The Railway Sleeper: 50 years of pretensioned, prestressed concrete
Transcription
The Railway Sleeper: 50 years of pretensioned, prestressed concrete
Paper: Taylor Paper The railwa sleeper: 50 years of pretensioned, prestresse concrete cy H. I? J. Taylor, BScTech, PhD, FEng, FIStructE,FICE Costain Building Products Ltd Synopsis This year,1993, marks 50 years fromthe introduction of the pretensioned, prestressed railway sleeper. This paper describes some history of the early development of prestressed sleepers and of our Codes. The success links the design development with that of the prestressed sleeper is self-evident, butthe paper does emphasise the successful way in which the prestressed sleeper copes with a most hostile working environment and provides an insight into the developmentof our design methodsfor prestressed concrete. Definitions Monoblock sleeper: a one-piece sleeper which may be made of timber, concrete or steel, supporting two rails and tying them together (Fig1). Rail block: a block of stone or concrete which supports a single rail (Fig l). Flat bottom rail Twin block sleeper:a sleeper consisting of two concrete blocks tied together with a metal bar (Fig 1). Bullhead track:rail track consistingof bullhead rail held in cast-iron chairs which are supportedby sleepers or blocks jointed in 60 ft lengths (Fig 2). Flat bottom rail track: rail track consisting of flat bottom rail frequently continuously welded into long lengths held by a fastening system directly to sleepers (Fig 2). On occasion, baseplates are used between the rail and the sleeper. Fastening system: the system of connection between sleeper andrail. This may be through a chair or with a tie or clip between the rail and sleeper. Modem fastening systems are elastic, incorporating a spring clip, a pad beneath the rail and insulators to provide electrical isolation betweenrails for track circuiting purposes. Bullhead rail track Introduction This paper discusses the introduction and use of pretensioned monoblock sleepers over the last 50 years. However, concrete has been used in block sleepers and, to a lesser extent, in reinforced concrete monoblock sleepers for 100 years in slow speed track. Monoblock sleeper L I L/ d 1 r 1 l Irl 0 - ,il Y Twinblock sleeper m Railblock Fig I . Sleeper types The Structural EngineedVolume 71/No 16/17 August 1993 Fig 2. Bullhead andflat bottom rail track The Stockton to Darlington Railway used stone blocks to support the rail in 1825 (Fig 3). Difficulties were experienced with the absence of ties to hold the rails together but, by the time that the Manchester and Liverpool Railway - the world’s first passenger railway - was opened in 1830, tied blocks were used. With the development of reinforced concrete, experiments were made in the first half of this century with RC sleepers. Some RC sleepers were tried in World War 1. The need to find a replacement for timber emerged at the start of World War 2 and, by then, early work on prestressed concrete led tothis being an option intrials. In 1941 two designs of reinforced concrete sleepers were produced by the Chief Civil Engineer’s Department of the London Midland and Scottish Railway, and some were made and put in a branch line near Derby. The results were inconclusive, but experience was gained in strain measurement and the difficulties of carrying out research on real track that was to prove valuable in the future. In the next year 100 RC sleepers were put in the mainline near to Watford and survived forjust 10 days! While this design development was taking place, work was being carried out by a separate group at Colwall in Worcestershire on prestressed concrete sleepers. This work used an experimental long line prestressing bed built by a Dowsett company and was carried out by engineers from the railway authorities, the Board of Trade, and Dowsett. These sleepers were designed by Dr Mautner of The Prestressed Concrete Company, later to become ofpart the Mouchel organisation. Fig 4 is a photograph of a painting by Dame Laura Knight exhibited at the Royal Academy in 1945 which shows the workat Colwall and gives a good impression of the excitement and pressureof those times. The next setof concrete sleeper tracktrials included both RC sleepers and prestressed sleepers from Colwall. These first prestressed sleepers were 281 Paper: Taylor B .r. c ! I - A c . t I I I side a wide range of support reactions are found. These depend on the nature of the ballast and formation below it aswell as the form of ballasting and the quality of maintenance. The point of loading also definesthe points of maximum moment and shear as coincidental. This was brought out very well inthe discussion of one of the earlyreview papers in the development of sleepers', when a Mr Henzell remarked: - 4575 I I I " 1 3 1 1 'A point to be noted about the sleepers described by the authors was that they