What Cylinder Drift Tells You About Which Seal Gave Up

Why this matters

A cylinder that will not hold its position gets one diagnosis in most shops, and it is piston seals. It is often the wrong repair, and more usefully, it is a repair that cannot stop drift on its own. On a cylinder with a rod on one side, fluid crossing the piston has nowhere to go unless there is also a path out of the cylinder circuit, and that second path is the one that decides whether the machine drifts. A shop that understands that spends one afternoon capping ports instead of two days pulling a cylinder that was never going to fix the complaint.

Lead with the load, because the test is where people get hurt

Every check below leaves a load held by nothing but trapped oil and a fitting somebody just installed. Before any of it:

• Put the load on mechanical supports rated for it, and leave them there for the whole test, including overnight. The caps are the thing being tested, so they are not the thing holding the load. Nobody works, stands, or reaches under the load at any point.
• Shut the machine down, relieve the circuit to zero and discharge any accumulator under the plant's energy control procedure before a fitting is touched. A charged accumulator holds full working pressure with the power off and the pump locked out, which is why stored energy is its own isolation step under 29 CFR 1910.147 in general industry. The sibling SOP on discharging an accumulator walks it.
• Never crack a fitting to relieve pressure. A stream escaping a partly-loosened joint injects through skin at pressures far below normal working pressure, and that is a surgical emergency rather than a cut. Relieve through the machine's own controls or its bleed provision, then open the joint with gloves and eye protection, expecting residual fluid that is hot.
• Trapped oil in a capped cylinder changes pressure with temperature. A warm shop overnight is enough to raise it. That is another reason the mechanical supports stay in and the caps are treated as instruments rather than as restraints.

Drift needs a path across the piston and a path out, and only one of those is necessary

Take a standard cylinder with a rod on one side. The cap end has full bore area; the rod end has that area minus the rod, so it is smaller. Now block both ports.

If the load pushes the cylinder toward retract, the cap end has to give up a volume set by the full bore area, and the rod end can only accept the smaller volume set by the annulus area. The surplus has nowhere to go, and oil does not compress meaningfully. The cylinder is hydraulically locked, no matter how badly the piston seal is passing. Push it the other way and the arithmetic reverses into a shortfall, which locks it just as hard.

Two conclusions come straight out of that, and they are not symmetrical:

• A path out of the cylinder circuit is necessary. Without one, there is no drift.
• A leaking piston seal is not necessary. With an exit path on the surplus side, a cylinder with a perfect piston seal still drifts, because the surplus leaves and the shrinking side is fed through the same leak path.

So the repair priority is the opposite of the usual instinct. Closing the exit path stops drift even with a bad piston seal. Replacing the piston seal cannot stop drift while the exit path is open. The piston seal still deserves attention, because fluid crossing it dissipates energy as heat inside the cylinder, which is what the sibling HowTo on finding a system's heat picks up as a hot cylinder body between cooler lines. But it is not what closes a drift complaint.

The four candidate paths

Only the last one loses fluid from the machine. That single fact separates one candidate from the other three before you touch a wrench, and the reservoir level over a week is the cheapest measurement in this whole article.

The cap sequence, and what each result eliminates

Each step is a separate overnight or shift-length hold with the load on its supports, the position marked on the rod with a scribe or tape at a fixed reference, and the elapsed time written down. Drift is a rate, so an unrecorded duration makes the number useless.

• Both cylinder ports capped. Drift means fluid is leaving the cylinder body: rod seal, port fitting or a cracked tube. No drift means the cylinder is not losing fluid externally, and nothing at all has yet been established about the piston seal.
• Rod-side port capped, cap-side line connected normally. Retract now requires the rod end to grow, and the only supply is across the piston. Drift here proves the piston seal is passing. No drift means it is holding.
• Cap-side port capped, rod-side line connected normally. This one only blocks when the load drives the cylinder to extend, because the cap end would then have to be fed a volume larger than the rod end can supply across the piston. Where the load drives retraction, as a settling boom does, a passing piston seal still lets it move: the cap end sheds its full bore volume across the piston, the rod end takes the annulus share, and the surplus leaves through the connected rod-side line. Run it once to confirm your cap actually seals, and read the result against the direction the load actually pushes.
• Both lines connected, block progressively further out. Cap at the valve's cylinder port with the hose and cylinder left intact and full. Holding there means the exit path is inside the valve block. Drifting there means the exit is in the hose or the cylinder, which the first step already spoke to.

