Statistical assessment methods for the sensitivity of energetic materials

Mohamed Alouaaniari, Michel H. Lefebvre, Christiaan Perneel, Michael Herrmann

Research output: Contribution to journalArticlepeer-review

Abstract

The joint research project Particle Processing and Characterization was started in March 2003 under the EUROPA ERG1 Arrangement, with the objective to study the influence of crystallization and processing techniques on particle quality and its implication for the formulation of PBX with IM behavior. As sensitivity assessment is a crucial task of the project, impact sensitivity tests of HMX and RDX samples have been performed at two different laboratories, and several statistical techniques have been tested in addition to the standard BAM "1/6" and Bruceton "up and down" methods. When the "go/no-go" response is plotted against the stimuli, the obtained curve is a step function while the actual behavior of the impacted explosive should exhibit a sigmoid function. Therefore, we have investigated statistical methods for analyzing the experimental results and to determine this sigmoid curve by using a statistic assessment. We compared the results obtained by the "1/6" method, Bruceton, the logistic regression, and Probit analysis for several explosives. The investigation shows that the proposed alternative methods exhibit a useful capability to classify any explosive samples according to their respective sensitivity. This is of importance particularly for qualifying explosive formulations for their later potential use in insensitive or reduced sensitivity applications.

Original languageEnglish
Pages (from-to)60-65
Number of pages6
JournalPropellants, Explosives, Pyrotechnics
Volume33
Issue number1
DOIs
Publication statusPublished - Feb 2008

Keywords

  • Binary logistic regression
  • Bruceton
  • HMX
  • Impact sensitivity
  • Probit analysis
  • RDX

Fingerprint

Dive into the research topics of 'Statistical assessment methods for the sensitivity of energetic materials'. Together they form a unique fingerprint.

Cite this