Mevcut çelik demiryolu köprülerinde yorulma ömrü tayini

Abstract

Ulusal demiryolu ağı üzerinde hizmet vermekte olan eski demiryolu köprüleri, yürürlükte olan boyutlandırma standartlarına göre, günümüzdeki demiryolu trafiğinin içerdiği ağır vagonları taşıyacak kapasitede olmamasına rağmen; bunları herhangi bir zorlanma işareti vermeden taşımaktadır. Bunun nedeni, teorik yöntemlerle belirlenen köprü yük taşıma kapasitesi ile gerçek olanı arasındaki büyük farktır. Öte yandan sözü edilen eski demiryolu köprüleri sınırlı yük taşıma kapasitelerine sahip olup, sözü edilen farka çok fazla güvenilemez. Bir köprüden maksimum yarar sağlanabilmesi için, var olduğu iddia edilen yük taşıma kapasitesinin, gerçek yük taşıma kapasitesine belirli bir güvenirlilik limiti içinde yakın olması gerekir. Köprü test yöntemleri, mevcut köprülerin gerçek yük taşıma kapasitelerinin saptanmasında kullanılabilecek güvenilir araçlardır. Bu nedenle, mevcut demiryolu köprülerinin gerçek yük taşıma kapasitelerini belirlemek kadar, elemanlarında birikmiş olması kuvvetle muhtemel olan yorulma hasarlarının varlığının ve yoğunluğunun saptanması, geriye kalan yorulma ömürlerinin belirlenmesi için gerekli analitik ve deneysel yöntemlerin geliştirilmesi de oldukça önemlidir. Bu sebeple, Haydarpaşa-Ankara hattı Klm.55 + 161 de bulunan, 1912 ile 1976 yılları arasında hizmet vermiş olan Diliskelesi demiryolu köprüsü ele alınmış, servis verdiği yıllardaki katar tipi, yük ve trafik durumu kullanılarak boylama kirişler için yorulma hasar birikimi, mevcut bilgisayar programları yardımıyla tespit edilmiş ve ayrıca sözkonusu köprüden enleme ve boylama kirişler alınarak TU-BRIDGES Projesi kapsamında Haydarpaşa'da kurulan Hasar Tespit ve Yorulma Laboratuvarında (HASLAB) yorulma deneyleri yapılmıştır. Elde edilen veriler çeşitli algoritmalardan geçirilerek değerlendirilmiş ve uluslararası yapılan diğer çalışmalarla olan dağılımı incelenmiştir. Bununla birlikte, 1992, 1993, 1994 yıllarında, TU-BRIDGES Projesi kapsamında test edilen ve Ankara- İstanbul demiryolu hattında bulunan Karacam, Bekdemir ve Cambazkaya köprülerinin orjinal projeleri ve arazi testlerinden elde edilen datalar esas alınarak üretilen üç boyutlu, rafine edilmiş bilgisayar modellerinde, günümüze kadar üzerlerinden geçen trafik yükleri gözönüne alınarak birikmiş yorulma hasarları çıkarılmış ve geriye kalan yorulma ömürleri tayin edilmiştir.Railway bridges are part of the Turkey's aging infrastructures. Older railway bridges, most of them steel, were built around at the beginning of the century. The safety of these bridges for transporting both freight and passengers is vitally important. Today there is a renewed interest in the safety of railroad bridges. Safety is of prime objective especially when railroads are used to transport passengers which is a increasing public concern. Öne of the majör concerns of railway bridge engineers today is the safety of old riveted structures and potential fatigue damage that has accumulated (Out et al. (1984), Fisher et al. (1987)). A riveted steel bridge which gave service on Ankara- istanbul main rail corridor between 1912 and 1976 (see Figüre 1) was considered in the fiili scale fatigue test program which was started in TU-BRIDGES Fatigue Laboratory to produce the test data needed for estimation of remaining fatigue life of the other similar railway bridges stili servicing on this railway such as Karacam, Bekdemir, Cambazkaya which have been tested in 1992, 1993, 1994. \ '':M?r^ f%i%jr'',,---',- "-,''.'.:k-"-;; ~^/N \ ir %-~ _.;"!." /,. "'J....',;> | ı. '.r. "'- - i "î v 'ı.- ''"' ',,.'-~w.'l _."*_ - i_L '--İL â _: "V ". " L* Figüre l. A view From the Bridge Selected for Full Scale Fatigue Test Both the American Association of State Highway and Transportation Officials (AASHTO 1989) [5] and the American Railway Engineering Association (AREA 1991) [6] use estimates for fatigue strength of different steel details based on experimental data. Fatigue strength of the bridge members and their riveted connections are expressed in terms of a S-N curve in logarithmic scale, as is exemplifıed in Figüre 2. Stress fy*î --A/\AAAMAAA._ SR^-SU "nin ^^\. Time ^^-^ OONSTANT _\* AMFUTUCE EAUGUEIIMT FINTIEIIFE <= > INHMTELIFE . o ImN Figüre 