Engineering Failure Analysis 13 (2006) 933945

www.elsevier.com/locate/engfailanal

Failure analysis of a motorcycle brake disc

M. Boniardi a, F. DErrico
a

a,*

, C. Tagliabue a, G. Gotti b, G. Perricone

b,1

Department of Mechanics, Politecnico di Milano, Via La Masa 34, 20156 Milano, Italy
b
Brembo S.p.A., Viale Europa 2, 24040 Stezzano (BG), Italy
Received 7 April 2005; accepted 30 April 2005
Available online 13 September 2005

Abstract
A certain number of motorbike disc brakes, made of stainless steel, shown the presence of small cracks only after a
few thousands miles. These cracks were mainly located nearby the holes placed on ange to ventilate and refresh pads.
According to results, the deterioration can be led back to thermal cyclic strain (related with the heatingcooling
cycles developed during the brake action) superimposed to the mechanical strain caused by braking torque.
This work analyses the aforesaid disc brakes investigating both the main causes and the evolution of its deterioration
in order to nd out possible solutions. The short lifespan of such discs has to be ascribed to the rapid decay of the
mechanical properties of the manufacturing material. Material decay is liable for starting cracks.
Several actions could be chosen to face this problem. The choice of a particular chemical composition, which will be
demonstrated to be unt for the purpose, produced an extreme tempering of the steel as a direct result of its protracted
exposure to high temperatures (a situation which can be considered usual referring to disc brakes).
In this work, we present the eect of choosing a dierent kind of steel, characterized by a greater resistance to the
tempering processes.
2005 Elsevier Ltd. All rights reserved.
Keywords: Fatigue; Thermal fatigue; Metallurgical examinations; Brake failure

1. Introduction
A motorbike brake disc is composed of two main parts, a ange and a bridge (Fig. 1). The former faces
pressure from pads and therefore is subjected to higher temperatures during braking; the latter works in
cooler conditions and has the function of transferring brake torque to the wheel hub.
*

Corresponding author. Tel.: +39 02 2399 8215; fax: +39 02 2399 8202.
E-mail addresses: fabrizio.derrico@polimi.it (F. DErrico), guido_perricone@brembo.it (G. Perricone).
Tel.: +39 035 605 700; fax: +39 035 605 605.

1350-6307/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2005.04.012

934

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

Fig. 1. Scheme of a motorcycle brake disc.

The degree of reliability of the braking action is directly proportional to the rmness of the entire braking system. This means that the friction coecient between the disc and the braking pad must be kept as
much as possible constant, under any condition.
Hence, the rotor is designed to withstand both the maximum possible deceleration (emergency braking)
and a series of braking cycles. The main problem to deal with in braking cycles is the control of the fading
phenomena. High frequency braking cycles lead a raising of the disc temperature. Therefore, after a period
of temperature growth, the disc will reach its balance between the heat generated by braking action and the
dissipated heat. At the balance a steady state temperature is reached. If the steady state temperature is too
high, several problems happen among which a decay of friction coecient and a not constant behaviour
between following brakings.
To face this problem, designers focus on disc performances operating with high temperatures (500
600 C or more). Discs are then checked on a test bench which simulates extreme braking conditions, such
as an alpine downhill or a sport race.
Other issues are wear and corrosion. To prevent problems with these phenomena, disc brakes are usually
made of martensitic stainless steel. Because of their narrow thickness (usually less than 10 mm) discs are
generally obtained by milling of at hot rolled sections. These sections are milled in the annealed (normalized) condition. Then, are subsequently quenched and nished o.

2. The case history

In our analysis, we worked on two brake discs (280 mm diameter, 8 mm thickness) made of martensitic
stainless steel. The discs were produced using two similar types of steel coming from two dierent suppliers
(from now on, named A and B). They could both be compared to the AISI 410 standards, except for some
chemical modications aimed at gaining the hardening response. In fact, in brake applications, a surface
hardness of 3236 HRC (in the quenched condition) is required.
The B type discs show the presence of small cracks (detailed in Fig. 2) after a few thousands miles of use,
whereas in the A type this deterioration is not noticeable [1]. These cracks were clearly visible even with the
naked eye. They triggered o from the ventilation holes, spreading for some millimeters in the radial
direction.
A deeper analysis was necessary to investigate and understand the dissimilar behaviour of the two kinds
of discs.

