The present investigation was motivated by a noticeable high-pressure pulsations and vibrations during commissioning of a centrifugal compressor on a high-pressure natural gas transmission system. During commissioning, the gas from the compressor discharge is recycled back to compressor inlet through a recycle line turning the gas flow from the main discharge line into a smaller diameter side branch, thus creating a deadleg due to the closure of a check valve farther downstream on the main discharge line. To better understand and characterize the flow generated pulsations due turning flow into side branches, an experimental setup was constructed on ambient air to test three side-branch to main pipe diameter ratios (d/D), nominally: 1.0, 0.75 and 0.5, over a wide range of mean flow velocities. Experimental results of the normalized pulsation pressure amplitudes (P*) vs. Strouhal number characterized the flow-acoustic field for the three d/D ratios and for acoustically tuned and detuned systems. It was found that P* decreases as d/D decreases for acoustically tuned systems. The highest P* was achieved with d/D = 1 which reached a value of 3.543 at St = 0.1376. At lower d/D = 0.762, the maximum value of P* decreased to 1.173 at a slightly lower St number of approximately 0.1. In the case of d/D = 0.5132, the normalized pressure P* was further reduced to a very low value of 0.2462 at a wider range of St number. Acoustically tuned system is characterized by having the highest anti-nodal point of acoustic velocity oscillation at the Tee-junction location in the axial direction of the main pipe, with minimum acoustic leakage into the side branch. When the system is acoustically detuned, i.e., when the axial acoustic velocity oscillations at the Tee-junction is reduced, the strength of the local acoustic source at the junction is also reduced resulting in a lower peak normalized pressure amplitude at the deadleg. The peak amplitudes of pressure pulsations occur at frequencies exactly matching the deadleg 1/4 wavelength resonance, whether it is tuned or detuned with respect to the inlet section. That is, the peak frequencies tracked the changes in the deadleg length and was not influenced by the inlet section acoustic resonance characteristics. This led to the postulation that the main driver of the flow-generated pulsations, i.e., the acoustic source is associated with the high-shear area of the flow facing the deadleg at the trailing edge of the Tee-junction rather than the ‘vortex bubble’ due to flow separation at the leading edge of the side branch.

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