Fatigue and Repair Cases in Steel Bridges

by Chitoshi Miki, Yuichi Ito and Eiichi Sasaki Department of Civil Engineering, Tokyo Institute of Technology 2-12-1 Oookayama, Meguro-ku, Tokyo 152-8552, Japan
Abstract Steel bridge fatigue damage cases and their repair are summarized. They are organized by the causes of fatigue damage such as the existence of weld defects, deformation, and so on. These fatigue damage cases and repair methods were collected and stored in a database. This paper also introduces the database which is accessible from the internet.

1. Introduction Many steel bridges have been constructed with the development of highway and railway networks. Welded girder bridges, which are sensitive to fatigue, have been positively adopted since 1960s. As a result, many fatigue failures have been observed in the U.S.A, Japan, and other countries since the 1960s[1-12]. Because the steels used will suffer damage year after year, and both the traffic volume and the usage requirements will go on increasing from now on, it is feared that the load for the structure will become too heavy. Therefore, fatigue failures will increase, and bridge maintenance will gain importance in order to steel bridges safe. Because the causes of failures and repair measures for the cases are often referred from the repair cases reported in the past, it is very useful to gain knowledge from the past case studies and it is expected that such accumulated information will be of use in the maintenance technology. Therefore, repair cases for past fatigue failures must be collected so that anyone can refer to them. As a result of the activities of IIW-XIII-WG5, many repair cases for fatigue failures observed at various welded joints have been collected. Recently, networks of information on the internet are very substantial throughout the world. We now consider providing through the internet the information on many steel bridge repair cases for concerned persons as well as bridge engineers who are engaged in bridge maintenance business. This paper introduces the cases of fatigue damage, the repair and retrofit method applied to these steel bridges, and the system of the database of repair cases. The evaluation of the repair methods is also discussed using this database. 2. Fatigue in Steel Bridges and Retrofit Works The causes of fatigue of steel bridges may be classified as follows: Welding defects were included at the time of fabrication. An inappropriate structural detail of low fatigue strength had been adopted. Stresses and deformations unforeseen in design occurred at joints of members. The structure behaved in a manner not expected, such as due to vibration. There are two approaches to repairing and reinforcing: To remove the cause of damage or to alleviate it.

 To increase the fatigue strength of the detail concerned. A thorough explanation of the causes of damage is indispensable in selecting methods of repair and reinforcement for these bridges. Depending on the damage, there are cases when it is permissible to leave the component untouched. On the other hand, some cracks which will be dangerous unless a remedial measure is immediately applied. 2.1 Existence of Welding Defect There are cases of girders failing due to fatigue cracks occurring from welding defects unintentionally left in the welds of the plate joints at the bottom flanges of plate girders. Normally, such a joint is subjected to X-ray inspection after welding so that occurrences of damage are rare. A fatigue crack, after propagating through the bottom flange, passes the fillet weld between the flange and web and penetrates to the web. Brittle failure will occur in the event the fracture toughness of the steel is low. Fig. 1 is a case of a bottom flange failing due to insufficient penetration of a butt weld [13]. There are also cases reported of cracking of welds being a cause besides lack of penetration [12]. Such structural details would be no problem at all if there were no defects, and it may be said that it would suffice to perform welding so that a joint with sound quality is again obtained. However, this welding would be done in the field, and in many cases it will be difficult to secure working conditions under which sound welded joints can be obtained. There are also many cases in which brittle cracking has gone through the web and has stopped immediately under the top flange for problems such as removal of cracks, groove preparation, or handling of the start-restart part of welding. Consequently, in such cases, distortion due to cracking is removed by a method such as jack-up, after which repair is done by splicing using high-strength bolts [12](Fig. 2).

Fig. 1 Fatigue crack propagated through girder flange

Fig. 2 Repairing by bolted splice

Fig. 3 Change of allowable fatigue stresses from 60 code to 92 code

It is close to impossible to predict the appearance of fatigue damage caused by a defect. However, after a single case has occurred, it is possible to take measures such as intensive inspection of joints made in the same factory around the same time using the same welding details. In the case of Fig. 1, as a result of inspecting butt welds made in the same factory for the same bridge, a large number of incomplete penetration defects were discovered. 2.2 Adoption of Joint Details of Low Fatigue Strength This would include cases of bridges not designed against fatigue in the first place being subjected to conditions of use where fatigue would be a problem, as often occurs with old structures, the allowable stresses used in fatigue design on subsequent review being found to have been inappropriate. Fig. 3 shows a comparison of the allowable fatigue stress of a typical joint in a steel bridge in 1964 and at present in the fatigue design codes in Japan [16]. As a result of fatigue tests carried out since the 1960s on many full-scale and large models, allowable fatigue stress ranges were reconsidered and made the present values. With regard to such joint details, there are numerous cases in which fairly large-scale measures are necessary, such as raising the fatigue strength of the detail or lowering the stress occurring in the zone. 2.2.1 Cover plate end Fig. 4 shows fatigue cracking produced in weld toes of fillet welds used for attaching cover plates[17]. This is a part where high stress concentrations occur, and the fatigue strength, as classified by AASHTO and JSSC as the joint type of lowest fatigue strength, is very low. Depending on the configuration at this part, there are cases of fatigue strengths being even lower than assumed in design guides, and as seen in the King's Bridge accident [18], this is also a detail in which welding defects are liable to occur. In the case of Fig. 4, fatigue cracks occurred at many places, and the method of repair differed according to the dimensions of the crack. With regard to cracks of surface lengths not less than 38 mm, as shown in Fig.5, drillholes are provided immediately above the cracks and splicing is done with high-strength bolts. For cracks smaller than this, TIG dressing or hammer peening is applied [23](Fig.6).