The case: a boom that settled overnight, twice

The signal. A machine whose boom settled about 5 inches over a 10 hour night, so 0.5 inches per hour. The customer's own history mattered: the piston seals had been replaced about six months earlier for this same complaint, and the drift was back inside a week.

Reservoir level first. Marked and read across a full week of nightly settling. It did not move. An external leak large enough to lower a boom 5 inches a night moves a reservoir level visibly, so this was already pointing away from the rod seal before anything was disassembled.

Step one, both ports capped. Boom on stands rated for it, machine locked out, circuit relieved, accumulator discharged, caps fitted with gloves and eye protection on a cold circuit. Ten hours: no measurable movement. The external candidate is dead, and this matches the reservoir reading rather than merely agreeing with it by luck. Note what this step did not do: it said nothing about the piston seal, because with both ports capped the cylinder is locked either way.

Step two, rod-side port capped only. Ten hours: 4.5 inches, so 0.45 inches per hour. Under the rule above, the rod end can only be fed across the piston, so the piston seal is passing, and it is passing enough to produce 0.45 of the 0.50 inches per hour seen in normal configuration, which is 90 percent of the complaint rate. The six-month-old seal job either did not hold or was not the whole story.

Step three, capped at the valve's cylinder port, hose and cylinder intact. Ten hours: no measurable movement. The hose is not the exit path and neither is the cylinder body. Whatever lets fluid out is inside the valve block.

What was true. A worn closed-centre spool in the directional valve, leaking its cap-side port to tank, plus a piston seal that was genuinely passing. The valve's tank port ran warm during a hold, which is the second symptom the thermal survey card would have predicted and which was checked after the cap tests rather than before, so it confirmed rather than led.

Why the previous repair failed. The shop before had the piston seal right. They had one of the two conditions, and it was the one that is not necessary. With the spool still passing, the boom kept settling: surplus cap-end oil left through the valve, the rod end was fed through the same leak path, and the fresh piston seal changed the rate rather than the outcome.

What the fix priority was, and why. The spool closed the complaint. The piston seal was done at the same time because the machine was open and because fluid crossing the piston is a heat source that shortens the fluid and the seals around it, but the two were reported to the customer as one repair and one improvement, not as two repairs, because only one of them was going to stop the boom settling.

What would have changed the conclusion. If step two had come back with no movement, the piston seal would have been cleared and this would have been a valve-only job. If step three had come back drifting, the exit path would have been in the hose assembly and the cylinder would still not have needed to come off. Both alternatives close in the same afternoon, which is the point of running the sequence rather than guessing.

Reading a drift rate rather than a drift

A cylinder that moves at all is not automatically faulty. Every real spool and every real seal passes something, machine builders publish an acceptable drift rate for held functions, and that rate belongs to the builder rather than to a rule of thumb. Ask for it before you condemn anything.

What travels without a specification is the trend. Mark the rod, record the hold time, and compare the rate to the same measurement taken the same way on the same function. A rate that has doubled since the machine was commissioned is a finding regardless of whether the absolute number would have failed anyone's threshold, and a rate you measured once is a number with nothing to compare it to.

How to verify before you pull a cylinder

• Confirm the caps seal. Run the control step above. A weeping cap turns every result into a false positive, and it is the one error that makes a healthy cylinder look guilty.
• Record every hold as a rate, with its duration. Two tests of different lengths compared as raw inches is the most common arithmetic error in this whole procedure.
• Get the second symptom before you spend. A leaking spool runs a warm tank port during a hold. A passing piston seal runs a hot cylinder body between cooler lines. If your cap sequence says one thing and the temperatures say another, run the sequence again rather than picking the one you like.
• Check whether the function is supposed to have a load-holding valve and does not. A boom that relies on a directional valve spool to hold a load is a design question as well as a drift question, and the sibling card on load-holding valves owns it.

References

• Machine builder documentation for the acceptable drift rate on each held function, the cylinder's bore and rod dimensions, and whether a load-holding valve is specified for that circuit
• 29 CFR 1910.147, for control of hazardous energy including stored hydraulic energy, in general industry, before any cap is fitted or any joint is opened
• Safety data sheet for the specific hydraulic fluid, governing skin and eye contact with hot fluid released when a joint is opened
• See related: What a Load Holding Valve Is Preventing and When It Is Not Enough; How to Find Where a Hydraulic System Is Making Its Heat; Discharging a Hydraulic Accumulator Before Any Work Begins