Nine Gallons: Rebuilding the Wash Loop Separator
Reading this because you are specifying, buying or selling containers? Send the details and a grader answers in writing, usually the same business day.
A closed-loop wash line's real instrument is the make-up water meter. Ours drifted from 9 to 23 gallons per tote over five weeks because the separator's coalescing media had blinded, and nothing else in the process complained.
We publish a figure: under 9 gallons of make-up water per tote, against an industry single-pass typical of 40 or more. In February last year that figure was 23 gallons and had been for about five weeks, and nothing in the plant told us except a meter reading somebody checks on a Monday.
This is what went wrong, how it hid, and what we changed.
What a closed loop actually is
Our caustic stage recirculates. A charge of sodium hydroxide solution at 1.5 to 2.5% is held at 160 to 180 °F, pumped through a rotating spray head inside the tote, collected, and returned through three stages before going back to the tank:
- A solids stage — a settling and screening section that drops out particulates: label fragments, grit, sediment, product solids.
- A coalescing stage — media that encourages small oil droplets to merge into large ones so they can float and be skimmed off.
- A carbon polish — removes dissolved organics that would otherwise accumulate and foul the charge.
If all three work, a single caustic charge washes dozens of units. Make-up water is only what evaporates and what leaves in the final potable rinse, which is where the 9 gallon figure comes from.
How it degraded
The coalescing media blinded. Over about eighteen months of oil cycles, the media had accumulated a soap-and-solids film that prevented droplets from coalescing, so the oil stayed emulsified and went back into the tank.
Emulsified oil in the charge does two things. It reduces the caustic's effectiveness, because some of the alkali is tied up in soap. And it carries over into the rinse stages, which meant the rinse needed more fresh water to reach a clean endpoint.
The operators compensated without really deciding to. Cycles that used to finish clean needed a bit longer, and the final rinse needed an extra fill to reach neutral pH. Everything still worked. Every unit still passed. The certificates were all honest.
Why nothing complained
This is the part worth paying attention to if you run any kind of closed loop.
- Wash quality did not drop. Units still came out clean, because the operators were adding time and water to compensate.
- Final rinse pH still hit neutral. That measurement is an endpoint, not a rate — it tells you the rinse finished, not how much water it took.
- Caustic consumption rose slightly, which looked like a run of oil-heavy intake.
- Cycle times rose slightly, which looked like a busy period.
- No alarm, no interlock, no gauge out of range. Every individual signal was within tolerance.
The only thing that showed it was the make-up water meter, because water consumption is the one measurement that integrates everything the loop is failing to do.
The rebuild
Five days, most of it waiting for media.
- Drained and neutralised the charge, which was overdue anyway.
- Stripped and replaced the coalescing media entirely. Cleaning blinded media is possible and not worth the labour.
- Re-sized the solids stage. The original screening was too fine and was loading the coalescer with material that should have dropped out earlier.
- Replaced the carbon, which was partly spent and partly channelled.
- Added a differential pressure gauge across the coalescing stage, which is the direct indicator we should have had from the start.
- Added a weekly make-up water log on the board rather than a monthly meter read.
Back to 8.6 gallons per tote within two shifts of restarting.
What it cost, and what it had been costing
| Measure | Figure |
|---|---|
| Make-up water, normal | ≈9 gal / tote |
| Make-up water, degraded | ≈23 gal / tote |
| Excess water over five weeks | roughly 45,000 gal |
| Extra heating energy | substantial — it was all heated |
| Extra caustic consumption | roughly 20% above normal |
| Rebuild cost | ≈$9,400 including media and the DP gauge |
| Extra cycle time | roughly 15% on oil cycles |
The water itself is cheap in Missouri. The heating is not — every one of those extra gallons went to 170 °F — and the extra caustic is not.
What we changed about how we watch it
Three things, in descending order of usefulness.
Make-up water per tote, weekly, on the board. It is a single number, it integrates every failure mode in the loop, and it is on a whiteboard where the operators see it rather than in a monthly report.
Differential pressure across the coalescer. The direct indicator of blinding. Should have been there from day one.
A rule against silent compensation. If a cycle needs more time or more rinse water than the standard, that gets written down rather than absorbed. The operators were doing exactly what good operators do — making it work — and that instinct is precisely what hid a system fault for five weeks.
Why we published this
Because we quote a 9 gallon figure on our sustainability pages and for five weeks it was 23. If we are going to publish operating numbers, the honest version includes the periods when they were wrong and what we did about it.
The figure in our cumulative water saving is computed from the metered log, so those five weeks are in it at the degraded rate. We did not smooth them out.
903 words · published 1 April 2025 by Yusuf Baran, Wash-line supervisor at IBC Tote Market LLC. We correct posts rather than quietly deleting them; if something here is wrong, tell us and we will say what changed.