Fig. 4 Fatigue crack initiated from the weld toe at the end of cover plate

Gas tungsten arc remelted weld toe

Fig. 5 Retrofit details of bolted splice

Peened weld toe

Fig. 6 Retrofit by TIG dressing and hammer peening

2.2.2 Rigid frame bridge bent Fig. 7 shows fatigue cracks in the bridge rigid bent in the Route 3 of Tokyo Metropolitan Expressway. In the design of rigid frame bents, the stress concentration due to shear lag behavior is taken into account based on Okumuras studies [24]. However, the measured stresses are much higher than calculated stresses in accordance with the standard design procedures for these structures. Detailed inspection revealed fatigue cracks initiated from the weld root of partially penetrated welds between the flange plate of the beam and the flange plate of the column, which are defects of incomplete penetration. Fig. 8 shows the variations of assembling method of rigid frame bent of box section column and beam. Because of the complicated crossing of plate elements of beam and column, welding work becomes almost impossible for some portions. These kinds of structural details lead to the occurrence of inherent weld defects and become the causes of fatigue accidents. The existence of these inherent defects and stress concentration due to the structural geometry are the reasons for this fatigue accident. All steel rigid bents in highway bridges in Japan were designed by applying the

Fig. 7 Fatigue Crack in Rigid Frame Bridge Bent

(a) Type A

(b) Type B

(c) Type C

Fig. 8 Inherent defects due to assembling procedures

Column

Beam

tw =26mm =26m tw

Fig. 9 Temporary Repairing by Bolted Splices

same design formula. Therefore, all rigid frame bents in Tokyo Metropolitan Expressway were inspected and fatigue cracks were observed in more than 300 rigid frame bents. From the viewpoint of preventing brittle fracture, the critical crack length of 30mm has been set and measures have to be taken for cracks which exceed this length. In order to investigate the suitable retrofit methods, wide research work has been performed, including large scale fatigue tests [25] and field measurements. Four bents with long cracks have been fixed by applying bolted splices as a temporary measure (Fig. 9).

2.2.3 Gusset Plate Joint As shown in Fig. 3, there is a large difference in the allowable stresses of gusset joints made on webs or flanges between the 1960 and the present design standards for railway bridges. As a preventive measure, improving the fatigue strength of this detail was studied. Fig. 10 shows one of the methods being proposed to gusset plate on flange edge, in which a specially developed circular cutting device (Fig. 11) is used to shape the radius of the gusset end large, and the fatigue strength is increased by reducing local stress concentration at this location which becomes the initiation point of fatigue cracking. Fatigue prevention measures using this method are presently being applied on a trial basis. Regarding the end of a gusset attached to a web, it is difficult to reduce stress concentration by changing the configuration as in Fig. 10. The methods conceivable are to raise fatigue strength by finishing the weld toe by grinding, TIG-dressing, or hammer peening. However, gussets or attachments of this kind are attached by fillet welding in many cases, and if the toe is finished, fatigue cracks will then occur from the root, and not much improvement can be expected.

Improve the figure (machine-cut or gas-cut)

Fig. 10 Improving works of flange gusset detail Fig. 11 Cutting works by applying the newly developed tool

Fig. 12 Improving works of flange gusset detail

Fig. 12 shows a method which has been proposed and is being used on a trial basis as a measure against fatigue cracks which have occurred in similar details in highway bridges, aiming to reduce stresses at gusset ends by splicing with high-strength bolts. Two of the holes for bolts have been drilled at the ends of a crack and serve as stop holes. Stresses near the gusset end can be reduced 20 to 30% by this method, and so this is an adequate countermeasure. 2.3 Occurrence of Unexpected Stress and Deformation at Joints of Members Various rules have been established in order to simplify calculations in the structural design of a bridge. Particularly, with regard to connections between perpendicularly crossing members such as main girders and cross beams or cross beams and stringers, simple supports or pin connections are often specified, and deformations and stresses obtained from design calculations differ considerably from those in actual structures, especially in the vicinities of connections. Rules are on the conservative side when determining cross-sectional dimensions, but fatigue is often caused by secondary stresses due to restraint moments. 2.3.1 Main girder-floor beam connection in through-type plate girder Fig. 13 shows a through-type plate girder railway bridge in which fatigue cracking occurred when the flanges of the floor beams were provided with cut-outs to facilitate joining at the connections attaching the floor beams to the main girders [19]. In the design, floor beams are provided simple support by main girders; thus only shear force, reactions of cross beams, are transmitted at the joints, and there is no problem if cut-outs are made in the flanges.

