In the present study, a subcritical and supercritical regenerative organic Rankine cycle (rORC) was designed. The designed rORCs assist a combined heat and power (CHP) engine, the fuel of which is biogas produced from anaerobic digestion of domestic wastes in Belgium. R245fa was selected as the working fluid for both the subcritical and supercritical rORC. During the parametric optimisation, the net power production, mass flow rate, exchanged heat in the regenerator, total pump power consumption, thermal and exergetic efficiencies of rORC were calculated for varying turbine inlet temperatures and pressures. After parametric optimisation of the rORC, the results were compared with the results of the previous study, in which only a simple ORC is analysed and parametrically optimised. Moreover, the effect of the regenerator was revealed by examining all results together. Finally, the exergetic analysis of the best performing subcritical and supercritical rORC was performed. Furthermore, the results of the present and previous studies were considered together and it is clearly seen that the subcritical rORC shows the best performance. Consequently, by using the subcritical rORC, the disadvantages of the using simple ORC (low performance) and supercritical cycle (safety, investment) can be eliminated and system performance can be improved.

In this study, simple and recuperative organic Rankine cycles (ORCs) were designed as the bottoming cycle to a simple gas turbine. After the design, the performance of the ORC engines was analyzed for varying turbine inlet temperature and pressure. During the study, eight different working fluids were selected for the design of the ORCs to identify the best-performing working candidate, namely, benzene, cyclohexane, ethanol, methanol, R21, R152a, toluene, and trans-2-butane. For all turbine inlet pressure values, the maximum performance was obtained from methanol for simple ORC (sORC) and from trans-2-butane for recuperative ORC (rORC). The maximum net power production, thermal efficiency, and exergy efficiency of the sORC were obtained with methanol at 45 bar 350°C, achieving 1967 kW, 24.5%, and 48.9%, respectively. For rORC, the maximum net power production, thermal efficiency, and exergy efficiency were obtained with trans-2-butane at 40 bar 240°C as 2,523 kW, 31.1%, and 62.4%, respectively. The highest net power production, thermal efficiency, and exergy efficiency of the gas turbine (GT)-ORC combined cycle are found at 40 bar and 240°C for rORC, reaching 8,723 kW, 47.63%, and 67.33%, respectively. This means that almost $\mathrm{1,605}\mathrm{kg}\text{-}{\mathrm{CO}}_2/\mathrm{h}$ reduction in ${\mathrm{CO}}_2$ emission is possible with the use of rORC as a bottoming cycle in the GT. When the total heat input from the fuel in the burner of the GT is considered, it is observed that the power production of the GT-rORC combined cycle can be increased up to 40.7%. In order to identify optimum performance under real operating conditions, the designed system can be used by ORC engine designers, manufacturers, and in facilities where simple gas turbines are available.

In this study, the possibility of using Stepwise multilinear regression and deep learning models to estimate the behaviour of the organic Rankine cycle (ORC) has been investigated. It was found that a number of parameters affects the performance of the turbine and hence the amount of power obtained by the ORC. Therefore, limited and simulative parameters might not be sufficient to obtain the best prediction expression. In the present study, the data obtained from a 10 kW ORC system was used as the basis for deep learning models. To this end, the variable selection, which represents the inputs to the neural network, is included in the first steps of a stepwise multilinear regression (SMLR). The aim of the deep learning (DL) models is to use the capabilities of dense layers, and then to strengthen SMLR contributions. The main aim here was to estimate the power generation of the expander, which has an important role in deciding the ORC’s performance. The present study is intended to act as a crucial resource for defining an active estimation procedure for the ORC system through the use of DL. Therefore, an interoperability framework is proposed to estimate ORC power production using SMLR and DL as a new approach in this study. The interoperability approach for the proposed models (SMLR and DL) was found to be successful.