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

935

Fig. 2. Detail of a crack placed on a type B disc.

To exclude from the analysis the potential problems ascribable to the applied milling process, some prototypes were obtained by laser cutting with regards to the cutting method, the behaviour did not show any
variation, since the B type discs failed independently from the cutting process.
The analysis includes both the unused samples of the A and B type, and some used and cracked B type
samples, which from now on will be named BU.

3. Observations and results

3.1. Visual and chemical analyses
Every BU sample showed cross cracks arising from the more external holes. As depicted in Fig. 3, cracks
were aligned with holes fastening the disc to wheel hub and they are the only type of cracks occurred.
The uniform colour of the braking surface demonstrated at least a 500550 C working temperature.
Quantometric analyses were carried out on the two rotors samples (A and B) to determine their chemical
compositions. Results are reported in Table 1. It can be observed that the A and B steels were two particular kind of martensitic stainless steel, similar to the AISI 410, except for some chemical dierences not
declared by manufacturers.
The main dierence was the greater content of molybdenum and vanadium of the A type steel, and the
higher manganese percentage of B steel. Usually, the least expensive means of increasing hardenability at a
given carbon content is by increasing the manganese content [2].
3.2. Selection and removal of samples
The A, B and BU discs were sectioned by band saw, paying attention to have a constant lubrication and
cooling. Some metallographic samples were obtained, polished and chemically etched (glyceregia, 3 parts
glycerol, 3 parts HCl, 1 part HNO3) to reveal their microstructure.

936

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

Fig. 3. BU rotor characterized by two cross cracks highlighted by the circles.

Table 1
Comparison of chemical composition of type A and B discs

Disc A
Disc B

Si

Mn

Cr

Ni

Mo

Al

Cu

Ti

Nb

Sn

0.078
0.066

0.38
0.36

0.86
1.51

0.031
0.019

0.003
0.004

12.30
12.16

0.124
0.084

0.100
0.026

tr.
tr.

0.068
0.066

0.003
0.002

tr.
tr.

0.100
0.038

tr.
tr.

0.003
0.001

The BU-1 and BU-2 samples were extracted from the BU disc nearby the holes involved by the two
cracks (Fig. 4).
The BU-1 sample was mechanically opened, paying attention not to damage the crack surfaces. Subsequently the fracture surface was investigated with a scanning electron microscope (SEM).
3.3. Metallography and microstructures
Fig. 5 shows an example of the microstructures revealed on the A, B and BU disc surface, nearby the
ange centre. From the metallographies reported in Fig. 5(a) and (b), it can be inferred that the microstructure was martensite not tempered. This microstructure is peculiar of a quenching heat treatment used for
giving hardness to the ange surface. A quite dierent martensitic structure can be noted in Fig. 5(c): it
appears less rened and with a certain amount of widespread carbides.
The area close to disc spokes (spokes are the building blocks of the brake bridge), was analyzed too and
the microstructures are shown in Fig. 6.
For the A sample, a ferritic microstructure was observed, as shown in Fig. 6(a). It shows a grain rening
(5 lm mean length) by recrystallization due to the hot rolling. The microstructures of the B and the BU

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

937

Fig. 4. BU disk sectioned; the BU-1 and BU-2 samples chosen for the metallographic and fractographic researches are highlighted.

