See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330324241

INITIAL FINDINGS ON RESIDUAL CAPACITY OF A DAMAGED STEEL RAILWAY

BRIDGE GIRDER

Conference Paper · January 2019

CITATIONS READS

0 30

2 authors, including:

Anna Maria Rakoczy

Transportation Technology Center
28 PUBLICATIONS   56 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Onboard technology to detect changes in bridge behavior View project

Railway bridges - Fitness for service View project

All content following this page was uploaded by Anna Maria Rakoczy on 18 January 2019.

The user has requested enhancement of the downloaded file.

 1 INITIAL FINDINGS ON RESIDUAL CAPACITY OF A DAMAGED STEEL RAILWAY
2 BRIDGE GIRDER.
3
4
5
6
7 Anna M. Rakoczy, Ph.D., Corresponding author
8 Transportation Technology Center, Inc. (TTCI)
9 55500 DOT Road, Pueblo, CO, 81001, USA
10 Telephone: 719-584-0782. E-mail: Anna_Rakoczy@aar.com
11
12 Duane Otter, Ph.D., P.E.
13 Transportation Technology Center, Inc. (TTCI)
14 55500 DOT Road, Pueblo, CO, 81001, USA
15 Telephone: 719-584-0594. E-mail: Duane_Otter@aar.com
16
17
18
19 Word count: 3,790 words text + 13 table/figures x 250 words (3,250) = 7,040 words
20
21
22 Submission Date: November 1, 2018
 1 ABSTRACT

2 Transportation Technology Center, Inc. (TTCI) is testing five steel railway bridge spans at
3 the Facility for Accelerated Service Testing (FAST) for fatigue performance. The bridges carry
4 approximately 150 million gross tons (MGTons) (136 million gross tonnes or MGTonnes) per year
5 of heavy axle load (HAL) traffic. TTCI is using these bridges to investigate service life of common
6 steel railway bridge spans.
7 This paper presents initial results of testing on a damaged steel girder span currently
8 ongoing at TTCI. The presence of multiple components in a built-up steel girder provides a level of
9 redundancy within a bridge member. If a single component develops a crack, there will be
10 redistribution of stresses into other components of the built-up member. The redundant load paths
11 within mechanically fastened girders allows the girders to continue to carry load in the event that a
12 single component is damaged (1).
13 TTCI designed the test to provide an in-service evaluation of the potential benefit thought
14 to be present in riveted girders fabricated from multiple components, including web plates, flange
15 angles, and cover plates. Three notch defects were installed in the bottom partial length cover plate
16 (tension flange) of the 32-foot (9.75-meter) span. The notched girder continues to carry load with
17 no change in deflection and more than 2.5 million cycles accumulated. Testing at FAST will
18 provide further information about this behavior.
19 This research is part of the Association of American Railroad’s (AAR) Strategic Research
20 Initiative on bridge life extension.
21 Keywords: Railway Bridges, Riveted Girder, Damaged Girder, Field Test, Crack Propagation
22
23
 Rakoczy & Otter 1

1 INTRODUCTION

2 To investigate member-level redundancy in the field, the test at the Facility for Accelerated
3 Service Testing (FAST) was designed to provide an in-service evaluation of the potential benefit in
4 riveted girders fabricated from multiple components, including web plates, flange angles, and
5 cover plates.
6 Three notch defects were installed in the bottom partial length cover plate (tension flange)
7 of the 32-foot (9.75-meter) span. The objective of this testing is to assess the behavior of a 32-foot
8 (9.75-meter) riveted steel deck plate girder (DPG) span, with tension flange defects intentionally
9 introduced, under heavy axle load (HAL) service at FAST. The specific objective of the member
10 level redundancy testing is (a) determine crack growth rate, which provides information regarding
11 suitable inspection intervals; (b) determine critical crack length at which fracture might occur; and
12 (c) determine whether fracture propagates to additional components of the built-up girder, or the
13 number of additional cycles required to do so. (2) This research is still in progress and will
14 continue as long as the bridge continues to carry the loads. If fracture of the bottom cover plate
15 occurs, the span will be evaluated to determine whether or not there is cracking in any other
16 elements of the tension flange. If not, the span will be evaluated to determine if train operations can
17 be continued until cracking is initiated in a second element of the tension flange. If or when the
18 span is no longer deemed to be suitable for continued FAST train operations, the following span
19 repair options will be considered: repair tension flange with a patch plate and sufficient number of
20 bolts, replace bottom cover plate in its entirety, patch or replace other elements as needed or
21 replace span.
22 Laboratory tests to date on full-scale specimens have shown the crack propagation and
23 fracture of a member element (cover plate, for example) to remain confined to that element rather
24 than to introduce defects into adjoining element upon fracture. (1) Testing at FAST on an actual
25 bridge span under actual train loadings will serve to provide further information about this
26 behavior, which has the potential to greatly extend calculated span life.
27 Multiple safety and monitoring systems are in place on this span to detect changes in bridge
28 behavior and limit deflection in case of changes.
29 LITERATURE REVIEW

