- Email: [email protected]
Analyzing the mechanisms of thermal fatigue and phase change of steel used in brake discs
Analyzing the mechanisms of thermal fatigue and phase change of steel used in brake discs Zhiqiang Li, Jianmin Han, Zhiyong Yang, Weijing Li PII: DOI: Reference:
S1350-6307(15)30001-7 doi: 10.1016/j.engfailanal.2015.07.002 EFA 2603
To appear in: Received date: Revised date: Accepted date:
28 March 2015 8 June 2015 1 July 2015
Please cite this article as: Li Zhiqiang, Han Jianmin, Yang Zhiyong, Li Weijing, Analyzing the mechanisms of thermal fatigue and phase change of steel used in brake discs, (2015), doi: 10.1016/j.engfailanal.2015.07.002
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Analyzing the mechanisms of thermal fatigue and phase change of steel used in brake discs
T
Zhiqiang Li, Jianmin Han*, Zhiyong Yang, Weijing Li
SC R
100044, P.R. China
IP
School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing
* Corresponding author. Tel.: +86 10 51683300; fax: +86 10 51683300. E-mail address: [email protected] (Jianmin Han)
NU
Abstract
Thermal cracks on the friction surface of railway brake discs can develop during
MA
their lifespan. Cracks often initiate after severe braking conditions along with the occurrence of hot spots. The cyclic thermal and mechanical loads causes high temperature, plastic strain and even phase change of the brake disc steel. In this paper,
D
full scale emergency braking tests were conducted and the peak temperature of
TE
localized area was found exceeding the austenitizing temperature of the steel. Thermal
CE P
cyclic tests was performed to simulate the temperature variation during braking. Volume change of the steel caused by microstructure transformation was taken into consideration in numerical simulation. Combining with the fracture behavior of brake
AC
disc steel in low cycle fatigue (LCF) tests in different temperature level, the simulation results show a good consistency with the results of microstructure observation and crack initiation. The occurrence of embedded crack could be well explained according to the simulation results and low cycle fatigue test results. Keywords: brake disc; phase change; thermal fatigue; numerical simulation.
1. Introduction Brake discs used on railway vehicles are crucial safety components which transform the kinetic energy created during braking into heat by the way of friction. After a period of usage, thermal cracks can be observed on the friction surface of brake discs. During repeated braking cycles, cyclic thermal and mechanical loads were applied on the friction surface[1-3]. For a severe braking condition, plastic strain
ACCEPTED MANUSCRIPT occurs on the friction surface of brake disc at high temperature levels and it causes irreversible damages for the material. Cyclic thermal plastic strain and residual stress are considered as dominant factors which initiate and cause growth of thermal cracks.
IP
T
Samrout and El Abdi[4, 5] have studied fatigue behavior of the brake disc steel for French T.G.V. under the thermal mechanical cyclic loadings and an anisothermal
SC R
elastoplastic model was proposed to predict the response of brake disc. Dufrénoy et al. [6, 7] investigated the damage mechanisms of brake discs. The cyclic stress-strain behavior during braking was studied and a thermo-mechanical model was proposed to
NU
describe the thermal fatigue behavior of brake disc material. Therefore, low cycle fatigue behavior of the studied steel is worthy to be paid attention for investigating the
MA
crack initiation and growth. In previous study[8], LCF tests under different temperature level and cyclic strain were conducted from room temperature (RT) to
D
600 ºC. LCF properties of the brake disc steel were obtained and fracture mechanisms
TE
of the studied steel were discussed.
In some cases, non-uniform contact conditions exist between the brake pad and
CE P
brake disc such that hot spots occur on the friction surface of the brake disc[9]. Instantaneous peak temperatures in these localized areas can exceed 900 ºC which is over the austenitizing temperature of the brake disc steel[10]. The great temperature
AC
fluctuation in the overheated area during braking and cooling causes microstructure transformation of the steel. Mori et al.[11] investigated the microstructure change of brake disc steel with the effect of the thermal load during wear and found the precipitated carbides were the initiation sites of the thermal cracks. Sakamoto[1] also found tempered martensite formed in the heated area on the friction surface. The stress intensity factor was calculated based on the finite element (FE) simulation results to investigate the crack propagation behavior. However, the resulting volume change due to microstructure transformation can also change the stress distribution on the friction surface of brake disc. Stress redistribution caused by microstructure transformation leads to different mechanisms result in crack initiation and growth. In this paper, the continuous braking tests were performed on a full scale dynamo testing machine in order to simulate the emergency braking (EB) processes. The
ACCEPTED MANUSCRIPT microstructure transformation of the brake disc steel was investigated by conducting the simulated thermal cyclic tests. The stress redistribution introduced by phase change were analyzed in finite element (FE) analysis. Combining with the low cycle
IP
T
fatigue (LCF) behavior and microstructure observations of the brake disc steel, the result of FE simulation can reflect the influence of microstructure transformation in
SC R
stress distribution and fracture behavior of brake discs. The related research works can provide a better understanding of potential threat caused by high energy braking. 2. Materials and experimental procedures
NU
For the study, a forged Cr-Mo-V low alloyed steel brake disc was tested on a full scale dynamo testing machine (shown in Fig. 1) to simulate the emergency braking
MA
(EB) processes. An electroslag remelted steel containing 0.32% C (weight percent) was used and other chemical composition is comparable to a 28CrMoV5-8 steel alloy.
D
The copper matrix powder metallurgy brake pads (shown in Fig. 2) were used in the
TE
full scale dynamo braking tests. For the brake pad material, the content of copper-tin alloy is 60%~70% as the matrix, the content of graphite as the lubrication component
is 15%~20%.
CE P
is 12%~18% and the content of the Al2O3 and SiC powder as the friction components
The braking conditions simulated were as follows. The running speed of the
AC
railway vehicle was equivalent to 200 km/h with an axle load of 16,000 kg. Braking pressure was chosen as 25 kN. In order to simulate a severe braking condition, three continuous emergency braking tests were conducted. For each braking cycle, there is a parking stage (P) and air-cooling stage following each emergency braking (EB). During the three continuous emergency braking tests, acceleration among each braking was 0.3 m/s2 and the convection cooling condition was considered. The variation of running speed and braking pressure of the three continuous braking tests is plotted in Fig. 3. Temperature at the mean friction radius position on friction surface of the brake disc was recorded continuously using an infrared thermometer and the video of braking process was recorded using a digital camera (shown in Fig. 1). Two thermocouples were embedded at the depth of 6 mm from the friction surface of the brake disc. Temperature at the mean friction radius (Position 1 in Fig. 4) and
ACCEPTED MANUSCRIPT the outer diameter of the friction surface (Position 2 in Fig. 4) were measured using these two thermocouples respectively. Temperature distribution on the friction surface of brake disc was recorded by an infrared thermal camera during braking process.
T
The continuous cooling transformation (CCT) curves of the studied steel were
IP
obtained using a Φ4 mm × 10 mm specimen on a thermal dilatometer (DIL805A).
SC R
Then thermal cyclic tests were conducted on the thermal dilatometer according to the temperature variation on the friction surface during full scale braking tests. Thermal
measured continuously during the test.
NU
strain could be calculated basing on the change in length of the specimen which was
Based on the measured temperature level obtained from full scale braking tests,
MA
strain controlled LCF tests were carried out from room temperature to 800 ºC. The total axial strain amplitudes were controlled at levels of ±0.8%, ±0.6%, ±0.4%, and a
D
lower level from 0.1% to 0.3%. The LCF tests were carried out using a
TE
servo-hydraulic machine (MTS-810) equipped with a temperature controlled furnace. The LCF test specimens (shown in Fig. 5) were machined according to Chinese
CE P
National Standard GB/T 15248-2008 ‘‘The Test Method for Axial Loading Constant Amplitude Low-Cycle Fatigue of Metallic Materials’’[12]. The specimens with a diameter of 6.35 mm and gage length of 25 mm were finely polished before testing.
AC
Two replications were tested at each strain amplitude with a fully push–pull mode (strain ratio R = -1). A triangular waveform signal was used and the strain rate was 5.0 ×10-3 s-1. The longitudinal strain was measured continuously with an extensometer and the plastic strain range was measured. The stress ranges were examined by the peak and valley stress values in each strain cycle. After the tests were completed, microstructures and fracture surfaces of selected specimens were examined using a scanning electron microscope (SEM). Cyclic strain-stress behavior and fracture mechanism of the brake disc steel were studied. The brake disc after braking tests was cut into pieces and the cross-section near the friction surface was polished and etched using a nital etchant (4% nitric acid 96% ethanol). Microstructure and Vickers hardness of both the picked specimens of brake disc and the thermal cycle test specimen were observed and measured. The fracture
ACCEPTED MANUSCRIPT mechanism of the brake disc was analyzed and compared with the fracture behavior of the LCF test specimen under the corresponding temperature level. 3. FE simulation methods
IP
T
The emergency braking process was simulated using ANSYS code and a 1/10 circumferential section model was used in the simulation considering the axial
SC R
symmetry of the brake disc structure. Sequentially coupled thermal-stress analysis was used for the simulation of braking process. For the studied brake disc steel, the mechanical and physical properties were obtained in previous work [8]. Mechanical
NU
properties of the steel obtained by tensile tests from room temperature to 800 ºC are shown in Fig. 6. The physical properties of the material used in the simulation are
MA
listed in Table 1.