wereentirelywithoutstirrupsortransversereinforcement.The engineers of the development work had to find out how to make a sleeper which would stand up to railway traffic and remain prestressed in service without stirrups or hooping, and that they had been able to do so was remarkable. Even the French engineer, Freyssinet,the founder of modem prestressing, rejected completely the idea of self-anchoring reinforcement for wires of one-fifth of an inch diameter or more, and in his sleeper patent before the outbreak of war he applied anchored wires, and a strong transverse hooping. The elimination of transverse reinforcement was still more remarkable when it was realised that a sleeper was not like an ordinary beam, with maximum bending and maximum shear occumng at different points. The ordinary beam has maximumbending at midspan and maximum shear at the supports, the maximum of the one occumng with the minimum of the other, but in a sleeper, which had cantilevered ends, the maximum bending and maximum shear were at almost the same point, and that point was not in the middle of the span but near the end of the sleeper; and, as there was no mechanical bonding of any kind, the wires had to be anchored in a short distance. To do so effectively with plain wires without any mechanical means was remarkable'. I "B Section Section A-A B-B Fig 3. Rail blocks put inthe west coast mainline at Cheddington near Tring to on 21 February 1943. That, and the building of the Dow Mac factory at Tallington in 194344, marks the start of the 50-yearperiod reviewed in this paper. British Rail now has 35M prestressed, pretensioned monoblock sleepers in track, and there are probably more than twice as many pretensioned monoblock sleepers elsewhere in the world - notably Russia and China, also in Canada and the USA and in small numbers in many other countries. From these beginnings more than 50M sleepers have been made in the last 50 years. Stages in sleeper design Railway sleepers are probably unique in the way in which they act as structures. First, they are not subject to self-load stresses during their working life as the staticself-weight of the rail is only in the orderof 0.1% of the total design load. They only ever see significant loads asdynamic loads. The impacts of a steel wheel on rail are absorbed to a certain extent by rail pads between the rail and the sleeper,but even these areso stiff that the demands on a sleeper are much more than a bridge deck loaded by pneumatic rolling wheels. Sleepers rest on the ballast but are not tied down to it,thus the impact load is able to make the sleeper oscillate. Sleepers are excited to their natural freqencies. This is a significant load effect and has been recognised in the past, but insufficient attention had been paid to ituntil relatively recently. The points of application of load are defined totally,but on the support Fig 4. Sleeper tests, 1943 I 282 The design process, with key steps and some significant variables, is shown in Fig 5 . The vehicle applies its load to the track through its axles and wheels. Axle loads depend on the design of the vehicle and on its maintenance. The axle loads the rail through its wheels. At this stage impact effects have to be'considered with significant variables: speed, vehicle type and the levels of sprung and unsprung mass, the track structure; straight or curving; and maintenance. From this stage to obtain sleeper rail reactions and the consequent ballast pressures involves a consideration of rail stiffness, sleeper spacing,ballast packing and maintenance and the shapeof the sleeper footprint. Design sleeper moments can be assessed from the loads and ballast pressures except that,because of dynamic effects, the unloaded resonant stresses have to be considered in developing a design moment envelope. Finally, a sleeper may be designed with given material strengths but with the remaining considerable freedom of varying the depth by profiling the top face to giveoptimum stresses throughout. With all these interactions and the need to consider effects that are so difficult to determine, quantify and maintain in the long term in practice, it is hardly surprising that much design development relies on an empirical approach taking into account what has gone before. The next section of this paper discusses some of the pioneering work on track from which the basic design rules on the loading side were derived and follows this through to thedesign process in use today. Assessment of load The two societies, theAmerican Railway Engineering Association and the American Society of Civil Engineers, set up committees in 1913 to instigate a scientific study of deformations and stresses in railway track. This committee, under the chairmanship of Professor A. N. Talbot, produced a steady stream of work from 1918 to 1940. The seven Talbot reports2 are important, not just for the data which were developed