s Quadro). In the HP Z820
Maximus system, however, the combined horsepower of dual
processors, a Quadro GPU, and a Tesla GPU are all available to
enable concurrent design and simulation. Previously, engineers
could only perform rendering and simulation tasks one at a time
because they tend to consume all the CPU cores available in a
system and slow down the workstation. But with a Maximusclass workstation, engineers may perform all these tasks in
parallel — without suffering from a system slowdown or
software performance degradation.
NVIDIA Maximus Features and Benefits for
ANSYS Mechanical
During Nov 2010 ANSYS introduced support for CUDA with
release of ANSYS Mechanical 13, which included static and
dynamic analysis using either the iterative PCG/JCG or direct
sparse solvers for SMP parallel. Later with release of ANSYS 14
during Dec 2011, performance enhancements were made along
with an extension for distributed parallel.
Parallel ANSYS Mechanical on a Maximus configured workstation
requires addition of an ANSYS HPC Pack license to unlock a base
configuration of 2 core use to 8 cores, and GPU use is included
in the price. Once implemented, a typical ANSYS simulation can
accelerate by ~4x over the base 2 core use.
Examples are provided of several product design and engineering
companies who have proven the benefits of HP Z800 series
workstations configured with NVIDIA Maximus technology.
NVIDIA Maximus Feature
Benefit to ANSYS Mechanical
Quadro 6000 GPU
GPU for accelerated visualization tasks of pre- and post-processing for
daytime use and accelerated computing for overnight use
Tesla C2075 GPU
GPU for maximum performance of accelerated computing at all times for
static and dynamic analysis, PCG and sparse solvers, SMP and DANSYS
Single Unified Driver
Intelligently allocates proper visualization and compute tasks to the proper
GPU, to ensure sufficient GPU (and CPU) resource utilization
Hardware update
Parametric Solutions: Maximus Provides
Productivity Boosts and Cost Cuts
Parametric Solutions (PSi) provides
engineering product development services,
with its primary specialization in gas turbine
applications for power generation and
aviation design. Much of PSi’s projects
require the use of ANSYS software for
complex analysis and design: thermal,
Newsletter EnginSoft Year 9 n°2 -
47
Liquid Robotics: Revolutionizing Ocean Research
with Maximus
Liquid Robotics’ ocean research robot, the Wave Glider,
is a solar and wave-powered autonomous design of an
unmanned marine vehicle (UVM) that offers a costeffective way to gather ocean data for commercial and
governmental applications. Critical ocean data can
help manage climate change on fish populations,
ANSYS Mechanical 14 performance on V14sp-5 for the Intel Xeon E5-2687W / 3.1 GHz CPU
provide earthquake monitoring and tsunami warning,
and Tesla C2075 GPU
monitoring water quality following an oil spill or
natural disaster, forecast weather, and; assess placement of wind
structural, fluid dynamics, vibration, kinematic synthesis and
or wave-powered energy projects for just a few major
optimization, and the coupling of these disciplines in fully
applications.
integrated simulations.
PSi works with large ANSYS simulations such as the structural
analysis of turbine blade that might contain up to 8 million
degrees of freedom (DOF). Before Maximus, the solving
requirements for a typical simulation were such that design and
analysis could not be conducted on the same system. The ANSYS
The Wave Glider is designed to operate continuously, without
intervention, for months and a year at a time, and ANSYS
software is used to simulate the complex mechanical designs
such as a structural assembly or the reduction of hydrodynamic
drag.
ANSYS Mechanical analyses of a shrouded blade and disk configuration (Images courtesy of
Parametric Solutions)
The Liquid Robotics Wave Glider ocean robot (Images courtesy of Liquid Robotics)
Astrobotic’s lunar lander and ANSYS Mechanical simulation (Images courtesy of Astrobotic
Technology)
analyses alone would require use of the entire system for 8 to 12
hours or more, and days if engineers wanted to conduct
interactive design on the system at the same time.
NVIDIA’s Maximus proved to be an extremely powerful
technology for their operations, with a remarkable 2x boost in
ANSYS Mechanical performance and increases in productivity for
all their ANSYS interactive and compute-intensive processes.
These gains were achieved while continuing to do other work,
such as CAD modeling, simultaneously on a single Maximuspowered workstation.
In the past, doing simulation or rendering required the
complete computational power of Liquid Robotics
systems, but HP workstations with NVIDIA Maximus
has changed the way its engineers operate. A limited
number of engineers are now able to do multiple tasks
at once, which has transformed the engineering
workflow. The design began with years of research
requiring millions of dollars, but now in just a few
weeks, design changes can be made that
incrementally increase performance and reduce costs.
Astrobotic Technology: Maximus Helps Ignite New
Era of Moon Exploration
Astrobotic develops lunar landers, rovers, and other
space robotic technology for lunar surface exploration
and imaging. The potential of the moon to contain
enough oxygen, rocket fuel, and other materials, to
build and fuel what’s needed to go to Mars and other
planets is motivation for Astrobotic technology
development. Robots could be used to set up a moon
base, extract materials, and even assemble equipment.
Remote robots working autonomously obviously
require extreme precision in design and direction, and
ANSYS Mechanical was deployed on HP Maximus
configured workstations.
Before using HP and NVIDIA Maximus technology, each step of
Astrobotic’s engineering process of 3D design, or analysis, or
rendering completely consumed their systems. For ANSYS
analysis that also meant restricting models to about ~500 K
degrees of freedom (DOF) for reasonable turn-around times. With
Maximus the ANSYS models were refined with up to 3,000 K DOF
to capture simulations more completely and in less time – all
without interruption of other applications on the workstation.
Contact HP or NVIDIA today to learn more about the benefits for
your ANSYS workflow and Maximus.
Hardware update
48 - Newsletter EnginSoft Year 9 n°2
Uno sguardo alle principali novità riguardanti
l’integrazione dei prodotti ANSOFT in bassa
frequenza in piattaforma ANSYS Workbench 14
Si realizza quindi un’analisi
multifisica di tipo sequenziale fra
Maxwell e gli ambienti di
simulazione di ANSYS.
L’import/export dei dati necessari
all’analisi multifisica
avviene
interamente in interfaccia ANSYS
Workbench e non necessità di
script dedicati.
Fig. 1 - Field (left) and circuit (right) coupling in ANSYS 14.
Con l’uscita della release 14 di ANSYS vengono introdotte
importanti novità anche per quanto riguarda la suite dei
prodotti ANSOFT in bassa frequenza: ci riferiamo in
particolare a Maxwell 15 ed a Simplorer 10.
Il seguente documento ha lo scopo di sintetizzare alcune di
queste novità per quanto riguarda l’integrazione di queste
tecnologie in interfaccia WB2 di ANSYS 14.
Field coupling e circuit coupling
Nella figura 1 viene riportato lo stato dell’arte per quanto
riguarda l’accoppiamento di campo (field coupling) e di
sistema (circuit coupling) attualmente implementabile in
ANSYS 14.
In relazione all’accoppiamento di campo (field coupling),
La figura 1 evidenzia come Maxwell possa scambiare dati
con le tradizionali tecnologie di casa ANSYS sia in ambito
meccanico strutturale che CFD.
Software update
La tipologia di dato scambiato
dipende dalla fisica che si sta analizzando.
Maxwell calcola, in relazione alla soluzione di campo
richiesta (magnetica od elettrica) ed al solutore
considerato (statico, armonico o transitorio), sia forze
elettromagnetiche che potenze dissipate; queste soluzioni
di campo possono essere trasferite rispettivamente alle
analisi strutturali e termiche di ANSYS, sia per gli ambienti
di analisi statica che armonica e dinamica. Allo stesso
modo le perdite di potenza possono essere esportate verso
un modello CFD di Fluent.
Per quanto riguarda l’analisi di sistema il software di
riferimento di casa ANSYS è Simplorer.
Simplorer può integrare in uno schema circuitale le
principali tecnologie di casa ANSYS, tipicamente attraverso
le seguenti tecniche: Cosimulazione e model order
reduction.
La Cosimulazione o “Co-simulation” (co-operative
simulation) è una metodologia di simulazione che consente
Newsletter EnginSoft Year 9 n°2 -
49
Nella figura 2 si riporta lo schema che in ANSYS
Workbench implementa questa analisi.
Fig. 2 - Accoppiamento bidirezionale termico fra Maxwell e Fluent.
