Low-leakage film-riding seals are a key enabling technology for utility-scale supercritical carbon dioxide (sCO2) power cycles. Fluid film-riding rotor-stator seals (operating with sCO2 as the working fluid) are designed to track rotor movements and provide effective sealing by maintaining a tight operating clearance (of the order of several microns) from the spinning rotor. Thin film-riding seals generate viscous shear heat during high-speed operation, and the reliable operation of such thin-film seals depends critically on the designer’s ability to control the thermal deformations of the seal/rotor bearing face, which in turn are tied to the designer’s ability to understand and predict the heat transfer across the seal bearing face. In this paper, we develop a simple axisymmetric thermal-mechanical model of a typical face seal to highlight how the uncertainty in heat transfer coefficient (HTC) on the seal bearing face drives uncertainty in seal deformation predictions, especially when the HTCs are an order of magnitude lower than those predicted with duct-based Dittus-Boelter correlations. This uncertainty in seal bearing face HTCs drives the need for an experimental quantification of HTCs in high-aspect ratio thin films associated with low-leakage film-riding seals. In this paper, we describe a non-rotating experimental test rig designed for estimating the HTCs on the seal bearing face using a shim-heater technique along with IR-camera-based temperature measurements. The experimental set-up consists of a thin metal shim (representing the seal bearing face) forming one wall of a pressurized duct with geometric similarity to a typical thin film of a face seal. Pressurized airflow past the shim is used to simulate the flow field expected in a non-rotating seal. The HTC test data for a non-rotating film (as against the actual seal film with rotating fluid) are lower than the actual seal, and establish a lower bound on the HTCs. This is especially useful for bounding the seal deformation uncertainty, which is vulnerable to the HTCs in the low-HTC regime. We present representative test data that is non-dimensionalized using radial-flow-based Reynolds number and compare these HTC estimates both with the predictions of Dittus-Boelter type correlations, and with the predictions of a 3D computational fluid dynamics (CFD) model. The purpose of the CFD model is to develop a HTC prediction tool for such thin-film surfaces, and the test data are used for validating this predictive model.
A rayleigh-step face seal with reverse steps.