but for the way in which laboratory tests and track trials werecarried out together. The work of the Talbot Committee was verythorough. Using fixed reference locations Talbot measured the displacement of the rail beneath single wheels, pairs of wheels and complete locosgiving valuable data onspread - the distribution of wheel loads along the rail to adjacent sleepers. The deformation of sleepers (always timber) was also measured, and this work was developed to determine ballast pressure distributions on sleeper soffitsand at depth. Fig 6 shows the results of a test of a sleeper inthe laboratory on a ballast bed with pressure measurements measured by pressure capsules embedded in the ballast. The work went on to consider stressesin rails and joints and even to determinethe effect of flat spots onwheels and the consequential stresses that they impose on the rails. Wheel flat spotsare still an important factor in sleeper lifetoday, and will be discussed later inthis paper. Talbot's work was available to the UK pioneers of RC and prestressed sleepers, and some of the experimental techniques used by Talbot were also used in the Cheddington trials. A paper by Johansen3describes the work by the LM&S Railway Research Department at Cheddington from 1942. In this work reinforced concrete The Structural EngineerIVolume 71/No 16/17 August 1993 Paper: Taylor Axle loads Variables: Vehicle type Vehicle maintenance Wheel load Variables: Axle load Vehicle maintenance Vehicle speed Track structure Track maintenance Ballast pressure Variables: Rail stiff ness Sleeper spacing Ballast packing Track maintenance Sleeper plan shape L Sleeper moments Variables: Sleeper shape Track maintenance Dynamic effects Sleeper design Variables: Materials chosen Manufacturing process used Fig 5. Design process sleepers fitted with strain gauges were placed in track and monitored. The sleepers, manyof which became severely cracked during the trials, yielded useful results, and many of Johansen’s conclusions are still very relevant today. The work showed that the cracking of the sleeper on the top surface beneath therail was more important than at ballast level, a point which was finally solved only in the early 1980s. The importance of damage being done to sleepers immediately after they are laid during ballasting, before the proper packing beneath them was completed, was recognised, as was the influence of wheel flats and corrugatedrail. Both of these latter two effects are of great significance today, and workis still going on to control them. 0 cu 10 I .E 20 3 Thomas4, carried out further tests at Cheddington where he used load cells and otheringenious means of measuring chair reactionsandballast pressures, developedby the Building Research Station where he worked. Dr Thomas was, of course, a major figure in the development of our concrete Codes in that era. The work on rail reactions led to the measurement of a frequency curve for loco and tender wheels and coach wheels. Figs7 and 8 reproduced from the paper show the results. The distribution of pressure beneath the sleepers was measured with a ‘ball sandwich’ which consistedof two metal plates separatedby regularly laid out ball bearings. The sandwich was fixed to the underside of the sleeper and sat on the ballast. The width of the indentation of the balls into the outer steel leaves ofthesandwichwasmeasuredandusedwith calibration information to estimate the pressures. Thomas produced data very like those of Talbot before him, and these were used, in a slightly modified form, in BS9865. BS986 gave design chair reactions and pressure distributions beneath Table shown insleepers as Fig 1 and 9. It is interesting that prestressed sleepers were to be designed to higher loads than reinforced ones, and the explanation was given RC thatsleepers may crack and can be designed to mean working loads whereas prestressed sleepers may not, as the bond may then be lost, and therefore they should be designed to the highest loads ‘likely to be incurred’. The 22 t load for mainline sleepers has been reduced since that time although theP-P/4 bearing pressure designis still used. TheP-P/2 diagram is not used nowadays, and mainline sleepers are used everywhere in the network. Itis important to recognise, however, that BS986 gave no require- \ 30 40 50 Pressure below centreline of tie Ballast depth 12 in Fig 6. Distribution of ballast pressure The Structural EngineedVolume 71/No 16/17 283 August 1993 Paper: Taylor 8 0 642 10 12 14 16 18 20 22 Chair reactions: Ton Fig 7. Frequency curve, chair reaction dueto locoand tender wheels ment for negative moment capacity underrail. theBS986 also defineda load test in whicha load was appliedat the rail seat position spread on a 125mmwide rubber pad.Two support points, 125 mm wide and 150 mm apart, were used, and a test load (which, for a class E sleeper, was 30 t) was defined before which