Influence Of Disc Topography On Generation Of Brake Squeal

Wear 225–229 Ž1999. 621–628

Influence of disc topography on generation of brake squeal

1

Filip Bergman, Mikael Eriksson ) , Staffan Jacobson ˚ Department of Materials Science, The Angstrom ¨ Laboratory, Uppsala UniÕersity, Box 534, S-751 21 Uppsala, Sweden

Abstract The influence of the disc topography on the generation of automotive disc brake squeal has been studied. Two brake discs were shot-blasted to produce small pits in the disc surface. The discs were then tested in a special brake squeal rig. During the tests, the coefficient of friction increased from about 0.3 to 0.45 as pits were gradually reduced in size as the discs were worn. A removable section in one of the discs made it possible to record the size and location of the surface defects by SEM-investigations before, during and after the test. For the tested padrdisc combination, there were no brake squeals generated as long as the friction coefficient was below a critical level of 0.4. The use of shot-blasted discs thus provides a unique possibility to investigate the correlation between brake disc topography, friction coefficient and brake squeal generation without changing neither the composition nor the macroscopic geometry of brake pad or disc. q 1999 Published by Elsevier Science S.A. All rights reserved. Keywords: Brake squeal; Coefficient of friction; Surface topography; Disc brake; Brake disc

1. Introduction Traditionally, brake squeal in automotive brakes has been studied as a phenomenon caused by the design of the brake system and the macroscopic friction properties. Many different mechanisms have been suggested as being responsible for brake squeal generation. Most of the work has concerned macroscopic frictional behaviour and commonly a stick–slip motion or a friction coefficient decreasing with sliding velocity has been blamed w1–7x. However, lately it has been shown that a brake system theoretically will generate squeal even with a stable friction coefficient, if it exceeds a certain level w4,8x. The question then turns over to why many commercial brake

)

Corresponding author. E-mail: [email protected] Prime NoÕelty of the Paper: It is shown that small pits in the brake disc surface lower the coefficient of friction. The friction coefficient gradually increases as the spits are worn away. Brake squeals are prevented as long as the coefficient of friction is below a critical value. This is a unique test for investigation of the correlation between disc topography, friction coefficient and brake squeal generation without changing any other test parameters. 1

systems do not generate squeal although the friction coefficient is above this critical level? The answer can probably be found within the microscopic friction characteristics of the brake, an aspect not included in the theoretical models and also almost neglected in the literature on brake squeal. In some technical areas, it is well known that rough surfaces excite less noise than do smooth. Despite this knowledge among practicioners, there is very limited information in the literature on its physical background. An earlier investigation has shown that the surface of the tested pads exhibit well-defined contact spots sliding against the disc. For an ordinary pad, it has been shown that the number of contact spots is as high as in the order of 10 5 w9x. These spots are considerably harder than the mean hardness of the pad and are composed mainly of structural fibres, abrasive particles and compacted wear debris. Obviously, the macroscopic friction force is the sum of the forces on the individual contact spots. The objective of this paper is to investigate the effect of a change in the microscopic friction conditions on the generation of brake squeal and the macroscopic coefficient of friction. The microscopic friction conditions were modified by the introduction of shallow pits evenly distributed over the disc surface. The gradual wear of the pits was repeatedly investigated in the SEM and correlated to the

0043-1648r99r$ - see front matter q 1999 Published by Elsevier Science S.A. All rights reserved. PII: S 0 0 4 3 - 1 6 4 8 Ž 9 9 . 0 0 0 6 4 - 2

F. Bergman et al.r Wear 225–229 (1999) 621–628

622

gradual change of squeal generation and macroscopic friction.

