Industries / Industrial equipment and services / Industrial machinery / Compressors
Seals and bearings
Making the parts that hold a compressor's spinning shaft in place and keep the gas inside where the shaft leaves the casing.
6
Milestones achieved, from 1910 to 2018.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1910
Pads that ride on oil
Albert Kingsbury's thrust bearing carries the load on a wedge-shaped film of oil between the shaft's collar and a ring of pivoted pads, giving very low friction and almost no wear. He tested the first in 1904 and was granted his patent in 1910; the first in hydroelectric service went in at Holtwood, Pennsylvania, in 1912. A. G. M. Michell of Australia invented the same principle independently, ASME says. Tilting-pad bearings later carried the rotors of high-speed compressors, such as the North Sea compressor in Fulton's 1984 case.
Credit
Albert Kingsbury; A. G. M. Michell
1976
Seals that run on gas
A dry gas seal holds the gas in with two rings, one turning, whose faces are kept apart by a thin film of gas drawn in by spiral grooves, so they never touch while running. Before them, centrifugal compressors mostly used "wet" seals lubricated with oil. John Crane says its spiral-groove work began in 1968, its first commercial installation came in 1975, and its Type 28 seal followed in 1976. By 2002 more than 80% of new centrifugal gas compressors had dry gas seals, wrote John Stahley of Dresser-Rand, and the US Environmental Protection Agency (EPA) put the share at about 90% in 2006. EPA reported that a pair of dry seals leaks 1 to 6 standard cubic feet of gas a minute, against 40 to 200 for a pair of wet seals.
Credit
John Crane
1985
Shafts held up by magnets
An active magnetic bearing holds the shaft in a magnetic field that sensors and amplifiers keep adjusting, with no oil at all. NOVA, an Alberta pipeline company, worked with S2M of Vernon, France, which developed the bearing, and Magnetic Bearing Inc. of Radford, Virginia, to fit one to a natural gas pipeline compressor that also had a dry gas seal. Writing in 2000, Ahrens of ABB and Sturgess of Revolve said this "World's First Oil-Free Compressor" was commissioned in 1985, and that by 1992, 34 compressors with about 60% of NOVA's installed power ran on magnetic bearings. Reliability problems with connectors, amplifiers, backup bearings, design and training then led NOVA to review the program, and other pipelines were slow to follow.
LessonA bearing run by electronics is only as reliable as its connectors, amplifiers and backup bearings.
Credit
NOVA; S2M; Magnetic Bearing Inc
1999
Seals that damp the rotor
Honeycomb seals, whose stationary surface is lined with small cells, had been used since the late 1960s at the balance pistons of high-pressure synthesis gas compressors because they were strong. In the 1990s makers began fitting them to calm whirl in higher-pressure compressors, since the rough surface adds damping. About 1999 Dresser-Rand began making damper seals of aluminum drilled with a pattern of holes, which were quicker to make and stood up to rubs better, and by 2011 it had built more than 400 compressors with damper seals, its rotordynamics engineer Edmund Memmott wrote. Takahashi and colleagues, describing a new damper seal for high-pressure compressors, write that honeycomb and hole-pattern seals are "popularly used".
Credit
Dresser-Rand
2008
Dirty gas wrecks seals
A compressor in natural gas gathering service had repeated dry gas seal failures at its discharge end. McCraw, Bakalchuk and Hosanna reported at the 2008 Turbomachinery Symposium that synthetic oil mist in the sealing gas was a significant source of the contamination, and that the fix changed the seals, the seal controls and the compressor. A 2021 tutorial by Tuffentsammer and Hopfner groups the main causes of dry gas seal failure into four kinds of incident, three of them contamination: from the process gas, from the seal gas supply, and from bearing oil.
LessonA seal that rides on a gas film is only as clean as the gas fed to it, so the supply system matters as much as the seal.
