NIPPON STEEL TECHNICAL REPORT No. 87 JANUARY 2003

UDC 624 . 2 Study on Connection of Crossbeams in Rationalized Plate-Girder Bridges

Nobuaki SAKURAI*1 Kouichi NAKAMURA*1 Atsuo OOTAKE*1 Yuuzou OKAMOTO*1

Abstract In Japan, many rationalized plate girders are constructed for the new Toumei/ Meishin Expressway Project and for others. They reduce the number of main gird- ers, using pre-stressed-concrete-slabs. The bending moment of the slabs, accord- ingly, is increased. Additionally, their reductions of diagonals and laterals cause stress concentration on the connection of cross-beams. Nippon Steel pursued ana- lytic studies and experiments on real bridges, in several projects contracted with Japan Highway Public Corp et al.

garding the design of the joint yet. 1. Introduction Since 1999 Nippon Steel Corporation has been awarded construc- In rationalized plate-girder bridges (see Fig. 1), the span of floor tion orders of rationalized plate-girder bridges, from Takahari Via- slab support is made longer than conventional bridges (see Fig. 2) duct of Higashi-Meihan Expressway to Sakae Viaduct, which is now thanks to the application of pre-stressed concrete floor slabs and other in the fabrication stage. During the period, the company has focused technologies, which leads to a larger bending moment in the direc- attention on the issue of the joint design, and through analyses, meas- tion of the floor span. Further, since lateral and diagonal bracings urements on real bridges and so forth, carried out a series of studies are omitted, a floor slab works as a main load distribution member regarding the design of the connection of a crossbeam with a main not only for normal traffic loads but also for lateral loads caused by girder in the rationalized plate-girder bridges. This paper reports the winds and earthquakes. results of the studies. For this reason, the joint of a girder with a floor slab is important; Japan Highway Public Corp., in its design guideline, instructs con- 2. Past Studies structors to verify the tensile force on the joint in the design of shear- Among various study reports, Ohgaki et al. presented a paper on connectors, even in the case of a non-composite girder1). However, analyses and tests2,3), paying attention to the studs and vertical stiff- although various study reports have been presented from various re- eners at the connection of the crossbeam, and there is a report of a search/design institutes, no unified philosophy has been formed re- loading test carried out on Hibakaridaira Bridge4). The results of these reports are essentially as follows. á When studs are arranged immediately above a vertical stiffener to which an intermediate crossbeam (a crossbeam between piers) is connected, a large axial force is imposed on the studs and stress concentration occurs at the upper end of the vertical stiffener, and as a result, fatigue problems may occur. á When studs are not arranged immediately above such a vertical Fig. 1 Rationalized plate-girder Fig. 2 Conventional bridge stiffener, the axial force on the studs is relaxed, but on the other bridge hand, a large bending stress is imposed on the upper flange of the