30 Connor et al. (2005) reported that, during a period surveyed from 1960 to 2005, no fracture
31 critical bridge with built-up members is known to have failed due to the fracture of one single
32 component propagating a fracture to an adjacent component. (3) In contrast, there are examples
33 of several highway and railroad bridges containing component failures (i.e., failures of one
34 component of a built-up member, such as an angle or plate) which continued to sustain service
35 loads in the “failed” state. Substantial fatigue life remains in a built-up steel girder with a failed
36 component thanks to member-level redundancy. Laboratory investigations at Purdue University
37 have shown that in at least 20 experiments, fracture of a single riveted girder component has not
38 resulted in propagation of a crack to an adjacent component. (1, 4) When a fatigue crack initiates
39 in one cover plate in a built-up section, redistribution of stresses into the uncracked components
40 can occur. The potential for stress redistribution during the crack growth process means the
41 service life of the member and its capacity of carrying the load is greater than the life to crack
42 initiation of a single element.
 Rakoczy & Otter 2

1
2 GENERAL INFORMATION ABOUT FACILITY FOR ACCELERATED SERVICE
3 TESTING (FAST) BRIDGE RESEARCH

4 TTC’s High Tonnage Loop (HTL) at FAST is 2.7 miles (4.3 kilometers) and is well-known
5 in the railway industry as the primary place for testing and evaluation of track components, from
6 rails and ties, to ballast and subgrade (Figure 1). Train operations at FAST include a train of
7 primarily 315,000 lbs. (143 tonnes) gross rail load (GRL) cars. The cars are approximately 10-
8 percent heavier than the maximum interchange standard of 286,000 lbs. (130 tonnes) GRL. Train
9 length is up to 110 cars. Wheels are well maintained, with few flat wheels. Locomotives for many
10 years were four-axle units. In recent years, locomotives are six-axle units with radial trucks.
11 Normal train operations at FAST are at 40 mph (64 km/h). The train operates mostly during the
12 night, with daylight hours used for track and train maintenance and measurements. Tonnage
13 accumulation in recent years ranges from about 120 MGTons, to 170 MGTons (108 MGTonnes to
14 154 MGTonnes). For the train and bridge spans at FAST, 1 million load cycles equates to about
15 158 MGTons (143 MGTonnes).
16

17
18 FIGURE 1. Bridge location at the HTL at FAST
19
20 On a night of undisrupted train operations, approximately 130 trains might pass over the
21 bridges, all in the same direction. Operations typically alternate directions each day. The train is
22 turned twice a week to equalize wheel flange wear.
23 Rail over all the bridges is 136 RE continuous welded rail.
24 The first open deck steel bridge was installed on the HTL at the FAST in 1997. Currently, TTCI
25 is testing five riveted steel deck plate girder (DPG) railway bridge spans for fatigue and safe
26 service life performance and two concrete spans. Other bridge related testing that is ongoing on
 Rakoczy & Otter 3

1 FAST are tests of bridge deck components, open deck bridge ties, deck fastening systems, and
2 bridge approach issues.
3
4 BRIDGE DESCRIPTION

5 The investigated railway bridge is a two span riveted steel DPG with an open deck. The
6 two short steel DPG bridge spans were installed in December, 2014. A 24-foot (7.3-meter) span
7 built in 1913 was donated by Norfolk Southern, from the former Norfolk & Western near Salem,
8 VA. A 33-foot span built in 1904 was donated by Canadian Pacific from near Fernie, British
9 Columbia. This span was shortened by about 7 inches (178 mm) at each end to fit the existing
10 opening. Figure 2 shows the two spans in the west steel bridge at FAST. The bridge is in a 5-
11 degree curve with 4 inches (102 mm)of superelevation. The open deck spans are also superelevated
12 by the same amount. (5)

