(79d) A Tray Column Design Method Incorporating a New Simple Efficiency Model

The concept of efficiency in distillation seems first to have been articulated in an article by W.K. Lewis published in Industrial and Engineering Chemistry in 1922. Since then that journal and its descendants have published many of the key papers on the topic of distillation tray efficiency (Murphree 1925, Lewis 1936, Chan & Fair 1984, Prado & Fair 1990, Garcia & Fair 2000). This paper starts with a very brief review of that history. This will bring us to a summary of our own recent work on efficiency in which we have first revised the famous and very simple O’Connell correlation to improve its accuracy as well as to broaden its applicability (AIChE Meeting, Spring 2017; CEP, July 2018).

Subsequently, Duss and Taylor (Distillation and Absorption 2018, Florence, September 2018) presented a new model for tray efficiency that uses the traditional concepts of numbers of transfer units for each phase but that is simpler than any analogous model in its demands for basic information and more accurate even than many more sophisticated models.

Any model based on the traditional concept of addition of resistances necessarily requires the slope of the equilibrium line. This, too, is a challenge for any models using the slope because it is the slope of a surface in hyperspace that is required and not so easy to estimate. Taylor and Duss (AIChE Meeting, Spring 2017) introduced the concept of Design Components as the basis for a new conceptually simple (but slightly computationally demanding) method of estimating the slope. In a more recent paper Taylor and Duss (Distillation and Absorption 2018, Florence, September 2018) provided an extremely simple equation to accurately approximate this slope that needs nothing more than the results typically immediately available from a column simulation.

What we attempt in this paper is to bring together these differing contributions to construct a complete and coherent tray design methodology that can take the designer from the early stages of a column simulation to preliminary sizing calculations all while using the least amount of information possible.

Missing from our earlier work is any discussion of physical property estimation. The simple efficiency model of Duss and Taylor requires only the gas and liquid phase viscosities. However, other popular models also require a diffusion coefficient. Diffusion in multicomponent systems is a complicated subject whose mathematical sophistication most often is ignored in column design work. This, more or less inevitably leads to the use of a pseudo-binary method of estimating diffusivity. Thus, we also compare and contrast some simple approaches to obtaining this quantity in a way that is as unambiguous as possible.

Our complete design procedure is illustrated with some practical examples.