Prediction of Solid Propellant Burning Rate Characteristics Using Machine Learning Techniques

Abstract

When formulating a new solid propellant, one of the most important aspects of its performance is the burning rate's response to a change in pressure. For this reason, it is useful to be able to predict the burning rate response of a given propellant before the propellant formulation is created such that experimental trade studies are minimized or reduced in scale. While many theoretical and phenomenological models have been proposed to explain various aspects of energetic material combustion, little work has been made publicly available in the application of machine learning models to predicting solid propellant burning rates. To facilitate model creation, the material formulation and burning rate parameters for over 600 publicly available propellant formulations have been collected into a coherent data set. This work utilizes the large amount of publicly available data to inform a random forest machine learning (ML) model in the prediction of solid propellant burning rate parameters. This ML model operates over a large parameter space including ammonium perchlorate composite, plastic bonded high explosive, and double-base propellants. The model's accuracy, adaptability, and prediction capabilities are presented and discussed. The effects of different materials on a propellant's expected burning rate are examined.

A review on the effect of ageing on the ballistic properties of solid gun and rocket propellants

“…Klinger et al [63] noted that it is important to measure how the solid rocket motor propellant burning rate changes in response to pressure. Ballistic instability, where there is a rapid rise and drop in pressure upon firing, can be a life limiting issue.…”

“…To achieve more realistic measurements of the propellant regression rates, two custom built microwave interferometers were used which operated at different frequencies and various configurations [63]. The dynamic regression rates for a propellant strand experiencing pressure sweeps in motor like conditions were also reported.…”