MINISTERO DELL'UNIVERSITĀ E DELLA RICERCA SCIENTIFICA E TECNOLOGICA
PROPOSTA DEFINITIVA PER LA COSTITUZIONE DI CENTRI DI RICERCA
RICHIESTA DI COFINANZIAMENTO
(DM n. 11 del 13 gennaio 2000)
Anno 2000 - prot. CE00482749
1. Name of the Centre and Principal Investigator
| Name of the Centre |
Centre for Computational Mechanics
|
| Principal Investigator |
NAPOLITANO Michele
|
| University |
Politecnico di BARI
|
2. Short- and long-term objectives
On October 1, 2000, the
Istituto di Macchine ed Energetica and the Dipartimento di
Progettazione e Produzione Industriale of the Politecnico di Bari have
merged to constitute the Dipartimento di Ingegneria Meccanica e
Gestionale(DIMeG), in order to provide an ideal environment for the
research activities in the areas of Mechanical and Management
Engineering. Inside this Department, we plan to build a
multidisciplinary Centre of Excellence for Computational Mechanics of
solids and fluids, which would put together the expertise and know-how
of most professors of the DIMeG in the areas of numerical methods and
information technology for supporting mechanical design (both
functional and constructive) i.e., aspects of kinematics, dynamics and
fluid-dynamics, as well as static and fatigue design, as applied to the
sectors of turbomachinery, automotive components, and mechanical
engineering, in general. This expertise would be now mature, provided
funding is granted, for a double leap forward:
(i) Integration/competition/synergy with the best similar centres, with
which the PPI already have links and cooperation programs (including
some of the most prestigious Universities/Research Centres in the
world, as MIT, Stanford, Harward, Oxford, Cambridge, Southampton,
Sheffield, AFWAL, Ohio, U.C. Berkeley, UCLA, University of Cincinnati,
ENS, Cachan (FR); Von Karman Institute for Fluid Dynamics, Virginia
Tech, etc.), towards further developments of computational methods and
of information technology for the state-of-the-art design in the
aeronautical and mechanical fields.
(ii) Technology transfer of the available technology for the sustain of
local industries, nowadays growing fast in these sectors, such as: a)
mechanical systems, in the Bari area, with extension in Gioia del
Colle, for production of mechanical transmissions, braking systems and
in general automotive and general mechanical components; b)
aeronautical, mechanical, propulsion, naval, chemical-pharmaceutical
and energy systems in the Brindisi area.
(iii) Educational activities, with the scope of facilitating the goals
listed above, in synergetic effort with the other teaching activities
at the Politecnico di Bari, at the degree level, but also at PhD level
(The Principal Investigator of the proposed Centre is also the Chairman
of the Ph.D. program in Mechanical Engineering of the Politecnico di
Bari), as well as continuing education programs.
At present, several resources in Computational Mechanics are
distributed between the two main buildings of the DIMeG so that
integrating such structures will be greatly favoured by the proposed
Centre.
The short term objective of the Centre is to develop, starting from the
existing resources, a supercomputing environment allowing to explore
engineering systems using advanced computational models which cannot be
handled by workstations of standard type. Moreover, the proposed
research projects will lead, in the medium-long term, to independent
support from industries and will achieve this goal by selecting
activities relevant for both the industrial and scientific aspects.
In summary, in the first three years of activity, the Centre will carry
on several advanced research projects in various areas of computational
mechanics and will build up a supercomputing environment which will be
available to both industrial and academic partners. At the end of each
year, a workshop will be organized which will present to all potential
partners the results of such activities. Also, research proposals will
be submitted to Italian and international industries as well as to
MURST, the UE etc., so as to quickly achieve financial independence.
The medium- and long-term objectives of the Centre are to provide a
venus for: i) performing state-of-the-art research in computational
mechanics; ii) helping Italian and international industries to solve
design, development and production problems; iii) improving the quality
of education in the area of mechanical and industrial engineering at
both the undergraduate and graduate levels. It is anticipated that at
the end of the first year of activities, the Centre will be able to
attract outstanding research as well as researchers from both Italy and
abroad.
REMARK. In case the Centre is funded, its activities will start most
likely at the beginning of the year 2001. Accordingly, for all research
lines of this proposal, the man-months indicated for the year 2002 are
also valid for the year 2003.
3. Research project of the first three years of activity
Workpackage 1
Title
|
Computational Fluid Dynamics |
Background
The Principal
Investigator (PI) has been working in Computational Fluid Dynamics
(CFD) for about 25 years. He has published more than 100 scientific
papers on the subject, more than 40 appearing on international journals
and books. At present, he and his younger colleagues (proff. Catalano,
De Palma, Pascazio and Verzicco) are involved in several research
projects on CFD with emphasis on: numerical methods for the Reynolds
averaged Navier-Stokes equations; large eddy and direct numerical
simulation of turbulence; turbomachinery fluid dynamics; fluid dynamic
design and optimization. Besides pursuing such lines of research, they
also intend to combine with their solid mechanics colleagues to start a
new research line on flutter. In preparation of such an activity, one
of the Ph.D. students in Mechanical Engineering has been sent to the
University of Oxford to work with prof. Mike Giles on such a very
challenging and interesting problem.
The motivation for the activities of the work package on CFD is described in the following.
In the last three decades, the speed of digital computers has
experienced an impressive growth; it has been estimated, in fact, that
the peak performance of processors double every 15-18 months (Moore's
law). At the same time the cost of the hardware has continuously
decreased giving further impulse to the numerical simulation of
physical phenomena. Within this scenario, computer simulations have
become competitive with respect to laboratory experiments especially in
those fields, like fluid dynamics, where the realization of prototypes
and field measurements are both complex and expensive. Of course, this
does not mean that experiments are useless or out of date, but only
that many of the rapid prototyping and trial-and-error phases of a
design can be substituted by less expensive numerical simulations. In
this way, detailed laboratory experiments can be restricted to the
final stage of the design, thus limiting the cost of the product.
These considerations have induced many research centres and industries
to the use of CFD, that nowadays has become a standard tool in both
research and industrial production.
Although being a well established discipline, CFD is however very far
from being mature, since there is a number of open problems whose
solution (at least partial) would give considerable benefits in the
field of applications.
One of the main problems in CFD is related to the decrease of the
smallest flow scale with the Reynolds number (Re) which is the ratio of
inertia to viscous forces. This phenomenon, known as turbulence
transition, implies that for increasing values of Re, smaller and
smaller flow scales are generated and their interaction produces a
time-chaotic highly unpredictable dynamics. Consistently, numerical
methods for solving high Re turbulent flows must have a space
resolution fine enough to capture the smallest flow scales and the time
integration must be carried out with small enough steps in order to
follow the dynamics of the smallest eddies. Todate, Direct Numerical
Simulation (DNS) of large scale turbulent flows is still far beyond the
capabilities of modern computers. Nevertheless these studies are of
great practical importance, since basic mechanisms of the flow dynamics
can be studied and this helps in the interpretation of more complex
flows. In addition, when turbulence models need to be tuned, or their
performance evaluated, DNS of simple flows provide widely used
benchmarks. A recent use of DNS is for instrumentation calibration,
since many measurement devices in fluid dynamics are highly non linear
(e.g., multi probe hot wires) and they need accurate calibrations in
order to give reliable results.
The alternative approach to solve turbulent flows of engineering
interest implies the use of turbulence models to simulate high Reynolds
number flows. Without describing in detail the differences among the
various models here we only distinguish between those based on
filtering (Large Eddy Simulation) and those based on averaging
(Reynolds Averaged Navier Stokes). Large Eddy Simulation (LES) models
compute all of the space and time scales of the flow which can be
resolved by the used grid, so that only the sub-grid-scales need to be
modelled. Since in turbulence (even in highly inhomogeneous flows) the
small scales have an homogeneous and isotropic behaviour, their
modelling is usually easier and quite general. The most promising LES
models are those of the ``dynamic'' class, i.e., those in which none of
the coefficients is externally provided but they are instead computed
as part of the solution. Although LES simulations are usually more
expensive than those based on the numerical solution of the Reynolds
Averaged Navier Stokes (RANS) equations, they have been seen to perform
very well also in strongly unsteady flows, rotating and stratified
flows, or in flows with extensive separations where RANS simulations
usually give less accurate results. On the other hand, RANS simulations
model all turbulent fluctuations in the field so that only the averaged
flow-field needs to be computed explicitly. Todate RANS models are the
only possible choice when high-Re, complex-geometry flows are to be
computed, so that they have become the standard in the industrial
design. In fact, although turbulence modelling is still far from being
satisfactory, for a given application, after tuning the parameters of
the turbulence model being employed, satisfactory results can be
obtained at a reasonable computational cost. Thus, all industrial CFD
codes use the RANS equations solved by means of finite volume or finite
element methods. Such methods have reached a good level of maturity and
are used with success in both the design and analysis of airplanes,
turbomachinery, etc.
On the other hand, besides the aforementioned need for improved
turbulence models, further progress is also needed with respect to the
accuracy, efficiency and robustness of numerical methods. In
particular, genuinely multidimensianal upwind methods are needed to
model the multidimensional nature of wave propagation phenomena more
accurately than standard methods based on a directional splitting.
Another very important aspect of CFD is the development of accurate and
efficient techniques for the aerodynamic design. The numerical approach
to the design process of fluid-dynamic components has the great
advantage of reducing the computational cost of the design phase, which
is based, to date, on subsequent improvements based on the experience
of the designers, each followed by expensive and time-consuming wind
tunnel tests. Two main approaches are currently followed by the CFD
community, the first one using stochastic algorithms and the second one
using gradient-based methods. Although such two methodologies have been
competing for the last years, it is now becoming more and more evident
that the two approaches could combine very efficiently, provided that
they work in a sequential manner, so as to identify good candidates,
first, and then, starting from these good configurations, to search the
optimum. Both methodologies require a particular care in the way they
are coupled with the CFD codes. A so-called black-box approach, in
which the CFD code is interfaced with the optimization code by means of
the flow solution only, requires a complete flow analysis every time
the objective function or its derivatives must be computed. In such a
case, the computational time required by the optimization process would
clearly become unaffordable. Thus, CFD and optimization scientists
involved in this research field are developing new techniques and
strategies, which contribute to reduce the time required by the design
process.
Finally, CFD is becoming a usable tool for studying flutter, which is a
very critical problem for today's aircrafts and advanced
turbomachinery. Such a topic will be the subject of the last research
line of this workpackage.
Objectives
The main objectives of
the first workpakage are to advance the state-of-the-art in the various
aspects of CFD which will be pursued during the three years of activity.
