Wallace B. Whiting

Sensitivity and uncertainty of process designs to thermodynamic model parameters : a Monte Carlo approach

Abstract Process simultalion relies heavily on the calculation of physical properties through thermodynamic models. The parameters in these models are subject to uncertainties that may, at first, appear to be small. However, the designs developed through use of these models can be significantly sensitive to these parameter uncertainties. Previous studies have considered special cases of design error or have involved simple sensitivity analyses close to the optimum design case. In the present work, we use Monte Carlo simulation with Latin Hypercube Sampling to quantify the probability of design errors. An example of binary distillation with phase-equilibrium calculations from a cubic equation of state is used to illustrate the technique.

INTERACTING JETS IN A FLUIDISED BED

Abstract This work extends the Pitot-tube probe technique to the study of jetting phenomena in a three-dimensional, high-temperature gas-solid fluidized bed. For an isolated jet, the jet height was defined as the intersection of two momentum flux profiles, one along the jet axis and the other in the emulsion phase. A study of two adjacent jets confirmed that the measured jet heights were close to the visual jet heights observed through a window in the bed. In two adjacent jets, the jets behave like two isolated jets at low nozzle velocities. As the nozzle velocity increases, the jet heights reach a maximum height in the transition zone. The jets begin to interact after the transition zone, and the jet height becomes a constant. Similar results are obtained for two kinds of perforated-plate distributors. The maximum jet height for multiple, interacting jets is a function only of distributor geometry and is well predicted by a simple geometric model.

The effect of fluid properties on ebulliometer operation

Abstract The simplicity and accuracy of the ebulliometric method for the measurement of vapor pressure, when applied to non-polar or moderately polar substances, is well documented. Unfortunately the accuracy of this technique decreases markedly for high boiling-point polar compounds, a fact not commonly discussed in the literature and not clearly understood. The decrease in accuracy for these compounds is manifest at least in part by a strong relationship between the measured equilibrium temperature and the power input for the electric heater on the ebulliometer. The knowledge of which fluid properties adversely affect the operation of the ebulliometer is important when assessing the accuracy of data obtained by this method and in determining the compatibility of the method for a specific fluid. This study on fourteen pure compounds was conducted to determine the more important fluid properties. It was found that proper operation of the ebulliometer is dependent upon the dipole moment and the degree and type of molecular association of the compound under study. This is in contrast to the commonly held view that the boiling temperature of the substance is of significant importance for proper operation. A correlation is given to screen compounds for ebulliometric study based on readily available data.

Fluid phase stability and equations of state

Abstract Many textbook explanations of fluid-phase stability and phase splitting are based on the assumption that the Gibbs energy of the mixture is a smooth, continuous function of composition that exhibits inflection points and a maximum. The true function, rather, is composed of branches which may not have inflection points and which typically meet at a cusp. Examples of these curves are shown and explained.

A COMPARISON OF DISTRIBUTION FUNCTIONS FOR CALCULATION OF PHASE EQUILIBRIA OF CONTINUOUS MIXTURES

Ill-defined fluid mixtures containing too many components for complete analysis are often described in terms of continuous distribution functions of boiling point or molecular weight or in terms of pseudo-components. For many systems, the choice of the pseudo-component method or one of the standard distribution-function approaches can significantly affect the calculated phase equilibria. For three sample systems, a comparison is made of the pseudo-component method, the method of moments, the Lobatto quadrature method, and our new method in which cubic-spline approximations are incorporated. The accuracy of representation of the composition distribution function is compared, as well as the calculated phase equilibria and CPU times. Although each method has advantages for specific types of systems and calculations, the cubic-spline method was found, in general, to be most accurate and adaptive, with only slight increases in computation times