Fig. 13 Through-type plate girder railway bridge, Fatigue cracks initiated at the connections between main girders

(a) Original detail

(b) Improved detail

Fig. 14 Retrofitting works at girder to cross beam connections

However, since the floor beams are fixed to main girders in the actual structure, fixed-end restraints moments are produced. As a result, fairly large direct stress components will occur in the flanges. Because of the large decrease in cross section and the change in configuration due to the cut-outs of the flanges, large stress concentrations occur at these locations. For repair, it will suffice to make a connection which is adequate against fixed-end moment, and as Fig. 14 shows, the structural detail of the connection was changed so that the force of a cross beam flange can be transmitted to a main girder flange. Also, holes were made at the ends of the fatigue cracks, and a fairly large area of the web was spliced. Connections attaching stringers to cross beams had fatigue cracks due to the same causes, and the repair concept is the same. 2.3.2 Main truss-floor beam Connection Fig. 15 shows fatigue cracking which occurred at the connection between floor beams and panel points of trusses in a deck-type truss girder highway bridge. This damage occurred because these detapass. Upon carrying out various examinations, it was decided that as the reinforcing measure the top flange of a floor beam and the flange of the upper chord member of a truss should be joined using a connection plate as shown in Fig. 16. In order to weld the connection plates to the flange plate of the truss, concrete at the top surface of the truss was blasted off by water jet.

Fig. 15 Fatigue crack at the connection between floor beam and truss panel point

Fig. 16 Retrofitting works: Floor beam and the flange of truss top chord was connected directly using a connection plate

Fig. 17 Fatigue cracks at the cross bracing connection details

Fig. 18 Various retrofitting methods for cross bracing connection details

Fig. 19 Cross bracings and stringers supplement work

2.3.3 Cross bracing connection in plate girder bridge[20] Fig. 17 shows fatigue cracking which occurred at welds between a vertical stiffener used attaching cross bracing of a plate girder bridge and an upper flange and web. Such fatigue is due to fairly large forces being produced at the various members of the cross bracing from the difference in deflection between the main girders caused by vehicle loads and the forcible deformation of the upper flanges of plate girders by deflection of the concrete deck in a direction perpendicular to the bridge axis. Various methods are employed for repair of this damage considering the degree of damage and ease of execution (Fig. 18). For the smallest cracks, in cases where there is no problem about size with fillet welds, rewelding the crack is done by TIG dressing. In case of small size of fillet weld with the possibility of crack occurrence from the root, after one to three passes of fillet welding on top of the crack, the toe is finished by TIG or grinding. When the

crack is large, it is removed by gouging after which rewelding is performed by full-penetration welding, with toe finishing done by TIG or grinding. The results of follow-up investigations indicated that most of the repaired spots remained sound, but fatigue cracks have occurred again where the stresses produced were large and where cracks had been large. When it is possible to remove the concrete deck on top of the girders, vertical stiffeners can be connected to upper flanges with high-strength bolts, and the forces at the various members of the cross bracings can be transmitted smoothly to the main girders. Improving the deformation behavior of the structure as a whole is also done by installing cross bracings and stringers (Fig. 19). In this case, the plate thickness of vertical stiffeners for attaching new cross bracings is changed from the 10 mm used in the past to about 16 mm, while their widths were made as large as possible. Further, to lower stresses at damaged parts, horizontal members of old cross bracings were removed. 2.3.4 Sole plate end Support points on a bridge may be said to be points of the greatest load concentrations on the bridge. Beam theory is normally employed in designing a bridge, but when sizes of girders and locations of supporting points are considered, the stresses occurring near support points differ considerably from those obtained by beam theory. Consequently, the regions around of supporting points are susceptible to fatigue damage. Fig. 20 shows fatigue cracking which occurred in a plate girder bridge where a sole plate was attached to the underside of a bottom flange by fillet welding [21]. For sole plates, when rotating function is lost, very high stresses occur at the front surface of the sole plate, and this local stress is made even higher by the deformation of the flange when the flange of the main girder and web are welded,