Trascurando la parte strutturale, trattata in
seguito, la figura 2 evidenzia come si imposta
un’analisi multifisica di tipo sequenziale fra gli
ambienti di simulazione Maxwell e Fluent.
I modelli vengono preparati separatamente nei
due ambienti di simulazione, ragion per cui il
modello nodi-elementi utilizzato in Maxwell e
Fluent risulta diverso.
Il trasferimento dei carichi viene impostato
manualmente in interfaccia Workbench e si
realizza attraverso interpolazione.
a componenti singoli di essere simulati in maniera
simultanea da software diversi. Questa tecnica permette
quindi ai due software di scambiarsi le rispettive soluzioni
in maniera collaborativa e sincronizzata.
Accoppiamento strutturale Maxwell-ANSYS
Maxwell calcola le forze elettromagnetiche. Queste forze
tipicamente afferiscono a due gruppi differenti:
La tecnica della Model Order Reduction (MOR) è una
disciplina della teoria dei sistemi e dei controlli che studia
le proprietà dei sistemi dinamici in modo tale da ridurne la
complessità, preservandone il comportamento agli ingressi
ed alle uscite (input ed output). Utilizzando proprio
questa tecnica è possibile trasferire nell’ambiente di
simulazione di Simplorer modelli a parametri concentrati
estratti da modelli agli elementi finiti realizzati tra gli altri
in ANSYS Mechanical, ANSYS Fluent e ANSYS ICEPack.
1. La densità di forza superficiale rappresenta la forza di
riluttanza magnetica e si genera all’interfaccia fra
materiali con permeabilità magnetica diversa. Tutte le
volte che si analizzano modelli aventi materiali con
permeabilità magnetica molto maggiore di 1 si deve
considerare questo tipo di forza.
2. La densità di forza volumetrica rappresenta la forza di
Lorentz e va considerata nel caso di conduttori percorsi da corrente immersi in un campo magnetico.
Nel seguito del documento vengono approfondite le
principali novità che riguardano l’accoppiamento di campo
e di sistema in ANSYS 14, in particolare:
Maxwell esporta entrambe queste categorie di forze verso
gli ambienti di simulazione strutturali statici, armonici e
dinamici e legge la mesh deformata output della
simulazione strutturale.
L’analisi descritta si realizza in ANSYS Workbench con lo
schema riportato in figura 3.
1) Accoppiamento bidirezionale termico Maxwell-Fluent.
2) Accoppiamento bidirezionale strutturale Maxwell-Ansys
Mechanical.
3) Cosimulazione Simplorer-Fluent.
Accoppiamento bidirezionale Maxwell-Fluent
L’accoppiamento termico già implementabile fra
Maxwell2D/3D e l’ambiente di simulazione di ANSYS
Mechanical diviene disponibile anche per la tecnologia
Fluent.
Il calcolo elettromagnetico di Maxwell fornisce in output le
potenze dissipate per unità di volume. Fluent legge in
input le potenze dissipate e calcola le temperature output
dell’analisi CFD.
Queste temperature possono essere quindi passate
nuovamente a Maxwell per una nuova soluzione
elettromagnetica. Affinchè la nuova soluzione di Maxwell
abbia senso, è necessario che siano state definite
opportune proprietà dei materiali in funzione della
temperatura.
Il processo si conclude allorché la differenza delle
temperature calcolate fra due iterazioni successive è
trascurabile.
Fig. 3 - Schema dell’accoppiamento bidirezionale strutturale
Maxwell-ANSYS Mechanical.
Software update
50 - Newsletter EnginSoft Year 9 n°2
Integrazione geometrica completa
In interfaccia Workbench ANSYS 14 si
realizza un’integrazione completa fra i
modelli geometrici di ANSYS ed i software
già ANSOFT in bassa ed alta frequenza
(Maxwell e HFSS).
Per modelli geometrici di Ansys si
intendono: le geometrie create dai
modellatori geometrici proprietari ANSYS (
DesignModeler e SpaceClaim) e le geometrie
eventualmente caricate in interfaccia
usufruendo dei numerosi plug-in a
disposizione per la quasi totalità dei CAD
commerciali, così come per i principali
formati geometrici neutri.
Fig. 4 - Simplorer: Schema di cosimulazione Simplorer-Fluent
Cosimulazione Simplorer-Fluent
Una delle novità più significative in ANSYS 14, riguardante
la simulazione di sistema, è rappresentata dalla tecnica
della Cosimulazione implementabile fra Simplorer10 e
Fluent.
Uno degli ambiti per i quali questa integrazione risulta più
significativa è l’analisi termoelettrica dei package delle
batterie.
Tipicamente è infatti necessario impostare un’analisi CFD
per il calcolo delle temperature di interesse, che in input
possa leggere sia la portata in massa del fluido
refrigerante, eventualmente controllata in temperatura,
che le potenze elettriche dissipate dalle batterie.
In particolare, una qualsiasi geometria
parametrica importata o creata in
interfaccia Workbench, può essere esportata
in Maxwell15 o HFSS14, sempre in formato
parametrico, così come una geometria
eventualmente creata in Maxwell15 o HFSS14 può essere
utilizzata in tutti gli ambienti di simulazione a
disposizione in ANSYS senza perdere le informazioni sui
parametri.
In figura 5, a mo’ di esempio, si illustra come 3 formati
geometrici parametrici, provenienti da fonti diverse (proE,
Maxwell e DesignModeler), possano essere trasferiti
tramite interfaccia al simulatore per le analisi in alta
frequenza HFSS.
Per ulteriori informazioni:
Emiliano D’Alessandro, EnginSoft
[email protected]
La tecnica della Cosimulazione consente infatti di
impostare per il caso specifico un’analisi dinamica con le
seguenti caratteristiche:
• Simplorer e Fluent eseguono insieme il run dei rispettivi modelli.
• Simplorer agisce come master, Fluent come slave.
• Ad ogni time step dell’analisi dinamica Simplorer passa i valori relativi alla potenza termica generati dalle
batterie ed alla portata di fluido refrigerante alla simulazione di Fluent. Fluent passa a Simplorer i valori di
temperatura di interesse calcolati.
• In Simplorer non c’è limite alla complessità per quanto riguarda la modellazione sia dei modelli elettrochimici delle batterie che del circuito di controllo.
In figura 4 è schematizzato un esempio di analisi come
descritta. In questo caso Il modello di Fluent si compone
di un'unica cella di batteria. Simplorer si interfaccia con il
modello di Fluent attraverso la definizione dei parametri
di input e di output delle grandezze di interesse.
Software update
Figura 5: Esempio di integrazione geometrica ANSOFT-ANSYS
in interfaccia Workbench.
HP recommends Windows® 7.
simulati
supercomputing power
HP Z400 and Z800 Workstations with NVIDIA® Tesla™, Intel® Xeon®
processors and HP performance displays.
HP Z Workstations utilize the latest computing technology to make innovation easier, through their
chipset, memory, multiprocessor and multi-core capabilities. See even simulation models with
daily increasing complexity come to life with NVIDIA® Tesla™, which enables pre/post-processing
to run simultaneously with analysis at up to 515 GigaFlops (DP) or 1030 Gflops (SP).
Visit hp.eu/workstations to learn more.
©2011 Hewlett-Packard Development Company, L.P. All rights reserved. The information contained herein is subject to change without notice. NVIDIA Corporation, NVIDIA, the NVIDIA logo, are trademarks
and/or registered trademarks of NVIDIA Corporation in the United States and other countries.
52 - Newsletter EnginSoft Year 9 n°2
Updates on the EASIT2 project: educational base
and competence framework for the analysis and
simulation industry now online
Project objectives and background
The EASIT2 project is a Leonardo da Vinci European Union cofunded project, part of the European Vocational Training
Action Programme. Leonardo da Vinci projects are aimed at
designing, testing, evaluating and disseminating innovative
vocational training and lifelong learning practices, and at
promoting innovation in training as well as methodologies,
contents, products.
EASIT2 is coordinated by the department of Mechanical
Engineering of the University of Strathclyde (Glasgow, UK),
and is partnered by EnginSoft, NAFEMS, EoN, EADS, Renault,
GEOFEM, Nokia, Nevesbu, TetraPak, AMEC, Selex Galileo.
Aim of the EASIT2 project is the development of 3 distinct,
but functionally related, competence-based tools:
• the educational base, a database of “standard” analysis
and simulation competencies;
• the competence framework, a software that enables
companies and motivates individuals to verify, track,
develop and attest competencies in the field;
• the new NAFEMS competence-based registered analyst
scheme, that will foster a transparent and independent
certification of individual analysis and simulation
competencies.
