Variable Amplitude Fatigue in Relation to Highway Bridges
With the introduction of welded construction in highway bridges it has become necessary to assess designs for fatigue. Current methods of calculating endurances involve constant amplitude data and assumption that the Palmgren-Miner law can be used to sum the damage caused by different stresses. In order to investigate the efficacy of this method laboratory tests have been conducted on representative welded connections using variable amplitude loading at endurances of up to 415 × 106 cycles. Using a Rayleigh spectrum of stresses it is shown that for endurances relevant to bridge service, the role of stresses below the constant amplitude fatigue limit is critically important. If they are ignored, calculated endurances can be many times too optimistic. An accurate allowance for the effects of low stresses can be made by representing the constant amplitude data by a curve having a higher stress exponent for stresses below the fatigue limit. At the longest endurance tested, only about 0.25 per cent of the stresses exceeded the fatigue limit and these caused 9 per cent of the calculated damage. Under an axle load spectrum, endurances were six times longer than calculated. This was found to be due to an interactive effect caused by small numbers of high stresses which effectively retard the rate of crack propagation. In practice, shapes of stress spectra differ from axle load spectra and situations in which measured stresses involve small numbers of high values are rare. In all cases, conservative endurances were estimated using the method given in the British Standard for bridge design.