2. Test material and experimental One set of brake pads and two standard cast iron brake discs, both made for a front wheel of a Volvo 850, were first subjected to a run-in procedure of 42 brakings identical to one sequence of the following test program. Both discs were then shot-blasted with angular SiC-particles Ž f 1–2 mm. causing about 100mm wide pits in the disc surface. The pads selected are a slightly modified, noncommercial version of the production pads used in the Volvo 850. These pads are suitable for fundamental investigations of brake squeal as they generate squeal over a wide range of brake pressures and temperatures. The nominal composition of the discs is given in Table 1 and the approximate composition of the pads are shown in Table 2. The exact composition of the pad material was not available from the manufacturer. All tests were conducted in a special brake squeal rig w10x. It is based on a Volvo 850 front-wheel suspension with an electric motor rotating the wheel via the original drive-shaft, see Fig. 1. The rotational speed is controlled separately and is not influenced by friction forces as in a normal brake dynamometer. The brake-line pressure is controlled with a servo-valve and the brake torque is measured using a torque gauge mounted on the drive-shaft. Two squeal tests were performed, one for each disc. Basically, both tests were based on a test sequence consisting of 42 brakings, see Fig. 2a. This sequence was repeated 30 times for disc 1 in Test 1 and 14 times for disc 2 in Test 2, making it a total of 1260 and 588 brakings, respectively. Each single braking lasted for 20 s during which the rotational speed was reduced from 3 rps down to 1 rps, see Fig. 2b. The brake-line pressure was held constant during each braking but was shifted from one braking to another, following the sequence shown in Fig. 2a. Between consecutive brakings, there was a 100 s idle period to avoid overheating of the brake components. This test sequence was designed to cover a wide range of pressure–temperature combinations, including high pressures at low disc temperatures and vice versa. During each single braking the brake pressure, brake torque, rotational speed and emitted sound were measured once every third second, see Fig. 2b. Sound was registered

Table 1 Nominal compositions of the cast iron discs Element

C

Si

Mn

Cr

S

P

Ni

Wt.%

3.5

1.9

0.60

0.15

0.10

0.03

- 0.05

Table 2 Approximate composition of the pad Structural component

Ingredient

Amount wwt.%x

Fibres Matrix Žorganic.

Steel, aramid and glass fibres Binder Other Metallic Žbrass, bronze. Graphite Metal sulphides Quartz Žabrasive. Clay minerals, iron oxide

30 8 11 15 15 8 5 8

Friction modifiers

Filler

only when the sound level exceeded 84 dB at frequencies between 500 and 20,000 Hz. Test 1 was interrupted five times Žafter 4, 9, 16, 23 and 30 sequences. to make silicon rubber replicas of the disc surface. These replicas were later investigated using a white light optical profilometer. Test 2 was interrupted four times Žafter 1, 5, 10 and 14 sequences. to study the disc surface using a scanning electron microscope ŽSEM.. Micrographs were made using both enhanced topographical mode and enhanced atomic number contrast mode in order to study the shape of the pits and how they were filled up with wear debris. The replicas and the SEM-investigations were made to monitor the running-in of the shot-blasted surface and to correlate this to squeal generation. In order to facilitate the repeated SEM-investigations of the disc surface in Test 2 the disc was equipped with a removable section, see Fig. 3. This section could easily be remounted without noticeably affecting the sliding surface of the disc.

3. Results The coefficient of friction and the accumulated number of squeals for the full test programs are shown in Fig. 4. As a comparison, the squeal generation prior to the shotblasting is shown. Fig. 4d shows the gradual increase in the fraction of smooth area on the brake disc surface, that is part of the area not occupied by the pits. The surface area increases as the pits are worn away, more rapidly at the earlier stages of the test and considerably slower at the end. After 1260 brakings, the pits still cover approximately 13% of the surface. Further, the surface profiles obtained from the replicas revealed a maximum pit depth of 30–40 mm after 378 brakings and roughly 20 mm after 966 brakings. The average diameter to depth ratio was generally estimated to 5, i.e., the pits were five times as wide as they were deep. The gradual reduction of the size of the blast pits during Test 2, due to the wear of the disc surface is shown in Fig. 5. During the initial period, the pits exhibited diameters

F. Bergman et al.r Wear 225–229 (1999) 621–628

623

Fig. 1. Ža. Schematic of the test-rig. An electric motor drives the wheel via the original drive-shaft. Žb. Photograph of the rig. The torque gauge is mounted on the drive-shaft. The wheel that was mounted during the tests is removed in the photo to improve visibility.

Fig. 2. Ža. Squeal test sequence of 42 brakings at predetermined brake-line pressures. This sequence was repeated 30 and 14 times, respectively, to form the two complete test programs. Žb. Pressure and rotational speed during one braking. Squeal measurements are made every third second.

Fig. 3. The brake disc and its removable wedge used to facilitate SEM-investigation of the disc surface in Test 2.