A guide through the analysis and simulation knowledge:
the EASIT2 Educational Base
At the foundation of the project there is the EASIT2
educational base: this is intended as guidance to those who
are engaged in continuing professional development, both at
a personal level and at an organisational level.
The educational base is a database of competence
statements covering most of the whole spectrum of analysis
and simulation competencies: a competence statement is a
sentence or paragraph that captures a specific competence,
Research & Technology Transfer
for example “List the various steps of the analysis and
simulation process”, or “Explain why strains and stresses are
generally less accurate than displacements for any given
mesh of elements, using the Displacement FEM”.
Statements express what an analyst should be able to do: the
emphasis on doing is what distinguishes the EASIT2
competence based approach from existing approaches based
The structure of the EASIT2 project, showing the interdependence between
the 3 competence-based tools.
Newsletter EnginSoft Year 9 n°2 -
53
Capturing analysis and simulation competencies:
the EASIT2 Competence Framework
Around the educational base, the EASIT2 competence
framework has been built. The competence framework is a
secure web based system, designed to be accessed over the
The EASIT2 project partners
Internet or a company Intranet.
The competence framework enables individuals or company
on more intangible ideas related to educational aims, such as
staff to actually record their competencies into a relational
for example a list of training objectives or course syllabus.
database. In fact, the competence framework integrates with
The EASIT2 educational base contains over 1400 competence
the educational base allowing its users to verify, attest and
statements, subdivided into 23 competence areas, starting
track their competence.
from basic topics “Introduction to Finite Element Analysis”,
For the individual user, the EASIT2 competence framework
and covering for more specialized topics, such as “Materials
helps tracking learning progresses and guiding further
Modelling, Characterization and Selection”, “Fatigue and
learning: the user can authenticate into the system, navigate
Fracture”, “Nonlinear Geometric Effects”, “Computational
the educational base and the related educational resources,
Fluid Dynamics”, etc.
assess his/her own competencies in the various areas of
In general, each competence area contains a varying number
competence available, and generate his/her own individual
of 50 to 100 statements. Inside a competence area,
competence report. For organisations, the competence
competence statements are presented in an order that
framework is designed to provide an open and highly
generally reflects increasing competence, that is, basic
customisable system. In fact, the competence framework
competencies are presented at the top of the list, while
allows the definition of groups of users, enabling companies
higher level competencies are presented at the bottom.
to define and track the competence development for both
Statements thus ideally guide the learner from mere
individual employees and teams. Furthermore, the
knowledge to more actionable abilities.
educational base itself is customisable, and can
Finite Elements Analysis
Thermo-Mechanical Behaviour
be modified or extended for example by adding
Mechanics, Elasticity and Strength of Materials Computational Fluid Dynamics
additional competence areas or statements of
specific interest. The competence framework is
Materials for Analysis and Simulation
Electromagnetics
designed to be capable of interfacing to existing
Flaw Assessment and Fracture Mechanics
Fundamentals of Flow, Heat and Mass Transfer
staff development or human resource
Fatigue
Multi-body Dynamics
management systems: in fact it is envisioned that
the ability it provides to define various levels and
Nonlinear Geometric Effects and Contact
Multi-physics
subsets of competencies will be useful for
Beams, Membranes, Plates and Shells
Multi-Scale Analysis
planning technical careers.
Dynamics and Vibration
Noise and Acoustics
Buckling and Instability
Optimization
Composite Materials and Structures
Probabilistic Analysis
Creep and Time-Dependency
Simulation Management
Plasticity
Areas of competence currently covered by the EASIT2 educational base.
Furthermore, each competence statement includes
information regarding the level of the competence relative
for example to the European Qualification Framework (EQF)
level.
The educational base can be used for example for educational
purposes: every statement is linked to appropriate
educational resources, such as books, papers, codes of
practice, etc., that will help an engineer to gain the
appropriate competence.
Each statement and educational resource included in the
educational base underwent a peer review process during the
EASIT2 project: it is anticipated that the educational base
itself will be further enriched and improved after the end of
the project. A public version of educational base will be
released to the public over the next months.
Enabling transparent analysis and simulation
qualification attestation: the new NAFEMS
competence-based registered analyst scheme
The project has developed and is currently
testing a new competence-based registered
analyst scheme, derived from the points-based
scheme currently offered by NAFEMS. The new
scheme will retain much of the sound set-up of the present
NAFEMS scheme, such as the requirements of workplace
experience, product and industry sector knowledge, etcetera,
but will also make use of the EASIT2 educational base and
competence framework. In the new competence-based
registered analyst scheme set-up currently being tested,
analysts will be required to access the competence
framework, attest their own competencies, and produce their
own individual competence report. The report will then be
attached to the new registered analyst application. The
information provided in the individual competence report,
will then be used for the assessment procedure.
Conclusions
The topic of the certification of competencies is a key issue
of the European debate over the relationship between the
Research & Technology Transfer
54 - Newsletter EnginSoft Year 9 n°2
rapid change of the technical knowledge and the
development of the job market. Over the last years, the
European Commission introduced a number of tools, such as
the Europass and the certificate supplement, aimed at
improving the transparency and transfer of competencies and
qualifications as part of the Bologna process.
The EASIT2 project will contribute a new set of tools,
specifically designed for the analysis and simulation industry,
that fit ideally into this paradigm shift from a curriculum
based qualification attestation, to a more sound and
transparent competence based certification.
EnginSoft values the EASIT2 project as strategic, and
therefore joined the project as a core partner to contribute
more closely to the project steering. EnginSoft has an active
role in the development of the innovative competence based
tools, and specifically it is in charge of the development of
the competence framework and is leading the competence
based registered analyst scheme workpackage.
For more information on the project please visit its website:
http://www.easit2.eu
The EASIT2 competence framework home page
To contact the author:
Giovanni Borzi - EnginSoft
[email protected]
EnginSoft in £3m EU Partnership to
Optimise Remote Laser Welding
The EU has awarded a £3.35m research grant to a consortium
involving EnginSoft to develop a technique for optimising
the use of Remote Laser Welding in assembly processes. The
consortium is led by Warwick Manufacturing Group (WMG) at
the University of Warwick, and also involves Jaguar Land
Rover, Comau, Stadco, Precitec and several important
academic institutions including Politecnico Milano, the
University of Molise, Ulsan NIST, the University of Patras,
Lausanne Polytechnic and SZTAKI Budapest. This is an
exciting project for the participants, of course - but why is it
important for the EU?
Vehicle assembly is a complex process involving the joining
of many subsystems by a variety of methods. For many years,
resistance spotwelding has been a key technology, with a
Research & Technology Transfer
welding head simultaneously bringing together metal
components (typically steel) whilst passing an electrical
current to locally re-melt the material and form a mechanical
connection. This is far from the only method of body
assembly, however. Recent years have seen an increasing use
of techniques such as gluing and riveting, for example. In
each case, the ability to determine the best joining method
for an assembly process is critical to assembly efficiency, and
therefore vital to the competitive position of many
companies within Europe. Designing an efficient assembly
process is frequently far from trivial - each method will have
its own limitations, power requirements, cycle times and so
forth, so determining the best configuration for an assembly
sequence will be a complex procedure.
Newsletter EnginSoft Year 9 n°2 -
55
That is where the "Remote Laser Welding System Navigator
for Eco & Resilient Automotive Factories" enters the picture.
Remote Laser Welding is a promising and relatively-new
joining method which involves an intense laser beam being
focused onto the material to be joined from one side only.
Local re-melting and fusing to the underlying material then
takes place. This can be very rapid, and since the beam can
be manoeuvred from place to place with small angular
adjustments of the welding head there is a great opportunity
for rapid cycle times at the welding stations. However, there
are also challenges to be met - in particular, the weld
locations must be "visible" to the welding head, and the gap
tolerances on the parts being fixed must be well-controlled.
The aim of the RLW project is to provide a software tool that
will enable the process designer to develop an optimal
configuration for the use of such processes. At the highest
level, it will take a series of conventional assembly
workstations and consider all the different ways in which
they could be combined to make use of RLW techniques.
This system level will propose an RLW-efficient assembly
process, leaving parts of the assembly sequence which are
unsuitable for RLW unchanged, but introducing RLW where it
is most effective. This may involve the introduction of
additional workstations or processes that are necessary for
the optimum use of RLW, as well as the combining of
workstations where RLW is able to perform in place of
multiple original locations.
