Fig. 1: Institut National des Sciences Appliquées of Lyon
The Institut National des Sciences Appliquées of Lyon
Celebrating last year its 50th
anniversary, INSA Lyon (www.insa-lyon.fr) is one of the top French
Engineering schools for offering one of the best preparation for the
professional world. Its strengths are:
quality of education and pedagogy, an emphasis placed on educational
staff and applied research, and international outlook. There are over 6
000 students taking first and higher degrees at INSA Lyon.
The scientific objectives of the Ampère Lab
The Ampère Lab has a general scientific objective, namely the
efficient management and the rational utilization of energy in systems
in relation with their environment, with respect to a wide range of
fields:
• Domains: System and Control, Electrical and Electromechanical Engineering, Microbiology,
• System range: from power transmission components to energy systems,
• Frequencies of the phenomena: from static to microwave (GHz),
• Application range: from transport systems to energy, environment and bioengineering,
• A general approach in engineering: understand
the physical phenomena in order to design complex systems.
Fig. 2: The Ampère Lab organisation
Scope of activities and key words are:
• dielectric and magnetic materials, power
electronics, high voltage, electromagnetic compatibility,
electromagnetic modelling,
• power transmission and control, mechatronics,
Fluid Power, robotics, diagnostics and operational security,
• ecological engineering, bacterial adaptation and gene transfer.
Among the 6 research teams of the Lab, the "Actuators and Systems", and
the "Control" research groups are collaborating in the "Fluid Power and
medical Robotics" scientific activity.
2 Research
The study of the transmission of power using a fluid
(oil, water or gas) constitutes, with electrical power transmission,
complementary aspects to one of the major research problem at the
present time, that is to say the design of energy efficient systems.
From this point of view, the Ampère Lab is unique in France
since it has a recognized expertise of all these technologies for power
transmission. This expertise is essential when considering the previous
economic and scientific issues. The research in Power Transmission and
Control is divided into 3 main scientific areas:
• multi-domain and multi-scale modelling,
• methods and tools for system design,
• system control.
The specificity of our research is to deal with multi-disciplinary
aspects on the theoretical viewpoint as well as on the experimental
side.
2.1. Multi-domain and multi-scale modelling
This research activity tackles theoretically as well as experimentally
the modelling of multi-energy power transmission systems, and
particularly when fluid energy is used.
On the theoretical side, a unified and structured approach of the
modelling for multi-physic and multi-scale systems (Bond Graph) is
developed, aiming at preserving the real nature of mathematical models
before any step of space and time discretization.
Concerning simulation aspects, the problem of coupling numerical codes
for fluid mechanics and for system dynamics is also considered, this in
order to globally optimize a component or a circuit, considering
simultaneously the mutual influence of control and local flow
phenomena.
Finally, on the experimental side with a strong link with industry, the
expertise acquired on the pneumatic technology is still reinforced.
Since 2007, research works have also been done coupling electrical
drives and conventional power transmission (hybrid vehicles), and
recently very original research projects have been initiated
(pneumatic-electric battery). Based on Bond Graph energy conservation
principles, complex models have been developed for automotive industry
and lead to the implementation of several libraries in AMESim
(LMS-IMAGINE).
Structured modelling and multi-scale systems
Models of Fluid Power transmission systems have to represent complex
behaviour based on nonlinear equations of infinite dimension with
strong interaction between local and macroscopic scales. In this
domain, computation time is still high. Local phenomena such as
turbulences, cavitations, or shock waves have influence on performances
at the macroscopic level (for example, turbulence may influence the
stability of a piloted poppet).
At the system level, the representation of this kind of relation is
more important than the characterization of these phenomena in any
point of space. To describe these interactions, a bond graph
representation based on boundary energy flow using port variables is
used. It relies on a geometric structure, called a Stokes-Dirac
structure, which encodes a relation between internal dynamics and
boundary exchanges (1), in a power-conserving manner (Fig. 3). The main
interest of this approach is that the mathematical structure is
preserved. It allows an easier interconnection of sub-models,
multi-domain modelling, mixing conservative and irreversible processes,
and structure analysis (controllability, observability, inverse
methods, and passivity).
Fig. 3: Decomposition of a Stokes-Dirac structure
System optimization with parameterization in CFD
In the design of fluid power components or systems, Computational Fluid
Dynamics (CFD) can play a key role for flow optimization, but also in
the performance analysis of the whole system. To study the sensitivity
of flow to parameter changes (geometry, boundary conditions), we apply
a novel approach based on the parameterization of the CFD computed
solution. It consists calculating an approximation of the complete
Navier-Stokes equations around a reference point.
This method is particularly efficient and more easily couples local
approaches to macroscopic system simulation (Fig. 4), avoiding long
computational schemes. This parameterization method can give all the
results corresponding to a continuous variation of design parameters
from one single design step, which means one single flow computation on
one single mesh.