624

F. Bergman et al.r Wear 225–229 (1999) 621–628

Fig. 4. Ža. The coefficient of friction and accumulated number of registered squeals in Test 1. Žb. The coefficient of friction and accumulated number of registered squeals in Test 2. The small step in friction after 420 brakings is a measurement artefact due to drift of the torque gauge between test sequences number ten and eleven and should be disregarded. Dashed vertical lines in Ža. and Žb. indicate where the tests were interrupted for replication or disc surface analysis. Žc. The run-in sequence prior to shot-blasting of the disc used in Test 1. Žd. Area fraction of disc surface not covered by pits and corresponding friction coefficient in Test 1.

from 10 up to 100 mm. Wear debris was collected and compacted in some of the smaller pits, as shown in the lower row of micrographs in Fig. 5. EDX-analysis showed that this compacted wear debris contained elements such as Al and Cu and a high concentration of carbon, all typical of the pad. As smaller pits were filled to a higher degree than larger ones, the share of filled pits increased as the disc was worn. In both tests, the friction coefficient was initially low Ž0.3. but quickly increased to 0.4 during the first 300 brakings. Even though the friction then started to stabilise,

it was still 25% below that of a nonprepared disc. As the tests continued, the discs became smoother with the friction gradually increasing. When the average friction level exceeded 0.4, squeal started to be generated more frequently. Fig. 6 shows that no squeals were generated at a coefficient of friction lower than 0.4. The same critical coefficient of friction was obtained in Test 2, see Fig. 4b. Squeals were first initiated at one specific brake-line pressure but with further friction increase the system started to generate squeals at other, neighbouring pressures as well. In Test 1, the first 50 squeals Žup to 350 brakings.

F. Bergman et al.r Wear 225–229 (1999) 621–628

625

Fig. 5. Gradual change of the disc surface and corresponding friction and squeal generation curves during Test 2. The top row of SEM-micrographs shows the topographical contrast and the bottom row the compositional contrast. Dark compositional contrast corresponds to wear debris from the pad that has become compacted into the pits. The test was interrupted for SEM investigation after 1, 5, 10 and 14 sequences, corresponding to 42, 210, 420 and 588 individual brakings, respectively.

were generated almost exclusively at a brake-line pressure of 8 bar, see Fig. 7. Between 350 and 450 brakings, squeal was mainly generated at 7 and 8 bar but also at 10 bar. After 450 brakings, squeal was generated also at 6, 9 and 11 bar. This should be compared to the run-in sequence prior to shot-blasting, where squeal was generated at all tested brake-line pressures except 3 and 4 bar.

4. Discussion The shot-blasted brake discs have proven an excellent tool for studying the influence of macroscopic friction on brake squeal generation. This technique allows the unique possibility to study a system where the friction coefficient gradually increases while the material combination and

other test parameters are kept constant. The results show three important features. Ž1. No squeals are generated when the coefficient of friction is below a critical value. The friction coefficient had to exceed 0.4 in both Test 1 and Test 2 before squeals were generated. This corresponds well to the rule of thumb, practiced within the brake industry, stating that brake pads with a high coefficient of friction also are more prone to generate squeals. In the literature, several reports have concerned the influence of macroscopic friction on squeal generation. Commonly a stick–slip motion or a friction coefficient decreasing with sliding velocity has been blamed w1–7x. Recently, a numerical model by Hulten ´ suggests that disc brake squeals can be generated also when the coefficient of friction is velocity independent w8x. This model suggests that the vibration in the pad is a nonsynchronous wave

626

F. Bergman et al.r Wear 225–229 (1999) 621–628

motion rather than a simple harmonic oscillation and that no instabilities Ži.e. brake squeals. occur when the coefficient of friction is below a certain value. The model is limited to two dimensions while optical measurements have shown brake pad vibrations to be rather complicated with both bending and torsional movements w11,12x. A full description would thus require 3D-modelling. However, within the limitations of Hulten’s model, our findings ´ basically support the assumption of a critical friction coefficient for brake squeal generation. However, we have previously shown that brake pads with similar friction coefficient values can have very different squeal propensities w13x. Thus, the value of the friction coefficient threshold must depend on other parameters such as pad and disc geometry, state of wear, stiffness, modulus of elasticity, damping and the frictional properties on both macroscopical and microscopical level. Accordingly, the present experimental threshold value of 0.4 should be regarded as unique for this special combination of pad and disc. Ž2. A very small friction increase may lead to a dramatic increase in squeal generation. The present investigation has clearly demonstrated the sharpness of the friction threshold for squeal generation. The increase in the average friction coefficient between braking no. 200 and 500 was less than 0.03 while the generation of squeals increased dramatically. At the threshold value, the system squealed only at one specific brake-line pressure. As the friction increased, squeals were generated at a growing number of pressures, as shown in Fig. 7. Ž3. The shot-blasted brake disc has a lower coefficient of friction than a regular disc. It is noteworthy that the uneven, shot-blasted discs exhibited lower friction coefficients than do normal smooth Fig. 7. Brake-line pressures that generated squeals. Ža. During the run-in sequence prior to shot-blasting of the disc used in Test 1. Žb. During Test 1 Žshot-blasted disc..

