Mechatronics design and optimisation methodology A problem formulation focused on automotive mechatronic modules Fredrik Roos, Jan Wikander Department of Machine Design, Mechatronics Lab KTH, 100 44 Stockholm e-mail:
[email protected]
Abstract
1
Introduction
The role of electromechanical or mechatronic subsystems will increase dramatically in future road vehicles. Two of the main driving forces for this development are new/improved functionality and modularisation, e.g. to make the auxiliary systems driveline independent. The foreseen very large production volumes of these sub-systems opens up new degrees of freedom in the design of the sub-system, i.e. integrated design and optimisation of all constituent components are possible. This paper is intended to give an overview of the complexity of the design of mechatronic modules and also to describe the authors’ approach to develop design and optimisation methods for automotive mechatronic modules. The idea is to develop a model based design and optimisation methodology for self-contained and physically integrated mechatronic modules. New optimisation methods for multi criteria optimisation across engineering domains are a necessity to find the best sub-system design for a given set of requirements. The methodology will focus on module design, but it will also be useful for comparing different design concepts. Some requirements engineering also need to be incorporated into the methodology since it is very important that the sub-systems are specified to such a level of detail that multi criteria optimisation can be performed. The problem and approach to an integrated mechatronics design and optimisation methodology is exemplified with a steer-by-wire system in the second part of the paper.
The supply and demand of electric energy will increase dramatically in future road vehicles. Many of the in-vehicle systems that today are mechanic, pneumatic or hydraulic will be replaced with electromechanical systems – in many cases under closed loop computer control. This development is primarily driven by new and improved functionality, but it is also necessary for the transition to electric or hybrid electric/internal combustion drive trains. In addition to the driving forces already mentioned the benefits of electromechanical or mechatronic sub-systems over purely mechanical systems are many; maybe the most important one is the new possibilities of modularisation. A mechatronic system can be modularised far more than a purely mechanical system, since the power and information flow between the modules can be transferred on electrical wires. One example is the braking system in a enger car; a mechatronic braking system (brake-by-wire) can be based on four mechatronic wheel brake modules. The information flow to the modules from the brake pedal (or other brake command device) is electronic and the power is transferred to the brake modules through the vehicle’s electrical system. This means that all brake functionality as for example, ABS and traction control system can be implemented in a more flexible way through control algorithms implemented in software.
1
in the first part of the design process. That way it will be possible to get aspects from for example a control or electrical point of view into the mechanical design at an early stage of the design process. This approach makes it possible to find an optimum design for the entire automotive sub-system, not only for a specific domain within the sub-system.
An automotive mechatronic sub-system is typically very complex to design and especially to optimise, mainly because of the multi domain characteristics of mechatronics, but also due to the safety critical nature of these systems. Further, the environmental disturbances will in many cases be extremely harsh, consider for example a break unit attached to an unsuspended structure and subject to extreme temperature and humidity variations. These characteristics and requirements in combination with also extreme cost requirements, call for new methods for specification, design and optimisation of automotive mechatronic sub-systems. Many general mechatronic systems are designed in a concurrent manner resulting in an hopefully optimised integration and configuration of standard components such as actuators, sensors, gearings and control units. The foreseen large production volumes of the new types of mechatronic sub-systems targeted in this work, opens up new degrees of freedom in the design in the sense that integrated and concurrent optimisation of all constituent components is possible. This both required and resulting freedom in design emphasises the need for new design methods and tools.
2
The key to an integrated mechatronics design methodology is modelling and simulation. Being able to analyse and simulate the sub-system at an early design stage is of great help to find the optimum design solution. Hence, one of the most important features of mechatronics design tools will be the inter-changeability of models between design tools from different domains.
Mechanicaldesign design Mechanical Mechanical design views views views
Mechanicaldesign design Mechanical Electromechanical views views design views
Mechanical design Controller and Mechanical design views Software design views views
Core model
Mechatronics development methodology
Mechanical design Electronic Mechanical design views hardware design views views
Concurrent model based engineering, one model many views
Figure 1. Different design views in model based mechatronics engineering
The mechatronics development process is different from other development processes in the sense that it spans over many closely coupled engineering domains. In order to optimise an automotive sub-system it is necessary to treat the entire sub-system in an integrative way during the design process. Parameters like inertia and electrical capacity depend on the sub-system geometry and can only be optimised in common [1], [2]. Further, the extensive functionality and complex structure of mechatronic systems means that it generally is not enough to optimise on a single criterion, often a multi-objective optimisation is needed.
Tool that for instance, easily enables changes made from an assembly point of view to be evaluated in control system simulations is desirable. This demands a ‘core model’ that represents the entire design and that can be viewed from various windows (views). It would be a great advantage if changes made in one view (window) were reflected immediately in other views [3]. There are however some problems arising just from the fact that different engineering domains use different models and modelling frameworks during the design work, e.g. control engineers are used to models in the form of transfer functions or state space descriptions. Without special precautions, such models generally do not have a direct relation to the physical parameters in the mechatronic system.
Optimisation within each domain separately will not result in the optimum system design; therefore all the domains of the automotive subsystem have to be treated concurrently, at least
2
3
be a necessity to facilitate integrated design across engineering domains and to facilitate optimisation. Further the models should different levels of abstraction and the different views typically used in the involved engineering domains. Modelling and simulation tools is not the focus here, but the modelling methodology will function partly as a requirement specification on a design tool chain. Many of the available domain specific design methods are based upon some optimisation methodology, e.g. as in systems identification, control design and fault detection. Extending the use of optimisation across engineering domains will involve parameterisation of an increased set of physical characteristics, widening the parameter space and increasing the complexity of the optimisation problem.
Approach
As depicted in figure 2, the realisation of advanced functionality in future vehicles entails a complex design process. The problem that is being approached in this research is hence strongly delimited to a methodology for deg optimal mechatronic modules. Such modules will be the low level corner stones to achieve advanced functionality such as vehicle stability control and collision avoidance. Given a vehicle’s high-level functional requirements and architectural platform, the design of a particular functionality, e.g. lateral stability control, is assumed to be initiated with a conceptual design phase evaluating different possible concepts. A selected conceptual design typically involves the usage of lower level mechatronic modules such as a break module in a break-by-wire system. From the point of view of modularity the hypothesis is here that the mechatronic modules we are aiming at are self-contained and physically integrated units for advanced and versatile actuation of Functional Requirements motion and force. The goal behind this problem formulation is to develop a model based design and optimisation methodology for such automotive mechatronic modules. The methodology will be based on dynamic component models and design by simulation. By facilitating the design/selection of all constituent components of a module, in the same integrated and concurrent design process, there is a potential of optimisation that goes across engineering domains. The main constituent components would typically be mechanical structure, actuator, transmission, sensor, control algorithm/software and embedded processing hardware. A modelling strategy and framework that allows to tly model the necessary characteristics of all constituent components will
System requirements (vehicle system)
Sub-system 1 requirements
Sub-system n requirements
Non-Functional Requirements Requirements identification
Functional structure (Functional model)
Conceptual design 1
Conceptual design 2
Evaluation and optimisation of conceptual design
Conceptual design 3
Conceptual design n
Etc.
Optimal concept
Electric view Geometric view Control view
Gear box model
Motor model
Controller model
Load model
Modelling of sub-system components
Analysis and optimisation
Final sub-system design Model database
Spring constant
EMF
Friction
Inertia
gear ratio Models of physical properties and phenomena
Figure 2. Model based design methodology - our focus is below the red line, and particular the analysis and optimisation concepts
3
Further, considering the requirements and the advanced functionality of the modules, the optimisation problem will involve several optimisation criteria.
Command from driver or from other system
Automatic Control
Software
Primary optimisation criteria that are targeted in this research are: • Weight / Envelope • Energy-efficiency
Electronics Sensors
Controller
Power electronics Converter /Driver
Electromechanical actuator
Transmission / Linkage
Load
Physical mechatronic module
Other important criteria for which the evaluation will be simplified by the developed methodology are for instance: • Cost • Noise / Vibrations • Dependability / Safety • Environmental Impact • Modularity / Flexibility
Energy Source / Buffer
Figure 3. Block diagram of a mechatronic actuation module.
One of the actuation modules will be the driver command/ module, attached to the steering wheel/joy-stick. The other will be the actual steering actuator that turns the front wheels of the vehicle. A conceptual block diagram of the system is shown in figure 4.
In the lower part of figure 2, the focus of the work is indicated. The models of constituent components and the physical phenomena that are important for the components’ separate as well as compound behaviour, will be developed and stored for easy access and use. The methodology will hence both design of components and selection from sets of already modelled components. Even though the research is focused on the module design it is inevitable to include the requirements engineering work on higher levels to ensure that requirements are specified in such a detail that the multi-criteria optimisation can be performed. This will also be very important when using the developed methodology for rapid evaluation of different conceptual designs.
4
Steering wheel / Joy-stick Driver
Mechatronic module TTP, communication bus
Stability Control Mechatronic Steering rack (wheel actuator) LF wheel
RF wheel
Figure 4. Steer-by-wire system
This small example is limited to the mechatronic steering rack module in figure 4. The mechatronic steering rack is here assumed to be a module (figure 3), with a linear electric motor as actuator (for simplicity).
Application; steer-by-wire
This section exemplifies the approach for the integrated mechatronic design methodology that is described earlier in this paper. The example is also intended to show the complexity and difficulties with highly integrated mechatronic design. As an example we will use a very simple steerby-wire system that consist of two mechatronic actuation modules interconnected with an electronic communication link (e.g. TTP or FlexRay).
The objective is to optimise the mechatronic steering rack module with respect to weight and/or energy efficiency given predefined yaw torque and velocity requirements on the front wheels of the vehicle. The four most important high-level views of the sub-system within this context are the geometrical, dynamic analysis, electro-mechanical and control views.
4
The different views of the subsystem can be classified as, for example, form or behaviour views. Where the behaviour views capture the mathematical description of the physical and informational behaviour of the component or sub-system [4], and the form views capture the shape of the sub-system or component. Each component in the sub-system can contain multiple behavioural ‘models’ with different level of detail. However, in this short example we will just show the top-level views of the subsystem.
