Techniques for Controlling Rail Corrugation

Transcription

1 32 Transportation Research Record 16 Techniqes for Controlling Rail Corrgation SO'N T. LAMSON ABSTRACT Corrgation on lo rails in crved track is a common and significant problem on North American railays. Research findings on probable cases are described and practical methods ot keeping this problem nder control are sggested. The role of heel and rail contact stresses in the plastic deformation of rail srface, a necessary condition for corrgation development, is examined, and the se of high-strength steel and rail profile grinding to control rail corrgation throgh redcing or preventing rail srface plastic deformation is discssed. Reslts of recent field trials of rail corrgation control sing a rail profile grinding techniqe are presented. c..:rrent interest in heavy LL~ighL o.11 high-spc~d passenger train operations has broght increasing attention to the centry-old problem of rail corrgation--a form of periodic rail srface deformation that cases severe vibration in vehicles and track. Vibration in trn generates noise, an environmental ~:::-~b!.e~ thet h~~ ~"?'.:'~~ ::- i:t,:t.n!q;t-ivp irrl1a in itp.nsp.ly poplated areas. In terms of vehicle and track maintenance, vibration can case severe damage to heels and rails as ell as to ties and ballast. Railay rail corrgation is generally classified into to grops according to avelength. Long-ave corrgation has avelengths on the order of one-half to one tie spacing, and a peak-to-peak amplitde from.2 to.7 in. or more. In North America it is common to heavy freight train operations sch as coal lines. At the trogh of the corrgation, here the heavily loaded heels (3, lb or more) repeatedly pond the rail, severe plastic flo and fatige damage are often observed. Short-ave corrgation has avelengths in the range of 2 to 4 in., and peak-to-peak amplitde is sally less than O. 5 in. It is more common in high-speed passenger train operations and in light rail transit systems. This problem (called "roaring" rail) has long been of considerable concern in Erope. With the recent stepped p development of light rail transit systems, this problem is receiving increased attention in North America. Short-ave corrgation does not sally case serios damage to vehicles and track strctres, at least in the short term. Hoever, the noise and vibration, especially in ndergrond transit systems, are particlarly troblesome becase of the increasing sensitivity of the pblic. In North America, long-ave corrgation is a common problem on lo rails in crved track and reqires periodic grinding. The cost of grinding cold rn into millions of dollars for a major railroad. Withot grinding, hoever, the conseqences old be far orse: speed restrictions or increased rail fatige failre, or both, de to severe heel and rail vibration. In some heavy-hal railay operations ith high axle loads (p to 33 tons) and high traffic density (p to 5 million gross tons (MGT) a year on single trackl, corrgation cold gro from a fe thosandths of an inch to beteen.2 and.4 in. in 6 months. At this level there is sally severe heel- and rail-contact-generated track vibration, and speed restrictions are sometimes necessary (actal imposition depends on, among other things, the sensitivity of a particlar track formati~~ t~ :ib~~ti~:!}. '!"c-!!!:int:in th'? rpqnir~d track capacity, these slo orders mst be removed as soon as possible. If rail grinding is not available, the rails ill have to be replaced. Ths, if slo orders are not acceptable and rail grinding is not available, corrgation cold drastically redce the service life of rails. A cost srvey by Roney (1) sggests that the cost of rail corrgation to tne Canadian railays is abot $3 million ( 1984 dollars) a year. The focs of this paper is mainly on the North American long-ave corrgation problem, ith identification of practical measres that cold help redce its magnitde. CORRUGATION THEORY To find a permanent soltion to the problem of rail corrgation, its case or cases mst be identified. To date there have been a nmber of theories--some based on general observations, others based on indepth stdy. Of the latter, there are to categor ies: contact resonance and stick-slip. The contact resonance theories consider resonance vibration of the heel and rail system as the primary case of rail corrgation. Sorces of resonance freqency may be fond in varios possible vibration modes of heels, axles, rails, and rail and tie systems. The stick-slip theories concentrate