AUTOMATIC GENERATION OF LOCAL THERMODYNAMIC APPROXIMATIONS IN PROCESS SIMULATION
Abstract
Methods of approximating the constituents of the K-value relation, namely vapor pressure, vapor phase fugacity coefficient, and liquid phase activity coefficients have been developed for use in dynamic and steady state simulations. The location and size of multiple approximating regions are automatically selected to cover the regions of composition temperature state space that are required for a given simulation so as to meet user specified error criteria. Emphasis is placed on using equations with thermodynamic character and with an appropriate number of parameters that are determined by matching rigorous property function values and first derivatives with those of the approximation equation. The error of approximation is estimated, with evaluation of a minimal number of additional rigorous function values, to adaptively control the size of the approximation region. One of the major advantages of using the approximation functions is the availability of analytical composition and temperature derivatives for use in an implicit integration method or for solution of algebraic equations via Newton's method. The approximation methods have been tested on batch and steady state distillation problems, a small steady state flowsheet with recycle and a liquid-liquid extraction problem. Results indicate that a savings of at least 80% can be made in the number of rigorous physical function evaluations. The work significantly advances the ideas initially proposed by Barrett and Walsh in its adaptive mechanism as well as the nature of the approximating and error estimating functions.
Degree
Ph.D.
Subject Area
Chemical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.