At a deeper level, the project will assist in developing
individual workstations. Here, the whole process of gap-
Fig. 2 - Remote laser welding uses a robot to direct a laser beam to a
location on one surface of a part, fusing it to an underlying part by the
local melting of the part materials
tolerances at the assembly interfaces must be managed by
part-screening and appropriate fixing, as well as defining the
process parameters for the laser and the geometrical
manipulation of the parts and the welding head. This
detailed model will also define the control requirements and
calculate the process timings and energy requirements
necessary to refine the higher-level definition of the
assembly system.
Finally, the software should assist the
designer in developing components that
are suitable for efficient laser welding by
providing appropriate feedback on the
part properties.
So the project will be an invaluable tool
in the development of state-of-the-art
assembly
processes
using
RLW
technology, and thereby play an
important role in maintaining Europe's
competitive advantage in systems
assembly.
Fig. 1 - Effective remote laser welding relies on carefully controlling the gap between the parts and the
control of the robot-mounted laser
For more information:
David Moseley, EnginSoft UK
[email protected]
Research & Technology Transfer
56 - Newsletter EnginSoft Year 9 n°2
DIRECTION: Demonstration of Very Low Energy
New Buildings
How to enhance the overall energy efficiency of a building in
order to achieve a consumption level of primary energy lower
than 60 kWh/m2 per year?
This four year EU-funded (Framework Programme 7 - FP7)
project, launched end of January 2012, aims at the creation of
a framework of demonstration and dissemination of very
innovative and cost-effective energy efficiency technologies for
the achievement of very low energy new buildings.
DIRECTION proposes five main progresses beyond the state of
the art in the following areas:
• Energy Efficiency Measures: Energy consumption reductions
of more than 50%.
• Low-Energy Buildings: CO2 emission reductions of more than
60%.
• Modeling and simulation.
• Building monitoring.
• Standards & regulations implemented by European and
national policy-makers.
An important impact for the building sector is expected in the
following four main areas:
• Energetic: drastic energy consumption reduction.
• Environmental: significant CO2 emissions reduction.
• Building sector business: encouraging a mass market for very
low energy new buildings, facilitating understanding of all
stakeholders.
• European policies: contributing to boost the implementation
of standards and regulations.
Demonstration activity
Based on the analysis of suitable energy efficiency technologies
and their technical and economic viability, the demonstration
Building and construction engineers, architects, energy
activity will be deployed at the sites in three new buildings, in
researchers, IT specialists and public authorities will work
which a set of very innovative measures such as constructive
together in order to show how the ambitious goal can be
elements for energy optimization, high efficient energy
reached.
equipment and advanced energy management will be applied.
Local, national and European
Spain – Valladolid
Italy – Bolzano
Germany – Munich
CARTIF III new building
New Technology Park of Bolzano
Nu-Office
stakeholders including public
authorities, users and citizens at
large will be kept up-to-date about
the progress and the outcomes of
the demonstration.
This research center will host offices
and test facilities (industrial
activities).
This building will host different
stakeholders (enterprises, research
institutes and public entities)
sharing the common goal to develop
and deploy energy efficiency
solutions.
Research & Technology Transfer
This office building, located in
Domagkstrasse, will be built to
“Sustainable Building” standards as
a non-private housing.
EnginSoft’s Contribution
EnginSoft will be in charge of
software development and will be
mainly committed in Building
Modeling & Process simulation to
improve the design of the energy
Newsletter EnginSoft Year 9 n°2 -
efficiency solutions. The objective is to optimize the
performance of the building as a whole and not only its single
components, accompanying the building with modeling and
simulation throughout its life cycle from design through
construction to operation. Software interfaces will be developed
to integrate the simulation tools necessary to perform the
envisaged numerical analyses.
• Building Information Modeling (BIM) will gather all the
information on the building taking into account the complex
interactions and interdependencies and allowing to transfer
information without loss from one actor in the design and
construction process to the other.
• Integrated design will be strongly supported by the
developed building models and dynamic simulation at
different stages of the process.
• Continuous commissioning will be integrated with modeling
and simulation and guarantee the smooth transition from
design to operation phase.
• Simulation aided automation and model based control will
allow to tap the full potential of energetically optimized
building operation.
• The building model will support the evaluation of the single
energy saving measures.
• Optimization analyses, based on dynamic simulation, will be
split as follows: building envelope, equipment and HVAC
concept, building energy management system (BEMS).
• A Black Box Model based on experimental data, by means of
suitable meta-modeling algorithms for the building and for
single parts/components, will be developed.
More information is available at:
www.direction-fp7.eu
Other Partners of the Consortium
SPAIN
• Centro tecnológico CARTIF
Research centre Project coordinator
• 1A Ingenieros
Engineering, architecture and technical consultancy
• Dragados
System integration Building contractor
GERMANY
• NUOffice
Domagk Gewerbepark - Project developer specialized in commercial real
estate
• Fraunhofer
Institute for Building Physics
• Facit
General contractor
ITALY
• EURAC Research
EURopean Academy - Research centre
• CL&aa
Claudio Lucchin & architetti Associated architecture studio
• Provincia Autonoma di Bolzano Alto Adige
Government partner
BELGIUM
• Youris.com
Media agency
57
Dissemination dei risultati del Progetto BENIMPACT
EnginSoft e l’Istituto per Geometri
Pozzo di Trento vincono il
Concorso "Tu Sei" 2012
Il progetto “Tu sei”, giunto quest’anno alla quarta edizione, è
nato da un Protocollo promosso
da Confindustria Trento e sottoscritto dalla Provincia Autonoma
di Trento per ridurre la distanza
tre mondo reale ed attività scolastiche attraverso percorsi
creativi ed interessanti, favorendo un positivo avvicinamento degli studenti al mondo delle
imprese per semplificare il futuro sbocco occupazionale dei ragazzi. Lo scopo è quello di
stimolare i giovani ad approcciarsi al mondo del lavoro
con responsabilità e intraprendenza.
Quest’anno sono stati presentati 21 progetti e hanno partecipato:
• 3 istituti comprensivi (scuole secondarie di primo grado);
• 12 istituti di istruzione secondaria di secondo grado;
• 600 STUDENTI circa;
• 27 AZIENDE.
Partnerships vincitrici del Premio “Tu Sei” 2012
• Istituto Comprensivo Tione (primo ciclo) di Tione con
Girardini Srl di Tione
• Istituto Tecnico per Geometri “A. Pozzo” di Trento con
EnginSoft SpA di Trento
• CFP Istituto Pavoniano Artigianelli di Trento con SILVELOX SpA di Castelnuovo Valsugana
La cerimonia conclusiva si è tenuta il 5 maggio 2012 presso l’auditorium Melotti del MART di Rovereto.
In quest’occasione i ragazzi hanno presentato i loro progetti e sono stati premiati i tre vincitori: uno per le scuole secondarie di primo grado e due, a pari merito, per gli
istituti di istruzione secondaria di secondo grado.
Il progetto di EnginSoft con l’Istituto per Geometri
Pozzo di Trento
Una decina di studenti del quinto anno dell’Istituto per
Geometri “A. Pozzo” di Trento hanno collaborato alla realizzazione di un database sui materiali utilizzati per la realizzazione di involucri edilizi e partecipato a un corso di
formazione sugli strumenti informatici sviluppati in
BENIMPACT. Hanno poi utilizzato alcuni di questi per
modellare il comportamento energetico in regime dinamico di un edificio da esporre in sede di esame di stato.
Per ulteriori informazioni:
Angelo Messina, EnginSoft
[email protected]
Research & Technology Transfer
58 - Newsletter EnginSoft Year 9 n°2
EnginSoft and Flow Design Bureau (FDB)
launch collaboration.
Best-in-class computer aided engineering (CAE) software and
consultancy services for the Offshore and Oil&Gas industries in Norway
Despite the fact that the renewable energy sectors are
gradually growing and are becoming more important for the
future of our energy supply and our earth, Oil&Gas will
continue to be the dominant energy sources for several
decades to come. This issue highlights Norway as a
responsible supplier of petroleum.
Today, Norway is one of the world’s leading producers of
fossil fuels, as well as one of the largest exporters of
natural gas and oil. Over the past 50
years, the country has developed an
advanced petroleum industry, that
today also encompasses some of the
front runners in subsea technologies,
which are considered vital for the future
supply of fossil fuels.