According to the hot spots distribution observed during the full scale dynamo
D
braking tests, heat flux applied onto the overheated area of the friction surface could
TE
be inversely calculated. By converting the kinetic energy into friction heat, the heat flux input can be described as:
CE P
q r , t P0 t r
(1)
where is the heat-partitioning factor representing the fraction of frictional heat
AC
flux entering the brake disc[13], is the friction coefficient between brake disc and pads, P0 is the braking pressure, and t is the angular velocity of brake disc. Convection heat transfer coefficient on the surface of brake disc model is expressed using a function of speed of the railway vehicle. Based on previous studies, CFD analysis was conducted to investigate the heat transfer coefficients of brake discs with different structures [14, 15]. The relationship between heat transfer convection coefficient and train speed for different surfaces is plotted in Fig. 7. For the 1/10 circumferential section model, symmetry constraints were applied on the section plane. Displacement constraints were applied on the surface of the bolt hole to restrain three degrees of freedom of the nodes in the mechanical simulation. When the peak temperature exceeds the austenizing temperature, volume change
ACCEPTED MANUSCRIPT of the brake disc steel should be considered due to phase transformation. Instantaneous coefficient of linear expansion can reflect the volume change of material during phase transformation under the specific thermal cycle during braking.
expansion could be calculated by the relationship:
T in T dT T0
SC R
Tn
IP
T
According to the results of thermal cyclic tests, the instantaneous coefficient of linear
(2)
where the is the total strain in the length direction and in is the instantaneous
NU
coefficient of thermal expansion. By using the ANSYS parametric design language (APDL), the in T during heating process was applied into the material properties
MA
of the brake disc steel in the transient FE model firstly. Peak temperature of every element and node was monitored during the FE simulations. It was programmed using
D
APDL that the area in which the peak temperature exceed the austenizing temperature
TE
were selected and activated as the phase changing region. Then the instantaneous coefficient of linear expansion in T during cooling were applied on the activated
CE P
elements in order to consider the phase changing expansion during cooling process. 4. Results and discussion
4.1 Emergency braking tests
AC
During braking, hot bands could be observed through digital pictures and infrared images (shown in Fig. 8 and 9). These hot bands represent the track of hot spots that rotating at a high speed during braking. The appearance of hot spots indicates that the contact condition between the friction surfaces of brake disc and brake pads was non-uniform. As a result, non-uniform temperature distribution in both radial and circumferential directions can be observed in the infrared images with high temperature gradients. Three kind of macro hot spots were observed on the friction surface during the three continuous braking tests. For the first braking test, hot spots are mainly located near the mean friction radius of friction surface on the brake disc (shown in Fig. 8). For the second braking test, macro hot spots were mainly located at the inner and outer diameter of the friction surface (shown in Fig. 9). Fig. 10 shows
ACCEPTED MANUSCRIPT the evolution of temperature distribution during the third braking test. The hot spots appears in the inner and outer diameter on the friction surface in the early stage and then migrates into the middle of the friction surface. The migration of hot spots in this
IP
T
braking cycle reflects the change in contact condition between the brake disc and brake pads.
SC R
The temperature variation near the mean friction radius of brake disc was recorded by the infrared thermometer and is shown in Fig. 11. During the emergency braking process, the peak temperature on the friction surface was found to be higher
NU
than 900 ºC. Temperature values recorded by thermocouples were around 600 ºC. The fluctuation of the temperature curve indicates the non-uniform distribution of
MA
temperature on friction surface in circumferential direction result from the localized friction between brake disc and brake pads.
D
4.2 Thermal cyclic tests
TE
Based on the envelope line of the temperature variation shown in Fig. 11, the curve of peak temperature on the friction surface of brake disc could be obtained. The
CE P
variation of peak temperature on the friction surface which represents the typical thermal cyclic of the material in the zone of hot spots was used for the thermal cyclic tests of the brake disc steel.
AC
For the three repeated thermal cycles, the results of the thermal deformation of the specimens show a good repeatability. Fig. 12a shows the specimen length change during the temperature variation which simulated the emergency braking (EB) cycle. In Fig. 12a, there is a fluctuation in length in the 1# marked box during the stage of emergency braking. The volume contraction reflects the austenizing process of the steel. In the air-cooling stage, there is an increase in length in the 2# marked box of the curve and it results from the volume expansion caused by phase change. In Fig. 12b, the peak temperature is lower than the austenizing temperature of the brake disc steel in a simulated routine braking (RB) cycle. The relationship between the length change of the material and temperature variation is nearly linear and no phase change occurred. CCT curve of the studied steel was obtained through continuous cooling tests
ACCEPTED MANUSCRIPT under different cooling rate. The peak temperature variation of brake disc steel during cooling process are plotted in CCT curve in Fig. 13. It can be seen that the microstructure of the brake disc steel mainly changes into martensite and bainite with
IP
T
the actual cooling rate on the friction surface. For the studied steel, perlitic transformation only occurs at a lower cooling rate which is around 0.2 ºC/s.
SC R
Based on the results of the thermal cyclic tests, the instantaneous coefficient of linear expansion in can be calculated according to Eq.(2). The in during heating
NU
and cooling process are calculated respectively and are shown in Fig. 14. By considering the volume change caused by phase transformation of the brake disc steel,
MA
the stress-strain response in the overheated area of the brake disc could be evaluated more accurately in the FE simulations. 4.3 FE simulation results
D
FE simulations have been widely used in investigating the temperature and stress
TE
variation during severe braking conditions. In a previous study, an elastoplastic FE model was used to analyze the plastic deformation and residual stress of the brake
CE P
disc material[16, 17]. In this study, the instantaneous coefficient of linear expansion of the material during heating and cooling processes was considered respectively. Circumferential stress evolution near the hot spot on the cross section of the
AC
brake disc is shown in Fig. 15. At the beginning of the braking process, the material on the friction surface was heated and thermal expansion occurred. The structure constraint causes compressive thermal stress on the friction surface (Fig. 15a). After braking, tensile stress can be observed in the overheated area. The occurrence of tensile stress in the hot spot area was dominated by the plastic compressive strain which cannot be released after braking (Fig. 15b). When the temperature of the overheated area decreases to M S temperature, the martensitic transformation of the material changed the tensile stress into compressive stress on the friction surface of the brake disc (Fig. 15c). The mismatch of coefficient of linear expansion between phase changing area and the origin material causes redistribution of thermal stress. After undergoing a phase change, the stress across the hot spot became stabilized and
ACCEPTED MANUSCRIPT the maximum tensile stress moved to the lower boundary of the phase changing area (Fig. 15d). The temperature and stress variation at three positions which are marked as A, B
IP
T
and C (shown in Fig. 15b) were observed and plotted in Fig. 16. The simulated peak temperature and thermal gradient are close to the experimental results shown in Fig.
SC R
11. It can be observed in Fig. 16a that the peak temperature at point A and B occurred at the same time and the peak temperature in the overheated area reached over 900 ºC. As the friction heat conducted inward, temperature at point C rose and the
NU
temperature of the brake disc tends to be uniform.
Comparing with the simulation results in previous studies[16, 17], the introduced
MA
phase changing stress causes a different stress strain behavior of the material near the hot spots. According to the stress distribution contour and variation curves in Fig. 15
D
and 16, the phase changing process causes stress gradient in great extent near the hot
TE
spots. At these 3 picked points, it shows a cyclic tensile-compressive stress and cyclic plastic strain during braking. The maximum tensile stress in circumferential direction
CE P
was moved into the lower boundary of the phase changing area. Therefore, the residual tensile stress beneath the phase changing area provides the driving force of crack initiation and propagation.
AC
4.4 LCF properties
As shown in Fig. 11 and 16, the maximum amplitude of the temperature variation of brake disc steel reaches over 800 ºC. The great temperature fluctuation causes significant thermal stress and cyclic plastic strain occurs around the hot spots during repeated braking cycles. Therefore, LCF properties at the temperature around 800 ºC need to be investigated and the related research work can provide significant information regarding the fracture behavior of the brake disc steel which undergoes the severe braking process. Fig. 17 shows the comparison of the stable hysteresis loops at different cyclic strain ranges at RT, 600 ºC and 800 ºC. The width of the stable hysteresis loop represents the plastic strain range ( p ). For the studied steel, the elasticity modulus
ACCEPTED MANUSCRIPT and the yield point drop along with the increasing temperature. As a result, it can be seen in Fig. 17 that for the LCF tests at 800 ºC, plastic strain is still evident when the total strain amplitude is controlled at 0.15%. The relationship between stress
n
(3)
SC R
t / 2 K p / 2
IP
T
amplitude ( t / 2 ) and plastic strain amplitude ( p / 2 ) can be written as:
where K is the cyclic strength coefficient and n is the cyclic strain-hardening exponent. Fig. 18 shows the linear fit relationship in logarithmic coordinates between
NU
t / 2 and p / 2 described in Eq.(3). The cyclic strength coefficient and the cyclic strain-hardening exponent at RT and elevated temperatures are listed in Table 2.
MA
The cyclic stress-strain curves at different temperatures reflects the tendency of the occurrence of plastic strain of the steel under a certain cyclic stress level. In the hot
D
spot region on the friction surface of the brake disc, the thermal stress causes plastic
TE
deformation and the cyclic stress-strain in repeated braking cycles leads to the rupture of the material. Therefore, thermal fatigue cracks can initiate in hot spot region easily.
CE P
Based on the Basquin[18] equation and Manson[19]–Coffin[20] equations, the relationship between total strain amplitude ( t / 2 ) and number of reversals to
AC
failure ( 2 N f ) can be expressed as: t e p f 2N f 2 2 2 E
b
f 2 N f
c
(4)
where f is the fatigue strength coefficient, b is the fatigue strength exponent, f is the fatigue ductility coefficient, and c is the fatigue ductility exponent. According to Eq.(4), the fitted t / 2 2 N f curves at RT and 800 ºC is plotted in Fig. 19. In double logarithmic coordinates, the elastic term and plastic term in Eq. (4) can be plotted by two straight lines with slopes b and c respectively. Intercepts with the vertical axis of the two straight lines represents the constants
f / E
and
f
. The
LCF parameters of the material in Eq. (4) are listed in Table 3. The transition fatigue life
2 Nt
locates at the intersection point of the two fitted strain lines which
represents the magnitude of the plastic strain amplitude was equal to that of the elastic
ACCEPTED MANUSCRIPT strain amplitude. In Fig. 19b, location of
2 Nt
moved to right and the
corresponding value was 10 times larger than that of RT in Fig. 19a. It is evident in the LCF test results that the plastic deformation tends to play a dominant role at
SC R
4.5 Fracture observations
IP
causes LCF cracks near the hot spots of brake discs.