*1 Civil Engineering & Marine Construction Division

- 75 - NIPPON STEEL TECHNICAL REPORT No. 87 JANUARY 2003

main girder and what is more, separation of the floor slab from the the above analysis. The bending and shearing composite stress on main girder or cracks between them may take place when the floor the studs was verified referring to the study of Hiraki et al.5) slab cannot follow deformation of the upper flange. á An X-shape section comprising a triangle rib, a main girder web Observations obtained through other studies are basically the and a vertical stiffener was examined as a member subjected to an same, despite the fact that numerical data are different to some ex- axial force and a bending force. Also examined was the bearing tent depending on conditions, such as the floor slab construction stress of the floor slab to which the X-shape section was to be sub- method (pre-cast or cast-in-place) and its structure (composite or jected. non-composite). The construction work of the Viaduct was completed in June 2002. No cracks have been found near the upper end of the vertical stiffen- 3. Nippon Steel’s Activities ers at the connections of crossbeams, and the structure is performing In the above situation, Nippon Steel took measures and conducted well. analyses and loading tests on real bridges as described hereinafter. 3.2 Nobuno viaduct 3.1 Takabari viaduct Nobuno viaduct is a continuous non-composite two plate-girder Takabari Viaduct is a continuous non-composite two plate-girder bridge having cast-in-place PC floor slabs, located in the western bridge having pre-cast PC floor slabs, located in an eastern suburb of part of Shimane Prefecture forming a part of Cross-Chugoku Ex- Nagoya forming a part of Higashi-Meihan Expressway (between pressway, and its order was placed by Matsue Construction Office of Osaka and Nagoya), and its order was placed by Higashi-Nagoya Japan Highway Public Corp. In this bridge, the triangle ribs intro- Construction Office of Japan Highway Public Corp. In this bridge, duced in Takahari Viaduct were arranged also on both the sides of studs were arranged immediately above a vertical stiffener to which the vertical stiffeners for the purpose of relaxing the bending stress an intermediate crossbeam was connected and, in addition, for the on the main girder upper flange (see Fig. 5). Usefulness of these purpose of relaxing the stress on the vertical stiffener, stiffening ribs triangle ribs was confirmed through an analysis by the three-dimen- called triangle ribs (see Fig. 3) were fitted to the upper flange and sional finite element method (FEM), and the result of the analysis web of main girders. The design of the portion is explained below. was verified through measurements on the real bridge. á An analysis was made on a frame (rahmen) model consisting of 3.2.1 Analysis by three-dimensional FEM main girders, a floor slab and crossbeams on two cases: Case-1 in The model for the three-dimensional FEM analysis was defined which live loads were imposed on the extended portions of a floor as shown in Figs. 6 and 7 to cover the entire bridge. The floor slab, slab and Case-2 in which live loads were imposed on the center of main girder webs, their flanges, crossbeam above a pier, vertical stiff- the floor slab support span (see Fig. 4). eners and triangle ribs were regarded as shell elements in the model, á The arrangement and number of the studs were determined based and all the other members as beam elements. Two cases, with and on the stress resultant (M: bending moment, S: shearing force, N: without the triangle ribs, were analyzed so as to evaluate their ef- axial force) at the upper end of the main girder web obtained from fects, and these two cases were combined with a case with a dead 55 55 200 55 55 load and another with a live load. Figs. 8 to 11 show the results of the analysis. The figures show the force in the axial direction of the studs, from the position of an Stud

250 intermediate crossbeam (shown as 0 m) to the center of the distance Upper flange to an adjacent crossbeam. The analysis made it clear that the tri- 350 23 300 50 Vertical stiffener

Triangle rib Triangle rib

350 16 350 (mm) 370 320 Fig. 3 Connection of intermediate crossbeam Outside face of (mm) Inside face of main girder main girder 1) Dead load + Live load (Case 1) Concrete barrier curb P PPPPaving Concrete C24 C24 Noise barrier Haunch Haunch barrier curb Floor slab Fig. 5 Detail of upper end of vertical stiffener at crossbeam connection Floor slab (shell element) 12 4 5678910 11 12 13 14 1618 3 1517 Member 101Member Member 102

310 Upper flange dummy member (beam element) 95 19 20 Upper flange (beam element) 100kN × 1.333 = 133.3kN Web (shell element) 3,000 3,000 250 2) Dead load + Live load (Case 2) Concrete barrier Lower flange (beam element) curb PPPaving P P Concrete barrier curb Noise barrie HaunchFloor slab Haunch Upper flange (beam element) Web (shell element) 12 4 5678910 11 12 13 14 1618 Crossbeam near vertical stiffener (stiff member) 3 1517 Intermediate crossbeam (beam element) 101Member Member 102 Crossbeam near vertical stiffener (stiff member) 19 20 Lower flange (beam element) 100kN × 1.333 = 133.3kN Fig. 4 Frame (rahmen) model Fig. 6 Analysis model (1)

- 76 - NIPPON STEEL TECHNICAL REPORT No. 87 JANUARY 2003 Inc: 0 Time: 0.000e+00

Z X Y

rib-death.dat 1 Fig. 7 Analysis model (2)

