SEPARATION SEQUENCING - USE OF INFORMATION DISTANCE

A fast process synthesis procedure is developed for generating prelimi nary separation schemes. The main approach is based on similarity indi ces obtained from chemical, physical and operating properties between two chemical species in a separation scheme. For the whole set of spec ies, a similarity matrix results. On this basis, algorithms for classi fying chemical species and generating separation sequences are develop ed. A distance between flowsheets is used in order to compare differen t separation schemes with one another or to some reference cases. The weights of the properties and heuristics are adjusted to account for w ell-known optimal cases. Changes in the number of considered separatio n properties and in their hierarchy are also used to improve the proce dure. The algorithm requires very little calculation and can even be i mplemented by hand in the simplest cases. Extensions of the algorithm include the use of actual values of the separation properties instead of boolean ones and the generation of non-sharp flowsheets.