(677d) Design and Implementation of Robust Multicomponent Working Fluid for Organic Rankine Cycle

The organic Rankine cycle (ORC) has been actively developed as a way to generate electricity, using waste or renewable heat resources. Many researches are concentrated on the study to enhance the performance of the working fluid for ORC as it is considered as the most important factor for the overall efficiency of ORC. The design of a working fluid is highly dependent on the characteristics of heat sinks or sources. One common issue on designing ORC utilizing heat sources or sinks, which is not isothermal in heat exchangers, is that it diminishes thermal efficiency by increasing the amount of exergy destruction in heat exchangers. Many studies have shown that it is inevitable to use a multicomponent working fluid to achieve high thermal efficiency [1-3]. This is because, with optimally selected working fluid substances and composition, using a multicomponent working fluid is advantageous in minimizing the exergy destruction in the heat exchanger, since it better fits the temperature glide of the heat source or sink, compared to single component working fluids. In this context, many studies have been performed to develop the method for designing the optimal working fluid mixture for ORC [1-3]. However, although the previous methods for designing the working fluid have succeeded in significantly improving the efficiency of ORCs, the multicomponent working fluid obtained as a result of the designing methods would not be suitable for an actual ORC operation due to its vulnerability to uncertainties in operating conditions and thermodynamic parameters.

Only a few studies have been performed to address the ORC design problem considering uncertainties [4-6] but these studies have common limitations to design robust working fluids. They used Monte Carlo (MC) simulations for evaluating the effect of uncertainties on ORC performance. MC simulations can provide the statistical moments required for stochastic optimization, but its slow convergence rate precludes the optimization from obtaining the robust design in a reasonable amount of time. Also, none of the above studies have considered the uncertainties in the composition of multicomponent working fluid, which can vary due to its thermal degradation or leakage [7].

In this study, a systematic method to design a robust ORC using LNG and ternary working fluid, which yields maximum power output even when the composition of the working fluid and the temperature of the heat source vary from the nominal points during operation of ORC and the values for thermodynamic parameters have ranges has been developed. Since it is computationally burdensome to consider all the uncertainties at once, the proposed method initially seeks the optimal combination for the working fluid mixture by considering only uncertainty in composition. It seeks the optimal composition and ORC operating condition giving the maximum weighted sum of the mean and the variance of the ORC power output. To suppress the factors that adversely affect the operation of ORC (violation of minimum temperature difference in heat exchanger and formation of liquid droplet in expander), the objective function is penalized when they occur. The procedure to derive the statistical moments consists of two steps. Initially, the required heat exchanger area is obtained by simulation of ORC model with a nominal operating conditions (composition, pump discharge pressure, and expander discharge pressure). At the next step, the simulation is carried out again with the obtained area and the varying composition and parameters. The mass fraction of each substance in the working fluid is assumed to follow uniform distribution centered at the nominal point to consider the worst case of ORC operation. To obtain the mean and variance with a small number of simulations, Polynomial Chaos Expansion (PCE) with sparse grid quadrature is employed. It has been shown that small changes in composition and parameters can have serious consequences for stable operation of ORC, and the design of working fluid by the proposed method allows flexible ORC operation despite the existence of uncertainty in the composition [8].

Finally, the proposed design strategy takes into account uncertainties in thermodynamic parameters and heat source in addition to composition. Using the selected working fluid which was turned out to be the most insensitive from the uncertainty of composition, optimization is carried out again when the critical temperature and pressure of each substance composing the working fluid vary within their measurement uncertainty, which can be found in the literature [9]. Also, the uncertainty in the temperature of the heat source is taken into account for further enhancing the operational flexibility of ORC. In sum, since the ORC uses ternary mixture as its working fluid, the design of ORC was performed assuming a total of the nine parameters or design variables under uncertainty (2 composition variables, critical temperature and pressure for each of the working fluid substance, and the temperature of the heat source), which requires excessive amount of computation using the method used when considering only the composition uncertainty. Therefore, the optimization using a surrogate model is devised to efficiently find the optimal and robust ORC design. Because the proposed surrogate model is constructed based on PCE, the statistical moments can be derived analytically, which leads to reduce the time for optimization drastically. Comparing the ORC, which was designed under more uncertainties, to the design obtained under only the composition uncertainty, the former design showed higher output even when the parameters and design variables varied from the nominal points.

[1] Lee, U.; Kim, K.; Han, C. Design and optimization of multicomponent organic rankine cycle using liquefied natural gas cryogenic exergy. Energy 2014, 77, 520−532.

[2] Kim, K.; Lee, U.; Kim, C.; Han, C. Design and optimization of cascade organic Rankine cycle for recovering cryogenic energy from liquefied natural gas using binary working fluid. Energy 2015, 88, 304−313.

[3] Liu, Y.; Guo, K. A novel cryogenic power cycle for LNG cold energy recovery. Energy 2011, 36, 2828−2833.

[4] Frutiger, J.; Andreasen, J.; Liu, W.; Spliethoff, H.; Haglind, F.; Abildskov, J.; Sin, G. Working fluid selection for organic Rankine cycles−Impact of uncertainty of fluid properties. Energy 2016, 109, 987−997.

[5] Mavrou, P.; Papadopoulos, A. I.; Seferlis, P.; Linke, P.; Voutetakis, S. Selection of working fluid mixtures for flexible Organic Rankine Cycles under operating variability through a systematic nonlinear sensitivity analysis approach. Appl. Therm. Eng. 2015, 89, 1054−1067.

[6] Bufi, E. A.; Camporeale, S. M.; Cinnella, P. Robust optimization of an Organic Rankine Cycle for heavy duty engine waste heat recovery. Energy Procedia 2017, 129, 66−73.

[7] Kruse, H.; Rinne, F. Performance and leakage investigations of refrigeration and airconditioning systems using refrigerant mixtures as working fluids. Int. Refrig. Air Condition. Conf. 1992, 621−630.

[8] Kim, K.; Kim, J.; Kim, C.; Lee, Y.; Lee, WB. Robust design of multicomponent working fluid for organic Rankine cycle. Ind. Eng. Chem. Res. 2019, 58, 10, 4154-4167.

[9] Chase Jr, M. W. and Tables, N.-J. T. Data reported in nist standard reference database 69, june 2005 release: Nist chemistry webbook. J. Phys. Chem. Ref. Data, Monograph, vol. 9, pp. 1–1951, 1998.