9. 040301_DTE-MS-JK_R3-Wear_Paper Published_Rev 0

Wear258(2005)1148–1155Railway noise and the effect of top of rail liquid friction modifiers:changes in sound and vibration spectral distributions in curvesDonald T.Eadie a,∗,Marco Santoro a,Joe Kalousek ba Kelsan Technologies Corp.,1140West15th St.,North Vancouver,BC,Canada V7P1M9b National Research Council of Canada,Centre for Surface Transportation,3250East Mall,Vancouver,BC,Canada V6T1W5Received13June2003;received in revised form28November2003;accepted1March2004Available online11November2004AbstractFor railway noise in curves,bothflanging and squeal noise can be environmentally significant.Rolling noise is dominant in tangent track. This paper examines the spectral sound distribution in curves for different wheel/rail system types,and compares spectra after the top of rail friction level is controlled with a special friction modifier.The friction modifier controls top of rail(TOR)friction at an intermediate level, and imparts“positive friction”attributes to the interfacial layer.A significant range of spectral characteristics was noted for the different wheel/rail system types.In all cases the friction modifier significantly reduced the sound levels at the frequencies associated with top of rail squeal,and also at the frequency bands related toflange contact noise.For some Metro systems a noticeable reduction was also recorded at the lower frequencies associated with rolling noise,possibly due to a reduction in“graunching”.There was some suggestion of reduction at low frequency vibration(frequencies down to30Hz)as well as in the high stick-slip oscillation frequencies.©2004Elsevier B.V.All rights reserved.Keywords:Railway noise;Metro systems;Liquid friction modifiers1.BackgroundRailway noise is an area of continuing research interest. Regulatory and public concern continues to drive technolog-ical initiatives to understand and define sources and means of mitigating the different types of noise emanating from rail-way systems.Vibration from trains is also an area of concern especially for Metro systems with nearby office or residential buildings.This paper is concerned with wheel squeal,flanging noise, rolling noise,and vibration.Squeal andflanging noise are associated with curves,particularly sharp curves(R<500m) whereas rolling noise is generally associated with tangent track.Spectral sound distribution levels are also considered for different types of trains.In particular we examine how noise and spectral sound distribution is altered by control ∗Corresponding author.Tel.:+16049846100;fax:+16049843419.E-mail addresses:deadie@(D.T.Eadie),msantoro@(M.Santoro),joe.kalousek@nrc.ca(J.Kalousek).of friction on the top of rail(TOR).Control of friction is achieved with a special friction modifier that does not impact traction or braking.1.1.Frequency range of different noise typesTable1summarizes the frequency range for the different types of railway noise of concern in this paper.These are either the ranges generally accepted within the industry,or those established from our own experience with sound mea-surements on many different systems.The identification offlanging noise as high frequency (5000–10000Hz)will be addressed in more detail later in the paper.1.2.Mechanism of noise generationThe sources and reduction possibilities of railway noise have recently been reviewed by Talotte et al.[1].0043-1648/$–see front matter©2004Elsevier B.V.All rights reserved. doi:10.1016/j.wear.2004.03.061D.T.Eadie et al./Wear258(2005)1148–11551149Table1Frequency range for different types of railway noiseNoise type Frequency range(Hz) Rolling30–5000Flat spots50–250(speed dependant) Ground borne vibrations4–80Structure–borne noise30–200Top of rail squeal1000–5000Flanging noise5000–10000 Rolling noise is established as originating from structural vibrations of the wheel,rail and sleepers resulting from the combined surface roughness of the wheel and running sur-faces.Roughness on wheels can be induced by factors such as the use of tread brakes,especially those made from cast iron.Ground borne vibrations and structure-borne noise mainly occur at low frequencies(<50Hz).Frequencies above this are attenuated increasingly rapidly[2].Vibration disturbance is usually caused by the large vertical dynamic forces be-tween wheels and rails.These forcesfluctuate in response to wheel and rail roughness over a wide range of frequen-cies.Wheel squeal originates from frictional instability