Issues in the design and planning of batch chemical plants

Sriram Subrahmanyam, Purdue University

Abstract

Chemical plants involving batch and semi-continuous processes have become important of late, due to the production of high value products and/or low production quantities (for example, pharmaceutical manufacturing). Batch processes are characterized by the use of general purpose equipment for multiple tasks and therefore, scheduling is vital in the design process. Design, scheduling and selection of an operating strategy must therefore be considered as a single integral problem. A single monolithic mathematical programming formulation for design and scheduling would result in a problem too large to solve using existing computational tools. The proposed approach decomposes the problem into the planning and scheduling stages. The planning and scheduling stages are posed as mixed integer linear programs (MILP). The planning stages consists of aggregate resource constraints which act as surrogate scheduling constraints. The uncertainty in demand and value of products is incorporated in the planning problem through discrete scenarios. First, the aggregate problem is solved for the design solution. The operability (and feasibility) of this design is tested by solving a series of scheduling problems implied by the design solution. A set of heuristics act as feedback to the planning stage if any of the scheduling problems are infeasible. Utilizing the monolithic design-scheduling model, the planning problem is derived and a theoretical relationship established between the aggregate problem and the monolithic model. A hierarchy of aggregate problems is derived, based on which an algorithm is developed to obtain optimal solutions to the overall design problem. The computation efficiency is greatly enhanced through the implementation of a sophisticated distributed algorithm to exploit the parallel structure exhibited by the design and scheduling MILPs. The methodology is validated through a large industrial case study and extended by applying it to other problems in the processing industry (waste treatment options integrated with design, supply chain management, strategic business planning).

Degree

Ph.D.

Advisors

Reklaitis, Purdue University.

Subject Area

Chemical engineering|Operations research

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS