How to Decide Between Repacking and Converting to a Seal

Why this matters

This decision gets made on preference, and preference is usually "seals are modern, packing is old." That reasoning puts a mechanical seal onto a shaft running six thousandths of runout, into a box too tight to circulate anything, on a service full of grit, and it fails inside a month, faster than the packing it replaced and with a bigger bill attached. It also leaves packing on services where a controlled leak was never acceptable in the first place. The decision is not about which technology is better. It runs through one hard check, three checks that convert into hours rather than a veto, and one judgement call, and the two pumps at the end of this card come out on opposite sides of it while failing at different checks.

Before you measure anything on the pump

Everything below requires the pump stopped. Lock and tag the disconnecting means under 29 CFR 1910.147 rather than switching it off, relieve pressure at the casing vent and confirm it on a gauge, then drain before the gland comes off.

Turning the shaft by hand to read runout means the guard is off. Do that only after the lock and tag is on, turn the shaft with a strap or by the outside of the coupling hub, and keep fingers out of the coupling gap and clear of the keyway.

Drain to a routed, closed container, not an open bucket, whenever the SDS names a vapor or aerosol route - and note that draining that fluid into an open container in an occupied room is a hazard your own procedure creates, not one the pump presented. Run the room's ventilation, follow the SDS, and where it calls for respiratory protection that puts you inside the employer's program at 29 CFR 1910.134, which requires fit testing before first use and at least annually under 1910.134(f)(2).

Assume sump and wet-well water is biologically active. Wet, that is a glove and eye problem. Once the packing and debris dry out and get picked at with a hook, the route becomes inhalation and the control is respiratory rather than a glove.

The shape of the gate

Five checks, and they do not carry equal weight, so running them as a scorecard where three passes beat two fails gives the wrong answer.

One is a hard check. It decides whether packing is an option at all. If it fails, packing is out and the remaining question is only which seal and what the box needs.

Three convert into work. They do not veto a conversion, they price it. Anything they fail can be bought with hours and parts, and knowing that is what stops a shop from calling a conversion impossible when it is only expensive.

One is a judgement. It has no threshold and it cuts both ways.

The hard check: can the service tolerate a controlled leak

Packing leaks by design, at a rate that is measured rather than eliminated, and a sibling card covers why the drip is the coolant and what happens when it is stopped. So the first question is whether that leak is acceptable, and it is answered from the fluid and the room, not from the pump.

A controlled leak is out where the fluid is toxic, flammable, regulated for discharge, corrosive to what it lands on, or expensive enough that losing a bucket a week matters. It is out where the room is occupied and the SDS names a vapor route, because a drip on a floor is a continuous evaporating source. It is out where the discharge would reach a drain that is not permitted to receive it, which is a question for the facility rather than for you. It is acceptable on ambient water into a floor drain, on services already contained by a bunded area, and on many industrial duties where nobody has ever thought about it because it has never been a problem.

Answer this one before any tool comes out of the truck. It is free, it is decided by information rather than measurement, and when it fails there is nothing left to decide.

The three that convert into work

Shaft or sleeve condition at the box. A face pair has to track a face that is moving with the shaft, and the tolerance is small: commonly a couple of thousandths of an inch of total indicated runout at the seal location, with the seal manufacturer's own figure governing. Surface finish and diameter matter too, and a sleeve grooved by previous packing is not a seal surface at any finish. Reading this takes a dial indicator, a magnetic base and ten minutes. Failing it costs a sleeve, or a shaft, or the bearing and alignment work that caused the runout in the first place.

Box geometry. A box designed for packing is a small, close bore, and its clearance around the shaft is deliberately tight. A seal needs radial room for the springs or bellows, axial length for the setting, and enough annular space for the flush to actually circulate rather than dead-end. Some pumps were supplied with a seal chamber and packed later, in which case there is room. Some were only ever packing boxes, and then the conversion needs an adapter, a bored-out chamber, or a seal specifically dimensioned for a packing bore. Measure the bore and the depth, and get those two numbers to the seal supplier before anything is ordered.

Flush availability. A seal needs its liquid film managed for temperature, cleanliness or margin above boiling, and each of those is a piece of piping. Is there a discharge tapping? A clean liquid source at a pressure above box pressure? Room to route a cooler? If none of those exists, the conversion includes making one, and on an abrasive or hot service it is not optional.

Each of these three has an hours number attached, and the honest conversation with an owner is that number, not a verdict.

