Synthesis of batch processing schemes for the production of pharmaceuticals and specialty chemicals

While the synthesis of continuous processes has been advanced to significant levels of effectiveness, through pure or hybrid implementations of rigorous optimization-based and heuristic approaches, corresponding progress in the synthesis of batch processing schemes has been lagging behind. Most of the research in the batch area has focused on the design of multipurpose or dedicated batch plants, and the optimal planning and scheduling of batch operations in multipurpose plant configurations. Unlike a continuous plant, which is composed of a well-defined network of (mostly) sing!e-function unit operations, a batch process is deployed through a series of batch-operating vessels, which accommodate varying sets of physico-chemical transformations of processing materials. As a result, while the synthesis of a continuous plant leads to a structured flow sheet of unit operations, the synthesis of a batch processing scheme leads to a structured sequence of operating steps. Consequently, we have approached the synthesis of batch processing schemes as a problem of synthesis of operating procedures. Such an operations-centered synthesis decouples the selection of equipment and allows efficient solution of a rather cumbersome combinatorial optimization problem, while allowing the synthesis of novel processes through the grouping of varying sets of operations in the same equipment. The methodology we have developed for selecting and ordering the operations to be performed is based on a combination of Means-Ends Analysis (MEA) and NonMonotonic Planning (NMP). This approach uses means-ends analysis to detect differences between the current process state and the desired product state. When a difference is detected, we apply a non-monotonic planning methodology, which incorporates a combination of heuristic and quantitative methods to construct a plan consisting of operations capable of resolving the difference. The first step is to determine a set of feasible operations that will eliminate the difference. This set of operations is known a,;; the Task Category. The operations are grouped into Task Categories based on the attributes of the tasks they perform. These Task Categories allow us to narrow the number of alternatives that need to be examined. From within the Task Category, a single operation is chosen to be incorporated into the design while all other operations in the set are retained as possible alternatives. In order to determine the applicability of the operation, the preconditions of the operation are assessed against the current process state. If any of the preconditions is violated, operations called "White Knights" are selected and applied before the current operation to alter the current state and remove the precondition violation. The elimination of the precondition violations through the nonmonotonic planning routine leads to a feasible ordering of operational steps which describe aspects of the evolving processing scheme. After all the precondition violations have been removed, the sequence of operations that has been generated is applied and a new current process state is generated. The Means-Ends Analysis with Nonmonotonic Planning (MEA-NMP) methodology is iteratively applied to resolve remaining differences between the new current states and the final product state. The MEA-NMP approach provides us with base case design, as well as, the search space of feasible alternatives by linking the Task Categories together to form the process superstructure. Since the alternatives have been maintained through the identification of the Task Categories, the problem can be formulated as a MINLP and a common MINLP solution strategy can be applied to determine the optimal design. This thesis describes both the MEA-NMP strategy, as well as, its methodological details. We will discuss how the combinatorial optimization problem, defining the synthesis of batch processing schemes, is decomposed into (a) a logic-based component, defining the feasible processing alternatives, and implemented through c:he MEA-NMP strategy, and (b) a reduced MINLP formulation whose solution identifies the optimal processing scheme. In addition, we will discuss the detailed mathematical formulation of the batch process synthesis problem using the finite automaton and the State Task Network, the models employed for the representation of Tasks, the logic that guides the selection of White Knights, and the metrics used for the evaluation of processing alternatives. In the thesis we also outline the computer-aided elements which implement the above ideas within the framework of the BatchDesign-Kit, and will illustrate the application for the MEA-NMP strategy on realistic case studies pertaining to the synthesis of processing schemes for the manufacturing of pharmaceuticals. Finally, we discuss the conclusions and contributions attributed to this work and possible directions for future research that have been brought to our attention during the development of this thesis.

Department dc:contributor.department

Massachusetts Institute of Technology. Dept. of Chemical Engineering

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

1999

Author dc:creator

• Ali, Shahin A

Advisor dc:contributor.advisor

• George Stephanopoulos.

• Chemical Engineering

Statement dc:rights

• M.I.T. theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission. See provided URL for inquiries about permission.

Licence dc:rights.uri

• http://dspace.mit.edu/handle/1721.1/7582

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/1721.1/9653

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/9653