2. S-N Representation of the Fatigue Strength The S axis corresponds to the stress cycle magnitude and N axis corresponds to the number of stress cycles to failure. Thus, the majör factors governing fatigue strength are the number of stress cycles, the magnitude of stress range, the type of stress range, and the type of construction details. The stress range is defıned as the algebraic differences between the maximum and minimum stress values. The AREA specifıcations state that fatigue need not be considered if both the live load and the dead load result in compressive stress in a member. Recently, Fisher verifıed that for riveted bridges a simple check for category D provides a good estimate and, category D provides a reasonable lower bound for crack development, for the number of cycles required to develop fatigue cracks (Fisher et al. 1987). Figüre 3 shows plot of the S-N curves for category C and D. Since some of the railway cars such as freight cars induce stress range which are below the constant amplitude fatigue limit for both category C and especially D, and therefore do not contribute toward cumulative damage of bridge members. it is important to clearly estimate the constant amplitude fatigue limit. Generally, full scale fatigue tests were conducted at high stress ranges, resulting in fatigue lives which are less than the number of stress cycles corresponding to the actual service lives of most old railway bridges stili giving service. However, in order to eliminate the uncertainty at constant amplitude fatigue limit level, it is necessary to perform the long-life testing of full scale bridge members. xiii Traditionally, full scale fatigue tests were stopped after 2,000,000 stress cycles and thus, results corresponding to stress cycles above this are seldom. On the other hand, most of the fatigue test results in the literatüre are taken from the tests of small specimens which are not able to reflect the actual case under dissimilar testing condition. Full scale fatigue tests for riveted members have been performed in Europe and USA (Figüre 3). Series of full scale tests were carried out by Reemsnyder (1975) on truss connections from riveted öre bridges. The tests were carried out at relatively high stress range levels. Six riveted built-up stringers were tested by Out, Fisher and Yen (1984) with primary focus of the high cycle fatigue behavior under constant cycle stress ranges bervveen 48 and 69 N/mm2. The fatigue tests carried out by Baker and Kulak (1985) were on portion of hanger angles removed from a highway truss bridge. The tests were carried out at a stress range of 165 N/mm2 and 188 N/mm2. Three types bridge stringers were tested by Brühwiler, Smith and Hirt (1988) which were removed from existing bridges, for high cycle fatigue behavior under constant cycle stress ranges between 60 and 120 N/mm2. The test results briefly mentioned up to here are summarized in Figüre 3, along with the others produced in this study. o* ^ O) N t/1 (fi ü U ü u Q Lü U cn " ^. j? O Q f m - ° ° ro t ı ı- h~ 01 ".- \ ' I X " İ" I Kil O) 00 m ı < < s - c ]<* J < < ^ * N ° * / /.a o ö) -L l / / 10 o 5 " P '« / / oı ° *. c l u / / 05 ^ w W ' V / / " r F "D ' l <="" xi="" lomctmcr="" l="" v<'v="" (d="" n="" ı="" \x,.="" c="" tu.-j="" +-"="" \.|="" cj="" -j="" v^="" °="" ! V / / - "İ v^ı ^ / l / l? v X«» /ı/ - -a \k l / / oo r ^. J^'l &. :>*İ Û? '^LM':.. ' ':.- N :- *. ] *. 4 l l '*.* İ -. / ^\&k' : ' ; j | f-.-.^tli.^V?>r-,!}j \ı ^"*J,*,,^5^:'^3''' ' ü; '.'' iK:1SW?i''^'i»«mi, «rrinnİno,.! v:imûi«â l|fi'».»,.- "«," İ a'ğ,,^,.,-^ 'Cr^ll|- »* *'".-? 4 "?'"'»"-.' '*.'-'' iSlfa '^g^^^ '. '. l' *a*ıSZ. '^...! -T. ..- l Ill*_l -T. ' "*~' .' '" . . '' l Figüre 4. A View From the Full Scale Fatigue Test Set-up vvi Figüre 5. A View From the Full Scale Test Set-up Figüre 6. The Stringer Which Has Been Tested Up to the Brittle Fracture Figüre 7. A Close View of the Crack / N. / N. / N. / \. 4000 3875 3875 3875 3875 3875 3875 3875 3875 Side Elevation V/K/K/K/K/K/^ T ^P\l/\l/\l/\l^^ Plan View P 1 r l^j U>j NP475 Gauge |^_ ^J Gauge / \^ Gauge l l y| 327 H845 815 0 11835 328 0 ı 1^ =(s ^ Top Flange ı Bottom Flange *..:\±r.+~..±... »"T\'*|.