<p>Tensile testing is a well-established method to assess the maximum strength of a material, while
relaxation tests are used to evaluate the viscoelastic behaviour of a polymer. Because of slow viscoelastic changes, significant measurement times are required for reliable descriptions. Therefore
the relaxation tests are usually combined with lifetime prediction models to reduce the experimental
load. Various traditional models use the time-temperature superposition principle while modificated
relaxation models are e.g. based on the time-strain superposition principle (TSSP). Both variations
require several measurement series to set up a relaxation master curve (RMC). The basic assumption
is that a higher strain corresponds to a higher temperature and a longer load duration, respectively.
The paper describes a new model approach which allows to predict the longterm behaviour by using
a reduced number of measurements as compared to widely models. The new model is based on the
well-known Maxwell model and assumes a mean relaxation time in combination with a relaxation
coecient. These parameters account for the inhomogeneity of the individual polymer chains. A
dimensionless number, similar to the relaxation coecient, has been successfully introduced for the
Weibull distribution and the particle size distribution. The new model allows to derive master curve
from one measurement series at a single strain by fitting the data to the model equation.<br></p>