Study of improved design methodology of $S-CO_2$ power cycle compressor for the next generation nuclear system application

Abstract

Since the supercritical $CO_2 (S-CO_2)$ Brayton cycle has advantages in economics due to high efficiency and compactness, various industries have been trying to develop baseline technology on the design and analysis of the $S-CO_2$ Brayton. As an application of $S-CO_2$ Brayton cycle to nuclear power system, the $S-CO_2$ Brayton cycle was coupled to a Sodium cooled Fast Reactor (SFR), which is one of the next generation nuclear reactors, to en-hance safety and economy. However, one of the challenging issues in the $S-CO_2$ Brayton cycle development is designing a compressor operating near the critical point of $CO_2$. Since thermodynamic property variation of $S-CO_2$ near the critical point shows non-ideal gas behavior, $S-CO_2$ compressor design and analysis methodology should be re-constructed to reflect the real gas effect properly. To overcome limitations on conventional ideal gas based design methodolo-gies real gas compressor design tool, KAIST_TMD, is constructed in this work. The KAIST_TMD is based on 1D mean stream line analysis and models for design and performance prediction are selected from open litera-tures. To reflect real gas effect in the KAIST_TMD, an enthalpy based calculation procedure is established and REFPROP thermodynamic property database is coupled to the code directly. The models in the $S-CO_2$ compressor design methodology are tested through KAIST_TMD and models were carefully selected through the comparison with experiment data. The KAIST_TMD is also utilized for non-dimensional design map generation to confirm the validity of the traditional design map used for the conceptual design phase. Due to flexibility of the design map generation method developed in this work, a realistic conceptual design and performance prediction can be supported for the future cycle design process. Lastly, a design study of $S-CO_2$ compressor for the SFR application is conducted with two different design options and design and performance prediction were successfully performed.