TY - JOUR TI - Stochastic approach for fine sediment erosion prediction DO - https://doi.org/doi:10.7282/T3Q81G10 PY - 2015 AB - This study aimed to characterize the erosion behavior of cohesive sediments in the Newark Bay, at flow velocities below 1 m/s based on their index properties. The experimental methodology and data interpretation scheme of this research were devised based on the critical analysis of previous literature and aimed to reduce uncertainty, subjectivity, and arbitrariness. A comparison of erosion measurements obtained in this study with the results of some in-situ experiments conducted by other researchers revealed a strong consistency between these studies. The fact that this ex-situ study has been as successful as in-situ studies is quite an achievement. The success of the devised experimental methodology was also highlighted when the results were compared to similar ex-situ studies because the range of erosion rates measured in this study was well beyond the capability of those methods. This research contributes to the literature on cohesive sediment erosion by offering new insights into three primary areas: regression, stochastic, and probabilistic analysis of erosion test results. First, this study employed the regression technique to obtain the best linear unbiased estimator of erosion rates based on sediment index properties. The analysis resulted in the development of two fairly valid models for both fine- and coarse-grained sediments of the Newark Bay: (1) Newark Bay Fine Model (NBFM) and (2) Newark Bay Coarse Model (NBCM). These models were evaluated through cross-validation and cross-model comparison, as well as validation against a new dataset. Second, a new methodology was developed for a stochastic analysis of erosion data by applying the Monte Carlo simulation technique. To the best of the author’s knowledge, this technique had not been previously used in sediment erosion studies. This robust stochastic method enabled the researcher to investigate erosion over many artificially generated samples, in lieu of measured data, and make more realistic predictions. The confidence interval provided by stochastic simulations has a significant application in sediment erosion risk analysis. Third, the framework developed for the probabilistic analysis of erosion data offers a standardized methodology for data analysis that paves the way for the comparison of different studies that use inconsistent methodologies. KW - Civil and Environmental Engineering KW - Cohesive sediments KW - Sediments (Geology) KW - Erosion LA - eng ER -