Renovation techniques for rib to deckplate fatigue cracking in orthotropic bridge decks

Steel orthotropic bridge decks possess 5 fatigue critical points. One of these is the longitudinal rib to deckplate weld. At that connection, a combination of large transverse bending stresses caused by concentrated wheel loads and residual stresses in local plasticized zones due to the welding procedure may initiate fatigue cracking at the weld root or toe. For new bridges, the normative documents provide minimum plate thicknesses, in combination with maximum rib spacings in order to avoid fatigue problems. However, a large number of these decks have been built around the world since the 1950’s, which do not comply to these recommendations, many of them showing fatigue cracks. For these bridges, renovation solutions are required since the deck system cannot be separated from the main structural system. A first part of this paper reviews the current state of the art of renovation techniques for poorly designed rib to deckplate connections including peening, stress relieving and overlaying. In a second part the results of a full scale fatigue test carried out at Ghent University College are presented and discussed. In this test, an epoxy bonded connection between the original steel plate and a repair plate is tested under fatigue loading conditions at 5000000 cycles. The test indicates no signs of crack initiation or debonding, confirming this method as a valid alternative.


INTRODUCTION
Steel orthotropic bridge decks contain at least 5 fatigue critical locations (Figure 1) (Uchida et al., 2008).Among these, the longitudinal rib to deckplate weld is one of the most important.At that connection, a combination of large transverse bending stresses caused by concentrated wheel loads and residual stresses in local plasticized zones due to the welding procedure may initiate fatigue cracking at the weld root or toe.Four crack types may occur 1) cracking from the weld root into the deckplate, 2) from the weld root into the weld, 3) from the upper weld toe into the deckplate, 4) and from the lower weld toe into the rib (Sim et al., 2009;Xiao et al., 2008;Ya et al., 2008).From the late 1970's up to today many cases of fatigue cracking at this location have been described.The main reasons for these many faulty bridge decks are a lack of understanding of the fatigue problem at the time of design and construction, the increase of traffic in the last 40 years, and the more intensive use of wide base tyres with increased internal pressure and edge stiffness.
Eurocode 1993-1-9 defines fatigue detail categories for the rib to deckplate connection, based on tests carried out in the 1980's (Figure 2) (CEN, 2007).These detail categories are rather conservative of nature, especially for the partial penetration welds as commonly used in modern practice.In addition, these fatigue classes cover only fatigue cracking in the rib.Fatigue cracks propagating into the deckplate are not considered.However, many such cases have been found in practice, and there is need for a fatigue detail for this type of fatigue failure as well.Impressive work in this field has been done by Kolstein, who has thoroughly analysed huge amounts of fatigue testing data from the 1980's, 1990's, and early 2000's (Kolstein, 2007)).His conclusions have lead to a comprehensive and more detailed proposal for classifycation of rib to deckplate weld details (Figure 3).These details include a category for deck plate cracking (∆σ c = 125 MPa), a category for rib cracking when using fillet

MATERIALS AND METHODS
Avoiding future fatigue cracking is possible in two ways.The first is the use of relief techniques of various kinds in order to improve the effective fatigue resistance of the connection above the values indicated in Figure 3.The second is the introduction of additional stiffness at the connection, in order to reduce the bending stresses.This can be done by increasing the plate thicknesses for new decks in combination with appropriate welding types, or for existing decks, using a suitable renovation technique.

Relief techniques
Possible relief techniques for welded connections of orthotropic decks have been described in detail by Qian and Abruzzese, and are summarized in Figure 4 (Qian and Abruzzese, 2009).Of these, Figure 4. Fatigue relief techniques for the rib to deckplate weld (Qian and Abruzzese, 2009).shot peening, fluid bed peening, and ultrasonic impact treatment, have been tested on orthotropic deck applications, and favorable fatigue results have been reported.However, practical applications remain very limited.

