Decking 



Steel surface plate 



Beam 



Wind 



Cavity 



Figure 31. — Cross section of approach span of bridge, showing supposed pattern of wind eddy formation 

 within cavities. 



Once the bridge had been ^ited, en- 

 hancement of the burning can also be attrib- 

 uted to the long rectangular cavities between 

 the beams located below the bridge. It is hy- 

 pothesized that wind blowing over these cav- 

 ities perpendicularly to the beam axis or 

 bridge direction produces eddies on the lee- 

 ward side of the beams within each cavity, 

 as illustrated in figure 31. As the windward 

 side burns, flames are entrained into the 

 eddies on the leeward side. Thus, combus- 

 tion is propagated into the adjacent cavity. 

 In this situation, only a limited amount of 

 air is entrained into the cavity with the 

 flames so that combustion of only a portion 

 of the volatilized gases from the inside sur- 

 face can be completed within the cavity. As 

 the heat builds up, an overabundance of the 

 volatilized gases is produced within the cavity. 

 Thus, the proper fuel air mixture is not at- 

 tained until the gases are swept out of the 

 cavity into the outside air. However, just as 

 these gases leave the cavity and begin flam- 

 ing combustion they are entrained into the 

 next cavity and so the cycle is repeated until 

 all the cavities are burning. Notice also, in 

 figure 31, that the two walls formed by the 

 beams are advantageously located for enhance- 

 ment of burning by reradiation between the 

 walls. 



There is no direct evidence to support the 

 hypothesis that eddies were formed in the 

 cavities below the deck. However, a recon- 

 struction of the bridge fire under laboratory 

 conditions gives us- a qualitative examination 

 of eddy production that could lead to support 

 or rejection of the hypothesis. For our pur- 

 pose a cross-sectional slice of the beam axis of 

 the bridge-approach span is sufficient. Since a 

 qualitative examination is all that is possible, 

 there is no strict adherence to scaling. The 

 deck is made up of IV2- by y2-inch white pine 

 lumber and the beams are replaced by 2- by 

 4-inch Douglas-fir. Two coats of diesel oil 

 were applied to the model to simulate the 

 creosote treatment. 



The simulated bridge section was placed in 

 the wind tunnel with the wind flowing at 

 right angles to the beam axis. To eliminate 

 burning on the exposed cross-sectional faces it 

 is necessary to cover both faces with sheet 

 metal. Two bridge sizes were burned in the 

 wind tunnel under identical conditions, the 

 only difference being in the width. In the first 

 test the bridge was 42 inches wide and in the 

 second 17 inches. In both cases ignition by 

 simulated ground fuels (excelsior) was on the 

 windward side. Because of the size difference 

 ignition was confined to the first few cavities 

 in case 1 and extended over the entire bridge 

 section in case 2. 



35 