In more details, the workpakage on CFD includes four research line:
1. Large Eddy and Direct Numerical Simulation (LES and DNS) of Turbulent Flows.
2. Advanced Numerical Methodologies for the Compressible Reynolds Averaged Navier-Stokes (RANS) Equations.
3. CFD Design and Optimization.
4. Flutter.
1. LES and DNS of Turbulent Flows.
The main objective of this research line is to create a library of well
established basic and industrially relevant flows which will allow the
validation of turbulence models and concepts relevant to practical
applications. This will be of great value for the second line of
research, which requires suitable turbulence models in combination with
accurate, efficient and robust numerical methods in order to solve flow
problems of engineering interest.
Another aim of this research is to set up a group with the know-how in
DNS and LES. This will complement and complete other research areas
like turbomachinery and fluid control.
2. Advanced Numerical Methodologies for the Compressible RANS Equations.
The main objective of this research is to develop state-of-the-art
methods for solving the compressible steady and unsteady RANS
equations. Such methods will allow the Centre to perform reliable
calculations for complex flows of industrial interest. Moreover the
most efficient time-dependent method developed in this research line
will also be an important tool to study flutter without resorting to
linearized models.
3. CFD Design and Optimization.
The main objective of this research is to devise accurate, very
efficient, and robust technologies for the automatic design of
fluid-dynamic components.
The state of the art and the current research in the field of the
fluid-dynamic optimization show that important contributions are still
required for reaching the following requirements: 1) capability of
analyzing a wide range of geometries:this would allow to investigate
non-standard configurations, which can offer large improvements with
respect to the existing ones; 2) improvement of the efficiency:the
optimization process should require a computational time comparable to
that required by the analysis of one configuration, regardless of the
number of parameters used. The proposed research aims at satisfying
both the above requirements. The first one will be approached by using
a slope-driven approach only, as well as by using both stochastic and
gradient-based optimization methods, in a sequential manner: in fact,
modern stochastic techniques, based on genetic algorithms, have
demonstrated their ability to explore innovative and unusual
configurations and to locate globally optimal solutions, which can be
used as starting points for the final, gradient-based, process of
refinement of the optimal configuration. Concerning the second
requirement, namely, efficiency, a progressive gradient-based
optimization strategy using adjoint equations for the evaluation of the
sensitivity derivatives will be developed and applied to the inverse
and the direct design of turbomachinery blades and of industrial
after-burners. Such technique is based on the idea of using different
levels of grid and of convergence, so as to obtain an increasing level
of accuracy of both the flow solution and the sensitivities while
converging the design problem to the optimum.
The use of adjoint equations will make the computational work
independent of the number of parameters used to define the
configuration under investigation. For the case of the combined
approach (stochastic+gradient based), an improvement of the
computational efficiency of the stochastic methods is needed, since a
large number of flow analyses is required. A possible approach to
reduce the impact of so many flow computations will be considered,
consisting in the development of a multi-level strategy, namely, in the
use of different levels of accuracy of the analysis tools involved in
the design process, and in its application to the inverse and the
direct design of airfoils.
Both approaches that will be considered clearly rely on the idea that
cheap, but lower accurate, analysis tools can be used in the first
steps of the optimization process, where globally optimal solutions are
found, whereas an higher accuracy is strictly required in the
refinement phase.
More in general, the idea is to combine lower and higher fidelity
models in the optimization process, so as to gain an overall reduction
in the computational time, compared with a strategy always invoking the
highest fidelity model.
4. Flutter.
The main objectives of this research are to develop efficient and
accurate numerical techniques for the analysis of flutter and for the
aeromechanical design of the blading, which can be integrated in the
design system of turbomachinery industries. An accurate numerical
definition of the flutter margin of bladed rotors would require the
solution of the coupled unsteady equations for both the structure and
the fluid (RANS) over the computational domain consisting of the whole
disc, all blades and passages between adjacent blades. This procedure
is computationally very expensive, as it requires cpu times of the
order of days and therefore it cannot be introduced into everyday
design practice. Nevertheless, one of the activities of this research
line will be the development of a very efficient solver for the RANS
equations, in order to speed up such computations so that they may be
used at least at the final stage of the design process.
An alternative approach for the analysis of flutter in turbomachines is
the linear method. Its implementation is a fundamental part of the
proposed research program and it is based (i) on the hypothesis of
cyclic symmetry of the bladed rotor, (ii) on the fact that the
frequencies of the structural modes are well apart each from the other
and flutter typically involves only one of such modes, (iii) on the
fact that the aerodynamic forces are much smaller than the elastic and
inertial ones and the difference between structural modes in vacuo and
aeroelastic modes is negligible and (iv) on the evidence that the
amplitude of the vibration is small in conditions of incipient flutter
and therefore the system behaves linearly.
Finally, in cooperation with the group pursuing the research line n.4
of the second workpackage, particular attention will be devoted to
provide an effective and user-friendly visualization of the
computational results. This is a very critical issues in all of the
proposed research lines, which require large scale computations and
thus the handling and interpretation of huge amounts of data.
Description of the activities
Research line n.1
Group leader
| Surname |
VERZICCO |
| Name |
ROBERTO |
| University |
Politecnico di BARI |
| Faculty |
Engineering |
| Affiliation |
DIMeG |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
VERZICCO |
ROBERTO |
Pol. Bari |
DIMeG |
1 |
4 |
4 |
| 2. |
SASSANELLI |
GIUSEPPE |
Pol. Bari |
DIMeG |
1 |
5 |
5 |
| 3. |
SELVAGGIO |
SERGIO |
Pol. Bari |
DIMeG |
1 |
5 |
5 |
| 4. |
FATICA |
MASSIMILIANO |
Stanford |
CTR |
1 |
1 |
1 |
| 5. |
MAGI |
VINICIO |
Basilicata |
DIFA |
1 |
3 |
3 |
Five recent papers
R. Verzicco and R. Camussi, ``Prandtl number effects in convective turbulence'', J. of Fluid Mech., vol. 383, (1999), pp. 55-73.
R. Verzicco and J. Jimenez, ``On the survival of nonuniform vortex
filaments in model turbulence'', J. of Fluid Mech., vol. 394, (1999),
pp. 261-279.
Verzicco, R., Mohd-Yusof, J., Orlandi, P. and Haworth, D., ``
Large-Eddy Simulation in Complex Geometric Configurations Using
Boundary Body Forces'', AIAA J., vol. 38, (2000), pp. 427-433.
Camussi, R. and Verzicco, R., ``Anomalous scaling exponents and
coherent structures in high Re fluid turbulence'', Phys. of Fluids,
vol. 12(3), (2000), pp. 676-687.
Fadlun, E.A., Verzicco, R., Orlandi, P. and Mohd--Yusof, J., ``Combined
immersed-boundary finite-difference methods for three-dimensional
complex flow simulations'', J. of Comp. Phys., vol. 161, (1999), pp.
35-60.
Research activity
Background and
motivation for this research line have been given already in the
introductory description of the work package. In this section we will
describe more in detail the objectives and research activity on Direct
Numerical Simulation (DNS) and Large Eddy Simulation (LES).
Concerning the DNS the first objective is to define and simulate a
number of test cases for free-shear and wall bounded flows. These cases
will be selected according to their technological relevance, the
availability of experiments for the validation of the results and their
interest in the scientific community. For example in most combustors of
gas turbines the flow is swirled in order to stabilize the flame.
Unfortunately the present RANS models does not give satisfactory
results for swirling and rotating flows since they do not properly
account for the extra curvature of the streamlines.
In the industry this problem is partly aleviated by ``ad hoc''
corrections that, however, are not general and must be modified from
case to case.
In the research centres, in contrast, an intense activity is presently
carriedout on the simulation of swirling and rotating flows with the
aim of understanding the mechanisms of the flow dynamics and producing
new turbulence models (or improving the existing ones). In this context
the direct numerical simulation of a swirling jet seeded with a passive
scalar could be a case simple enough to be accurately simulated but
still retaining many features of the original technological problem.
Following the same phylosophy, additional test cases will be selected
in agreement with the other members of the excellence centre and these
will include also wall bounded and free shear flows. All simulations
will be carried out according to the present state-of-the-art thus
using grids of the order of 10-100 millions of nodes and therefore
requiring paraller computer resources available at the centre or at
external computing centres. The results of each simulation will form a
database available to all members of the centre and, upon agreement,
also to external researchers. Information on integral quantities (like
force coefficients) as well as mean and turbulent profiles will be put
on the web site of the centre to give maximum visibility to the
performed activity. Finally a detailed report on the simulations and on
the advancement of the activity will be published on the annual
activity report of the centre.
It should be stressed that DNS is a research tool rather than an aid to
industrial design and, even if every fifteen years the computing power
grows by a factor O(10^3), the application of DNS to practical
applications is not foreseen in the near future.
The situation is different for LES simulations which already have a
niche for industrial applicability and continuously gain interest as
design tool. The reason for this is twofold; the first is that accurate
LES simulations are currently carried out at Re=O(50000) and few
applications of internal fluid dynamics are just in this range. The
second reason is that for several flows, once the turbulence transition
has happened, the flow does not change much with Re (think of the drag
coefficient of the cylinder beyond Re=200000). In this respect LES can
give good predictions even when the full scale Re simulation is not
possible. In this case the objective is to carry out LES of flows that
include effects of walls, scalar dispersion and strong separations.
These can be decided in agreement with other groups of the centre (for
example in a common project) or proposed by industrial affiliates in
case experiments for validation are available.
This activity will be initially useful in order to establish
performances and accuracy of different sub-grid-scale models and
filtering procedures. These will be additionally evaluated also by ``a
priori'' tests performed on the DNS database previously described.
Similarly to DNS also the LES activity will be published in the annual
report of the centre and on its web site.
A few words should be spent about the flow analysis since the
visualization of three-dimensional turbulent fields has always been a
difficult task. In the present case, we will take advantage of the
virtual reality research line of this centre for new rendering and
visualization procedures. These will be used in addition to the
standard techniques with the aim of making easier and more intuitive
the investigation of complex flow details.
The main goal of this research line is to set up a group with the
know-how in DNS and LES. This will certainly complement and complete
other research areas like turbomachineries and fluid control. Another
aim of this research is to create a library of well established basic
and industrially relevant flows which will allow the validation of
turbulence models and concepts relevant to practical applications.