Fig. 20 Fatigue damage in sole plate connection detail at support

Fig. 21 Retrofitting method for sole plate detail

Fig. 22 Fatigue cracks at the corner of reduced height of girder

Fig. 23 Retrofitting Works by Reinforcing Plates

An example of repair is shown in Fig. 21. The sole plate was changed to a longer one. Joining with the bottom flange was done using high-strength bolts. As the cracks had penetrated into the web, holes were drilled at the tips along with which splicing was done using high-strength bolts. The holes at the tips of cracks were also stopped with high-strength bolts. When cracks of webs are large, welding may also be done. 2.3.5 Cut-out web Fig. 22 shows fatigue cracking which occurred where a cut-out had been made at the end of a girder of a plate girder bridge. Regarding this part, bending stress is extremely small according to the beam theory, but since a cut-out is provided, a high stress component is produced in the normal-line direction at the corner of the curved part. This stress component becomes higher the smaller the curvature ratio of the curve. Since the flange and the web are joined by fillet welding, when looked at locally, this fillet weld is supposed to transmit load. Fatigue cracks often are initiated from the roots of welds because of this. In repairing, in order to lower stress at the bottom flange as much as possible, and to transfer stress smoothly to the supporting point, integrated reinforcing plates were attached to the flange and web with high-strength bolts (Fig. 23). In bridges newly constructed, ribs are attached in a manner to be continuous from the flange, the curvature ratios of cut-outs are increased, and welds are all made full-penetration welds to reduce stresses at this part. 2.3.6 Orthotropic steel bridge decks The orthotropic steel deck system is light-weight compared with concrete deck slabs and is suited to long span bridges. Because decks support traffic loads directly, and the thinner orthotropic steel bridge decks are flexible, actual stresses due to traffic loads in elements are different from those in design calculations, and fatigue is very severe. Fig. 24 shows the parts of orthotropic steel decks in which fatigue cracks developed. Fatigue cracks have developed most frequently from field welded joints of trough ribs used backing strips, and scallops at the intersections of the longitudinal and transverse ribs. Fatigue cracks initiated from the root of welds between trough ribs and deck plate and penetrating into the deck plate are one of the most serious examples of fatigue damage which have been observed recently.

Corner plate

Longitudinal Rib

Deck plate

Transverse Rib

Vertical Stiffener

Fatigue Crack

Fatigue Crack

Fatigue Cracks (Type E) Fig. 24 Fatigue cracks in orthotoropic steel decks

2.4 Vibration 2.4.1 Girders in the high speed train system When a train travels at high speed, there may be a case of vibration of a kind unforeseen occurring in a bridge. Fig. 25 is an example of fatigue damage due to vibration in bridge structures of the Shinkansen Line of Japan. Since the start of operation in 1964, speeds have been increased, and from about the time that 200 km/h was exceeded, vibrations of bottom flanges of stringers in plate girder bridges and truss girder bridges in directions perpendicular to the bridge axes began to appear prominently when crossed by trains. Also, vibrations in out-of-plane directions occurred in diaphragms of box-section girder bridges. Fatigue cracks have occurred in vertical stiffeners and ribs restraining such vibrations. Repair methods differ depending on the degree of damage. In case a crack has gone through the thickness of web plate, holes are drilled at both ends of the crack, and splicing is done using a plate provided with ribs (Fig. 26). When a crack is small with length at the surface about 20 mm, one to three passes of fillet welding is done on top of the crack, followed by TIG dressing (Fig. 27). Similar steps are taken as preventive measures for parts not yet cracked. The results of surveys 10 years after implementing such measures indicated cracks have not reoccurred, indicating that the measures have been appropriate.

End of Vertical Stiffener [Vibration]

Fig. 25 Fatigue damage due to the out of plane vibration of bottom flange of girder in bullet train systems

Fig. 26 Adding plates with rib

Fig. 27 Retrofitting the end of longitudinal rib plates by applying TIG dressing method

2.4.2 Sign poles on highway bridges There are various types of poles on highway bridges such as sign poles, signal poles, and light poles. Fig. 28 shows fatigue cracks developed at the end of rib plate of base plate of sign pole on elevated highway structures. Severe vibration is induced by the passage of vehicles, and combined with the low fatigue resistance of this detail, fatigue cracks penetrate through the whole section of pole. After this accident, a nationwide survey was carried out, and many similar fatigue accidents have been reported. There are two reasons for such vibrations: one is due to traffic, and another is due to wind, i.e. Karman vortex shedding. Fig. 29 shows the newly developed rib detail with high fatigue resistance [26].

Fig. 28 Fatigue cracks at the end of rib plate of base plate of sign pole

Fig. 29 Newly Developed Detail (U-shaped rib)

3. Database of Repair Cases 3.1 Format of summarized repair cases Repair cases of failed steel bridges which are included in this database were collected from technical reports, proceeding, journals, and publications which have been reported in the past [12, 27-60]. Repair cases were summarized using the same format as much as possible so that the user could compare each case. Table 1 provides the format of the summary in these repair cases. Each summary provides for figures and photographs on the bridge design, crack, repair procedures, and so on. These figures and photographs were linked with descriptions in the representative summaries. Pointing to and clicking on the underlined part such as Fig. 30 will automatically show the figure or photograph. In addition to repair cases of failed steel bridges, this database also includes some repair cases for fatigue failures in other structures. These repair cases are also summarized by using the same format as that of steel bridges. After this, repair cases for structures such as machines, ships, ocean structures, pipelines, and so on will be added in this database.
Table 1 Style of Repair Cases Field of application Circumstances of repair Types of structure Details of loading Description of damage Repair method applied

1 2 3 4 5 6

3.2 The composition of database system The homepage address of this system is
Fig. 30 Head of Homepage

u http://iiw.wg5.cv.titech.ac.jp/v .

