Fatigue resistance of concrete: influence of time-dependent scattering of compressive strength

Abstract In terms of sustainability and resource efficiency, concrete structures such as bridges and wind turbines should be used as long as possible and—in the case of new constructions (as a replacement)—the cross-sections should be as slender and thin-walled as possible using high-performance or ultra-high performance concrete. A further development of the fatigue design would be useful both for the verification of a possible longer remaining service life and for a safe, but also for economical and sustainable design of these engineering structures, which are particularly exposed to fatigue. The verifications of structural safety for non-static loading of concrete in the national and international design codes and standards provide for high safety margins, particularly for concretes with high strengths. These result, among other things, from the large scatter of the number of cycles to failure in experimental fatigue tests. In this article, current verifications of structural safety for non-static loading of concrete are presented, results of compressive tests on concrete specimens of different strengths, geometries and test boundary conditions are summarised in a database and the scatter of the experimentally determined number of cycles to failure is statistically evaluated. In addition, the compressive strength of concrete, which significantly influence the scatter of the numbers of cycles to failure, are statistically analysed for concretes of different ages. From this, a continuous description of the strength development and its scatter is derived. Finally, the compressive stress levels of the previously analysed fatigue tests are adjusted using a stochastic approach in order to take into account the scatter of the compressive strength of concrete as a function of the concrete age. By applying the time-dependent scatter of the compressive strength of concrete, a significant reduction in the scatter bandwidth of the analysed numbers of cycles to failure in the S–N curve is achieved.