“Introduction:

There are different centrifugal compressor designs. The machines in this paper all are of the same basic arrangement described as follows. They are powered by either an induction motor or a large synchronous motor. Synchronous motors are fairly common on the larger units because they can be used to control the power factor, thus reducing the cost of electricity. The motor drives a large bull gear at typically 1200 or 1800 RPM. The bull gear drives the pinion shafts which have an impeller on either one or both ends. The diameter of the bull gear is several times that of the pinion shafts so the impeller speeds are several times the speed of the input shaft. The bearings are a tilt pad design. Aluminum labyrinth seals provide the shaft sealing. On the bigger compressors, the span between the motor and the bull gear will often be several feet in length. All of these design features play a key role in the vibration response of these compressors.

In some cases, the level would reach a peak, then after several cycles disappear. In other instances, the amplitude would continue to build until the unit shut down on a high vibration trip.

A National Instruments 20 channel analyzer was connected to the Bently Nevada outputs on the entire machine. The compressor had fortunately been set up with X-Y probes on each of its 8 stages. Each impeller also had a key phaser output, so it was possible to monitor the phase along with the amplitude when the transients occurred.

The pattern was very repeatable. As the vibration started to change, the phase also changed along with the amplitude. The pattern looked exactly like what occurs with a rub, but in this instance, the problem had been present for several months. The air seals were made of soft aluminum, so any rub that might have occurred would have quickly cleared rather than lasting for an extended period of time. The vibration when put into polar format appeared like that of a snail shell with the amplitude and phase slowly changing and the amplitude getting larger on each cycle (Figure 3).

This problem was determined to be the Morton effect. When this problem occurs, a complex series of events takes place. It starts with the residual unbalance causing a high spot on the shaft as it goes by the tilt pad bearing. The viscosity of the oil shearing causes a localized heating effect on the shaft at the high spot.”