Choosing pressure-viscosity relations

One obvious minimum requirement for any temperature-pressure-viscosity correlation is that the pressure-viscosity relation be capable of describing all possible trends in the pressure dependence of viscosity. Empirical models are employed because theoretical models do not provide experimental accuracy. If an empirical model is selected to determine outliers from any data set, failure to accommodate any aspect of the pressure response may lead to incorporating flawed data and to the exclusion of accurate data in the regression of correlation parameters. Two simple models have been thoroughly investigated, one for faster-than exponential pressure response and another for slower-than exponential pressure response. A test for the applicability of the McEwen model has been derived, analogous to the derivative Stickel analysis. A new model, a hybrid of the McEwen model and the Paluch (Johari and Whalley) model, has been derived and tested on two liquids. The purpose of this article is not to construct a reference correlation, but rather, to show the pit-falls of assuming a pressure-viscosity relation of limited applicability.