For glassformers we propose a binary mixture relation for parameters in a cluster kinetics model previously shown to represent pure compound data for viscosity and dielectric relaxation as functions of either temperature or pressure. The model parameters are based on activation energies and activation volumes for cluster association-dissociation processes. With the mixture parameters, we calculated dielectric relaxation times and compared the results to experimental values for binary mixtures. Mixtures of sorbitol and glycerol (seven compositions), sorbitol and xylitol (three compositions), and polychloroepihydrin and polyvinylmethylether (three compositions) were studied.

A prior correlation model for glass formation based on cluster-size distribution kinetics is here extended to account for pressure effects as well as temperature effects. The model describes how rapidly cooling or compressing a liquid or colloid leads to structural arrest and a consequent sharp rise in viscosity or dielectric relaxation time. In addition to activation energies, we include activation volumes in the rate coefficients for monomer-cluster addition and dissociation and cluster aggregation and breakage. The approach leads to scaled pressure correlations and plots for viscosity that reveal strong and fragile glass behavior, and agree with experimental data. A simple relationship among viscosity, attractive interparticle energy, and particle volume fraction displays how hard spheres with attractive forces can vitrify at small particle densities.

No abstract

Since the quality of the raw upstream operating data can be poor in many instances (due to errors, incompleteness, and inconsistency), there is often an urgent need to cleanse the data in real time before using the data for process and business decision-making. In this paper, we present the design and development of an integrated data cleansing framework that can address a variety of upstream operating data quality issues systematically. Our proposed framework, dubbed SDPF (short for Scalable Data Processing Framework), benefits from the following features: Online/Real-time Data Cleansing, Scalability, Configurability, Reusability, and Comprehensiveness. We have tested and verified the performance of our data cleansing system with both real upstream operating data collected and synthetic data.

No abstract