13
14 FIGURE 2. 24-foot (7.3-meter) and 32-foot (9.75-meter) riveted DPG
15
16 Both spans are purposely overloaded by the HAL train at FAST. The normal rating of the
17 32-foot (9.75-meter) span is Cooper E-54 as provided by the CP. The FAST train, primarily
18 315,000 lbs. (143 tonnes) GRL cars, loading on the high rail girder of this span is E-72 at the
19 normal operating speed. To date these spans have performed well with no maintenance required
20 during the 476 MGTons (432 MGTonnes) of HAL traffic accumulated.
21 Figure 3 shows mid-span tension flange stresses as measured using strain gages in the 32-
22 foot (9.75-meter) span, under normal FAST train operations at 40 mph (64 km/h). The train on the
23 day of these measurements consisted of three locomotives (6 axles each) and 108 cars (4 axles
24 each), with a total weight of 17,395 tons (15,780 tonnes). A few of the cars near the end of the train
25 were only loaded to 286,000 lbs. (130 tonnes), while most were loaded to 315,000 lbs. (143
26 tonnes). The peak stresses measured under train operation correspond to predict bending stresses
27 calculated using gross section and zero impact.
 Rakoczy & Otter 4

32-foot Span, South Girder, 40mph

12 80
10 70
60
8

Stress, MPa
Stress, ksi
50
6 40

4 30
20
2
10
0 0
0 20 40 60 80 100 120
1 Time, sec
2 FIGURE 3. Tension flange stress at mid-span of 32-foot (9.75-meter) span under FAST train
3 loading
4
5 The measurements shown are for the south girder or high rail girder of the span. Note that the span
6 unloads completely (stress goes down to zero) between the lead and trail trucks of each car. This is
7 expected as the span length is less than the inside axle spacing on the cars, so there is a brief period
8 beneath each car when there are no axles on the span. Since fatigue is governed by stress range or
9 magnitude of the stress cycles, these short spans can be more susceptible to fatigue as compared to
10 longer spans that do not experience full unloading as the train traverses.
11
12 CALCULATION FOR BUILT-UP GIRDER IN 32-FOOT (9.75-METER) SPAN

13 The 32-foot (9.75-meter) DPG span at FAST has two main girders that are built-up sections
14 including web plates, flange angles, two top cover plates, and two bottom cover plates as shown in
15 the Figure 4.
16
 Rakoczy & Otter 5

Top partial length cover plate 15” x 7/16” PL x 22’4” (381 x 11 x 6807 mm)
Top full length cover plate 15” x 3/8” PL x 31’10” (381 x 9.5 x 9700 mm)
Top flange angles 6” x 6” x ½” L x 31’10” (152 x 152 x 13 x 9700 mm)

Web 36” x 3/8” PL x 31’10” (927 x 9.5 x 9700 mm)

Bottom flange angles 6” x 6” x ½” L x 31’10” (152 x 152 x 13 x 9700 mm)

Bottom full length cover plate 15” x 3/8” PL x 31’10” (381 x 9.5 x 9700 mm)
Bottom partial length cover plate 15” x 7/16” PL x 22’4” (381 x 11 x 6807 mm)

1 FIGURE 4. Built-up section of main girder of 32-foot (9.75-meter) span and position of the
2 gages.
3
4 Mid-span moment due to 53-foot (16-meter) cars of 315,000 lbs. (143 tonnes) on the 32-
5 foot (9.75-meter) span is about 1,522.5 kip-ft (2,064 kN-m) under static load conditions. The
6 calculated stresses using gross section due to bending moment are presented in Table 1. AREMA
7 (6) recommends the use of net section; however, the gross section stresses are closer to the
8 measured values.
9 TABLE 1. Calculated live load stresses using gross section
10
Moment of Inertia, in4 Centroid, Ys, in Stress, ksi
Gross Section
(mm4) (mm) (MPa)