cracking was not allowed. A similar test is still used as an acceptance criterion on randomly selected sleeperson eachproduction line cast. 0 1 2 3 4 5 6 7 8 910111213 Chair reactions: Ton Prestressed concrete design Fig 8. Frequency curve, chair reaction due to coaches In the discussion to ref. 4, Dr Mautner gave the basis of the design of the ‘Dow Mac’ sleepers used in the trials. He covered permissible stresses in some detail and referred back tests to carried out by Freysinnet in 1936. His other reference to design theory was toa paper in The Structural Engineer 1498 + in July 19406. In the appendixto this paper Dr Mautner presented as clear -.I an exposition of prestressed theory as can be found today. In two pages he covered elastic theory, treatmentfor losses, and principal stresses for shear. The design Codes, in terms of materials data, have developed more slowly. The data given in ref. 5 were quickly followed by those in BS986 I I I I and from then, through the Institution of Structural Engineers’ first report of 1951 and C P l l Y , we are led to BS8l 10 in use today. The development ofthe criteria is shown in Table 2. In the table, because of different approaches, it is difficult to make direct comparisons but, in developing the table, data relevant to the highest strengths of concrete were used as these I most exactly suited the strengthsof concrete used in sleepers. The Chairman of the CPl15 committee was Dr Thomas, and his ViceChairman was Dr F. Walley who is Chairman of the committee that looks after BS8110 today. The continuity of our Codes and their progression is as much a human issue as a technical one! Whether the success of the preDistribution of pressure under sleepers stressed railway sleeper was central to the development of the design process, or whetherit was just one of the many usesof the technique in the For class A, B or C sleepers pc = P/2 1940s and O OS, can be answered only by a historian. Without doubt, the survival of prestressed railway sleepers in their arduous environment must For class D sleepers pc = P/3 have given great confidence toall early practitioners in prestressing. For class E sleepers pc = P/4 except for that l = Development of the manufacturing system The early methodof manufacture used for pretensioned sleepers was highly automated. The Tallington factory was designed on foundry systems with l I for sleepers to be used with tracks with ash ballast, pc = P/3 Fig 9. Distribution of pressure under sleepers(BS986, 1945) TABLE I - Chair reactions, BS986 Chair reaction R in ton for: Class of sleeper 5.5 For design at centre of sleeper 5.5 9 10 11 10 12.5 15 Intermediate sleepers Lightly worked sidings Heavily worked sidings and refuge sidings, goods, loops, and the like, over which the speed is limited to 30 mile/h Tertiary Secondary Primary 284 sleepers Type of track ~ Joint sleepers 14 For design at section under rail 22 The StructuralEngineer/Volume 71/No 16/17 August 1993 Paper: Taylor TABLE 2 - Prestressed concrete design data Reference Criteria BS986 1945 First report [StructE 1951 Working load Comp. stress&, (N/fMl2) Tensile stress&, (N/mm2) Principal lo&, (N/mm2) Transfer strength 0.33LU 2.07 I BS8110 1985 0.33.L. 0.45K 0.05f C u 1.8 0.045 fCu 0.5.L Ultimate load Global safety factor Losses Elastic strain*/ 1 N/mm2of stress @ transfer 6 " t = 35) Creep strain/ 1 N/mm2offcc Shrinkage CP115 1959 2.5 0.24K 0 . 5 ~ 1.5 x DL +2.5 x LL 1.4 x DL +1.6 x LL 36 x 10-6 30 X 10-6 30 x lo6 36 X 10-6 44 x 10-6 58 X 10-6 48 X 10-6 55 x 10-6 300 x 10-6 300 x 10-6 300 X 100 x 10% fa concrete characteristic strength (cube) * this is basically a consequence of the recommendedor allowableE values of concrete # outdoor exposure the moulds being brought to the casting station on rollers and then, after filling with concrete, they were winched on rollers along the already stressed tendons. The method is described fully in ref. 1.The process hadthe advantage of having fixed work stations with a higher productivity than would be achieved if the men moved from mould to mould to demould, clean and refill them.These systems remained inplace until the end of the 1970s when fixed gang moulds with machines that passed over them were developed. The current automated gang mould systems are very efficient and the considerable investment in the factory and machines results in individual sleepers being made with very low labour content and with much more consistent quality than under the previous method. Figs 10 and 11 contrast the early and current manufacturing systems. Development of sleeper systems From 1950 to about1985 the sleeper systemwas under a constant process of development. Through this time theA to D classes were dropped, theE range was sometimes replaced by a F range, an intermediate EF range and even a G range. Each change was inresponse to experiencein track but, by the 1980s, the exact definition of