James M. Douglas

A prototype expert system for synthesizing chemical process flowsheets

Abstract PIP (process invention procedure) is an hierarchical expert system for the synthesis of chemical process flowsheets. It uses a combination of qualitative knowledge, i.e. heuristics and quantitative knowledge, i.e. design and cost calculations, arranged in a hierarchic The heuristics are used to select the unit operations, to identify the interconnections between these units, to identify the dominant design variables and to identify the process alternatives at each level of the hierarchy, while the quantitative models are used equipment sizes and costs, the raw material and utility costs and the process profitability at each level. A hybrid, expert system control architecture was developed for PIP that allows these two types of knowledge-bases to interact in such a way that the he PIP attempts to invent a flowsheet using a depth-first strategy, where one of the goals is to see if there is some reason why none of the alternatives Thus, it attempts to complete a design before any alternatives are considered. If profitable operation is observed over some range of the design variables at a particular level, then PIP proceeds to the next level and adds more de with the ones found at the earliest level, (these normally have the greatest impact on the processing costs) and if no profitable alternative can be fo Hence, PIP allows a design engineer to invent initial chemical flowsheet structures rapidly (in about 1 h vs 2 days), to estimate the optimum design co

A hierarchical procedure for the conceptual design of solids processes

Abstract A systematic procedure for the conceptual design of vapor—liquid—solid processes has been developed. In an evolutionary manner, it leads the user to select process units, to identify the equipment configurations and to determine the important design variables and the associated economic trade-offs. At present, the procedure is limited to low-molecular-weight solids and does not handle vapor—solid processes. This hierarchical procedure has been implemented in a computer code to aid in the development of preliminary process flowsheets for solids processes. A number of case studies have been analyzed, including the manufacture of bis (α-hydroxy ethyl)terephthalate from terephthalonitrile—selected for reporting in this article. Design variables unique to this solids plant—dilution ratio, wash ratio and crystallizer temperature—have been identified and their effect on overall plant economics quantified.

Automation in Design: The Conceptual Synthesis of Chemical Processing Schemes

Publisher Summary This chapter shows how one can use ideas and techniques from artificial intelligence, such as symbolic modeling, knowledge-based systems, and logic, to construct a computer-implemented model of the design process. By using the Douglas hierarchical approach as the conceptual model of the design process itself, the chapter shows how to generate models of the structure of design tasks, design decisions, and the state of design, thus leading to automation of large segments of the synthesis of chemical processing schemes. The result is a human-aided, machine-based design paradigm, with the computer “knowing” how the design is done, what the scope of design is, and how to provide explanations and the rationale for the design decisions and the resulting final design. The chapter argues that the human-aided, machine-based design is the paradigm that will characterize future design systems, where rapid conceptualization and prototyping of engineering artifacts are the source of competitive edge. The chapter further provides a formalized restatement of the hierarchical procedure, in an effort to cast it more closely to a computational process for its computer-based automation. The structure of a conceptual model that can be used to represent a design methodology has been discussed but the value of this model rests with the effectiveness of the representation schemes that one employs to describe the declarative and procedural components of the model in a way that the computer can “understand.” It has been emphasized that hierarchical design language is a formal framework for the development of the computational process that emulates the Douglas methodology for conceptual process design and its specific characteristics have been discussed.

The interaction between separation system synthesis and process synthesis

Abstract The problem of selecting a separation system for a process cannot be uncoupled from the problem of determining the optimum process flow rates, since the optimum flows normally involve a trade-off between raw materials costs and recycle (which includes separations) costs. An efficient way of obtaining a first solution of the coupled problem is to use shortcut procedures to get into the neighborhood of the optimum design conditions and to eliminate undesirable alternatives. Some of the available shortcut procedures and design heuristics are reviewed.

Analysis of process operability at the preliminary design stage

Abstract A hierarchical procedure is presented for assessing the steady-state operability of a process. The method can be applied as new processes are being designed, or it can be used as a decomposition procedure for studying the operability of existing processes. The results indicate that operability alternatives are often encountered, i.e. different problem formulations are obtained depending on the assumptions that are made. In addition, the analyses indicate that the equipment design should be based on the mean value of the disturbances, the “worst-case” conditions or the expected profit, depending on the particular equipment in the flowsheet.

25 The Paradigm After Next

Publisher Summary It is anticipated that in the future, concepts from artificial intelligence will play a major role in chemical engineering. Thus, process synthesis, design, and control can be exciting research areas in the years ahead. This chapter presents a different view of chemical engineering paradigms, to extend Westerbergs discussion of process synthesis and describes some additional research opportunities. It also proposes some curriculum modifications. It is expected that the emerging paradigm 20 or 30 years from now will be process engineering, which will include industrial chemistry, industrial biology, and industrial materials.

Operating Heuristics for the Control of Complete Chemical Plants

The use of heuristics has become commonplace in the design of a chemical process, but only rarely have heuristics been published to characterize the trade-offs for optimum steady-state control. Evaluation of a number of case studies has led to operating heuristics for an important class of reaction systems and common equipment types. Often, these heuristics suggest a logical set of control structure elements based on process economics.

Estimating Bounds for Process Optimization Problems. An Analogy to Rules of Thumb and Structural Modifications

The cost surfaces for many process design problems are often very flat in the neighborhood of the optimum. Thus, it might simpler to bound the minimum cost from above and below, instead of determining the exact optimal value. Two procedures for bounding the minimum cost of a train of waste-heat boilers are discussed in this paper. The first approach merely represents an approximate solution of the classical calculus problem. The second approach is a modification of the two-level method of optimization, which always enables us to find a tight lower bound on the minimum cost and is useful for evaluating some process alternatives. The predicted bounds depend on the design and cost parameters, so that the bounds are simple to update as energy prices change. These techniques provide better estimates of the optimum approach temperatures in waste-heat boilers than previously published rules of thumb. 13 refs.