Impact Effects on Pipelines Beneath Railroads

Transcription

1 Impact Effects on Pipelines Beneath Railroads HARRY E. STEWART AND MICHAEL T. BEHN Design method being developed for ncased cros ings of hihpressre gas pipelines e impact ractors to accont for the m crease jn live load respon, e de to rhe effects f velncle speed, track stiffness, vehicle sspension characteristics, or irreglarities in the mooing srface. Field experiments to mea re impact effects were condcted on an instrmented -pipeline 36 in. (914 mm) in diameter bried 5.75 ft (1.75 m) below the Facility for Accelerated ervice Testing track at the Transportat1n Test Center in Peblo, Colorado. Ranges of vehicle speeds and srface geomerry conditions were inve tigared, and impact factors bas.ed on measred pipeline strains were delermined. The reslts rndicated that train speed of 5 to 4 mph ( kmhr) had a relatively minor inflence on impact response,.whereas chan.ges in srlace geometry reslted in a range of dynamic p1pelme strams, with the maximm vales nearly 1.6 times larger than previosly recorded nder baseline operating conditions. When high-pressre gas pipelines cross beneath railroads, the owner of the railroad generally reqires that the carrier pipeline be installed within a metallic casing. The main design criterion for the cased carrier is that the circmferential (hoop) stress de to internal pressrization be less than some percentage of the specified minimm yield strength. The allowable percentage is based on the poplation density in the vicinity of the pipeline, the type of pipeline welds, and the operating temperatre. Becase the casing is designed to carry the earth and live loads, the carrier design for cased pipelines is naffected by additional live load effects de to impacts at the srface. Research focsed on the development of design procedres for ncased gas pipelines is nder way. Uncased pipelines mst be designed to withstand live load stresses imposed by vehiclar traffic as well as stresses de to internal pressre and earth load. Rational methods to accont for impact forces are an important part of design procedres. Two instrmented high-pressre steel pipelines were installed withot casings, sing ager boring methods, at the Transportation Test Center (ITC) in Peblo, Colorado. Field experiments were condcted to measre pipeline response to train loading. The effects of vehicle speed, internal pressre, and time since pipeline installation were investigated dring a 2-year period. Figre 1 shows profiles of the two pipelines. The pipeline 12 in. (35 mm) in diameter has a wall thickness of.25 in. (6.4 mm) and a specified minimm yield strength of 42, psi (29 MPa). The pipeline 36 in. (914 mm) in diameter has a wall thickness of.61 in. (15.5 mm) and a specified minimm yield strength of 6, psi (414 MPa). School of Civil and Environmental Engineering, Cornell University, Ithaca, N.Y The depth from the top of the railroad crossties to the crown of both pipes at the track centerline is 5.75 ft (1.75 m). Both pipelines were instrmented before field installation. Instrmentation consisted of strain gages, both internal and external, on the pipes, accelerometers, pressre transdcers, and temperatre sensors. Strain gages also were monted on the rails directly above the pipes to measre the applied wheel loads. The strain gages on the pipes were oriented to measre both circmferential and longitdinal strains at the inside and otside crown, springlines, and invert. The locations of the instrment stations are shown in Figre 1 as solid circles. The gage locations correspond to locations on the pipelines directly beneath the otside rail, track centerline, inside rail, and other locations along the pipe's long axis sfficient to measre the distribtion of strains along the pipeline. Testing of the pipelines began in Jly Measrements were made at 4- to 6-month intervals throgh the spring of 199. Althogh measrements of live load response were recorded for both pipes, special impact testing was condcted only with the 36-in. (914-mm) pipeline. The remaining discssion focses on the 36-in. (914-mm) pipeline data. BASELINE TESTING Field data were measred for a range of train speeds and internal pressres from the smmer of 1988 throgh the spring of 199. After the installation of the 36-in. (914-mm) pipeline, the annls left by the 1.5-in. (38-mm) ager overbore remained partially open and did not collapse flly. The reslting pipeline strains were small, becase contact between the pipe and the soil was limited. To replicate long-term loading conditions, the remaining annls arond the pipe was injected with a slrry of native sand and water in May Field data indicated that the annls had collapsed partially between Jly 1988 and May 1989, and strains had been increasing. The decision to fill the annls and increase live load transfer was necessary, becase long-term response was desired and the field testing program had a dration of 2 years. There is little dobt that, given several years, the annls wold have collapsed flly becase of repeated traffic. Between May and Jne 1989, the field measrements increased and stabilized at a consistent level. Measrements in Jly 1989 confirmed that the annls arond the pipeline was in a steady-state condition. Figre 2 shows the longitdinal pipeline strains at the crown and invert of the 36-in. pipe measred in May 1989 before the annls was filled, in May 1989 jst after the annls was