Up to 20 years ago, Norwegian oil
companies and most of their suppliers and service providers
concentrated their efforts on the Norwegian continental
shelf. In the recent past, however, we can notice
Norwegian-headquartered Oil&Gas related industries all
over the world. In fact, small and medium-sized Norwegian
companies seek to compete for market shares
internationally, and they are succeeding.
EnginSoft and Flow Design Bureau (FDB) are well aware of
this scenario, so much that they have decided to sign a
collaboration agreement for computer aided engineering
(CAE) software sales and consultancy for the Offshore and
Oil&Gas industries in Norway.
FDB is a technology development and consultancy company
that specializes in fluids engineering and heat transfer for
the energy sectors and businesses. FDB has a track record
of delivering high quality consultancy services to
Norwegian businesses within the Oil&Gas fields using
computational-fluid-dynamics.
EnginSoft is an engineering software and
consultancy organization which has the
know-how and resources
to provide
services based on a large variety of CAE
software, for fluid-dynamics, mechanical,
structural,
electromagnetic,
process
simulations in the aerospace, automotive,
chemical and Oil&Gas industries.
Based also on its own competencies developed over the
past 25+ years and its ambitions for international growth
(EnginSoft supports business operations in France, the UK,
Germany, Spain, Sweden, as well as in the USA, in the
Houston area, and with Standford University), EnginSoft
has identified the Oil&Gas sector as a core business for its
activities into which it can deliver services for most of the
required engineering applications.
In this light, EnginSoft and FDB agreed that by combining
efforts both companies can benefit and grow
business.
EnginSoft now has a partner in Norway through
which the Norwegian Oil&Gas industry can be
approached while FDB can count on EnginSoft,
to offer services and to participate in multidiscipline engineering projects. Eventually, the
goal of the collaboration is to formalize the
partnership, and for FDB to become a member,
a node in EnginSoft’s Network of engineering
companies.
For more information:
Livio Furlan, EnginSoft
[email protected]
EnginSoft Network
Newsletter EnginSoft Year 9 n°2 -
59
Dinamica esplicita:
nuovo competence
center EnginSoft a Torino
Da Gennaio 2012 è operativo, con sede Torino, il nuovo
Competence Center di EnginSoft specializzato soprattutto nel
calcolo strutturale, nella fluidodinamica e nella simulazione dinamica esplicita.
Di grande attualità, i codici di simulazione detti “espliciti”, diffusi e consolidati in primis per analisi di sicurezza passiva, si
prestano a molteplici applicazioni industriali. Sviluppati, in origine, per ridurre costose campagne sperimentali di verifica prodotto in ambito Automotive ed Aerospace, (immaginate le risorse necessarie ad un Crash Test!) sono oggigiorno applicate in
tutti i settori industriali, elettronica compresa. Il telefono cellulare che possediamo e che è diventato d’uso comune, con molta probabilità, è stato progettato, sviluppato e raffinato anche
attraverso una campagna di Drop Test Virtuale prima ancora di
test di caduta con prototipi “Fisici”.
In generale, in tutti i casi di forti non linearità (spostamenti,
deformazioni e contatti) e di dinamica veloce (collisioni, impatti, esplosioni) l’approccio Esplicito è la soluzione tecnicamente
ed economicamente ideale ai fini di una fedele rappresentazione del fenomeno fisico o multifisico.
Il team di ingegneri del Competence Center di Torino, vanta
competenza ed ampie esperienze nello sviluppo di applicazioni
di Passive Safety avanzate quali:
• Sistemi di ritenuta innovativi (sedile, cinture, pretensionatori, limitatori di carico).
• Airbag con logiche di apertura, gonfiaggio e attivazione di
ultima generazione.
• Accoppiamento prodotto-processo (es. stampaggio, saldatura e crash).
• Simulazione di materiali innovativi.
• Ottimizzazione multiobiettivo.
Come logico immaginare, per tecnici che si sono formati in
un’area industriale che, per tradizione, è fortemente
“Automotive Oriented”, metodi, strumenti ed esperienze sono
state originariamente sviluppate e raffinate proprio per far fronte alle tematiche dell’Auto e dei veicoli in generale. Tuttavia le
tematiche del mondo Auto, in apparenza differenti da quanto richiesto per altri settori industriali, presentano molte analogie
ed affinità dal punto di vista della simulazione numerica. Anzi!
Molto spesso le necessità di elevare qualità, sicurezza ed affidabilità del prodotto che si contrappongono all’esasperata rincorsa alla riduzione di tempi e costi stimolano la creatività degli ingegneri EnginSoft verso lo sviluppo di metodologie innovative
che determinano l’innovazione del prodotto e/o del processo.
Un esempio concreto è l’articolo pubblicato nello scorso numero della Newsletter e riguardante la riduzione dell’imballo di un
frigorifero INDESIT. Temi specifici affrontati da EnginSoft in cui
i solutori espliciti sono impiegati sono: elevate deformazioni di
materiali iperelastici, fenomeni di danno e rottura, impatti e
collisioni di ogni genere e natura (es. misuse nel settore machinery), multibody, sistemi elastocinematici, progettazione di serbatoi tramite l’approccio FSI (SPH), analisi termo-strutturali di
guarnizioni, etc etc. Non di rado le analisi esplicite ed implicite vengono accoppiate, qualora necessario, al fine di elevate
qualitativamente la rappresentatività del modello.
Al Competence Center di Torino è demandata la promozione,
vendita e supporto tecnico specialistico agli utenti, compresa la
formazione, per ciò che concerne il solutore LS-DYNA che oggigiorno è la soluzione Software più diffusa al mondo per l’analisi
dinamica: dell’Esplicito appunto.
Per ulteriori informazioni:
Alfonso Ortalda, EnginSoft
[email protected]
EnginSoft aderisce all’Unione Industriale di Torino
Lo scorso 13 Aprile la sede piemontese di EnginSoft si è
associata all’Unione Industriale di Torino.
L’adesione si inserisce in un crescente interesse di EnginSoft
per le associazioni territoriali di questo tipo, nella
convinzione dell’importanza di aggregarsi alle realtà
industriali locali, di comprenderne le specifiche
problematiche di sviluppo e le richieste di innovazione. In
un mercato sempre più globale, la comprensione del tessuto
sociale ed economico in cui i nostri diversi centri di
competenza si trovano ad operare sono un’imprescindibile
asset ed uno strumento vincente per rendere le nostre
aziende competitive sul piano internazionale.
EnginSoft è associata anche all’Unione Industriali di Trento,
il cui presidente, Ing. Paolo Mazzalai, è anche presidente di
SWS Group, holding del gruppo a cui appartiene EnginSoft,
e all’Unione Industriali di Brindisi.
Per maggiori informazioni:
www.ui.torino.it
EnginSoft Network
60 - Newsletter EnginSoft Year 9 n°2
Avvio del Project Management Office di EnginSoft
La crescente complessità di EnginSoft, conseguente alla
costante crescita del portafoglio Clienti e del fatturato del
gruppo, dell’insieme di prodotti e servizi offerti al mercato, del
numero di dipendenti e di sedi aziendali, in seguito anche al
processo di internazionalizzazione avviato nel 2006, ha portato
ad una notevole crescita del numero e della rilevanza dei
progetti interni ed esterni che l’azienda si trova ad affrontare.
I progetti, che sempre più spesso rappresentano una prassi delle
organizzazioni, sono iniziative per loro natura temporanee,
coinvolgono un team di lavoro interfunzionale e sono sovente
finalizzate a realizzare prodotti, servizi o cambiamenti
organizzativi cruciali per l’innovazione e la competitività delle
imprese. La gestione progetti, disciplina di management
operativo che coordina le risorse per consentire il
raggiungimento di obiettivi prefissati secondo tempi e costi
controllati, è quindi un asset organizzativo fondamentale per le
imprese. Coerentemente con questa visione, la direzione di
EnginSoft ritiene che la gestione progetti richieda uno specifico
supporto organizzativo: EnginSoft ha pertanto avviato la
costituzione dell’ufficio gestione progetti aziendale (PMO,
Project Management Office). Il modello organizzativo del PMO,
ampiamente collaudato e diffuso a livello internazionale, è
efficace per favorire l’adozione e lo sviluppo di buone pratiche
di project management in azienda: queste sono fondamentali per
garantire il raggiungimento degli obiettivi che i progetti
singolarmente si prefiggono e per una corretta gestione della
qualità dei risultati e dei rischi che ogni progetto porta con sè.