Fig. 6. Friction level for the registered squeals during Test 1. Note that all squeals were registered at a friction coefficient above 0.4.

discs. As the discs were gradually worn, the pits diminished and the friction coefficient slowly approached the level of smooth discs. Typically, the coefficient of friction also increases during each individual braking, see Fig. 8. This behaviour reflects that it takes some time for the pad and disc surfaces to adapt to the new contact situation, and that the adapted contact will exhibit a higher friction force. The explanation to these two friction variation phenomena is to be found in the nature of the contact between disc and pad. An earlier investigation has shown that the surface of the tested pads exhibit well-defined contact spots sliding against the disc w9x. Each spot exhibit a very inhomogeneous structure with widely varying mechanical properties. The spots are composed mainly of structural fibres Žtypi-

F. Bergman et al.r Wear 225–229 (1999) 621–628

Fig. 8. Example on the increase in the coefficient of friction during individual brakings. ŽThe examples are from brakings no. 279 and 280 in Test 1..

cally steel., abrasive particles Žtypically alumina. and compacted wear debris Žmainly polymeric.. They typically have a diameter of 100 mm, a height of 1 to 10 mm and are separated by a few hundred microns, see Fig. 9. The total number of spots on a pad is around 10 5 and hence contact pressures are typically very low Že.g., 4 MPa at a brake-line pressure of 10 bar.. The spot size is not constant, but can grow by the addition of more compacted debris, etc. The ‘steady state’ size depends on the contact situation Žbrake pressure, temperature, surface topography, etc.. and the pad composition. The main mechanism of friction is believed to be shearing of the top layer of these contact spots in their motion relative to the disc. Given this general description of the contact situation, the mechanism explaining the lower friction against the

627

shot-blasted disc surface with shallow pits must be either a decreased contact area or a reduced shear stress within the contact area. Is there a mechanism by which the presence of pits could reduce the shear stress within the areas of real contact? Probably, the pits could increase the amount of wear debris or alter the debris morphology. This would change the conditions in the sliding interface and possibly reduce the friction. However, simple tests of cleaning the disc surface from wear debris and surface contaminants by rubbing with a cloth soaked in acetone or alcohol did not result in any noticeable friction increase. This indicates that any difference in amount of wear debris in the contact area has limited effect, or only a short-lived effect. As reported by the present authors w14x, decreasing the nominal brake pad area down to 50% only has marginal effect on the friction coefficient. Thus, the brakes behave as expected for materials following the classical friction laws. This indicates that the presently achieved friction reduction is not due to the pits causing a reduced nominal contact area. Based on this discussion, both the low initial friction in each individual braking and the low friction against the shot-blasted surface, in general, are proposed to be due to a reduced real contact area. It is believed that for each new braking the pad surface has to adapt geometrically to the new position on the disc. This geometrical adaptation will take some time and results in a gradual increase of the real contact area and subsequently an increased friction force. The low friction against the shot-blasted surface, in general, is proposed to be due to the frequent encounters between the contact spots and the pits in the disc surface. These encounters damage the contact spots thus preventing them from growing to their normal ‘steady state’ size. By this mechanism, the real contact area is kept smaller than

Fig. 9. Ža. SEM-micrograph of the disc surface taken after 420 brakings in Test 2. Žb. SEM-micrograph of a pad surface. The contact spots, the flat islands in the picture, are mainly constituted by metal fibres, abrasive particles and compacted wear debris.

628

F. Bergman et al.r Wear 225–229 (1999) 621–628

against a smooth disc and the friction force thereby stays lower.

ing test materials and Claes Kuylenstierna for valuable discussions.

