Abstract

Commercially available dry gas seals (DGS) are the seal-of-choice for shaft-end sealing locations in compressors used in the oil and gas industry, and supercritical carbon dioxide (sCO2) power cycles. Typically, a DGS operates with a very thin gas/supercritical fluid film (typically 2 to 7 microns thin). DGS reliability is tied to how well this ultra-thin film is sustained under varying pressure, thermal, and speed conditions. Recent sCO2 turbomachinery development efforts have experienced a couple of catastrophic DGS failures at the compressor shaft-end locations. A suspected root cause of these DGS failures is thermal deformations caused by the excessive windage heating expected with the supercritical CO2 working fluid when the compressor operates at high pressures and high rotational speeds. In this paper, the thermal behavior of a DGS operating in a typical sCO2 compressor is investigated. Specifically, test data are presented for a specially instrumented, commercially available DGS operating in the GE-SwRI sCO2 compressor. The test data include temperatures measured on the seal stationary ring and DGS housing locations using several embedded metal thermocouples to monitor the thermal behavior of the DGS during typical sCO2 compressor missions. This paper describes how an existing GE-SwRI sCO2 compressor housing was modified to accommodate and route temperature sensors to the instrumented DGS. Test data shows higher than expected temperatures (about 180 to 190 °C) in the cavities surrounding the DGS, which provides useful insights to turbomachinery designers for designing seal cavities. The main reason for these high temperatures is the low sCO2 flow purging the seal cavity. Furthermore, the measured temperatures are compared with the predictions of a steady-state thermal model of the GE-SwRI sCO2 compressor. The thermal model uses a 1D flow advection network (accounting for cavity swirl, windage and heat transfer coefficients for sCO2 flow, air flow and bearing oil flow) exchanging heat with the surrounding structure, along with heat transfer through the structure. The thermal model uses CFD-based flows to account for the sCO2 flow past the DGS. The predictions of the thermal model match reasonably well with the measured temperature data, thereby providing validation to the modeling assumptions. The high structural temperatures in the regions around the DGS that are predicted by the model as well as measured during the tests point to the importance of devising cooling strategies for reliable operation of the DGS.

This content is only available via PDF.
You do not currently have access to this content.