-*"* » -====:::: "":":::::: Y Gauges Figüre 8. Schematic Description of the Test Set-up Fatigue Rating for Railway Bridges \ r T] ;; ı; [T Current Traffıc Estimated Past and Load Future Traffic Load Tİ T rn i Use Field Use Modified Measurements Comp. Model l Produced in BIM Modüle [2"! _~ l / Computer \^ f Generation of ^ M Member Load r \^ Histories / T] \[ Laboratory Fatigue Tests of Full Scale Bridge Components to define valid S-N Curves ~6] ^ Evaluation Remaining Fatigue Life Prediction for the Total Accumulated Fatigue Damage Brief explanation for these steps are given below : l Estimate of Rail Traffıc Data on the average number of trains per day on the Istanbul-Ankara main rail connection between 1912 and 1976 was determined from the Turkish State Railways Administration records. Ali steam locomotives were phased out for both freight and passenger service on 1965 and replaced by diesel locomotives. This date is very important because the steam locomotives were significantly heavier than the diesel locomotives. Additionally, diesel units allowed for the combining of engines in order to increase horse power and load- pulling capacity, increasing the number of locomotive loading cycles per train. On 1993 electrifıcation of Istanbul-Ankara rail line was completed and electrical locomotives were started to operate on this line. No direct data was available for the car load distribution and car passenger trains for each year betvveen 1912 and 1995. 2 Computer Generation of Member Load Historics in order to perform fatigue rating for individual bridge members starting from the most critical deck elements (stringers and floor beams) against to fatigue, the load histories experienced by critical members due to the past train traffıc were assembled. This was necessary to determine which members experienced tensile forces as well as the number of cycles within each stress level. For this purpose field-testing-based refıned computer model was used which is produced for each tested bridge in TU-BRIDGES project. Member loads under current traffıc were obtained from the individual bridge tests. Then, after having assembled the both results and employing the rainflow procedure presented by J.W.Fisher, B.T. Yen and D. Wang (1989), member load spectrums were obtained for each tested bridges. 3 Full Scale Fatigııe Ttxt* As was indicated in the preceding sections, full scale fatigue tests were started and being conducted to produce the needed amount of data to assess the fatigue limits more confidently for categories C and D. As was indicated before, category D fatigue curve was found to provide good estimate of the cycles for fatigue crack development, and previous studies (J.W. Fisher, B.T. Yen, D. Wang and J.E. Mann (1987)) has shown that fatigue cracking of a riveted built-up steel member can be detected and observed in öne ör more elements of the riveted member when the stress range exceeds 48 N/mm2 and failures at a high number of cycles below the constant stress range fatigue limit for category C, but above the constant stress range fatigue limit value for category D (see Figüre 3). 4 Prediction for the Total Accumulated Fatigue Damage After having completed the needed fatigue data by the full scale fatigue tests mentioned in Step 3, the linear damage accumulation rule proposed by Miner and presented by J.M. Barsom and S.T. Rolfe (1987) will be used to define the total damage accumulated from the past and present traffic. 5 & 6 Use Field Measurements & Use Modificd Computer Model Captured data from a railway bridge test using test equipment and testing procedure are processed to separate the dynamic component from the recorded data to obtain the static component. The second step in evaluating the collected data is to produce a refined computer model which will give a high correlation to the measured strains. Refınement of the three dimensional initial computer model which covers ahnost ali structural irregularities and stiffness changes obtained from bridge documents, construction plans and design calculations, to yield high correlation to the experimentally obtained bridge members' stress has been done employing Manual Remodelling Technique developed in the framework of the TU-BRIDGES project [7] and (Piroglu, F., Uzgider, E., Rahmatian, P.