Inoltre, il PMO rappresenta una risposta al crescente bisogno di
gestione integrata dei progetti: in tal senso esso diviene
strumento fondamentale per supportare efficacemente le
decisioni strategiche del management aziendale.
Il PMO si colloca in modo ideale in EnginSoft, che è già definita
dal proprio sistema di gestione qualità come una organizzazione
“a matrice”, ed è quindi già organizzata per processi e,
secondariamente, per progetti. EnginSoft è, inoltre,
culturalmente in grado di sostenere il cambiamento
organizzativo che la costituzione del PMO comporta, perchè
guidata da un gruppo dirigente altamente motivato e focalizzato
al raggiungimento di risultati, dove la gestione progetti è, al
EnginSoft Network
tempo stesso, una esigenza ed una pratica quotidiana. Il Project
Management Office di EnginSoft, pertanto, riporterà
direttamente alla Direzione, con le seguenti responsabilità:
• assistere la Direzione nella gestione progetti, in riferimento
a singoli progetti rilevanti, sia nelle valutazioni strategiche
preliminari che per quanto riguarda la gestione operativa. Il
PMO garantirà in particolare l’allineamento tra le scelte strategiche aziendali e gli obiettivi progettuali;
• fornire supporto metodologico alla gestione progetti, inclusa la definizione di standard di lavoro e la formazione interna. Allo scopo il PMO farà riferimento alla metodologia sviluppata dal PMI, Project Management Institute;
• fornire consulenza e supporto ai Project Manager aziendali;
• costituire, mantenere aggiornato e monitorare il portafoglio
progetti aziendale;
• integrare le diverse funzioni aziendali in modo da facilitare
il collegamento tra processi e progetti, la comunicazione tra
gli attori coinvolti e la gestione del portafoglio progetti.
Gli ambiti di intervento iniziali e privilegiati del PMO EnginSoft
saranno i seguenti:
• supporto alle nuove iniziative di business;
• sviluppo delle iniziative di business già lanciate (spin off,
start up, joint venture);
• facilitazione dei progetti interfunzionali (di ambito tecnico);
• metodologie e best practice.
Tramite la formalizzazione del PMO aziendale, EnginSoft intende
fattivamente allinearsi alle migliori pratiche e standard
internazionali, adeguate e funzionali alla propria proposta di
valore e al mercato, perseguendo la visione di gruppo
multinazionale, basato sulla specificità delle competenze e
coordinato in modo unitario da un management diffuso,
secondo precise direttrici individuate dalla Direzione
Generale.
Per maggiori informazioni:
Giovanni Borzi, EnginSoft
[email protected]
Newsletter EnginSoft Year 9 n°2 -
61
The Japan Association for Nonlinear CAE:
a New Framework for CAE Researchers and
Engineers
The joint industry-academia CAE association “The Japan
Association for Nonlinear CAE (JANCAE)” is a nonprofit
organization that mainly provides the “Nonlinear CAE
training course” which was introduced and started in
December 2001 by the founder Professor Noboru Kikuchi,
University of Michigan, USA. As of January 2012, it is
organized by its chairperson, associate prof. Kenjiro Terada
of Tohoku University, 10 executive board members and 49
staff members from universities and companies. Over the
past 10 years, a total of 460 teachers and 3300 participants
joined the training courses which have been held 20 times
since their introduction. Additionally, JANCAE organizes
committees for specific themes and special seminars, and
aims at achieving and realizing maximum effects for the real
CAE business by referring to participants’ opinions and
feedback, to develop training, seminars and planning
accordingly for the future.
The purpose of JANCAE
The change in the needs of CAE users and the expansion of
the fields in which CAE is used, along with the rapid
development of computer technologies, and the obvious
trend that nonlinear simulation is “a must” are clear
requirements today. The movement from linear to nonlinear
is regarded as a step-up to the second generation of CAE.
However, there is a major hurdle with which past experiences
can’t help!
CAE is not just about using CAE software. It means creating
correct models, choosing appropriate analysis methods,
evaluating result outputs from software properly and giving
feedback to the design team after understanding the
surrounding environments and possible phenomena. This is
really a requirement from industry. However, there is a gap
between this clear “requirement from industry” and the
general “university curriculum”. To fill this gap, it cannot be
said often enough, that we rely on the backup from academia
and the support of the software vendors. It would be
dangerous if we ignore this gap while the usage of nonlinear
CAE increases in industry.
This being the situation, JANCAE started its activities from
establishing a forum including CAE training courses, which
offer participants the opportunity to learn nonlinear CAE
intensively, to work hard and learn from each other - as an
initial target. Of course, it is impossible to fill all gaps just
with this training course. For this reason, the long term
target of JANCAE is to establish a new framework for CAE
researchers and engineers.
(JANCAE website: http://www.jancae.org)
Nonlinear CAE training course
A major activity of JANCAE is the CAE training course, which
gathered a total of 3300 participants at its past 20 editions
and about 150 to 200 at each of the courses recently. Fig.1
shows the participants classification by industry. The
Fig 1 - Participants classification by industry
Japan Column
62 - Newsletter EnginSoft Year 9 n°2
universities, material manufacturers and software
vendors, JANCAE has developed a major rubber
material database and simulation templates, which
have been uploaded to their web site in the meantime.
Now, the association also develops a material model
subroutine which can be used for different commercial
FEM codes - This topic was introduced in the EnginSoft
Newsletter 2011 Winter Edition.
Interview
I had the pleasure to conduct the following interviews
with the executive board members of JANCAE.
Fig. 2 - The nonlinear CAE training course curriculum from 2001 to 2010
numbers reflect the overall industrial structure of Japan with
its major sectors: automotive/automotive components,
electrical equipment, general machinery and beyond these,
participants from a variety of fields, such as
materials/chemicals,
steel/metal
and
civil
engineering/building who also attended the training courses.
The attendees from research/academia and software/software
vendors account for about 10%.
Each training course runs over 4 days, the courses are held
twice a year. Despite the fact that the program always starts
on the week-end, the attendance is very good and many
students attend repeatedly. It is very important and
recommended to regularly attend different courses, because
CAE has changed a lot over the last 10 years and evolves
constantly. The program of the training courses is structured
in a first and second half, 2 days for each half. The first half
is for basics. The lessons teach the basics of each topic, such
as material models and elements, and analysis methods. The
second half focuses on the applications. It covers the main
themes of each course, such as coupled fields and multiscale.
Fig.2 shows the training course topics from 2001 to 2010.
The courses are conducted by professional researchers and
industry experts. For sound practical experiences in CAE, it is
necessary to understand coupled fields at various levels. As
you can see in fig.2, the curriculum is well-thought-out, so
that the participants can learn a wide variety of what CAE
covers today. Aside from classroom lectures, the courses also
provide time for hands-on training in the review and
discussion parts of the program.
Committee activities
Independently from the CAE training course, which mainly
consists of classroom lectures, JANCAE organizes “The
Material Modeling Committee” as a practical approach to the
study of nonlinear materials. The Committee was originally
established in 2005 as “The Rubber Committee”. In the
following years, its research activities have diversified into
all material nonlinearity topics including metal plasticity. In
the frame of the Committee, members learn about typical
nonlinear material modeling by studying the basic theory of
the constitutive equations, material testing methods, and
how to handle test data. Moreover, in collaboration with
Japan Column
What changes do you see in the use of CAE in Japan?
H Takizawa, PhD, Mitsubishi Materials Corporation says:
In the past, the engineers who want to use a numerical
simulation had to program it by themselves. Nowadays CAE
software is commonly used in manufacturing companies, also
because the general purpose codes, which are mainly
developed in western countries, have become widespread.
Well-established graphical user interfaces and visualization
tools offer the benefits to simulate many different problems
and gain results quickly. Capabilities like these support
engineers in many companies to reduce time and costs of
development cycles.
However, this environment which delivers results very easily,
also brought along a negative aspect, as we tend to lose our
preparedness to think about the objects of the analysis from
different aspects and to understand it deeply. Indeed, just
getting results is not enough, we should decipher the
necessary information from simulation results and reflect it
with the idea of the design. I have doubts that the
quantitative aspect of substituting prototype testing by
simulation for cost reduction, is standing out in some recent
CAE work. I fear that it has weakened our sense of value, to
see the reality of interest which can’t be seen in prototype
testing, and to understand it deeply. I strongly believe that
we need to get back to such a sense of value.
