Experimental Testing and Computational Analysis of Viscoelastic Wave Propagation in Polymeric Split Hopkinson Pressure Bar

摘要

The use of polymeric bars in the traditional Kolsky or Split Hopkinson Pressure Bar (SHPB) has been suggested by several authors as a means of improving coupling to low impedance materials and to increase incident wave rise time to assist in achieving dynamic equilibrium when testing soft materials. However, one aspect that must be addressed in this application is viscoelastic wave propagation leading to wave attenuation and dispersion. The amount of dispersion and attenuation depends on the bar material selection and incident wave signal. Viscoelastic wave propagation has been successfully addressed in Polymeric SHPB through experimental determination of the wave propagation coefficients, and has been investigated through analytical techniques; however, there is no widely accepted method for computationally modeling these events, which would benefit test apparatus design and optimization. To address this challenge, experimental impact tests were undertaken on a polymeric SHPB utilizing 2.5 m long acrylic (PMMA) bars with various strikers (0.712 m bar, 0.459 m bar, steel sphere) and impact velocities. The response was measured using the incident bar strain gauge signal and a Photon Doppler Velocimetry apparatus to measure the incident bar end velocity. The apparatus was computationally modeled using a commercial explicit finite element code and a viscoelastic Ogden material model. Parameter identification was undertaken using optimization software. The experimental test results were repeatable for all strikers and impact velocities, with the amount of attenuation and dispersion being highest for the sphere impact, compared to the striker bars. The computational model and optimization was applied to the incident bar strain gauge data to identify the material coefficients. The dominant material properties were dependent on the nature of the striker, but were able to accurately represent the incident and reflected wave signals. The resulting models were then validated using the PDV data using cross-correlation, demonstrating numerical robustness.