David Darwin, University of Kansas
Research dating back over 25 years has established the key factors that control bridge deck cracking – age, bridge deck type, concrete material properties, site conditions, curing, and even date of construction. An understanding of these factors has been put to good use in a two-phase pooled-fund study under the direction of the Kansas Department of Transportation in conjunction with 18 other state departments of transportation and the Federal Highway Administration.
Evaluation of over 150 bridge decks, most supported by steel girders, has demonstrated that even in the best preforming bridges, crack density will increase to some degree with increasing age. The general observation, however, is that those decks that perform well during the first three years after construction will perform well throughout the life of the deck. Monolithic decks tend to perform much better than decks that are constructed with overlays that are placed as part of initial construction, largely because cracks in the subdeck tend to reflect through the overlays and shrinkage in the overlays tends to be restrained by the subdeck, resulting in additional cracking in the overlay.
The choice of concrete mixture proportions has a large impact on cracking performance, in some cases in unexpected ways. Mixtures with higher volumes of cement paste (that is, more water, more cement, or the combination of the two) exhibit greater drying shrinkage, which results in greater transverse cracking due to restraint provided by the girders. This restraint tends to be greater for steel-girder bridges than for precast, prestressed girder bridges, although the latter can result in increased cracking if camber is not controlled and continues to increase over time. Increasing concrete slump leads to increased settlement cracking, the principal reason that transverse cracks form directly above and parallel to the top reinforcing steel. Increasing air content tends to reduce cracking because entrained air acts as a workability agent and air bubbles do not shrink. As somewhat of a surprise to many in the field, increasing compressive strength correlates with increased cracking. This has been observed in a number of states, where higher-strength concretes, whether used for early strength or to reduce permeability, correlate with increased cracking. This increase in cracking results because high-strength concrete creeps less than lower-strength concrete, even at the same ratio of stress to strength. Reduced creep is a positive property for high-strength concrete in compression, such as used in columns in high-rise buildings, but reduced creep tends to limit the relief of tensile stresses in bridge decks, resulting in increased cracking.
Extra finishing of the deck can lead to increased plastic and drying shrinkage cracking as it works the coarse aggregate below the surface while increasing the thickness of the (high shrinkage) paste at the surface, while at the same time delaying the initiation of curing.
Site conditions that lead to rapid evaporation of bleed water from the surface of bridge decks or concrete temperatures that exceed those the bridge girders at the time of placement, result in, respectively, increased plastic shrinkage and thermal cracking.
One of the most interesting aspects of the study of older bridge decks involves the observation that decks that were cast 30 years ago exhibit less cracking than those that were cast 10 or even 5 years ago. The principal changes in construction over that period have involved the use of more finely ground cement, higher-slump concrete, and the switch from buckets and conveyor belts to pumps as the principal method for placing concrete. The more finely ground the cement, the smaller the pores in the hardened cement paste within concrete and the greater the drying shrinkage. Higher slump, whether it is obtained with more cement and water or with a plasticizer, results in more settlement cracking, and pumping concrete usually involves the requirement for higher slump and higher paste content, both of which can add to the potential for settlement and drying shrinkage cracking.
Stresses in the bridge decks resulting from either the order of placement of the concrete during construction or traffic loads has been shown to play a much smaller role in cracking than factors dealing with material properties or construction.
With this understanding bridge decks have been constructed as part of the multi-state pooled-fund study. The specifications for these bridge decks involve an overall approach aimed at reducing plastic, settlement, thermal, and drying shrinkage cracking. This approach involves the use of low cement and water contents; low slump; moderate, not high strength; temperature control of the concrete; minimum finishing; and an early start and extended curing.
The specifications for low-cracking high-performance concrete (LC-HPC) involve the use of concrete with increased aggregate content and aggregate size, along with an optimum aggregate gradation to allow the use of concrete with the cement content of 540 lb/yd3 (320 kg/m3) or less. Water cement ratios have been in the range of 0.43 to 0.45 to help limit concrete compressive strength. Air contents range from 6.5 to 9.5%, and the designated slump range is 1.5 to 3 in. (40 to 75 mm). Not unexpectantly, one of the challenges has been to get contractors to use slumps in this low range. To help limit both plastic shrinkage and thermal cracking, the temperature of concrete, as delivered to the site, is specified as 55 to 70°F (13 to 21°C). In cold weather, the temperature must be maintained for both the girders and the deck.
The use of buckets and conveyor belts (the latter with a low drop from the belt to the deck) is emphasized, although the majority of the decks have been placed using pumps. Vertically mounted internal gang vibrators, spaced at 1 ft (300 mm), are used to improve consolidation and thus reduce settlement cracking. To help limit the cement paste at the surface of the deck, concrete finishing is minimized through the use of a single-drum roller screed (including double-drum roller screeds with one roller immobilized). This has worked well for the concretes with an optimized-aggregate gradation and controlled temperature, even at low slump.
After concrete placement, fully saturated, presoaked burlap is placed within 10 minutes of strike-off (see in Figure 1) and kept constantly wet with spray hoses until the concrete has set. Soaker hoses are then placed and the burlap is covered with white plastic. Curing continues for 14 days. To allow the concrete to dry slowly, the deck is sprayed with a curing compound upon removal of the burlap. The curing compound is maintained for seven days. The deck forms are removed within two weeks of termination of curing so that the deck can dry from both sides. The use of stay-in-place forms has a disadvantage in that the deck dries from only one side, which doubles the moisture gradient.
The results of the study, which includes an equal number of control decks constructed using conventional procedures, are summarized in Figure 2, with crack density shown in linear meters per meter of bridge deck as a function the age of the bridge. The control decks were selected to match the LC-HPC decks based on structure type and traffic loading. As shown in Figure 2, the LC-HPC decks have performed far better as a group than the control decks. What is not evident from the figure is that the LC-HPC decks have performed better than the matching control decks in every case. Full comparisons are available at https://iri.drupal.ku.edu/node/43.
In the next phase of the pooled-fund study, additional techniques are being applied to reduce cracking. These techniques include the use of fibers, potentially to reduce plastic shrinkage cracking; internal curing using pre-wetted lightweight aggregate combined with slag cement as a replacement for portland cement combined with a small additional replacement with silica fume; and the use of shrinkage-reducing admixtures. Bridges using these techniques are both in the planning and construction stages, and the effectiveness of these additional techniques will become apparent over time.
Further Information
For further information about this project, please contact the author at [email protected].