Renovation techniques
Possible renovation techniques for rib to deckplate connections of orthotropic bridge decks have been investigated in detail by De Jong (2006).These techniques should allow for an addi-tional life time of at least 25 years.Of the 13 mentioned renovation techniques, 3 have been considered for further investigation.These are: a reinforced concrete overlay, a rib injection technique, and the addition of a second steel plate bonded to the existing layer.The first solution consists of a reinforced high performance concrete overlay (RHPC).Reduction of stresses of 60% in the ribs and 80% in the deckplate are reported (De Jong, 2006).The major challenges are the shear resistance of the bonding interlayer, the concrete quality, reinforcement corrosion, and the correct application.Practical applications have been realized with this technique in the Netherlands with alternating success, especially due to the high reinforcement quantities.An alternative, using steel fiber reinforced concrete (SFRC) has been reported by Murakoshi (2008).
However, the results of this research are inconclusive at the moment.Thin (20 mm) reinforced high performance concrete overlays (TRHPC), suitable for moveable bridges, have been reported by Boeters et al. (2009).In such conditions, concrete cover is insufficient, and CFRP or stainless steel reinforcement becomes necessary.Although some practical problems have been mentioned, good overall performance has been reported.The second solution, rib injection, has been investigated by De Jong (2006).However, although substantial stress reduction was achieved, poor fatigue resistance was observed.This technique is, according to De Jong, not suitable for orthotropic deck renovation.The third solution consists of adding a second steel plate on top of the existing deck plate, using either a thin epoxy adhesive layer (bonded steel plate : BSP) or a thick PUR adhesive layer (sandwich steel plate : SSP).
In both solutions, the adhesive material is polymer based, exhibiting a temperature, strain velocity, and pressure based behavior.However, in the BSP solution, this influence is negligible, whereas the influence is important for the SSP solution (De Jong, 2006;de Freitas et al., 2010).Based on these data, it was concluded that BSP is a technique suitable for further study.It is especially important to know whether the achieved reduction of transverse bending stresses in rib and deckplate, which will relieve these connections from fatigue cracking, will remain present under fatigue loading, since the bonding layer itself may fail as well.In order to investigate this, a BSP renovation technique was tested at 5 million cycles on a test specimen at Ghent University College.In the subsequent paragraphs, the test setup and the test results will be presented.

Original specimen
The test panel comprises of a 12 mm deckplate measuring 1800 by 2000 mm (Figure 5).There are 3 trapezoidal 200 mm high 5 mm thick ribs with a 100 mm lower rib spaced 600 mm apart.The ribs are supported at midspan by a 500 mm high and 10 mm thick floor beam.The steel quality is S355.Figure 6 shows the test setup in the laboratory at Ghent University College, together with the location of the strain gauges, placed transversely to the longitudinal ribs on the middle rib, longitudinally at 330 mm from the floor beam, except for strain gauge number 5 placed longitudinally on the middle rib at the floor beam location.Details of the gauge locations respective to the weld toes are given in Figure 7.These locations are symmetrical to the middle rib axis.For   comparison purposes, calculations are done by a full 3D linear FE analysis (Figure 8).The model uses Mindlin type shell elements of adequate thickness.By interpolation and extrapolation, the FE stress values are related to the strain gauge positions, as well as to the weld toe and root positions.
During static testing, a load of 50 kN was applied by hydraulic jack on a 250 by 250 mm loading plate.This load corresponds to a pressure of 0.8 MPa, provided the load is evenly distributed over the contact surface.This last condition is very difficult to achieve, and requires introduction of a polychloroprene interlayer between the load surface and the deckplate.As an example, Figure 9 shows stress values, as calculated from measured strain values, for strain gauge number 4 (Figure 6), for different interlayer thicknesses, in relation to the FE calculated values.The results show that, considerably lower measured strains can be found when applying a too shallow interlayer (5 mm).This effect was already mentioned by De Corte, ( 2005) and De Jong (2006), but is overlooked in many other contributions.This uneven distribution is caused by the difference in flexural stiffness between the loading plate and the deckplate, attracting more load to the rib to deckplate joint, which can be considered an elastic support.Using loading conditions with loads extending over more than one rib to deckplate joint, may only increase this phenomenon.The results of all measurements, for varying transverse load locations, at all strain gauges are extrapolated to the critical points at toe and root, and presented in Table 1.