Fig. 4: Parameterization in CFD for system simulation and optimization
New experimental procedures in pneumatics
Complementary to previous approaches, our research group has developed
a strong experimental knowledge of the pneumatic technology, crucial in
relation with industry.
Fig. 5: Simulated vs. measured pressure and temperature for a tank discharge
Since 2006, in the framework of the
ISO , we have worked on new experimental procedures for the
characterization of pneumatic components. The proposed method relies on
the discharge or the charge of a tank in order to identify the mass
flow rate. This identification procedure naturally requires a good
knowledge of the heat transfer (convection) and experimental procedures
to identify model parameters. Good results have been obtained (Fig. 5)
and this allows a wide range of applications to be explored now.
Beyond the multi-disciplinary approach for the modelling of power
transmission systems, the originality of these works relies mainly on
the continuity between theory, virtual prototyping and experimentation.
This expertise plays a crucial role in the quality of our research and
its application in industry.
2.2. Methods and tools for system design
The main objective of this research work is the theoretical and
methodological development of tools for studying generic and intrinsic
properties of systems. Our approach, based on an inverse methodology,
is a break with the conventional simulation methods applied in
engineering since it tries to avoid the conventional
try/error/correction iterative procedure (Fig. 6). Because inverse
approaches are the natural way to define design problem (generally,
output are specified variables while inputs are unknown), it leads to
more efficient and relevant methods (a single calculation is required).
The final goal here is to give simple rules and methods that can easily
be used in industry for design purposes. These works are based on Bond
Graph energy conservation principles, causality concepts, model
inversion techniques, algebraic approach, and operational analysis in
infinite dimension.
Fig. 6: Inverse approach for system design
Design model / design problem adequacy
The purpose here is to define properly the design problem before any
numerical work. Structural analysis tools for studying system
controllability, inversion, I/O decoupling, are useful to show the
right choice. With our approach based on causal and bicausal concepts,
this information can be obtained at the early stages of the design
project (on the Bond Graph model). This indicates how to change the
architecture, the output specifications or the component choice in
order to reach the required performances for the system (Fig. 6).
Recently, for MIMO linear systems, essentiality orders have efficiently
been applied to determine if there is a control law that enables the
system to be statically decoupled and inversed.
Software developments within MS1 (Lorenz Simulation) and Modelica
(Modelica Association) are now available and have been tested in
industry.
Help for defining design specifications
The purpose of this second aspect is to couple inversion techniques to
dynamic optimization in order to tackle the energy efficiency problem
in system design. It consists of simultaneously solving the component
choice and the optimal control problems. Due to numerical difficulties
for computing the initial conditions on the co-state of the
optimization problem, this method has presently been implemented for
MIMO linear systems only.
The Ampère Lab is one of the world’s leading research
groups in this domain. The expected results should solve a wide range
of problems in engineering, such as system design, choice of component
technology and system architecture, robustness, and control law
structural analysis. A wide range of application is covered by this
approach: mechatronics, hybrid vehicles, Fluid Power, etc.
2.3. System control
Started 20 years ago, this activity focuses on the adaptation,
synthesis, implementation and comparison (performance,
economical… criteria) of modern control laws, issued from the
new development in control theory of the international automatic
community. The Lab has wide theoretical and experimental experience on
gains scheduling, nonlinear I/O linearization, adaptative, first order
and higher sliding mode, flatness, backstepping control, with aims to
adapt these different strategies in prototypes or experimental test
rigs.
During the last three years, high order sliding mode control,
techniques based on strictly positive real systems (SPR) and control
strategies without sensors have mainly been studied. Generally, the
applied methodology can be described as follows:
• first, simulation models are used and analyzed in order to obtain reduced models,
• second, the control strategies are adapted and
implemented with co-simulation tools (Fig. 7): system simulation with
AMESim and control with Matlab/Simulink),
• third, the control strategy are implemented
using rapid control prototyping tools (dSPACE) and tested on
experimental rig,
• the last step, the implementation of the
control algorithm on the appropriate electronic will be developed in
the future with an other research group of the Lab.
Fig. 7: Virtual bench test about rudder control
High order sliding mode Control
A drawback of the sliding mode is that the discontinuous control signal
may excite high frequency dynamics neglected at modelling stages such
as non modelled structural modes, time delays and so on. This causes
fast, finite-amplitude oscillations known as chattering, which would
result in loud noise, high wear of moving mechanical parts and thus
should be definitely eliminated. The consequence of this phenomenon
concerns the wearing effect of power modulators and actuators. Excess
of energy consumption and high frequency excitation can be very harmful
to the system.
A classical solution to reduce the chattering consists of using a
threshold in control law; however, the controller robustness is
inhibited theoretically and introduces a static error experimentally.
