Work log — injection moulding shop, hydraulic leak hunt week
Engineer: M. Okafor (fluid power)
Project: IM-4 320 tonne injection press hydraulic station
Log period: Day 01 to Day 06
Document ref: WL-2026-0911-SL10
Day 01 — the pan that started the hunt, and a pump that was loud
Stand-up 07:50. The IM-4 press has been losing hydraulic oil for three weeks, roughly ten litres a week by the sight-glass baseline on the 250 litre tank, and the floor pan under the station is slick again after a weekend wipe. Two symptoms sit beside the leak: the clamp cylinder has gone soft through the fast-advance phase, taking 1.9 seconds where the cycle spec says 1.4, and the pump set has a new high whine that the maintenance crew swore, honestly, they had been ignoring for a fortnight because the press was still making parts.
Morning I did the discipline that finds a leak before it has a drama: I cleaned the whole station, dried the pan, marked a chalk box around the pump, the manifold, the clamp hose bundle, the drain line and the return-line filter, and started the press for a twenty minute baseline run. The oil level dropped 0.9 cm in the sight glass across that run, which is a leak and not a transfer, and the whine pitched up when the clamp valve shifted, which is a clue that points at the high-pressure standby circuit rather than the tank breather.
Evening note: a hydraulic leak is either dirty oil, hot oil, or a tired component, and I have not touched any of that oil yet, I have only made the machine tell me where to look.
Day 02 — the stethoscope, the chalk box, and the first suspect that was innocent
Tuesday I ran the classic hydraulic leak hunt in order, because skipping the order is how a crew tightens a fitting while the real leak drips behind them. First I checked every static joint hot, with the system at 160 bar on the clamp stroke and the machine locked out, working the hex on high-pressure fittings with a torque wrench set to the drawn values, and the check caught two loose unions on the pump manifold, each about a quarter turn, which on their own explain a slow weep but not ten litres a week across three phases. I re-torqued both, marked them, and cleaned the area to see if the leak moved.
Second came the dynamic hunt, with the press cycling, and here is where the stethoscope earns its keep. The mechanical probe on the clamp hose bundle picked up a damp hiss exactly at the swaged coupling behind the fast-advance valve, and the chalk on that coupling turned from a clean ring to a wet smudge inside forty minutes of cycling. The coupling is a 25S swage fitting on a 20 mm hose, and the hose at that swage shows a soft collar and a flattened hex, classic signs of an assembly that was over-torqued at the last rebuild and has now crushed its own ferrule seat, so the oil is weeping at the swage face, not at the thread.
Afternoon I swapped that hose assembly off the line, sent it for re-swaging, and fitted a fresh 20 mm bore, 500 mm long hose with a new swaged coupling and a spiral reinforcement rated for 210 bar. The re-fit took forty minutes, and on the first cycle the fast-advance time came back to 1.5 seconds, most of the soft clamp returned, and the whine dropped by a clear notch, which pointed a finger I had not expected at the blocked return-line filter that the same crushed swage had been hiding.
Evening note: the first suspect is rarely the whole story. Two loose unions explained a weep, one crushed swage explained a slow loss, and the machine is telling me there is a third chapter, and the return line is where it writes.
Day 03 — the blocked return line, the hot tank, and the leak that was never a leak
Wednesday I went after the third chapter. The return-line filter’s differential pressure gauge sat at 2.6 bar, against a 0.5 bar healthy reading on its tag, and the tank felt hot at the filler, 52 C against a 42 C summer normal, so the oil is being pushed hard through a clogged return filter and then past a tank wall it usually rinses. I isolated the filter, dropped its bowl, and the element came out plastered in a brown sludge that smelled like burnt oil, blocked enough that paper would not let air through the pleats, and the sludge is the fingerprint of the over-torqued swage, because a crushed ferrule edge shaves a fine metallic glitter into the flow that the pump then cooks into varnish on the filter pleats.
The insight that stopped the week cold: the ten litre weekly loss was never one leak. Two loose unions wept a little, the crushed swage wept a little more, and the blocked return filter turned the station into a pressure cooker, pushing oil out past seals and breathers that would never weep on a clean system. The soft clamp, the high-pitched pump, the hot tank, and the daily pan puddle were four symptoms of one sludge-loaded return path wearing every seal in the station at once.
Afternoon I changed the return element to the OEM 10 micron beta 200 rating, flushed the tank with a clean charge of ISO VG 46 and drained the pump inlet strainer, and I logged the oil lab sample for water and particle count, because the varnish always leaves a second-generation story in the oil. The differential pressure after the change sat at 0.3 bar at full flow, and the tank dropped from 52 to 44 C across a two hour run, so the pressure cooker is switched off.
Evening note: a leak is not always a leak, sometimes it is a blocked return path that taxes every seal, and the ten litres a week the plant budgeted for “normal” was the machine asking for a new filter element.