50 48.52 Between girders 40 Center 30 Outside girders Fig. 12Stress on vertical stiffener (without triangle ribs, under dead 20 load) 10 6.26 Ð0.68 2.46 1.69 0.61 0 Ð0.95 Ð0.98 Ð5.57 Ð0.25 Ð7.72 Ð11.85 Ð6.47 Ð10 Ð9.2 Ð17.31 Ð14.35 Axial force (kN) Ð20 Ð24.29 Ð19.99 Ð30 Ð40 Ð50 012345 Distance from crossbeam (m) Fig. 8 Axial force on studs (without triangle ribs, under dead load) 50 Between girders 40 34.91 Center 30 Outside girders 20

10 5.01 1.41 1.32 0.88 Ð5.65 0 Ð1.7 Ð5.39 Ð0.74 Ð6.83 Ð2.77 Ð10.54 Ð10 Ð6.45 Ð12.46 Ð13.34 Ð9.47 Fig. 13 Stress on vertical stiffener (with triangle ribs, under dead load)

Axial force (kN) Ð20 Ð19.77 Ð19.9 Ð30 angle ribs reduced the force in the axial direction of studs and the Ð40 fluctuation of the axial force by 30% or so. Figs. 12 and 13 show Ð50 012345 the stress condition at the upper end of a vertical stiffener under a Distance from crossbeam (m) dead load. It is clear in the figures that stress concentration was Fig. 9 Axial force on studs (with triangle ribs, under dead load) relaxed by about 20%. 50 3.2.2 Measurements on real bridge Between girders 40 Center For verifying the validity of the above analysis, measurements 30 Outside girders were carried out on Nobuno Viaduct using a roughter crane corre- 20 sponding to the live load B specified in the roadway bridge specifi- 10 cation. The measurements were done automatically in every 30 min. Ð0.2 Ð0.34 0.72 0 Ð1.09 Ð0.52 Ð1.49 Ð2.86 0 Ð1.14 Ð1.52 Ð4.64 using a personal computer for a four-month period from before the Ð10 Ð3.19 Ð1.44 Ð1.82 Ð5.28 Ð5.33 Ð11.98 concrete casting of the floor slabs to the casting of concrete barrier Axial force (kN) Ð20 Ð19.74 curbs and loading tests (from late June to early November), focusing Ð30 on the axial force on the studs, the stress concentration at the upper Ð40 end of vertical stiffeners, the bending stress on the upper flanges and Ð50 012345 so on. Distance from crossbeam (m) As a typical example of the measurements, Fig. 14 shows the Fig. 10 Axial force on studs (without triangle ribs, under live load) axial force on the studs under a wheel load imposed from above a 50 Between girders crossbeam connection. Compared with Fig. 11, which shows the 40 Center analysis result under the same loading condition, the tensile axial 30 Outside girders force on the studs actually measured was about one half that of the 20 analysis. This is presumably because, while the analysis assumed 10 3.61 0.86 0.63 Ð0.63 Ð0.59 Ð1.42 0 that studs alone were responsible for the load transmission from a Ð3.21 Ð0.24 Ð1.42 Ð1.55 Ð4.1 Ð10 Ð4.02 Ð1.2 Ð1.52 Ð5.58 Ð5.54 floor slab, in the actual load transmission bearing stress, adhesion of Ð14.07

Axial force (kN) Ð20 Ð15.93 concrete and other factors were involved. For the same reason, the Ð30 measured stress at the upper end of a vertical stiffener was roughly Ð40 1/3 that of the analysis. Ð50 012 345 As a result, it was confirmed that the measures taken in the de- Distance from crossbeam (m) sign of Nobuno Viaduct were sufficient against tensile force caused Fig. 11 Axial force on studs (with triangle ribs, under live load)