T
temperatures around 800 °C. As a result, cyclic strain in repeated braking cycles
In order to investigate the fracture mechanisms of the brake disc steel during the LCF tests at high temperature level, fracture surface of tested LCF specimens were
NU
observed by using SEM (Hitachi S-530). The morphology of fracture surfaces obtained from specimens tested at RT and 800 ºC were observed and SEM pictures of
MA
the fracture surfaces after tested under the strain amplitude of ±0.4% were also compared.
As shown in Fig. 20a and Fig. 21a, fatigue cracks initiated from the external
D
surface of both specimens and propagated inwards perpendicular to the principal
TE
stress axis. For the fracture surface shown in Fig. 21a, the size of fatigue crack
CE P
propagation area is smaller than that showed in Fig. 20a. It indicates a shorter fatigue life for the steel when it undergoes the same cyclic strain at higher temperature level. Fig. 20b and Fig. 21b show the crack initiation sites near the border of the
AC
fracture surface. At higher temperature level, the fracture surface was oxidized rapidly when exposed in an air atmosphere. The oxide film generated on the fracture surface at higher temperature level tends to accelerate the crack propagation and initiation of secondary cracks under the cyclic load. In Fig. 20c and Fig. 21c, fatigue striations are clearly visible and secondary cracks can be observed. In the stage of steady fatigue growth, the average striation space was about 2~3 m at RT and 5~7 m at 800 ºC, respectively. Every striation on the fracture surface represents a tensile-compressive load cycle. At higher temperature, plastic deformation happens much more easily and the larger space between fatigue striations indicates the number of cycles to fatigue failure is relatively lower. Moreover, it can be observed in Fig. 21a and Fig. 21c that more parallel secondary cracks initiate and propagate with the fracture of the oxide film. The alternating ridges
ACCEPTED MANUSCRIPT or strips of steps on the facture surface can be observed, as well. It results from the oxide film on the fracture surface causes stress concentration and fracture easily under the action of cyclic loads[21].
IP
T
As the fatigue cracks propagate inward, the load bearing area of the specimen keeps decreasing and finally fracture apart. The dimples on the final rupture area
and density of dimples is lager in Fig. 21d. 4.6 Microscopic observations
SC R
reflects the ductility of the material and it can be distinctly proved by that the number
NU
A brake disc with hot spots and thermal cracks was cut into pieces and the cross section of the specimen with hot spots were polished and etched with 4% nital. Fig.
MA
22 shows the macroscopic picture of an etched cross section of a specimen with hot spots. There are several hot spots which are in shape of semi-ellipse and overlap with
D
each other which can be observed. These hot spots represents overheated regions in
TE
the brake discs. The complicated braking history of the brake disc leads to the formation and migration of hot spots. The Vickers hardness on the cross section was
CE P
measured along two paths which start from the friction surface. The hardness at different depth from the friction surface of brake disc was plotted in Fig. 23. In order to evaluate the level of hardness under different thermal cycles, three kinds of thermal
AC
cyclic tests were conducted to simulate three kinds of braking conditions: emergency braking (EB) only, emergency braking following with routine braking (EB + RB), and routine braking (RB) only. The corresponding Vickers hardness after different thermal cyclic tests was measured. The Vickers hardness in the overheated area was found to be at a same level with the braking condition of EB + RB and was higher than that of the material inside. It illustrates that once the peak temperature exceeds the austenizing temperature of the brake disc steel during an emergency braking, microstructure transformation occurred and the hardness in the hot spot region rises. For the original material below the hot spots, the Vickers hardness was not increased. It indicates that the thermal cycle in a routine braking with lower braking energy causes insufficient heat and no microstructure transformation occurs. In the higher magnification view of the cross section with hot spots and
ACCEPTED MANUSCRIPT cracks(shown in Fig. 24), the picture can be divided into four sections. The microstructure in these four sections which are marked as S1, S2, S3 and S4 was quite different from each other and they were observed respectively using SEM and are
IP
T
shown in Fig. 25. The microstructure near the friction surface is shown in Fig. 25a and Fig. 25b. Martensite mixed with a small amount of bainite can be observed and
SC R
the grain size is relatively small. The Vickers hardness in the hot spot was measured and it was 436 HV. The lower boundary of the hot spots, which is marked as S3 in Fig. 24, is light grey in color and the microstructure is shown in Fig. 25c. Isometric crystal
NU
consists of ferrite mixed with carbides could be observed in this section. The microhardness in this section S3 is measured to be 285 HV. It indicates the
MA
recrystallizing process at the lower boundary of hot spots. Fig. 25d shows the original microstructure of the brake disc steel which is tempered martensite in relatively larger
D
grain size and the microhardness in this section was 308 HV. Comparing with the
TE
microstructure pictures from Fig. 25a to Fig. 25c, phase change of the steel can be easily distinguished in the hot spot region.
CE P
In previous study[16], depth of the thermal cracks was found to be consistent with the depth of tensile residual stress. Stress variation during braking process was simulated using an elastoplasticity FE model and the peak temperature was lower than
AC
the austenizing temperature of the steel. According to the simulation result in this study, severe friction process in localized area generates lots of friction heat and the peak temperature in the hot spot region can exceed the austenizing temperature. Therefore, microstructure transformation occurred in the overheated area during cooling. The phase changing process leads to localized stress concentration and higher tensile stress locates near the lower boundary of the hot spots after braking. Fig. 26 shows the secondary crack and embedded crack locates at the lower boundary of the hot spot. It can be figured out according to the propagation path of the cracks that there are embedded cracks initiate beneath the friction surface and propagate towards inward and outward simultaneously. Some of the embedded cracks coalesce with the surface cracks and then propagate inward in the following braking cycles (shown in Fig. 26a). Other embedded cracks initiate and grow individually
ACCEPTED MANUSCRIPT without coalescence with surface cracks. These isolated embedded cracks propagate outwards towards the friction surface and stop at the recrystallized section (shown in Fig. 26b). The recrystallized region is a transition boundary between residual tensile
IP
T
stress to compressive stress. According to the simulation results, the initiation of these embedded cracks can result from the cyclic stress-strain response around the hot spots.
SC R
During one emergency braking, plastic strain around the hot spot is induced by the high stress which exceeds the yield point of the steel at the corresponding temperature. In the following repeated braking cycles, for the hot spots which are reheated to
NU
austenizing temperature, cyclic strain occurs at the lower boundary and contributes the fracture driving force. Low cycle fatigue cracks initiate at the position with
MA
maximum cyclic strain amplitude. Moreover, micro-cracks can initiate around inclusions or impurities of the material and void also may form on grain boundaries.
D
Once the crack initiates, it propagates under the action of cyclic stress and cyclic
TE
strain. Although the hot spots and surface cracks were considered as the important factor that causes premature failure of brake discs in previous work[22, 23].
CE P
According to the simulation results and the microscopic observations in this work, the existence of the embedded cracks are proved to be invisible threaten to the brake discs with hot spots.
AC
5. Summary and Conclusions In this paper, the mechanisms of initiation and propagation of embedded thermal cracks under severe braking process was studied. Based on the experimental observation of temperature variation during simulated emergency braking process on a full scale dynamo testing rig, the typical temperature cycle on the friction surface of brake disc was obtained. Thermal cyclic tests were conducted to investigate the volume expansion of the studied brake disc steel induced by microstructure transformation. The instantaneous coefficient of linear expansion was calculated and it was considered in the numerical simulation in FE analysis. The cyclic stress-strain behavior near the hot spot was observed and the stress redistribution caused by phase change was investigated according to the simulation results. LCF tests of the studied brake disc steel under the corresponding temperature level were carried out. LCF
ACCEPTED MANUSCRIPT properties and fracture mechanisms under different temperature level were discussed. Microstructure and crack locations on the cross section of the friction surface of a brake disc with thermal crack was observed. The effect of hot spots on microstructure,
IP
T
stress distribution and crack initiation were discussed in this paper. The following conclusions of this work were obtained:
SC R
(1) During emergency braking, the peak temperature in the overheated area can exceed the austenizing temperature of the brake disc steel. Based on the temperature variation obtained in the full scale braking tests, thermal cyclic tests were conducted
NU
and microstructure transformation of the studied brake disc steel occurs under the simulated thermal cycle and the volume expansion can be observed in the thermal
MA
cyclic tests.
(2) According to the thermal cyclic tests, the instantaneous coefficient of linear
D
expansion which considering the phase change of the material was obtained and used
TE
in FE analysis. After braking, the stress distribution near the hot spots were changed by the phase changing process and the maximum tensile stress in circumferential
CE P
direction moved from the friction surface into the lower boundary of the phase changing area.
(3) For the LCF tests at higher temperature level, plastic deformation occurs more
AC
easily in the material and the fatigue life is lower. Space between fatigue striations on the fracture surface of specimen which was tested at 800 ºC was larger than that tested at lower temperatures. Alternating ridges and strips of steps on the facture surface covered with oxide can be observed and secondary cracks are larger in amount after testing as 800 ºC. (4) In hot spot region, the microstructure of martensite mixed with bainite can be observed and the microhardness is evidently higher than the original material below the hot spots. There is a recrystallized belt at the lower boundary of hot spot and the microhardness is lower than that at other positions. The embedded cracks could be observed which initiate and propagate under the action of cyclic tensile stress below the hot spot. The embedded cracks can grow individually or coalescence with surface cracks. The existence of embedded cracks are potential threaten for the operational
ACCEPTED MANUSCRIPT safety of brake discs. Acknowledgement
IP
SC R
Science Foundation of China (Grant No. 51271014).