mainly on the frictional contact instability beteen heels and rails dring vehicle movement throgh crves. These theories, althogh not yet proven, cold be sed to explain the corrgation phenomenon in crved track nder heavy axle load. These theories are revieed in the light of the North American rail corrgation problem. Theories on the stick-slip phenomenon explain that the contact beteen heel and rail is not stable bt alternates from stick to slip. Dring the stick oyole rail ear is small, bt dring the slip cycle significant ear occrs becase of relative heel and rail sliding nder very high contact stresses. King and Kalosek (l_) are among a nmber of researchers ho explain the stick-slip theory in the context of heelset crving behavior. rhis theory hypothesies that dring crve negotiation, the effective rolling radis is the same on both heels, even thogh the heel on the high rail (oter) side has a l onger traveling distance. To accommodate this difference, the heel on the bigh rail side mst

2 Lamson 33 rotate faster than the one on the lo rail side. Becase the heelset is an integral component, this instantaneos rotational speed differential places the axle in torqe ntil the torsional reaction becomes greater than the adhesion forces at the heel and rail interface. The axle then tists back, reslting in heel and rail slip. The stick-slip cycles set the axle into torsional vibrations hich in trn perpetate the stick-slip cycles, casing corrgation to increase. The assmption of eqal effective rolling radis on high and lo rail sides does not hold, becase, for a conical heel tread in a flanging configration, the heel on the high side travels on a larger heel radis. The stick-slip theory is, hoever, still valid if a torqe of significant magnitde exists on the axle. varios stdies ~nto the crving behavior of three-piece freight trcks have shon that large creep forces e1ist beteen heel and rail interfaces, and that for a given heelset, especially the leading one, the creep force orientation is sch that a large torqe is applied to the axle (3-5). Figre 1 shos an example of a force diagram fo; a leading heelset as predicted by a steady-state, trck-crving compter model (6). The magnitde of torqe on the axle is abot- 22, lb/ft in a 3-degree crve. Apart from the existence of a sfficiently large torqe, the occrrence of stick-slip frther reqires that heel and rail friction have a falling characteristic at high creepage and that the heel and rail creepage condition be severe enogh to give an operating point in the falling portion. 6.7 kips.14 3 crve 1-lon ccr 1" over sperelevotion - -- FIGURE 1 An example of force diagram for the leading heelset in a crve. > CURVE CENTER Figre 2 shos some sample friction-creepage crves for varios heel and rail contact conditions. It is evident that a falling characteristic de to creep satration occrs at a fe percent creepage for all cases except those ith ater and oil contamination. Hoever, most experimental reslts for the loer range of creepage, sch as those in Figre 3, sho a somehat different characteristic. These indicate that creep satration begins at abot 1 percent creepage bt sho no falling characteristic p to 3 percent creepage. Figre 4 shos the resltant creepage on lo rail in a range of crves. These data are derived from compter simlation of the crving behavior of a three-piece freight trck ith 35-ton axle load. Generally creepage increases ith the degree of crvatre and is abot 1 percent in 8-degree crves. These data (from Figres 3 and 4) indicate that stick-slip as a reslt of friction and creepage falling character is tics is not likely. It may be sspected, hoever, that in the field other conditions, sch as heel and rail contamination and vibration, may exist bt are insfficiently acconted for in the laboratory. These may create the necessary falling characteristics for stick-slip to occr. In a recent paper, Clark <1l reported experi- i= 3 " FRICTION-CREEP CURVE 4 SAND WATER-SAND EFFECT OF RAIL CONDITION TANGENT TRACK AVERAGE F-C CURVES TRACTION MOTOR I ALL TEST SITES PERCENT CREEP FIGURE 2 Average friction-creep crves. I CLEAN i.:.4 ".2 j:: ~ "- I 2 3 CREEPAGE S (%)- ~ MEASURED CURVES CURVES CALCULATED WITH THE MEASURED INITIAL GRADIENTS AND COEFFICIENTS OF SLIDING FRICTION FIG URE 3 Comparison of the longitdinal traction/creep relationship of clean rollers and rollers et ith ater. ;;; 12 I -;; 1 Vl 8 C> <t.. 6 I- 4 <t ~ 2 :::> Vl 2 4 s a DEGREE CURVE FIGURE 4 Resltant creepage verss degree crve at balance speed. mental ork to verify the stick-slip theory. The reslts sho that for a creepage condition greater than l percent, considerable vibration occrred, and in some cases a corrgation-like pattern appeared on the rail srface. PLAS'rIC DEFORMATION stick-slip vibration may create periodic variation in the tangential force component at heel and rail

