An edited internal thread from the high-speed gearbox line. The topic is a small oil leak that grew until it stopped a line, the labyrinth seal, the sling ring, and the one disassembly photo that settled a week of arguments.
From: K. Stefan, Maintenance — To: J. Ruiz, Design; cc production
Subject: oil mist at the high-speed shaft — gearbox 40, input end
Morning. Gearbox 40, the 3000 rpm input end, is weeping oil mist onto the coupling and the floor, enough to form a wet ring after two shifts, and the bearing temperature is normal at 62 C and the level gauge sits mid-window, so this is not a fill problem and not a bearing heat problem. The drawing says labyrinth seal on the input shaft with a sling ring, two-piece, and I have asked the line to clean and watch before we pull the cover. What I would like from design, before I touch it: is the labyrinth supposed to breathe, and is the sling ring supposed to throw the oil back or simply keep it away from the seal mouth? The machine is our reference for a customer audit in ten days and I would rather open it once with the right parts.
[Inline reply from production: the wet ring is on the coupling side of the sling ring, not inside the hub. Photo attached showing the ring wiped clean on the outboard face, oil on the inboard.]
From: J. Ruiz, Design — To: K. Stefan; cc production
Subject: RE: oil mist at the high-speed shaft — how the sling ring works
Short answer: the labyrinth is a non-contact throttling path and it breathes on the outboard face by design, and the sling ring is meant to sling oil away from the seal mouth radially, back down to the return gallery, before it can climb the shaft. On a 40 mm shaft at 3000 rpm the centrifugal head at the ring is substantial and if the ring is facing the wrong way, or if the return hole behind it is plugged, oil rides the shaft, past the ring, and out the labyrinth mouth, which sounds exactly like your symptom, wet on the outboard, clean on the inboard. These rings are stamped and easy to install backwards, the drain side has to face inboard toward the oil. Pull the cover and send me a photo of the ring orientation and the return hole, with the shaft static, and we will know in five minutes whether it is assembly or a plugged gallery.
From: K. Stefan, Maintenance — To: J. Ruiz; cc production
Subject: RE: oil mist — photos attached, it is better than we hoped
Cover off at 09:40, photos in the attachment, and the answer is embarrassingly simple: the sling ring is in correct, drain face inboard, the return hole is clear, a 6 mm gallery open end to end. So the ring is not backwards and the gallery is not plugged, and oil still climbs the shaft. There is one more detail in the photos that I do not like: the labyrinth lands are worn on the lower half, uneven wear, and the radial clearance at the bottom measures about 0.9 mm against what the drawing file says should be 0.25 to 0.40. I checked with feeler gauges at four points around the bore, bottom, both sides, top, and the bottom gap is the big one, the bore is no longer concentric with the shaft. That smells like the housing bore settling or the labyrinth carrier distorting, not like the seal laughing at us.
From: J. Ruiz, Design — To: K. Stefan; cc production, quality
Subject: RE: oil mist — 0.9 mm on the bottom, that changes the physics
0.9 mm bottom radial on a 40 mm shaft at 3000 rpm is the story, thank you for the feeler work. A labyrinth works as a throttling path because the radial gap is small, tight clearance keeps the flow velocity high and the pressure drop high across each land, and the design intent is 0.25-0.40. At 0.9 on the bottom the gap area roughly quadruples at the bottom quadrant, the throat opens, the oil film bridges the gap, and the centrifugal action of the sling ring alone cannot push against a wide-open throat, the mist simply crawls out on the low-side gap. The uneven wear tells me the rotor is rubbing the lands at some cycles, likely a shaft sag or a housing distortion, and that rubbing locally opened up the same side. Can you tell me whether the bottom wear corresponds to the installed side, and whether the coupling half bearing endfloat has changed since the last runout record? The labyrinth being the victim rather than the villain is the lead, because a seal does not wear itself.
From: Production — To: K. Stefan; cc J. Ruiz
Subject: RE: runout record and installed orientation
Filed the last three runout records, all within 0.02 mm TIR at the coupling end, so the shaft itself is straight and the bearing preload is normal. The housing, we noticed at the pull that the foot bolts around the input end were only finger tight, two of four, and there is a witness mark on the housing foot and the mounting pad, which someone has been fighting for a while. That is our confession: the line had a soft housing foot for the last audit cycle, and the bore walked down with the vibration. The labyrinth carrier did not distort, the gearbox sagged and took the bore with it, and the seal was only the messenger.
From: J. Ruiz, Design — To: K. Stefan; cc production, quality
Subject: RE: soft foot — repair plan and the one change on the drawing
With the soft foot confirmed, the repair is defined and I am not blaming the seal. Re-mount the gearbox on the machine base with four new hardened washers and torque the foot bolts to the drawing 240 N·m in a two-pass criss-cross, face the pad flat (0.03 mm flatness over the foot), and take the housing bore runout at the input end after re-torque, target under 0.05 mm TIR. Replace the sling ring and the labyrinth lands with the original 0.25-0.40 radial, and re-run the breather, because a labyrinth also vents around the shaft and a blocked vent builds pressure that pushes the oil film out, and that pressure was half the leak. One drawing change: add a ‘torque pattern and check bore concentricity’ note to the gearbox installation sheet, because a soft foot is invisible to every bearing gauge until the housing bore walks.
