Optimization-Based Analysis of a Simplified Ozone Model

Bialon, P.M. (1996). Optimization-Based Analysis of a Simplified Ozone Model. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-96-134

Preview

Text WP-96-134.pdf Download (288kB) | Preview

Abstract

This paper describes the development of a software optimization tool that aims at finding the optimal strategy for decreasing the ozone concentration in Europe. A simplified ozone model has been used in the research. This model describes the relations between reductions of certain industry pollutants and the resulting ozone concentration abatement, as well as the costs associated with given emission reductions. The task of finding the most cost-effective strategy results in a difficult nonlinear problem comprising thousands of variables and constraints. Moreover, other difficulties can arise, such as issues connected with problem conditioning and solution sensitivity to parameter changes. These potential difficulties lead to the question of software robustness. It was decided to apply three different optimization algorithms in order to ensure that the tool would be able to solve different optimization problems generated during different scenario analyses. The task of joining three different nonlinear solvers (CFSQP, Conopt, and MINOS) with completely different interfaces also turned out to be interesting from the software-engineering point of view. The object-oriented programming technique provided a simple way to divide the software parts into those that were common to all three solvers and those that were solver-specific. The model and the methodological and engineering difficulties in its analysis are summarized in this paper. The paper also contains a description of the software tool developed, as well as conclusions and further suggestions motivated by the first results.

Actions (login required)

View Item