T
The authors would like to acknowledge the support of the National Natural
References
AC
CE P
TE
D
MA
NU
[1] H. Sakamoto, K. Hirakawa. Fracture analysis and material improvement of brake discs. JSME International Journal Series A. 48 (2005) 458-64. [2] D.J. Kim, C.S. Seok, J.M. Koo, W.T. We, B.C. Goo, J.I. Won. Fatigue life assessment for brake disc of railway vehicle. Fatigue & Fracture of Engineering Materials & Structures. 33 (2009) 37-42. [3] F. Bagnoli, F. Dolce, M. Bernabei. Thermal fatigue cracks of fire fighting vehicles gray iron brake discs. Engineering Failure Analysis. 16 (2009) 152-63. [4] H. Samrout, R. El Abdi. Fatigue behaviour of 28CrMoV5-08 steel under thermomechanical loading. International Journal of Fatigue. 20 (1998) 555-63. [5] H. Samrout, R.E. Abdi, J.L. Chaboche. Model for 28CrMoV5-8 steel undergoing thermomechanical cyclic loadings. International Journal of Solids and Structures. 34 (1997) 4547-56. [6] P. Dufrénoy, D. Weichert. A Thermomechanical Model for the Analysis of Disc Brake Fracture Mechanisms. Journal of Thermal Stresses. 26 (2003) 815-28. [7] P. Dufrénoy, G. Bodovillé, G. Degallaix, Damage mechanisms and thermomechanical loading of brake discs, in: L. Rémy, J. Petit (Eds.), European Structural Integrity Society, Elsevier, 2002, pp. 167-76. [8] Z. Li, J. Han, W. Li, L. Pan. Low cycle fatigue behavior of Cr–Mo–V low alloy steel used for railway brake discs. Materials & Design. 56 (2014) 146-57. [9] J. Kim, B.C. Goo, S.C. Yoon, S.T. Kwon, Thermographic Investigation of Hot Spots in Railway Brake Discs, in: H.S. Lee, I.S. Yoon, M.H. Aliabadi (Eds.), Advances in Fracture and Damage Mechanics Vii, 2008, pp. 669-72. [10] T.J. Mackin, S.C. Noe, K.J. Ball, B.C. Bedell, D.P. Bim-Merle, M.C. Bingaman, et al. Thermal cracking in disc brakes. Engineering Failure Analysis. 9 (2002) 63-76. [11] H. Mori, T. Tominaga, M. Matsui, H. Qiu, T. Tsujimura. Observation of Thermal Cracks of Microstructural Size Scale by Thermal Effect during Wear in Ni-Cr-Mo Alloy Seel Brake Discs. Journal of the Japan Institute of Metals. 70 (2006) 785-9. [12] GB/T 15248-2008. The test method for axial loading constant amplitude low-cycle fatigue of metallic materials., Chinese National Standard Publishing House, Beijing, 2008. [13] C.H. Gao, X.Z. Lin. Transient temperature field analysis of a brake in a non-axisymmetric three-dimensional model. Journal of Materials Processing Technology. 129 (2002) 513-7. [14] P. Like, H. Jianmin, L. Zhiqiang, Y. Zhiyong, L. Weijing. Numerical simulation for train brake disc ventilation. Journal of Beijing Jiaotong University. 39 (2015)
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
IP
T
118-24. [15] L. Wallis, E. Leonardi, B. Milton, P. Joseph. Air flow and heat transfer in ventilated disc brake rotors with diamond and tear-drop pillars. Numerical Heat Transfer Applications. volume 41 (2010) 643-55. [16] Z. Li, J. Han, Z. Yang, L. Pan. The effect of braking energy on the fatigue crack propagation in railway brake discs. Engineering Failure Analysis. 44 (2014) 272-84. [17] Z. Yang, J. Han, W. Li, Z. Li, L. Pan, X. Shi. Analyzing the mechanisms of fatigue crack initiation and propagation in CRH EMU brake discs. Engineering Failure Analysis. 34 (2013) 121-8. [18] O.H. Basquin, The exponential law of endurance tests., Proceeding of ASTM 10, 1910, pp. 625-30. [19] S.S. Manson, Behavior of materials under conditions of thermal stress, NACA Technical Note 2933, Washington, DC, 1953. [20] L.F. Coffin, Jr. A study of the Effects of Cyclic Thermal Stresses on a Ductile Metal. Trans. ASME. 76 (1954) 931-50. [21] Fatigue and Fracture, Metals Handbook, Vol.19, American Society for Metals, 1996. [22] J.G. Kim, S.T. Kwon, S.C. Yoon. Analysis of hot spots evolution on brake disc using high-speed infrared camera. Key Engineering Materials. 417 (2010) 317-20. [23] H. Kasem, J.F. Brunel, P. Dufrénoy, M. Siroux, B. Desmet. Thermal levels and subsurface damage induced by the occurrence of hot spots during high-energy braking. Wear. 270 (2011) 355-64.
ACCEPTED MANUSCRIPT List of figure captions Fig.1 Full scale test rig for repeated emergency braking process at a simulating
T
running speed of 200 km/h.
IP
Fig.2 Configuration of the copper matrix powder metallurgy brake pad that was used
SC R
in the experimental braking tests.
Fig.3 Running speed and braking pressure profile for three continuous braking cycles. Fig.4 Schematic showing of temperature measurement positions in the brake disc
NU
using thermocouples (unit: mm).
Fig.5 Geometry and dimension of the LCF test specimens (unit: mm). Fig.6 Mechanical properties of the brake disc steel at RT and elevated temperatures.
MA
Fig.7 Relationship between heat transfer convection coefficient and running speed of railway vehicles for different surface groups used in FE simulations.
D
Fig.8 Comparison of hot spots between digital images and infrared pictures (a) during
braking cycles)
TE
braking and (b) at the end of the braking process. (First braking of three continuous
CE P
Fig.9 Comparison of hot spots between digital images and infrared pictures (a) during braking and (b) at the end of the braking process. (Second braking of three continuous braking cycles)
AC
Fig.10 Evolution of hot spots and temperature distribution at different braking time: (a) 7 s, (b) 15 s, (c) 30 s, and (d) end of the braking process 50 s. Fig.11 Temperature variation recorded by infrared thermometer at the mean friction radius of brake disc during three continuous emergency braking. Fig.12 Temperature variation and change in length of the specimen in the thermal cyclic tests that simulate (a) emergency braking and (b) routine braking. Fig.13 CCT diagram of the brake disc steel and the cooling curve of the brake disc steel after emergency braking. Fig.14 Calculated instantaneous coefficient of linear expansion of the brake disc steel during (a) heating process and (b) cooling process for phase changing area. Fig.15 Circumferential stress distribution near the hot spot at different time during
ACCEPTED MANUSCRIPT braking and cooling (a) 2 s, (b) end of braking 60 s, (c) starting time of phase change 80 s, and (d) finishing time of phase change 140 s. (Unit: Pa) Fig.16 Simulated results of (a) temperature variation and (b) circumferential stress
IP
T
variation on the cross section at point A, B and C.
Fig.17 Comparison of the stable hysteresis loops of LCF tests at different temperature
SC R
levels (a) room temperature, (b) 600 ºC and (c) 800 ºC.
Fig.18 Cyclic stress–strain behavior of the brake disc steel at room temperature and elevated temperatures.
NU
Fig.19 Comparison of strain-life curves between LCF tests at (a) room temperature and (b) 800 ºC.
MA
Fig.20 SEM fractograph of LCF specimen tested at RT and t / 2 =±0.4%. (a) crack origin sites and strips of steps on the fracture surface, (b) higher magnification view
D
of the fatigue crack initiation area “A”, (c) fatigue striations in area “B” (d) final
TE
rupture area.
CE P
Fig.21 SEM fractograph of LCF specimen tested at 800 ºC and t / 2 =±0.4%. (a) secondary cracks and strips of steps on the fracture surface, (b) higher magnification view of the fatigue crack initiation area “C”, (c) fatigue striations in area “D” (d) final
AC
rupture area.
Fig.22 Hot spots and cracks on an etched cross section of the brake disc. Fig.23 Comparison of microhardness at different distance from the friction surface on the cross section and microhardness after different braking conditions. Fig.24 Higher magnification view of the cross section with hot spots and four sections with different microstructures. Fig.25 Microstructure at different depth from the friction surface on the cross section in (a) S1, (b) S2, (c) S3, and (d) S4. Fig.26 Embedded cracks beneath the hot spots on the cross section which (a) coalesce with surface crack during propagation and (b) grow inwards and outwards individually.
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 1
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
Fig. 2
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 3
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 4
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
IP
Fig. 5
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 6
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 7
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 8
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 9
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 10
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 11
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 12b
CE P
TE
D
MA
NU
Fig. 12a
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 13
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 14b
CE P
TE
D
MA
NU
Fig. 14a
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
Fig. 15
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 16b
CE P
TE
D
MA
NU
Fig. 16a
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
Fig. 17
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 18
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 19b
CE P
TE
D
MA
NU
Fig. 19a
D
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
Fig. 20
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
Fig. 21
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
Fig. 22
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 23
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 24
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
Fig. 25
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 26b
CE P
TE
D
MA
NU
Fig. 26a
ACCEPTED MANUSCRIPT
Table 1 Physical properties of the brake disc steel used in the FE simulation Specific heat (J/kg·K)
Thermal conductivity (W/m·K)
25
473
44.1
200
523
400
7840
607 754
800
1130
NU MA D TE CE P AC
43.6
SC R
600
T
Density (kg/m3)
IP
Temperature (°C)
41.6 38.0 44.7
ACCEPTED MANUSCRIPT
K (MPa)
n
RT
1175
0.0488
600
712
0.0536
800
217
T
Temperature (°C)
IP
Table 2 The cyclic strength coefficient and the cyclic strain-hardening exponent at RT and elevated temperatures
AC
CE P
TE
D
MA
NU
SC R
0.1076
ACCEPTED MANUSCRIPT
Table 3 LCF parameters of the material at RT and elevated temperatures
f (MPa)
b
f
c
RT
1241
-0.0684
2.5573
-0.9657
600
705
-0.0474
0.5898
800
210
-0.1432
0.5783
IP
SC R NU MA D TE CE P AC
T
Temperature (°C)
-0.7714 -0.6792
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
IP
T
Highlights: Phase changing in hot spots were observed after high-energy braking. Temperature variation during braking was simulated by thermal cyclic tests. Instantaneous coefficient of linear expansion was considered in FE simulation. LCF tests of brake disc steel were conducted from room temperature to 800ºC. Embedded fatigue cracks were found beneath the hot spots with phase changing.
S1350-6307(15)30001-7 doi: 10.1016/j.engfailanal.2015.07.002 EFA 2603
To appear in: Received date: Revised date: Accepted date:
28 March 2015 8 June 2015 1 July 2015
Please cite this article as: Li Zhiqiang, Han Jianmin, Yang Zhiyong, Li Weijing, Analyzing the mechanisms of thermal fatigue and phase change of steel used in brake discs, (2015), doi: 10.1016/j.engfailanal.2015.07.002
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Analyzing the mechanisms of thermal fatigue and phase change of steel used in brake discs
T
Zhiqiang Li, Jianmin Han*, Zhiyong Yang, Weijing Li
SC R
100044, P.R. China
IP
School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing
* Corresponding author. Tel.: +86 10 51683300; fax: +86 10 51683300. E-mail address: [email protected] (Jianmin Han)
NU
Abstract
Thermal cracks on the friction surface of railway brake discs can develop during
MA
their lifespan. Cracks often initiate after severe braking conditions along with the occurrence of hot spots. The cyclic thermal and mechanical loads causes high temperature, plastic strain and even phase change of the brake disc steel. In this paper,
D
full scale emergency braking tests were conducted and the peak temperature of
TE
localized area was found exceeding the austenitizing temperature of the steel. Thermal
CE P
cyclic tests was performed to simulate the temperature variation during braking. Volume change of the steel caused by microstructure transformation was taken into consideration in numerical simulation. Combining with the fracture behavior of brake
AC
disc steel in low cycle fatigue (LCF) tests in different temperature level, the simulation results show a good consistency with the results of microstructure observation and crack initiation. The occurrence of embedded crack could be well explained according to the simulation results and low cycle fatigue test results. Keywords: brake disc; phase change; thermal fatigue; numerical simulation.
1. Introduction Brake discs used on railway vehicles are crucial safety components which transform the kinetic energy created during braking into heat by the way of friction. After a period of usage, thermal cracks can be observed on the friction surface of brake discs. During repeated braking cycles, cyclic thermal and mechanical loads were applied on the friction surface[1-3]. For a severe braking condition, plastic strain
ACCEPTED MANUSCRIPT occurs on the friction surface of brake disc at high temperature levels and it causes irreversible damages for the material. Cyclic thermal plastic strain and residual stress are considered as dominant factors which initiate and cause growth of thermal cracks.
IP
T
Samrout and El Abdi[4, 5] have studied fatigue behavior of the brake disc steel for French T.G.V. under the thermal mechanical cyclic loadings and an anisothermal
SC R
elastoplastic model was proposed to predict the response of brake disc. Dufrénoy et al. [6, 7] investigated the damage mechanisms of brake discs. The cyclic stress-strain behavior during braking was studied and a thermo-mechanical model was proposed to
NU
describe the thermal fatigue behavior of brake disc material. Therefore, low cycle fatigue behavior of the studied steel is worthy to be paid attention for investigating the
MA
crack initiation and growth. In previous study[8], LCF tests under different temperature level and cyclic strain were conducted from room temperature (RT) to
D
600 ºC. LCF properties of the brake disc steel were obtained and fracture mechanisms
TE
of the studied steel were discussed.
In some cases, non-uniform contact conditions exist between the brake pad and
CE P
brake disc such that hot spots occur on the friction surface of the brake disc[9]. Instantaneous peak temperatures in these localized areas can exceed 900 ºC which is over the austenitizing temperature of the brake disc steel[10]. The great temperature
AC
fluctuation in the overheated area during braking and cooling causes microstructure transformation of the steel. Mori et al.[11] investigated the microstructure change of brake disc steel with the effect of the thermal load during wear and found the precipitated carbides were the initiation sites of the thermal cracks. Sakamoto[1] also found tempered martensite formed in the heated area on the friction surface. The stress intensity factor was calculated based on the finite element (FE) simulation results to investigate the crack propagation behavior. However, the resulting volume change due to microstructure transformation can also change the stress distribution on the friction surface of brake disc. Stress redistribution caused by microstructure transformation leads to different mechanisms result in crack initiation and growth. In this paper, the continuous braking tests were performed on a full scale dynamo testing machine in order to simulate the emergency braking (EB) processes. The
ACCEPTED MANUSCRIPT microstructure transformation of the brake disc steel was investigated by conducting the simulated thermal cyclic tests. The stress redistribution introduced by phase change were analyzed in finite element (FE) analysis. Combining with the low cycle
IP
T
fatigue (LCF) behavior and microstructure observations of the brake disc steel, the result of FE simulation can reflect the influence of microstructure transformation in
SC R
stress distribution and fracture behavior of brake discs. The related research works can provide a better understanding of potential threat caused by high energy braking. 2. Materials and experimental procedures
NU
For the study, a forged Cr-Mo-V low alloyed steel brake disc was tested on a full scale dynamo testing machine (shown in Fig. 1) to simulate the emergency braking
MA
(EB) processes. An electroslag remelted steel containing 0.32% C (weight percent) was used and other chemical composition is comparable to a 28CrMoV5-8 steel alloy.
D
The copper matrix powder metallurgy brake pads (shown in Fig. 2) were used in the
TE
full scale dynamo braking tests. For the brake pad material, the content of copper-tin alloy is 60%~70% as the matrix, the content of graphite as the lubrication component
is 15%~20%.
CE P
is 12%~18% and the content of the Al2O3 and SiC powder as the friction components
The braking conditions simulated were as follows. The running speed of the
AC
railway vehicle was equivalent to 200 km/h with an axle load of 16,000 kg. Braking pressure was chosen as 25 kN. In order to simulate a severe braking condition, three continuous emergency braking tests were conducted. For each braking cycle, there is a parking stage (P) and air-cooling stage following each emergency braking (EB). During the three continuous emergency braking tests, acceleration among each braking was 0.3 m/s2 and the convection cooling condition was considered. The variation of running speed and braking pressure of the three continuous braking tests is plotted in Fig. 3. Temperature at the mean friction radius position on friction surface of the brake disc was recorded continuously using an infrared thermometer and the video of braking process was recorded using a digital camera (shown in Fig. 1). Two thermocouples were embedded at the depth of 6 mm from the friction surface of the brake disc. Temperature at the mean friction radius (Position 1 in Fig. 4) and
ACCEPTED MANUSCRIPT the outer diameter of the friction surface (Position 2 in Fig. 4) were measured using these two thermocouples respectively. Temperature distribution on the friction surface of brake disc was recorded by an infrared thermal camera during braking process.
T
The continuous cooling transformation (CCT) curves of the studied steel were
IP
obtained using a Φ4 mm × 10 mm specimen on a thermal dilatometer (DIL805A).
SC R
Then thermal cyclic tests were conducted on the thermal dilatometer according to the temperature variation on the friction surface during full scale braking tests. Thermal
measured continuously during the test.
NU
strain could be calculated basing on the change in length of the specimen which was
Based on the measured temperature level obtained from full scale braking tests,
MA
strain controlled LCF tests were carried out from room temperature to 800 ºC. The total axial strain amplitudes were controlled at levels of ±0.8%, ±0.6%, ±0.4%, and a
D
lower level from 0.1% to 0.3%. The LCF tests were carried out using a
TE
servo-hydraulic machine (MTS-810) equipped with a temperature controlled furnace. The LCF test specimens (shown in Fig. 5) were machined according to Chinese
CE P
National Standard GB/T 15248-2008 ‘‘The Test Method for Axial Loading Constant Amplitude Low-Cycle Fatigue of Metallic Materials’’[12]. The specimens with a diameter of 6.35 mm and gage length of 25 mm were finely polished before testing.
AC
Two replications were tested at each strain amplitude with a fully push–pull mode (strain ratio R = -1). A triangular waveform signal was used and the strain rate was 5.0 ×10-3 s-1. The longitudinal strain was measured continuously with an extensometer and the plastic strain range was measured. The stress ranges were examined by the peak and valley stress values in each strain cycle. After the tests were completed, microstructures and fracture surfaces of selected specimens were examined using a scanning electron microscope (SEM). Cyclic strain-stress behavior and fracture mechanism of the brake disc steel were studied. The brake disc after braking tests was cut into pieces and the cross-section near the friction surface was polished and etched using a nital etchant (4% nitric acid 96% ethanol). Microstructure and Vickers hardness of both the picked specimens of brake disc and the thermal cycle test specimen were observed and measured. The fracture
ACCEPTED MANUSCRIPT mechanism of the brake disc was analyzed and compared with the fracture behavior of the LCF test specimen under the corresponding temperature level. 3. FE simulation methods
IP
T
The emergency braking process was simulated using ANSYS code and a 1/10 circumferential section model was used in the simulation considering the axial
SC R
symmetry of the brake disc structure. Sequentially coupled thermal-stress analysis was used for the simulation of braking process. For the studied brake disc steel, the mechanical and physical properties were obtained in previous work [8]. Mechanical
NU
properties of the steel obtained by tensile tests from room temperature to 800 ºC are shown in Fig. 6. The physical properties of the material used in the simulation are
MA
listed in Table 1.