The judgement: who is there when it lets go

This one cuts both ways and it is worth stating in both directions, because it is regularly used as a one-sided argument.

Packing needs somebody. It is adjusted, checked and eventually repacked on a cadence. A remote or unattended pump is a poor packing candidate simply because nobody is going to walk past it.

But packing fails gracefully and a seal does not. A stack past its life leaks more, then more again, and gives days or weeks of warning. A mechanical seal holds essentially nothing visible right up until it holds nothing at all, and then the full leak arrives at once, unattended, at whatever hour it chooses.

So an unattended pump argues for a seal on the maintenance ground and against a bare single seal on the failure-mode ground. Where both apply, the answer is usually a seal plus containment: a dual arrangement with a barrier fluid and an alarm on the reservoir, or a containment seal with leak detection, so the sudden failure is bounded rather than absorbed by the room.

Pump A, run through the gate

A transfer pump lifting from a below-grade sump. Ambient water carrying sand and silt. Plant room with a floor drain, millwrights on shift. Currently packed, and it has been repacked twice this year.

The hard check passes, so packing stays on the table and the three work checks now price the alternative. All three fail, and they fail expensively: correcting 0.006 in of runout is a rotating-element job, not a sleeve swap, because six thousandths at the box points at a bent shaft, worn bearings or a coupling problem rather than at surface wear. Add a machined or adapted chamber and a clean-flush line the room does not have, and the conversion runs to several times the hours of a repack, with parts that are not on the truck.

Decision: repack, and fit a lantern ring with a clean flush above box pressure. That last part is the actual fix for the twice-a-year repacking. The grit is what is eating the stack, and an outward flow across the inner rings keeps it out. Then find the runout, because six thousandths is chewing the packing too and it will chew the next stack the same way.

What would flip this: if the runout came back at a couple of thousandths after a bearing job that was happening anyway, two of the three work checks would be down to a chamber adapter and a flush line, and on a pump being repacked twice a year the conversion starts to pay for itself in visits.

Pump B, run through the gate

A transfer pump in an occupied room, handling a fluid whose SDS names its vapor as a respiratory hazard. Currently packed, by a previous owner. Runs unattended overnight.

The gate closed at the first check, and everything after it changed meaning. Packing is not an option on this service and never was; the pump has been running out of compliance with its own fluid's handling requirements since somebody packed it. The remaining three work checks are not deciding anything, they are just telling you the conversion is cheap: the surfaces pass, the chamber is already there, the tapping is already there.

The judgement check is what shapes the specification rather than the decision. Unattended plus a vapor hazard plus a seal's sudden failure mode means a bare single seal puts an unbounded release into an empty occupied space at three in the morning. Decision: convert, to a dual arrangement with a barrier fluid and an alarm on the reservoir, so that the first seal failing is an alarm rather than an event.

Two pumps, one gate, opposite answers, and they did not fail at the same check. Pump A was decided by three expensive work items after the hard check let packing stand. Pump B was decided at the hard check, in the first minute, from a document rather than a measurement.

How to verify you got this right

Write the failing check on the ticket, not the conclusion. "Repacked" tells the next tech nothing. "Repacked: runout 0.006 at the box, no flush source, grit service, lantern ring fitted" tells them what to re-test and what would change the answer. On a conversion, "converted: SDS vapor route, packing not permissible on this service" tells them not to un-do it.

Re-measure the runout after any conversion, with the new sleeve fitted. A sleeve is a part with its own runout, and fitting one to a shaft that was within tolerance can put the assembly outside it. Reading the shaft and then not reading the assembly is the most common way a correctly specified conversion still fails in its first month.

On a conversion, take the first seal's life as data rather than as a verdict. If it goes early, pull it and read the faces against the failure-signature key in the sibling card before ordering another. A conversion that fails on abrasives, on heat, or on installation are three different mistakes, and only one of them means the gate was run wrong.

References

• 29 CFR 1910.147, control of hazardous energy, for lock and tag plus pressure relief before removing a gland or turning a shaft with the guard off
• 29 CFR 1910.134, respiratory protection, including fit testing before first use and at least annually under 1910.134(f)(2), where the SDS names a vapor or aerosol route
• Seal manufacturer's data for maximum shaft runout, surface finish and required chamber dimensions, and the pump manufacturer's drawing for the box bore and depth actually fitted
• See related: What Packing Does That a Mechanical Seal Cannot; What a Mechanical Seal Is Actually Sealing Against; How a Mechanical Seal Fails and What You See