Axial Position Monitoring: Setting Up Thrust Protection That Actually Protects

A thrust bearing failure on a turbine or compressor can turn into a rotor-to-stator rub within seconds. Axial position monitoring is the only protection that sees it coming. But that protection is only as good as its setup: where the probes look, where zero is set, and how the setpoints relate to the actual bearing.

This article walks through the practical side of thrust protection: float zones, the two common methods of setting zero, calculating setpoints from babbitt thickness, and what to check in the field when the thrust alarm goes off.

Why axial position is a protection measurement

The thrust bearing holds the rotor in its axial position against the net thrust from the process: steam or gas pressure across the stages, balance piston leakage, coupling forces. The axial clearances in the flow path are often only a few millimetres, and sometimes much less.

Radial bearing problems usually develop over days or weeks and show up in vibration trends. A thrust bearing can fail much faster. Once the babbitt on the active pads is wiped, the rotor moves axially until something else stops it: labyrinth seals, impellers, blades or the diaphragms.

This is why axial position is wired to a shutdown, not just displayed. The goal is to trip the machine after the bearing is damaged but before the rotor contacts the stator. Everything in the rest of this article is about placing that trip point correctly.

How axial position is measured

Axial position is measured with eddy-current proximity probes mounted parallel to the shaft axis. The probe observes a surface perpendicular to the shaft (the shaft end or an integral collar) and outputs a DC voltage proportional to the gap.

For a standard 8 mm probe system the key numbers are:

• Scale factor: 200 mV/mil (7.87 V/mm), calibrated on AISI 4140 steel
• Linear range: about 80 mils (2 mm) of gap
• Output: negative DC voltage; a larger gap gives a more negative voltage

Three rules about probe placement matter more than any configuration setting:

1. Observe a surface integral with the shaft. A shrink-fit or keyed thrust collar can shift on the shaft. If the probe watches a collar that has moved, the reading no longer represents rotor position.
2. Mount the probes close to the thrust bearing. API 670 calls for axial probes within 305 mm (12 in) of the thrust collar. Further away, thermal growth of the shaft between the bearing and the probe shows up as false axial movement.
3. Use two probes with voting logic. One probe means one failed cable, connector or Proximitor can either trip the machine or blind the protection. API 670 requires redundant axial probes, normally with 2-out-of-2 voting for the shutdown.

If the shaft material differs significantly from 4140 steel, the actual scale factor differs too. This is checked during calibration (see below).

Cold and hot float zones

The rotor is free to move axially between the active (loaded) and inactive thrust pads. This total axial clearance is the float zone, and it is the reference for everything else.

The cold float zone is what you measure at standstill: push the rotor against one set of pads, then against the other, and read the travel on a dial indicator. Typical values are in the range of 10 to 20 mils (0.25 to 0.5 mm), but the OEM drawing is the authority.

In operation the picture changes. Under load the rotor sits on the active pads, but the oil film, the pad and the support elastically deflect. Thermal growth of the bearing housing also shifts the pads. The result is the hot float zone: the band within which the rotor actually moves in normal operation. It is typically shifted towards the active side compared with the cold measurement.

Two practical consequences:

• A reading slightly beyond the cold active-side position in normal operation is not necessarily wear. It may be oil film and deflection.
• Setpoints based only on the cold float, with no allowance for the hot condition, tend to give nuisance alarms.

Two methods of setting zero

The monitor needs a reference point: a rotor position that it displays as zero. Two methods are common in practice. Both work, as long as everyone who reads the screen knows which one was used.

Method 1: zero at the centre of the float zone

The rotor is pushed fully to one side, then to the other, and the probe gap is set so that the monitor reads zero midway. With a 16 mil (0.4 mm) float, the rotor against the active pads reads +8 mils (0.2 mm) and against the inactive pads reads −8 mils (−0.2 mm).

• Pros: symmetrical readings and setpoints; easy to compare both directions; the active side is immediately visible as "plus".
• Cons: depends on an accurate float measurement; in normal operation the reading is not zero, which sometimes confuses operators.

Method 2: zero at the active side

The rotor is pushed firmly against the active thrust pads, and that position is set as zero. Movement towards the inactive side reads negative, up to the full float value.

• Pros: in normal operation the reading sits near zero; any positive increase means the rotor has moved beyond the active pads, which points directly to pad wear or excessive deflection.
• Cons: asymmetrical setpoints; a small error in pushing the rotor during setup goes straight into the zero.

The mistake to avoid

The most common field error is not choosing the "wrong" method. It is setting zero with the rotor in an arbitrary position, for example wherever it stopped after shutdown, without pushing it against the pads. The float is then unknown on both sides, and every setpoint built on that zero is wrong by an unknown amount.

Record the method used, the measured float and the gap voltage at zero in the commissioning report. The next engineer will need it.

Setting the probe gap and checking calibration

The probe gap at zero must leave enough linear range in both directions. The rotor has to travel through the full float, plus the wear allowance on each side, plus a margin, without leaving the linear part of the transducer curve.

In practice:

1. Push the rotor to the zero position of the chosen method, with a dial indicator on the shaft end as an independent reference.
2. Adjust the probe so that the gap voltage at zero sits where the required travel in both directions stays inside the linear range. With Method 1 this is usually close to the middle of the range; with Method 2 it is offset towards the inactive side.
3. Push the rotor to the opposite side and compare the change in gap voltage with the dial indicator. At 200 mV/mil, a 16 mil float should give about 3.2 V of change.
4. Lock the probe and record the gap voltage.

