Comparison of a Normal and Logistic Probability Distribution for the Determination of the Impact Resistance of Polycarbonate Vision Panels

International standards for the safety of machinery define requirements for the design of safeguards in machine tools. An essential requirement of the safeguard consists of retaining ejected workpiece or tool fragments in case of an accident. Appropriate protective performance of the guard is demonstrated by means of an impact test carried out against a standardized projectile. The impact resistance (IR) is generally used as quantitative measure of an appropriate protective performance. It is defined as maximum kinetic projectile energy a safeguard is able to withstand. The standard procedure for determining the IR is the so-called bisection method, which involves narrowing a wide interval through a series of impact tests. However, this approach is associated with considerable uncertainty since it depends solely on the last two impact tests. In the present study, an alternative approach based on a probabilistic description of failed impact tests is proposed. A normal and a logistic distribution are compared in terms of their suitability for modeling the probability of failed impact tests. Both distributions are well suited, albeit the normal distribution requires considerable data preparation, which also affects its results. In contrast, the logistic distribution does not require any data preparation, providing an advantage over the normal distribution. This new approach can reduce the uncertainty associated with the determination of the IR providing more accurate and reliable results.