3 34 Transportation Research Record 16 contact. Hoever, for this force to generate periodic plastic deformation as observed in typical North American corrgated rails, the force magnitde mst exceed the elastic limit of the rail steel. Ths, it is necessary to kno the magnitde and natre of the heel-loading environment at the heel and rail interface, as ell as the resistance of the rail steel to plastic deformation nder actal track operations. A sample calclation that illstrates the magnitde of the heel-loading environment relative to the strength of the rail steel follos. For a 1-ton car ith 36-in. diameter heels, rnning at slightly nder balance speed (1 in. over-sperelevation) in a 3-degree crve, the heel loading on the lo rail is: 1. Vertical: Static heel load Vertical load transfer Steady-state heel load Dynamic load factor (assmed) Dynamic heel load Compressive contact stress: average heel and ne rail (tread is slightly hollo ith a concave radis of 3 in.) ne heel and ne rail (tread is l in 2) orn neei ana ne ra1i (tread is hollo ith false flange radis of 4 in.) 32, lb 3,2 lb 35,2 lb 1% 38,72 lb 195, psi 227,1 psi 43,4 psi 2. Tangential (estimation based on steady-state crving simlation of three-piece trck) : Lateral creep force Longitdinal creep force Resltant creep force 3. Ratio of tangential to vertical 8, lb 7, lb 1,6 lb force.275 Both standard carbon and lo-chrome alloy rails are sed in North America. Carbon rails are sed mainly in tangent track and shallo crves (e.g., less than 3 degrees) Lo chrome rails are sed in sharper crves as a measre against the more severe heel-load environment. Yield strengths of standard carbon and chrome rails are abot 75, psi and 1, psi, respectively. Even after taking into accont a constraint fr actor of 1. 7, these yield strengths are mch loer than the estimated range of contact stresses (195, to 43, psi). Therefore, plastic deformation of rail old occr nder the first heel passage. For sbseqent heel passages, the residal stress created nder earlier heels improves the resistance of the rail steel to frther plastic deformation. This phenomenon, called shakedon, has been the sbject of intensive research in the field of contact mechanics. Johnson (_!!), in a recent comprehensive revie of this sbject, proposed an estimate for the elastic and shakedon limits applicable to rail steel in rolling contact as shon in Figre 5. According to this estimation, the shakedon limit for standard carbon and chrome rails, in the absence of tangential force, is 22, psi and 27, psi, respectively. Hoever, if a tangential force of O. 275 times the vertical is also present, these shakedon limits are redced to abot 15, psi and 2, psi for standard carbon and chrome rails, respectively. These estimates indicate that: 1. Standard carbon rails old be plastically deformed nder the passage of most heels, and 2. Chrome rails old be plastically deformed nder the passage of ne and orn (i.e., ith false flange) heels. For average heels, the likelihood SUB-SURFACE} A LOWER (HILLS) BOUNDS B UPPER (JOHNSON) - 3 SURFACE : C (RYDHOLMJ ~ B Q. :;J ~ 2 >< <{ a:: :r I "' VJ..._o o~ ;;::: ~o a:: I I :::;; 3: :r ol- SUB SURFACE SHAKEDOWN ----LINE CONTACT --- POINT CONTACT ~ '----'-~-'--~--'-~--'-~-'---' TIP RATIO OF TANGENT FORCE TO VERTICAL FORCE FIGURE 5 Elastic and shakedon limits in rolling contact. of plastic deformation is less than certain, oecase the shakedon limit nder.275 creepage is only slightly belo the contact stress. For a track carrying 5 mgt of 32-ton axle loads per year, and assming a heel condition distribpercent orn, the plastic deformation cycle for standard carbon rail and chrome rail old be abot 1,5, and 5, times a year, respectively. Ths, crved track is generally too hostile an environment for the standard carbon and lo allo rail steels crrently in common se. These rails old likely corrgate in the presence of vibration. PRACTICAL SOLUTION The preceding discssion indicates that lo rails in crves are sbject to sch a severe loading environment that nless hardened rails are sed, there is potential for idespread corrgation. High heel loads (de to heavy cars) and creepage (de to poor