Today, parallel processing supercomputers allow simulations of up to 10(6) discrete elements. and advances in processing power should allow 10(9)-10(10) discrete elements. The availability of such large data sets offers real scope for statistically accurate calculations to describe the long-range evolution of the bulk flow and stress fields during industrial handling and processing of material in granular and powder states. We show that careful post-processing analysis of granular dynamics simulation results in terms of time series/signal processing can lead to a greater understanding of the micro-dynamics of granular systems and the defining events in the evolution of such systems. In particular, we apply the Fourier transform to investigate periodicity throughout the duration of the simulation and the wavelet transform to investigate disturbances which are localised in time and which are much more difficult to model with Fourier methods. Whilst we are essentially pointing to post-simulation/experiment techniques which are generic, we illustrate their importance for the investigation of segregation phenomena in poured binary granular heaps, with a particular focus on the diffusive mechanism of kinetic sieving.