4.1
A related view that not is shown here is the multi body system (MBS) view of the mechanical linkage, including the motor mover (permanent magnet). This view is important for simulating forces and torques on the structure.
4.2
Electromechanical view
The electromechanical view describes the transformation from electrical to mechanical energy in the electromechanical actuator.
Geometric view
R
L
+
F= Cf * I E= Cv * v
Converter/ Driver G
Uin -
F, mm, v
+ Iref
Figure 7. Electromechanical view of the sub-system
Figure 7 shows a simple high level model of the linear motor. Important variables and parameters are, the motor force and voltage constants (Cf , Cv), necessary size (mass, mm) of the mover etc. Figure 5. 3D geometric view
A geometric view (CAD view) of a system describes all physical dimensions of all the constituent mechanical parts.
4.3
Control view
The control view shows the sub-system from a control engineer’s perspective, i.e. a mathematical representation of the plant and the controller. Load
p ref
e
e
I ref
Iref
PD controller
I ref
I
Current
Cf
Force
Gain1
G inverter
Integrator position
Figure 6. Geometric view of the steering rack
1 s
v elocity
1 s
acceleration
Integrator1
The geometric view of the mechatronic steering rack can at the top level be an assembly drawing looking something like the one shown in figure 6 with 3D visualization in figure 5. Important parameters from the geometric view are size and shape of moving parts (l, w, h, dm, lm). Combined with information of the material used, parameters like mass and inertia of the moving parts can be calculated.
ml + mm Gain
Figure 8. Control view
The control engineer is dependent on data (parameters) both from the electromechanical and geometric/MBS views.
5
4.4
Table 2 shows a design structure matrix (DSM) with some selected system parameters. As seen in the DSM and table 1, one of the most central parameters in the design and optimisation of a sub-system like this is the motor mover geometry and mass (mm). The mover makes it hard (impossible?) to isolate the mechanical load from the electromechanical actuator design.
Parameters and optimisation
Linkage mass, m Mover mass Stator geometry Number of Windings Max current, Im Motor Resistance Force constant, C f Voltage constant, C v Controller design
X X X O O X O X O X
ue Tor q
Conclusions and future work
The dependence of parameters from different engineering domains (and views) is large in mechatronic systems. The only possibility to optimise a mechatronic sub-system is to treat the system in an integrative way, across the different domains.
effi
cen cy Tot al m ass
X X X O O O O O X X
5
E ne rg y
eer ing . St
Req
Linkage geometry
Req
Elektromechanics
Motor geometry CAD
. St
eer ing
velo city
A selection of the most important sub-system parameters is listed in table 1. An X marks a direct relation between parameter and requirement/optimisation criterion. An O indicates an indirect relation. As seen in the table all of the parameters listed affect the performance (torque and velocity) either direct or indirect. In other words there exists a vast number of different sub system designs that will fulfill the subsystem requirements on torque and speed.
O O O O O O X
Now, when the problem has been described, we will focus on finding design and optimisation methods. To start with for a very reduced subsystem, maybe just for a motor and a fixed load, then more and more parts will be added to the problem. As an hypothesis it will be possible to find isolated parts of a system that are not so dependent on other parts, and in that case they can be treated separately.
O X X X O O O O O
X
References
Table 1. Parameter relations to requirements and optimisation criteria.
[1] Gausemeier J., Flath M. and Möhringer S. (2000), Modelling of Functions of Mechatronic Systems, exemplified by tyre pressure control in automotive systems, Int. J. of Vehicle Design, Vol 28, Nos. 1/2/3, 2002, pp. 5-17
If we study the parameters that affect the mass of the sub-system directly, i.e. linkage mass, mover mass and stator geometry; the optimisation problem of minimising the weight of the mechatronic steering rod, does not appear to be very complex. But if we consider the parameters that affect the mass indirectly the optimisation problem grows rapidly. L
Cf
Cv Im
X X
X X
X X
Cf
O X X O X X O
X
Cv X Im
L
[3] van Amerongen Job, Coelingh Erik, de Vries Theo J.A. (2000), Computer for mechatronic control system design, Robotics and autonomous systems no 30 2000 pp. 249260.
Depend →
Provides → w h l ml dm lm mm R Linkage witdh w X Linkage hight h X Linkage length l X Linkage mass ml Mover diameter dm X X Mover length lm X X Mover mass mm Motor Resitance R Motor Inductance Motor Force Const. Motor Voltage Const. Rated Motor Current
[2] Rothfuss, R. et al. (2002), Systems Engineering in the Design of Mechatronic Systems, Int. J. of Vehicle Design, Vol 28, Nos. 1/2/3, 2002, pp. 18-36
[4] Rajarishi Sinha, Paredis Christiaan J.J., Khosla Pradeep K. (2002), Behavioral Model Composition in Simulation-Based Design. Proceedings of the 35th annual simulation symposium IEEE 2002
X = direct relation O = inderect relation (e.g. sets requirments on)
Table 2. Design Structure Matrix (only a few of a vast number of parameters are shown here)
6