in curves between the wheel and rail.Stick-slip oscillations (more accurately referred to as roll-slip)excite a wheel resonance;the wheel vibration radiates noise efficiently. The accepted model involves TOR frictional instability un-der lateral creep conditions leading to excitation of out of plane wheel bending oscillations.These are radiated and heard as squeal.The most recent developments involve more rigorous mathematical modeling by Heckl[3].The start-ing point for squeal is lateral creep forces that occur as a bogie goes through a curve and the wheel/rail contact patch becomes saturated with slip(creep saturation).A crit-ical component in all the modeling work is the require-ment that beyond the point of creep saturation,further in-creases in creep levels lead to lower coefficient of fric-tion.This is known as negative friction,referring to the slope of the friction creep curve at saturated creep con-ditions.In more general tribological terms,this would be equated to changes in sliding velocity,rather than the rail-road term creep.This leads to roll-slip oscillations be-tween the wheel and the rail which excite a wheel res-onance,and the wheel web radiates the noise.Most re-cently,De Beer et al.have developed a frequency domain model of squeal noise in terms of lateral contact position, and supported the results with laboratory measurements [4].“Graunching”is another form of noise associated with curving.Unfortunately little is known in the literature about this phenomenon.This is a low frequency phenomenon,of perhaps several,probably associated with stick-slip of a dif-ferent form than that of top of rail squeal.It may possibly be due toflange rubbing.2.Field experimentation2.1.Wheel/rail system types and characteristicsFor this purpose the wheel/rail systems examined have been classified into three general types:1.Trams(Europe and Japan)and light rail(North America),with axle loads typically9–11t.2.Metro(Europe)and heavy rail(North America).Theseare characterized by moderate axle loads and heavier rail sections than in trams(11–15t).3.Heavy haul freight(axle loads20–35t).Each test is further characterized according to the follow-ing parameters:1.Curve radius.2.Speed.3.Number of axles per car.4.Number of cars.5.Track type.6.Brake type.7.Gauge face lubrication practice.8.Friction modifier application method.2.2.Sound measurementsSound level measurements were carried out using a Bruel &Kjaer2260Sound Level Meter.The microphone wasfitted with a foam windscreen and handheld on the outside(high rail side)of the test curve.The sound meter was held7.5m from the center of the track,with the microphone1.2m above the height of the rail.The maximum sound level range was preset to between65and130dB for train passes(35–100dB for ambient).Sound measurements were made in the center of the curve.The sound level meter was programmed for event recording,enabling the instrument to automatically measure and stores the event data.The logged data was downloaded via a serial interface to a laptop computer.In all but one case the data reported is the averaged L Leq values for each of sev-eral trains under the particular conditions evaluated.L Leq is the linear weighting of the average noise level.Measurements were made for the residence time of the train in the curve.For trams,the residence time was approximately15s,for Metros about30s,and for freight trains about2min.Five or more trains were measured and the data in this paper is always the average for all the trains measured.For System8,L Amax was measured with a Rion NA-24m set for A weighting,which measured only at certain selected frequencies.2.3.Vibration measurementsVibration measurements were carried out in a track build-ing located about10m from the trackside friction modifier delivery system,near the beginning of the test curve.This location was less than ideal,but access to local office build-1150 D.T.Eadie et al./Wear258(2005)1148–1155ings was not available.A Norsonics Model121analyzer was used to collect the data.The analyzer was set to collect1/3 octave band spectra for1s periods and was operated continu-ously during the testing.The vibration results were converted from acceleration to velocity,and all the vibration results in this paper are in terms of velocity decibels using a decibel reference of dB re2.5×10−8m/s(1␮in./s).2.4.Friction modifier characteristicsThe top of rail friction modifier used in this work is a water-based liquid material