An alternative neural network approach to calculate the molecular weight distribution from dynamic rheological properties of i-PP resins

The calculation of the molecular weight distribution (MWD) of a polymer from its rheological properties is an attractive method since rheological measurements are comparatively faster and cheaper than the classical gel permeation chromatography technique (GPC). The calculation, however, still has some drawbacks, such as the sensitivity of the mathematical solution involved (ill‐posed problem) and the limited frequency range covered by commercial rheometers, which can be especially critical for crystalline polymers, for which the time–temperature superposition is of limited worth. In this article, a new approach for evaluating the MWD from the storage modulus and the relaxation modulus curves is proposed. The method, based on the use of a neural network model, was employed to evaluate MWD from rheological data obtained with different isotactic polypropylene resins. The results show that this approach can be successfully used to compute MWD curves and should expand the range of application of the rheological technique. © 2000 John Wiley & Sons, Inc. J Appl Polym Sci 75: 1416–1423, 2000