Spatiotemporal characterization of wave-augmented varicose explosions.

• Simulations of transonic non-Newtonian, wave-assisted slurry atomization in novel injector. • Unraveled visualization of annular wave cycle. • Nozzle exit marks dramatic increase in azimuthal variation of slurry waves. • Instabilities are excited in self-amplification cycle. • Adaptive mesh refinement found to be insufficient for this modeling application. The onset of three-dimensional instabilities during "Wave-Augmented Varicose Explosions" (WAVE) liquid disintegration is characterized for an annular flow of a shear-thinning slurry that is fed into a central transonic steam flow. Droplet production inside the nozzle is enhanced by ligaments radially flicking up from the slurry wave into the steam flow with radial:axial velocity ratios exceeding 0.5. The wave also leaves residual ligaments in its wake, which facilitate further disintegration. After birth, a wave spends 80% of the wave cycle period building up to peak height at the nozzle exit. Two effervescent mechanisms are provided as 1) steam penetrates the rising wave and surface deformation allows steam fingers to force through, and 2) the wave collapses on itself, trapping steam. Baroclinic torque drives the development of Rayleigh-Taylor (RT) instabilities and reaches values on the order of 1 × 10 13 1/s2. Both RT and Kelvin-Helmholtz instabilities are self-amplified in a viscosity-shear-temperature instability cycle because the slurry is non-Newtonian. Velocities inside the nozzle (wave formation region) are generally azimuthally similar (two-dimensional), but those outside the nozzle (radial bursting region) are azimuthally uncorrelated (three-dimensional). Inter-variable correlations show significant decoupling of quantities beyond the nozzle exit, and local strain rate fluctuations were found to correlate particularly well with bulk system pulsation. Although adaptive mesh refinement (AMR) can provide computationally efficient resolution of gas–liquid interfaces, this technique produced different results than an equivalent non-dynamic mesh when modeling WAVE. Gradients were particularly affected by AMR, and turbulent kinetic energy showed differences greater than 150% outside the nozzle. [ABSTRACT FROM AUTHOR]