Several parking structures in Biloxi, MS and Gulfport, MS were examined between 7 and 9 September 2005, shortly after the 29 August 2005 landfall of Hurricane Katrina in the Gulf of Mexico near New Orleans. Five of the parking structures were constructed of precast concrete with pretensioned double-tee beams for decks; the other three were of cast-in-place reinforced concrete. All were either on the shore of the Gulf of Mexico or facing the Gulf across the street; all were subjected to storm surge estimated to be 20 to 30 ft in height. While none of the cast-in-place structures suffered any structural damage (with the exception of one such structure that suffered partial collapse probably because of impact from a casino barge), all of the precast concrete structures suffered partial collapse of the second-floor deck. Storm surge seems to have reached the level of this deck, which was typically 10 to 15 ft or so above grade (grade being roughly 3 ft above normal sea level), but not the third-level deck, which was generally 20 to 25 ft high. In cases where only part of the deck collapsed, it was typically on the side from which the waves approached.
The concentration of collapsed deck on the seaward side, where wave action would be strongest, implies that wave action played a role in the collapse, either vertically, through uplift of the deck beams, or horizontally via the spandrel beam. Evidence of horizontal pounding by waves can be seen in Figure 1 and Figure 2, note the holes in spandrel beams where the shear key in end of the deck-beam web notched into the socket of the spandrel beam; the shear key appears to have punched the holes. But this horizontal action may not have been the principal cause of the failure. In some cases, the collapsed deck beams had been oriented parallel to the seashore, as in Figure 3; in others, perpendicular to it, as in Figure 4. In some cases, the spandrel beam rested on the outside of the column (Figure 5), and could only have moved inward at mid-span, in flexure. (Some flexure, of course, is clearly possible, as shown in Figure 6).
Vertical loading was clearly substantial. Shear failure was observed in deck-beam shear keys (Figures 7 and 8) and in supporting girders (Figure 9, the interior girders in the background of Figure 10, or the three bites in the 3rd-level spandrel beam in Figure 11). Such loading could have been caused by pounding resulting from deck beams being lifted and then dropped, or by the weight of water accumulated on the top of the deck before it could pour down a ramp. If the latter occurred though, why would damage tend to be concentrated at the seaward side or corner of the deck? Water would probably flow quickly enough to distribute around the entire second-floor deck and cause more general damage.
If it were uplift and dropping of the deck beams, rather than the weight of water, what caused the uplift? Wave action alone might conceivably have lifted the deck beams, but buoyant forces would have contributed strongly. The deck beams and their concrete topping weigh approximately 150 pcf; seawater weighs approximately 66 pcf. By Archimedes’ Principle, buoyancy alone would have reduced the downward vertical load on the deck beams by almost half, a loading condition for which these pretensioned members were probably not designed. In addition to buoyancy of the deck beams and topping, the shape of the double-tees, whose ends were enclosed by their supporting girders, would have allowed for the creation of air pockets (Figure 12). As the storm surge rose to the soffit of the second-level deck, the concrete topping over the deck would have greatly limited the ability of air to escape, and additional buoyant force equal to the weight of water displaced by the air pocket would have been applied upward to the deck beams, potentially reaching or exceeding the self-weight of the deck beams and their topping.
Under such a condition—buoyancy caused by air pockets and the volume of the deck concrete itself neutralizing the self-weight gravity loading—the negative bending moment induced by the pretensioning would have been unopposed by gravity-induced positive bending in these simply supported beams. The negative bending induced by pretensioning is greatest in the region of the midspan, so if uplift were the cause of the collapse, one would expect to see evidence of negative-bending flexural damage at the midspan, such as concrete spalling at the bottom fiber at midspan and diagonal cracks radiating outward and downward from the top fiber. If on the other hand it were the weight of water on top of the deck that caused failure, one would expect to see the opposite: concrete spalling at the top fiber at midspan and diagonal cracks radiating outward and upward from the bottom fiber.
The damage observed here supports the negative bending hypothesis: spalling of concrete at the bottom fiber rather than the top (Figure 13) and diagonal cracks radiating outward and downward from the top fiber (Figure 14). In Figure 14, note the damage to the deck flange every three feet or so at the points where angle clips (visible on the right-hand side of Figure 7 as well) connected to the spandrel beam: the concrete cover spalled downward rather than upward, particularly at the midspan, implying that the midspan of the deck moved up relative to the spandrel beam.
This review of fairly widespread damage to precast concrete garage structures in Biloxi, MS, and Gulfport, MS, caused by Hurricane Katrina is based on limited information: brief (1 to 2 hr) visual examination of only five structures, no material testing, examination of structural drawings, mathematical modeling of the structures, nor analysis of the imposed hydrodynamic forces. We did not search for or examine precast concrete structures where collapse did not occur but that were similarly exposed to storm surge and wave action (we didn’t see any). We therefore draw no firm conclusions about the safety of other, similar structures that could be exposed to storm surge, nor do we offer any advice for mitigating potential risk. Because this is a quick reconnaissance study, we have not yet performed a literature review to check whether this phenomenon has already been observed elsewhere and addressed by standards-writing authorities such as the Precast Concrete Institute, American Concrete Institute, International Code Council, or American Society of Civil Engineers. Such additional steps would be required before firm, general conclusions about safety and risk mitigation could be made.
Submitted by Keith Porter and Gilberto Mosqueda
September 11, 2005