Dry gas seal (DGS) hang-up detection in a centrifugal compressor

Case study · Primary dry gas seal hang-up due to oil contamination

ApplicationLNG

Lead time> 9 monthsAddressed 9 months post detection in a planned outage

OutcomeDGS cartridge replaced in a planned maintenance outage. Primary seal contamination confirmed on disassembly.

DGS hang-up · Fault mode

Oil migration across the intermediate labyrinth can contaminate the retainer ring and dynamic gasket causing the stator ring to get “hung-up”

The asset

Service	Mixed Refrigerant
Seals	Tandem dry gas seals
Primary vent	Differential pressure across a needle valve, flow

How a dry gas seal works

Dry gas seal cross-section showing the sleeve, rotor, stator, spring and dynamic gasket

The rotor ring turns on the shaft sleeve. The stator ring runs a few microns from it, held off by the seal gas film.

The spring pushes the stator toward the rotor, and the dynamic gasket lets it slide axially to follow the shaft. A small, steady leak passes to the primary vent.

How hang-up develops

Oil reaches the seal. Lube oil migrates from the bearing housing past the separation seal, most often during starts, stops or a loss of separation gas.
The primary ring sticks. Oil and the deposits it leaves add friction at the secondary seal, so the ring can no longer slide freely.
The faces stop tracking. When the rotor moves axially, the ring lags behind. The gap opens and primary vent pressure (or orifice/needle valve differential pressure) and flow rise.
Left in service. A hung seal can run with high leakage, or close suddenly and contact, leading to seal failure and an unplanned shutdown.

Why a vent alarm alone misses it. Primary vent pressure or orifice/needle valve differential pressure and flow move with suction pressure and speed. A fixed alarm set wide misses the early rise, and one set tight can cause a spurious vented trip.

DGS hang-up · Anomaly identification · The rule

Fourteen Cases tier the response, and a standard-limit Case fires first.

Coce Cas1 NDE Pri Dgs Issues Config 8

Default

Equipment Type

CENTRIFUGAL_COMPRESSOR__1_CASING

Seal configuration

Vent DP and flow

Cases, in evaluation order

1 · Dp F Hh 12 · Dp F Hh 23 · Dp Hh F Ll 14 · Dp Hh F Ll 25 · Dp Hh 1 6 · Dp Hh 27 · Dp F H 18 · Dp F H 29 · Dp H F L 110 · Dp H F L 2 11 · Dp H 112 · Dp H 213 · Dp H 314 · F Hh Catchall 1

Dp differential pressure, F flow, H and L standard high and low, Hh and Ll critical high and low.

Case 7 · Primary NDE Dry Gas Seal IssueMedium

ALL

Adaptive Threshold Standard High

coce_mass_flow_nde_seal_gas_primary_vent

True

Adaptive Threshold Standard High

coce_dp_nde_seal_gas_primary_vent

True

Adaptive Threshold Standard High

coce_case1_diff_pressure_primary_vent_perc

True

Positive Deviation

coce_dp_nde_seal_gas_primary_vent_ml

True

Two independent measurements

Primary vent flow and the differential pressure across the needle valve come from separate transmitters. Both must read high, so one drifting transmitter cannot raise the event.

Compares DE with NDE

The third test is the percentage difference in needle valve differential pressure between the two seals. Suction pressure and speed move both seals together. A seal problem moves one.

Compares with the ML model

The fourth test is the percentage difference between the actual and machine learning model predicted needle valve differential pressure. The model is trained on ‘good’ data and account for normal changes from speed or supply pressure variations

Tiers the priority

Cases 1 to 6 test the critical limits. Cases 7 to 13 test the standard limits and fire earlier, at lower priority, so the site is warned before the event escalates. Case 14 is a flow-only catch-all.

Case 7 raised this event at Medium priority, on the standard limits, before any critical limit was reached.

DGS hang-up · Anomaly identification · The data

Primary NDE dry gas seal hang-up

07:50

Press play

Four and a half hours of primary vent data from the NDE seal, replayed at speed.

NDE primary vent: differential pressure and flowMP/HP MR compressor · 2-minute samples

07:50Normal operation

09:21DE and NDE diverge as rotor axial position changes

09:40Vent differential pressure and flow rise

~09:50Case 7 raised

10:06Peak leakage

12:13Return to normal operating condition

DGS hang-up · Alert · What was sent

The Service Request sent to the site included the diagnosis and the plots.

XXXX-UNITXX-YYYY-### High [Status] Asset ID : 100XXXXX Created at : [date, time] External Id: 27XXXXX REOSMSMD

Service Request #[###]Discussion ThreadEdit

One-month window · every Primary NDE Dry Gas Seal Issue on this asset

Week 1Week 2Week 3Week 4Week 5

Description Duration: 34 minutes

Potential NDE Primary Dry Gas Seal hangup detected

Insights

Between 09:21 AM CT and 12:13 PM CT on DD-MMM-YY, the following were detected on the NDE primary dry gas seal of the MR Compressor:

The NDE primary vent differential pressure rose to 8.78 psi (60.5 kPa), against a model prediction of 5.61 psi (38.7 kPa).
The NDE primary vent flow rose from 14.9 lb/h (6.8 kg/h) to 22.2 lb/h (10.1 kg/h).
The DE primary vent differential pressure changed by less than 0.4 psi (2.8 kPa), so the DE vs NDE difference moved from −11% to −74%.

Leakage rose on the NDE seal only and followed a change in rotor axial position. This is indicative of the NDE primary seal stator ring hanging up.

Recommendations

Verify the NDE primary vent flow and needle valve differential pressure transmitters against local indication.
Check the seal drains for signs of oil.
Continue to monitor. If the event recurs or escalates to a critical-limit Case, plan a DGS cartridge replacement at the next planned outage.

Plots

Event

Work Orders

Work Logs

Who was informed

RERotating Equipment Engineer
OSOperations Supervisor
MSMaintenance Supervisor
MDMechademy diagnostics engineer

How it was delivered

Turbomechanica notification on the platform
Email to watchers and assignees
CMMS notification

Timeline

Each capsule is one occurrence of the event on this asset, so recurrences read on one line.

DGS hang-up · Execution and result

Nine months later, disassembly confirmed oil contamination on the primary seal.

Execution

Month 0
Case 7 raised, Service Request sent
Medium priority, on the standard limits.
Month 0
Site response
Outage work order created.
Months 0–9
Monitored in service
Multiple recurrence detected with a gradual increase in primary vent differential pressure excursions during rotor axial movements.
Month 9
Planned maintenance outage
DGS cartridge replaced.
Month 9
Disassembly
Oil contamination confirmed on the primary seal.

Primary seal on disassembly, with oil contamination indicated by an arrow — Primary seal on disassembly. The arrow marks the oil contamination.

Result

Diagnosis	Confirmed on disassembly
Found	Oil contamination on the primary seal
Action	DGS cartridge replaced in a planned outage
Lead time	> 9 months
Unplanned downtime	None
Value	One or more vented trips avoided. Unscheduled outage for DGS replacement avoided.