Historicidad e identidad

The measured temperature data, from January 24 to July 13 in 2004, plotted in Figure 2-2, is selected and used throughout the following discussions. Figure 2-3 and Figure 2-4 show the detailed hourly-varying temperatures during the warmest and coldest week, respectively. In this section, the temperature distribution is observed and analyzed with the purpose of clarifying the following uncertainties, including (1) the temperature variation patterns of the GFRP panel, (2) the differences between the measurements and the available AASHTO LRFD (2007) temperature design specifications for concrete decks, and (3) the effects of the environmental and material parameters on temperature variations. Since the wind speed, also one of another important influencing factors, has not been measured in the field, then, the data from the nearest weather station, Konza Prairie Biological Station, located at Manhattan, KS, with the latitude and longitude of 39.1027N, 96.6098W (http://www.ncdc.noaa.gov, 2012), is obtained for reference. Based on the observation, some of the findings can be summarized as follows:

Fig. 2-2 Measured Temperature from January 24 to July 13, 2004

Fig. 2-4 Measured Temperature from Feb.7 to Feb.14, 2004

Fig. 2-6 Temperature Linear Fitting Results of the Bottom Surface and the Air

(1) Figure 2-2 shows the temperature variations at the panel surfaces from January 24 to July 13 in 2004. It can be observed that the measured temperatures are more significantly fluctuated at the top surface as opposed to the less varying ones on the bottom surface. This phenomenon indicates that the GFRP top surface temperature, evidently being higher than that of the air, is decisively affected by the solar radiation together with the material energy absorption ability; while the GFRP bottom surface temperature, almost being consistent with that of the air, is basically affected by the air convection below the panel. At the same time, the high and varying top surface heat energy cannot be easily transferred through the panel depth, so that it only makes negligible effects on the temperatures at the bottom surface, and this behavior may be attributed to the lower GFRP thermal conduction properties and the hollow section configurations.

The night sky radiation behavior, for the case when the bridge temperatures are lower than that of the ambient, is also observed in Figure 2-2. Specifically, during the night, when the bridge surface faces the night sky, it loses heat by radiation to the sky and gains heat from the surrounding air by convection. If the surface is a good radiation emitter or the convection is weak, it will tend to radiate more heat to the sky than it gains from the air, and the net result is the surface temperature dropping below to that of the air. Since the night sky radiation behavior often induces negative thermal gradients, thus the frequent occurrence of this behavior on GFRP panels in this project may indicate that the negative thermal gradients are common for GFRP panel bridges and the attentions should be given in design.

(2) The hourly-varying sinusoidal temperature variations are clearly observed from the plots of the warmest and coldest week temperature distributions in Figure 2-3 and Figure 2-4. Therefore, the sinusoidal fitting algorithm discussed in the previous analytical modeling section is proven to be reasonable even though the lagging effect is not evident for this GFRP panel due to its shallower configurations and the smaller thermal inertia.

In addition, the measured maximum positive and negative temperature differences between the top and bottom surfaces during the warmest and coldest week are 28 (51 ) and 7 (12 ), respectively. Considering the design temperature stipulated for concrete slabs in AASHTO LRFD (2007), it specifies a 25 (46 ) for positive gradients, and multiplied the positive gradients by -0.3 for plain concrete decks and -0.2 for asphalt overlaid decks, respectively, to obtain negative gradients. It can be calculated that the temperature differences of

the current GFRP panel, though not the worst case yet, are already extending the range that is specified for concrete slabs. Thus the available temperature design guidelines are no longer effective for GFRP slabs.

(3) Figure 2-5 shows the measured environmental conditions at both the bridge site and the weather station. The environments at the weather station can approximately represent that at the bridge site since the measured air temperatures are almost the same. The measured wind speed at the weather station is generally lower than 5 ⁄ (16.4 ⁄ ). Under this wind speed, the thermal coefficient of convection for a concrete deck will be approximately less than 20 ⁄ (3.5 ⁄ ) (Elbadry and Ghali 1983). In this sense, even though it is still unclear about the actual convection coefficient for a GFRP panel, yet considering the fact of almost consistent temperatures between the deck bottom surface and the air, it can be estimated that evident convection behaviors may happen under current wind speeds and the value of 20 ⁄ (3.5 ⁄ ) will be a good reference to be used in the following numerical models as the convection coefficient of the GFRP panel.

(4) The relationship between the bridge surface and the ambient temperatures are linearly fitted. Obviously, the bottom surface shows almost the same temperature as that of air, shown in Figure 2-6, while it is not the case for the top surface where a discrete pattern exists, shown in Figure 2-7. Therefore, for this specific GFRP panel, it is reasonable to use the ambient temperature to represent the bottom surface temperature. For the top surface temperature, however, it can be further divided into two scenarios. During the night without external solar radiation, the bridge surface temperature tends to approach the ambient temperature, and the linear trend is still valid; while during the daytime with solar radiation, the linear trend no longer exists.