Table 2 Cause of Fatigue in Steel Bridges

1 Welding defects were included at the time of fabrication. 2 An inappropriate structural detail of low fatigue strength had been adopted. 3 Stresses and deformations unforeseen in design occurred at joints of members. 4 The structure behaved in a manner not expected such as due to vibration. A. B. C. D. E. F. G. H.

Table 3 Repair Method

Removal of crack Re-weld Surface treatments such as TIG dressing and Peening Re-weld + post weld surface treatments Bolted splice Shape improving Stop hole Modification of connection detail

From the address, the user can access the homepage of this web-site (Fig. 30). The causes of fatigue in the steel bridges were classified in four categories as shown in Table 2 [23]. The repair measures are classified into eight types from A to H as shown in Table 3. Repair F or shape improving means to repair the weld part and any nearby part in order to improve the joint itself such as a full penetration weld. On the other hand, repair H corresponds to modifying the connection details where spreads from the joint environment to the whole structure. Therefore, repair H is different from repair F due to the detail of the repair target and the degree of repairing scale. Generally, there are many cases where repair was done by a combination of measures rather than only a single measure. Depending on the repair cases included in this database, stop hole is commonly adopted as an emergency repair with a view to arrest the crack propagation, and there are many cases where permanent repair permits other methods. In such cases, the repair method adopted as a permanent measure is recommended here. At the bottom of this homepage, each repair method for each fatigue failure is evaluated by the authors judgment. The detailed explanations are as follows. a) Repair method for Cause 1 When the cause of failure is a weld defect unintentionally left in welds, the repairing method first supposed is to have sound joints. Therefore, it is effective to repair locally such as repair A -removal of crack, repair B -re-weld and repair D - re-weld + post weld surface treatment. In case the crack propagates in a first member, it is recommended that repair E - bolted splice should be used together with the above repair procedures. Because there will be a higher possiblity that weld defects during fabrication and in the field are left in many weld joints using the same details, it is very effective to modify the connection detail (repair H) and improve the shape (repair F), although these are large-scale undertakings. b) Repair method for Cause 2 There is a higher possiblity that a fatigue crack may re-initiate from the weld if repair for the failure owing to this cause is done to restore the weld joint before cracking. Consequently, two measures should be taken: lower the stress occurring in the weld zone by adding to members.

 increase the fatigue strength of the detail and lower the stress concentration by dealing with the weld zone directly. The former corresponds to repair E and H. On the other hand, the latter which repairs the weld directly, corresponds to the repair method of lessening the stress concentration by finishing the weld toe such as grinding or TIG dressing, i.e., repair C, of which the idea is to strengthen in advance, and repair D, where the idea is to retrofit after cracking. The latter also includes peening, which introduces the compressive residual stresses. c) Repair method for Cause 3 This fatigue failure resulted from secondary stress due to fixed-end restraint moments at the connection of the members, the deformation due to behavior of the structure as a whole and the difference in deflection between the members. Therefore, there are two concepts conceivable in repairing such a failure: increasing stiffness to make possible adequate resistance of the joint against moments. reducing moments. From the viewpoint of the whole bridge structure, repair H, which is to modify the structural detail, is the most effective method. However, repair E, which corresponds to the stress reduction due to increasing the cross section, is recommended if there are structural restrictions. There are a few cases where the localized repair such as repair C or D and the releasing secondary stress such as repair F or G are taken, for example, the failure at the bottom ends of vertical stiffeners due to out-of-plane deformation of the girder web. d) Repair method for Cause 4 For the fatigue failure resulting from the vibration due to vehicle or train passage or strong wind, it is difficult in many cases to restrain such vibration by itself. Consequently, it is effective to take a measure so that the stress occurring from vibration can be reduced. Because only localized repair cannot lessen the stress occurring from vibration, the repaired part will continue to have cyclic response to stress. Therefore, it is effective to reduce the stress according to modification of structural detail (repair H) or in case of structural restriction increasing the cross section (repair E), , or increasing the fatigue strength (repair C or D). From the above description, suggestions for selecting a method of repair and reinforcement are shown in Table 4. The grade of repair methods A to H as shown in Table 4 is subjectively judged by the authors. This evaluation will be improved as we accept many professionals opinions from the comment function of this system. Taking advantage of Table 4, the users can see for each repair case, what action was actually carried out for each cause of fatigue failure. This web site also contains a web page to search for words and phrases within the site contents. By inputting of a keyword, it will be possible to search for the users expectations. This database uses Microsoft Index server as its searching software.

Table 4 Applicability of Combination

A 1 2 3 4 G F F F

B G F F F

D G

E E E E E F: Fair

F G E G G

H G

G G G

E G G

G G G

E E E

E: Excellent

G: Good

N:No good

Fig. 31 Distribution of the repair cases in Japan.

Fig. 32 Distribution of the repair cases in other countries.

Fig. 33 Relationship between years of service and causes.

