Academic

Topics:

Current status

Earned an academic degree Doctor of Sciences, Dr., ETH Zurich in 2007. The title of the dissertation is: Deterministic and stochastic batch design optimization techniques.

Abstract of the dissertation

Andrej Mošat`, 2006, Deterministic and Stochastic Batch design optimization techniques, PhD Thesis abstract

Custom chemicals and pharmaceuticals manufacturing is performed very often in multipurpose batch plants. Multipurpose plants offer a high degree of flexibility for the design and operation of chemical processes. As a consequence, problems related to mathematical models of different scenarios such as: design of a single process, design of a new plant, retrofit of an existing plant or scheduling a given portfolio of processes in a given plant are complex and demand substantial computational resources. The mathematical models of a given problem related to batch processing are usually very difficult to formulate. Typically these problems are highly nonlinear, and unless simplified, are believed not to be solvable in polynomial time. The problem complexity even increases if multiple objectives are to be considered. On top of the multiobjective decision problems, the uncertain factors inclusion further complicates the mathematical models.

A solution to the discussed problems is a design, i.e. an allocation of recipe tasks to batch plant equipment. The aim of the proposed algorithms is to optimize the design of a single chemical process to be implemented in an existing multipurpose batch plant. Various objective functions used in the multiobjective optimization are defined as quantitative measures of the quality of such process designs. In this work a Tabu Search metaheuristic method was successfully applied to a wide range of problems.

The main goals of the research posed in this thesis can be summarized as:

1. Formulation of the mathematical problems and development of methodologies supporting the batch plant engineering development team during the process design phase.

2. Proposing a set of solving algorithms for tackling the presented problems.

3. Finding and demonstrating a practical method of reducing significantly the size of the batch design optimization domain. A Superequipment concept was developed as a mathematical model of an equipment class capable of performing any chemical operation class. A Superequipment unit must additionally fulfill the reality criterion, that means it must be transformable into a real equipment unit during and after the optimization. This concept reduces the combinatorial complexity in the solution space significantly.

4. Stipulating and investigating a stochastic mathematical model, related to batch process development, which can be handled by multiobjective optimization in a reasonable time frame allowing rapid result output. Productivity robustness related to a design is defined in the stochastic approach.

5. Automatically selecting a set of feasible and good-performing designs as a basis for the decision making in the early batch process development. The problem domain of early stage batch process development is extended in the presented formulations.

In addition to the optimization of a single deterministic design, the new optimization algorithms assist in: retrofitting of equipment, grass-root design of a new plant, automated production plant line selection or in evaluating a design robustness by an automated stochastic method. The results compilation is also automatically presented as a selection of individual designs, sorted according to prioritized list of objective function values.

The methods, demonstrated on various case studies, show feasibility of the resulting designs, a broad applicability of the methods for automatized integrated process development. The discussed stochastic batch process design approach presents important measures related to the risks in the preliminary design stages.

Research in MPBP Optimization

2002-2006

As a Ph. D. Student of ETH Zurich http://www.ethz.ch, working in the group of Prof. Hungerbuehler, cooperating with Dr. Laurent Cavin and Dr. Ulrich Fischer, we performed research of Multipurpose Batch Plants Optimization.

The research is distantly related to the work of D. W. Rippin, working at the ETH Zurich some 20 years ago.

Programs, tools, technologies

Some of the exciting technologies, algorithms and tools have been used during the members of our group. As we gathered some experience throughout the years, I`ll add time-to-time more information about them, discuss pro`s and con`s, review.

Some of the topics:

Tabu Search

BPD Tool

MPBP Optimization

Economic data

XML

VRML, etc.

Papers

Some publications, posters we've written/published

Slovak Society of Chemical Engineers, Conference in Slovakia, High Tatra`s 2004

Abstract of the conference proceeding:

pdf, Abstract file, 89kb

Full-text of the proceeding:

pdf, Full file, 266kb

About this page

This page contains different sub-categories related to chemical engineering, computer research, batch plant processing, multiobjective optimization, Tabu Search, XML, XSLT, programming.