Financial support
| Amount (ML) |
10 |
| Source(s) |
Ateneo |
Research line n.2
Group leader
| Surname |
DE PALMA |
| Name |
PIETRO |
| University |
Universita' degli Studi di ROMA "Tor Vergata" |
| Faculty |
Engineering |
| Affiliation |
DIM |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
DE PALMA |
PIETRO |
Roma "Tor Vergata" |
DIM |
1 |
3 |
3 |
| 2. |
CINNELLA |
PAOLA |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
| 3. |
BONFIGLIOLI |
ALDO |
Basilicata |
DIFA |
1 |
3 |
3 |
| 4. |
NAPOLITANO |
MICHELE |
Pol. Bari |
DIMeG |
1 |
1 |
1 |
| 5. |
PASCAZIO |
GIUSEPPE |
Pol. Bari |
DIMeG |
1 |
2 |
2 |
Five recent papers
De Palma P., Pascazio
G., Napolitano M.,``A validation study of a hybrid fluctuation
splitting scheme for transonic inviscid flows'', Lecture Notes in
Physics, Vol. 515, pp.373-378, Springer-Verlag, Berlin Heidelberg, 1998.
De Palma P., Pascazio G., Napolitano M., ``A hybrid multidimensional
upwind scheme for compressible steady flows'', AIAA Paper 99-3513, 30th
AIAA Fluid Dynamics Conference, Norfolk (VA), 1999.
De Palma P., Pascazio G., Napolitano M.,``A multidimensional upwind
solver for steady compressible turbulent flows'', to appear in
Computational Fluid Dynamics Journal, Japan Society of CFD, Japan, 2000.
Bonfiglioli A., De Palma P., Pascazio G., Napolitano M., ``An Implicit
Fluctuation Splitting Scheme for Compressible Flows'', 1st
International Conference on Computational Fluid dynamics, Kyoto, 2000.
Bonfiglioli A., De Palma P., Pascazio G., Napolitano M., ``A hybrid
fluctuation splitting scheme for the steady Euler equations in three
dimensions'', ECCOMAS Congress, Barcelona, 2000.
Research activity
Numerical methods for
the solution of the Navier-Stokes equations have reached such a
maturity level that numerical simulation is assuming a more and more
important role in the aerodynamic and thermo-fluid-dynamic design of
advanced aircraft and turbomachinery. Nevertheless, two main
difficulties make such methodologies not yet fully satisfactory. The
first one concerns the lack of robust models for transition and
turbulence, which makes a correct prediction of heat transfer in
turbomachinery blades still very difficult to achieve, particularly in
transonic flows. The second difficulty is related to a theoretical
limit of the numerical methods for the solution of the Navier-Stokes
equations currently employed in practically all industrial
applications. In such methods, in fact, the advection terms, accounting
for the wave propagation phenomena and the presence of discontinuities
in transonic and supersonic flows, are modelled by either a centred
discretization scheme with artificial viscosity or by upwind methods
based on a grid-aligned one-dimensional model of wave propagation
phenomena. Such models are unsuitable to correctly simulate the
genuinely multidimensional physics of wave propagation. This second,
methodological, aspect is considered in the present Research Line 2
(RL2) following one of the approaches that, in the last decade, have
proven more promising: the Fluctuation Splitting (FS) method. Such a
method has been studied by some of the best Computational Fluid
Dynamics (CFD) specialists in the world (Roe, van Leer et al in the
USA; Deconinck, Koren et al in Europe). In particular, in a first
effort to achieve a genuinely multidimensional approach, the present
research group has developed, within the framework of two research
projects supported by the European Community, a very efficient
local-adaptive multigrid method based on a simple-wave decomposition of
the Euler equations. Such a decomposition leads to the solution of an
equivalent set of scalar advection equations, suitable for the
application of accurate upwind FS discretizations and very efficient
time integration schemes based on optimal smoothers and multigrid.
Unfortunately, such a methodology, although robust and very efficient,
is limited to first-order accuracy and is thus inadequate for viscous
flow calculations. Therefore, a characteristic decomposition of the
Euler equations has been considered to develop a more accurate
genuinely multidimensional methodology for compressible flows. On the
other hand, the coupling terms in the compatibility equations render
the application of standard FS schemes less appropriate than in the
case of pure scalar advection equations and the design of optimal
smoothers impossible. In fact, both stability and accuracy problems may
arise from a straightforward extension of such schemes to scalar
equations with source terms. Therefore, a new class of matrix FS
schemes has been introduced, by generalizing the procedure proposed for
scalar advection equations to the discretization of systems with non
commuting Jacobians. Following this path, an accurate and robust method
for subsonic, transonic and supersonic flows has been recently
proposed, which has the only drawback of depending on a parameter which
is in general function of the solution. Therefore, such a method is, in
our opinion, a likely candidate to be the future state-of-the-art
method for solving the compressible Navier-Stokes equations. Developing
and validating such a method in two and three dimensions, as described
in the following, is the aim of the research proposed by the present
group. The present RL2 aims at developing a matrix Fluctuation
Splitting (FS) method for the Navier-Stokes equations which is accurate
and robust for all turbomachinery flow regimes (subsonic, transonic,
supersonic). Recently, two linear matrix FS schemes, based on a
characteristic decomposition of the Euler equations, have been
introduced: the monotone and robust matrix N scheme, which is
first-order accurate and satisfies the (local) positivity (P) property,
and the accurate matrix LDA scheme, which satisfies the linearity
preservation (LP) property. The goal of designing a nonlinear scheme,
satisfying both the P and LP properties, is a formidable task since no
TVD scheme can exist for hyperbolic systems in more than one space
dimension, energy being the only functional known to be bounded. In a
previous effort to circumvent such a paramount difficulty a nonlinear
scheme has been proposed which is already at the state of the art for
subsonic, transonic and supersonic flows, but has the only drawback of
depending on a parameter which needs to be tuned. Two strategies are
considered in order to design an improved scheme. The first one
consists of the combination of two already existing linear matrix
schemes: the N scheme and the LDA scheme. The former is only first
order accurate and is capable of providing monotone solutions whereas
the latter is second order accurate, thus it is suitable for computing
viscous flows but it produces spurious oscillations in high-gradient
regions. Such schemes will be combined according to a non-linear
strategy based on the control of the local extrema of the solution. The
second strategy consists of the developement of a nonlinear matrix FS
discretization obtained by generalizing the optimal scalar PSI scheme
to systems so that the resulting nonlinear matrix scheme be energy
bounded. Firstly the two methods will be developed in two dimensions
and will be employed for computing well documented turbomachinery flow,
such as rotor and stator cascade flows. In particular, accuracy and
robustness of the new methodology will be validated versus experimental
data and numerical solutions obtained with the state-of-the-art
methods, widely employed today for solving such problems. In particular
it will be assessed if it is worth to extend both schemes to 3D or only
one of them. Also viscous flow computations will be performed, the
viscous terms in the Navier-Stokes equations being evaluated by a
standard Finite Element discretization, which has already proven fully
satisfactory. It will be shown that the proposed methodology is
superior to current state of the art methods (Roe's FDS method) with
respect to both accuracy and robustness. Furthermore, the turbulence
modelling will be improved by considering two-equation turbulence model
of the k-epsilon and k-omega type. The methodology will be employed to
compute well documented three-dimensional flows of aerodynamic and
turbomachinery interest.
Financial support
| Amount (ML) |
40 |
| Source(s) |
COFIN 99 |
Research line n.3
Group leader
| Surname |
CATALANO |
| Name |
LUCIANO |
| University |
Politecnico di BARI |
| Faculty |
Engineering |
| Affiliation |
DIMeG |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
CATALANO |
LUCIANO |
Pol. Bari |
DIMeG |
1 |
4 |
4 |
| 2. |
MANODORO |
DARIO |
Pol. Bari |
DIMeG |
1 |
6 |
6 |
| 3. |
PETRUZZELLI |
NICOLA |
Southampton |
DIM |
1 |
2 |
2 |
Five recent papers
Catalano L. A., Dadone A., "Progressive optimization for the design of 2D transonic cascades", paper IS-066, 14th ISABE, 1999.
Catalano L. A., Dadone A., "Progressive optimization for the inverse design of 2D cascades", AIAA Paper 2000-3204.
Dadone A., Petruzzelli N., Mohammadi B., "3d aerodynamic shape design
using Hessians based on incomplete sensitivities", accepted for
publication by IJCFD.
Medic G., Mohammadi B., Petruzzelli N., Stanciu M., "3D optimal shape
design for complex flows: application to turbomachinery", AIAA Paper
99-0130.
Catalano L. A., Dadone A., Manodoro D., Saponaro A., "Sviluppo di
post-bruciatori per impianti combinati: parte 1a: confronto tra
indagine numerica e sperimentale; parte 2a: progettazione automatica
mediante ottimizzazione progressiva", 55mo ATI, 2000.
Research activity
The purpose of the
research is to devise accurate, very efficient, and robust technologies
for the automatic design of fluid-dynamic components. In particular,
the efforts will be concentrated on developing one or more technologies
capable of analyzing a wide range of geometries in a computational time
comparable to that required by the analysis of one configuration,
regardless of the number of parameters used. The fulfillment of these
two points would make the proposed optimization technology an efficient
and practical tool for the inverse and the direct design of
fluid-dynamic components of industrial interest, also allowing the
investigation of non-standard configurations, which could offer larger
improvements with respect to the existing ones. The proposed research
includes investigations and developments on both stochastic and
gradient-based techniques, using, as starting point, three research
codes for the inversedesign of 2D fluid-dynamic components, already
developed (and/or under development) by the researchers of the staff.
In the first year, only the second approach will be considered. In
particular, the activity will concentrate in the development of a
progressive optimization strategy, for the direct design of 2D
airfoils, turbomachinery blades and for the inverse design of
industrial after-burners. The proposed approach aims at generating an
optimal configurationin in a time comparable to that required by one
single flow analysis. The proposed optimization technique is based on
five ingredients. First, the use of an appropriate basic flow solver
allows to compute all important features of the flow field, and thus
the objective function, very accurately. Second, approximate but
efficient design sensitivities are obtained using a discrete adjoint
formulation: the adjoint problem is based on an auxiliary flow solver
with a proper level of artificial viscosity, which smooths the
sensitivity derivatives, in presence of shocks and numerical noise. The
third ingredient consists of the addition of a time-derivative in the
adjoint equations, which allows the use of a robust iterative solver.
The fourth element relies on the idea of bringing to convergence
simultaneously all of the iterative solutions involved in the entire
algorithm: the strategy performs a sequence of operations, consisting
of a partially converged flow solution, followed by a partially
converged adjoint solution, followed by an optimization step. The fifth
ingredient is the use of progressively finer grids for the progressive
solution of the flow field.
It is noteworthy that the smoothing introduced by the auxiliary flow
solver also allows to solve the adjoint equations always very cheaply
on the coarsest mesh, without a significant loss of accuracy. The
proposed optimization technique will be tested versus the inverse
and/or the direct design of 2D airfoils, turbomachinery blades and
industrial after-burners. In the latter case, a commercial CFD code
will be used, which does not allow a complete interface with the
optimization code: clearly, this will obstruct and/or prevent the use
of some of the ingredients cited above.
