Influence of physical-property variations on leakage for supercritical CO2 seal

Abstract

In the supercritical CO2 Brayton cycle, dynamic sealing of rotating machines is one of the critical technologies. Under drastic variations of physical properties, the flow characteristics, sensitivity of parameters to leakage, and feasibility of theoretical models are not well understood. Experimental parameters of labyrinth seals include upstream parameters (302.15 K to 379.15 K, 7 MPa to 14 MPa) and pressure ratios (1.05 to 1.60, below the critical flow) covering trans-critical and pseudo-critical conditions. A comparison between experiments and traditional models leads to the following conclusions: In terms of linear correlation with the leakage, the upstream parameters in the density form are significantly superior to those in the temperature form. For the expansion factor part, the isentropic form can lead to errors ranging from -25% to 45%. In model prediction, the use of both temperature-based and isentropic-based parameters can result in significant prediction errors ranging from -41% to 67%. This paper recommends Martin's model with the density-based upstream parameters, the iterative isentropic model, and the Childs model, each with an error of ±10% or less.