16,395.8 19.1
With all bottom cover plates 10.6 (73)
(6,824,447,208) (484)
13,784.95 20.9
With one bottom cover plate 13.9 (96)
(5,737,729,390) (532)
11,095.43 22.9
Without both bottom cover plates 18.9 (130)
(4,618,266,647) (582)
11
12 The stresses get higher if the section loses both bottom cover plates. However, losing a
13 cover plate is not common. Even with a fatigue crack on the plate, the plate can still carry some
14 load. If the crack is detected in a timely manner and the condition is repaired, the bridge can
15 continue to be used without additional restrictions.
 Rakoczy & Otter 6

1 MODIFICATIONS TO 32-FOOT (9.75-METER) RIVETED STEEL SPAN AND SAFETY

2 MEASURES

3 Defects were installed in the bottom partial-length cover plate (tension flange) of the 32-foot (9.75-
4 meter) span. The girder cross section at mid-span is shown in Figure 4. Defects were introduced in
5 the partial-length bottom cover plate of the most heavily loaded girder (south side, high rail) in
6 order to provide stress concentrations in which cracks might initiate and propagate. Defects were
7 installed at locations near centerline, and at distances of about 33 inches west and east of the center
8 notch. Defects consist of three notches ground in the bottom cover plate using a 4 inch cut-off disk,
9 extending 1.5 inches into the width of the flange. Approximately 1/32-inch remained at the top of
10 the bottom flange at the outside edge. (Figure 5). In order to replicate the Purdue laboratory tests as
11 closely as possible, Dr. Hebdon and Dr. Connor came to TTC and performed the notching using
12 the same tooling that was used in their lab tests (1, 4).

13
14 FIGURE 5. Three notches cut into bottom cover plate of 32-foot (9.75-meter) span south
15 girder
16
17 Safety cribbing, shown in Figure 6, was installed to limit the deflection of the span should a
18 fracture occur. In addition, a deflection limit switch was installed and connected to the signal
19 system. Deflection indicators also are used for periodic monitoring of maximum deflection during
20 train operations. A series of three crack length indication gages were also installed. They are wired
21 to an independent signal that activates if a crack grows beyond any one of the gages.
 Rakoczy & Otter 7

1
2 FIGURE 6. Safety cribbing and instrumentation installed under 32-foot (9.75-meter) span of
3 FAST west steel bridge
4
5 MEASUREMENTS AND DATA RESULTS TO DATE

6 For the 32-foot (9.75-meter) span, the following measurements are collected under normal
7 FAST train operations:
8 • Mid-span tension flange strain in the axial direction, on the bottom cover plates, for each
9 girder.
10 o For the south (high rail) girder which is most heavily loaded, and containing the
11 introduced defect(s), strain gages are applied to each tension flange element (flange
12 angles, each side of the full length cover plate, and both sides of the web about 1
13 inch above the flange angles).
14 o For the south and north girder, tension flange strain gages at the quarter span
15 location are also used.
16 • Mid-span vertical deflections, on the south girder.
17 Data collection is triggered automatically as the train approaches the bridge.
18 The defects on the bottom flange were introduced in September 2015. Since that time, 362
19 MGTons (328 MGTonnes) has been accumulated. There are cracks visible in each of the three
20 notches (Figure 7). In Notch 1 the crack has not yet propagated beyond the notch. The cracks
21 beyond the notches were reported to be 0.10 inch (notch 2) and 0.08 inch (notch 3) in June 2017.
22 The cracks are inspected weekly with visual inspection, and monthly using either magnetic
23 particles or dye penetrant. The cracks have not shown any growth during the last several months of
24 train operation.
 Rakoczy & Otter 8

1
2 FIGURE 7. Crack growing out of notch 2
3
4 The stresses near the center of the bottom cover plate at mid-span are remaining at the same
5 level since February 2015 (Figure 8). Data from February, March, April and May of 2015 are from
6 the girder before notches were introduced while others are measurements from the bridge with
7 defects.
12 80
10 70
60
Stress, MPa
8
Stress, ksi

50
6 40
4 30
20
2
10
0 0
North mid- South mid- North South
span span quarter quarter
02_2015 03_2015 04_2015 05_2015 09_2015
10_2015 11_2015 02_2016 05_2017 09_2017
8 02_2018 03_2018 04_2018

9 FIGURE 8. Mid-span and quarter location stress comparison of north and south girder,
10 train speed 40 mph
11
 Rakoczy & Otter 9

1 Figure 9 shows mid-span stresses on the south girder since September 2015, and they are at
2 the same level since the defects were introduced. The bridge will continue to be monitored,
3 inspected frequently, and evaluated for repair should crack growth or a fracture occur.

