On the structure and accuracy of programmed burn

Accurate computation of the evolution of a (typically curved) detonation front in a complex geometry, and of the state behind it, is a practical problem in the design of devices that use high-energy explosives. Direct numerical simulations are infeasible: accuracy demands high resolution of the smallest scale (the reaction zone), which is typically several orders of magnitude smaller than the device scale. Programmed burn is an engineering alternative that has been shown to produce acceptable results at a fraction of the cost. The underlying algorithm prescribes the trajectory of the detonation front a priori and replaces the actual reaction zone by a mock up that is a few computational cells thick and in which the reaction rate is taken to be a constant. The state of the explosive at the end of the reaction zone is thereby computed at a relatively modest cost, and the bulk of the computational effort is reserved for the region behind the reaction zone wherein the products of detonation perform useful work. The reasons for the remarkable fidelity to which the physical situation is captured by the programmed burn are not well-understood. This investigation, aimed at achieving such an understanding, considers a model problem for a steady, curved detonation propagating down a rate stick. It examines the pseudo-reaction-zone structure of the programmed burn, studies the sensitivity of the state of the reaction products to the choice of the reaction zone length, and compares the results to those for the actual, physical reaction zone. Conclusions are drawn as to the causes behind the success of the programmed-burn algorithm. The analysis is based on the asymptotic limits of small front curvature and small departures from the Chapman - Jouguet speed. Results are presented for ideal as well as nonideal explosives.