Wear 273 (2011) 93 99

Contents lists available at ScienceDirect

Wear

jou rna l h om epage: www.elsev ier .co

A study romand bra

Saeed Ab Lara KTH Machineb KTH Buildingc Bombardier T

a r t i c l

Article history:Received 20 AReceived in reAccepted 21 AAvailable onlin

Keywords:Railway brakeAirborne partiWear

akes a, bothnd th

gainbutio-on-dize dilationamet

in dMaterials such as iron, copper, aluminum, chromium, cobalt, antimony, and zinc have been detected inthe nano-sized particles.

2011 Elsevier B.V. All rights reserved.

1. Introdu

The maimental. Heastudied extbustion proinvestigate

Gustafssthermore, iBudapest [5centrationsupper limit[6].

The purpimentally emorphologytypical orgations on a pcontact pre

CorresponBrinellvgen 8

E-mail add

0043-1648/$ doi:10.1016/j.ction

n concerns about airborne wear particles are environ-lth effects of the inhaled nano-sized particles have beenensively but most studies have been focused on com-cesses [1]. Only a few studies have been carried out tothe emission of wear particles in rail transport.on recently presented a review of these works [2]. Fur-nvestigations in the Stockholm [3], London [4] and] underground systems have shown particle mass con-

in the range of 3001000 g/m3 much higher than the for urban trafc in the EU, which is 50 g/m3 per day

ose of the research presented in this paper is to exper-valuate the number, concentration, size distribution,, and element analysis of airborne wear particles fromnic brake pads. A series of eld tests and their simula-in-on-disc machine using the same sliding velocity andssure have been performed.

ding author at: KTH Machine Design, Royal institute of Technology,3, SE 10044 Stockholm, Sweden. Fax: +46 8202287.ress: sabbasi@kth.se (S. Abbasi).

2. Experiment set-up

Two different set-ups were considered for the experiments. Aseries of full-scale eld tests were performed with a Regina X54 testtrain. The main reason for conducting the pin-on-disc laboratorytests was to clarify the results from the eld test (e.g., to be ableto distinguish the airborne wear particles that originate from thebrake disc from other particles in the surrounding environment).

In both eld and laboratory tests, typical organic brake pads(Becorit 950-1) were tested against steel brake discs. The chemi-cal compositions of these braking components are reported in [7].Airborne wear particles were collected on lters during testing andsubsequently analysed with a scanning electron microscope (SEM)and energy-dispersive X-ray spectroscopy (EDX).

2.1. Field tests

A Regina X54 train was equipped with particle measurementinstruments at two different sampling points. The eld tests wereconducted in normal trafc conditions on a regular Swedish inter-city track over the course of three days.

The test route is shown in Fig. 1. The maximum allowable oper-ational speed of the train was 200 km/h when both mechanical andelectrical brakes were active (although the speed was reduced to180 km/h when the electrical brake was deactivated on purpose).The train followed normal trafc operation when it was on the main

see front matter 2011 Elsevier B.V. All rights reserved.wear.2011.04.013 of airborne wear particles generated fke discs

basia,, Jens Wahlstrma, Lars Olanderb, Christina Design, SE 10044 Stockholm, Sweden

Service Engineering, SE 10044 Stockholm, Swedenransportation Sweden AB, SE-721 73 Vsters, Sweden

e i n f o

ugust 2010vised form 12 April 2011pril 2011e 17 June 2011

padscles

a b s t r a c t

Brake pads on wheel-mounted disc brerties and robustness. During brakingparticles that may become airborne aparticles in Rail transport project is tocontrolling the number and size distritests and laboratory tests with a pinmatter of particles, along with their sof results from the pin-on-disc simuultra-ne peak for particles with a diparticles with a size of around 350 nmm/locate /wear

organic railway brake pads

ssonc, Ulf Olofssona, Ulf Sellgrena

re often used in rail transport due to their good thermal prop- the disc and the pads are worn. This wear process generatesus affect human health. The long term purpose of Airborne

knowledge on the wear mechanisms in order to nd means ofn of airborne particles. In this regard, a series of full-scale eldisc machine have been conducted. The morphology and thestribution and concentration, have been studied. The validity

has been veried by the eld test results. Results show aner size around 100 nm in diameter, a dominant ne peak foriameter, and a coarse peak with a size of 37 m in diameter.