In the second year of activity, the progressive optimization technology
for the design of airfoils and turbomachinery blades, which will be
more mature at this stage, will be extended to three dimensions, with
application to the inverse design of turbomachinery blades in inviscid
transonic flow conditions. Part of this activity will be devoted to the
determination of a set of parameters able to well represent standard as
well as non-standard blade profiles. Moreover, the progressive
optimization code developed in the first year for the inverse design of
a 2D industrial after-burner will be extended to the direct design of
this component. Also in this case, part of the activity will be devoted
to the determination of a set of parameters able to well represent
standard as well as non-standard configurations. Finally, a stochastic
optimization technique, based on genetic algorithms, will be partially
developed, with possible application to one or more of the 2D
fluid-dynamic components cited above.
In the third year of activity, the development of the stochastic
optimization technique will be completed, with application to the
inverse and the direct design of one or more of the 2D fluid-dynamic
components cited above. Moreover, two major improvements are expected:
first, the gradient-based, progressive optimization strategy will be
extended to the direct design of 3D turbomachinery blades in inviscid
transonic flow conditions, to demonstrate its maturity for complex 3D
design applications; second, the combination of the stochastic and the
gradient-based techniques will be considered, with application to the
inverse and the direct design of 2D airfoils and/or turbomachinery
blades: the first technique must be able to explore innovative and
unusual configurations and to locate globally optimal solutions, which
will be used as starting points for the final, gradient-based, process
of refinement of the optimal configuration. Concerning the efficiency
of this combined, sequential approach, the large number of flow
analyses required by the stochastic method would make this approach
unaffordable for complex applications, where the flow solution is very
computationally demanding. For such a reason, a multi-level strategy,
consisting of the use of different levels of accuracy of the analysis
tools involved in the design process, will be investigated, so as to
increase the efficiency of the overall optimization process. The idea
is to combine lower (but cheaper) and higher fidelity models in the
optimization process, so as to gain an overall reduction in the
computational time, compared with a strategy always invoking the
highest fidelity model. Several lower accurate approaches to the flow
computations will be considered, including the use of coarse grids
and/or simplified flow models.
Financial support
| Amount (ML) |
16 |
| Source(s) |
University research funds and contracts |
Research line n.4
Group leader
| Surname |
PASCAZIO |
| Name |
GIUSEPPE |
| University |
Politecnico di BARI |
| Faculty |
Engineering |
| Affiliation |
DIMeG |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
PASCAZIO |
GIUSEPPE |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
| 2. |
CAMPOBASSO |
SERGIO |
Pol. Bari and Oxford |
DIMeG |
1 |
3 |
3 |
| 3. |
DE PALMA |
PIETRO |
Pol. Bari |
DIMeG |
1 |
2 |
2 |
| 4. |
NAPOLITANO |
MICHELE |
Pol. Bari |
DIMeG |
1 |
1 |
1 |
| 5. |
CINNELLA |
PAOLA |
Pol. Bari |
DIMeG |
1 |
4 |
4 |
| 6. |
LAMBERTI |
LUCIANO |
Pol. Bari |
DIMeG |
1 |
2 |
2 |
Five recent papers
M. Giles, "An approach for multi-Stage Calculations Incorporating Unsteadiness", ASME Paper 92-GT-292.
S. Campobasso and M. Giles, "Analysis of the effect of mistuning on Turbomachinery Aeroelasticity", to be published.
Research activity
Flutter is an
aeroelastic instability, which occurs when the forces acting on a
structure give rise to rapidly growing free oscillations. This happens
if the energy transfered by the working fluid to the structure is
greater than the energy dissipated by the friction forces and may lead
to mechanical failure if the stress corresponding to these deformations
exceeds the material strength.
In the field of gas turbines for industrial and aeronautical
application, flutter is well known, since it often occurs in bladed
rotors with high aspect ratio blades (fans of aeronautical engines and
last stages of low pressure turbine of both aeronautical and industrial
engines). The occurrence of flutter in service may lead to the loss of
the whole engine and its analysis in the design phase is often carried
out with empirical correlations, which generally lead to weight
increments and performance penalties in the final design.
The main goal of the proposed research is the development of efficient
and accurate numerical techniques for the analysis of flutter and for
the aeromechanical design of the blading, which can be integrated in
the design system of turbomachinery industries.
An accurate numerical definition of the flutter margin of bladed rotors
would require the solution of the coupled unsteady equations for both
the structure and the fluid (RANS equations) over the computational
domain consisting of the whole disc, all blades and passages between
adjacent blades. This procedure is computationally very expensive, as
it requires CPU times of the order of days and therefore it cannot be
introduced into everyday design practice. Nevertheless, one of the
present research activities will be the development of a very efficient
RANS solver capable of performing time-accurate solutions on vibrating
bodies within hours of CPU time on a 32 processor parallel machine.
An alternative approach for the analysis of flutter in turbomachines is
the linear method. Its implementation is a fundamental part of the
proposed research program and it is based (i) on the hypothesis of
cyclic symmetry of the bladed rotor, (ii) on the fact that the
frequencies of the structural modes are well apart each from the other
and flutter typically involves only one of such modes, (iii) on the
fact that the aerodynamic forces are much smaller than the elastic and
inertial ones and the difference between structural modes in vacuo and
aeroelastic modes is negligible and (iv) on the evidence that the
amplitude of the vibration is small in conditions of incipient flutter
and and therefore the system behaves linearly. The hypothesis (i)
allows one to restrict the analysis to the computational domain
consisting of a single blade, of the corresponding disc sector and the
passage in which the blade works, making use of a proper cyclic
boundary condition on the radial boundaries. The evidences (ii) and
(iii) lead to a substantial reduction of the structural analysis, since
the vibrating motion of the assembly is assumed equal to the in vacuo
natural mode whose stability is being assessed (for example first
bending). Finally, the observation (iv) legitimates the linearisation
of the unsteady RANS equations about the steady state and the
assumption of periodic unsteadiness.
The flow variables are expanded in Fourier series and the linearised
equations are solved for each harmonic, the frequency of the base one
being that of the structural mode under investigation. Once the forces
acting on the blades have being determined, the work done for the
prescribed displacements can be computed and its sign determines the
aeroelastic stability of the bladed rotor.
The linear method requires considerably less computing resources than
the solution of the coupled structure and fluid unsteady equations and
therefore is well suited for everyday design practice in industry.
Additionally this technique can be easily extended to the optimisation
of the aeromechanical design of the blading, aimed at maximising the
aeroelastic stability. This can be achieved modifying the shape of the
blade, while fulfilling mechanical and aerodynamical constraints like
material strength and aerodynamic performance.
The milestones of this research program are the implementation of the
linear method for the analysis of flutter, its use for improving the
knowledge on turbomachinery aeroelasticity and its extension to the
optimisation of the aeromechanical design.
This work will be carried out in collaboration with Rolls-Royce plc, a
worldwide leading industry of aircraft engines, which will actively
contribute to the validation of the developed codes and will include
them in their multi-disciplinary design systems.
Financial support
Workpackage 2
Title
|
Computational Mechanics of Solids, Structures and Mechanisms |
Background
In the DIMeG there are
various research teams working on mechanics of solids, structures and
mechanisms. The main investigators of the DIMeG in these fields
involved in the CE project are Prof. Giuseppe Demelio, working in solid
mechanics and machine design, Prof. Luigi Mangialardi, working in
applied mechanics, Prof. Luigi Tricarico, working in the field of
manufacturing.
There are solid links with the "Dipartimento di Disegno Tecnico
Industriale e della Rappresentazione" of the Polytechnic of Bari, where
Prof. Giuseppe Monno works with his research team, with the "Universitā
della Basilicata", in particular with Prof. Giacomo Mantriota, working
in applied mechanics and with the computational mechanics team of the
CNR-IRIS Institute in Bari, in particular with Dr. Michele Ciavarella.
All the research groups are working on challenging problems with good
results in terms of advance of the knowledge and engineering
applications.
There is a strong interaction with local and national industry, i.e. Calabrese Engineering, MERMEC, Piaggio, BOSCH, GETRAG.
The solid mechanics and machine design group of the DIMEG developed its
researches in collaboration with the CNR-IRIS computational mechanics
team, working in various challenging problems and becoming a reference
group at very high level in the contact mechanics and fretting fatigue
fields. In particular, in contact mechanics, various analytical and
semi-analytical techniques have been developed in order to solve
non-Hertzian contact problem with friction. The analytical solution of
the 2-dimensional contact problem of second order spline profiles has
been obtained and, using the extension of the Cattaneo problem to non
Herztian problem, developed by Ciavarella, has been extended to the
partial slip problem. The solutions obtained have been used for a
preliminary analysis of results obtained in fretting fatigue tests and
available in the literature. On the basis of the knowledge developed,
some suggestions on the pads geometry and on the improvement of the
experimental techniques in fretting fatigue have been given and the
set-up of an original test rig is actually in progress.
The effect of surface roughness in contact mechanics has been pointed
out in the studies concerning the properties of the solution of the
elastic contact of Weierstrass profile with a flat surface adopting a
multiscale approach.
The Applied Mechanics research group, in the last years, following the
general recommendation of using efficient, reliable and economical
technologies in wind power systems, investigated whether a V-belt CVT
with automatically adjustable speed was compatible with a wind power
system for electric energy production and in waterpumping windmills.
The automatic CVT option is a simple and economical solution. With an
appropriate design it may control the velocity ratio between the wind
turbine and the electric generator thus optimising the efficiency of
the whole wind power system with an increment that can reach the 50%.
The previous results suggested to investigate closely the CVT
transmission by means an experimental approach. A test rig had been set
up to study the experimental behaviour of an automatic CVT equipped
with a rubber V-belt and two spring-loaded pulleys, one at the driving
shaft and one at the driven shaft. The results had been compared with
the theoretical estimations.
The optimal results obtained up to this time in stationary conditions,
suggest to examine further the dynamical behaviour of CVT but now in
transient condition, namely, during the speed ratio changing. Our
research unit has undertaken this work since few months and intends to
carry on this project by means a theoretical and an experimental
approach.
The main research activity of the Industrial Design group is concerned
with the techniques for the "transformation" of data into meaningful,
computer-generated images. Application areas for visualization
technology are, for example, the analysis of data encountered in
computational fluid dynamics, computational analysis, digital imaging,
or the simulation of complex physical processes.
