Oyebolu, Folarin Bolude (2019) Meta-heuristic & hyper-heuristic scheduling tools for biopharmaceutical production. PhD thesis, University of Warwick.

Preview

PDF WRAP_Theses_Oyebolu_2019.pdf - Submitted Version - Requires a PDF viewer. Download (3641Kb) | Preview

Request Changes to record.

Abstract

The manufacturing of biopharmaceuticals requires substantial investments and necessitates long-term planning. Complicating the task of determining optimal production plans are large portfolios of products and facilities which limit the tractability of exact solution methods, and uncertainties & stochastic events which often render plans obsolete when reality deviates from the expectation. This thesis therefore describes decisional tools that are able to cope with these complexities.

First, a capacity planning problem for a network of facilities and multiple products was tackled. Inspired by meta-heuristic approaches to job shop scheduling, a tailored construction heuristic that builds a production plan based on a sequence — optimised by a genetic algorithm—of product demands was proposed. Comparisons to a mathematical programming model demonstrated its competitiveness on certain scenarios and its applicability to a multi-objective problem.

Next, a custom object-oriented model was introduced for a manufacturing scheduling system that utilised a failure-prone perfusion-based bioprocess. With this, process design decisions such as cell culture run time and process configuration, and single-product facility scheduling strategies were evaluated whilst incorporating simulations of stochastic failure events and uncertain demand.

This model was then incorporated into a larger hyper-heuristic to determine optimal scheduling policies for a multi-product problem. Various policy representations are tested and a few policies are adapted from the literature to fit this specific problem. In addition, a novel policy utilising a look-ahead heuristic is proposed. The benefit of parameter tuning using evolutionary algorithms is demonstrated and shows that tuned policies perform much better than a policy that estimates parameters based on service level considerations. In addition, the disadvantages of relying on a fixed or rigid production sequence policy in the face of uncertainty is highlighted.

Item Type:	Thesis or Dissertation (PhD)
Subjects:	R Medicine > RS Pharmacy and materia medica T Technology > T Technology (General) T Technology > TS Manufactures
Library of Congress Subject Headings (LCSH):	Metaheuristics, Pharmaceutical biotechnology, Manufacturing processes
Official Date:	July 2019
Dates:	Date Event July 2019 UNSPECIFIED
Institution:	University of Warwick
Theses Department:	Warwick Business School
Thesis Type:	PhD
Publication Status:	Unpublished
Supervisor(s)/Advisor:	Branke, Jürgen, 1969-
Sponsors:	Engineering and Physical Sciences Research Council
Format of File:	pdf
Extent:	xiv, 171 leaves : illustrations, charts
Language:	eng

Request changes or add full text files to a record

Repository staff actions (login required)

View Item

Downloads