Artificial technologies in sustainable braking system development

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/234052439

Artificial technologies in sustainable braking system development Article in International Journal of Vehicle Design · February 2008 DOI: 10.1504/IJVD.2008.017185

CITATIONS

READS

9

43

2 authors: Dragan Aleksendrić

Cedomir V. Duboka

University of Belgrade

University of Belgrade

62 PUBLICATIONS 276 CITATIONS

64 PUBLICATIONS 182 CITATIONS

SEE PROFILE

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

26th JUMV Intl. Automotive Conference Belgrade 2017 View project

All content following this page was uploaded by Cedomir V. Duboka on 06 February 2015. The user has requested enhancement of the downloaded file.

2

1

D. Aleksendrić and Č. Duboka

Introduction

As the automotive industry requires products of ever increasing quality in a shorter time to market, and as the number of confronted requirements is also rapidly increasing, engineers are turning to advanced technologies for help. In order to enable technical systems to support after-sales activities and provide better customer satisfaction in order to become successful in the market place, updated technologies, technical skills and methods should be applied in all phases of product development, like concept analyses, design, simulation, prototype testing, manufacturing, verification and certification, marketing and service. Fulfilment and tuning of ever-increasing mutually opposed, and often even confronted requirements is necessary, and that is why a systems approach is needed in solving the problems, even in the earliest phases of development. Systems Engineering should create a set of alternatives in order to satisfy imposed requirements for the product, because quality should be ‘in-built’ into each system, and not only ‘added’ to it. Management of engineering changes has to be skilfully implemented in a cost effective manner in order to satisfy the basic requirements of sustainable development i.e., price, quality of service, and time to market. That is why sustainable development of braking systems, for example, needs application of flexible technologies, being able to create and check new engineering solutions, without physical prototyping, in order to respond faster to customers’ needs and changing regulations under compressed development cycles. Regarding complexity of braking systems, according to Figure 1, the overall performance of braking systems interrelates with the performance of its subsystems and parts. The main goal of new technologies which could satisfy aforementioned development conditions is related to possibilities for establishing correlations between the braking system performance and performance of its subsystems and parts (service, secondary, and parking braking systems). Furthermore, development and testing of each component of the braking system should be done according to allocated requirements simultaneously with the cost and time to market prediction. Implementation of such an environment for braking system development demands knowledge and experience related to: •

fulfilment of requirements with regard to performance, reliability (or safe operation) and cost

•

test results obtained with many brake types following different test procedures for evaluation of tribological behaviour, and/or certification and verification purposes

•

life prediction method (Todorović et al., 1995)

•

modelling method for frictional behaviour of brakes (Todorović et al., 1987)

•

identification of tribo-mutations in brakes (Duboka et al., 1997)

•

investigation in the field of friction stochastic character, contact phenomena, friction mechanisms reliability performance, conformity of production, and quality of friction brakes, and other (Todorović et al., 1993; Duboka, 1996; Arsenić et al., 1986)

Artificial technologies in sustainable braking system development

3

•

development of digital mock up of brakes (Aleksendrić, 1999; Aleksendrić et al., 2000, p.10)

•

virtual brake testing (Aleksendrić et al., 1998, 2003; Džipković et al., 2001, 2003)

•

analysis and prediction of friction material characteristics (Aleksendrić et al., 2004, 2006)

•

identification and modelling of braking systems operation (Aleksendrić et al., 2006)

•

intelligent controlling of braking systems operation (Aleksendrić et al., 2006)

•

conducting the analysis of failure mode effects (FMEA)

•

the functional safety of computer control systems characterisation (Safety Integrity Level (SIL) of Safety Instrumented Systems (SIS)) etc.

Figure 1

2

Braking system requirements evaluation

Advanced system engineering

Systems engineering provides a conceptual framework for Integrated Engineering (IE) – a concept integrating ideas, methodologies and tools towards Total Quality Management. The effects of interactions among many parts of the overall system must be taken into account during the engineering process (Blanchard et al., 1981). Building physical prototypes to see how an assembly operates and running tests to see how, if, and where the parts fail would not be enough. This build-and-break method can take an extremely long time, and may cost millions of dollars. Systems approach starts with the problem formulation, system requirements, specifications, and practice to state the problem in terms of the top-level function that the system must perform.

4

D. Aleksendrić and Č. Duboka

Braking systems in road vehicles depend on a hierarchy, as shown in Figure 1: subsystems (two independent service braking systems, emergency braking system, parking braking system), assemblies (control, transmission, wheel brakes, for example), subassemblies (calliper including brake piston), and components (brake rotor which may be drum or disc depending on the brake kind and type, friction material which may be brake pad or brake lining, and brake control device, which may be wheel cylinder, etc.). Figure 1 also shows that analysing only system-level requirements will not be sufficient to enable identification of the mission imposed on the sub-system, assembly or individual components because each of them has to fulfil own-level requirements in order to the braking system-level requirements can be satisfied. Systems engineering has been employed in this paper, as shown in Figure 1, for braking system design or redesign from the point of view of imposed requirements to its subsystem and parts. The complexity of braking system developing and the trade-off procedure regarding the final performance of braking systems in this case may be demonstrated by changing the characteristics of the friction material. It will be shown that changing only one brake element i.e., friction material type, as a part of disc pad assembly, substantially affects braking system operation. That is why selection of friction material characteristics, their achievement and ‘virtual’ verification with respect to predefined braking system performance is a very complex task. This is even more complex in the case where the same braking system is going to be used on vehicles with different weights, different maximum speeds, suspension systems etc. As shown in Figure 1 there are a number of requirements which have to be satisfied by the friction material, as a part of disc/pad assembly, and all of the following requirements differently affecting the final braking system performance: •

bedding-in period

•

stability of cold friction coefficient

•

stability of friction coefficient vs. pressure, speed, and temperature variations

•

mechanical characteristics (compressibility, bending strength, hardness) etc.

The brake’s performance is primarily influenced by the contact situation between metal brake disc and friction material pads. The contact situation is differently affected by the wide range of mechanical properties of the friction material ingredients (Eriksson et al., 2000). Automotive friction materials are complex composites which may contain more then 20 different ingredients. The synergetic effects of all these ingredients, as shown in Figure 2, determine final friction material characteristics. The complex contact characteristics of brakes’ tribological system are mostly affected by the physicochemical properties of the friction materials ingredients. Moreover, the same friction material formulation can be differently affected by the manufacturing procedure set-up, defined by conditions of dry mixing, pre-forming, hot moulding, and heat treatment (post curing).

Artificial technologies in sustainable braking system development Figure 2

5

Influencing factors on friction material characteristics

From Figure 2, it can also be seen and concluded that the most complex task relates to establishing relationships between friction material compositions and manufacturing process parameters vs. imposed friction materials requirements. This is particularly difficult taking into consideration complexity of following requirements imposed to the friction materials: •

sensitivity of the friction coefficient to the application line pressure and/or sliding speed and/or temperature variation

•

wear characteristics under different thermal regimes of operation

•

vehicle weight

•

braking systems’ transmission characteristics.

3

Integrated engineering in virtual environment

It is evident that brake performance is affected by the complex inter-related tribo-processes occurring during braking in the contact of brake friction pair elements, especially regarding the variation of the friction coefficient in different operating regimes. The influence of the changing friction coefficient on braking system performance, expressed by adhesion utilisation diagrams (for vehicle laden and unladen conditions), is shown in Figures 3 and 4 (for the case of disc brakes used in the braking system of a passenger vehicle). Obviously, adhesion utilisation curves i.e., braking system performance curves are differently influenced by friction coefficient or brake factor C variations. Therefore, prediction of braking performance over different friction material types that might be built-in in brakes for a given vehicle is particularly important. Furthermore, prediction of brake performance has to be done for different

6

D. Aleksendrić and Č. Duboka

operating conditions, because friction material behaviour should be known in advance for different initial speed, application line pressure and temperature variations. Figure 3

Adhesion utilisation vs. brake factor ‘C’ variations: laden vehicle condition

Figure 4

Adhesion utilisation vs. brake factor ‘C’ variations: unladen vehicle condition

Therefore, system performance should be based on identifying and achieving previously allocated requirements to all subsystems and parts. This is especially related to operation of brakes and, in particular, to requirements allocated to friction material and the brake rotor. It has already been mentioned that trade-off procedures and adjustment of performance between system and subsystem and parts is not that easy to be done on physical models but might be usefully performed by means of virtual models of braking systems.

Artificial technologies in sustainable braking system development

7

Implementation of systems engineering procedures on the braking system in a virtual environment demands fulfilment of the following steps: •

defining top-level braking system mission-set up of adhesion utilisation curves for vehicle laden and unladen condition

•

setting-up and calculation of braking system transmission responses for possible driver inputs

•

development of digital mock up of brakes

•

providing preconditions for digital brake mock up testing

•

virtual brakes testing according to calculated braking system’s transmission outputs

•

using virtual test results for analysis of thermo-mechanical phenomena in brakes

•

modelling of friction material characteristics vs. formulation and/or manufacturing and/or operating condition variations.

Development of a digital mock-up of brakes is the most important step in implementation of integrated engineering in a virtual environment. Digital mock up development implicates integration of methods which should enable •

modelling of contact between disc and disc pads

•

defining of size, shape and position of contact region during braking process

•

defining of size and distribution of the surface pressure

•

enabling computation of value of thermal expansion in friction pair contact to achieve sharply rendered initial working conditions i.e., analysis of the hot-spot effect

•

research of impact of pressure distribution in friction pair contact in order to define the method of transfer of contact from one group of contact regions to another during braking process

•

procurement of possibility for computation of values of critical sliding speed for specific working conditions of the disk brake

•

computation of induced thermo-mechanical stresses and displacements under conditions of development of thermo-elastic instability, etc.

In order to enable accurate analyses of the friction pair contact phenomenon, there is a need to ‘digitalise’ braking cycle with regard to the strong influence of the load history over changes in initial speed, application line pressure, kinetic energy, temperature, friction material type, size of contact area in every discrete moment of braking cycle, etc. Moreover, the digitalisation method associated with finite element analysis helps to determine the value of the affected temperatures during the contact of a friction pair, and to analyse thermo-mechanical phenomena in brakes. The main task of digital mock-ups developing is to provide preconditions for virtual brake testing in line with many different ‘if – then’ scenarios. The basic pre-requirement for defining usefulness of a digital mock-up of brakes and its quality is to enable simulation of a braking cycle as it would be performed on an inertia dynamometer.

8

D. Aleksendrić and Č. Duboka

In this case, because of compatibility between virtual and real testing of brakes, a digital mock-up of the disc brake has been developed in such a way as to provide testing of brakes in digital form, taking into consideration real features of the single-ended full scale inertia dynamometer (see Figure 5), developed at the Frimeks Laboratory of the Faculty of Mechanical Engineering (Automotive Department), University of Belgrade. Generally, the aim of testing a brake by means of an inertia dynamometer is to provide a controlled method for transformation of kinetic into thermal energy by reducing the speed of revolving masses and to dissipate it to the environment as shown in Figure 5. The DC motor (1) via coupling (2), drives a set of flywheels (3) independently mounted on the driving shaft (4), thus providing a range of rotational inertia from 10 kgm2 to 200 kgm2. The flange (5) firmly joined to the shaft (4), bears rotating part of the tested brake (disc) while immobile flange (6), being firmly connected to the bench foundation (7) is used for mounting stationary parts of the tested brake (callipers). During brake testing, temperature, application line pressure, rotational speed and braking torque are recorded by PC – based data acquisition system at a sampling rate of 50 Hz. Figure 5

Single-ended full scale inertia dynamometer

Therefore, based on real single end full scale inertia dynamometer a ‘digital test bench’ has been developed which also represents a digital mock-up of a disc brake (see Figure 6). This was achieved by developing the digital mock-up, which would enable ‘testing of a disc brake’ under conditions which would be the same to those realised with the same brake while tested by means of the real brake inertia dynamometer or in the real vehicle. In the real test bench situation, a flywheel is used to duplicate that part of the vehicle total mass which corresponds to the vertical wheel load of the wheel to which the tested brake is associated. In the virtual test situation, instead of selecting a ‘flywheel mass’, it is possible to increase the density of the disc to achieve the necessary initial inertia. That is how the ‘virtual’ disc differs from the real one, because in the digital form disc does not only represent the brake rotor, but also the inertial mass by which the required initial kinetic energy for brake testing would be provided. Therefore, when the disc is accelerated to the prescribed initial brake speed, it also

Artificial technologies in sustainable braking system development

9

provides the required kinetic energy for given test conditions. One can also create other initial test conditions regarding brake initial speed, and brake application pressure, as well as the initial brake interface temperature. Figure 6

Digital mock-up of the disc brake

After the disc has been accelerated to reach the prescribed brake initial speed, the brake should be activated with the brake application pressure whose rate is defined in advance (see Figure 6). During virtual testing of brakes it is allowed to measure not only variations in kinetic energy, brake speed, and braking torque but also the size of contact area between disc and disc pads. As it is shown in Figure 6, a sphere of exceptionally large diameter is used to approximate the engaged disk pad area entering into contact with the disc. Simultaneously with the pressure increase during brake activation, the spheres, shown in Figure 6, will be engaged to provide flat surface contact. Depending on the instantaneous value of pressure activation, the total contact area will be changed. The advantage of such a digital approach reflects the fact that in the case when one knows the size of the total contact area it would be possible to investigate complex thermo-mechanical phenomena by means of an unlimited number of plans of distribution of thermal and/or mechanical loads on the predefined contact regions that correspond to the real ones as shown in Figure 7. The distribution of thermal and/or mechanical loads at every contact region has different functional dependency over time and depends on the working conditions and characteristics of the materials in contact (see Figure 8). This scenario can be applied to the disc pad and/or brake rotor.

