Abstract
Compiled in sufficient quantities, the effectiveness of design heuristics (knowledge based rules) to help identify pollution prevention alternatives has been demonstrated. In this paper, a core structure of associated heuristics and procedures developed using chemical engineering principles are presented for use during conceptual as well as retrofit design. The approach is applicable for continuous bulk processes and has been encoded in the software P2TCP (Pollution Prevention Tool for Continuous Processes), a prototype expert system, to help develop inherently cleaner processes.

Unlike hierarchical or step-wise design techniques, heuristics are presented that help identify potential unit operations and analyze their interactions independently within the reaction and separation systems of a chemical process. Interactions between the reaction and separation systems are then taken into account to further reduce the number of overall flow diagrams requiring subsequent evaluation. Future extensions to the knowledge base and combination with complimentary tools like process simulation packages are envisioned.
 

Introduction
Opportunities can be identified to reduce environmental impacts caused by the consumption of energy and the generation of wastes in chemical processes. Associated modifications are typically most "effective" during the conceptual stages of design. However conceptual design is often limited by resource constraints and driven principally by economic considerations. The identification of pollution prevention opportunities is still not typically practiced consistently or routinely (Fromm 1992; Hethcoat 1990). Similarly, available pollution prevention texts (e.g. IChemE 1995; Mulholland and Dyer 1999) focus predominantly on methodology but provide only limited depth in terms of the knowledge used to actually generate alternatives.

To generate alternatives for pollution prevention, a significant number of skilled man-hours are usually required, opportunities may not be systematically identified, energy consumption may not be considered and opposing effects may be overlooked. The use of an appropriate computer based tool can help reduce these problems. In this paper, a core structure of heuristics (knowledge based rules that result in qualitative advice) and procedures are presented that can help systematically identify process flow diagram alternatives in the context of pollution prevention. These heuristics and procedures were encoded to provide an initial prototype expert system, the Pollution Prevention Tool for Continuous Processes (P2TCP) (Pennington 1997).

Design knowledge is often represented in the form of heuristics, as outlined in concept by Smith (1995), the IChemE (1995), Rossiter (1995), El-Halwagi (1997), Allen and Rosselot (1997), Mulholland and Dyer (1999), Berger (1999) and others. Douglas (1985) demonstrated that, once sufficiently developed, sets of heuristics can be an effective means for the rapid identification of process options to help synthesize design in terms of potential capital and operating cost optima. Similar knowledge can be used to analyze a process for pollution prevention opportunities (Douglas 1992). However, as the focus is changed to the context of pollution prevention, the objective becomes the generation of alternatives for reduced waste and energy consumption. Hence, the heuristics are not necessarily identical with those used for economic synthesis or analysis.

An initial prototype set of heuristics and approaches is presented in this paper that were developed using chemical engineering principles for continuous processes in the context of pollution prevention (Pennington 1997). These heuristics can help address the identification of unit operations and their potential interactions, as in the development of a conceptual process flow diagram. Both energy consumption and waste generation are considered. Input requirements include basic physical-chemical properties commonly used in design and reaction pathway synthesis. However, no base case flow sheet is required, hence the resultant expert system can be used during conceptual as well as retrofit design, facilitating the development of inherently cleaner processes. Complimentary heuristics, for example to identify opportunities associated with the detailed design of unit operations (e.g. Butner 1997) or the selection of treatment techniques that do not transfer wastes from one medium to another (e.g. Berger 1999), are addressed elsewhere in the literature. An illustration of how the proposed pollution prevention heuristics can compliment existing design methodology and tools is presented in Figure 1.

The heuristics and procedures presented are represented in a framework similar to step-wise or hierarchical approaches often used for the synthesis of process flow diagrams, as described by Douglas (1985), in the context of overall synthesis, and Barnicki and Fair (1990, 1992), for the design of separation systems. However using these hierarchical approaches may not result in the identification of all feasible options, even if used iteratively. This is predominantly associated with limitations of the design knowledge represented but also the methodology. Performing the analysis independently for the reaction and separation systems in a process followed by consideration of potential interactions can increase the number of alternatives identified with the same knowledge base. Iterative application is not required. An analogous approach can be used to identify separation system options.

Pollution prevention opportunities have been successfully identified using the procedures and heuristics presented in this paper (Pennington 1997). Alternatives for a number of continuous chemical process flow diagrams and separation systems were subsequently compared using simulation tools. Processes considered included the production of allyl chloride, the manufacture of chlorobenzenes and the separation of a mixed hydrocarbon stream generated in naphtha reforming units of oil refineries. However, it should be noted that the heuristics and procedures presented only provided a basis for the construction of an initial prototype expert system. Further research areas and validation needs are highlighted in this paper.
 

Conclusions
A prototype expert system has been developed to help designers of continuous chemical processes systematically identify process flow diagram alternatives that may result in reduced waste generation and energy consumption. Advantages of the expert system include:

• the analysis performed is not resource intensive and does not require the prior specification or simulation of a process flow diagram; it will therefore help identify pollution prevention opportunities during conceptual as well as retrofit design.

• the heuristics were derived systematically using fundamental chemical engineering principles; the structure developed can therefore provide a greater depth of analysis and more structure than key-word approaches or other similar brainstorming techniques.

• each system (reaction and separation) of a process is analyzed independently, not sequentially, and then potential interactions are considered to help ensure all options are identified without iterative application.

It should be noted that the scope of the prototype was restricted to the development of a core knowledge base. As with any tools, the results should not be accepted blindly. Further enhancements and research can transform this prototype into a computer-based system for the identification of pollution prevention options with significant commercial potential. Efforts to enhance the knowledge base include validation of the separation heuristics, extension of the reaction system heuristics to non-homogenous pathways, expansion of the system interaction heuristics and the addition of heuristics to identify potential treatment options that do not result in pollutant transfer. Focus on the development of heuristics to identify opportunities for the use of "advanced" technologies, for example combined reaction-separation techniques, is considered a high priority. Peripheral research efforts include the combination of the expert system with chemical process simulation packages and other associated pollution prevention tools, including multi-objective design making techniques that account for both environmental impacts as well as economics.