As Kolsky Bars (Split Hopkinson Pressure Bars) have grown in popularity, so too has the need for high rate material characterization. To achieve higher rates in excess of 10,000 strain per second, smaller samples and therefore smaller bars are needed. These smaller bars, approximately 3mm in diameter, are susceptible to additional errors due to their size. For example, a thinner bar is more susceptible to bending and gauges cannot always be placed on both sides of the bar to remove additional error in post processing. Impacts from sample tolerancing also become more prominent as the size of the sample decreases. The community needs to be confident that Kolsky bars of smaller sizes are capable of generating accurate and reproducible results. Given this we have developed a statistically rigorous analysis to capture the system error of a Kolsky apparatus. This involved in excess of 30 tests without a sample present to characterize system attributed errors. Following this, over 30 samples of aluminum 2024 were tested and the statistical spread was calculated. Following this methodology allows us to identify what is sample and system variance, allowing for a better identification of material properties.
Having experimental bounds established for the system also makes model generation more straight-forward. Experimental data to date indicates that variance between load histories is maximum at compressive and tensile peaks with lesser variance on loading and unloading sections. So long as results from codes such as Finite Element Analysis packages fall within the ranges identified in this experimental technique, the model can be considered accurate. Data from simulations need not exactly match an experimental run so long as the simulation fits within the experimentally determined error bounds. To that end, simulations of the system have been generated using finite elements and the results are compared to those collected from experimentation.