strength grades was lost.The current F40 sleeper was developed in the mid-l980s, and some of the thought which went into this will be described later. Fig 11. Production system, 1993 In parallel with the changes in strength of sleepers, great changes took place in the rest of the track system. Bullhead jointed track gave way to continuously welded flat bottom rail and chairs gave way to baseplates which themselves gave way to elastic fasteningsystems with pads beneath the rail. Manyfastening systems were tried, often with unsatisfactoryresults. The difficulty of providing a fixing to the concrete that will hold the spring fastening was nearly always underestimated. Early systems often failed by fatigue in the fasteningitself or by the fixing to the concrete pulling out or fracturing. The major reason for replacement of most early sleepers is because of a fracture of the interface between fastening and concrete. In 1960 the Pandrol clip was first introduced into the BR network, and this has proved to be, both in theUK and worldwide, an excellent fasteningsystem. The clip, illustrated in its latest form in Fig 12, is familiaralltowho travel by rail and has hold down force-displacement properties which are excellent in holding the rail to a vibrating sleeper and at thesame time restraining the pad in place. (Fig 13.) The Pandrol system is also one which, unlike previous systems, requires minimum maintenance and, as it does not rely on a screw fastening, can be quickly checked visually. Design process The simple steps of the design process have been described. This section covers them in more detail and gives an idea as to how much of current design is science, art, and experience. Sleepers always see dynamic loads. With the normal sleeper spacing the time taken from a wheel to pass over three successive sleepers at 200 km/h is about 0.025 S. On heavy haul lines, sleepers are more closely spaced and speeds are lower so that, typically, the time would be 0.095 S. Within this time the peak value would be applied for some0.01 or 0.04 S, respectively. Wheel flats, typically the size of a 50p piece, may occur if a wheel skids under braking. Over time these can develop to 75mm or longer, imposing an additional hammer blow as theyhit therail. Wheel flats are common, as any regular train traveller can easily recognise. Out of round wheels and frozen suspensions, the latter being not uncommon in trucks with certain suspension types sometimes carrying cement or coal, can also cause loads F irg 12. Pandrol railfastening Fig IO. Production system, 1943 The Structural EngineerIVolume 71/No 16/17August 1993 285 Paper: Taylor 1000 800 600 400 Vibration mode found in practice 200 0 6 04 2 8 161014 12 Clip toe deflection (mm) 1 Fig 13. Pandrol rail fasteningpeqormance to be impulsive. Finally, as shown by Thomas4, the wheel passing over a joint or aweld in welded track causesa further impact effect. Fig 14 gives a summary of 12 different formulae for deriving the impact factor for the wheel onto the rail. The factor is speed related but is also greatly influenced by maintenance of the whole railway. Clearly, we must be led by experience in deciding on this factor! The distribution of load between sleepers or spread has been defined mathematically by Zimmermdn9. Zimmerman’sformula is: ( R S L = W 1-- W Bending moment from sideforces Fig 15. Moments.from dynamic effects andside loads 3r2:2) where W is the enhanced wheel load(for impact) i = 6 EI/L3C E is the rail modulus I is the rail inertia L is the sleeper spacing C is the support spring rating 2.0 - 1.5 - 10 150 50 100 Speed km/h 1 Steam loco, Peterson formula 2 Clark formula highend of range 3 Diesel loco, Peterson formula 4 Indian formulafor light track 1 5 AAR formula for diesel locos 1 6 Indian formulafor light track 2 Fig 14. Impact factor formulae 286 7 German formula 8 Schram 9 Clark formula 10 AAR formula for steam locos 2 11 Ore 071 formula 1 12 Ore 071 formula 2 Once again the value of C to be used is a matter of conjecture and, before a distribution factor can be assumed, some allowance must be made for combinations of loads from adjacent wheels on the vehicles. On balance the combination of the increasing impact factor and the reducing spread factor comes to somewhere between 1 .S and 2.5, depending on the typeof railway, axle loads, maintenance,etc. The temptation to derive thesecombination effects backwardsfromtheperformance of sleepers of known strength with known vehicle types and speedsis almost irresistible. The rail seat reactions derived in this way are applied to the pressure diagram in Fig 9. The centre pressure may be zero immediately after the sleeper has been tamped beneath the rail, it maybe P/4 during service running and tend towards P/2 as the track becomes in need of retamping. The effect of this will be to reduce the positive (sagging) moment beneath the rail and increase