spokes, reported in Fig. 6(b) and (c), respectively, were ferritic too. They revealed the same recrystallization
phenomenon but with a grain average dimension between 20 and 30 lm. The recrystallization phenomenon
is better visible at lower magnication. A more accurate hot rolling could explain the dierence between the
A and B sample grain size. Cracks happened however faraway from spokes where microstructures are
martensitic.
Therefore, going from ange to spokes, substantial changes in microstructures were noticeable. Such
changes suggest that only the ange was subjected to quench hardening: probably an induction hardening.
In fact, the ange had a martensitic structure, while the bridge had a ferritic and normalized one. Finally,
Fig. 7 shows a metallography performed on the BU-2 sample which highlights the crack features: the pattern is essentially transgranular and at, as expected in fatigue crack propagation.
3.4. Hardness tests and results
Hardness was measured on all the three kinds of discs. The measure was carried out both on the entire
ange and spokes area. Lower hardness values were measured on the spokes, due to the softer ferritic structures: this is in line with the previous metallography analyses.
Figs. 8 and 9 report the hardness values and the relative measurement positions for the brand new A and
B discs. Flange areas of A and B types exhibited similar values, proving that both the discs types were
subjected to analogous heat hardening treatments. Instead, the analyzed BU sample shown a considerable
decrease of hardness values in several positions, as summarized in Fig. 10.
To check that the hardness decay registered in the centre of the ange was not due to a local over-heating, the hardness measurements on the BU disc was extended in the circumferential direction too. This
analysis conrmed that the phenomenon was not local, but rather symptomatic.
3.5. Fractographic analysis
Fig. 11 reports a microfractography of the BU-1 sample fracture surface. It is possible to distinguish
clearly into two dierent zones. The former is planar, in which the crack generates a fatigue pattern (the
direction of which is indicated by the pointer in the illustration). The latter nail-shaped was caused by
the nal mechanical opening of the crack exerted in the laboratory.

938

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

Fig. 5. Microstructure of A (a) and B (b) an BU (c) samples revealed in correspondence of the center of the ange (positions 3).

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

Fig. 6. Microstructures of A (a), B (b) and BU (c) ange discs.

939

940

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

Fig. 7. Microstructure nearby the crack in the BU-2 sample.

Fig. 8. Hardness measurements of the A-1 and A-2 samples.

Some remarks can be inferred from the examination of the surface: oxides can be noticed in the A zone,
strengthening the hypothesis that identies the heat fatigue as the prior damage mechanism.
Fig. 12 shows a magnication of two details, A and B, taken from Fig. 11: here are highlighted the crack
stopping lines. Between them there is an average distance of about 40 lm.
3.6. Test of resistance to tempering
In order to investigate the residual stress eld, X-ray diractometric analyses were performed on the
three disc ange surfaces. The goal was to be sure that the failure was not inuenced by prior tensile stresses. A not detrimental compressive state of stresses was revealed. The average value was 270 MPa for the
disc A, 277 MPa for the disc B, 540 MPa for the disc BU.

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

941

Fig. 9. Hardness measurements of the B-1 and B-2 samples.

Fig. 10. Hardness measurements of the samples and of the BU-2.

The resistance to tempering of the A and B discs was analyzed by means of experimental tests carried out
in a laboratory furnace. In order to do that, some samples were cut from the hardened ange of the two
discs.
The tests consisted of putting two samples of A and B materials into the furnace, exposing them to three
constant temperature values (450, 500 and 600) with variable time intervals (0, 5, 6, 12, 24 h). After every
test, the samples were pulled out from the furnace to allow air cooling and then their hardness was
measured. Fig. 13 shows the obtained results. The study of the diagram shown that, if exposed to
450 C, neither the rst nor the second type of discs, reveal tempering phenomena. But after 12 h of
exposure to 500 C, the B sample was tempered, whereas the A sample managed to keep almost the same
original hardness.

942

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

Fig. 11. BU-1 sample crack surface: (a) comprehensive view of the three area A, B, C (crash breaking caused by the mechanical
opening of the crack); (b) detail of area A, in which can be clearly noticed the presence of widespread heat oxidation phenomena; (c)
detail of area B, involved in the nal phase of the crack propagation.

4. Analysis of results and discussion

During braking, the disc surface is subjected to strains which are sum of two components. The rst component is a mechanical strain, due to brake torque and pad friction. The second one is induced by the thermal strains produced by unavoidable temperature gradients present on the disc.
A schematic representation of the mechanical loading of a disc is reported in Fig. 14. When the disc is
moving during a brake, the compressive circumferential stresses reach their highest values in the small area
approaching pads. Highest values are located in the centre of the ange due to the particular pressure distribution. As soon as ange passes across pads, the material is loaded in tension. Increasing the distance
from pads, mechanical stresses rapidly decrease.
Therefore in braking, at each disc rotation, areas around holes are subjected to alternative (tension to
compression) cyclic stresses. Moreover, in traction situation, the stress intensication factor on the hole

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

943

Fig. 12. Magnication of the B and C areas from Fig. 11 (the crack arrest lines are underlined).