According to the hot spots distribution observed during the full scale dynamo
D
braking tests, heat flux applied onto the overheated area of the friction surface could
TE
be inversely calculated. By converting the kinetic energy into friction heat, the heat flux input can be described as:
CE P
q r , t P0 t r
(1)
where is the heat-partitioning factor representing the fraction of frictional heat
AC
flux entering the brake disc[13], is the friction coefficient between brake disc and pads, P0 is the braking pressure, and t is the angular velocity of brake disc. Convection heat transfer coefficient on the surface of brake disc model is expressed using a function of speed of the railway vehicle. Based on previous studies, CFD analysis was conducted to investigate the heat transfer coefficients of brake discs with different structures [14, 15]. The relationship between heat transfer convection coefficient and train speed for different surfaces is plotted in Fig. 7. For the 1/10 circumferential section model, symmetry constraints were applied on the section plane. Displacement constraints were applied on the surface of the bolt hole to restrain three degrees of freedom of the nodes in the mechanical simulation. When the peak temperature exceeds the austenizing temperature, volume change
ACCEPTED MANUSCRIPT of the brake disc steel should be considered due to phase transformation. Instantaneous coefficient of linear expansion can reflect the volume change of material during phase transformation under the specific thermal cycle during braking.
expansion could be calculated by the relationship:
T in T dT T0
SC R
Tn
IP
T
According to the results of thermal cyclic tests, the instantaneous coefficient of linear
(2)
where the is the total strain in the length direction and in is the instantaneous
NU
coefficient of thermal expansion. By using the ANSYS parametric design language (APDL), the in T during heating process was applied into the material properties
MA
of the brake disc steel in the transient FE model firstly. Peak temperature of every element and node was monitored during the FE simulations. It was programmed using
D
APDL that the area in which the peak temperature exceed the austenizing temperature
TE
were selected and activated as the phase changing region. Then the instantaneous coefficient of linear expansion in T during cooling were applied on the activated
CE P
elements in order to consider the phase changing expansion during cooling process. 4. Results and discussion
4.1 Emergency braking tests
AC
During braking, hot bands could be observed through digital pictures and infrared images (shown in Fig. 8 and 9). These hot bands represent the track of hot spots that rotating at a high speed during braking. The appearance of hot spots indicates that the contact condition between the friction surfaces of brake disc and brake pads was non-uniform. As a result, non-uniform temperature distribution in both radial and circumferential directions can be observed in the infrared images with high temperature gradients. Three kind of macro hot spots were observed on the friction surface during the three continuous braking tests. For the first braking test, hot spots are mainly located near the mean friction radius of friction surface on the brake disc (shown in Fig. 8). For the second braking test, macro hot spots were mainly located at the inner and outer diameter of the friction surface (shown in Fig. 9). Fig. 10 shows
ACCEPTED MANUSCRIPT the evolution of temperature distribution during the third braking test. The hot spots appears in the inner and outer diameter on the friction surface in the early stage and then migrates into the middle of the friction surface. The migration of hot spots in this
IP
T
braking cycle reflects the change in contact condition between the brake disc and brake pads.
SC R
The temperature variation near the mean friction radius of brake disc was recorded by the infrared thermometer and is shown in Fig. 11. During the emergency braking process, the peak temperature on the friction surface was found to be higher
NU
than 900 ºC. Temperature values recorded by thermocouples were around 600 ºC. The fluctuation of the temperature curve indicates the non-uniform distribution of
MA
temperature on friction surface in circumferential direction result from the localized friction between brake disc and brake pads.
D
4.2 Thermal cyclic tests
TE
Based on the envelope line of the temperature variation shown in Fig. 11, the curve of peak temperature on the friction surface of brake disc could be obtained. The
CE P
variation of peak temperature on the friction surface which represents the typical thermal cyclic of the material in the zone of hot spots was used for the thermal cyclic tests of the brake disc steel.
AC
For the three repeated thermal cycles, the results of the thermal deformation of the specimens show a good repeatability. Fig. 12a shows the specimen length change during the temperature variation which simulated the emergency braking (EB) cycle. In Fig. 12a, there is a fluctuation in length in the 1# marked box during the stage of emergency braking. The volume contraction reflects the austenizing process of the steel. In the air-cooling stage, there is an increase in length in the 2# marked box of the curve and it results from the volume expansion caused by phase change. In Fig. 12b, the peak temperature is lower than the austenizing temperature of the brake disc steel in a simulated routine braking (RB) cycle. The relationship between the length change of the material and temperature variation is nearly linear and no phase change occurred. CCT curve of the studied steel was obtained through continuous cooling tests
ACCEPTED MANUSCRIPT under different cooling rate. The peak temperature variation of brake disc steel during cooling process are plotted in CCT curve in Fig. 13. It can be seen that the microstructure of the brake disc steel mainly changes into martensite and bainite with
IP
T
the actual cooling rate on the friction surface. For the studied steel, perlitic transformation only occurs at a lower cooling rate which is around 0.2 ºC/s.
SC R
Based on the results of the thermal cyclic tests, the instantaneous coefficient of linear expansion in can be calculated according to Eq.(2). The in during heating
NU
and cooling process are calculated respectively and are shown in Fig. 14. By considering the volume change caused by phase transformation of the brake disc steel,
MA
the stress-strain response in the overheated area of the brake disc could be evaluated more accurately in the FE simulations. 4.3 FE simulation results
D
FE simulations have been widely used in investigating the temperature and stress
TE
variation during severe braking conditions. In a previous study, an elastoplastic FE model was used to analyze the plastic deformation and residual stress of the brake
CE P
disc material[16, 17]. In this study, the instantaneous coefficient of linear expansion of the material during heating and cooling processes was considered respectively. Circumferential stress evolution near the hot spot on the cross section of the
AC
brake disc is shown in Fig. 15. At the beginning of the braking process, the material on the friction surface was heated and thermal expansion occurred. The structure constraint causes compressive thermal stress on the friction surface (Fig. 15a). After braking, tensile stress can be observed in the overheated area. The occurrence of tensile stress in the hot spot area was dominated by the plastic compressive strain which cannot be released after braking (Fig. 15b). When the temperature of the overheated area decreases to M S temperature, the martensitic transformation of the material changed the tensile stress into compressive stress on the friction surface of the brake disc (Fig. 15c). The mismatch of coefficient of linear expansion between phase changing area and the origin material causes redistribution of thermal stress. After undergoing a phase change, the stress across the hot spot became stabilized and
ACCEPTED MANUSCRIPT the maximum tensile stress moved to the lower boundary of the phase changing area (Fig. 15d). The temperature and stress variation at three positions which are marked as A, B
IP
T
and C (shown in Fig. 15b) were observed and plotted in Fig. 16. The simulated peak temperature and thermal gradient are close to the experimental results shown in Fig.
SC R
11. It can be observed in Fig. 16a that the peak temperature at point A and B occurred at the same time and the peak temperature in the overheated area reached over 900 ºC. As the friction heat conducted inward, temperature at point C rose and the
NU
temperature of the brake disc tends to be uniform.
Comparing with the simulation results in previous studies[16, 17], the introduced
MA
phase changing stress causes a different stress strain behavior of the material near the hot spots. According to the stress distribution contour and variation curves in Fig. 15
D
and 16, the phase changing process causes stress gradient in great extent near the hot
TE
spots. At these 3 picked points, it shows a cyclic tensile-compressive stress and cyclic plastic strain during braking. The maximum tensile stress in circumferential direction
CE P
was moved into the lower boundary of the phase changing area. Therefore, the residual tensile stress beneath the phase changing area provides the driving force of crack initiation and propagation.
AC
4.4 LCF properties
As shown in Fig. 11 and 16, the maximum amplitude of the temperature variation of brake disc steel reaches over 800 ºC. The great temperature fluctuation causes significant thermal stress and cyclic plastic strain occurs around the hot spots during repeated braking cycles. Therefore, LCF properties at the temperature around 800 ºC need to be investigated and the related research work can provide significant information regarding the fracture behavior of the brake disc steel which undergoes the severe braking process. Fig. 17 shows the comparison of the stable hysteresis loops at different cyclic strain ranges at RT, 600 ºC and 800 ºC. The width of the stable hysteresis loop represents the plastic strain range ( p ). For the studied steel, the elasticity modulus
ACCEPTED MANUSCRIPT and the yield point drop along with the increasing temperature. As a result, it can be seen in Fig. 17 that for the LCF tests at 800 ºC, plastic strain is still evident when the total strain amplitude is controlled at 0.15%. The relationship between stress
n
(3)
SC R
t / 2 K p / 2
IP
T
amplitude ( t / 2 ) and plastic strain amplitude ( p / 2 ) can be written as:
where K is the cyclic strength coefficient and n is the cyclic strain-hardening exponent. Fig. 18 shows the linear fit relationship in logarithmic coordinates between
NU
t / 2 and p / 2 described in Eq.(3). The cyclic strength coefficient and the cyclic strain-hardening exponent at RT and elevated temperatures are listed in Table 2.
MA
The cyclic stress-strain curves at different temperatures reflects the tendency of the occurrence of plastic strain of the steel under a certain cyclic stress level. In the hot
D
spot region on the friction surface of the brake disc, the thermal stress causes plastic
TE
deformation and the cyclic stress-strain in repeated braking cycles leads to the rupture of the material. Therefore, thermal fatigue cracks can initiate in hot spot region easily.
CE P
Based on the Basquin[18] equation and Manson[19]–Coffin[20] equations, the relationship between total strain amplitude ( t / 2 ) and number of reversals to
AC
failure ( 2 N f ) can be expressed as: t e p f 2N f 2 2 2 E
b
f 2 N f
c
(4)
where f is the fatigue strength coefficient, b is the fatigue strength exponent, f is the fatigue ductility coefficient, and c is the fatigue ductility exponent. According to Eq.(4), the fitted t / 2 2 N f curves at RT and 800 ºC is plotted in Fig. 19. In double logarithmic coordinates, the elastic term and plastic term in Eq. (4) can be plotted by two straight lines with slopes b and c respectively. Intercepts with the vertical axis of the two straight lines represents the constants
f / E
and
f
. The
LCF parameters of the material in Eq. (4) are listed in Table 3. The transition fatigue life
2 Nt
locates at the intersection point of the two fitted strain lines which
represents the magnitude of the plastic strain amplitude was equal to that of the elastic
ACCEPTED MANUSCRIPT strain amplitude. In Fig. 19b, location of
2 Nt
moved to right and the
corresponding value was 10 times larger than that of RT in Fig. 19a. It is evident in the LCF test results that the plastic deformation tends to play a dominant role at
SC R
4.5 Fracture observations
IP
causes LCF cracks near the hot spots of brake discs.