10

D. Aleksendrić and Č. Duboka

Figure 7

Thermo-mechanical loads modelling on the disc pad digital mock up

Figure 8

Thermal loads and expansions calculation according to virtual testing results

The critical point of virtual brake testing relates to friction variations during braking for specific working conditions. It is clear that during virtual testing friction changes have to be known in advance in order to start testing according to pressure, speed and temperature variations. That is why a virtual environment for brake testing needs embedding intelligence capabilities. It means that artificial intelligence technology can be employed for friction material characteristics prediction in the contact of the friction pair, for a wide operating conditions range. As explained above, very complex and highly non-linear phenomena are involved in the field of tribology. This is the reason why analytical models are difficult, even impossible, to obtain. The understanding of friction change requires accurate modelling and prediction vs. friction material formulation, manufacturing and testing conditions.

Artificial technologies in sustainable braking system development

11

Artificial intelligence consists of various technologies. One of them is artificial neural networks as a method for functional approximation between input and output parameters. The mechanism of parallel processing of input signals towards outputs as a way for learning input/output relationship is used here for predicting of friction material characteristics in the contact of the friction pair. The abilities of a neural model of friction material behaviour related to cold performance are illustrated in Figure 9. Figure 9

Cold performance comparison: real and predicted

Implementation of the artificial neural network for the friction materials behaviour modelling is a complex task because, besides all other factors, the architecture of the neural network needs to be properly determined. The architecture of a network consists of the description of how many layers a network has, the number of neurons in each layer, each layer’s transfer function and how the layers are connected to each other. Neural network learning ability to extend its prediction power on data out of training data set is essential for successful implementation of artificial neural networks on friction characteristics prediction. Back-propagation training, used here, begins by presenting an input pattern vector to the network sweeping forward through the system to generate an output response vector and computing the errors at each output.

12

D. Aleksendrić and Č. Duboka

The error signal is a product of the difference between desired and actual output values. This error signal is then back-propagated to the lower layers in order to find out the global minimum of the error surface area. The results from Figure 9 show that the artificial neural networks can be used for developing of the neural model of friction material behaviour, not only for cold performance tests but also for fade and recovery behaviour and wear performance prediction. The developed neural models of the friction material behaviour have been able to generalise complex non-linear, multi-dimensional relationships which in this case relates 26 input parameters with only one output parameter. The influences of the composition of all the friction materials with 18 different ingredients, five most important manufacturing conditions, and three testing conditions have been integrated with variations of friction coefficient or brake factor C. Using neural models of the friction material behaviour in different working conditions of brakes offers possibilities for investigating braking system characteristics. It is especially important for introducing new sophisticated control systems for braking systems operation based on which operation of braking system transmissions can be further improved.

4

Conclusions

In recent years, much attention has been given to new computer technologies in an attempt to provide conditions for sustainable development of braking systems. The basic request is related to simulation of mechanical systems operation under real working conditions, and then to optimise them through a series of iterative procedures in a virtual environment. Based on such an approach, the methods for simulation of operation, visualisation of loads and prediction of braking systems performance in real working conditions have been developed in the FRIMEKS Laboratory for Friction Mechanisms and Braking Systems of the Faculty of Mechanical Engineering, University of Belgrade. Application of artificial technologies, as shown here, should enable integration of theoretical and experimental knowledge not only to ensure ‘virtuality’ in the process of braking system development, but also to attain a sufficiently high level of ‘reality’ in order to justify the reasons for integrated technologies application, i.e., to perform sustainable development. New systems engineering tools presented in this paper based on integration of artificial environment and intelligence were applied: •

to create and assemble parts into the brake digital mock-up, and later on into the braking system of a motor vehicle

•

to ‘instrument’ such Braking System Virtual Prototype by asking for certain outputs – putting together 3D design models and FEA models in such a way to enable not only virtual testing, but also introduction of friction and wear behaviour using artificial intelligence

•

to run a standard set of parametric design simulations or design-of-experiment tests, in addition to the drive through simulation tests which should correspond to expected braking cycle behaviour under inertia dynamometer conditions

•

to make design decision in coordination with requirements by means of the systems approach.

View publication stats