Simple New Algorithm for Distillation Column Design Rafiqul Gani and Erik Bek-Pedersen CAPEC, Dept. of Chemical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark Separation ofa mixture into two or more products through distillation is very common in the process industry. For a given feed mixture and specified product purity requirements, the design of distillation columns typically involves determination of the number of plates, feed-plate location, reflux, and re- boil ratio. Preliminary values of these design variables are often determined by trial and error using the McCabe-Thiele Ž . method McCabe and Thiele, 1925 . Since distillation is an energy intensive process, it is desirable to determine the val- ues of these design variables corresponding to a minimum in terms of cost of operation. Rigorous simulation and optimiza- tion are commonly employed to determine the optimal de- sign. In this article, a simple new method, which is visual Ž . graphical and has a similar starting point as the McCabe- Ž . Thiele method that is, use of vapor  liquid data , is proposed for determination of the distillation column design variables. A novel feature of this method is that the determined values Ž of the design variables correspond to a near-optimal or opti- . mal solution with respect to cost of operation, without re- quiring any rigorous simulation or optimization. The new method is based on identification of the largest driving force, defined as the difference in composition between the vapor and liquid, and its relation to feed-plate location. This driv- ing-force concept is similar to the idea of separation power Ž . Seader and Henley, 1998 . In its present version, this method is applicable to distillation columns with one feed and two products for binary as well as multicomponent mixtures. Theoretical Background Ž . Like the McCabe-Thiele method 1925 , the starting point is the graphical representation of the vapor  liquid data. Ž . However, instead of plotting the vapor composition y i Ž . Ž . against the liquid composition x , the driving force F is i Di Ž . plotted as a function of liquid or vapor composition. F is Di Correspondence concerning this article should be addressed to R. Gani. defined as x  i ij F s y y x s y x . 1 Ž. Di i i i 1q x  y1 Ž . i ij Since energy needs to be added or removed to maintain the existence of the vapor and liquid phases, the value of F Di is indirectly related to the energyadded or removed. If F is Di large, less energy is involved, while if F is small, more en- Di   ergy is involved. A plot of F vs. x for a constant  s3 is Di i 12 shown in Figure 1. It can be seen that F is a concave func- Di tion with respect to x with a well-defined maximum. The i BD and AD lines represent operating lines corresponding y y to minimum reflux, while BD and AD represent operating lines intersecting on the line D  D for a refluxgreater than y x   the minimum. As x ™0 or 1, F ™0. This phenomenon is i Di also observed for other binary mixtures. Figure 2 shows plots of four different types of binary mixtures, with  calculated ij Ž as a function of temperature, pressure, and composition em- . ploying a suitable model . The plots in Figure 2 confirm that the F vs. x function for a nonazeotropic mixture is Di i Figure 1. Driving-force-based separation efficiency for constant  s3. June 2000 Vol. 46, No. 6 AIChE Journal 1271