Day 04 to Day 05 — the settling, and the numbers that line the bottom of the log
Day 04 the press ran a full eight hour shift after the filter change, and the sight-glass baseline told the real story: the level held flat through the whole shift, no loss measured on the dipstick at 16:00 against the morning 07:30 line. The fast-advance held at 1.4 seconds, the clamp no longer slips during the hold phase, and the pump whine has dropped to the flat, even tone of a station that is not pushing against a clog. I topped the tank to the cold full line with one litre of ISO VG 46 and marked the fill, and that one litre is the entire top-up the machine has needed in three days, against the ten litres a week it was asking for before.
Day 05 I walked the full circuit once more with the stethoscope and the torque wrench, because a leak hunt is only finished when the second pass finds nothing. Every high-pressure joint came back dry at 160 bar, every swage held its chalk ring clean through a full cycle set, and the return filter differential sat at 0.3 bar under load. I also wrote the week’s hard-won rule into the PM card: check the return-filter differential pressure before replacing another seal, because a blocked return filter taxes every gland in the station and sells itself as a slow consumable leak.
Evening note: the plant has been re-sealing this press for two years on the strength of that phantom leak, and the true cost was not the oil in the pan, it was the repeated down-time and the seals yanked out of a perfectly healthy station.
Day 06 — handover, and the station that ran dry all day
Handover with the shift supervisor at 09:00. I showed him the sight-glass graph of the week, the dipstick line that dropped 0.9 cm in twenty minutes on Day 01 and held flat all of Day 04, and he told me, with the kind of dry pride a plant man saves for a good repair, that the pan under the station has been dry for two whole shifts, a first in three weeks. The station is logged with the new return element, the re-swaged hose, the re-torqued manifold unions, and the oil sample sitting in the lab to keep watch on the varnish story.
Closing the week with the numbers: ten litres a week of loss traced to two loose unions, one crushed swage, and one sludge-loaded return filter; fast-advance back to 1.4 seconds from 1.9; pump whine returned to its healthy flat tone; tank temperature back to 44 C from 52 C; and the dipstick line that no longer moves, which is the friendliest graph a fluid power engineer sees all quarter.
Attachment A — leak hunt record
| Check | Day 01 | After repair |
|---|---|---|
| Sight-glass loss | 0.9 cm / 20 min | flat all shift |
| Return filter dP | 2.6 bar | 0.3 bar |
| Tank temperature | 52 C | 44 C |
| Fast-advance clamp | 1.9 s | 1.4 s |
| Top-up demand | 10 L / week | 1 L / 3 days |
Attachment B — hydraulic leak hunt checklist
1. Clean, mark and baseline the station before touching anything. 2. Re-torque every high-pressure static joint to drawing value. 3. Hunt dynamic leaks with a stethoscope on hoses, swages and valves. 4. Check the return-filter differential pressure; a clog taxes every seal. 5. Watch tank temperature; a hot tank points at restricted return flow. 6. Flush and re-filter when varnish or sludge shows up in an element. 7. Re-verify with a full shift and a flat dipstick before signing off.
Glossary
Swaged coupling: a hose end held by a ferrule squeezed onto the hose. Return-line filter: the filter element through which oil returns to the tank. Differential pressure: the pressure drop across the filter, a rising dP signals a clogging element. Fast-advance: the low-pressure high-speed phase of a clamp cylinder stroke. ISO VG 46: a common hydraulic oil viscosity grade.
Week-end reflection
Friday evening. Three bullets for the file. First, clean the machine before you accuse it, the chalk box and the dry pan turned a three-week mystery into a twenty minute map of suspects. Second, a blocked return path is a pressure cooker, the crushed swage and the sludge-loaded filter sold a slow leak that was really every seal in the station working overtime, and the return-filter dP gauge was the cheapest instrument on the whole press. Third, the fast-advance clock is a leak detector, when the clamp took 0.5 seconds longer than spec the machine was telling us the circuit was losing flow hours before the pan looked wet, and the quiet fix was one hose, one element, and two quarter turns with a torque wrench. The IM-4 now runs a cold, flat, stethoscope-quiet station, and the pan under the machine has stopped collecting evidence.
Field note — the two instruments that did the real detective work
One more honest note before the week closes. The dramatic tools this week were cheap: a mechanical stethoscope at forty euros and the return-filter differential pressure gauge that came with the machine. The stethoscope found the crushed swage in forty minutes of cycling, while a visual inspection alone would have missed a weep hidden inside a hose bundle, and the dP gauge called the blocked return filter at 2.6 bar when the pan still looked like a routine slow leak. The expensive logic, by contrast, is the habit of changing an element before suspecting a seal, because a filter that molts varnish will eat a station full of new seals and hand the plant the same leak next month. When the lab sample comes back with the particle count and water content, the second-generation story will be on paper, and the PM card will carry the new rule: read the dP before you touch the seals, and let the dipstick baseline tell you when the leak has really stopped. The IM-4 has been running dry-pan for two full days now, its whine back to the flat tone of a pump that is not begging, and the shift supervisor has stopped asking whether it will leak again this week, which is the quietest compliment the whole project got.