2 2 lnslrmantollon mal\llold 1V=====---- 1H In1trmented section Weld neck and blind flange o) 36 in. Diomeler Test Pipeline 111 Insl,menJed seclion Weld neck and blind flange 15.fl 15,5 fl.1 25 fl 14.5 fl bl 12 in. Diameter Test Pipeline FIGURE 1 Profile views of test pipelines (looking west). (a) (b) c ] IJl e - 15:!-Lo...,,.5....,,oo...,so2 Distance (in.) to as 125-ton cars and are representative of the heavy loadings anticipated in the near ftre on U.S. revene lines. As shown in Figre 2, the strains before the annls was filled were sbstantially smaller than those after the annls was filled in May Strain decreased from May 1989 to Jne 1989, as any locked-in injection pressres dissipated. The Jne 1989 and Jly 1989 data indicate that the contact conditions between the pipe and soil had stabilized and were taken to represent the long-term condition. The relative changes in pipe strain from May 1989 to Jly 1989 shown in Figre 2 are representative of the changes of circmferential strain at the pipe crown, invert, and springline over time. Train speeds above the instrmented pipeline were varied from a slow roll of roghly 5 mph (8 kmhr) to 4 mph (64 kmhr). The pper limit was based on the maximm speed that the train cold achieve throgh the test section. Figre 3 shows dynamic longitdinal strains at the crown and invert of the 36-in. (914-mm) pipe, at the gage station directly below the centerline of the track. The data indicate that at the pipeline depth of ft (1.75 m), there was no measrable effect of train speed for the baseline field condition, withot srface irreglarities. Ths, for the normal track conditions at the Facility for Accelerated Service Testing (FAST) track at TIC, impacts were not measred. Figre 4 shows the dynamic wheel loads measred sing the strain gage instrmentation installed on the rails directly 1 (kmhr) Distance (in.) FIGURE 2 Changes in longitdinal strains at crown and invert over time. injected with the native sand and water slrry, in Jne 1989, and in Jly (Distance corresponds to the track centerline.) The rail srface at this time was level, withot irreglarities. Trnin loading was generated by slowly rolling loaded freight cars weighing 315, lb (14 kn), prodcing 39.4-kip (175-kN) wheel loads. The freight cars are referred c 5 e i Lonq "...J - 5.._ > ::i Inver I rc:own -1 O<\I Train Speed (mph) FIGURE 3 Longitdinal strains al crown and invert verss speed, Jly 1989.

3 Stewarl and Behn 3 -;;; 5 a. :.;;: 4 "'O 3 _J 1111 I li I a :: 1 ;; z -"' then to increase the predicted stresses by a factor that is greater than nity at the srface and that decrea. es with depth. This method acconts for the attenation of dynamic stre es with increasing depth. The two most common formlations for this variable depth impact factor are those recommended by the American Society of Civil Engineers (ASCE) (7) and the American Petrolem Institte (API) (8). The impact factor recomrr. 1ded in the ASCE method eqals 1.5 between and 5 ft ( and 1.5 m), decrease. linearly to LO at a deptb of 22 ft (6. 7 m), and remains con tant below that depth. T he API impact factor is 1.75 between and 5 ft ( and 1.5 m) and decrea es linearly by.3ft (.1 m) between 5 and 3 ft (1.5 and 9.1 m). Below 3 ft (9.1 m), the API method ses an impact factor of 1.. IMPACT TESTING 2.5 5, Time (sec.) FIGURE 4 Dynamic wheel loads and longitdinal strains at invert for train speed of 3 mph. above the pipeline for a trnin peed of3 mph l48 kmhr) and the corre ponding longitdin1tl strains at the pipe invert beneath the track centerline. The train ed for this data rn consisted of one locomotive and five freight cars. There is some variation in the dynamic wheel loads from the freight cars. The dynamic loads are approximately 4 ± 4 kip (178 ± l kn). The nominal static wheel load f r tbe freight car was 39.4 kips {175 kn). T hi indicate that at 3 mph (48 km hr) the srface impact factor i. ±.1. Figre 4 al o how that for ax! reslt in a. inglc tr s plse at the pipeline depth. IMPACT FACTORS Design methods for pipelines sbjected to traffic load generally se some factor to accont for the increa e in live loading eff ct de to vehicle dynamics and the qality of tbe rnning rface. For railroad loadings on bried pipelines, two approaches are often sed. The fast is to se an impact factor a a mltiplier of the static wheel load and calc late pipe response on the basis of the increa ed rface loading. Thi approach is also s d for conventional track design and several methods are available for stimating the srface impact factor. Typical method arc ba ed n a combination of vehicle peed, wheel diameter, track tiffness, track