94 S. Abbasi et al. / Wear 273 (2011) 93 99

Fig. 2. Fig. 1. Field test route, with the industrial track between Nyk

Thermocouples positions in the brake pad (main brake pad).

track. Somethe maximwas rather was gathereduring test

The comwas used aThe trains 150% durinment devicmechanicalwas 10 Hz. main brakeinvestigate

Two setferent samp145 mm farexposed to

Fig. 3. Two sampling points, brake pad sampling point (right) anping and Flen highlighted.

tests were conducted on an industrial track, whereum operational speed was only 90 km/h. As that areaisolated from disturbance and noise, most of the datad from that region. The climatic conditions of this route

runs are reported in [7].pact brake caliper used was RZS, and Becorit 950-1

s the brake pads. The brake disc was made from steel.weight was 62,500 kg and the brake percentage wasg operation. The test train was equipped with measure-es to measure and record speed, and total electrical and

brake force on each axle. The data acquisition frequencyFig. 2 shows four K-type thermocouples inserted in the pad. Particles generated from pad-disc contact wered by particle measurement devices.s of DustTrak, Grimm and P-Trak were used in two dif-ling points (see Fig. 3). One sampling point was located

from the main brake pad. During braking, it was highlythe particles generated by the main brake pad. We refer

d global sampling point (left).

S. Abbasi et al. / Wear 273 (2011) 93 99 95

Fig. 4. Schematic of the test equipment [8]. A: room air; B: fan; C: ow rate measure-ment; D: lter; E: exible tube; F: inlet for clean air, measurement point; G: closedbox (chamber); H: pin-on-disc machine; I: pin sample along with thermocouple; N:air inside box, well-mixed; J: air outlet, measurement points; L: dead weight; M:rotating disc sample, N: air inside chamber.

to this poinlocated in tfrom concrrefer to this

2.2. Labora

The labmachine wipin (Fig. 4). stant appliespeeds of upgential forcK-type therthe nominaon-disc mathe cleanlinby Sundh aet al. [11] twheelrail

The lte(according ciency of 99

The 110piece of whby using a wmechanical

Table 1Contact conditions in the laboratory tests.

No. Load (N) Sliding velocity (m/s) Time (min)

1 60 12.4 202 40 12.4 203 20 12.4 20

imens were cleaned ultrasonically for 20 min with both heptaneand methanol. The test conditions are presented in Table 1.

2.3. Particle measurement devices

In this study, four different types of particle measurementinstrument were used. The main instrument was a Grimm 1.109aerosol spectrometer. The second device was a P-Trak particlecounter. The P-Trak was a condensation nuclei counter that mea-sured the number concentration of airborne particles between 0.02and 1 m in diameter. The third instrument was a scanning mobil-ity particle sizer (SMPS) which used only on laboratory tests. The

ombined an electrostatic classier (TSI 3071) with a parti-nterr, wh

me techs we

ults

. 5ond

on pll of tremem/h.

GrimtTraorcelustrvalueresetionaelemcted

partn in

on G

Fig. 5. Effects (brake samplint as the brake pad sampling point. The other point washe middle of the axle. The effect of generated particlesete sleepers and ballast was more traceable in it. We

point as the global sampling point.

tory tests

oratory tests were performed using a pin-on-discth a horizontal rotating disc and a dead-weight-loadedThe machine ran under stationary conditions with con-d normal forces of up to 100 N and at constant rotational

to 3000 rpm. A load cell was used to measure the tan-e acting on the pin. Each pin was also equipped with amocouple inserted by drilling, and placed 1 mm froml contact area between the pin and the disc. The pin-chine was operated in a sealed box in order to controless of the incoming air. This setup was previously usednd Olofsson [8], Olofsson et al. [9,10] and Wahlstrmo study the airborne particles generated by simulatedcontact and passenger car brakes.r used to ensure particle-free inlet air was of class H13to standard EN 1822), with a certied collection ef-.95% at maximum penetrating particle size.

mm diameter disc specimens were cut from a usedeel-mounted steel brake disc from a Regina X54 trainater jet, while the 10 mm diameter pins were sawn out

SMPS ccle coucounteused to

Thedevice[9,10].

