Paula Bettio Staudt

Phase stability analysis using a modified affine arithmetic

Abstract Phase stability analysis is a crucial step in the determination of multiphase equilibrium. This analysis by the tangent plane distance (TPD) minimization is a well-known technique, as well as the difficulties in providing guarantees that the global minimum has been found. On this regard, interval methods are powerful tools since they provide such guarantees. In this work, an interval Newton method plus generalized bisection, based on a modified affine arithmetic, is used to reliably find all possible stationary points of the TPD function. Additionally, an improved convergence test is suggested as well as a special treatment for mole fraction weighted averages. Several mixtures with up to 5 components, including LLE island type ternary systems, were studied. Both activity coefficient models and cubic equations of state were considered. For all the cases tested, the proposed modified affine arithmetic method was superior to other interval-based methods.

Reliability vs. Efficiency When Solving Multiphase Equilibrium Problems with Hybrid Optimization Codes

Abstract Computation of phase equilibrium is a very important and frequently encountered problem in process systems engineering. Accurate and robust flash routines are at the core of the major part of chemical engineering design applications, ranging from pipelines to distillation columns, chemical reactors, and oil and gas production. The multiphase equilibrium calculation by the direct minimization of the Gibbs free energy has some potential advantages over classical methods, mainly because the equality of chemical potentials is a necessary but not a sufficient condition for equilibrium. Recent developments in global optimization and the availability of very fast computers have stimulated the research on the direct minimization approach. In this work, a hybrid method (genetic algorithm for the global search and interior point for refinement) was implemented and the efficiency/reliability relation was studied. For some systems the global minimum can be found in a NFE compatible with the methods currently available in process simulators. But, it was also found that, for some systems, there is an asymptotic behavior. Apparently, a reliability of 100 %, when using global search algorithms with some random nature, would be found only for an infinite number of function evaluations.

Performance of predictive models in phase equilibria of complex associating systems: PC-SAFT and CEOS/GE

Cubic equations of state combined with excess Gibbs energy predictive models (like UNIFAC) and equations of state based on applied statistical mechanics are among the main alternatives for phase equilibria prediction involving polar substances in wide temperature and pressure ranges. In this work, the predictive performances of the PC-SAFT with association contribution and Peng-Robinson (PR) combined with UNIFAC (Do) through mixing rules are compared. Binary and multi-component systems involving polar and non-polar substances were analyzed. Results were also compared to experimental data available in the literature. Results show a similar predictive performance for PC-SAFT with association and cubic equations combined with UNIFAC (Do) through mixing rules. Although PC-SAFT with association requires less parameters, it is more complex and requires more computation time.

DYNAMIC SIMULATION OF REACTIVE DISTILLATION PROCESSES TO PREDICT START-UP BEHAVIOR

Abstract In the last years, the integration of reaction and distillation in one process, has gained more interest. The mixing of these two operations is especially convenient and shows many advantages. The startup procedure of a conventional distillation column spends a lot of time and energy besides producing offspec products. In reactive distillation, this problem becomes worse once the product cannot be easily recycled, making the start-up procedure very costintensive. In this work, the dynamic simulation of reactive distillation columns is studied in order to analyze different startup procedures. Different strategies were implemented and simulated. Furthermore, an evaluation of uncertainties in model parameters on the startup time was carried out.