This thesis contributes to the knowledge and the characterization of small-scale Organic Rankine Cycles (ORC). It is based on experimental data, thermodynamic models and prospective studies.
The experimental studies include:
1. A prototype of small-scale waste heat recovery ORC using an open-drive oil-free scroll expander, declined in two successive versions with major improvements.
2. A prototype of hermetic scroll expander tested on gas test rig designed for that purpose.
The achieved performance are promising, with expander overall isentropic effectivenesses higher than 70% and cycle efficiencies comparable or higher than the typical efficiencies reported in the scientific literature for the considered temperature range.
New steady-state semi-empirical models of each component are developed and validated with the experimental data. The global model of the ORC prototype allows predicting its performance with a good accuracy and can be exploited to simulate possible improvements or alternative cycle configurations.
Dynamic models of the cycle are also developed for the purpose of evaluating the cycle behavior under transient conditions. These models are used to define and compared different control strategies.
The problematics of cycle optimization and fluid selection are treated using the steady-state semi-empirical models. The thermodynamic optimization of such cycles is first demonstrated by practical examples. Furthermore, three different methods for fluid selection are proposed, investigated and compared. Their respective advantages and fields of application are provided.
Finally, two prospective studies of small-scale ORC systems are proposed. The first one is solar ORC designed for the rural electrification of remote regions in Africa. This prototype aims at competing with the photovoltaic technology, with the advantage of generating hot water as by-product.
The second prospective study deals with the recovery of highly transient heat sources. Advanced regulation strategies are proposed to address the practical issues of such systems. These strategies are compared with the state-of-the-art strategies and show a non-negligible potential of performance improvement.