ANALYSIS OF WASTE VITRIFICATION PRODUCT-PROCESS SYSTEMS

Integrated product and process design has been approached by represent ing product specifications and process operations in a common space of mixture compositions. Interval reasoning can be a useful preprocessin g step prior to detailed design in situations where there is uncertain ty in feed composition, or it is sufficient to be within a given range of values for product properties. This paper presents new interval re asoning algorithms for abstract representations of unit operations suc h as reactors and separators. These methods enable the propagation of product specifications, given as ranges on properties and compositions , to ranges on feed conditions, and vice versa. Comparisons of differe nt processes may also reveal relative preferences between systems. In this paper, we demonstrate these concepts and illustrate their applica tion through a waste vitrification system example in which feed stream composition, product composition, viscosity and durability are all co nstrained. Two different vitrification systems are compared in terms o f their ranges of acceptable feed compositions and the range of possib le product mass. The determined ranges on variables provide a superset of system capabilities; points outside this space are proven infeasib le, while points inside the space are possibly feasible. This paper de monstrates that interval propagation schemes that make specific use of the structure inherent in chemical engineering processes, such as con servation of mass, can be very effective in limiting the spurious incl usion of infeasible points. In addition, redundant equations are valua ble in reducing spurious solutions-as has been shown elsewhere for oth er qualitative and semi-qualitative reasoning schemes. (C) 1998 Elsevi er Science Ltd. All rights reserved.