
3.2.6 Statistical Modeling for Polymer Temperature Measurement
Mark G. Vangel
Stefan D. Leigh Statistical Engineering Division, CAML
Anthony J. Bur Polymers Division, MSEL The objective of this project is to develop a method for indirectly measuring the temperature of a polymer during fabrication. A compound is added to the polymer which fluoresces when excited by a laser. Empirically, it has been shown that the ratio of the intensities of the resulting spectrum at two wavelengths is, for a practically useful range of temperature, nearly linearly related to the temperature within the material. Thus, a calibration curve can be determined which relates this ratio to temperature. By measuring the intensity at two wavelengths, a user can then use this curve to estimate temperature.
Statistical models have been developed for
experimental spectra in an attempt to explain
this empiricallybased calibration curve. A
mixture of four Gaussians where the mixture
proportions are allowed to vary with temperature
was fit to data, with good results. From the figure,
it is clear that the proportions vary linearly
with temperature. Some hysteresis is also
apparent, probably because of a change in
the additive when it is heated. This model
will hopefully provide insights leading
eventually to a physical explanation of
the phenomenon. Figure 18: Proportions of Components of a Gaussian Mixture as Functions of Temperature
Date created: 7/20/2001 