known as KELTRACK®[5–7]. After water removal,a thin dryfilm remains that provides an intermediate coefficient of friction in the range0.30–0.35 as measured with a push tribometer.This friction level does not compromise braking or traction.The differences betweena true friction modifier and a lubricant have been reviewed[8].Friction levels have not been specifically reported in this paper,but the reader is referred to a previous publication for typical friction values[11].Typical top of rail friction levels will be0.5–0.6,and after application of friction modifier will be controlled in the range0.35±0.4.The friction modifier is a suspension of engineered solids that provide the required frictional properties.Other compo-nents are present in order to:(1)Control application performance in the delivery mecha-nism.(2)Maximizefilm strength and endurance,a parameterknown as“retentivity”.The positive friction characteristics of this friction modi-fier thinfilm have been established in two roller rig studies in Japan[9]and in other studies[6].The frictional properties of the thinfilm are the net result of the friction modifier and the other Third Body compo-nents.(The Third Body refers to the all the components of layer between wheel and rail.)The friction modifier has been designed for optimal interaction with the iron oxide wear components that dominate the Third Body under normal con-ditions.Thefilm characteristics have also been developed to provide the appropriate durability so thefilm lasts for as many axles as possible.2.5.Friction modifier applicationThe friction modifier can be applied by a number of al-ternative methods such as spray from on-board locomotives or transit cars[10],using a trackside applicator,or from a track maintenance vehicle.In this work the water-based liq-uid friction modifier was applied to the top of rail either by hand application using paint rollers,or automatically using a special applicator from Portec Rail,the Protector Top of Rail system[7,11].For hand application,a target application rate of0.3g/m per rail was used.The friction modifier was applied to the tangent immediately before the curve and through the body of the curve.In all cases except for System8,frictionFig.1.Typical trackside top of rail applicatorinstallation.Fig.2.Close up of top of rail application bar.modifier was applied to the top of both rails(inner and outer). Tests on System8examined the difference in sound reduction for application to the top of the low(inner)rail only versus both rails.Fig.1shows the automatic trackside application system.Two specially designed wiping bars are mounted on thefield side of each rail,and these are shown in Fig.2. 3.Test results3.1.Spectral sound patterns across different rail systemsSpectral sound patterns in curves werefirst compared without friction modifier between the three general types of wheel/rail systems,Fig.3.This data is the same as the“baseline”spectra for each separate system shown in Figs.4–8,10and11.Corresponding information on curve radius,train speed,etc.can be found in Tables2–4.The spec-tral sound patterns for the three systems are rather different. The tram systems have the lowest overall sound levels.Nev-ertheless there are distinct maxima for the trams within the wheel squeal range,e.g.at1250and2500Hz.This is con-sistent with the human experience at these sites,where thereD.T.Eadie et al./Wear 258(2005)1148–11551151Fig.3.Sound spectrum:different wheel/rail systemtypes.Fig.4.System 1,site A,tram.Fig.5.System 1,site B,tram.was a substantial amount of annoying wheel squeal in all cases.The Metro systems show a generally higher overall sound level and less distinction between rolling and squeal noise contributions.(All the Metro sites are underground.)The freight systems generally show a larger contribution fromtheFig.6.System 2,tram.Fig.7.System 3,Metro.Fig.8.System 4,Metro.flanging frequency range than the other systems,presumably because of the larger lateral and flanging forces associated with the high axle loads.3.2.Effect of friction modifier:tram systemsThe effect of friction modifier was examined on the spec-tral sound distribution of two tram systems.Measurements were taken at two curves on System 1and one curve on System 2.Characteristics of the tram systems are shown in Table 2below,which also provides average sound levels with and without the friction modifier.Spectral sound characteristics with and without friction modifier are shown