Glossary term

Orbit Plot

X-Y vibration display showing shaft-relative motion from orthogonal probes, used to interpret rotor motion, rubs, whirl, runout and bearing behaviour.

Definition

method

An orbit plot is an X-Y display of shaft-relative motion, usually built from two orthogonal displacement probes at the same bearing plane.

Orbit plots show the path followed by a rotor centreline relative to nearby bearings or probe brackets. They are used in rotor dynamics and condition monitoring to interpret synchronous vibration, oil whirl, rubs, looseness, misalignment, runout, bearing anisotropy and transient machine behaviour.

An orbit plot is an X-Y vibration display that shows the path of a rotating shaft centreline relative to a bearing, casing or probe bracket. It is usually built from two orthogonal non-contact displacement probes installed at the same axial station. One probe provides x(t) and the other provides y(t), so the orbit is the parametric curve:

\mathbf{r}(t)=x(t)\mathbf{i}+y(t)\mathbf{j}

The plot is commonly used on turbines, compressors, generators, pumps, marine shafting, high-speed test rigs and other machines where shaft-relative motion is more informative than casing vibration alone.

Engineering Role

Orbit plots help engineers see how the shaft moves inside its available clearance. They can reveal circular or elliptical synchronous motion, oil whirl, oil whip, rubs, looseness, coupling effects, bearing anisotropy, shaft bow, rotor instability and measurement problems. When a once-per-revolution phase mark is available, the plot can also show phase angle and precession direction relative to shaft rotation.

An orbit should be interpreted with the machine layout in mind. Probe orientation, bearing type, oil-film behaviour, support stiffness, load, speed, thermal growth and filtering all affect the shape. A clean ellipse is not automatically acceptable, and a distorted orbit is not automatically a fault. The engineering question is whether the observed motion is repeatable, physically plausible and consistent with speed, load, phase and other evidence.

Raw, Filtered and Compensated Orbits

An orbit plot must state what was removed before plotting. A raw proximity-probe orbit may include DC shaft position, mechanical runout, electrical runout, broadband vibration, subsynchronous motion and synchronous 1x response. If the mean position is removed, the plotted channels become:

x_{ac}(t)=x(t)-\overline{x}
y_{ac}(t)=y(t)-\overline{y}

This can make the orbit easier to read, but it hides shaft centerline movement. That matters when bearing clearance, oil-film lift, rub risk or thermal growth is part of the decision.

A filtered orbit may show only 1x, a subsynchronous band or another selected component. The filter choice should be documented because a 1x-filtered ellipse can look acceptable while broadband motion, rub impacts or a growing oil-whirl component are present in the raw data.

Slow-roll compensation is another separate step. It attempts to remove geometric or electrical runout observed at low speed:

x_{comp}(\theta)=x_{meas}(\theta)-x_{slow}(\theta)

where \theta is shaft angle from the phase reference. Compensation is useful only when the slow-roll reference is stable and the probe/shaft material condition has not changed. A defensible orbit review therefore compares raw, filtered and compensated views rather than relying on one clean-looking plot.

Worked Example: Estimate Orbit Size and Clearance Fraction

A sleeve-bearing fan has two orthogonal displacement probes at one bearing plane. After calibration and removal of the stated DC gap offset, the synchronous 1x orbit is approximated by:

x(t)=28\cos(\Omega t)\ \mu\text{m}
y(t)=16\sin(\Omega t)\ \mu\text{m}

The orbit is an ellipse with semi-major displacement:

a=28\ \mu\text{m}

and semi-minor displacement:

b=16\ \mu\text{m}

The peak-to-peak motion in the probe directions is:

X_{pp}=2a=56\ \mu\text{m}
Y_{pp}=2b=32\ \mu\text{m}

If the minimum radial running clearance at this bearing is:

c=150\ \mu\text{m}

then the largest orbit radius relative to that clearance is:

\displaystyle \frac{a}{c}=\frac{28}{150}=0.187

or about 18.7\% of the clearance.

Engineering comment: the orbit is not close to the assumed radial clearance in this simplified calculation. That does not prove the machine is healthy. The engineer still checks the DC shaft centreline position, runout compensation, probe linear range, bearing temperature, oil condition, phase stability, speed trend and whether subsynchronous components are hidden by filtering.

Now suppose the same data contain a subsynchronous component near 0.45x running speed. If that component grows with speed and the orbit precesses in the forward direction, the evidence may suggest oil whirl or fluid-induced instability. The orbit plot helps form that hypothesis, but the conclusion requires a waterfall spectrum, order tracking, bearing data and repeatable operating evidence.

Orbit plot is not runout. Runout is geometric or functional variation measured relative to a datum during rotation. An orbit plot shows dynamic shaft-relative motion and may include runout, vibration, oil-film motion, thermal position and probe effects.

Orbit plot is not a tachometer. A tachometer gives speed or phase reference. The orbit uses displacement channels; a tachometer or keyphasor may be added to mark angular position.

Orbit plot is not a waterfall spectrum. A waterfall shows spectral amplitude versus speed, time or condition. An orbit plot shows two displacement channels against each other in the time or angle domain.

Orbit plot is not order tracking. Order tracking organizes vibration by shaft order. An orbit may be filtered to 1x, subsynchronous or broadband motion, but it remains an X-Y path.

Orbit plot is not an orbital mechanics trajectory. In rotating machinery, “orbit” refers to shaft centreline motion relative to the machine, not a spacecraft or planetary path.

Validation and Common Mistakes

A defensible orbit plot states probe locations, angular orientation, probe sensitivity, gap voltage, displacement units, shaft material, calibration, DC offset treatment, filter settings, sampling rate, tachometer or phase reference, speed, load, bearing condition and whether the plot is raw, compensated or filtered.

Common mistakes include:

  • plotting two channels that are not orthogonal or not at the same bearing plane;
  • forgetting probe scale factor, probe angle, gap voltage or shaft material effects;
  • removing DC position without stating it, hiding shaft centreline movement;
  • comparing filtered 1x orbits with unfiltered broadband orbits as if they were the same measurement;
  • treating electrical runout, mechanical runout and dynamic vibration as the same contribution;
  • ignoring aliasing, clipping, low signal-to-noise ratio or poor probe bracket stiffness;
  • inferring rub, whirl or misalignment from orbit shape alone without phase, speed trend and supporting machine evidence.
REF

See also