Renovated specimen
As a repair solution, 2 steel plates measuring 600 x 300 x 6 mm have been bonded to the steel plate, directly above the instrumented rib (Figure 6).The choice for 2 plates placed longitudinally one after another results in the introduction of a loaded splice, a location pointed out as a source of premature failure by De Jong (De Jong, 2006).The different steps during this bonding procedure are 1) sanding the top surface, 2) applying the epoxy adhesive, 3) installation of the top plates, and 4) applying pressure during curing.The procedure and the result are shown in Figure 10.This repair solution (BSP) requires the use of a thin epoxy coat.For this purpose PC5800/BL epoxy is used.The main properties are given in Table 2. Figure 11 to 19 show stress values, as calculated from measured strain values for strain gauges 1 to 9, for both the original and the renovated situation.Dynamic tests with an amplitude of 50 kN (F min = 5 kN /F max = 55 kN) at a frequency of 3.4 Hz, corresponding to the maximum frequency of the dynamic actuator relative to the occurring stroke were carried out on the renovated orthotropic deck at load location 90 (Figure 9).This location corresponds to a load applied transversely, symmetrical to the longitudinal axis of the instrumented  rib.During these tests, at predetermined intervals, the bonding quality was tested with ultrasonic testing equipment.In this method, the wave speed between upper and lower surface of the sandwich plate is measured.Debonding should translate into a sharp decrease of the wave speed.As can be seen from Figure 20, debonding was not observed during the tests.This result was confirmed by the strain results remaining constant at all locations.

RESULTS AND DISCUSSION
The extrapolated results from Table 1 are compared to the maximum allowable stress cycles according to Eurocode 3.1.9(CEN, 2007).However, the fatigue detail categories from Eurocode 3.1.9represent values for 2 million cycles.For bridge with substantial numbers of heavy traffic, the number of cycles to be taken into account largely exceeds this value of 2 million.A calculation based on fatigue load model 4 of Eurocode 1.2 proves that for most situations with heavy traffic, the effective number of loads, from a cumulative damage calculation with Miner's rule, exceeds 100 million.
Consequently, the detail can be designed, based on the variable amplitude fatigue limit (VAFL), which corresponds to 0.405∆σ c .Table 1 proves that for most transverse load positions, fatigue life will not reach the design life of the deck, even when considering Kolstein's detail category proposal.This is as expected, since a 12 mm deckplate in combination with 5 mm ribs spaced 300 mm has indeed proven to be inadequate for heavily trafficked situation in practice.
When analyzing the results from Figure 11 to 19, one can observe that, both the maximum compressive, as well as tensile values are reduced.Remarkably, the locations of maximum stress before and after renovation do not coincide anymore.When analyzing the stress reduction of the original maximum, reductions of 70% can be found.As explained, the load position for the new maximum does not coincide with the old location.Obviously, the strain result from gauge number 5 should not -or very little -be influenced by the renovation technique.It acts therefore as a reference gauge.In addition, the results for the various gauges correspond rather well with those calculated with the FE model, after introduction of the bonded plate as a plate of increased thickness.Only in a rare number of situations, stresses exceeding the

Stress [MPa]
Load location relative to specimen edge [cm] Strain gauge n°2 without plate with plate     Figure 18.Stress results from strain gauge 8 (before and after renovation).
variable amplitude fatigue limit are discovered.Finally, from Figure 20 it may be concluded that the BSP repair solution, remains intact up to 5 million cycles, which is the constant amplitude fatigue limit for steel structures.BSP is therefore a viable alternative for renovation of poorly designed orthotropic plated bridge decks.

Conclusion
After a review of the current state of the art of renovation techniques for poorly designed rib to deckplate connection of orthotropic plated bridge decks, the static and dynamic effects of a bonded steel plate (BSP) solution were analyzed.Static tests revealed that indeed large stress reductions can be achieved, although it is important to notice that the load positions corresponding to maximum stresses before and after renovation do not coincide.A fatigue test to 5 million cycles revealed no structural failure, neither in the rib to deckplate welds, neither in the bond layer.It can therefore be concluded that the bonded steel plate solution is a viable renovation alternative for poorly designed orthotropic plated bridge decks.Nevertheless, more practical research is needed into the on-site application parameters which may differ from the laboratory conditions on which the conclusions from this paper are based.Cycles: 0 100,000 650,000 1,750,000 2,900,000 3,250,000 3,771,000 5,000,000 Wave speed (m/s)

Figure 6 .
Figure 6.Experimental setup and strain gauge locations.

Figure 11 .
Figure 11.Stress results from strain gauge 1 (before and after renovation).

Figure 12 .
Figure 12.Stress results from strain gauge 2 (before and after renovation).

Figure 14 .
Figure 14.Stress results from strain gauge 4 (before and after renovation).

Figure 17 .
Figure 17.Stress results from strain gauge 7 (before and after renovation).

Table 1 .
Comparison of measured stress values, extrapolated to critical points, to variable amplitude fatigue limit values.
position Left

of rib Right of rib Deckplate Rib Deckplate Rib Weld toe Weld root Weld toe Weld toe Weld root Weld toe
Bold values indicate that the ∆σ< 0.405.∆σccriterion is not met