is about 3 times the nominal stress in ange [3]. As already said above, another phenomenon is superimposed to enable the crack formation under fatigue-in-temperature condition.
In fact, thermal strains develop because heating by friction are not uniform. Smaller or greater thermal
gradients depend on many factors like the disc geometry, braking pressure, caliper type, stopping interval
between two consecutive brakes, etc.
This means that holes are also subjected to heatingcooling cycles at each rotation. This phenomenon
produces thermal dierential strains of contiguous material portions. Therefore at each rotation, the portions of material nearby the holes are subjected to thermal-fatigue phenomena.
It sounds clear that dierent boundary conditions may anticipate crack start and shorten crack propagation period. For example, this may occur as a consequence of an increase in torque brake (similar to an
emergency brake), an increase in braking frequency (as it happens in ABS conditions) or for material
fatigue limit decay. Decay in the fatigue limit may be caused when high temperatures are developed or when
the material itself exhibits poor resistance to the operating conditions. This last case can be well related to
the evidences obtained on the tests carried out on the A and B discs.
In fact, the comparison between the microstructures observed on the braking anges revealed a greater
concentration of carbides for the B type disc, according to its higher sensitivity to tempering.

944

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

Fig. 13. Hardness diagram, considering time and exposure temperature.

Fig. 14. Scheme of the mechanical loading exerted during the braking action.

So, the protracted high temperature exposure caused the tempering of the B type disc, while the A type
did not show any kind of damage linked to the temperature exposure. Tempered steel loose its mechanical
property given by the heat treatment because the microstructure was changed. It is useful to underline that
all the kinds of fully hardened steels are potentially inclined to tempering phenomena when heated, but
each of these may exhibit dierent behaviours at the same time-temperature value [4].
Among the martensitic steels, the AISI 410 type is the most used. Unfortunately, together with the 403
and the 416 types, is the one which reveals the fastest decay of mechanical properties associated with the
raising of the temperature; whereas the 422 type retains a good mechanical behaviour up to 650, a resistance which is mainly due to the addition of molybdenum, vanadium and tungsten [2].

M. Boniardi et al. / Engineering Failure Analysis 13 (2006) 933945

945

According to these elements, the choice of the most appropriate material for the construction of brake
discs must take into account surely an elevated hardness, necessary to avoid premature wear damages, but
also it must keep this hardness constant during high temperature working cycles. So, it is required the use of
materials with a good resistance to tempering, at least in the conditions in which they will be likely to work.

5. Conclusions
The lifespan increase of a motorcycle brake disc depends strictly on the geometry (position of holes,
shape of spokes, etc.), the material properties at high temperatures and operating conditions.
In particular, operating under stable loading conditions and disc geometry, the inadequate material can
drastically reduce the disc lifespan.
With regard to the work presented herein it is possible to conclude that:
 The cracked disc was made of steel similar to the AISI 410 type, apart from some chemical dierences
and was subjected to the full heat tempering treatment on the ange.
 The cracks were a consequence of the excessive tempering of the martensite, due to the high temperatures
working conditions, and proved by the precipitation of carbides and hardness decay. Consequently the
steel was subjected to a worsening of its mechanical properties.
The inadequacy of the B type martensitic steel, with such strongly marked tendency to tempering is evident. According to this fact all the other components of the same material should be replaced to avoid the
same problem.
The A type steel, which is another variation of the AISI 410, can be chosen as a suitable substitute. The
reason why it must be considered an improvement derives from its greater amount of alligants such as
vanadium and molybdenum, which enable the material to strengthen its resistance to tempering.

References
[1] Sasada M, Okubo K, Fujii T, Kameda N. Eects of hole layout, braking torque and frictional heat on crack initiation from small
holes in one-piece brake discs. Seoul 2000 FISITA World Automotive Congress # F2000G331; 2000.
[2] ASM Handbooks. Properties and selection: irons, steels, and high performance alloys. vol. 1. Materials Park, Ohio: ASM
International; 1990.
[3] Pilkey WD. Petersons stress concentration factors. 2nd ed. Wiley; 1997.
[4] Thelning KE. Steel and its heat treatment. 2nd ed. London: Butterworths; 1984.