Improving the Current Elevated Temperature Assignment Model

As mentioned previously, the existing elevated temperature assignment methodology

could be further improved by:
– Using individual retention test results in the end of life regression analysis rather
than the average values
– Defi ning end of life other than 50 % retention (as appropriate)
– Drafting a method for forcing closure of programs at 10,000 h of aging
– Assigning provisional ratings faster based on a robust review of available data
Regardless of whether the single temperature 2,000-h heat aging program is used
or the full four-point traditional UL 746B [ 2 ] program, it is imperative that a better
statistical analysis be conducted on the data to analyze material performance. But,
to design a test program with a robust statistical method, it requires a thoughtful
understanding of the material property to be tested and the variation in test results
when the plastic compound is tested for a given property. For example, it is well
understood in the plastics industry that test properties to determine material toughness,
such as impact strength test methods like Charpy or Izod, sometimes produce
test results that are not very consistent. Because the typical sample size suggested
by most standards is fi ve specimens per set, it is sometimes diffi cult to determine the
sample variation. And if this sample variation is not properly taken into consideration,
it can result in a lack of fi t to a regression curve necessary for determination
of end of property life. Thus, one potential improvement would be for sample size
to refl ect the variation in the property test results for the specifi c material. Either the
test property needs to be selected so that variation is reduced or the sample size
needs to be selected large enough so that the variation is appropriately taken into
consideration in the analysis.
In terms of defi ning end of life with 50 % retention, it is challenging to evaluate
polymeric compounds for performance at elevated temperatures using the
Arrhenius activation energy model when the individual behavior of different compound
classes and formulations does not necessarily fi t the model. The Arrhenius
model may work very well when the compound exhibits fi rst-order kinetics (Feller, 147).
However, plastic compound formulations are complex chemical mixtures of multiple