NUMERICAL PROCESS AND PLANT-SIMULATION METHODS

This paper represents an attempt io explicitly map the conceptual acti vities which constitute a process design task into a series of well-po sed, complete and general numerical procedures. In spite of the fact t hat there is a substantial number of design manuals and design procedu res which process engineers can consult to choose the most proper appr oach to a particular problem, there is a remarkable lack of generality on the one side (procedures which apply to the design of a wastewater treatment process cannot be used even in principle to design a fluidi zed bed system), and an obvious repetitivity on the other side (all pr ocedures involve mass and energy balances at some point). With the adv ance of numerical techniques, virtually every procedure has been compu terized, so that engineers can avail themselves of a multitude of comp uter tools in the majority of their process design activities: as a re sult of the lack of coordination among different procedures though, th e situation for what codes are concerned is very confusing: there are many codes which per form nominally the same task, giving (sometimes s ubstantially!) different results when applied to the very same problem ; each code uses its own set of property tables, its own I/O format, e tc. Finally, with very few exceptions, these codes are not mutually co mpatible, i.e. the output from any of them cannot be used as the input to any of the other, not only because of the respective formats, but rather because the quantities taken to represent a certain physical pr ocess are not the same in different codes. This is a very unsatisfacto ry stare of affairs, both for the final user and for the software prod ucer. Unfortunately, this seems ro be a problem for which commercial, technical and historical reasons make it very difficult to find a solu tion in the short term, The author is convinced that an Parry exposure to this problem and to a new approach to its solution can only benefi t our engineering students, and has therefore endeavoured, together wi th his coworkers, to devise a ''modular approach'' to the solution of process simulation problems. The material presented here has been orig inated by a series of lectures and seminars developed in the last thre e years for master and doctoral level students in Mechanical Engineeri ng. The result of this ''distillation'' process maybe yet: unripe, as they are definiteIy not complete: but the implications, also in terms of practical applications, are very promising, and the approach deserv es more attention in the future. The order of presentation of the mate rial is historical/logical: it begins with the old-fashioned slide-rul e calculations and proceeds towards the most recent developments of AI -based methods. The path is made clear from the very beginning: we are trying to extract from the various engineering activities all the ess ential knowledge which pertains to the engineer himself, with the fina l goal of transforming this body of knowledge - in some form suitable to machine communication - to a ''universal-process simulator'', which can then be applied with a high degree of confidence to a variety of particular process simulations.