Troubleshooting the bundle assembly of a centrifugal compressor

Published on: 

A centrifugal compressor, packaged by Cooper Industries (now part of Rolls-Royce), was used in an offshore facility to compress natural gas. It was a three-stage assembly: from 6 kg/cm3 to 90 kg/cm3 with two stages combined in the low pressure (LP) bundle and a third stage separately operating in the high pressure (HP) bundle assembly.

Efforts were taken to reduce the time required in overhaul once the maintenance staff identified unusual noises emanating from the compressor. With a rigid commitment in place to supply gas to users downstream, the management demanded urgent completion of an in situ repair.

The HP bundle assembly was a horizontally split Kawasaki RB-7B with seven stages. It was oil lubricated, and oil sealed having internal stage separation of aluminum alloy labyrinths. The maximum throughput of natural gas was about 2.3 million standard cubic feet per day (mmscm/d).


The inlet pressure of the HP compressor was 30 kg/cm and discharge pressure was about 92 kg/cm. The compressor was driven by a Rolls-Royce MK-1535 gas turbine aerodynamically connected to a Cooper Power Turbine and a step-up helical box by Philadelphia Gears. 

Sealing of the shaft was achieved through sets of aluminum labyrinth and a peripheral honeycomb combination with high-grade oil between the sealing surfaces. The compressor had a stationary casing with a horizontally split rotor assembly pushed and secured by a split-ring retainer system bolted on from the coupling side.

Problem detection

During operation on load, an unusual humming sound came from the HP and LP coupling area. This sound appeared to be emanating with a higher pitch from the HP bundle side. It began about 45 minutes after loading the unit or as soon as it reached its operating temperature. In addition, vibration parameters of the HP compressor had nearly doubled.

Engineers checked on the condition of the coupling shim. When found intact, they opened the bundle assembly and noticed a failure of the HP bundle O ring, which seals the space between the bundle and the casing. This generated abnormal noise due to HP gas at the discharge end seeking to find a way through the failed O ring area towards the HP suction side. To fix the problem, they removed the drive-end seal assembly. The bundle sealing O-ring was damaged on the periphery over a length of about six inches.

(Read more in the September/October issue of Turbomachinery International)