Numerical modelling of short pulse shock initiation of ultrafine Hexanitrostilbene (HNS)

The study of short duration shock pulses is relevant in many high-energetic applications. One such application is the initiation of insensitive high explosives, often preferred as booster charges in military and civilian applications due to superior safety and timing properties. Ultra-fine Hexanitrostilbene (HNS) is a material that meets the requirements of such booster charges. Critical flyer impact velocity for a given flyer thickness and material, and run-up to detonation distance as a function of flyer velocity for a given flyer, becomes key observables in impact experiments designed to characterise a specific grade of HNS. Numerical modelling can serve as an important additional tool in such studies.

In this work, we investigate the use of the numerical method smoothed particle hydrodynamics (SPH), and a particular extension to that method called Regularized SPH (RSPH), for modelling the interaction of a plastic flyer with an HNS pellet. This report describes the basic features of the model used for studying the initiation process of the explosives. It presents a detailed discussion of the importance of using an appropriate amount of artificial viscosity to control numerically the fluctuations resulting from the impact. Details are given on how the Ignition & Growth model describing the detonation process is implemented. Resolution requirements for the specific application is formulated, first for the simpler case of non-reactive HNS, then for reactive HNS. It is concluded that 20-40 calculation nodes (referred to as particles) are needed across the width of the flyer in order to secure a reasonably accurate detonation threshold and run-up to detonation description. This represents a severe resolution requirement which makes it very challenging to simulate the problem in full 3D. Two different options for reducing the CPU-cost, using either variable resolution or a time-varying simulation domain, are discussed.

The results indicate that using time-varying simulation domain is a better strategy than using variable resolution in this case because of numerical stability issues. Comparisons of numerical and experimental results of a 75 𝜇m thick flyer reveal a good fit when the flyer width is roughly 600 𝜇m or larger. As the flyer width is reduced below 600 𝜇m, the numerically obtained critical flyer velocity increases faster than the experimental data suggests. The reason for this discrepancy has as of yet not been established.