Simulations for evaluation of kinetic methods in differential scanning calorimetry .2. Effect of additional noise on single-peak methods

Theoretical data with various amounts of random gaussian errors have been used to test the validity of single-peak methods. This procedure was applied to various multiple linear regression algorithms with one or two kinetic exponents, and to the Achar-Brindley-Sharp, the Freeman-Carroll and the Ellerstein methods. Simulations were performed according to the results previously obtained for an experimental polymerization of an epoxy-novolac resin. The test of validity of the methods was performed using the LSM parameter which represents the fit between the mathematical treatment used in the kinetic model and known data, relative errors and standard deviations. The study shows in our case the lack of robustness of the Freeman-Carroll and of the Ellerstein methods in the presence of noise, and their sensitivity to temperature errors. The effect of smoothing has also been studied.