Comparison of methods for predicting incipient motion for sand beds

A comparison of four methods for predicting the incipient motion conditions of a uniform sand bed is presented. The four methods are: (1) the Shields diagram, (2) an empirical approach, (3) a method derived from resolution of rotational forces, and (4) a simplified resolution of rotational forces with a variable lift force coefficient. The four methods are used to predict the incipient motion conditions for 97 experimental runs taken from seven independent experimental flume studies. The effectiveness of predicting depth averaged incipient motion velocity for each of the four methods are compared. The simplified resolution of rotational forces model (4) and Shields method (1) were most successful in predicting the incipient motion velocity [R-2 = 0.77, 0.74 and root mean square error (RMSE) = 0.18, 0.15, respectively]. The slope of line of best fit for plots illustrating predicted versus measured incipient motion velocity were similar (slope = 0.63, 0.65, respectively), illustrating that both methods provide a similarly justifiable prediction of depth averaged incipient motion. The empirical method was the least successful at predicting the measured incipient motion conditions (R-2 = 0.49, RMSE = 0.41).