T
temperatures around 800 °C. As a result, cyclic strain in repeated braking cycles
In order to investigate the fracture mechanisms of the brake disc steel during the LCF tests at high temperature level, fracture surface of tested LCF specimens were
NU
observed by using SEM (Hitachi S-530). The morphology of fracture surfaces obtained from specimens tested at RT and 800 ºC were observed and SEM pictures of
MA
the fracture surfaces after tested under the strain amplitude of ±0.4% were also compared.
As shown in Fig. 20a and Fig. 21a, fatigue cracks initiated from the external
D
surface of both specimens and propagated inwards perpendicular to the principal
TE
stress axis. For the fracture surface shown in Fig. 21a, the size of fatigue crack
CE P
propagation area is smaller than that showed in Fig. 20a. It indicates a shorter fatigue life for the steel when it undergoes the same cyclic strain at higher temperature level. Fig. 20b and Fig. 21b show the crack initiation sites near the border of the
AC
fracture surface. At higher temperature level, the fracture surface was oxidized rapidly when exposed in an air atmosphere. The oxide film generated on the fracture surface at higher temperature level tends to accelerate the crack propagation and initiation of secondary cracks under the cyclic load. In Fig. 20c and Fig. 21c, fatigue striations are clearly visible and secondary cracks can be observed. In the stage of steady fatigue growth, the average striation space was about 2~3 m at RT and 5~7 m at 800 ºC, respectively. Every striation on the fracture surface represents a tensile-compressive load cycle. At higher temperature, plastic deformation happens much more easily and the larger space between fatigue striations indicates the number of cycles to fatigue failure is relatively lower. Moreover, it can be observed in Fig. 21a and Fig. 21c that more parallel secondary cracks initiate and propagate with the fracture of the oxide film. The alternating ridges
ACCEPTED MANUSCRIPT or strips of steps on the facture surface can be observed, as well. It results from the oxide film on the fracture surface causes stress concentration and fracture easily under the action of cyclic loads[21].
IP
T
As the fatigue cracks propagate inward, the load bearing area of the specimen keeps decreasing and finally fracture apart. The dimples on the final rupture area
and density of dimples is lager in Fig. 21d. 4.6 Microscopic observations
SC R
reflects the ductility of the material and it can be distinctly proved by that the number
NU
A brake disc with hot spots and thermal cracks was cut into pieces and the cross section of the specimen with hot spots were polished and etched with 4% nital. Fig.
MA
22 shows the macroscopic picture of an etched cross section of a specimen with hot spots. There are several hot spots which are in shape of semi-ellipse and overlap with
D
each other which can be observed. These hot spots represents overheated regions in
TE
the brake discs. The complicated braking history of the brake disc leads to the formation and migration of hot spots. The Vickers hardness on the cross section was
CE P
measured along two paths which start from the friction surface. The hardness at different depth from the friction surface of brake disc was plotted in Fig. 23. In order to evaluate the level of hardness under different thermal cycles, three kinds of thermal
AC
cyclic tests were conducted to simulate three kinds of braking conditions: emergency braking (EB) only, emergency braking following with routine braking (EB + RB), and routine braking (RB) only. The corresponding Vickers hardness after different thermal cyclic tests was measured. The Vickers hardness in the overheated area was found to be at a same level with the braking condition of EB + RB and was higher than that of the material inside. It illustrates that once the peak temperature exceeds the austenizing temperature of the brake disc steel during an emergency braking, microstructure transformation occurred and the hardness in the hot spot region rises. For the original material below the hot spots, the Vickers hardness was not increased. It indicates that the thermal cycle in a routine braking with lower braking energy causes insufficient heat and no microstructure transformation occurs. In the higher magnification view of the cross section with hot spots and
ACCEPTED MANUSCRIPT cracks(shown in Fig. 24), the picture can be divided into four sections. The microstructure in these four sections which are marked as S1, S2, S3 and S4 was quite different from each other and they were observed respectively using SEM and are
IP
T
shown in Fig. 25. The microstructure near the friction surface is shown in Fig. 25a and Fig. 25b. Martensite mixed with a small amount of bainite can be observed and
SC R
the grain size is relatively small. The Vickers hardness in the hot spot was measured and it was 436 HV. The lower boundary of the hot spots, which is marked as S3 in Fig. 24, is light grey in color and the microstructure is shown in Fig. 25c. Isometric crystal
NU
consists of ferrite mixed with carbides could be observed in this section. The microhardness in this section S3 is measured to be 285 HV. It indicates the
MA
recrystallizing process at the lower boundary of hot spots. Fig. 25d shows the original microstructure of the brake disc steel which is tempered martensite in relatively larger
D
grain size and the microhardness in this section was 308 HV. Comparing with the
TE
microstructure pictures from Fig. 25a to Fig. 25c, phase change of the steel can be easily distinguished in the hot spot region.
CE P
In previous study[16], depth of the thermal cracks was found to be consistent with the depth of tensile residual stress. Stress variation during braking process was simulated using an elastoplasticity FE model and the peak temperature was lower than
AC
the austenizing temperature of the steel. According to the simulation result in this study, severe friction process in localized area generates lots of friction heat and the peak temperature in the hot spot region can exceed the austenizing temperature. Therefore, microstructure transformation occurred in the overheated area during cooling. The phase changing process leads to localized stress concentration and higher tensile stress locates near the lower boundary of the hot spots after braking. Fig. 26 shows the secondary crack and embedded crack locates at the lower boundary of the hot spot. It can be figured out according to the propagation path of the cracks that there are embedded cracks initiate beneath the friction surface and propagate towards inward and outward simultaneously. Some of the embedded cracks coalesce with the surface cracks and then propagate inward in the following braking cycles (shown in Fig. 26a). Other embedded cracks initiate and grow individually
ACCEPTED MANUSCRIPT without coalescence with surface cracks. These isolated embedded cracks propagate outwards towards the friction surface and stop at the recrystallized section (shown in Fig. 26b). The recrystallized region is a transition boundary between residual tensile
IP
T
stress to compressive stress. According to the simulation results, the initiation of these embedded cracks can result from the cyclic stress-strain response around the hot spots.
SC R
During one emergency braking, plastic strain around the hot spot is induced by the high stress which exceeds the yield point of the steel at the corresponding temperature. In the following repeated braking cycles, for the hot spots which are reheated to
NU
austenizing temperature, cyclic strain occurs at the lower boundary and contributes the fracture driving force. Low cycle fatigue cracks initiate at the position with
MA
maximum cyclic strain amplitude. Moreover, micro-cracks can initiate around inclusions or impurities of the material and void also may form on grain boundaries.
D
Once the crack initiates, it propagates under the action of cyclic stress and cyclic
TE
strain. Although the hot spots and surface cracks were considered as the important factor that causes premature failure of brake discs in previous work[22, 23].
CE P
According to the simulation results and the microscopic observations in this work, the existence of the embedded cracks are proved to be invisible threaten to the brake discs with hot spots.
AC
5. Summary and Conclusions In this paper, the mechanisms of initiation and propagation of embedded thermal cracks under severe braking process was studied. Based on the experimental observation of temperature variation during simulated emergency braking process on a full scale dynamo testing rig, the typical temperature cycle on the friction surface of brake disc was obtained. Thermal cyclic tests were conducted to investigate the volume expansion of the studied brake disc steel induced by microstructure transformation. The instantaneous coefficient of linear expansion was calculated and it was considered in the numerical simulation in FE analysis. The cyclic stress-strain behavior near the hot spot was observed and the stress redistribution caused by phase change was investigated according to the simulation results. LCF tests of the studied brake disc steel under the corresponding temperature level were carried out. LCF
ACCEPTED MANUSCRIPT properties and fracture mechanisms under different temperature level were discussed. Microstructure and crack locations on the cross section of the friction surface of a brake disc with thermal crack was observed. The effect of hot spots on microstructure,
IP
T
stress distribution and crack initiation were discussed in this paper. The following conclusions of this work were obtained:
SC R
(1) During emergency braking, the peak temperature in the overheated area can exceed the austenizing temperature of the brake disc steel. Based on the temperature variation obtained in the full scale braking tests, thermal cyclic tests were conducted
NU
and microstructure transformation of the studied brake disc steel occurs under the simulated thermal cycle and the volume expansion can be observed in the thermal
MA
cyclic tests.
(2) According to the thermal cyclic tests, the instantaneous coefficient of linear
D
expansion which considering the phase change of the material was obtained and used
TE
in FE analysis. After braking, the stress distribution near the hot spots were changed by the phase changing process and the maximum tensile stress in circumferential
CE P
direction moved from the friction surface into the lower boundary of the phase changing area.
(3) For the LCF tests at higher temperature level, plastic deformation occurs more
AC
easily in the material and the fatigue life is lower. Space between fatigue striations on the fracture surface of specimen which was tested at 800 ºC was larger than that tested at lower temperatures. Alternating ridges and strips of steps on the facture surface covered with oxide can be observed and secondary cracks are larger in amount after testing as 800 ºC. (4) In hot spot region, the microstructure of martensite mixed with bainite can be observed and the microhardness is evidently higher than the original material below the hot spots. There is a recrystallized belt at the lower boundary of hot spot and the microhardness is lower than that at other positions. The embedded cracks could be observed which initiate and propagate under the action of cyclic tensile stress below the hot spot. The embedded cracks can grow individually or coalescence with surface cracks. The existence of embedded cracks are potential threaten for the operational
ACCEPTED MANUSCRIPT safety of brake discs. Acknowledgement
IP
SC R
Science Foundation of China (Grant No. 51271014).
