The flocculation of cohesive sediments represents a critical process in coastal sediment transport, with its appropriate representation in numerical models crucial for the prediction of contaminant transport, coastal morphodynamics and engineering problems. In this study, a flocculation model considering the effects of multiple fractal dimensions is incorporated into a two-phase numerical modelling framework and used to investigate the effects of spatio-temporal variations in sediment concentrations on the temporal evolution of local floc sizes. Initially, the model is applied to simulate the aggregation of clay suspensions in a vertical grid-stirred settling column, with results confirming the importance of multiple fractal dimensions when predicting the time evolution of floc sizes. The adoption of multiple fractal dimensions, in particular, allows the two-phase numerical model to better match the measured settling column data with improved overall correlation. This is especially the case when predicting initial floc size growth during the early period of settling when the flocs tend to adjust more rapidly to their equilibrium sizes. The two-phase model is then applied to simulate field measurements of mud resuspension process in a tidally driven channel. Again, by considering multiple fractal dimensions within the flocculation model, better agreement is obtained between observed and modelled suspended sediment concentrations, while predicted floc sizes are also in general accord with previous field measurements made within the same estuary.