the negative, hogging moment at the centre. The effects of curved track and of vibration in one mode both lead to tensile stresses at the top of the sleeper under the rail (Figl S ) , particularly towards the inside fastening location. One of the major causes of vibration damage to sleepers is rail corrugation. In the late 1970s an increase of incidence of corrugated rail caused some looseningof the iron shoulders that hold the fastening into the sleeper and caused top crackingof sleepers. The trend to raising the position of the prestress tendons came from this time, as did a redesign of the embedded stem of the shoulder. Rail corrugation consistsof the development and propagation aofregular wave on therail top. Passengers can easily recognisea corrugated section, as it ‘sings’ when a train passes overit. Corrugation is still not fully understood; it is a feature of all road systems supporting wheels and is even found in highways, but with a very different pitch and amplitude. Rail corrugation appears to depend on rail metallurgy, wheel and vehicle type brake system (rim or disk) and line speed. Researchin this area continues. Rail grinding with special machines which reprofile the rail headused is as a maintenance measure, but even then the corrugation can recur over a period of months or years. The current BR mainline standard sleeper shows the effect of these demands in its bending moment diagram (Fig16).This sleeper is somewhat shorter than its predecessors for operational reasons, to enable it tobe laid with machinery which itself is within a safe working gauge for relaying to The Structural EngineerNolume 71/No 16/17 August 1993 Paper: Taylor ' proceed on a weekday with the adjacent line being usedby normal traffic. The importanceof a dialogue between the railway users of sleepers in many areas, the installer, the designer and the manufacturer cannot be overemphasised if a satisfactory design conclusionis to result. The under rail sagging capacity in the F40 is less than that of previous sleepers, as it is shorter and picks up less load from the cantilever and because it has thicker and much more resilient 10 mm rail pads than the previous standard5 mm to6 mm pads. It was also decided, in modern wellmaintained high speed lines, that it was important to raise the prestressing tendons in the section to improve the reverse hogging moment capacity, as this negative moment demand was speed related. The F40 sleeper was developed in the early 1980s in a programme of test and trial application which givesa good ideaof the levelof loading actually experienced by sleepers. In thelate 1970s problems were experienced with the then standard F27 sleeper in the most arduous parts of the London to Glasgow West Coast Main Line (WCML). The problems were on isolated sections and showed up as cracks in the sleepersat the top, at each side of the cast in rail fastening housings. This resulted in a new design of the housing with better pullout resistance and torsional grip and also stimulated a new look at the sleeper design. After a study of various design options and measurement of sleeper movements under traffic, it was decided to raise the prestressing strands in the section to givea larger prestress under therail seat and effectively hold any cracks together. 1 C Rail 1210+ ~ 1210 I I I 15 E 0 5 .-K v) 15 5a, > t 30 1210 0 605 605 1210 Dimensions in (mm) Fig 16. Moment strength envelope, F40sleeper I Point (6) I I I I 80 40 0 -40 l 0 1 I I 0.01 0.02 0.03 I 0.04 1 L 0.05 0 I 0.01 Time in S - 0 start arbitrary Time in I I 0.02 S 0.03 I 0.04 1 0.05 - 0 start arbitrary Cycles recorded from a loco type 87 travelling at speedsof 100 mile/h measured at two points -40 h 0 40 Y v t a 80 0 0.01 0.02 0.03 0.04 0.05 0 Time inS - 0 start arbitrary 0.01 0.02 0.03 0.05 0.04 Time in S - 0 start arbitrary Cycles recorded from a carriage at the centre of the same train travelling at100 mile/h measured at two points Fig 17. Measured strain readings, West Coast Main Line,1984 The Structural EngineedVolume 71/No 16/17 August 1993 207 Paper: Taylor The criticality of this element of the sleeper designis interesting. Fig 17 TABLE 3 - Hypothetical distribution of axle loads, West CoastMain Line shows the result of tests on instrumented sleepers in the WCML where strains were measured using resistance strain gauges and a high speed % of total Axle load (t) datalogger. The class 87loco, which already had achieved some notoriety 5 12.5 in causing more rapid track deterioration than previous locos, imposed 80 17.5 strainchangesof 100 x compressionand 10 x tensioninthesleeper 10 under the rail. The passenger coaches of the same train give 50 x 22.5 compressionand 50 x tension. 