Use of this function is explained as follows. When searching the repair cases done involving TIG adopted for the Shinkansen train, the user needs to input the appropriate keywords, for example, TIG and Shinkansen or TIG & Shinkansen, or TIG & Shin*. After submitting the search, a list of appropriate cases will be produced. In this function, users can search appropriate cases with differential keywords similar to internet search engines. 3.3 Analysis of Repair Cases It is possible to analyze the cases by making use of the search system. Regarding

reported repair cases of failed steel bridges, the distribution of all the combinations in the Japanese cases in shown in Fig. 31 and all others are shown in Fig. 32. Depending on the number of each case, the tendency of the distribution showed a differential relationship between the cases in Japan and in other countries. Compared with the cases in other countries, the cases in Japan included few cases for fatigue failures occurring from a weld defect. There are many cases repaired by bolted splice for Cause 2 or Cause 4. No cases are repaired only by a stop hole, which is well known as an emergency repair for cracking. However, there are many cases that stop holes were used together with other permanent repair methods. The relationship between years of service until observation of fatigue failure and the cause of fatigue failure is shown in Fig. 33. If the cause of fatigue failure is a weld defect which is Cause-1 shown in Table 2, fatigue cracking is apt to occur in the early years of operation [61]. On the other hand, if the cause of fatigue failure is an inappropriate structural detail of low fatigue strength which corresponds to Cause-2, the occurrence of fatigue cracking concentrates in the range over ten years of service [23]. Fatigue cracking caused by stress and deformation unforeseen in design, which corresponds to Cause-3, the most frequent cause in all the repair cases, has been found regardless of years of service. The occurrence of fatigue failure due to Cause-4 showed the same tendency as Cause-3. However, fatigue failure due to wind vibration occurred in the early years, and the fatigue failures occurring after ten years of service were caused by vibration of the members due to traffic or train passage. Fatigue failures due to this cause were observed at the bottom ends of the web vertical stiffeners in Shinkansen bridge structures and at the base joints of sign poles installed in Japanese highway bridges. It is characteristic that Cause 4 as shown in Table 2 also includes the loss of support function and the fixity of pin joints. The relationship between the years of service and repair methods is shown in Fig. 34. The cases that adopted Repair E(bolt splice) through Repair H(modification of connection detail) are dealt with here. As years of service increase, the number of repair cases performed by repair E - bolt splicing increases because of a permissible method. On the other hand, repair H - modifying the connection detail and repair F - shape improving were adopted extensively despite years of service. In general, modifications of connection detail become large-scale and more difficult. Repair is costly because of performing the measure in advance for the same detail as crack location. In cases where this repair method is adopted to steel bridges where cracking occurred soon after opening, it is suggested that it should be examined if this measure will be appropriate or not and how much this measure will depend on the Total Life Cost of the steel bridge. On the other hand, one needs to consider execution and quality control such as trial testing in the case of repair F shape improving. The relationship between years of service and detection year of fatigue cracking is shown in Fig. 35. The occurrence of fatigue cracking has been rapidly increasing since the first half of the 1970s when many welded girder bridges were opened. Supposing that the repair for

Fig. 34 Relationship between years of service and repair cases.

Fig. 35 Relationship between detection year of cracking and repair method

fatigue failure was done at once after the detection of fatigue cracking, repair E bolt splicing was frequently adopted until the first half of the 1980s. On the other hand, repair H modifying the connection detail and repair D adding surface treatment such as TIG dressing have become popular since the second half of the 1980s. This indicates that the system of the structural analysis progresses and the improvement of fatigue strength for weld joints such as filled weld toe become supported. 4. Conclusions As mentioned at the beginning, in repair of fatigue damage, the basic principles in repair and reinforcement, accurate investigation of causes is always an absolute condition, to eliminate the cause, improve fatigue strength, lower overall stresses of members, lower local stress concentrations, and increase or decrease stiffness at parts. An appropriate method must be selected for each case. This is a difficult problem since design is predicated on structures of the past. Acknowledgement The work described in this paper forms part of the study program of IIW-XIII-WG5. The members of IIW-XIII-WG5 contribute to this paper. We express whom it may concern gratitude.
References 1) Nishimura T., Miki C.: Fracture of steel bridges caused by tensile stress, JSCE, pp.55-64, 1975, 11. (In Japanese) 2) Miki C.: The cases of fatigue damage occurred in welded bridges in the United States, Bridge and Foundation Engineering, vol.16, No.10, pp.18-24, 1982. (In Japanese) 3) Abe H., Taniguchi N., Abe M.: Repairing, reinforcing and fatigue problem in steel railway bridge, Bridge and