If step 3 does not match, do not adjust the monitor until it does. Check the system on a calibrator (for example a micrometer fixture such as the TK3) with the actual probe, extension cable and Proximitor, ideally on a target of the shaft material. A wrong probe/cable length combination or a non-standard shaft material changes the scale factor, and the protection then trips at the wrong displacement.

Alarm and danger setpoints

Setpoints should be calculated from the bearing, not copied from a default configuration. The logic is simple: start from the position where the rotor rests on the pads, then add the amount of babbitt loss you are prepared to accept before alarm and before trip.

A common rule of thumb is to alarm after about 5 mils (0.13 mm) of babbitt loss and trip after about 10 mils (0.25 mm). The final values must respect two limits from the OEM: the babbitt thickness on the pads and the minimum axial clearance in the flow path. The danger setpoint must always trip the machine before the rotor can reach the stator.

Example: cold float 16 mils (0.4 mm), alarm at 5 mils and danger at 10 mils of babbitt loss, both directions.

Method 1 — zero at the centre of the float zone:

• Danger, active direction: +18 mils (+0.46 mm)
• Alarm, active direction: +13 mils (+0.33 mm)
• Rotor on active pads: +8 mils (+0.20 mm)
• Rotor on inactive pads: −8 mils (−0.20 mm)
• Alarm, inactive direction: −13 mils (−0.33 mm)
• Danger, inactive direction: −18 mils (−0.46 mm)

Method 2 — zero at the active side:

• Danger, active direction: +10 mils (+0.25 mm)
• Alarm, active direction: +5 mils (+0.13 mm)
• Rotor on active pads: 0
• Rotor on inactive pads: −16 mils (−0.41 mm)
• Alarm, inactive direction: −21 mils (−0.53 mm)
• Danger, inactive direction: −26 mils (−0.66 mm)

The physical trip point is the same with both methods. Only the numbers on the screen differ, which is exactly why the method must be documented.

After the first run at full load, compare the actual operating reading with the active-pad position. If the hot reading already sits a few mils beyond it because of oil film and deflection, review the active-side alarm with the OEM rather than living with a permanently active alarm.

Configuration notes for BN 3500

In a BN 3500 system, axial position is typically configured as a thrust position channel on a 3500/42M Proximitor/Seismic monitor. A few settings deserve particular attention:

• Normal thrust direction. Defines whether movement towards or away from the probe is the active (normal) direction, and therefore which way the reading goes positive. Get this wrong and the active-side setpoints protect the inactive side.
• Zero position. The gap voltage that the monitor displays as zero. It must match the gap voltage recorded at the zero position of the chosen method.
• Setpoints and delays. Over and under setpoints for alert and danger, each with a time delay to ride through short transients. Keep danger delays short; a thrust failure does not wait.
• Voting. The shutdown should come from the relay logic combining both thrust channels, normally 2-out-of-2 danger.
• OK limits and Not OK handling. Decide in advance what a Not OK on one channel does to the voting, and make sure it raises an alarm in the DCS. A silently bypassed channel turns 2-out-of-2 protection into no protection.

After configuration, verify each channel end to end: apply a known gap change (or push the rotor) and confirm the reading, the alarm, the relay and the signal in the DCS.

When the thrust alarm goes off: five field checks

An axial position alarm raises one question first: is the rotor really moving, or is the measurement lying? These checks, in this order, usually answer it.

1. Compare both channels and the trends. Real axial movement shows on both probes at once and in the same direction. One channel moving alone points to the measurement chain. A sudden step is more suspicious than a gradual drift.
2. Check the gap voltage and the transducer chain. A reading near the end of the linear range or a jump to supply voltage suggests a cable, connector or Proximitor fault. Check the −24 V supply and look for moisture or oil in junction boxes.
3. Check thrust pad temperatures. If the bearing has RTDs or thermocouples in the pads, a real increase in thrust load shows up as rising active-pad temperatures. Axial movement with flat pad temperatures deserves a closer look at the measurement.
4. Review process conditions. Thrust load depends on the process: load changes, steam or gas conditions, compressor operation near surge, fouling, and worn balance piston or labyrinth seals, which increase leakage and net thrust. Correlate the alarm with what the process was doing.
5. Analyse the lube oil. Babbitt particles (tin, lead) in the oil confirm pad wear. Also check oil temperature and pressure at the bearing: poor lubrication reduces the load the thrust bearing can carry.

If checks 1, 3 and 5 all point the same way, treat the reading as real and act on it. Do not raise the setpoints to make the alarm go away.

Axial position vs. differential expansion

On steam turbines the two are often confused, but they answer different questions:

• Axial position measures the rotor relative to the thrust bearing. It protects the thrust bearing and responds in seconds.
• Differential expansion measures the rotor relative to the casing, usually far from the thrust bearing. It tracks the difference in thermal growth between rotor and casing during start-up, load changes and shutdown, and protects the axial clearances in the flow path.

They use different probe locations, ranges and setpoints. One cannot replace the other.

Summary

Thrust protection works when four things are right: probes on an integral surface close to the bearing, a zero set with the rotor pushed against the pads, setpoints calculated from the actual float and babbitt, and a documented method that every engineer on site understands.

Advance Technology supplies, installs and commissions Bently Nevada vibration monitoring and protection systems for industrial plants across Central and Eastern Europe. Our engineers hold Bently Nevada training certifications. If you need an audit of existing thrust protection settings or commissioning of a new system, contact us.