I would also like the maintenance regime to record the bottom radial gap of the labyrinth at the next two oil changes, 0.25-0.40 target, and flag a growing gap early, because a gap that doubles quietly is the audit failure that announces itself as a wet coupling. The sling ring stays, the labyrinth stays, the geometry returns to drawing, and the machine should behave as designed when the seals are allowed to do their job on a shaft that stays where it belongs.
From: K. Stefan, Maintenance — To: J. Ruiz; cc production, quality
Subject: RE: gearbox 40 — closed after 48 h run
Closing with the numbers. Gearbox re-mounted with four new hardened washers, foot bolts torqued 240 N·m two-pass criss-cross, pad faced to 0.03 mm flatness, bore runout at the input end after re-torque 0.04 mm TIR, labyrinth lands and sling ring replaced to the 0.25-0.40 radial, vent cleared. Ran 48 hours including two full start-stop cycles, bearing stays 61-63 C, coupling flange dry, no wet ring on the floor, and the audit passed with the leakage photo showing a clean flange.
Attachment: before-and-after bore concentricity records, labyrinth gap measurement sheet, foot torque record with pattern, gearbox 40 repair closure report.
Honest note after the thread — the leak was never really a seal problem. The labyrinth and the sling ring were engineered fine, and maintenance even installed the ring correctly, but a soft gearbox foot let the bore settle 0.9 mm out on the low side, quadrupling the throttling gap, and the oil film walked out the wide-open mouth. A non-contact seal is a geometry device, it performs only as well as the concentricity the rest of the machine gives it, and a 0.05 mm housing TIR is as important to the seal as the gap itself. When the mist is on the outboard face, check the gap first, the ring second, and the mounting bolts third, the bolts usually being the ones nobody wants to admit were loose.
Reference data — labyrinth radial gap design 0.25-0.40 mm on 40 mm shaft; measured 0.9 mm bottom before repair, 0.04 mm TIR bore after re-torque; foot bolt torque 240 N·m two-pass; bearing temperature 61-63 C.
Glossary — labyrinth seal: a non-contact shaft seal using a series of close-clearance lands; sling ring: a rotating ring that throws oil away from the seal mouth by centrifugal action; radial gap: the clearance between shaft and land; TIR: total indicated runout; soft foot: a mounting condition where the base pad is distorted and the housing bore settles under load.
Field note — a short way to think about labyrinth clearance
Anyone joining this thread late can hold onto one idea: a labyrinth seal converts shaft-side pressure into a throttled flow, and the throttle strength depends on the gap to the fourth power in the ideal laminar model, so a doubling of the radial gap does not leak twice as much oil mist, it leaks an order of magnitude more. That is why the 0.9 mm bottom on a 0.25-0.40 design was not a minor tolerance slip but a completely different device, one that had lost its throttling character on the low side. The sling ring still did its job, it threw oil off the ring as designed, but the oil film bridged the wide-open gap on the sagged quadrant before the ring could act on it. In practical terms this means the fastest interrogation of any non-contact seal leak is not to order a new seal, it is to measure the gap at four points around the bore and compare bottom to top: a symmetric gap that is simply too big points to wear or an oversize replacement, an asymmetric gap points to a shaft or housing that moved, and an asymmetric gap with the big side at the bottom almost always points to a mounting that settled. Measure the soft foot before you buy the seal kit, and the kit, when the time comes, will be the last part you need.
From: M. Okafor, Production — To: K. Stefan, J. Ruiz; cc quality
Subject: FW: oil mist — closing the loop after the 48-hour dry run
Closing the loop with both of you, because the thread deserves an ending. We re-torqued the foot bolts in two passes to 240 N·m on all four corners, in the bolt order we agreed, then wiped the shaft, the labyrinth lands and the sling ring and started the box. First 24 hours: bearing steady at 62 C, level gauge mid-window, and the coupling stayed dry, not a film. Second 24 hours: dry again, and the audit ran it in front of the customer team today, a half-hour loaded run at 3000 rpm, with no oil mist, no wet ring, nothing to explain away. Total top-up over the 48 hours was about 0.2 litres of oil, which is normal breathing, not a leak.
Worth one sentence for the file: the parts that fixed it were not replaced, no new labyrinth, no new sling ring, no seal kit. The repair was three bolts re-torqued, which is why I want this in the maintenance standard and not in the spare-parts list. Any gearbox with a soft foot gets the feeler-gap check and the two-pass torque before anyone orders a seal. I have asked the line to add it to the PM sheet for the three other 3000 rpm boxes of the same build.
[Attachment: audit photos, coupling side, labelled: wet ring 12/03, clean 12/05.]