Foundation Engineering, vol.17, No.8, pp.24-29, 1983. (In Japanese) 4) Miki C.: Repairing, reinforcing and fatigue problem in foreign countries bridges, Bridge and Foundation Engineering, vol.17, No.8, pp.30-34, 1983. (In Japanese) 5) Nishikawa K.: Repairing, reinforcing and fatigue problem in roadway bridge, Bridge and Foundation Engineering, vol.17, No.8, pp.19-23, 1983. (In Japanese) 6) Subcommittee for investigation of fatigue damage of steel structures, Committee for steel structures: Survey of fatigue damages in steel bridges, Proc. of JSCE, No.368/I-5, pp1-12, 1986.4. (In Japanese) 7) Iwasaki M., Natori T., Fukazawa M., Terada H.: Fatigue cracking in steel bridge members and recommendation of retrofitting methods, Yokogawa Bridge Works, No.18, pp.36-52, 1989. (In Japanese) 8) Subcommittee on fatigue of steel orthotropic deck: Fatigue of orthotropic steel bridge deck, Proc. of JSCE, No.410/I-12, pp.25-36, 1989.10. (In Japanese) 9) Nakai H.: The cases of fatigue failures in city highway bridge, Bridge and Foundation Engineering, vol.24, No.2, pp.23-31, 1990. (In Japanese) 10) Natori T., Asaoka T., Inada I.: Repair practice of steel bridges, Yokogawa Bridge Works, No.21, pp.63-90, 1992.1. (In Japanese) 11) Ichikawa A.: Problems in maintenance of railway steel structures and examples of retrofitting and reinforcement, Bridge and Foundation Engineering, vol.28, No.8, pp.17-21, 1994. (In Japanese) 12) Fisher J.W.: Fatigue and Fracture in Steel Bridges (Case Studies), Wiley- Interscience, 1984. 13) Fatigue Handbook, Japan Road Association, (in Japanese) , 1997. 14) L. P. Schwendeman and A. W. Hedgren,Jr: Bolted Repair of Fractured I-79 Girder, ASCE, ST10 ,Oct., 1978. 15) J. W. Fisher , B. T. Yen and D. Wang: Fatigue of Bridge Structures, A Commentary and Guide for Design, Evaluation and Investigation of Cracking. ATLSS Report No.89-2, July 1989. 16) Fatigue Design Recommendations for Steel Structures, Japanese Society of Steel Construction. 1993 (in Japanese) 17) J. W. Fisher, R. E. Slockbower, H. Hausaman and A. Pense: Long Time Observation of Fatigue Damaged Bridge, J. of Tech., Conunsel, ASCE, 107, April 1981 18) R. B. Mandison, G, R, Irwin : Fracture Analysis of Kings Bridge, Melbourne, Journal of Structural Engineering , ASCE ST97 Sept, 1971 19) Miki C.: Maintaining and Extending the Lifespan of steel Bridges in Japan, IABSE Symposium. San-Francisco, 1995 20) Miki C., Takenouchi H., Mori T and Ohkawa S.: Repair of Fatigue Damage in Cross Bracing Connections in steel Girder Bridge, Structural Eng. / Earthquake Eng. Vol.6, No. 1, 319-395, JSCE, April 1989 21) Tateishi K., Natori T. and Miki C.: Fatigue Damage of Shoes in Plate Girder Bridges and Improvement of Their Details. Structural Eng./ Earthquake Eng., Vol.27, pp.117-176, April, 1994 (in Japanese) 22) Sakamoto K., Ichikawa A., Miki C., and Abe M.: Vibration Fatigue of steel Bridges of the Bullet Train System. IABSE, Workshop. Lausanne 1990. 23) Miki C.: Repairing and reinforcing of fatigue damaged steel bridges, WRC Proceedings IIW, pp. 286-298. 1998. 24) Okumura T. and Ishizawa N.: The Design of Knee Joints for Rigid Steel Frames with Thin Walled Section, Proc. of JSCE , No.153, pp.1-18, 1968. (In Japanese)