12 80
10 70
60

Stress, MPa
8
Stress, ksi
50
6 40

4 30
20
2
10
0 0
Outside Inside Outside Inside Web
Cov. Plt. Cov. Plt. Angle Angle
09_2015 10_2015 11_2015 02_2016 05_2017

4 09_2017 02_2018 03_2018 04_2018

5 FIGURE 9. Stress comparison of various components on south girder, train speed 40 mph
6
7 BRIDGE FATIGUE LIFE ESTIMATES FOR UNDAMAGED CONDITION

8 The fatigue life of this span in the undamaged condition can be estimated in a number of
9 ways. For short spans, with span length less than the inside axle spacing of the cars, the spans
10 experience a complete loading and unloading cycle for each car that traverses the span. The FAST
11 train consists primarily of 53-foot (16-meter) rotary dump cars with an inside axle spacing of about
12 34.5 feet (10.5-meter). The effective span length (center to center of bearings) of the span is
13 approximately 31 feet (9.5 meters). For typical short steel girder spans, the governing cross section
14 will be near mid-span and bending moment will be the governing load case.
15 Bridge fatigue life estimates are normally first made using theoretical calculations. The
16 quick and simple calculations using basic load rating information are conservative. More detailed
17 calculations, taking full advantage of the AREMA rating provisions, can provide a better estimate.
18 Where actual strain gage data is available, it provides a more accurate estimate of fatigue life for a
19 particular structure. (6) Several fatigue life estimates are demonstrated using both theoretical and
20 measured stress ranges. Past traffic history is also necessary to determine the remaining life of a
21 bridge. However, it is often difficult to estimate past traffic due to inadequate records and variable
22 train types on many lines. It is believed that this particular span experienced approximately 20
23 years at 45 MGTons (41 MGTonnes) per year of unit trains with 263,000 lbs. (119 tonnes) cars and
24 20 years at 40 MGTons (36 MGTonnes) per year of unit trains with 286,000 lbs. (130 tonnes) cars.
25 (5)
26 First the fatigue life is calculated using theoretical stress range and FAST loading of 53-
27 foot (16-meter) rotary dump cars (72-inch (1.8 meter) axle spacing, 40.5-foot (12.3 meter) truck
28 centers) with a gross rail load of 315,000 pounds (143 tonnes). The maximum bending moment
 Rakoczy & Otter 10

1 from this loading is increased to account for impact. The design impact is computed and reduced
2 for the normal train operating speed of 40 mph (64 km/h) at FAST. The bending moment also is
3 adjusted for the effects of superelevation since this bridge is on a 5-degree curve with 4 inches (10
4 centimeter) of superelevation. Maximum stress is computed using the net section modulus of the
5 girder. These calculations are all made in the same manner as is done for an AREMA load capacity
6 rating. For a fatigue life estimate, the amount of impact can be further reduced per Table 15-1-7 in
7 the AREMA Manual for Railway Engineering. (6) Girder spans are calculated using 35 percent of
8 the impact used for load capacity rating. This results in a reduced bending moment to use for
9 calculating the stress range. For short spans (spans that experience full unloading under each car),
10 the stress range is simply equal to the maximum live load stress. And the number of load cycles per
11 train will be approximately the same as the number of cars in the train, with a few miscellaneous
12 cycles due to locomotives and end of train. The theoretical stress range under the FAST train is
13 13.85 ksi (95 MPa) (net section stress, adjusted for fatigue impact and superelevation at 40 mph).
14 For stress ranges above 9 ksi (62 MPa), AREMA (6) and the American Association of State
15 Highway and Transportation Officials (AASHTO), per the National Steel Bridge Alliance
16 (NSBA), (7) both recommend the same life estimation procedure for riveted spans, i.e., the
17 standard Category D fatigue life (S-N) curve. Using fatigue Category D for riveted details, the
18 fatigue life is estimated at 0.82 million cycles for the 32-foot (9.75-meter) span. This number
19 corresponds to 130 MGTons (118 MGTonnes) of HAL traffic for the 315,000-pounds (143-tonnes)
20 cars at FAST. A bridge life this short is likely acceptable only on lines carrying minimal amounts
21 of traffic.
22 Then, the fatigue life estimate is obtained using accurate stress range information by means
23 of strain gage measurements under actual traffic. The measured stresses are considerably lower;
24 thus, use of the measured stress range should produce more acceptable life estimates for this span.
25 There are several possible factors contributing to measured stress range being less than that
26 calculated using basic beam theory: 1) distribution of wheel loads by the rail and deck; 2) partial
27 fixity of the bearings in translation and rotation; 3) partial section contribution from the rail and
28 deck; 4) actual impact (due to smooth continuous welded rail) being less than that assumed; 5) all
29 components of the girders acting in unison; and 6) effective span length. The peak stresses vary
30 from car to car. In order to use the data from a typical train pass, the cycles should be counted
31 using a rain flow cycle counting method. (7) Distribution of the stress ranges for the 32-foot (9.75
32 meter) span is shown in Figure 10. This distribution represents two passes of the FAST train, one
33 in each direction (clockwise and counterclockwise), to provide a full representation of the traffic.
 Rakoczy & Otter 11