crving of three-piece trcks) place the present rail steel on the threshold of plastic deformation. With heel and rail contact being steel on steel (lo damping), and ith limited adhesion (stickslip), there is potential for vibration. Uniformly spaced tie spport cold add to the vibration prob.lem as ell. Knoing that these to factors, that is, pla.stic deformat ion and contact vibration, are essential in the development of corrgation, the soltion obviosly is to eliminate or redce their severity. To achieve this goal hen there is no option of replacing existing trcks ith ones of steerable design, or replacing existing track ith a continosly spported strctre, or redcing axle load, the athor believes the folloing remedial actions cold be taken: 1. Profile grinding. At present, rail is grond periodically to remove corrgation. This grinding operation can be adapted to prodce an optimal rail profile as ell, ithot additional cost. The objective is to create a rail profile that conforms to existing heel profiles, given the existing range of track gage error. On the lo rail, the primary aim is to ensre that heel and rail contact takes place aay from the false flange shold the heel be orn and the track gage iden. If this is achieved, a potential contact stress of 43, psi nder orn heels can be redced to 195, psi, making the lo

4 Lamson 35 HEAD HARDENED RAIL ~. 4 ~. 'J :;.o ~. 7.o SILICON CHROME RAIL.O'H , , ,... CHROME MOLYBDENUM (A) RAIL "" ~2 >- ~ (FEET ) FIGURE 6 Typical initial rail profiles. 3 4 o <;~::.,...~~~~~~ A ( INCHES) 5 rail less ssceptible to plastic deformation. This techniqe ill be described in the case stdy. 2. Grinding of ne rail. The srface of ne rail normally is not smooth. Apart from mill scales, there are irreglarities ith avelengths from a fe inches to several feet. An example is shon in Figre 6. This roghness can excite heel and rail vibration at a time hen the rail steel has not yet hardened to its potential. Early removal of srface roghness old redce the initial plastic deformation and prodce a smoother hardened rail srface at a later stage. 3. Sperelevation. Track shold be elevated at balance or at a slightly deficient level to avoid heel load transfer to lo rails. In a mixed-traffic sitation, sperelevation shold be designed to accommodate the heavier traffic categories. 4. Heat-treated rail. Heat-treated rails sch as the head-hardened type cold be sed to improve rail resistance to plastic deformation. With a yield strength (2 percent proof stress) of abot 125, psi, they old be able to resist plastic deformation nder most heel loads except the false flange type. When sed ith a profile grinding program, hich is effective in avoiding false flange contact, heattreated rails old provide satisfactory resistance against plastic deformation, and hence rail corrgation. 5. Rail lbrication. The standard practice of lbricating the gage face of the high rail effectively transfers some of the creepage from lo rail to high rail. Figre 7 is a three-piece trck crving simlation reslt shoing the redction of longitdinal creepage and creep force on the lo rail as the friction factor on the high rail gage face decreases. This has the dal effects of redcing the likelihood of heel slip and increasing the shakedon limit of the lo rail. Daniels and Blme (9) reported a correlation stdy at FAST in hich ratl corrgation groth as mch loer dring the periods hen rail lbrication as sed than dring the...j.6 ~.5 It: :i: ~.9 <[ ~.oe ::!.7.6 -;;; a. ~ 1 gg ir e ~ 7...J 6 5 l.j 4 It: i CURVE, 25mm (l in.) OVER-BALANCE SUPERELEVATION 1 TON CAR AT Q. 2 6 km/h (35mph) l.j l.j It: I. I LUBRICATED-- o----- ORY AVERAGE HIGH RAIL GAUGE FACE FRICTION FACTOR FIGURE 7 Effect of rail lbrication on creepages and creep forces. periods ithot lbrication (Figre trackside lbricators are already being railays, a field trial to validate cold be carried ot qite readily. 8). Becase sed on most this effect There are also other measres that may be less readily achievable bt are nonetheless feasible: 1. Wheel remachining. More freqent heel inspection and remachining old redce the percentage of defective heels sch as those ith false flange or flat spots. If this action is coordinated ith a program of profile grinding and the se of heattreated rail, the problem of rail corrgation old be greatly redced. For several years the Hamersley Iron and Mont Neman Mining Railays in Astralia (33-ton axle loads) have been experiencing minimm rail corrgation problems sing the preceding strategy.

