At LLNL we have been evaluating shock-initiation thresholds for various explosives for a number of years, using thin flyer plates accelerated by an electric gun as the shock stimulus (1 to 3). It is advantageous in time and expense to minimize the number of test specimens required for a threshold determination. We have recently improved our velocity calibrated and our experimental protocol in selecting firing voltages so that we can usually obtain a good estimate of the threshold using only 8 to 10 samples.

Improvements in thin-pulse shock initiation threshold measurements

In this second part of a series of publications dealing with the application of the Robbins-Monro method to the sensitivity testing of explosives, the experimental aspects are treated. The results of an experimental program with a primer drop-test apparatus confirmed the efficiency which was concluded from the computer simulation study, also for extreme percentage points. Further a method is described to determine the reliability of the outcome after the execution of the Robbins-Monro process. It is shown that for an efficient application of the Robbins-Monro method the level distance which can be adjusted should be about 1/2 to 1/3 of the standard deviation (of a normal distribution) dependent on the percentage point.

Application of the Robbins‐Monro method to the sensitivity testing of explosives. II. The experimental practice

Influence of sensitizer thermochemical properties on stab initiation of sensitized stab mixes

Sensitivity tests provide data that help establish guidelines for the safe handling of explosives. Any sensitivity test is based on assumptions to simplify the method or reduce the number of individual sample evaluations. Two common assumptions that are not typically checked after testing are 1) explosive response follows a normal distribution as a function of the applied stimulus levels and 2) the chosen test level spacing is close to the standard deviation of the explosive response function (for Bruceton Up-Down testing for example). These assumptions and other limitations of traditional explosive sensitivity testing can be addressed using Transformed Up-Down (TUD) test methods. TUD methods have been developed extensively for psychometric testing over the past 50 years and generally use multiple tests at a given level to determine how to adjust the applied stimulus. In the context of explosive sensitivity we can use TUD methods that concentrate testing around useful probability levels. Here, these methods are explained and compared to Bruceton Up-Down testing using computer simulation. The results show that the TUD methods are more useful for many cases but that they do require more tests as a consequence. For non-normal distributions, however, the TUD methods may be the only accurate assessment method.

https://iopscience.iop.org/article/10.1088/1742-6596/500/5/052007/pdf

Improved sensitivity testing of explosives using transformed Up-Down methods

A computer simulation study of the Robbins-Monro method is presented. This method can be used to determine percentage points of a distribution function describing the probability of explosion as a function of some stimulus level. Calculations were performed to establish the influence of several parameters involved in the process. It is concluded that the method is suitable for determining 50, 10 and 1 percent points for several distribution functions. The efficiency for a restricted number of experiments and for a reasonable choice of the parameters is similar to that of a situation where experiments are performed at a fixed level, however with the advantage that the desired percentage point is obtained by convergence.

A. A. Schilperoord

Papers

Application of the Robbins‐Monro Method to the Sensitivity Testing of Explosives. I. A computer simulation study