What are the main characteristics of the way CAE is used
in Japan and what kind of CAE tools are required?
Y Umezu, JSOL Corporation says,
For operations within the organization which require lower
dependence on individual abilities, CAE managements seek to
automatize CAE tools as much as possible, even if this
requires considerable investments. To do this, scripting
capabilities, which make customization easy for matching it
to their own operation processes, are required. On the other
hand, to take full advantage of the individual abilities, CAE
tools are used for the purpose of checking the effects of
his/her own ideas (design change, countermeasure, etc.)
rather than looking for a new idea, because the users know
the risks of using CAE like a black box. In such cases, CAE
tools, which show the effects of their idea through
differences in the results, are required. That’s called KAIZEN
style.
Newsletter EnginSoft Year 9 n°2 -
What do you expect from CAE and its surrounding
environment to contribute to the growth of the Japanese
manufacturing sectors?
T Kobayashi, Mechanical Design & Analysis Co. says,
Japanese manufacturers had been good at making higher and
added-value products with less components (for example,
when we think of cameras and motorcycles). In recent days,
CAE is intended for reasonable simulations for large assembly
products. This tendency/trend can be called emulation rather
than simulation. I think not only emulation, but also the
methods which close in on the essentiality more intuitively
(for example First Order Analysis) are suited to Japanese
engineers.
10 years after the establishment of JANCAE, how do think
has it contributed to the manufacturing in Japan and
what are the remaining challenges?
K Terada, PhD, Tohoku University says,
There are always conscientious CAE engineers who are aware
of the importance of gaining an understanding of both the
basic principles and software usage for simulations. Although
the activities of JANCAE were something grabbing and
immature, I believe that they must have helped satisfying
the engineers’ desire to learn. At the same time, we
succeeded in creating a unique community that enables them
to share the information, mindsets as well as the sense of
63
value with other participants and instructors, by providing an
endorsement to their aforementioned problem consciousness.
It seems that such accomplishments are well recognized and
being taken over steadily by the younger generation.
However, the activities necessarily have limitations because
all the staffs in the JANCAE are volunteers. We are not
satisfied with the current situation and would like younger
supporters, especially staffs from the academia, to be
involved and proactive in expanding our activities. Creating
such an environment and organization is an urgent problem
to be solved.
Conclusions
The environment around nonlinear CAE will change further in
the future. Now, a polarization of “concentrated CAE for
structures” and “distributed CAE for each person” can be
observed in industry. What we should learn is different in
both cases, and we will be required to have better skills in
many different situations. JANCAE will be a new framework to
raise the whole level of CAE users and their individual skills
– both increases when we work with conscious people.
This article has been written in collaboration with the Japan
Association of Nonlinear CAE
Akiko Kondoh, Consultant for EnginSoft in Japan
easy and efficient
coordination of your FP7
EU research projects
24 hours
2
ho
a day, from
m any platfo
platform
fo
orm
www.EUCOORD.com
powered by
European Projects Coordination Tool
[email protected]
64 - Newsletter EnginSoft Year 9 n°2
METEF Foundeq 2012: EnginSoft soddisfatta della
partecipazione all’esposizione
Lo stand di EnginSoft al salone internazionale dedicato al mondo dell'alluminio e delle macchine per la lavorazione
dei metalli, raccoglie consensi e cattura l’attenzione dei visitatori
METEF Foundeq, expo internazionale dedicata alla filiera produttiva dell'alluminio e dei metalli non ferrosi, è uno degli
eventi di maggior rilievo e di richiamo internazionale che si
occupa di tecniche e tecnologie innovative per l'industria fusoria. L’evento, che si è svolto per la prima volta presso la
Fiera di Verona dal 18 al 21 aprile 2012, ha registrato 15.000
presenze, un incremento degli operatori professionali presenti e della presenza di buyer esteri: un risultato positivo, nonostante la non facile congiuntura economica.
Fin dalla prima edizione, EnginSoft partecipa alla manifestazione con un proprio spazio espositivo: quest’anno ha riscosso molto interesse la presentazione su una pedana girevole
di alcuni getti, ottimizzati con l’uso del software di simulazione dei processi di fonderia MAGMA5.
I visitatori hanno potuto apprezzare componenti di alta qualità meccanica come il forcellone di una moto Ducati, il sottobasamento di un dodici cilindri Ferrari o il canotto di uno
sterzo, vicino alle tecnologie di simulazione che contribuiscono alla perfetta realizzazione di questi capolavori dell’ingegneria italiana. Componenti realizzati grazie alla passione
EnginSoft diventa socio di
AMAFOND: Associazione
nazionale fornitori macchine e
materiali per fonderie
AMAFOND è l’Associazione Italiana dei Fornitori di Macchine,
Prodotti e Servizi per la Fonderia, nata nel 1946, punto di
riferimento per gli operatori del settore offrendo servizi
tecnici, normativi, economici e legislativi.
Dalla sua costituzione è aderente all’Unione del Commercio e
dei Servizi della provincia di Milano. L’Amafond è inoltre
socio fondatore del Cemafon-Comitato Europeo dei
Costruttori di Macchine ed Impianti per Fonderia.
L’Associazione ha lo scopo di coordinare, tutelare e
promuovere gli interessi tecnici ed economici del settore
macchine e prodotti per fonderia, in generale di tutti i
fornitori delle industrie metallurgiche. Le Aziende associate
sono circa settanta, divise in quattro grandi gruppi: gruppo
Events
di fonderie come Perucchini, GFT e modellerie come CPC.
EnginSoft ha anche presentato le anime che si impiegano
nella realizzazione di questi componenti: esempi che dimostrano le potenzialità di MAGMA Core and Mold, un software
innovativo che finalmente permette di simulare il processo di
generazione delle anime (Core). Il modulo propone un modello di lavoro facile e intuitivo, che aiuta a comprendere a fondo il processo produttivo delle anime e consente di analizzarne tutte le fasi: strumento indispensabile per individuare le
migliori strategie progettuali per la definizione delle casse
d’anima.
All’interno della manifestazione si è svolto anche
l’International Forum on Stuctural Components by High
Pressure Die casting HPDC - Prospettive e sfide dell'industria
della fonderia ad alta pressione, organizzato da AIM - Die
Casting Study Group e Alfin-Edimet Spa in cooperazione con
Amafond, Assofond, Assomet, Cemafon e Caef. Al forum, coordinato dal nostro Piero Parona, presidente del Centro di
Studio Pressocolata dell’AIM, Associazione Italiana
Metallurgia.
prodotti, gruppo macchine ed
impianti, gruppo forni per
fusione e trattamenti termici,
gruppo pressofusione.
Queste Aziende rappresentano
un volume di affari di circa 1
miliardo di euro con una quota
export superiore al 60% ed
impiegano quasi 5000 addetti.
EnginSoft si è associata ad Amafond per rafforzare e
sviluppare ulteriormente i propri contatti con i titolari e i
decision maker delle Fonderie Italiane e per promuovere
EnginSoft anche internazionalmente come fornitore di
servizi, grazie alla presenza istituzionale di Amafond alle
principali fiere Internazionali con la diffusione della
directory delle proprie aziende associate.
Per ulteriori informazioni:
www.amafond.com
The Fisker Karma 2012
Electric Luxury Car will be
showcased at the CAE
Conference 2012
Henrik Fisker will be the
Keynote Speaker of the
Automotive session
22 ber
to
Oc 2012
CAE
The CAE Poster Award is an EnginSoft initiative which is part of the program of the
International CAE Conference 2012 that will take place in Lazise (Verona) - Italy from
22 to 23 October.
INTERNATIONAL
The CAE Poster Award is a competition dedicated to the best posters that show
original and relevant CAE applications.
The CAE Poster Award is part of the EnginSoft CAE Culture Promotion Program to
improve the correct use of simulation tools, both in industry and the academia, and
to foster the growth of the CAE analysts community.
The Poster Award is divided into two categories:
- industry: all types of companies are welcome to take part
- accademia: students, graduated students and researchers are welcome to take part.
A Scientific Committee will select the 10 best posters for each category, and the first three posters will be
awarded during the Conference evening on 22 October.
The participation in the competition is free. It includes a free pass to the International CAE Conference 2012 for
the nominated authors.
Deadline for the poster submission: 28 September 2012.