3. Res

Figsbrake cbrakestests. Athe afoof 70 kby theby Dusbrake fbeen ilimum been ppropor

An is depitypicalis showbased ly from a Becorit brake pad. Before testing, the disc spec- disc machin

of the different brake levels on the concentration and total number of the recorded particlg point). The train speed was 70 km/h and the electrical brake was deactivated on purpo (TSI CPC 3010). The fourth instrument was a DustTrakich reported the mass concentration in mg/m3. It was

asure particles between 0.1 and 10 m.nical specication and set-up of all of the measuringre akin to those in previous studies by Olofsson et al.

7 show eld test results based on applying differentitions. Different brake levels and deactivating electricalurpose were the main concern of these series of eldhese results were registered by running the test train onntioned industrial track (Fig. 1) at an operational speed

Every 6 s, the total number of particles was recordedm spectrometer and the concentration was recorded

k. In all of these graphs, the magnitude of train speed,, brake pad temperature and particle concentration haveated as a normalised value in the vertical axis. The max-s of these factors in the illustrated time interval have

nted using the number 1, and other values are scaledlly.ent analysis by EDX from a piece of the new brake pad

in Fig. 8. Figs. 9 and 10 present the SEM images foricles. A typical result by SMPS from a pin-on-disc test

Fig. 11. Figs. 12 and 13 show particle size distributionrimm measurement results. In Figs. 1316 the pin-on-e had been used to simulate braking force. Loads 60 N,es, brake pad temperature and train speed reduction in normal trafcse.

96 S. Abbasi et al. / Wear 273 (2011) 93 99

Fig. 6. Effects (global sampli

Fig. 7. Effectssampling poin

40 N and 2velocity of in a train, atrain. One rpin-on-discwere based

4. Discussi

In any uwith frictioof the different brake levels on the concentration and total number of the recorded particlng point). The train speed was 70 km/h and the electrical brake was deactivated on purp

of the different brake levels on the concentration and total number of the recorded partt). The train speed was 70 km/h and the electrical brake was activated.

Fig. 8. A typical spectroscopy result from a part of Beco

0 N reproduced brake levels 7, 5 and 3, and a sliding12.4 m/s in pin-on-disc simulated a speed of 70 km/hccording to disc size and wheel and brake radius in aepetition was conducted for each test condition during

simulation. The particle volume distributions in Fig. 15 on an assumption of a spherical shape for the particles.

on

nlubricated contact, the loss of material is associatednal heat, vibration, wear debris, and tribochemical reac-

tions in thepotential tocess [13]. Athe airbornanism whicprocesses. Cdict the parproject whi

As depictests conrincreasing bes, brake pad temperature and train speed reduction in normal trafcose.

icles, brake pad temperature and train speed in normal trafc (brake

rit brake pad by EDX.

contacting surfaces. Actually, these factors have the interact each other and even inuence the wear pro-

fraction of the wear debris can be transformed intoe particles. Fig. 17 shows a schematic view of a mech-h lead to the generation of airborne particles in wearoncerning these factors and providing a model to pre-ticle formation is a long-term objective of the currentch needs further studies.ted in Figs. 5 and 13, both eld tests and laboratorym that wear rate and particle generation increases withraking force. This result is in line with previous studies

S. Abbasi et al. / Wear 273 (2011) 93 99 97

Fig. 9. A typical image by SEM from a particle that was collected during the eldtests in the local sampling point lter.

Fig. 10. A typisimulation.

by Olofssonand concenmechanicalpoint was locontact aresurface andthe discrep

Fig. 11. A typiby SMPS (load

Fig. 12. Particle size distribution in different brake levels when the electrical brakewas deactivated on purpose. Results were recorded by Grimm (local sampling point).cal image by SEM from a particle that was collected during pin-on-disc

et al. [9,10]. Figs. 5 and 6 illustrate particle generationtration during mechanical braking. Effects of different

brake levels are evident in Fig. 5. As only one samplingcated near the brake pad, and the fact that the apparenta was only a portion of the nominal brake pad contact

also moved in the nominal contact area during braking,ancies were small. In contrast, the global point (Fig. 6)

cal size distribution for particles sized less than 520 nm in diameter: 60 N, sliding velocity: 12.4 m/s).