South Girder of 32-foot Span

25%

20%
Percentage

15%

10%

5%

0%
6.4
6.7
7
7.3
7.6
7.9
8.2
8.5
8.8
9.1
9.4
9.7
10
10.3
10.6
10.9
11.2
11.5
1 Stress Range Cycle, ksi

2 FIGURE 10. Stress range cycles for the FAST 32-foot (9.75-meter) steel span
3
4 Following AREMA guidelines, an equivalent stress range can be calculated using the root-
5 mean-cube method. For the 32-foot (9.75-meter), all the cycles are above 6 ksi (41 MPa), so all
6 cycles should be included. Figure 11 presents equivalent stress ranges for various locations on the
7 span due to the test train consist for 120 days of measurements. The equivalent stress range for the
8 32-foot (9.75-meter) span is 9.85 ksi (68 MPa) for bottom cover plate and 10.3 ksi (71 MPa) for
9 inside angle. This is significantly lower than the theoretical value of 13.85 ksi (95.5 MPa). At 9.85
10 ksi (68 MPa), the fatigue life estimate using Category D is 2.3 million cycles, or 355 MGTons (322
11 MGTonnes) for the HAL train. This life estimate is about three times greater than the estimate
12 based on simple theoretical calculations. However, a bridge life this short is likely still only
13 acceptable for lower tonnage lines.

Equivalent stress
12 80
Equivalent stress, MPa

70
Equivalent stress, ksi

10
60
8
50
6 40

4 30
North mid-span South mid-span 20
2 Cov.Pl. outside Cov.Pl. inside
Angle outside Angle inside 10
Web
0 0
0

20

40

60

80

100

120

Day #
14
15 FIGURE 11. Equivalent stress ranges for the FAST 32-foot (9.75-meter) steel span
 Rakoczy & Otter 12

1
2 The AREMA rating guidelines offer a further life increase for stress range cycles below 9
3 ksi (62 MPa). In that case, stress cycles can be compared against what is essentially a Category C
4 fatigue life curve. In order to properly implement this AREMA provision, the cycles need be
5 evaluated against different portions of the S-N curve using the Palmgren-Miner rule as described in
6 the NSBA Fatigue Primer. (7) The equivalent stress range equation given in AREMA cannot be
7 used. Because there are so few stress range cycles below 9 ksi (62 MPa) for this span, the resulting
8 fatigue life estimate is improved only slightly, to 2.4 million cycles (368 MGTons, 334
9 MGTonnes). Figure 12 shows a comparison of the fatigue life estimates for the 32-foot (9.75-
10 meter) span. The rating provisions discussed (simple Category D and AREMA Rating with NSBA
11 implementation) are included.
12

MGT Estimate Fatigue Life, 32-foot Span

500

400

300

200

100

0
Theoretical Measured Measured Current FAST
Stresses Stresses Stresses Tonnage
Category D Category D AREMA Rating
13 & NSBA