For more details, please visit: www.caeconference.com
POSTER AWARD
Promoted by
www.caeconference.com
66 - Newsletter EnginSoft Year 9 n°2
International modeFRONTIER
Users’ Meeting 2012
From innovative wind generators to the calibration of a diesel
oxidation catalyst and the optimal assembly of a genome, ESTECO
successfully celebrated the fifth edition of the International
modeFRONTIER Users' Meeting, on 21st and 22nd May 2012 in
Trieste, Italy. Speakers and participants travelled from different
parts of the world to share their experiences and knowledge,
address some of the most relevant issues of the sector and to
learn about the potential, the latest use and applications of the
software.
In his welcome speech, Prof. Carlo Poloni, President of ESTECO,
introduced the meeting’s primary theme: collaboration, and
highlighted how nowadays sharing knowledge and resources is
ranked high as a key success factor in many companies.
Technology can be of paramount importance in the process of
cooperation, for example within the application of
multidisciplinary methodologies, such as the ones supported and
enhanced by the ESTECO technology. The guest of honor was
David Edward Goldberg, Director of the Illinois Genetic
Algorithms Laboratory (IlliGAL) and professor at the Department
of Industrial and Enterprise Systems Engineering (IESE) of the
University of Illinois. As the leading expert of genetic
algorithms, Prof. Goldberg focused in his speech on the concept
of "innovation" and "collaboration" from the perspective of
genetic algorithms (GAs), search procedures based on the
mechanics of natural selection and genetics. “GAs may be
thought of as computational models of innovation, - said Prof.
Goldberg - but may also be thought of as inducing a kind of
collaborative process between human and machine and also as
models of certain kinds of social systems.”
Among other specialists, Prof. Alberto Tessarolo showed the
results of a recent work completed with Ansaldo Sistemi
Industriali on the application of modeFRONTIER as an aid to the
design of innovative wind generators of different size and
conception. Prof. Tessarolo explained how the use of genetic
optimization techniques is applied to the design of complex
systems and components in industrial applications. He outlined
how in the particular example, interfacing electromagnetic and
thermal finite-element computation programs in the
modeFRONTIER multi-objective constrained optimization
environment made it possible to identify the most promising
design configuration in absence of previous industrial
experiences of electric machines with similar characteristics. Mr.
Luciano Mariella from Ferrari GeS, unveiled encouraging results
about an innovative project for the London 2012 Olympic Games
carried out by Ferrari Gestione Sportiva and C.O.N.I., aiming at
improving the hydrodynamic performance of the rudder of the K1
Events
and K2 Kayak. The optimization loop was focused on CFD
simulations to properly evaluate the rudder's, the fin's
hydrodynamic performance and identified new “optimum” shapes
which were built and tested by Italian national athletes in
preparation for the upcoming Olympic Games. Several other
experts from leading companies and prominent international
research centers, provided exciting results on application cases
using the modeFRONTIER software across a wide range of hightech industrial sectors during the two busy days.
For more information, please visit:
um12.esteco.com
Graz Symposium Virtual Vehicle
EnginSoft Germany participated as an exhibitor in the 5th
edition of the Graz Symposium Virtual Vehicle on April 17th
and 18th 2012. Organised by the competence center "Das
virtuelle Fahrzeug Forschungsgesellschaft mbH" it hosted
around 150 particpants from companies like Daimler and
Audi coming together presenting and discussing processes,
methods, tools and best practices in interdisciplinary
vehicle development. EnginSoft Germany showcased
multiple disciplines such as structural analyses and
aerodynamic simulations driving a vehicles' development
can be connected through the workflow automation and
design optimization software modeFRONTIER in order to
enhance virtual vehicle development.
www.gsvf.at/cms/
Constructive Approximation
and Applications
EnginSoft will participate in and sponsor the 3rd Dolomites
Workshop on Constructive Approximation and Applications
(DWCAA12). The conference will take place in Alba di
Canazei (Trento, Italy), from September 9 to 14, 2012.
DWCAA12 aims to provide a forum for researchers to present
and discuss ideas, theories, and applications of
constructive approximation. The conference will host
several interesting keynote speakers and presentations.
EnginSoft will sponsor the Best Paper Award for young
researchers. This award (worth up to 500 euro) will be
given for the best paper submitted for the proceedings of
the conference by either a student, PhD student or postdoc. In case of more than one author (with all co-authors
in one of the above categories), the amount will be divided
equally among the co-authors. The selection of the best
paper will be made by the Scientific Committee of
DWCAA12.
Further information on the themes and sessions of the
conference and the Call for Paper can be found at:
http://events.math.unipd.it/dwcaa2012/
By participating in DWCAA12, EnginSoft is supporting new
studies on approximation and the importance that these
techniques have to tackle real-case industrial problems.
Newsletter EnginSoft Year 9 n°2 -
67
Event Calendar
ITALY
International CAE Conference
22-23 October 2012
www.caeconference.com
CALL FOR PAPER IS OPEN!
_________________________________________________
19,20,21.06.2012 - Corso di formazione "La simulazione elettromagnetica di apparati elettrici - caratterizzazione avanzata dei materiali magnetici"
Padova c/o Competence Center EnginSoft
26.06.1012 - Workshop "Massimizzare le performances nei
processi fusori della ghisa e dell’acciaio per ottenere getti di
qualità superiore"
Padova c/o Competence Center EnginSoft
27.06.2012 - Workshop "Simulazione dei Processi di
Stampaggio a Caldo di Metalli non Ferrosi: Nuovi Sviluppi,
Vantaggi e Prospettive"
Brescia c/o API
04.07.2012 - Workshop "Sperimentazione virtuale come strumento srategico per la competitività delle aziende"
Altavilla Vicemtina (VI) c/o CUOA
11.07.2012 - Workshop "Sperimentazione virtuale nella
Dinamica della frattura e caratterizzazione dei materiali";
Torino c/o Centro di Formazione FIAT del Lingotto
18.07.2012 - Workshop "Progettare strutture e componenti
con materiali compositi - scelta dei materiali e simulazione"
Bergamo c/o Competence Center EnginSoft
18.07.2012 - Workshop "Lo stato dell'arte delle tecnologie di
simulazione dei settori strategici: Oil&Gas, Power e Chemical"
SanDonato (MI) c/o Hotel Crown Plaza
Per maggiori informazioni e dettagli sugli eventi:
www.enginsoft.it/eventi
[email protected]
FRANCE
6-7.06.2012 NAFEMS French Conference, Paris.
27-28.06.2012 TERATEC Forum, Paris
meet us at EnginSoft booth.
13-14.11.2012 Virtual PLM, Reims
EnginSoft will be exhibiting and presenting.
GERMANY
24-25.05.2012 CST European User Conference. Mannheim.
EnginSoft presented: "Microwave Bandpass Filter MultiObjective Optimization using modeFRONTIER & CST MWS".
18-22.06.2012 Achema Conference 2012. Frankfurt/Main.
25-26.06.2012 FLOW 3D European Users Conference. Munich.
03.07.2012 JMAG Users Conference, Frankfurt/Main.
24-26.10.2012 ANSYS User's Conference & CADFEM Users
Meeting, Kassel.
10.2012 GT Power user's conference. Frankfurt/Main.
UK
modeFRONTIER Workshop at the University of Warwick
* 13 June
* 18 July
* 5 September
* 16 October
* 8 November
* 10 December
30-31.05.2012 NAFEMS UK Conference - EnginSoft gave a presentation.
04.07.2012 modeFRONTIER for InfoWorks CS Workshop
New Interface demonstration day, University of Warwick
11.07.2012 NAFEMS - Using Variability in Simulation:
A Practical Workshop, Teddington -EnginSoft will be attending
SPAIN
10.05.2012 NAFEMS Awareness Seminar on Numerical
Methodologies and Modeling of Coupled Systems, Madrid.
For more information:
www.enginsoft.com
Events
INTERNATIONAL
22-23
OCTOBER
2012
CONFERENCE
www.cobalto.it
Presentations in
The new voice of CAE
Join us at the International CAE Conference from the 22nd-23rd October
Key industry leaders, solutions and insight; further your
professional experience and expand your industrial network
LAGO DI
GARDA
Hotel Parchi del Garda
Via Brusá, località Pacengo
Lazise (VR) - Italy
Tel. +39 045 6499611
www.hotelparchidelgarda.it
INTERNATIONAL CAE CONFERENCE - INFOLINE
[email protected] - Tel. +39 0461 915391
Special Guest:
Professor Parviz Moin
Mechanical Engineer
Professor at Stanford University;
Worldwide expert in fluid dynamics
www.caeconference.com