25) Morikawa H., Shimozato T., Miki C. and Atsushi I.: Study on Fatigue Cracking In Steel Bridge Piers with Box Section and Temporary Repairing, Proc. of JSCE, 2002.4. 26) Miki C. and Masakazu S.: A Study on U-Shaped Rib Configuration with High Fatigue Resistance, IIW Document No.XIII-1885-01. 27) Miki C., Sakano M., Tateishi K., Fukuoka Y.: Data base for fatigue crackings in steel bridges, Proc. of JSCE, No.392/I9, pp.403-410, 1988. (In Japanese) 28) Demers C.E., Fisher J.W.: FATIGUE CRACKING OF STEEL BRIDGE STRUCTURES, Volume I, U.S.Department of Transportation, FHWA-RD-89-166, 1990. 29) William Wright P.E.: Post-weld treatment of a welded bridge girder by ultrasonic impact treatment, Preliminary Report, F.H.W.A Turner-Fairbank Highway Research Center.6300 Georgetown Pike. McLean, VA22101.1996. 30) Engineering News Record, Aug., 1,1974. 31) Engineering News Record, Sep., 20,1962. 32) Engineering, Sep., 21,1962. 33) Schmerber L., Hofmann u. R.: Die Haseltalbrucke, Der Stahlbau 33, Hsft 6, pp.168-177, 1964. 34) Nather Fr.: Erfahrungen mit der Sanierung und Verstarkung stahlerner Stra enbrucken, Heft 5, pp.44-50, Baukultur 1990. 35) Repairing and reinforcing in roadway bridge: Civil Engineering journal, 36, 5, pp.275-285. 1995 (In Japanese). 36) Horikawa K.: Failures and repair of bridges, Journal of the Japan Welding Society, vol. 52, No.7, pp.13-22, 1983 (In Japanese). 37) Zwerneman F. J.: Fatigue damage to steel bridge. Dia-phragms, J. of Perform. Const. Facil., vol.7, No.4, 1993. 38) Chatterjee S.: Strengthening and refurbishment of Severn Crossing. Part 1: introduction, Pro. Instn Civil Engrs Structs & Bridges, 1992, 94, Feb., 1-5. 39) Flint A. R.: Strengthening and refurbishment of Severn Crossing. Part 2: design, Pro. Instn Civ. Engrs Structs & Bridges, 1992, 94, Feb., 7-22. 40) Cuninghame J.R., Beales C.: Strengthening and refurbishment of Severn Crossing. Part 4:TRRL research on Severn Crossing, Pro. Instn Civ. Engrs Structs & Bldges, 1992, 94, Feb., 37-49. 41) Elentenstain A. G.: The Silver bridge collapse recounted. J. of Perform. Const. Facil., vol. 7, No. 4, 1993. 42) Akesson B., Edlund B., Shen D.: Fatigue cracking in a steel railway bridge, Struct. Engineering International, vol. 7, No 2, pp. 118-120, 1997. 43) Miki C., Mori T., Tuda S., Sakamoto K.: Retrofitting Fatigue-Cracked Joints by TIG Arc Remelting, Proc. of JSCE, No. 380/I-7, pp. 111-119, 1987. (In Japanese). 44) Sakamoto K.: Retrofitting fatigue-cracked joints by TIG arc remelting, Bridge and Foundation Engineering, vol. 28, No.8, pp. 55-58. 1994. (In Japanese). 45) Ito Y., Tsutsumi Y., Nakamura S.: Improvement of fatigue performance of steel bridge 1; Countermeasures and their applicability, Proc. of Annual conference of JSCE, vol. 50, 1995. (In Japanese). 46) Miki C., Sugimoto I., Kaji H., Negishi H., Ito Y.: A study on fatigue strength improvement of flange gusset in existing steel railway bridges, Proc. of JSCE, No. 584/I-42, pp.67-77, 1998. (In Japanese).

47) Nishimura T.: Damages in Pin connected railway truss bridges and rational methods of inspection, repair and improvement, Railway Technical Research Report, No. 483, July, 1965. (In Japanese). 48) Tsurymaki E., Nozawa S., Abe M., Aoki T.: Maintenance of Pin-connected railway truss bridges, Bridge and Foundation Engineering, vol. 27, No.8, pp.68-71, 1993. (In Japanese). 49) Takenouchi H., Tanikura I., Ohtani S., Miki C.: Retrofitting works for fatigue cracks at floor beams of steel truss bridges, Proc. Of JSCE, No.570/I-40, pp.141-151, 1997. (In Japanese). 50) Japan road association: Fatigue of steel bridges, Maruzen CO. LTD., 1997.5. (In Japanese). 51) Hanshin Expressway Public Corporation, Hanshin Expressway Management Technology Center: Repair cases of civil structure in Hanshin Expressway, 1982,3. (In Japanese). 52) Hanshin Expressway Public Corporation, Hanshin Expressway Management Technology Center: Repair cases of civil structure in Hanshin Expressway-II-, 1987,3. (In Japanese). 53) Hanshin Expressway Public Corporation, Hanshin Expressway Management Technology Center: Repair cases of civil structure in Hanshin Expressway-III-, 1991,3. (In Japanese). 54) Yamaguchi T., Shiraki B., Umemura N., Tanaka H.: Vibration caused by Karmans vortex on bridge members and its countermeasure, Mitsubishi Juko Giho, vol.7, No.7, pp.122-135, 1970. (In Japanese). 55) Haneoka J., Nagamuna T., Maruyama S., Yonekura T.: Actual stress of intersection between cross beam lower flange and girder web in steel highway bridge, Proc. Of Annual Conference of JSCE, vol.49, 1994. (In Japanese). 56) Natsumoto M., Tanikura I., Okawa S., Nanasaki Y.: The effect of retrofitting work to the connecting corner between stiffening girder and vertical member of arch bridge, Proc. of Annual Conference of JSCE, vol.51, 1996. (In Japanese). 57) Takada H., Mizumoto M.: Report on strengthening construction of Sozuka-Ohhashi, Yokogawa Bridge Works, No.19, pp.126-136, 1990. (In Japanese). 58) Narita N.: Vibration of bridge members due to wind, Bridge and Foundation Engineering, vol.5, No.9, pp.1-5, 1971.9. (In Japanese). 59) IIW-XIII-WG5-12-97. 60) Hargrave R. E., Common failure mechanisms in BOF hoods, Iron and Steel Engineer, pp. 22-28. Nov. 1996. 61) Kratena J., Hrejci H., Collapse of a warehouse and its repair, Struct. Eng. International, vol.5, No.3, 1995.