5 36 Transportation Research Record 16 a_ Ell "' "'<1J "' "'.c: > t: "'"' V> - t: "' ><.. "' ,.,,, o.k.:'.:..._~~~~~- --LUBRICATED NON-LUBRICATED ~---~~ c: +' ~4l ;-... -' 3 I ~:;: 2 / ~ ~ }i --.< I I I< I I I >-==<" ~ 5 1 MILLION GROSS TON (Adjsted) FIGURE 8 Rail corrgation groth by lbrication condition; standard carhon rail, 5 degree crve, hardood ties. 2. Track gage control. The biggest problem ith timber track, as far as corrgation is concerned, is gage idening, hich cases false flange contact on lo rails. The se of concrete ties solves this problem bt introdces high dynamic load becase of l e1..i._y t:: L :..le it1iioo ~r;.:1 ::.t!!'f = ::!.!. ~~.1~ F~!' ~ imh~r t rack, a practical soltion cold be to insert one steel tie, for example, every five tie spacings. The spacing, hoever, s hold not be niform (i.e., nc al.ays one every f ive ties.i as this may introdce a periodic stiffness variation. Gage rod is another, perhaps less practical, option. 1,..--~~~~~~~---,-~~~~~~~~~ 5 MAG MPH - BEFORE GRINDING 2 CURVE 8 CURVE 3 MPH - AFTER GRINDING - 2 CURVE 5 '- MAG ' HZ CURVE HZ FIGURE 9 Measrements of rail vibration. gage idening and heel ear creates an effective total gage Hlening nder heel. passage of abot.5 to. 75 in. {.25 in. flange ear +.25 in. gage face ear + p to. 25 in. dynamic gage id ~~i:;;~. Th:!.:: l~~ t~.:!. ~i~~ n~ f~l RP. flange contact problem, as shon in Figre 1. FALSE F ~ ANG E CONTACT CASE STUDY: PROFILE GRINDING TEST ON CP RAIL (1) Otline In May and Jne 1982, a rail profile grinding test as condcted on a section of crved track on the Canadian Pacific (CP) main line throgh the Rocky Montains. The first objective as to redce the problem of corrgation on lo rails in crves. For this, to crves ere selected: an 8-degree crve ith severe corrgation (average depth O. 595 in.) (1.44 nn) on the lo rail (chrome), and a 2-degree crve ith typical corrgation (average depth.335 in.) (. 84 nn) on the lo rail (standard carbon) Measrements of rail vibration nder a 3-mph test train {three locomotives, five 1-ton cars) are shon in Figre 9. These indicate the severity of the problem, especially in the 8-degree crve. Investigation Track gages and heel profiles ere srveyed in order to design a sitable rail profile to redress the problem of corrgation on lo rails in crves. This srvey revealed that in crves here rail corrgation occrs, the static track gage is typically.25 in. ider than sal. Under traffic, there ill be p to. 25 in. of additional gage idening de to lateral heel forces (typical magnitde of 4 to 12 kips otard on both high and lo rails). The srvey of heel profiles shoed that there as a large percentage of heels ith abot.25 in. flange ear, and a small bt significant -percentage of heels having false flange (i.e., reversed tread radis on the field side) The combined effect of FIGURE 1 Wheel and rail contact geometry in crves ith ide gage. Design of the Grond Rail Profile The preceding investigation led to the conclsion that false flange contact is a primary contribting factor to the problem of corrgation on lo rails. To avoid false flange contact {ithot resorting to the remachining of orn heels and the prevention of track gage idening), the field side of the lo rail mst be grond don so that heel and rail contact takes place only on the gage side. Figre 11 shos diagrannatically hat mst be done. In reality, grinding of the field side had to be redced becase of the high cost of grinding and short track time. A cost-effective grinding program is one that INCH 3 ii MM ii It INCH 1 3 MM HIGH RAIL LOW RAIL FIGURE 11 Proposed grond rail profile for crves on CP Rail.

