Improvement of extrapolation of traffic load effect on highway bridges based on Rice's theory

It is a primary strategy in assessment, maintenance and rehabilitation of existing bridges in China to analyze structural safety using real traffic load data. For medium and large span bridges, the most common approach to extrapolate extreme load effect takes advantage of the Rice's theory. The method, however, has a problem of determining optimal starting effect interval for tail histogram curve fitting of crossing rate. At present, the problem is generally solved using the approach proposed by Cremona. But the approach is essentially subjective in computing procedure. This paper carries out the test and verification of Cremona's approach, discovering remarkable discreteness of extrapolated extreme load effects, and in-depth study reveals further flaws-improper understanding of statistical concept, miscalculation of significance level, and deviation of extrapolation by linear fitting method. To solve these issues, this paper develops a new method of determining optimal starting interval for tail histogram fitting. Instead of direct application of the Kolmogorov theory to crossing rate histogram, the new method starts analysis with the sample empirical distribution of effect at arbitrary time in stochastic process. By characteristics analysis of Kolmogorov distribution variable, it figures out whether the stochastic process variable complies with the hypothesis of the Rice's theory or the fitting curve tails with its theoretical curve (identification of optimal starting interval). Through in-depth analysis and case study, it can be seen that the proposed method can best reduce subjectivity in analysis procedure and improve on the load effect extrapolation based on Rice's theory.