in Figs.4–6.Table 2Test site characterization,tram systemsSystem 1,site ASystem 1,site B System 2System type Tram Tram Tram Curve radius (m)191935Speed (kph)161616Axles per car 886Cars per train 111Track typeImbedded U-rail Imbedded U-rail Imbedded U-rail Brake typeDisc Disc Disc Gauge face lubrication No Yes No FM application method Manual Manual Manual Average L Leq baseline,dB re 2×10−5Pa 83.392.483.3Average L Leq friction mod,dB re 2×10−5Pa71.280.672.51152 D.T.Eadie et al./Wear258(2005)1148–1155Table3Test site characterization,Metro systemsSystem3System4System type Metro(underground)Metro(underground) Curve radius(m)9790Speed(kph)832Axles per car44Cars per train1010Track type Wooden ties/concrete Wooden ties/concrete Brake type Tread TreadGauge face lubrication Yes YesFM application method Manual Trackside applicator Average L Leq baseline,dB re2×10−5Pa101.5105.9Average L Leq frictionmod,dB re2×10−5Pa91.998.6Figs.5and6indicate that noise in the wheel squeal and flanging parts of the spectra have been significantly reduced with the friction modifier.At frequencies below the squeal range there is little change in the spectra.3.3.Effect of friction modifier–Metro systemsTable3details the system characteristics for the two Metro systems selected for reporting,and Figs.7and8show the effect of friction modifier on the spectral sound distribution for these systems.The average sound levels in the Metro systems are rather high across the entire spectrum.Although there are no spe-cific peaks in the part of the spectrum assigned to squeal,to the human ear at both Metro sites the presence of squeal was very apparent.Based on this,although the spectra are rel-atively featureless there must be a considerable broad-band contribution which may be due to rolling noise,squeal,or flanging noise.It may be that the relatively high overall sound levels are obscuring the spectral maxima apparent in the tram e of narrow-band spectra rather than1/3octave would help clarify this point,and future studies will use this methodology were appropriate.Application of friction mod-ifier to the top of rail in these cases affects a broader range of the spectrum than in the tram case.One possibleexplana-Fig.9.Average vibration spectra,System4,Metro.tion is that there is significant contribution from“graunching”noise at low frequency,and that these have a stick-slip oscil-lation mechanism.If so,the positive friction characteristics of the friction modifier might be expected to mitigate this noise source.Reductions in sound levels in these cases are observed right down to30Hz.3.4.Effect of friction modifier on vibrations on System4 (Metro)During the sound testing described for System4,vibration was also measured in the curve.Vibration levels before and after friction modifier application are shown in Fig.9.Although the test site was less than ideal,the measure-ments suggest some reduction in vibration levels in the 30–60Hz frequency range.The measurements at frequencies below30Hz are of limited value as the trains were too close to the measurement site.The test results also show vibration reduction at the high frequencies(>1500Hz)where the fric-tion modifier is effective as a result of controlling slip-stick interaction at the wheel/rail interface.Further investigation of this area is clearly warranted.3.5.Effect of friction modifier–heavy haul freightTable4outlines system characteristics for three different heavy haul freight systems.Table4Test site characterization,freight systemsSystem5System6System7System type Heavy haul freight Heavy haul freight Heavy haul freight Curve radius(m)291200148.5Speed(kph)323232Axles per car444Cars per train1006080Track type Concrete ties/ballast Concrete ties/ballast Wooden ties/ballast Brake type Tread Tread TreadGauge face lubrication No Yes VariableFM application Trackside applicator Trackside applicator Trackside applicator Average L Leq baseline,dB re2×10−5Pa90.6102.4Average L Leq friction mod,dB re2×10−5Pa81.886.9D.T.Eadie et al./Wear 258(2005)1148–11551153Fig.10.System 5,heavy haul freight.Results before and after friction modifier applications by automated trackside delivery system are shown below in Figs.10and 11Systems 5and 6,respectively.The results show a distinct reduction in squeal and higher frequency flanging noise.Changes in rolling noise are marginal at best.3.6.Identification of flanging noise spectral pattern Because flanging noise sounds to the human ear like a grinding or