Scientific visualization is becoming an increasingly important tool in
engineering, medicine, and computational science. One of the main
reasons for such a growing interest is that acquisition devices and
computational techniques producing large volumetric datasets are
becoming more and more common.
Data analysis for complex dataset is currently done mostly by analyzing
static images of 2D slices. Users are thus forced to mentally combine
multiple views for constructing a mental model of 3D shapes, adding
further complexity to what is often an already difficult task. The
creation of a virtual environment for the direct analysis in three
dimensions of volumetric data overcomes these problems. To achieve this
goal, advances are necessary both in data visualization algorithms and
in the techniques used for interacting with the volume.
The team that works on Technology and Manufacturing Systems has an
intensive scientific activity in the following fields: Selection of
materials, treatment and production processes, simulation techniques
for design and optimisation of production systems, application of
artificial intelligence to automated process planning, CAD/CAM
integration, analytical and numerical techniques for optimisation and
design of manufacturing processes such as welding, forming, non
traditional manufacturing processes (laser welding and heat treatments;
electron beam).
With reference to the research line presented in this project, the team
works in the problem of the formability evaluation in sheet metal
forming and in laser welding technology. The approach followed in the
first activity combines Finite Element Method, Design of Experiments
and neural networks techniques, experimental test. Within CNR, MURST,
and industrial projects, equipment have been designed and realised to
predict the material Forming Limit Curves (FLC) with Swift and Nakazima
test, to perform conventional, hydro-mechanical and hydro-forming deep
drawing. Using commercial Finite Element code (Ansys, Dyna3d, Abaqus)
some FE models have been developed for the FLC prediction thought the
numerical simulation of standard tensile test, Swift and Nakazima test.
For the determination of the plastic instability actually the team is
exploring the use of ductile damage models inside the FE method.
As concern laser welding technology, the team is collaborating with the
Centro Ricerche Fiat - TO, investigating trough numerical and
experimental approaches, the weld quality of aluminium alloys using
CO2, NdYag and high power Diode Laser.
Objectives
The main objective of
the second workpackage is to press forward the state of the art in
computational mechanics of solid, structures and mechanisms, sheet
metal forming, to integrate and to share the acquired knowledge in
computational mechanics in order to offer to the local industry a set
of competencies and services oriented to the improvement of the product
quality and reliability in all the engineering aspects, starting from
the design phase up to the manufacturing one.
The main goals pursued by the Solid Mechanics and Machine Design Group can be summarized in:
1) the integration of numerical and analytical techniques to allow the
opportunity to use commercial FEM codes for the evaluation of global
structural behaviour and the calculation of the forces globally
exchanged in the contact zones and analytical or semi-analytical method
for the evaluation in detail of the contact stresses;
2) the analysis and the comparison of the results obtained using
different fretting fatigue test rigs, in order to assess the more
reliable damage parameters and to obtained fretting fatigue design
charts.
3) the evaluation of the reliability of the different methods used for generating thermal barrier coatings.
Regarding the Applied Mechanics Group, till now the studies on CVT
transmission have been conducted considering an almost steady state
operative conditions, hence it was possible to utilise the stationary
model of CVT also in transient conditions. Naturally this approach
gives good results as long as the physical quantities, that influence
the dynamical behaviour of transmission, change slowly during the time.
Actually, in order to avoid the possibility of global sliding between
the belt and the pulley, the axial thrust on the half-pulley are, in
some operative conditions, too large with negative effects on the
transmission's efficiency.
These considerations suggest to further investigate the mechanics of
CVT in transient condition with the aim to choose the better solutions
and particularly the better axial thrust in order to improve the
transmission's efficiency. If these results will be achieved then it
will be possible to propose this transmission in industrial and
automotive fields in order to improve the energy utilization and to
reduce the environmental impact of energy usage: for example the
vehicles' fuel consumption performance could improve as well as
pollutions could reduce.
Consequently the most important phases, which the research unit will face, will be the following:
1. Development of a first simplified theoretical model of the shifting dynamics of the CVT transmission.
2. Improvement of the previous model by considering the influence of some parameters which the former model neglects.
3. Theoretical evaluation of the optimal axial thrusts on the moving
half-pulleys in order to improve the CVT's performance in transient
conditions.
From the point of view of the Industrial Design Group, the full
potential of Scientific Visualization is exploited when allowing users
to view and interact in real-time with the data in a natural way. The
ability to render perspective views of a meaningful dataset at
interactive speeds dramatically improves final user perception, thanks
to motion parallax effects, makes it possible to use direct interaction
techniques to naturally perform complex 3D tasks, such as finding the
best viewpoint or positioning cutting planes, and opens the door to the
usage of data rendering in visual simulation applications for training
and testing.
For those reasons this research unit will focus its attention on the following points:
1) Evaluate the state-of-the-art and develop new interaction techniques
for data analysis and manipulation in Virtual Environments.
2) Adapt and improve data visualization algorithms and techniques for
defined applications to provide skills and solutions useful to the
other units.
The main objectives pursued by the Technology and Manufacturing system
group is to study the formability of innovative materials and sheet
metal forming processes. On the base of experimental activities in
process in the DIMEG Lab and in collaboration with the steel maker Ilva
- Laminati Piani TA (Italy) and the research centre Centro Ricerche
Fiat - TO (Italy), the objectives of the team are: (i) Finite Element
implementation of damage mechanism in sheet metal forming to predict
plastic instability; (ii) Study with Finite Element approach of the
Tailored Welded Blank formability; (iii) Investigation of the Tailored
Welded Blank formability in hydro-mechanical and hydro-forming processes
Description of the activities
Research line n.1
Group leader
| Surname |
DEMELIO |
| Name |
GIUSEPPE |
| University |
Politecnico di BARI |
| Faculty |
ENGINEERING |
| Affiliation |
DIMeG |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
DEMELIO |
GIUSEPPE |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
| 2. |
CIAVARELLA |
MICHELE |
CNR-Bari |
IRIS |
1 |
2 |
2 |
| 3. |
DECUZZI |
PAOLO |
Pol. Bari |
DIMeG |
1 |
2 |
2 |
| 4. |
PALMISANO |
MARIO |
Pol. Bari |
DIMeG |
1 |
4 |
4 |
| 5. |
LAMBERTI |
LUCIANO |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
Five recent papers
1. Ciavarella M,
Demelio, G., Barber, JR, Jang, YH, The contact of fractal rough
surfaces, Proceedings of the Royal Society, 2000, 456, 387-405
2. Ciavarella, M, Demelio, G, On the extraction of notch stress
intensity factors (N-SIFs) by post-processing of stress data on the
free-edges of the notch, J. Strain Analysis, vol. 35, part 3, 221-225,
2000
3. Ciavarella M, Demelio G, On non-symmetrical plane contacts, Int. J. of Mech. Sci., 41, 1999, 1533-1550
4. Ciavarella, M, Decuzzi, P, Demelio, P, Monno, G, Hills, DA, The
design of hydrodynamically lubricated Journal bearings against yield,
J. Of Strain Analysis, vol.34 no 3, 1999, 165-173
5. Ciavarella M, Demelio G, Hills DA, "The use of almost complete
Contacts for Fretting Fatigue Tests, Fatigue and Fracture Mechanics:
Twenty-Ninth Volume, ASTM STP 1332, T.L. Panontin and S.D. Sheppard,
Eds., American Society for Testing and Materials, West Conshohocken,
PA, 1999., 696-709
Research activity
Fretting and thermal fatigue in gas turbine components.
Two important problems affecting the reliability and the efficiency of
gas turbines engines are fretting fatigue and thermal fatigue.
Fretting Fatigue is a phenomenon of fundamental importance for the
reliability of machinery and mechanical structures, and results at
present one of the strategic areas for aeronautical turbine engines.
Fretting manifests itself especially in the region of connections (e.g.
the dovetail or fir-tree region, or riveted joints) undergoing
mechanical contact where varying normal and tangential loads are
present. Within the region of partial sliding of the contact, an
increase of friction coefficient is found and a rapid nucleation of
fatigue cracks, leading to a abrupt reduction of safe life of the
component (up to 90%). At present, several theoretical models have been
proposed by different researchers in the area for the evaluation of
fatigue life under such conditions, but it is necessary to conduct ad
hoc experimental investigations, now under progress in prestigious
research centres in USA some of which in collaboration with us, while
we are developing efficient numerical methods for studying frictional
non-Hertzian contact.
All the theoretical models require a detailed evaluation of the normal
and tangential contact stresses shared at the contact interface and,
subsequently, of the internal stress. The degree of accuracy required
is not easily available with commercial Finite Elements codes, due to
the mismatch between the mesh to be used for modelling large structural
components and the mesh required for modelling localised contacts.
Moreover, the dependency of solution upon the load history leads to a
sensible increment of the number of iterations required in the solution
scheme and then of the amount of the run times, which are often
incompatible with the analysis of complete stress history. The aim of
the research to be performed is to integrate the computational
facilities offered by the new Centre with the analytical methods
developed and in progress.
While the fretting fatigue affects the joints between the turbine
blades and the rotor disks, the thermal fatigue affects directly the
turbine blades.
In order to increase the efficiency, it is needed to increase
temperatures in the first few stages of gas turbine engines. This
requires, however, materials of characteristics more and more
demanding, so that recently, several kind of coatings are under
development giving locally properties of thermal insulation, without
having the low resilience characteristics of the monolithic ceramic
material. The proposed activity would concern the study of thermal
fatigue behaviour of such coatings, which can have several kind of
problems such as early detachment and microcracking, due to residual
stresses arising because of differential dilatation.
From the computational point of view, a proper study of the stress
fields in real components subjected to thermal fatigue requires a
relevant amount of computational resources. It is necessary to solve a
coupled thermal-stress analysis problem and to simulate singular stress
fields.
In both the problems the visualisation of the results is an open issue
due to the complex geometry of the connections and the overlapping
fields to show (e.g. normal and tangential loads and stresses).
Financial support
| Amount (ML) |
20 |
| Source(s) |
Research contracts |
Research line n.2
Group leader
| Surname |
CIAVARELLA |
| Name |
MICHELE |
| University |
Politecnico di BARI |
| Faculty |
ENGINEERING |
| Affiliation |
CNR-IRIS, BARI |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
CIAVARELLA |
MICHELE |
CNR-Bari |
IRIS |
1 |
3 |
3 |
| 2. |
DECUZZI |
PAOLO |
Pol. Bari |
DIMeG |
1 |
2 |
2 |
| 3. |
PETRUZZELLI |
NICOLA |
Southampton |
DIM |
1 |
2 |
2 |
| 4. |
DEMELIO |
GIUSEPPE |
Pol. Bari |
DIMeG |
1 |
2 |
2 |
Five recent papers
1. Ciavarella, M,
Demelio, G, On the extraction of notch stress intensity factors
(N-SIFs) by post-processing of stress data on the free-edges of the
notch, J. Strain Analysis, vol. 35, part 3, 221-225, 2000
2. Ciavarella M, Demelio, G., Barber, JR, Jang, YH, The contact of
fractal rough surfaces, Proceedings of the Royal Society, 2000, 456,
387-405
3. Ciavarella M, Demelio G, On non-symmetrical plane contacts, Int. J. of Mech. Sci., 41, 1999, 1533-1550
4. Ciavarella, M, Decuzzi, P, Demelio, P, Monno, G, Hills, DA, The
design of hydrodynamically lubricated Journal bearings against yield,
J. Of Strain Analysis, vol.34 no 3, 1999, 165-173
5. Ciavarella, M, Demelio, G, Numerical methods for the optimisation of
specific sliding, stress concentration and fatigue life of gears, Int.
J. Fatigue, 21, 1999, 465-474
6. Ciavarella M, Demelio G, Hills DA, "The use of almost complete
Contacts for Fretting Fatigue Tests, Fatigue and Fracture Mechanics:
Twenty-Ninth Volume, ASTM STP 1332, T.L. Panontin and S.D. Sheppard,
Eds., American Society for Testing and Materials, West Conshohocken,
PA, 1999., 696-709
Research activity
Multi-level design of brakes/clutch/gearbox systems and other mechanical
components.
The purpose of the research program is to devise effective and
efficient optimization methodologies for the solution of
multidisciplinary design problems of engineering interest. It is
intended to investigate advanced optimizers based on stochastic and
slope driven methods coupled with multi-level strategies. The use of
hybrid techniques is required by an enhanced ability of modern
stochastic methods to locate globally optimal solutions and to explore
innovative and unusual configurations, whereas classical,
gradient-based methods prove to be more suitable in a final process of
refinement of the solution. On the other hand, stochastic search
methods tend to be profligate in terms of number of analyses required
to converge to an optimal solution. This becomes a crucial issue when
each analysis is computationally demanding. A possible approach to
address such a practical problem is the use of multi-level strategies.
Multi-level strategies consist of the use of different levels of
accuracy of the analysis tools involved in the design process. The
attempt is to combine lower and higher fidelity models to gain an
overall reduction in the computational time compared with always
invoking the highest fidelity model.
This multiresolution approach matches exactly the need-for-speed in the
visualization and rendering of the results and allows the final user to
interact with the models/results even in realtime in a virtual
environment like the CAVE or the WORKBENCH.
The application of the proposed optimisation techniques are strongly
oriented to create a link with multinational industry groups which have
local research and manufacturing centres, Bosch Sistemi Frenanti and
Getrag (gearbox).
In fact one of the field of application will be the design of brakes
and clutch systems. The growing demand for higher performances, the
recent introduction of new asbestos free friction materials and the new
environmental laws for noise reduction require more detailed analysis
of brakes and clutches in order to define new optimal design criteria.
Particular attention must be paid to the phenomenon of thermoelastic
instability which can appear in automotive and railway applications.
Such a phenomenon leads to in service reduction of contact area with
localization of load and heat over small regions of high pressure and
temperature (hot spots), causing degradation of the friction material,
and under severe conditions, even structural failure of the device
because of vibrations or thermal cracking. The main objective is to
determine the influence of design parameters, such as dimensions and
material's properties, on the critical sliding speed of the system in
order to define new optimal design criteria, which can be performed
efficiently only employing high-performance computers.
Also for the interactive visualization of the whole brake/clutch model
with the time-variant analysis results it is imperative to use
high-performance graphics workstations.
Financial support
Research line n.3
Group leader
| Surname |
MANGIALARDI |
| Name |
LUIGI |
| University |
Politecnico di BARI |
| Faculty |
ENGINEERING |
| Affiliation |
DIMeG |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
MANGIALARDI |
LUIGI |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
| 2. |
MANTRIOTA |
GIACOMO |
Basilicata |
DIFA |
1 |
2 |
4 |
| 3. |
CARBONE |
GIUSEPPE |
Pol. Bari |
DIMeG |
1 |
2 |
4 |
| 4. |
SORIA |
LEONARDO |
Pol. Bari |
DIMeG |
1 |
3 |
4 |
Five recent papers
Mangialardi L.,
Mantriota G.: Continuously Variable Transmission with Torque-Sensing
Regulators in Waterpumping Windmills . Renewable Energy. Vol. 7, No. 2,
pp. 807-823, 1994.
Mangialardi L., Mantriota G.: Automatically regulated C.V.T. in wind
power systems . Renewable Energy. Vol. 4, No. 3, pp. 299-310, 1994.
Mangialardi L., Mantriota G.: "Dynamic behaviour of wind power systems
equipped with automatically regulated Continuously Variable
Transmission.". Renewable Energy. Vol. 7, No. 2, pp. 185-203, 1996.
Mangialardi L., Mantriota G.: "Comments on "Maximum mechanical
efficiency infinitely variable transmissions". Mechanism and Machine
Theory. Vol. 33. No. 4, pp. 443-447, 1998.
Mangialardi L., Mantriota G.: "Power flows and efficiency in infinitely
variable transmissions". Mechanism and Machine Theory. Vol. 34, No. 7,
pp. 973-994, 1999.
Research activity
The rubber V-belt CVT
transmission cannot transmit power exceeding 10 kW in order to avoid
low reliability. Over this power threshold the metal V-belt is commonly
used. The dynamics of metal pushing V-belt CVT transmission was
extensively studied by several researchers, whose theoretical and
experimental works are concerned, especially, with steady - state CVT
dynamics. Today more and more attention is placed on CVT shifting
dynamics, but not many works concerns this subject. Most of the works
on CVT transient dynamics are only experimental works, moreover the
researchers have to overcome several difficulties in order to develop a
suitable theoretical model of CVT shifting dynamics.
The methodology, until now adopted, to consider an almost steady-state
evolution of the transmission during the shifting phases, is no more
acceptable in several practical cases. For example for a CVT
transmission installed on a vehicle it is not possible to assume almost
stationary the drive's behaviour. Moreover, recent studies have evinced
doubts on the possibility to correctly evaluate the power dissipation
in transient conditions by means of a steady state model of
transmission. Different authors have been investigating experimentally
the continuously variable transmission during speed ratio changes, but
no theoretical model exist, for this case, capable to foresee the CVT's
dynamical behaviour especially for what concerns the axial thrust on
the moving half-pulleys and the power dissipation.
The formulation of a theoretical model of metal pushing V-belt CVT
transmission in no steady-state conditions will be conducted, first,
without the effect of the deformability of the belt and pulley and of
the clearance among the metal plates. This first approach will be
compared with the experimental results in order to evaluate its
reliability. The expected result is to give to the designers a
simulation tool, easy to use, which allows a correct design of the
transmission.
Afterwards the clearance among the metal plates and the strain of the
belt and pulley will be considered in the model, and their effect on
the system response to a variation of torque, power and speed ratio,
will be calculated. Experimental tests will be made, by using a test
rig, which, at the moment we are developing in our Department, in order
to evaluate the influence of the several parameters on the dynamical
response of the system and different operative condition will be
simulated:
1. Acceleration phase with an increment of power request as it happens when overtaking another vehicle.
2. Increment of torque transmitted to driven pulley: the velocity will
be kept constant and the speed ratio will be changed as it happens in a
climbing manoeuvre.
Naturally all these phases will be numerically simulated by means the
theoretical models previously developed. We have to underline that the
complexity of the simulation model needs a great computational effort.
In effect the control strategy of the transmission has to link in a
unique simulation software the dynamical model of the transmission, the
engine features of the vehicle (represented by its engine torque-speed
map), the hydraulics of the transmission and the interaction model
between the vehicle and the environment which is needed in order to
evaluate the power demand.
At the end of the work the simulation software will be experimentally
checked and will be utilised to evaluate the optimal solutions
necessary to reduce the power dissipation and the response delay.
Financial support
Research line n.4
Group leader
| Surname |
MONNO |
| Name |
GIUSEPPE |
| University |
Politecnico di BARI |
| Faculty |
ENGINEERING |
| Affiliation |
DIP. DI DISEGNO TECNICO INDUSTRIALE E DELLA RAPPRESENTAZIONE |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
MONNO |
GIUSEPPE |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
| 2. |
UVA |
ANTONIO |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
| 3. |
FIORENTINO |
MICHELE |
Pol. Bari |
DIMeG |
1 |
3 |
3 |
| 4. |
DELLISANTI |
MASSIMILIANO |
Pol. Bari |
DEE |
1 |
2 |
2 |
Five recent papers
1. Bjoern Heckel,
Antonio E. Uva, Bernd Hamann and Kenneth I. Joy, "Surface
Reconstruction using Adaptive Clustering Methods," in Guido Brunnett,
Hanspeter Bieri and Gerald Farin, eds., Geometric Modeling, Supplement
to the Journal Computing, (to appear), 2001
2. Kuester, F., Duchaineau, M.A., Hamann, B., Joy, K.I. and Uva, A.E.,
(1999), 3DIVS: 3-dimensional immersive virtual sculpting, in: Ebert,
D.S. and Shaw, C.D., eds., Workshop on New Paradigms in Information
Visualization and Manipulation (NPIV '99), ACM Press, New York, New
York, pp.~xxx--xxx
3. Kuester, F, Uva, A, Hamann, B, Monno, G.:"3DIVS: 3-DIMENSIONAL
IMMERSIVE VIRTUAL SKETCHING", Proceedings of the 12th International
Conference on Engineering Design, ICED 99, pp. 1407-1412, Munich,
Germany, august 1999.
4. Hamann, B, Kreylos, O., Monno, G., Uva, A.: " Optimal Linear Spline
Approximation of Digitized Models", Proceedings of the 1999 IEEE
Intermational Conference on Information Visualization, CAGD'99,
Computer Aided Geometric Design Symposium, pp. 244-249, London,
England, july 1999.
5. Heckel, B., Uva, A. and Hamann, B. (1998), Clustering-based
generation of hierarchical surface models, in: Wittenbrink, C. M. and
Varshney, A., eds., Visualization '98, Proc. Late Breaking Hot Topics,
IEEE Computer Society Press, Los Alamitos, California, pp. 41-44
presented at: Visualization '98, Research Triangle Park, North
Carolina, October 1998.
Research activity
Interaction Techniques for data analysis and manipulation in Virtual Environments.
The starting point is to define and establish a correct
hardware/software framework setup for the visualization of data
produced by engineering analysis.
In the last years Prof. Monno and Doct. Uva set up a multi-platform
virtual reality framework for Computer Aided Design applications.