- 77 - NIPPON STEEL TECHNICAL REPORT No. 87 JANUARY 2003

50 of 100 years, it is necessary to study the issue of the tensile stress on Between girders 40 Center studs in consideration not only of live loads but also of temperature 30 Outside girders changes. 20 Fig. 17 shows the bending stress on the upper flange of a main 10 0.34 0.11 0.07 0.64 girder. The measured stress in the portion around an intermediate Ð0.05 0.03 0.15 0.64 0 0.11 2 Ð0.89 Ð0.51 Ð0.08 Ð0.05 Ð1.73 Ð0.12 crossbeam connection was roughly 10 N/mm and the stress ampli- Ð10 Ð2.57 2 Ð3.49 tude was also about 10 N/mm , showing that the measures taken in Axial force (kN) Ð20 Ð7.43 the design of the viaduct were sufficient. However, considering the Ð30 2 Ð40 fact that a bending stress of roughly 35 N/mm occurred on a flange Ð50 at the center between two crossbeams during the concrete casting of 0 1 2 3 4 5 a floor slab, attention must be paid in the case that composite girders Distance from crossbeam (m) are used. Fig. 14 Axial force on studs (actual measurement under live load) 3.3 Matsudate viaduct by live loads. Watching the aging with the passage of time in the Matsudate Viaduct is a continuous non-composite two plate-girder period of 2 months or so after the concrete casting of the floor slabs bridge having pre-cast PC floor slabs, and its order was placed by (see Figs. 15 and 16), however, it became clear that temperature Aomori Construction Office of the Ministry of Land, Infrastructure changes caused an axial force and a bending force on the studs im- and Transport. There is a study report on the relationship between a mediately above an intermediate crossbeam connection and they vertical stiffener to which a crossbeam is connected and cracks of formed a tensile force amounting to an allowable stress, and that the concrete at haunch portions of a rationalized plate-girder bridge hav- stress amplitude was very large. Therefore, in view of a service life ing cast-in-place PC floor slabs6). The conclusion of the study is that, for suppressing the force to cause concrete cracking, it is desir- Stress Secular change of axial stress on studs (above crossbeam connection) Temperature Concrete casting Tensioning of PC Concrete casting Removal of able to increase the width of the vertical stiffeners to somewhere of floor slabs of barrier curb concrete forms Inner side Above web Outer side close to the flange width. In view of this, the authors carried out a study using the analysis by the two-dimensional FEM, for the pur- pose of confirming if the same thing occurred in the case of a pre- cast floor slab having adjustment mortar and a different shape of the haunch and determining an optimum width of vertical stiffeners. The result of the analysis is as follows. Fig. 18 shows the stress distribution when the width of vertical stiffeners is small and Fig. 19 the same when the width is increased to the flange width. Whereas Time (day) laminar tensile stress distribution is seen in the joint of the upper Fig. 15 Secular change of axial stress on studs (above crossbeam flange with the floor slab in Fig. 18, it is confirmed in Fig. 19 that the connection) occurrence of the tensile stress can be suppressed. The tensile stress Secular change of axial stress on studs (above crossbeam connection) Temperature in the above is considered to cause the upper flange to separate from Concrete casting Tensioning of PC Concrete casting Removal of Inner side of floor slabs of barrier curb concrete forms Above web the adjustment mortar, leading to a problem of corrosion in a long Outer side

Time (day)

Fig. 16 Secular change of bending stress on studs (above crossbeam connection)

Fig. 18 Stress on vertical stiffener (narrow width) 0.0 m from crossbeam 0.2 m from crossbeam 1.0 m from crossbeam 4.0 m from crossbeam

Stress

Temperature

Jul. 2001 Aug. 2001 Sep. 2001 Oct. 2001 Time (day) Fig. 17 Bending stress on flange Fig. 19 Stress on vertical stiffener (increased width)

- 78 - NIPPON STEEL TECHNICAL REPORT No. 87 JANUARY 2003

studs and vertical stiffener. According to an analysis by FEM re- garding another bridge project, under a certain loading condition, a stress amounting to roughly 410 N/mm2 is formed in the area above Posiotion of a vertical stiffener. In Sakae Viaduct, however, no reinforcement is intermediate crossbeam 1,800 ＝ provided on the bases that an analysis by FEM tends to yield larger 200

