Matthias Bendig
Developing a methodology that allows identifying maximumelectricity production with the help of Organic Rankine Cycles (ORC) from the waste heat of an industrial process, at the lowest specific cost, without jeopardising the increase of the industrial process’s thermal efficiency. Such is the goal of this thesis. In order to reach this goal, a software tool which is able to identify the most suitable cycle for a heat source regarding electricity production and cost efficiency is developed and explained in detail. Threemain areas of research can be identified, the definition of waste heat, the analysis of experimental data with the help of data reconciliation and the identification of a suitable working fluid and operation parameters of an ORC for any given heat source. A method for identifying, characterising and quantifying the available waste heat, which can be converted into a useful form, for any industrial system, is presented. A distinction is made between avoidable and unavoidable waste heat. Combined pinch analysis and exergy analysis is used to characterise the waste heat potential of a given process. Based on the study of two industrial processes, it will become clear how the constraints of a waste heat analysis influence the outcome and the potential for the integration of ORCs. The studies illustrate how the increase in energy efficiency and degree of heat recovery and integration of a process can be contradictory to the production of electricity with ORCs. The concept of data reconciliation is needed for the measurements, collected from two ORC demonstrators. Apart from "classical" data reconciliation, a new method is presented, which consists in including parameters as virtual measurements and time dependency of measurements. This increases the redundancy of the system and thus the overall accuracy of the reconciled values. A methodology, capable of choosing the designpoint, a suitable working fluid and a cycle configuration, for the lowest specific investment cost while at the same time maximising the electricity output, for a given process environment is developed. The new methodology uses a multi objective optimisation (master optimisation) andMixed Integer Linear Programming (MILP) problem (slave optimisation). A novel approach of combining multi objective nonlinear master optimisations with single objective linear slave optimisations is introduced, increasing the reliability of the algorithm.
EPFL2015