Starting from this base we want to built a new virtual environment
specifically designed to work with a new generation of stereo
projection systems currently marketed under names like
ImmersiveWorkbench, ResponsiveWorkbench and ImmersaDesk. Those systems
are drafting table-style, projected displays that enable researchers to
view and manipulate 3-D models in real time. The images appear to have
a real presence in space when viewed through stereoscopic shutter
glasses. The user integrates head-tracking for multiple points of view
and uses a set of pinch gloves combined with a stylus device for
interaction with the VE. The spatial data describing the user's head
position and hand movements is fully incorporated into the VE.
Manipulation of objects in virtual environments is often awkward and
inconvenient. A lack of a tactile feedback, tracker noise, poor design
of interaction techniques, and other factors can make the simple task
of grabbing and moving a virtual object a frustrating experience.
Numerous studies have focused on how humans manipulate objects in the
real world and how tools, workplaces, etc., should be designed to
achieve more effective scientific data visualization and manipulation,
but still significant issues have to be studied and understood in order
to exploit the full potential of VR technology. Finding the right kinds
of interaction metaphors by which VR devices can be used to interact
with scientific datasets, however, requires experimentation with many
of the various possibilities. This is one of the main goals of this
research which, we believe, can lead to an application-oriented, but
high-quality, form of virtual reality, giving a compelling sense of
immersion but at the same time be expressive and ergonomic enough for
an actual use.
Data visualization algorithms and techniques.
For particular types of datasets like the ones produced by the
computational efforts of the research groups participating in the
Center "ad hoc" analysis techniques and real-time visualization
algorithms (based on multiresolution) have to be investigated.
For example, our research involves the development of an application
for the visualization of CFD (Computational Fluid Dynamics) data in the
Virtual Environment. The application will be designed based on existing
visualization software packages but evaluating new paradigms. Users
will be able to investigate the flow by placing streamlines, cutting
planes, and rakes the flow field by using natural hand movement. Future
work will involve computing approximate scalar and vector properties to
visualize changes in the flow field with geometry and time. This will
facilitate better understanding of the fluid flow and speed up the
design process.
On the other side, the techniques to be developed will provide a smooth
level-of-detail control and aims at guaranteeing a uniform, bounded
frame rate even for widely changing viewing conditions. The
optimization algorithms will be independent from the particular data
structure used to represent multiresolution meshes and results.
Financial support
| Amount (ML) |
20 |
| Source(s) |
CNR |
Research line n.5
Group leader
| Surname |
TRICARICO |
| Name |
LUIGI |
| University |
Politecnico di BARI |
| Faculty |
ENGINEERING |
| Affiliation |
DIMEG |
Staff
| nš |
Surname |
Name |
University |
Affiliation |
Month/
man
2000 |
Month/
man
2001 |
Month/
man
2002 |
| 1. |
TRICARICO |
LUIGI |
Pol. Bari |
DIMeG |
1 |
4 |
4 |
| 2. |
SPINA |
ROBERTO |
Pol. Bari |
DIMeG |
1 |
3 |
4 |
| 3. |
PALUMBO |
GIANFRANCO |
Pol. Bari |
DIMeG |
1 |
4 |
3 |
| 4. |
GALANTUCCI |
LUIGI |
Pol. Bari |
DIMeG |
1 |
2 |
2 |
Five recent papers
1. L.M. GALANTUCCI, R.
SPINA, L. TRICARICO: "A quality evaluation method for laser welding of
Al alloys through neural networks", ANNALS OF THE CIRP, VOL. 50/1, PP.
131-134, 2000
2. M. DE COSMO, L. TRICARICO: "Analysis of an Industrial Net Shape
Forming Application through Numerical and Experimental Approach",
ADVANCED TECHNOLOGY OF PLASTICITY 1999, Edited by M.. Geiger, SPRINGER,
Berlin, VOL. 2, pp. 759-764, 1999
3. CHEP, L. TRICARICO: "Object-Oriented analysis and design of a
manufacturing feature representation", INTERNATION JOURNAL OF
PRODUCTION RESEARCH, VOL. 37, NR. 10, PP. 2349-2376, 1999
4. M. DE COSMO, L.M. GALANTUCCI, L. TRICARICO: "Design of process
parameters for dual phase steel production with strip rolling using the
finite element method", JOURNAL OF MATERIALS PROCESSING TECHNOLOGY,
ELSEVIER SCIENCE, VOL. 92 - 93, pp. 486 - 493, 1999
5. L.M. GALANTUCCI, L. TRICARICO: "Thermo-mechanical simulation of a
rolling process with a FEM approach", JOURNAL OF MATERIALS PROCESSING
TECHNOLOGY, ELSEVIER SCIENCE, VOL. 92 - 93, pp. 494 - 501, 1999
Research activity
Study of the Tailored-Welded blank formability through the finite-element implementation of ductile damage models
In the last decades the attention towards sheet metal materials and
technologies has continuously increased. Through the introduction of
innovative materials into the market place have been obtained products
with enhanced properties; new technologies like Tailor Welded Blank
(TWB) and Hydro forming have been adopted to satisfy the increasing
demand of quality customized products with low costs.
Nowadays the use of Computer Aided (CA) technologies can represent an
important means to reduce the development cycle of a new product,
optimising its process parameters and improving its quality. The CA
techniques are very useful in the sheet metal industry to simulate
forming operations, design equipments, plan the operation sequence and
estimate the product cost.
The great contributions obtained through Finite Element Analysis (FEA)
are oriented to the study of the product behaviour during process and
service. Using these systems, a deeper analysis of stress and strain
distributions in tools and sheets can be performed to predict process
forces and evaluate the presence of defects on the products. In fact
numerical simulation of a sheet forming process allows to study the
sheet formability at the early stage of development cycle, without
fabricating tool prototypes and so minimizing experimental test. A
simulation model able to predict the plastic instability during a
forming operation, can give information about the forming process
feasibility: this requires a criterion for the onset prediction of the
localised necking in the FEA.
With the aim to educate young researchers in the FEA of sheet metal
forming processes, this research is focused to: (i) implement ductile
damage models for the prediction of plastic instability; (ii) study the
formability of TWB; (iii) improve the formability of TWB with
hydro-forming processes.
1. Finite element implementation of damage models in sheet metal forming
In sheet metal forming processes, the blank is subjected to plastic
deformation that can overcome the necking onset, compromising the
component functionality. The study of load-displacement curves obtained
with standard tensile tests allows to characterize the material
behaviour of cold formed sheets using general constitutive laws; these
laws lose their validity in the plastic region after necking and are
unable to predict the plastic instability because they do not consider
damage mechanisms.
Metallographic studies evidence that the ductile damage is
characterised by mechanisms of void nucleation, void growth,
coalescence or micro-crack linking of neighbouring voids. An approach
for modelling these damage mechanisms is to formulate a new yield
function; in Tvergaard and Needleman model the yield function
contemplates the void evolution during the material deformation using a
set of micro-mechanical damage parameters.
For the identification of these parameters, in this step of the project
a macroscopic approach based on tensile tests and their finite element
simulation is proposed. Damage model is implemented in the FE
simulation while damage parameters are obtained using an inverse
characterization method.
Basing on previous researches, the proposed approach has been tested
for low carbon steel using Tvergaard and Needleman damage model and
ABAQUS solver. The aim of this project is to extend the analysis
towards different kind of materials (HSLA steels, aluminium alloys), FE
solvers and damage models.
A FEM model of the deep drawing process, which also implements the
damage parameters evaluated, will be successively created to evaluate
the limit drawing ratio (LDR). Numerical results will be compared with
experimental ones obtained performing deep drawing tests.
2. Study with a Finite Element approach of Tailored welded Blanks formability
A Tailored-Welded Blank is realized through the welding and the
subsequent forming of flat sheets with different local properties
(material, thickness, coating type). The approach is actually used for
TWB steel in automotive industries, replacing conventional production
phases realised by: (i) definition of initial blanks, (ii) individual
forming of blanks, (iii) spot welding of formed blanks.
TWB process, like classical stamping ones, agrees with the design need
of combining materials with antithetic behaviours (formability,
strength, corrosion resistance); in addition this technique increases
component quality and reduces manufacturing costs and scraps, because
decreases stamping equipment and eliminates downstream spot welding
operations. In spite of these potentialities, TWB process involves a
careful analysis of operations in order to: study interactions of
formability behaviors of the weld joint and sheets, make an accurate
design of the stamping equipment, identify the optimal setup for
welding (laser type, speed, power). Thus some crucial activities can be
the formability evaluation of a TWB and the optimal positioning of the
welding line.
The research objective is to analyse the laser weld influence on the
TWB formability for steel and aluminium alloy blanks. The mechanical
characterization of the weld line will be investigated through the
methodology specified in the first research step. In this phase the
experimental approach consists of standard tensile test on TWB
specimens with welded lines differently oriented with respect to the
maximum deformation direction. Successively FE models will be used to
simulate a deep drawing process on TWB and to identify the optimal weld
position and characteristics. FE models will be validate through deep
drawing tests.
3. Investigation of TWB formability in hydro-mechanical and hydro-forming processes.
The adoption of TWB present difficulties connected to their formability
and to the tooling equipment design. A hydro-forming process can
overcome these limitations, producing parts by fluid pressure instead
of traditional tools (punches, dies, holders). In this way, welded
blanks with different thickness can be formed in a simple way,
eliminating tool shape constraints by the adoption of a fluid. Other
advantages of a hydro-forming process are: (i) the increase of TWB
formability, (ii) improvement of surface quality due to the metal/fluid
interaction, (iii) reduction of forming steps on complex shapes (a
lower working time and a higher production rate), (iv) reduction of die
making costs. For a successful process and a good product surface
quality, the study of relations between formability and process
parameters is necessary to eliminate product defects. The process
parameters are: the chamber pressure and its variation law; the
friction between blanks and tools; tool profile radii and the
pre-bulging pressure.
The aim of this research phase is to analyse the improvement of the TWB
formability using a hydrostatic stress state. A FE model, implemented
with the ductile damage criteria, can be used to evaluate the
manufacturability and the possibility to obtain well formed product by
the variation of process parameters. A sensibility analysis of optimal
parameter set will be performed to evaluate the effect of blank
thickness, mechanical characteristics of welded joints, the position of
welding line, the pressure variation law, the deformation speed. The
numerical results will be then compared with experimental ones.