qality and nsprng mass of the wheel et. (1--). Ln general these meth els predicl rface impact factors on the order.f 1. to 1.6 for track in good condition at vehicle peed from 5 to 4 mph (8 to 64 kmhr). Tmpact factor bas don the e methods increase to approximately 2. to 2.5 at high train peed t r track in poor condition. The s con I approach for impact loadings, which is more common for pipeline de. ign is to predict stresses within the soil mass that are based on a nominal design wheel load and The observation that negligible speed-indced impacts occrred throgh the test section is consistent with wheel load data reported previo ly for FAST (9), in that only a small percentage of wheel loads at the well-maintained FAST track were significantly larger than the nominal static vales. Becase the primary prpose of the field experiments was to provide data to sbstantiate a pipeline de ign procedre, it wa. important to replicate typical field conditions and to generate realistic pper bond load ing conditions. Proj ct advisors from the gas and rail1 oad ind tries, the Association of American Railroads, and the American Railway Engineering Association also were concerned that the loading conditions at FAST might not represent those of revene lines, becase the track maintenance standards are high, and irreglar train wheels are removed when they are detected. Ths, a serie of impact tests intended lo ca e increased dynamic loadings representative of les er-qality track wa initiated. In addition, impact loading measrements cold be sed ro sbstantiate crrent impact formlations sed commonly in pipeline design. Impact testmg consisted of progressively degradmg the track qality above the pipeline and operati1jg the train at a range of peeds. The degradation procedre inclded in tailing a rail joint directly above the pipeline. The installation of the joint reqired removal of the \\heel l ad detection circit.. Wood hims were placed between the top of the ties and tie plates at b th rail over a distance of roghly f1 (24.4 m) o that a niform rail rai e of 3 in. (76 mm) was achieved over the central 3 ft (9.1 m). The wood shims over the central portion of the elevated track were removed in stages beneath the inside rail to prodce a dip in one rail and a cro -level variance of p to 3 in. (76.2 mm) between the inside and otside rail. The joint at the rail above the pipeline also cold be adj ted to prodce either a tight joint or a plled joint. The gap ca ed by the plled joint was approximately. in. (2.3 mm). In addition, the end of the pstream rail at the joint was progre sively grond to simlate a battered joint. The mi match ranged from to approximately.3 in. (7.6 mm) and was iacrea ed with increa ing ross-level variances. The test condition were selected to correlate wiu1 track class designation specified by the Federal Railroad Administration (FRA) (JO) so that the track irreglarities cold be related to revene track conditions at other sites.

4 4 Eight test steps were investigated, for FRA Class 6 + standards down to FRA Class 1 standards. For each test step, train speeds were varied from a slow roll to the maximm permissible or safe train speed, with both a tight and a plled rail joint. Table l smmarizes the impact test conditions along with the associated FRA class limits for cross-level and rail mismatch. The maximm joint gaps are also given in Table 1. The joints for the tight joint conditions were made as close as possible, not exceeding V16 in. (1.6 mm). Figre 5 shows the measred cross-level variances between the otside and inside rail variances verss tie nmber for the test steps given in Table 1. Shims were removed from the inside rail, which cased the dip in the rail profile shown. The test pipeline was located beneath Tie 53, corresponding to the center of the rail dip and maximm cross-level variance. Test Step 1 represents the nominally smooth track that had been shimmed to provide a niform 3-in. (76mm) raise thr gh the test ection. A slow roll of the train throgh the te t se ti on indicated that the installation of the shims and rail joint did not case a change in the strains measred in the 36-i. (914-mm) pipeline from those recorded dring the baseline measrements. Ths, the slow roll at Test Step 1 was representative of the baseline test conditions. Impact testing proceeded for each test step by increasing the train speeds from 5 mph (8 kmhr) p to the maximm attainable speed given in Table 1 with the rail joint tight, and then repeating the speed seqence with a plled joint. After the completion of a test step, the shims were removed as necessary, and the rail was grond to the rail mismatches given in Table 1. Figre 6 shows the variation in dynamic train in the pipe beneath the track from Test Step Sb for several important pipeline locations. Figre 6 indicates that train increa es only slightly as peed increa e from 5 to 4 mph (8 to 64 kmhr). There is a light sbpeak in the dynami.c strains near 2 mph {32 kmhr) which correspond to a resonant effect that freqently ha been observed in other te ti ng at TT sing 39-ft (11.8-rn) jointed rail ections and train traveling at 18 mph (29 kmhr). Also the longitdi nal straill are not symmelril:al abot the nrestrained pipe's netral axis. This trend is also shown in Figre 2. The circmferential strains at the springline have a greater absolte magnitde than at invert. This trend was observed consistently in all of the experimental data. J c -=- 2 - o; > S1ep_ I_ ' ' ' ' ' ' ' 1 e FIGURE 5 Tie Nmber Cross-level variances for impact tests. (kmhr) ITI _ Hoop Sring\ine c--, ' ' ;: Long. Invert -5 {.Hoop Invert _L rown - 1 c,o, ,...