(460a) A Simple New Concept for Chemical Process Flowsheet Design and Analysis

Synthesis, design and analysis of chemical products involve the identification and analysis of molecules and their mixtures with desired (target) properties and performances, as in drugs, pesticides, solvents, aroma and food products. In each case, molecules that are likely to match the target properties and performances are identified, usually in experiment-based trial and error solution approaches. Application of techniques such as computer aided molecular design (CAMD) are also becoming increasingly popular for solution of these problems. The main features of these techniques are the following: building blocks (atoms and functional groups) to generate and represent molecules; group-contribution (GC) based property models; a standard molecular structure notation system, such as SMILES, to store and visualize the molecular structure information; and a screening system to identify the molecules that match the target (design) properties. The objective of this paper is to present a simple new concept for process flowsheet synthesis, design and analysis that works in a similar fashion as CAMD. That is, like molecules, flowsheets are generated and represented by functional process-groups; process-group contribution based property models (for example, for energy consumption) is used flowsheet property estimation; similar to SMILES, a notation system for storing the flowsheet structural information, called SFILES is used; and, the same screening methods used in CAMD are also employed here to find the feasible flowsheets. Like functional groups in molecules, each process-group is characterized by an atom-like parameter (type of unit operation) and a molecular weight like parameter (driving force related to the operation). As in pure component property estimation with GC-methods, once a flowsheet is represented by a set of process-groups, the types of unit operations present in it are immediately known together with the driving forces for each unit-operation. This information is then used by a GC-like property estimation model (summation of contributions of all groups present to obtain the desired property). Using a table of regressed contributions for each group, the property (such as the energy consumption of a flowsheet) is very easily estimated. Thus, given a set of process-groups and product specifications, they are combined to generate feasible flowsheets, the flowsheet information are stored as a simple one-line code (SFILES, from which flowsheets are visualized), and, its property is estimated to identify the optimal flowsheet (with respect to the flowsheet property), without practically any simulation of the process flowsheet. By definition, the energy consumption is lowest when the driving force is the largest. Finally, a software for computer aided flowsheet design (CAFD) has been developed. The main limitation, like CAMD, is the availability of enough process-groups and the corresponding property models. Current and future work is creating more process-groups and property estimation models. Through CAFD, a large range of process flowsheet design and analysis can be solved. An additional feature, different from CAMD is that using the driving forces for each process-group and the corresponding inlet-outlet information, a reverse design technique is used to back-calculate the set of design variables that will define the operation of the corresponding unit operation. For example, in the case of conventional distillation, the driving force is related to the relative volatility and using these as known values, the reflux ratio, number of stages and feed location is determined for specific values of the inlet-outlet streams. Thus, for different values of relative volatility, the corresponding driving forces are calculated together with reflux ratio, number of stages and feed location for different sets of product specifications. Note that these calculations are independent of the identities of the chemicals involved. The problem can be solved for fixed values of driving forces (when the chemical identities are not known), or, for fixed values of relative volatilities (when the chemical identities are known). Using the relation between relative volatilities and driving forces, the unknown variable is calculated. Once both these values are known, a table of precalculated values of design variables is used to identify the values matching the design target. Similar methods have been determined for design of solvent-based extractions (distillation, liquid-liquid extraction, crystallization), reactors and membrane-based separations. This has also generated process-groups representing these unit operations (involving 2 or more compounds). The paper will present the main features of the process-group based concept for chemical process flowsheet design through illustrative case studies (covering a wide range of chemical processes), some of which correspond to well known examples reported by other workers. These case studies help to confirm that the new concept is able to find the same results as those reported earlier. In addition, it is also able to find new alternatives not reported earlier. The application examples will also highlight the use of the CAFD software and the visualization/storage of the flowsheet through the SFILES notation system. A great advantage of the SFILES notation system is that thousands of flowsheets can be stored, their data retrieved and visualized almost instantly. A unique feature of this design concept is that it is simple, easy to apply and is truly predictive.