14 FIGURE 12. Comparison of fatigue life estimates for FAST 32-foot (9.75 meter) span in
15 undamaged condition using stress cycle ranges and rating provisions
16
17 In presented methods, the predicted fatigue life estimate is showing that the span should
18 have failed due to fatigue years ago. New methods of fatigue life estimation are needed to provide
19 more realistic fatigue life assessment. TTCI is working on a probabilistic method for fatigue
20 evaluation that will provide a broader picture of fatigue life. Using a probabilistic method, the
21 number of cycles or accumulated traffic is estimated in terms of probability of fatigue crack
22 initiation. The reliability analysis also can be used for estimating predicted years of service life (or
23 MGT) of the bridge with different levels of safety. The probabilistic method can be very beneficial
24 for railroad bridge owners who have control over inspections, and know the traffic that has and
25 will continue to operate on their structures.
26
 Rakoczy & Otter 13

1 CONCLUSIONS AND FUTURE RESEARCH

2 To date, the 32-foot (9.75-meter) span at FAST continues to carry train loads with no
3 maintenance required, no additional defects noted, and total accumulated tonnage of at least 1,926
4 MGTonnes (2,176 MGT) with over 500 MGTons (450 MGTonnes) of HAL traffic. The presence
5 of more than one cover plate in a built-up steel girder provides a level of redundancy within a
6 bridge member. If a single component develops a crack, there can be redistribution of stresses into
7 other components of the built-up member. The notched girder continues to carry train loads with
8 389 MGTons (352 MGTonnes) and over 2.5 million load cycles accumulated after the notches
9 were introduced. There are cracks visible in each of the three notches. However, only minimal
10 propagation of cracks beyond the notches is noticeable in two of the notches. The crack is still
11 within the third notch. The stresses near the center of the bottom cover plate at mid-span are
12 remaining at the same level since February 2015. The bridge will continue to be monitored,
13 inspected frequently, and evaluated for repair should excessive crack growth or a fracture occur.
14
15 ACKNOWLEDGEMENTS

16 The authors are indebted to Dr. Robert A.P. Sweeney, CN (retired), and Dr. Robert J.
17 Connor of Purdue University and Dr. Matthew Hebdon from Virginia Tech for their assistance
18 with interpretation and implementation of the various fatigue evaluation techniques available to the
19 bridge engineering community.
20
21 AUTHOR CONTRIBUTION STATEMENT

22 The authors confirm contribution to the paper as follows: study conception and design: A.
23 Rakoczy, D. Otter; data collection: A. Rakoczy, D. Otter; analysis and interpretation of results: A.
24 Rakoczy, D. Otter; draft manuscript preparation: A. Rakoczy, D. Otter. Both authors reviewed the
25 results and approved the final version of the manuscript.
26
27 REFERENCES

28 1. Hebdon Matthew H., C. Korkmaz, and Robert J. Connor, “Member-level redundancy of built-up
29 steel girders.” Purdue University, West Lafayette, IN, 2015
30 2. Otter, Duane. “Old Steel Spans for New Bridge Research at FAST.” Railway Track & Structures
31 112, no. 6 (2016).
32 3. Connor, Robert J., Robert Dexter, and Hussam Mahmoud, “NCHRP Synthesis 354: Inspection
33 and Management of Bridges with Fracture-Critical Details.” Washington DC, Retrieved, 2005.
34 4. Hebdon, Matthew and Robert Connor. “Towards a Better Estimation of Inspection Intervals for
35 Cracking Steel Railroad Bridges.” In Proc. AREMA Conference, 2015.
36 5. Otter, Duane, Anna M. Rakoczy, and Stephen Dick. “Fatigue Life and Fitness-for-Service
37 Analysis: 32-foot Steel Deck Plate Girder Bridge Span at FAST.” Technology Digest TD-16-025.
38 AAR/TTCI. Pueblo, CO. 2016.
39 6. American Railway Engineering and Maintenance of Way Association, Manual for Railway
40 Engineering AREMA. Chapter 15 – Steel Structures, Lanham, MD. 2016.
 Rakoczy & Otter 14

1 7. John W. Fisher, Geoffrey L. Kulak, and Ian F.C. Smith. A Fatigue Primer for Structural
2 Engineers, National Steel Bridge Alliance, USA, 1998.

View publication stats