6 Lamson 37.s::.7 c:.6 :r: ~.5 ::::; a. ::::> ::::;.4 x <{.3 ::::; i== <{.2 (!) ::::> er.1 er I KEY: D BEFORE PROFILE GRINDING TEST JUNE ABOUT 5 MGT AFTER PROFILE GRINDING TWICE MAY 83 LOW RAIL - STD CARBON - 2 TEST CURVE : I~ 1~ 1 1 ~ :~ :: :: :: :~ :! ~~ 2 ~~ 2 ~ ~~ 23 ~! ~: ~~ ~~ 27 ~: 2 9 CORRUGATION SITES (BASED ON TIE No.) FIGURE 12 Effect of profile grinding o n the formation of rail corrgation. removes only sfficient rail metal to avoid heel and rail contact in the false flange area beteen grinding sessions. At the time of the next session, this grinding can be repeated, hence excessive grinding is not necessary each time. Test Reslts Corrgation regroth as measred periodically after the test grinding. Figre 12 shos the corrgation levels in the 2-degree test crve 1 year after the test grinding. The rail had been sbjected to a total of to profile grinding sessions. The open bars represent corrgation levels existing abot 6 months (abot 25 mgt) after conventional (symmetrical) grinding. The fll bars represent corrgation levels 6 months after profile grinding. Except at 4 locations here minor increases occrred, the remaining 26 locations shoed a significant redction in rail corrgation levels, Frther Work Folloing the previosly described test, CP Rail adopted the designed profile rail as a provisionary standard for the main line. A more comprehensive field test involving 12 crves is being carried ot. The objective of this test is to frther examine the performance of the designed rail profile and to explore its refinement. ACKNOWLEDGMENTS This paper is based on a nmber of research projects that ere fnded either by the BHP Melborne Research Laboratories; Hamersley Iron Company; Mont Neman Mining Company joint research program; or by the Canadian Institte of Gided Grond Transport, Transportation Development Centre, Canadian Pacific Railay, Railay Advisory Committee; or Speno Rail Services joint research program. The athor is gratefl to these organiations for their spport. The athor also is very appreciative of the contribtion of Sherr ill Arnold in the preparation and editing of this paper. REFERENCES 1. M.D. Roney. Economic Aspect of the ear of Rail on Canadian Railays. Presented at International Symposim on Contact Mechanics and Wear of Rail/Wheel Systems, Vancover, B.C., Jly F.E. King and J, Kalosek. Rail Wear and Corrgation Stdies. AREA Blletin 656, Vol. 77, 1976, p D. Boocock. Steady State Motion of Railay Vehicles on Crved Track. Jornal of Mechanical Engineering Science, Vol. 2, No. 6, D.E. Neland. Steering a Flexible Railay Trck on Crved Track. Trans. ASME, Jornal of Engineering for Indstry, P.P. Marcotte. K.J.R, Matheson, and R. Yong. A Practical Model of Rail Vehicle Crve Negotiation. Presented at 7th Symposim of International Association of Vehicle System Dynamics, Cambridge, England, Sept s. Marich and S.T. Lamson. A Strategy and Procedre for the Development and Application of a Bogie Crving Model. Presented at Railay Engineering Conference, Sydney, Astralia, Sept R.A.P, Clark. Rail Corrgation--Recent Theories. Track Technology for the Next Decade. University of Nottingham, Nottingham, U.K., Jly K.L. Johnson. Inelastic Contact: Plastic Flo and Shakedon. Presented at International Symposim on Contact Mechanics and Wear of Rail/ Wheel Systems. Vancover, B.C., Jly , L.E. Daniels and N. Blme. Rail Corrgation Groth Performance. Presented at Second Heavy Hal Railay Conference, Colorado Springs, Colo., Sept S.T. Lamson. Rail Profile Grinding. CIGGT Report Canadian Institte of Gided Grond Transport, Qeen's University, Kingston, Ontario, Canada, Pblication of this paper sponsored by Committee on Track Strctre System Design.