Another complex solution consists of using high order sliding mode
algorithms. This technique is studied at the Ampere Lab and was
demonstrated on a missile rudder pneumatic positioning application;
this was also compared to a gains scheduling approach. This work leads
to the development of specific observers (extended or not) using only
one mechanical sensor. Figure 8 shows the test bench developed for
testing specific control algorithms.
Fig. 8:Experimental test bench for rudder control
Control without sensor
This work concerns the study of electro-pneumatic actuators in an
environment with high perturbations, for example the observation of
perturbation in aeronautic context using a minimum of physical sensors.
The robust nonlinear control can be a solution to reject perturbations,
but this approach leads to knowledge of all variable states. Our work
aims at the development of specific control strategies with high
performances, without position sensors or pressure sensors but using
observers to obtain the unknown information.
Strictly positive real system and stick-slip phenomenon
From 2007, some works have led to proposals of
new methods to study the stability of an equilibrium set of nonlinear
models. It has been applied to path planning of electro-pneumatic
actuators. Moreover, using an appropriate switching strategy between
different control laws (path planning or pressure regulation), the
undesirable stick slip occurrences have been reduced to zero.
Control of electro-hydraulic system
A high performances electro-hydraulic test
bench has recently been developed. It offers high acceleration, high
precision, micro-displacement at very low velocities, with high
bandwidth servovalves (Fig. 9).
Classical control strategies have been
implemented in simulation and experiments. An accurate virtual model is
developed including servovalve, hydraulic lines and pressure supply
nonlinear characteristics and dynamic.
Control of hybrid Fluid Power systems
An actuator controlled by on/off valves is
structurally considered as a hybrid dynamic system. This class of
system presents simultaneously continuous dynamic (actuator) and
discrete behaviors (on/off valves). Due to recent developments of fast
on/off valves, it is now possible to replace classical servovalves by
low cost servosystems using an appropriate control algorithm. A first
application of this technique in the Lab is the development of a
specific electro-pneumatic haptic system for learning perfect movements
in medico-chirurgical context.
Fig. 9: Electro-hydraulic test bench (actuator on right hand)
These
know-how and original contributions leads to an important expertise in
term of Control in Fluid Power, from theory to application. Today,
researches focus on energy efficiency, stick-slip in pneumatic systems
and robustness. The main aim is still to be able to specify which
control law is the more appropriate for a specific Fluid Power system.
3 Experimental Facilities
The Ampère Lab has 7 laboratory facilities located at the different Engineering schools supporting the Lab:
• Electromagnetic Compatibility test centre,
• Environmental microbiology laboratory,
• High Voltage test centre,
• Diagnostic platform for electrical systems,
• Characterization and Reliability platform for passive components,
• Characterization platform for power SC components and systems,
• Fluid Power Centre.
The Fluid Power experimental facilities are located at INSA Lyon on 140
m2. It consists of a range of electro-pneumatic and electro-hydraulic
rigs used for different purposes such as virtual prototyping
(modelling) and rapid prototyping (control implementation). The main
characteristics are given below:
• Fluid energy
supply: air supply (2500 Nl/mn at 7 bar), hydraulic supply (18,5kW, 46
l/mn à 210 bar), vacuum pump,
• high performance hydraulic drive (300 mm
stroke, 20 kN maximum static force, 2 servovalves @ 1kHz +/- 5%
control),
• electro-pneumatic drive benchmark for the performance evaluation of control algorithm,
• flow/pressure rigs for pneumatic component and circuit characterization (ISO6358), isothermal tank,
• sensors, acquisition system, rapid prototyping equipments (dSpace)
4 International links
The Fluidtronic group is a
founder member of the Fluid Power Centres of Europe (FPCE ), a grouping
of the foremost research institutions in the field. Each year
international scholars, mainly from North Africa, North and South
America, are visiting the Lab for long or short stays. It welcomes also
PhD students from scientific programs with China, South America, and
North Africa.
5 Relation with industry
The Fluidtronic research
activities are mainly industry focussed. These collaborations are built
around PhD, undergraduate and MSc projects of the Department of
Mechanical Design Engineering. Recent collaborations on research and
consultancy projects include:
PSA PEUGEOT CITROËN, VOLVO TRUCKS, VOLVO POWERTRAIN, DGA (Ministry
of Defence), CETIM, CNES (French Aerospace Agency), BOSCH,
BOSCH-REXROTH, EDF, NUMATICS-ASCO JOUCOMATIC, LMS-IMAGINE, SHERPA
ENGINEERING, LORENZ SIMULATION, FLUOREM SA, PLASTIC OMIUM AE, IRISBUS,
LOHR INDUSTRY, ALSTOM, KNORR, HCL (Hospice civil de Lyon), …
