<p>The work applies the Tavares Breakage Model in Rocky DEM to simulate size reduction of a quarry rock in a laboratory jaw crusher. Initially, particles were modeled as a variety of polyhedral shapes to capture shape variability in the feed. From breakage parameters that were previously estimated from tests with single particles, and bench-scale tests for selection of preliminary contact parameters, the sliding friction was reduced to approximately match crusher throughput and power consumption for a base case test. Results show that fluctuations in throughput and power in individual experiments are also observed in the simulations, being even greater in the latter. Simulations were then carried out with different feed size ranges and closed-side settings. Comparisons to experiments show that the simulation approach captured reasonably well the effect of these operational variables on crusher throughput, product size distribution and power consumption, with greater differences observed when the crusher was fed with particles that are finer than the open-side setting. A final set of simulations was then carried out to investigate the impact of different sources of variability in the results. It showed that, for the rock studied, representing the feed either with uniform particle size and shape did not reduce variability in key performance measures, whereas removing the variability in particle fracture energies and representing it solely with the mean values resulted in reduction in variability in net power and statistically equal values of mean net power, throughput and product size.</p>