Fig. 13. A compoints and pin

Fig. 14. The tin 2 repetitionwere recorded

shows no cdetected paof differentwere believ

Fig. 15. A comsimulations evity: 12.4 m/s).parison between particle size distribution in the eld test sampling-on-disc simulation (results were recorded by Grimm).

otal average number of particles when different loads were applieds (the grey and black column) during pin-on-disc simulation. Results

by Grimm (sliding velocity: 12.4 m/s, time: 20 min).

onsiderable changes between the total amount of therticles and their concentration during the application

mechanical brake levels. Re-suspension and dilutioned to be the main reasons for this phenomenon.

parison between volume distribution in eld tests and pin-on-discery 3 min. Results were recorded by Grimm (load: 60 N, sliding veloc-

98 S. Abbasi et al. / Wear 273 (2011) 93 99

Fig. 16. A typical running-in effect on the size distribution of particles in pin-on-disc simulations every 3 min. Results were recorded by Grimm (load: 60 N, slidingvelocity: 12.4 m/s).

Fig. 7 presents a comparison between different brake levelswhen bothneously. Foand the brato stop theforces wererequested dnitude of thlevels, the tforce, mechtrain. This pof the applously, applyat these levwe assumewas one quaxle.

Figs. 9 anthe laboratocopper werysis resultsmetals suchminum, ant

A typicatri-modal daround 100280 and 35peaks in th(Figs. 12 anciples in meindecency oas shown in

Fig. 17. A schairborne parti

Fig. 13 presents a typical comparison between size distributionof two sampling points during both the eld test and laboratorysimulation. Although the coincident of their peaks are acceptable,there are some discrepancies between their frequencies. This canbe explainemixture of ratory condparticles oraccording tto size enlaratio and vasize enlargedrawing in ne region,illuminatedment with Wautomobile

Typical vtest are pre

ith tion. 15s oveutiondencad mas trcks ahange dat

clus

follis of

pin-icle g

distditionctivahanie pa

theiame

the 00 nm

the ter.

wled mechanical and electrical brakes were used simulta-r the rst four levels, there was no particle generationke force applied by the electrical system was sufcient

train. However, in levels 5, 6 and 7, the mechanical added to the total brake system in order to satisfy theeceleration rate. Although the driver dictated the mag-e deceleration rate manually by using different brakerain computer struck a balance among electrical brakeanical brake force, and the speed and weight of therocess caused only slight differences in the magnitudesied mechanical brake forces for levels 6 and 7. Obvi-ing mechanical brake forces caused particle generationels. As four brake pads were mounted on each train axle,d the total amount of the brake force in each brake padarter of the total recorded applied brake force on that

d 10 illustrate a typical particle from the eld test andry test. Besides their similarity in appearance, iron and

e the main elements in both particles. The element anal- of brake pads, in particles and in bulk (Fig. 8), detected

as: iron, copper, cobalt, chromium, molybdenum, alu-imony and zinc, which is in line with results in [2,12].l result of SMPS is shown in Fig. 11. According to theistribution, a peak occurred near the ultra-ne region

nm in diameter, and two other peaks occurred around0 nm in diameter in the ne particle region. These twoe ne particle region were also detected by Grimmd 16), although SMPS and Grimm used different prin-asuring particles. It is also noteworthy to mention thef the particle size distributions at different brake levels,

Fig. 12.

ticles wapplica

Figsbutiondistribdepenbrake pcess wtest trathese cof thes

5. Con

Theanalys

1. Thepartsizecon

2. Deamec

3. Threa. In

db. In

6c. In

e

Acknoematic illustration of mechanisms which lead to generation of thecles in the unlubricated contacts wear processes.