＠ stress figures than is likely to actually occur, that non-composite gird- 500 500 9

Upper flange ers are used in the viaduct, and that the design presupposes the most unfavorable situation little likely to actually take place in terms of loading conditions. Stud 187.5 375 187.5 50 50 4. Conclusion Fig. 20 Stud arrangement Analytical and experimental studies related to the design of crossbeam connections of several rationalized plate-girder bridges term. were outlined in this paper. At present, since the specification re- It was confirmed, based on the above, that for inhibiting the sepa- garding the tensile fatigue endurance of studs is not clearly set forth, ration of the upper flange from the adjustment mortar, it was effec- it is considered desirable that the studs be arranged avoiding the po- tive, in the case of pre-cast PC floor slabs, to increase the width of sition just above a crossbeam connection. With regard to the bend- the vertical stiffeners at the crossbeam connections to somewhere ing deformation of the upper flange of a main girder, which leads to near the width of the flange, although the cracking force was not as the separation of a floor slab from the main girder, it is considered large as in the case of cast-in-place floor slabs. necessary to give a certain extent of stiffness to the flange or provide 3.4 Sakae viaduct reinforcement at adequate portions, especially in the case of a com- Sakae viaduct is a continuous two plate-girder bridge of inte- posite girder. Any contribution to increasing the durability of ratio- grated super- and sub-structures having cast-in-place PC floor slabs, nalized plate-girder bridges would be a source of great pride for forming a part of Outer Circular Expressway around Tokyo, and its Nippon Steel. order was placed by Saitama Construction Office of Japan Highway References Public Corp. The design of the crossbeam connections of the via- 1) Japan Highway Public Corp.: Bridge Construction Section, Design Guide- line, Book 2. Jul. 1998 duct, which is presently in the design stage, is explained hereafter as 2) Ohgaki, K., Kawaguchi, Y., Yabe, J., Nagai, M.: Design Method for Shear- an example of the latest design trend of this kind of connections of a Connectors of Continuous Composite Two Plate-Girder Bridges. Steel rationalized plate-girder bridge. Construction Engineering. 4(15), Sep. 1997 In view of past experiences and study results in other bridge 3) Yabe, J., Yamamoto, A., Ohgaki, K., Saito, H.: Simple Analysis Method projects, for reducing the axial force on the studs and preventing for the Stresses of Shear-Connector of Continuous Composite Two Plate- Girder Bridges. Kawasaki Heavy Industries Technical Report. (139), floor slabs from separating from the flanges, the following philoso- (1998) phy was adopted in the stud arrangement of the viaduct (see Fig. 4) Highway Technology Center: New Tomei Expressway - Investigation 20). and Study of Latest Steel Bridge Technologies, A Follow-up Study of á No studs are arranged just above an intermediate crossbeam con- Draft Design and Construction Guideline of Rationalized Plate-Girder Bridges. Mar. 2000 nection. 5) Matsui, S., Hiraki, H., Miyoshi, E.: Latest Approval of Headed Studs in á The outermost row of studs is 50 mm from an edge of the flange, West Germany and Calculation Examples. Brides and Foundations. (Sep. and the center row is as close to above the web as possible. The 1986) center row studs are arranged exactly above the web particularly in 6) Highway Technology Center: New Tomei Expressway - Investigation the range within 500 mm from a crossbeam connection7). and Study Report on Design and Construction of Rationalized Plate-Girder Steel Bridges (Part 3). Mar. 1997 In a design wherein studs are arranged not on the centerline of a 7) Highway Technology Center: Design and Construction Manual of Two vertical stiffener but on both the sides of it, local bending force is Continuous Steel Composite Plate-Girder Bridges Having PC Floor Slabs. imposed on the flange owing to the restriction between the floor slab, Mar. 2002

- 79 -