T
The authors would like to acknowledge the support of the National Natural
References
AC
CE P
TE
D
MA
NU
[1] H. Sakamoto, K. Hirakawa. Fracture analysis and material improvement of brake discs. JSME International Journal Series A. 48 (2005) 458-64. [2] D.J. Kim, C.S. Seok, J.M. Koo, W.T. We, B.C. Goo, J.I. Won. Fatigue life assessment for brake disc of railway vehicle. Fatigue & Fracture of Engineering Materials & Structures. 33 (2009) 37-42. [3] F. Bagnoli, F. Dolce, M. Bernabei. Thermal fatigue cracks of fire fighting vehicles gray iron brake discs. Engineering Failure Analysis. 16 (2009) 152-63. [4] H. Samrout, R. El Abdi. Fatigue behaviour of 28CrMoV5-08 steel under thermomechanical loading. International Journal of Fatigue. 20 (1998) 555-63. [5] H. Samrout, R.E. Abdi, J.L. Chaboche. Model for 28CrMoV5-8 steel undergoing thermomechanical cyclic loadings. International Journal of Solids and Structures. 34 (1997) 4547-56. [6] P. Dufrénoy, D. Weichert. A Thermomechanical Model for the Analysis of Disc Brake Fracture Mechanisms. Journal of Thermal Stresses. 26 (2003) 815-28. [7] P. Dufrénoy, G. Bodovillé, G. Degallaix, Damage mechanisms and thermomechanical loading of brake discs, in: L. Rémy, J. Petit (Eds.), European Structural Integrity Society, Elsevier, 2002, pp. 167-76. [8] Z. Li, J. Han, W. Li, L. Pan. Low cycle fatigue behavior of Cr–Mo–V low alloy steel used for railway brake discs. Materials & Design. 56 (2014) 146-57. [9] J. Kim, B.C. Goo, S.C. Yoon, S.T. Kwon, Thermographic Investigation of Hot Spots in Railway Brake Discs, in: H.S. Lee, I.S. Yoon, M.H. Aliabadi (Eds.), Advances in Fracture and Damage Mechanics Vii, 2008, pp. 669-72. [10] T.J. Mackin, S.C. Noe, K.J. Ball, B.C. Bedell, D.P. Bim-Merle, M.C. Bingaman, et al. Thermal cracking in disc brakes. Engineering Failure Analysis. 9 (2002) 63-76. [11] H. Mori, T. Tominaga, M. Matsui, H. Qiu, T. Tsujimura. Observation of Thermal Cracks of Microstructural Size Scale by Thermal Effect during Wear in Ni-Cr-Mo Alloy Seel Brake Discs. Journal of the Japan Institute of Metals. 70 (2006) 785-9. [12] GB/T 15248-2008. The test method for axial loading constant amplitude low-cycle fatigue of metallic materials., Chinese National Standard Publishing House, Beijing, 2008. [13] C.H. Gao, X.Z. Lin. Transient temperature field analysis of a brake in a non-axisymmetric three-dimensional model. Journal of Materials Processing Technology. 129 (2002) 513-7. [14] P. Like, H. Jianmin, L. Zhiqiang, Y. Zhiyong, L. Weijing. Numerical simulation for train brake disc ventilation. Journal of Beijing Jiaotong University. 39 (2015)
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
IP
T
118-24. [15] L. Wallis, E. Leonardi, B. Milton, P. Joseph. Air flow and heat transfer in ventilated disc brake rotors with diamond and tear-drop pillars. Numerical Heat Transfer Applications. volume 41 (2010) 643-55. [16] Z. Li, J. Han, Z. Yang, L. Pan. The effect of braking energy on the fatigue crack propagation in railway brake discs. Engineering Failure Analysis. 44 (2014) 272-84. [17] Z. Yang, J. Han, W. Li, Z. Li, L. Pan, X. Shi. Analyzing the mechanisms of fatigue crack initiation and propagation in CRH EMU brake discs. Engineering Failure Analysis. 34 (2013) 121-8. [18] O.H. Basquin, The exponential law of endurance tests., Proceeding of ASTM 10, 1910, pp. 625-30. [19] S.S. Manson, Behavior of materials under conditions of thermal stress, NACA Technical Note 2933, Washington, DC, 1953. [20] L.F. Coffin, Jr. A study of the Effects of Cyclic Thermal Stresses on a Ductile Metal. Trans. ASME. 76 (1954) 931-50. [21] Fatigue and Fracture, Metals Handbook, Vol.19, American Society for Metals, 1996. [22] J.G. Kim, S.T. Kwon, S.C. Yoon. Analysis of hot spots evolution on brake disc using high-speed infrared camera. Key Engineering Materials. 417 (2010) 317-20. [23] H. Kasem, J.F. Brunel, P. Dufrénoy, M. Siroux, B. Desmet. Thermal levels and subsurface damage induced by the occurrence of hot spots during high-energy braking. Wear. 270 (2011) 355-64.
ACCEPTED MANUSCRIPT List of figure captions Fig.1 Full scale test rig for repeated emergency braking process at a simulating
T
running speed of 200 km/h.
IP
Fig.2 Configuration of the copper matrix powder metallurgy brake pad that was used
SC R
in the experimental braking tests.
Fig.3 Running speed and braking pressure profile for three continuous braking cycles. Fig.4 Schematic showing of temperature measurement positions in the brake disc
NU
using thermocouples (unit: mm).
Fig.5 Geometry and dimension of the LCF test specimens (unit: mm). Fig.6 Mechanical properties of the brake disc steel at RT and elevated temperatures.
MA
Fig.7 Relationship between heat transfer convection coefficient and running speed of railway vehicles for different surface groups used in FE simulations.
D
Fig.8 Comparison of hot spots between digital images and infrared pictures (a) during
braking cycles)
TE
braking and (b) at the end of the braking process. (First braking of three continuous
CE P
Fig.9 Comparison of hot spots between digital images and infrared pictures (a) during braking and (b) at the end of the braking process. (Second braking of three continuous braking cycles)
AC
Fig.10 Evolution of hot spots and temperature distribution at different braking time: (a) 7 s, (b) 15 s, (c) 30 s, and (d) end of the braking process 50 s. Fig.11 Temperature variation recorded by infrared thermometer at the mean friction radius of brake disc during three continuous emergency braking. Fig.12 Temperature variation and change in length of the specimen in the thermal cyclic tests that simulate (a) emergency braking and (b) routine braking. Fig.13 CCT diagram of the brake disc steel and the cooling curve of the brake disc steel after emergency braking. Fig.14 Calculated instantaneous coefficient of linear expansion of the brake disc steel during (a) heating process and (b) cooling process for phase changing area. Fig.15 Circumferential stress distribution near the hot spot at different time during
ACCEPTED MANUSCRIPT braking and cooling (a) 2 s, (b) end of braking 60 s, (c) starting time of phase change 80 s, and (d) finishing time of phase change 140 s. (Unit: Pa) Fig.16 Simulated results of (a) temperature variation and (b) circumferential stress
IP
T
variation on the cross section at point A, B and C.
Fig.17 Comparison of the stable hysteresis loops of LCF tests at different temperature
SC R
levels (a) room temperature, (b) 600 ºC and (c) 800 ºC.
Fig.18 Cyclic stress–strain behavior of the brake disc steel at room temperature and elevated temperatures.
NU
Fig.19 Comparison of strain-life curves between LCF tests at (a) room temperature and (b) 800 ºC.
MA
Fig.20 SEM fractograph of LCF specimen tested at RT and t / 2 =±0.4%. (a) crack origin sites and strips of steps on the fracture surface, (b) higher magnification view
D
of the fatigue crack initiation area “A”, (c) fatigue striations in area “B” (d) final
TE
rupture area.
CE P
Fig.21 SEM fractograph of LCF specimen tested at 800 ºC and t / 2 =±0.4%. (a) secondary cracks and strips of steps on the fracture surface, (b) higher magnification view of the fatigue crack initiation area “C”, (c) fatigue striations in area “D” (d) final
AC
rupture area.
Fig.22 Hot spots and cracks on an etched cross section of the brake disc. Fig.23 Comparison of microhardness at different distance from the friction surface on the cross section and microhardness after different braking conditions. Fig.24 Higher magnification view of the cross section with hot spots and four sections with different microstructures. Fig.25 Microstructure at different depth from the friction surface on the cross section in (a) S1, (b) S2, (c) S3, and (d) S4. Fig.26 Embedded cracks beneath the hot spots on the cross section which (a) coalesce with surface crack during propagation and (b) grow inwards and outwards individually.
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 1
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
Fig. 2
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 3
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 4
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
IP
Fig. 5
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 6
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 7
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 8
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 9
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 10
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 11
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 12b
CE P
TE
D
MA
NU
Fig. 12a
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 13
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 14b
CE P
TE
D
MA
NU
Fig. 14a
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
Fig. 15
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 16b
CE P
TE
D
MA
NU
Fig. 16a
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
Fig. 17
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 18
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 19b
CE P
TE
D
MA
NU
Fig. 19a
D
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
Fig. 20
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
Fig. 21
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
Fig. 22
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
Fig. 23
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
Fig. 24
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
Fig. 25
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 26b
CE P
TE
D
MA
NU
Fig. 26a
ACCEPTED MANUSCRIPT
Table 1 Physical properties of the brake disc steel used in the FE simulation Specific heat (J/kg·K)
Thermal conductivity (W/m·K)
25
473
44.1
200
523
400
7840
607 754
800
1130
NU MA D TE CE P AC
43.6
SC R
600
T
Density (kg/m3)
IP
Temperature (°C)
41.6 38.0 44.7
ACCEPTED MANUSCRIPT
K (MPa)
n
RT
1175
0.0488
600
712
0.0536
800
217
T
Temperature (°C)
IP
Table 2 The cyclic strength coefficient and the cyclic strain-hardening exponent at RT and elevated temperatures
AC
CE P
TE
D
MA
NU
SC R
0.1076
ACCEPTED MANUSCRIPT
Table 3 LCF parameters of the material at RT and elevated temperatures
f (MPa)
b
f
c
RT
1241
-0.0684
2.5573
-0.9657
600
705
-0.0474
0.5898
800
210
-0.1432
0.5783
IP
SC R NU MA D TE CE P AC
T
Temperature (°C)
-0.7714 -0.6792
ACCEPTED MANUSCRIPT
AC
CE P
TE
D
MA
NU
SC R
IP
T
Highlights: Phase changing in hot spots were observed after high-energy braking. Temperature variation during braking was simulated by thermal cyclic tests. Instantaneous coefficient of linear expansion was considered in FE simulation. LCF tests of brake disc steel were conducted from room temperature to 800ºC. Embedded fatigue cracks were found beneath the hot spots with phase changing.