This woof the Royavaluable asof Technolo

References

[1] A. Madl, nanopart

[2] M. GustaOlofsson

[3] C. JohansAtmos. E

[4] A. Seatonundergro355362d by particle characteristics. Airborne particles are aliquid droplets or solid particles in the air. As the labo-itions were set up to provide air ow without any solid

droplets, these discrepancies are reasonable. Besides,o Hinds [14], sub-micron-sized particles are susceptiblergement due to a high magnitude of surface-to-volumen der Waals forces. During eld tests, the probability ofment was higher than in laboratory conditions due tounltered air and using longer tubes. Two peaks in the

of around 350 nm and 600 nm in diameter, have been in Figs. 15 and 16. These observations are in agree-ahlstrms work [11], although his studies focused on

brake pads.olume distributions from a eld test and a pin-on-discsented in Fig. 15. All results conrm that coarse par-sizes of 37 m in diameter are generated during the

of similar braking force. and 16 show typical particle size and volume distri-r time. The size distribution started from a tri-modal

and ended with a bimodal distribution. This time-y may be explained by a mild-severe transition and theaterial stabilisation process during operation. This pro-

eated in the eld test by rst running the train on thend applying the brakes before starting the eld tests. Ases can affect laboratory simulations, the average valuesa have been used in Figs. 13 and 14.

ion

owing general conclusions can be made based on anthe test results:

on-disc test is a robust method for studying airborneeneration, based on the similarity of morphology and

ributions, with the prerequisite that the same contacts are used in the laboratory tests as in the eld tests.

ting electrical brakes or applying higher levels ofcal brakes increase particle generation from brake pads.rticle size regimes were identied:

ultra-ne particle region, a peak of around 100 nm inter,ne particle region, three peaks of 280 nm, 350 nm and

in diameter, with the 350 nm peak dominating,coarse particle region, a peak of around 36 m in diam-

gments

rk formed a part of the activities of the Railway Groupl Institute of Technology. The author acknowledges thesistance from Dr Wobushet Sahle of the Royal Institutegy, and Dr Anders Jansson from Stockholm University.

K. Pinkerton, Health effects of inhaled engineered and incidentalicles, Crit. Rev. Toxicol. 39 (8) (2009) 629658.fsson, Airborne particles from the wheelrail contact, in: R. Lewis, U.(Eds.), WheelRail Interface Handbook, CRC Press, 2009, pp. 550575.son, P. Johansson, Particulate matter in the underground of Stockholm,nviron. 37 (2003)., J. Cherrie, M. Dennekamp, K. Donaldson, J. Hurley, C. Tran, The Londonund: dust and hazard to health, Occup. Environ. Med. 62 (6) (2005).

S. Abbasi et al. / Wear 273 (2011) 93 99 99

[5] I. Salma, T. Weidinger, W. Maenhaut, Time-resolved mass concentration, com-position and sources of aerosol particles in a metropolitan underground railwaystation, Atmos. Environ. 41 (2007).

[6] European Commission (EC), Air quality stds. 2008-07-01.[7] S. Abbasi, U. Olofsson, L. Olander, C. Larsson, U. Sellgren, A. Jansson, A eld test

study of airborne wear particles from a running regional train, J. Rail RapidTransit., in press.

[8] J. Sundh, U. Olofsson, Wear testing in relation to airborne particles generatedin a wheelrail contact, Lubr. Sci. (2009) 135150.

[9] U. Olofsson, L. Olander, A. Jansson, Towards a model for the num-ber of particles generated from a sliding contact, Wear 267 (12) (2009)22522256.

[10] U. Olofsson, L. Olander, A. Jansson, A Study of Airborne Wear Particles GeneratedFrom a Sliding Contact, J. Tribol. 131 (2009), 044503-1.

[11] J. Wahlstrm, A. Sderberg, L. Olander, L. Olofsson, A. Jansson, Airbornewear particles from passenger car disc brakes, Eng. Tribol. 224 (2010) 179188.

[12] E. Fridell, M. Ferm, A. Bjrk, A. Ekberg, On-board measurement of particulatematter emissions from a passenger train, J. Rail Rapid Transit. 225 (1) (2011)99106.

[13] I. Hutching, Tribology: Friction and Wear of Engineering Materials, CRC Press,1992.

[14] W.C. Hinds, Aerosol Technology: Properties Behavior, and Measurement ofAirborne Particles, 2nd ed., Wiley, NY, 1999.

A study of airborne wear particles generated from organic railway brake pads and brake discs1 Introduction2 Experiment set-up2.1 Field tests2.2 Laboratory tests2.3 Particle measurement devices

3 Results4 Discussion5 ConclusionAcknowledgmentsReferences