hissing noise,it is sometimes assumed that this is associated with lower spectral frequencies.The following examples illustrates the distinction between top of rail squeal and flange contact sound spectral patterns.The first example is from a North American freight rail sharp curve which has been monitored for several years (Sys-tem 7,Table 4).In this curve the trackside gauge face grease lubrication is used to control wear,and the friction modifier,applied by an atomized spray to the top of rail [11],controls the wheel squeal noise.On one particular occasion,local res-ident complaints prompted us to carry out additional sound level monitoring.To the trained ear,it was clear that the sound (which was indeed quite irritating)was the hissing sound typ-ical of flange contact.It turned out that the trackside grease lu-brication units had become non-functional.When these were repaired,the hissing flange noise disappeared.The spectral sound patterns are shown in Fig.12.This figure shows (A)trains with some TOR squeal,(B)TOR application with no gauge face lubrication (flanging noise),and (C)full control of train noise with both TOR friction control and gauge face lubrication.This result does not however address thequestionFig.11.System 6,heavy haulfreight.Fig.12.(A)Some squeal,(B)TOR only,(C)TOR plus gauge face lubrica-tion.of the spectral frequency of “graunching”or grinding which may also be associated with flange contact.Further work is needed in this area.The results on Metros and trams although limited,suggest that in some cases top of rail friction control alone can elim-inate both top of rail squeal and higher frequency flanging noise.(This conclusion is undoubtedly site specific.)How-ever in heavy haul freight applications,the higher lateral and flanging forces in sharp curves require gauge face/flange in-terface lubrication in addition to top of rail friction control to fully control noise.This conclusion is also probably site spe-cific,and further testing is needed to explore the limits and track conditions that affect these conclusions.The dual top of rail and gauge face application may only be required in sites with specific conditions of curvature,track superelevation,axle load,bogie type,etc.The second example is from a tram system in Japan.Sys-tem details are shown in Table 5.Fig.13shows that with application of friction modifier to the top of the low rail only,there is a reduction in sound levels in the range from 1000to 4000Hz,but little change at frequencies above 4000Hz.In this case the figure shows the spectra of the 1s highest A-weighted level.However with application of friction modifier to the top of both rails,the high frequency component is further reduced.The fric-tion levels on the top of the low rail are primarily respon-sible for reducing gauge-spreading forces,regardless of top of high rail friction levels.However,the friction levels on both rails affect the reduction in flanging force [12].Hence providing controlled reduced friction on the high rail willTable 5Test site characterization,System 8System 8System type Tram Curve radius (m)160Speed (kph)24Axles per car 4Cars per train 1Track type Wooden ties/ballast Brake typeTread Gauge face lubrication NoFM applicationTrackside applicator Average L Amax baseline (dBA)90.8Average L Amax friction mod (dBA)85.71154 D.T.Eadie et al./Wear258(2005)1148–1155parison of friction control on low rail only vs.application to both rails.further reduceflanging forces,and consequentlyflanging noise.4.DiscussionThefield data described shows considerable variation in spectral sound characteristics between different wheel/rail systems in curves.All systems considered show contribu-tions from bothflanging noise and top of rail wheel squeal. Application of top of rail friction modifier reduced these noise sources in all systems considered.The reduction inflanging noise in most systems tested in transits(trams and Metros)is interesting.In systems where conventional gauge face lubrication is used,there is still a reduction inflanging noise with the application of friction modifier(see Figs.5,7and8).In cases where there is no gauge lubrication,the friction modifier reducesflanging noise to an even lower level(see Figs.4and6).Reduction inflanging noise is because the friction modifier reduces TOR friction from an average“dry”state of0.45–0.6 down to an optimum level around0.35as measured with a push tribometer.The result is a corresponding reduction in