4 27.5 The pattern of oscillation shows 6 to 7 cycles of near identical amplitude 1 32.5 followed by a very rapid decline. This was put down to the fact that the sleeper comes away from the ballast and vibrates freely momentarily after a wheel passes and thensits back onto the ballast whereit is well damped. TABLE 4 - B R system track miles(simplified table) The frequency is approximately 500 Hz and is different from that caused by rail corrugations which, typically, is 900 Hz at 160 km/h. 41 2 Tonnage x lo6 Category An estimate of the fatigue life of the concrete gives interesting results. <2 2<t<5 5<1:2 12<t speed k m h Assuming that the WCML carries approximately10 M gross t p.a. and that the axle loading falls between 12.5 t and 32.5 t, in the pattern of traffic 1248 A 15 957 > l 60 0 proposed inTable 3, it is possible to come to some simple conclusions. The loco gives the compression worst case and the coach the tension worst case. 2949 B 120< <l60 578 2249 563 Looking at thetension case, 430000 x 7 cycles are experiencedp.a., 3.2 x 106 cycles. Assuming that this causes the majority of the fatigue damage, 80< < l 20 218 1213 C 3599 3529 it ispossible to predict the fatigue life of a sleeper under therail, using data produced more recentlyby Cornelissen & Reinhardt’O (Fig 18). This, for the D 4 0 329 957 21 2694 F27, suggests a life of 6.2 x lo6 cycles or approximately 2 years. This life beforecrackingwasconsistent with experienceatthemostarduous locations on the WCML, although it should be realised that a cracked technical problem, as it involves nearly all aspects of the running of the ra network. One interesting development in the last few years demonstrates this sleeper survives for some years beyond the time of cracking before it needs quite clearly. In 1986 an ‘intelligent rail’ was put into the west coast mainline to be replaced. at Braidwood, southof Glasgow. The installation consistedanofinstrumented The revised F40 has sufficient prestress not to go into reverse stress under this level of cyclic loading and has therefore given excellent durabilitylength of rail with a datalogger, minicomputer and telemetry module at the track side. The systemwas capable of measuring and recording all axle loads in track to date. as they passed by. The system was set up with a trigger level of a 30 t wheel The compression side also has a reassuringly high fatigue life. The load which rang an alarm in the Glasgow Control Office. Glasgow was able average prestress levels, with the cyclic loading imposed on them, still give to question the computer system, to identify the guilty train, vehicle and axle. compressive fatigue lives of the order of 1020cycles. Statistical data were collected weekly by the British Rail Centre at Derby As we have seen from 50 years of satisfactory use, compressive fatigue of concrete in prestressed sleepers is not a problem and is never likely to using a modem link to download the data. Initially, up to 10 such excursions a week were detected and, after directed maintenance, this level is now less th become one. one/week. The vehicles at fault were a combination of some locos, many freight vehicles (some of which were privately owned) with a set of faults 50 years on which would not have been always readily visible during normal maintenanc The replacementof bullhead chairedrail on timber sleepersby continuously Such was the success of the Braidwood experiment that now BR has three welded flat bottom rail on concrete sleepers is practically complete on the such monitoring stations in operation and is installing a fourth. mainlines of the BR network. Table 4 gives an idea of the extent of the The final most interesting area to be considered is how long the sleepers current network, with the track categorised in terms of speed, A-D, and gross last and howdo they fail? The second partof this question can be answered. annual tonnes carried. The only area of the network where there is still The vast majorityof sleepers that are replacedfail because of the difficulty timber sleepered track in any significant amount is in categories C1 and D 1, of controlling the arduous environmentinwhichtheyoperate.Track with lesser amounts in C2 and D2. These are rural lines where it is harder maintenance, rail corrugation, joints, wheel and vehicle defects, and even to commercially justify the levels of investment needed to relay. derailment, all cause damage leading in some cases to replacement. A very Further developments of complex ‘sleepers’ or ‘bearers’ for switch and small number of sleepers of less than adequate strength find their way into crossing work has taken place, after a thorough research programme. Currently, prestressed concrete bearersup to 6 m long are used for switches track, as could be expected from any product accepted for use following a and crossings where they offer greater stability than timber and overcome statistical procedure based on random test. The experience of BR is that the problem of sourcing the high quality hardwood previously used in theseprestressed concrete sleepers last for 30 years