Test Results for the Static and Rotordynamic Characteristics of a Long (L/D = 0.75) Smooth Seal in Two-Phase (Mainly Gas) Conditions With a 62-Bar Inlet Pressure

   Recent multiphase-pump developments encountered several rotordynamic issues with smooth balance-piston seals, creating a need to better understand the performance of annular seals under multiphase-flow operation. This paper presents measurements of static and dynamic characteristics of a long smooth seal (L/D = 0.75, D = 114.686 mm, and C_r = 0.200 mm) operating under pure- and mainly air condition with a mixture of air and silicone oil (PSF-5cSt). Tests are performed at a supply pressure of 62.1 bars-a, 3 rotation speeds (5, 10, 15 krpm), 3 pressure ratios (PRs) (0.6, 0.5, 0.4), for a range of inlet liquid volume fraction (LVF_i) from 0% to 8%. The test fluid is injected radially upstream of the test seal with no intended inlet preswirl. The results are then compared to: (1) the previous test reported in 2017 by Zhang et al. with similar testing condition but a different seal geometry and (2) the prediction from a bulk-flow model developed in 2012 by San Andrés. The 2017 test geometry was L/D = 0.65, D = 89.306 mm, and C_r = 0.188 mm.

   The present test results mandated the use of frequency-dependent direct K_Ω and cross-coupled k_Ω dynamic stiffness coefficients. On the other hand, the imaginary components of the dynamic-stiffness coefficients could be fitted with frequency-independent direct C and cross-coupled c damping coefficients.

   Results show a significant increase of K_Ω as LVF_i increases, especially at low PR. Test results reported in 2017 by Zhang et al. has an opposite tendency of K_Ω as an impact of increasing LVF_i, namely, K_Ω decreases as LVF_i increases. Measured K_Ω drops significantly, as a result of decreasing PR, and the drop is more significant at low LVF_i.

   Concerning k_Ω and c, the results from Zhang et al. (2017) and the present results agree to the effects of changing speed, PR, and LVF_i under pure- and mainly air conditions.

   Measurement shows a peak of k_Ω at LVF_i of 2%. As LVF_i further increases from 2% to 8%, k_Ω drops slightly or remains unchanged. The value of k_Ω increases as speed goes up. However, the impact of changing PR on k_Ω is insignificant.

   As LVF_i increases from 2% to 8%, C increases by 10-15%. Test results reported in 2017 by Zhang et al. showed no significant increase.

   Effective damping (C_eff) switches from negative to positive as excitation frequency Ω increases. Cross-over frequency Ω_c is defined as the Ω value where C_eff changes from negative to positive. Increasing Ω_c decreases stability. Results show little impact of changing PR on Ω_c, but as speed increases from 5 to 15 krpm, Ω_c increases by 3.5-4 times. Ω_c increases by 50-80% for LVF_i increasing from 0 to 2%, but then decreases or remains unchanged as LVF_i further approaches to 8%.

   Except for the direct dynamic stiffness and the impact of changing LVF_i on the cross-coupled dynamic stiffness and cross-over frequency, the bulk-flow model of San Andrés (2012) does a generally decent job in predicting the trends and magnitudes of the rotordynamic coefficients.