Financial support
| Amount (ML) |
20 |
| Source(s) |
Research Contracts |
4. High education activity of the first three years
PhD and other High Educational Programs
PhD programs
Program 1
Title
|
Ph.D. program in Mechanical Engineering (Dottorato di ricerca in Ingegneria Meccanica) |
Co-ordination
| Surname |
NAPOLITANO |
| Name |
MICHELE |
| University |
Politecnico di BARI |
| Faculty |
ENGINEERING |
| Affiliation |
DIMeG |
Description of activity
The Ph.D. program in
Mechanical Engineering of the Politecnico di Bari offers several
curricula to graduate students willing to pursue a Ph.D degree in
Mechanical Engineering. The curricula offered are: Computational Fluid
Dynamics; Fluid Dynamics of Turbomachinery; Applied Mechanics;
Mechanical Design; Virtual Reality.
Every year, there is a national competition open to all Italian and UE
graduates, offering 6 to 8 positions. Each graduate students winning
one such position is allowed to choose the curriculum he prefers and
starts a research activity under the guidance of one of the faculty
members involved in the program. The students also take advanced
courses to complete their engineering and scientific knowledge.
Many of the research lines proposed in this project fall within the
areas of this Ph.D. program. Therefore, most future students will be
involved in such research topics, which will be the subjects of their
theses. It is noteworthy that only half of the positions offer a full
scholarship, so that the Centre for Computational Mechanics could also
provide funds for additional scholarships.
More importantly, many of the faculty members participating to the
Centre have close ties with foreign universities offering Ph.D.
programs, so that the Centre could co-finance the scholarships of
outstanding students willing to pursue a Ph.D. program abroad, while
working at a research project involving both the foreign University and
the Centre.
N. of available positions for year
Program 2
Title
|
Ph.D. program in advanced production system (Dottorato di ricerca in Sistemi avanzati di produzione). |
Co-ordination
| Surname |
DIOGUARDI |
| Name |
GIANFRANCO |
| University |
Politecnico di BARI |
| Faculty |
Engineering |
| Affiliation |
DIMeG |
Description of activity
The Ph.D. program offers
several curricula in the areas of advanced manufacturing systems as
well as in Management engineering. It works similarly to the Ph.D.
program in Mechanical Engineering. Although only one of the research
lines proposed by the Centre is related to this Ph. D. program, it is
anticipated that one doctoral student will choose to work to such
research line.
N. of available positions for year
Other High Education Programs
Program 1
Title
Description of activity
The Politecnico di Bari
offers every years several post doctoral positions in the various
research areas pursued by its Departments. It is anticipated that
outstanding Ph.D. students, after obtaining their degrees, will compete
to obtain such positions and work in one of the researchh lines
proposed by the Centre. At present, two such positions have been
obtained by: Dr. A. Uva, who is involved with the research line on
Virtual Reality; Dr. P. Cinnella, who is involved with the research
lines on Advanced Numerical Methods for the RANS Equations and Flutter;
Dr. Lamberti, who is involved in advanced research on structural
optimization. Dr. Uva has spent more than two years at the University
of California at Davis, working on developing very advanced Virtual
Reality software. Dr. Paola Cinnella has a Ph.D. degree from the
Politecnico di Bari and another one from the ENSAM of Paris, France,
where she has developed a very accurate and efficient finite volume
method for the time-dependent Euler and RANS Equations, which will be
one of the basic computational tools to be applied in the research line
on flutter. Dr. Lamberti, as a specialist on structural optimization,
will provide a valuable contribution to the coupling of the structural
and fluid dynamics equations required by a non linear flutter analysis.
5. Interaction with industrial research and/or spin-off on public services
Many professors from the
DIMeG have close ties with local and national industries, such as
GETRAG and BOSCH. About half of the theses discussed by our mechanical
engineering students are actually projects carried out inside such
industries. Also, some industries are currently funding Ph.D.
scholarships in addition to those funded by the Politecnico and the UE.
The proposed Centre will be the ideal institution to make such a
cooperations more regular and more effective. Each Industry willing to
participate to the activities of the Centre will pay a yearly fee and
will have a representative in the Center Research Council, without
voting power. Such a Council will discuss and decide the research
directions which will be followed by the Centre. It is hoped that
research engineers from local and non-local industries will participate
to the Centre activities both in person and by computer connection.
As far as spin-off on public services is concerned, given the research
area of the Centre, it is difficult to anticipate any significant
spin-off in the short term. Nevertheless, the public administrations
(Comune, Regione, Provincia) will be regularly invited to the seminars
offered by the Centre, so as to immediately pinpoint possible common
interests and start up cooperative programs.
6. Scientific and administrative management of the Centre
The Centre will have a structure very similar to that of the interdepartment Centres of the Politecnico di Bari.
From the scientific viewpoint, the Centre will have a Director who, for
the first three years will be the Principal Investigator of the Centre.
The Centre will also have a Scientific Council made up of all
research-line leaders (with voting power) and industrial
representatives (without voting power).
Finally, the Centre will have a Managing Board, made by the Director
and three components of the Scientific Council, one being an industrial
representative.
After the funding of the Centre, the Scientific Council will propose a
Statute in accordance with the Italian laws and the rules of the
Politecnico di Bari, which will become operative after being approved
by the Academic Senate of the Politecnico.
From the administrative viewpoint, the Centre will have an
administrator, who has to be under the direct supervision of the
administrative secretary of the DIMeG. For the first three years, it is
anticipated that the administrator can be an administrative officer of
the DIMeG, working part time for the Centre. Needless to say, the
administration of the Centre will have to obey both the Italian laws
and the administrative and accounting rules of the Politecnico.
7. Justification of the costs
First year - Costs
| Durable equipment (ML) |
700
(361520 Euro) |
| Consumables (ML) |
50
(25823 Euro) |
| Computing (ML) |
110
(56810 Euro) |
| Personnel (ML) |
225
(116203 Euro) |
| Travel and subsistence (ML) |
55
(28405 Euro) |
| Other costs (ML) |
40
(20658 Euro) |
|
| TOTAL (ML) |
1180
(609419 Euro) |
|
| Money requested to the MURST (ML) |
920
(475140 Euro) |
| University co-financing (ML) |
260
(134279 Euro) |
Description
Thanks to a COFIN99
research grant, the PI has acquired a computer with 4 CPUs. Rather than
acquiring a supercomputer, which requires a full time computer analyst,
we plan to acquire 2 parallel-architecture 32-node distributed-memory
computers, as well as 2 graphic workstations and ten personal
computers, all of them being connected on a local network as well as to
Internet.
It is important to remark that the items indicated in the following do
not include the expenses to be made with the funds already available to
some of the research lines.
Here follows a detailed description of the first year costs:
Durable Equipment: 700Ml.
2 graphic workstations, 2 parallel-architecture 32-node
distributed-memory computers, 1 laser color printer, 1 high-speed b/w
laser printer, 1 workbench for virtual reality. Air conditioning for
the computer room.
Consumables: 50Ml.
Paper, toners, pens, pencils, electricity and telephone costs.
Computing: 110Ml.
Computer maintainance, licences for structural, fluid dynamics, visualization, and virtual reality commercial codes.
Personnel: 225Ml.
90Ml for Ph.D. positions, 90Ml for post doctoral positions, 15Ml for one part-time secretary, 30Ml for a system manager.
Travel & Subs. 55Ml.
Scientific missions for the presentation of the results.
Other costs: 40Ml.
External seminaries, organization of workshops, publications.
TOTAL: 1180Ml.
Second year - Costs
| Durable equipment (ML) |
30
(15494 Euro) |
| Consumables (ML) |
50
(25823 Euro) |
| Computing (ML) |
110
(56810 Euro) |
| Personnel (ML) |
225
(116203 Euro) |
| Travel and subsistence (ML) |
55
(28405 Euro) |
| Other costs (ML) |
40
(20658 Euro) |
|
| TOTAL (ML) |
510
(263393 Euro) |
|
| Money requested to the MURST (ML) |
415
(214330 Euro) |
| University co-financing (ML) |
95
(49063 Euro) |
Description
Durable Equipment: 30Ml.
Desks, bookshelves, books, journals, PCs, etc.
Consumables: 50Ml.
Paper, toners, pens, pencils, electricity and telephone costs.
Computing: 110Ml.
Computer maintainance, licences for structural, fluid dynamics, visualization, and virtual reality commercial codes.
Personnel: 225Ml.
90Ml for Ph.D. positions, 90Ml for post doctoral positions, 15Ml for one part-time secretary, 30Ml for a system manager.
Travel & Subs. 55Ml.
Scientific missions for the presentation of the results.
Other costs: 40Ml.
External seminaries, organization of workshops, publications.
TOTAL: 510Ml
Third year - Costs
| Durable equipment (ML) |
30
(15494 Euro) |
| Consumables (ML) |
50
(25823 Euro) |
| Computing (ML) |
110
(56810 Euro) |
| Personnel (ML) |
225
(116203 Euro) |
| Travel and subsistence (ML) |
55
(28405 Euro) |
| Other costs (ML) |
40
(20658 Euro) |
|
| TOTAL (ML) |
510
(263393 Euro) |
|
| Money requested to the MURST (ML) |
415
(214330 Euro) |
| University co-financing (ML) |
95
(49063 Euro) |
Description
Durable Equipment: 30Ml.
Desks, bookshelves, books, journals, PCs, etc.
Consumables: 50Ml.
Paper, toners, pens, pencils, electricity and telephone costs.
Computing: 110Ml.
Computer maintainance, licences for structural, fluid dynamics, visualization, and virtual reality commercial codes.
Personnel: 225Ml.
90Ml for Ph.D. positions, 90Ml for post doctoral positions, 15Ml for one part-time secretary, 30Ml for a system manager.
Travel & Subs. 55Ml.
Scientific missions for the presentation of the results.
Other costs: 40Ml.
External seminaries, organization of workshops, publications.
TOTAL: 510Ml
Overall costs
| First year (ML) |
1180
(609419 Euro) |
| Second year (ML) |
510
(263393 Euro) |
| Third year (ML) |
510
(263393 Euro) |
|
| Total costs (ML) |
2200
(1136205 Euro) |
|
| Money requested to the MURST (ML) |
1750
(903800 Euro) |
| University co-financing (ML) |
450
(232406 Euro) |
|
|
|
Data 16/10/2000 18:25
|
|
|
|
(del sistema alla chiusura da parte del proponente)
|
|
Firma del Rettore .................................
|
|
Data 17/10/2000 12:18
|
|
(PROF. Antonio CASTORANI)
|
|
(del sistema alla chiusura da parte del rettore)
|
(per la copia da depositare presso
l’Ateneo e per l’assenso alla diffusione via Internet delle
informazioni riguardanti i programmi finanziati e la loro elaborazione
necessaria alle valutazioni; legge del 31.12.96 n° 675 sulla "Tutela
dei dati personali")