in the mainline, followed by replacement and cascading down to less heavily trafficked lines where a locations. further 10-20 years’ life may be possible. This paper demonstrates that the design of track is not just an ongoing 1 1 1 1 Conclusion 1.o 0.8 E 0.6 5 c \ X 2 b \ \ 0.4 compression - tens> 0.2 l 6 Hz 0.0 0 1 2 3 4 5 6 Log N Fig 18. Fatigue of plain concrete with stress reversals 288 I bry specimens 7 8 9 The prestressed, pretensioned railway sleeper has been a success-story both for prestressed concrete and for British Rail. The sleeper has been central to the development of the BR network from a railway in needof modernisation in 1945 to the network that we use now. BR’s achievement is that its modernisation went from steam on bullhead jointed track to high-speed mixed trafficked routes with continuously welded rail. The robustness of prestressed sleepers allows passenger trainsof 250 k m h to share the same track with slower goodstraffic. In Japan and France the high speed network has been developed on new dedicated lines, with the advantage that the track design has only one loading type. Britain’s challenge tofit high-speed railways into an existing network is much harder, and our success is insufficiently emphasised. Even now, train speeds are limited more by the need to ensure passenger comfort on the original curving alignments thanby inadequacies of hardware. The challenge for the future is in financing the secondary lines and in maintaining the high standards of track in the primary network. In both of these areas, prestressed concrete will surelystill have a decisive role. Continued ot1 page 295 The Structural EngineedVolume 71/No 16/17 August 1 9 9 3 Paper: Bloomer Paper: Taylor Paper: Taylor continuedfrom page288 Acknowledgement The author is pleased to acknowledge the assistanceof his colleague John Surtees for his gift of historical and technical knowledge essential to the paper and also his now retired colleague John Snasdell who taught him much about sleepers. The helpful coments of Dr D. Cope, of British Rail, are also gratefully acknowledged. References Barker, R. S. V., Lester, D. R.: ‘The development and manufacture of prestressed concrete units’, Society of Engineers, May 1946, pp41-75 discussion contribution, Henzell, J. S. 2. Stresses in railroad tracks - the Talbot Reports 1918-1940, reprinted by American Railway Engineering Association, Washington, DC, 1980 3. Johansen, F. C.: ‘Experiments on reinforced concrete sleepers’, Proc. ICE, Railway Division, May 1944, pp3-20 4. Thomas, F. G.: ‘Experiments on concrete sleepers’, Proc. ICE, Railway Division, 1944, pp21-66 5 . BS986 Concrete railway sleepers, London, British Standards Institution, first pub. 1941,2nd rev. 1945 6. Everitte, T. J.: ‘Recent developments of prestressed concrete construction with resulting economy in the use of steel - Technical Appendix by K. W. Mautner’, The Structural Engineer, July 1940, pp626-642 7. Firstreport on prestressedconcrete, London, IStructE, September 1951, pp31 8. CP115 The structuraluse of prestressed concreteinbuildings, London, British Standards Institution,1959 9. Zimmerman, H.: Die Merechnung des Eisenbabnoberbaues (original pub. c. 1890) W. Ernst & Sons, 1941,3rd ed. 10. Cornelissen and Reinhardt: ‘Fatigue of plain concrete with stress reversals’, CEB Bulletin No. 188, Lausanne, CEB, 1984 1. Fig 8. End of project Acknowledgements The author expresses hisappreciation to allhis colleagues who worked on the stack project. He is particularly grateful for the support of George Maddison (M) (Allott & Lomax). References 1. BS CP3: Chapter V: Part 2: Wind loads ,London, British Standards Institution, 1972 2. Bolton, A: ‘Design against wind-excited vibration,’ The Structural Engineer, 61A, No. 8, August 1983 1: 3. ESDU 82026 Strong winds in the atmospheric boundary layer: Part mean hourly wind speeds, London, Engineering Sciences Data Unit, August 1984 4. BS 4076 Specification for steel chimneys, London, British Standards Institution, 1978 5 . ESDU 85038 Circular-cylindrical structures: dynamic response to vortexshedding:Part I : Calculation procedures and derivation, London, Engineering Sciences Data Unit, May 1986 6. ESDU 86035 Calculation methodsfor along-wind loading: Part I : Response of line-like structures to atmospheric turbulence, London, Engineering Sciences Data Unit, December 1987 7. Hirsch, G: ‘Control of wind-induced vibrations of civil engineering structures’, Aachen, 4th Colloquium on Industrial Aerodynamics, June 1980 8. Roshko, A: ‘Experiments on the flow past a circular cylinder atvery high Reynolds number’, Journal of Fluid Mechanics, May 1961 9. CICIND Model Code for steel chimneys, International Committee on Industrial Chimneys, - . May 1988 10. Design and constructionof steel chimney liners, American Society of Civil Engineers, 1975 - The Structural Engineer/Volume 71/No 16/17 August 1993 295