5. Conclusions

References

The run-in of a shot-blasted disc has been shown to be a unique test where the effects of a slowly increasing friction coefficient on squeal generation can be studied for one specific disc–pad combination while other test parameters are kept constant. Brake squeal will not be generated if the coefficient of friction is kept below some critical level. For the tested padrdisc combination, this level was found to be 0.4. Also a small increase in the coefficient of friction in the close vicinity of the critical level causes a dramatic increase in squeal generation, indicating some threshold behaviour of the excitation mechanism. The small impressions caused by shot-blasting of the disc lowers the macroscopic friction coefficient and can thus help to prevent brake squeals. The friction reduction is proposed to be due to a reduction of the real contact area. The frequent encounters with the pits in the disc surface damage the contact spots, preventing them from growing to their normal ‘steady state’ size. By this mechanism, the real contact area is kept smaller than against a smooth disc and the friction force thereby stays lower. When the pits are worn down to become small and shallow, material originating from the brake pads is compacted in the pits making the surface virtually flat. This surface still reduces the friction by 25% as compared to a conventional, smooth disc. The squeal reducing effect from shot-blasting is limited in time due to the wear of the disc surface.

w1x S.K. Rhee, P.H.S. Tsang, Y.S. Wang, Friction-induced noise and vibration of disc brakes, Wear 133 Ž1989. 39–45. w2x M.R. North, Disc brake squeal, in: Braking of Road Vehicles, Loughborough, United Kingdom, 1976. w3x S.W.E. Earles, M.N.M. Badi, Oscillatory instabilities generated in a double-pin and disc undamped system: a mechanism of disc-brake squeal, Proc. of the Institution of Mechanical Engineers 198 Ž4. Ž1984. 43–50. w4x J.O. Hulten, ´ Friction phenomena related to drum brake squeal instabilities, in: ASME Design Engineering Technical Conferences, ASME, Sacramento, CA, 1997. w5x T. Borchert, Dynamical behaviours of the disc brake pad, SAE Technical Papers Series 912656, 1991. w6x R.A. Ibrahim, Friction-induced vibration, chatter, squeal, and chaos: Part II. Dynamics and modeling, friction induced vibration, chatter, squeal and chaos, American Society of Mechanical Engineers, Design Engineering Division 49 Ž1992. 123–138. w7x N. Millner, An analysis of disc brake squeal, SAE Trans. 87 Ž1978. . w8x J. Hulten, ´ Drum brake squeal—a self-exciting mechanism with constant friction, in: SAE Truck and Bus Meeting, SAE, Detroit, MI, USA, 1993. w9x M. Eriksson, F. Bergman, S. Jacobson, Surface characterisation of brake pads after running under silent and squealing conditions, in: S.S. Eskildsen ŽEd.., Proc. Nordtrib’98, Ebeltoft, Denmark, DTI Tribology Centre, Aarhus, Denmark, 1998, pp. 657–663. w10x F. Bergman, M. Eriksson, S. Jacobson, A software based measurement system for test and analysis of automotive brake squeal, Submitted to TriboTest, 1997. w11x A. Felske, G. Hoppe, H. Matthai, ¨ Oscillations in squealing disk brakes—analysis of vibration modes by holographic interferometry, SAE Technical Paper Series 780333, 1978. w12x J.D. Fieldhouse, T.P. Newcombe, Double pulsed holography used to investigate noisy brakes, Optics and Lasers in Engineering 25 Ž1996. 455–494. w13x F. Bergman, L. Gudmand-Høyer, M. Eriksson, S. Jacobson, The effect of Cu 2 S, PbS, Sb 2 S 3 solid lubricants on the occurence of brake squeals for three automotive brake pad matrix types, in: S.S. Eskildsen ŽEd.., Proc. Nordtrib’98, Ebeltoft, Denmark, DTI Tribology Centre, Aarhus, Denmark, 1998, pp. 665–672. w14x F. Bergman, M. Eriksson, S. Jacobson, The effect of reduced contact area on the occurence of brake squeals for an automotive disc brake pad, To be submitted to Journal of Automobile Engineering.

Acknowledgements The authors gratefully acknowledge the financial support from the Swedish board for Technical Development ŽNUTEK., Volvo Technological Development for provid-