lateral forces andflanging forces.It is well known that the presence of gauge face lubrication increases the angle of attack and the magnitude of lateral forces.The top of rail friction modifier is effective enough to mitigate the increase in lateral forces caused by gauge face lubrication[12].The other outcome is,in some cases,elimination of high rail gauge face contact with associated dramatic reduction in rail wear[11].The reduction in rolling noise in some cases but not in others is of interest.The mechanism for this noise reduction is not yet entirely clear.Future work will investigate the possible role of rail roughness changes with friction modifiers and how this might be related to the results reported in this paper. 5.ConclusionsThis paper has shown that friction modifiers can reduce overall noise in curves across a wide range of wheel/rail sys-tems,as shown in Fig.14.Fig.14.Summary of average sound level reductions.This work also shows that in practical railways there is a large variation in absolute sound levels and spectral pat-terns.These have been characterized across trams,Metro, and heavy haul freight.The results show that:•Friction modifiers reduce squeal noise across all systems considered.•Friction modifiers reduceflanging noise in all transit sys-tems tested,but not necessarily in freight,where effective gauge face lubrication may also be required because of the higher lateral andflanging forces,especially in sharper curves.•For systems with highest overall noise levels,the noise tends to be reduced across a broader part of the spectrum with friction modifiers.•In one case,some reduction in low frequency vibration has been observed with friction modifier application.Ongoing work is examining in more detail the mechanisms by which friction modifiers reduce low frequency noise and vibrations,as well as the effects on wheel and rail rough-ness.Future work should also consider the potential impact of friction modifier on roughness levels on tangent track.The controlled intermediate friction is expected to reduce the rate of roughness increase because of reduced longitudinal and friction forces.AcknowledgmentsThe following individuals are acknowledged for their helpful contributions:Mr.David Elvidge for the freightflang-ing noise measurements(System7),ATS Consulting(Dr. Hugh Saurenman)for the vibration measurements,Dr.Y.Oka for the sound measurements on System8,One of the referees for suggesting that“graunching”was responsible for some of the low frequency noise.References[1]C.Talotte,P.E.Gautier,D.J.Thompson,C.Hanson,Identificationmodeling and reduction potential of railway noise sources,in:Pro-D.T.Eadie et al./Wear258(2005)1148–11551155ceedings of the Seventh International Workshop on Railway Noise, Portland,Maine,October2001.[2]C.Esveld,Modern Railway Track,2nd ed.,TU Delft,Netherlands,2001,ISBN90-800324-3-3.[3]M.A.Heckl,Curve squeal of train wheels.Part3.Active control,J.Sound Vibrat.229(3)(2000)709–735,reference therein.[4]F.G.De Beer,M.H.A.Janssens,P.P.Kooijman,Squeal noise of railbound vehicles influenced by lateral contact position,in:Proceedings of the Seventh International Workshop on Railway Noise,Portland, Maine,October2001.[5]K.S.Chiddick,Solid lubricants and friction modifiers for heavy loadsand rail application,US Patent6136757(October2000).[6]D.T.Eadie,J.Kalousek,K.S.Chiddick,The role of high positivefriction(HPF)modifier in the control of short pitch corrugations and related phenomena,Wear253(1–2)(2002)185–192.[7]D.T.Eadie,M.Santoro,W.Poll,Local Control of Noise and Vi-bration with KELTRACK Friction Modifier and Protector TracksideApplication:An Integrated Solution,Seventh International Workshop on Railway Noise,Portland,Maine,October2001.[8]D.T.Eadie,J.Kalousek,Railway Age(2001)48.[9]A.Matsumoto,Y.Sato,H.Ono,Y.Wang,Y.Yamamoto,M.Tani-moto,Y.Oka,Creep force characteristics between rail and wheel on scaled model,Wear253(1–2)(2002)199–203.[10]M.Tomeoka,N.Kabe,M.Tanimoto,E.Miyauchi,M.Nakata,Fric-tion control between wheel and rail by means of on-board lubrica-tion,Wear253(1–2)(2002)124–129.[11]D.T.Eadie,N.Hooper,B.Vidler,T.Makowsky,Top of rail frictioncontrol:lateral force and rail wear reduction in a freight application, in:Proceedings of the International Heavy Haul Conference,Dallas, May4–8,2003.[12]J.Kalousek,Rolling radius difference:Do appreciate it’s signifi-cance?National Research Council of Canada Centre for Surface Transportation Report CSTT-VTD-54-AAR,June2001.。