Credit
James McCraw, Vladimir Bakalchuk and Richard Hosanna; Martin Tuffentsammer and Thomas Hopfner
2018
A standard for gas seals
The first edition of API 692 sets requirements for the design, testing, installation, commissioning and start-up of dry gas seals and their support systems on axial, centrifugal and rotary screw compressors and expanders, replacing part 4 of the fifth edition of API 614; it was taught as a short course at the 2018 Turbomachinery Symposium. Stahley had noted in 2002 that no accepted industry standard for gas seal support systems existed, and that API had only recently addressed them in API 614 (1999).
Credit
American Petroleum Institute
Watch for
Separation seals that save nitrogen
A separation seal sits between a dry gas seal and the oil-lubricated bearing next to it, and is usually fed nitrogen to keep bearing oil out of the gas seal. John Crane says "test data indicates" its Type 93AX coaxial separation seal cuts nitrogen use by up to 80% compared with conventional radial separation seals, and keeps working without separation gas.
Achieved whenan operator or another outside party publishes measured nitrogen flow, or dry gas seal failure counts, before and after fitting it.
Claimed by
John Crane, June 2025
Open
01
A public seal failure tally
One open table of every published dry gas seal failure (case studies, tutorials and papers), coded by root cause (process gas contamination, seal gas supply contamination, bearing oil, vibration, other), service, seal type and outcome, with a link to each source.
The briefCredit · unclaimed02
Fault data for oil bearings
A public, labeled set of shaft-motion recordings from oil-film journal bearings on a bench rotor kit: normal running, oil whirl, rub, misalignment, unbalance and starved oil supply, at several speeds, with a baseline classifier.
The briefCredit · unclaimed
Open 01 · the brief
A public seal failure tally
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
One open table of every published dry gas seal failure (case studies, tutorials and papers), coded by root cause (process gas contamination, seal gas supply contamination, bearing oil, vibration, other), service, seal type and outcome, with a link to each source.
Why it is still open
The figures in circulation come from papers that cite unnamed records. Tuffentsammer and Hopfner (2021) say "recent incident reports and studies" show dry gas seals cause only 39% of compressor shutdowns and that three of the four main failure types are contamination; John Crane says "research has shown" contamination is a significant cause. Searches of Zenodo, GitHub and CORE found individual case studies but no public tally.
Built with what exists
CORE's free search API (api.core.ac.uk), which indexes the Turbomachinery and Asia Turbomachinery & Pump Symposium proceedings; the terms of API 692 for the coding scheme; a small AI model built for extraction from long documents, such as Meta's Muse Ember (see the Note "The jobs agents repeat"); a spreadsheet.
First step
Run a CORE search for "dry gas seal failure", list every case study from 2001 on, and code the first ten.
Done looks like
At least 40 published failures coded and linked; a second person codes 15 of them and agrees on the cause for at least 90%; a published share of failures by cause, with the count behind each share.
Open 02 · the brief
Fault data for oil bearings
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public, labeled set of shaft-motion recordings from oil-film journal bearings on a bench rotor kit: normal running, oil whirl, rub, misalignment, unbalance and starved oil supply, at several speeds, with a baseline classifier.
Why it is still open
Public machine-fault sets are mostly for rolling-element (ball and roller) bearings, such as the Case Western Reserve data and a 2025 rotating-machinery set from the ISAC lab, both on Zenodo. In 2015 a Seoul National University and KEPCO team recorded normal, rubbing, misaligned and oil-whirl states on a Bently Nevada RK4 rotor kit, but the paper gives no way to obtain the data. Searches of Zenodo, GitHub, Kaggle, Mendeley Data and IEEE DataPort found no fluid-film fault set.
Built with what exists
A guarded bench rotor kit with oil journal bearings, such as the RK4 used in the 2015 study; two Bently Nevada 3300 XL proximity probes per bearing, set at right angles; a USB data acquisition card; Python and scikit-learn.
First step
Record normal running and oil whirl at two speeds on one kit, publish the first 50 recordings with a data sheet, and ask a second lab to repeat them.
Done looks like
At least 500 labeled ten-second recordings across six states and four speeds, released under an open license, with a baseline classifier's accuracy reported on runs recorded on a different day from its training data.