Community Research and Development Information Service - CORDIS

Abstract

The objective of this project was to develop an expert system that will: a) propose the Best Available Technologies (BATs) by taking into account the energy requirements and the constraints of an industrial system, b) optimize the design parameters for each chosen BAT in a way that they comply with the energy requirements, and c) integrate the BATs in the model of the whole industrial process and select the Best Achievable process Configuration (BAC).

The expert system that was developed in this project consists of: i) the Expert System Shell, ii) the Energy BAT Database that is the "background knowledge" of the Expert System, and iii) the optimization modules that attached to the Expert System will perform the necessary optimization tasks.

The expert system shell is a classic one based on rules of the form "if ... then ...". In order for the shell to be able to select the BATs, the proper rules for the expert system were specified and inserted in it. These rules were identified from the analysis of existing studies conducted by experts for finding the BATs.

Databases of currently available technologies in several areas like combustion, utility networks, diesel engines, gas turbines, etc. were developed. These databases were not developed to be exhaustive, but rather being a good starting point for the development of an exhaustive database. They were developed both in ASCII format, and Microsoft EXCEL format, for higher transportability between different computer systems.

Supporting algorithms were developed for the expert system, that: a) identify the optimal design parameters for the chosen BATs, and then b) find the BAC. These algorithms are based on existing optimization techniques (e.g., MINLP, EMO, Pinch Analysis, etc.). Moreover, a module that provides basic explanations for the choices of the expert system shell was developed.

An alternative expert system shell was also developed, that is based on the optimization method of Genetic Algor

Download application/pdf (47000)

Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top