-3,,--4---_,,.so Train Speed (mph) FIGURE 6 Variation of live load strain with train speed, Impact Test Step Sb E' _s Impact factors for the field tests were defined as the ratio of pipeline strain nder impact conditions to the strain measred at the same gage location from the baseline tests. Table 2 gives the measred impact factors at three critical pipeline locations. determined for the worst srface geometry case and maximm attainable train speed. The impact factors TABLE 1 SUMMARY OF IMPACT TEST CONDITIONS Rail Mismatch Maximm Test FRA Cross Lnvel (in.) ( ln. ) Joint Step Class Speeds (mph) Test f'ra Max. Tnsc FM. Max. Gap (in.) 6+ o - 4oa o a.so o 4oa.42 o.so oa 1. 1., S Sa 4 o - 4oa Sb - 4oa S S o 2sh S. 2S. 8-11> D. 2S.7S a - Maximm attainable at test section b - Maximm allowable for FRA Class

5 Stewart and Behn 5 TABLE 2 IMPACT FACTORS FOR PULLED JOINT TEST CONDITIONS Train Cross Rail Joint Test Speed Level Mismatch Gap leect Factor at Station 8 Step (mph) (in. ) (in.) (in.) Hoop, lnvcrc Hoop, Sp<ingllne Longitdinal, Invert (2) 1.13 (3) 1.12 (2) (1) (3) 1.1 (1) (2) 1.1 (3) 1.12 (1) (2) 1.19 (3) 1. OS (2) Sa (2) 1.17 (3) 1.12 (1) Sb 4 1.S8.19.7S 1. S2 (2) (3) ( 1) 2S (3) 1.48 (3) 1.17 (1) S (1) (3) 1. 7 ( 1) a - Nmbers in parentheses refer to pipeline station: 1 - otside rail; 2 - centerline; 3 - inside rail from Test Steps 1 throgh Sa did not show a clear trend of increasing with worsenea rrack condition. Test Steps Sb throgh 7 had increased cross-level variance and rail mismatch, bt the maximm allowable test train speeds decreased from 4 mph (64 kmhr) to 1 mph (16 kmhr). The data given in Table 2 sggest that larger impact factors wold have been achieved for Test Steps 6 and 7 if the train speeds had been higher. In general, Test Steps Sb and 6 reslted in the highest measred impact factors. Figre 7 shows comparisons of the pipeline strain from the impact tests with the strain from the initial condition or baseline cases at the same gage location. Figre 7a shows data from Test Step Sb, and Figre 7b shows data from Test Step 6. The strains at the inside invert, otside crown, and otside springline are shown, sing data taken from all instrmented sections along the pipe, as shown in Figre 1. As indicated in Figres 7a and 7b, impact factors can be determined by the ratios of the impact strains to the initial condition strains. There is a distribtion of impact factors from roghly.8 to 1.6 for both test steps. Impact factors of less than nity are possible de to wheel bonce, load transfer between inside and otside rails, and dynamic interaction effects of the trains passing throgh the irreglar track section. As described previosly, both ASCE and API recommend design impact factors dependent on depth below the track. Figre 8 shows the design impact crves for ASCE and API, along with the maximm impact factor determined from the field testing. Althogh only one experimental pipeline depth was investigated, the datm shown in Figre 8 sggests that the ASCE recommendation may be nconservative. The field testing had limitations on the maximm train speeds, particlarly for the most severe geometric irreglarities. Ths, it is likely that greater impacts are possible with revene train speeds of p to 8 mph (129 kmhr). The API design crve has an impact of l.7s for the pper S ft (1.S m), which is larger than the maximm field test vale of 1.6. Given that higher impacts than measred dring the field tests may be possible, the API crve wold be preferred for ncased pipeline design. "' 1 :i.. c: :g 8 a: 8. 6 E - 4 c= :::!: 2 (a) "' 1 :i.. c: :g 8 a: 8. 6 E - 4 ; c= :::!: 2 (b) FIGURE 7 Cross level'l.6in. (41 mm) fo l)<j 'V CJ Roilmosmalch'2in(Smml ' )',,,.... Join! gap'o.sin, (2mml <c<>t}o Train speeds' 5-4 mph,, 'O <> CJ,, (8-64kmhry,, I I...,. I I o I {;; 8 411:( J;;t>,gl;V # r I 1 sv, lo.id B " -,,B ' c'.own,)' ln,ert,,. o Longildinal al Inver! I..,, a Circmferential at Springline!:"" 6 Circmferential at Invert Magnitde of Initial Condition Reading, µ.e Cross level,2in (Simm),ro,)', Roil m15malch'o 2Sin (6mml,i' Jami gapo Sin (2mml r,.<>v Trom speeds' 5-25 mph I I I a (8-4kmihrl () '\ I I ;o I B I I I, {!, cfo n, l;,;ll.} - 'lf Ll Sprongline,..ti. 4%; II, O Long1ld1nal at Inver! Inwt D Circmferential at Springline 6 Circmferential at Invert o "'---'----'-'--"-'---'---'-'--"--..,,_ Magnitde of Initial Condition Reading, µ.e Measred impact factors from field tests.

6 6 Design Impact Factor b 1. 5 g 1 F Maximm Measred Impocl 15 Cl 2!O eration that high r impact might hav been developed had higher test peed been pos ible the impact formlation given by APT is recommended for the design f nca ed ga pipelin e crossing beneath railroads. ACKNOWLEDGMENTS This research was spported by the Gas Research Institte (GRI). Kenneth B. Brnham is the GRI project manager. The participation of gas and railroad indstry advisors and TIC personnel is appreciated. REFERENCES SUMMARY FIGURE 8 ASCE and API design impact factors and maximm measred impact factor from field tests. Field testing of live load response on well-maintained track at TIC indicated that negligible dynamic impact effects occrred dring bas fo1e field testing. ln respon e to coi1cems from the gas and railroad indstries, and to replicate pper bond condition t lhe extent practically possible, a cries of pecial impact te t wa condcted to investigate live load response for changes in track qality consistent with FRA class standards. Track qality was degraded progressively by increasing the cross-level variance between the inside and otside rai ls prodcing a condition representative of a dipped joint. Rai l mi match on the order of.25 in. (6.4 mm) wa incl.ded along with a plled joint prodcing a gap of approximately.8 in. (2.3 mm). Heavy-axle fre ight cars were operated over an in trmented teel pipeline 36 in. (914 mm) in dfrimeter b ricd 5.75 ft ( ) below the top of the tie. lmpact factors increased slightly with peed for each of the test configrations. Impact factors based on pipeline strain were measred and ranged from.8 to 1.6. On the basis of the maximm measred impact factor of 1.6, and the consid- 1. Ma1111afor Railway E11gi11eeri11g. American Railway Engineering As ocii1tion. Wa hington, D.., A. N. Talbot. tresse in Rai lroad Track, Repon of the pccial ommittee on Strcs.es in Railroad Track. l'roceedings, Amcri can Railway Engineering sociation. First Progress Report. Vol. 19, 191, pp econd Progress Report. V I , pp C. W. lark. T rack Loading Fndamentals-Parts 1-7. Railway Gazelle, 16, London, M. Srinivasan. Modern Permanent Way. Somaiga Pblications, Bombay, India, J. Eisenmann. Germany Gains a Better Understanding of Track Strctre. Railway G(lzee lwenional, Ag. 1972, pp C.. f'rederick and D. J. Rond. Vertical Track Loading. Pro ceedings, Track Technology for the Next Decade. Thomas TelCord Ltd., L ndon, Jly pp ; discssion pp. LSl Commiee a n Pipdine Crossings for Railroads and Highway.. Interim Specijicatio11s for the De ig11 of Pipeline ro Sings of Railroacls and Riglrways. A E New Y rk, Jan. l Recommended Practice for Liqid Petrolem Pipelines r s ing R ailroads and Highways. AP Recommended Practice I 12 5th ed. American Petrolem Jn titte, Wa hingcon, D.., ov. 19 L 9. H. E. Stewart :ind T. D. O ' Rorke. Load Fncror Method fo r D ynamic Track Loading. Jomal of Tra11 portation 11gi11eeri11g. AS E. Vol. 11 4, No.!,.Ian. 1988, pp. 2l Track Safe1y t1111dard. Office or afery. Federal Railroad Administration, U.,. Department of Transportation Pblication of this paper sponsored by Committee on Clverts and Hydralic Strctres.