Start With the Impulse Lines, Not the Transmitter
A differential-pressure flow or level reading upstream of a separator that drifts, dies, pins upscale, or lags the process by minutes is almost always a tap or impulse-line problem, not a cell problem. Sand, scale, wax, asphaltene and hydrate settle into the dead leg between the tap and the manifold and stay there, because there is no flow in that tubing to sweep them out. On oil and gas inlet separation duty this is chronic, and instruments pulled during shutdown routinely come out with slurry packed in the taps.
Swapping the transmitter is the fix that wastes the shift. Prove the cell with the manifold in two minutes, then work back to the taps.
Read the Symptom Before You Open Anything
Capture a trend of DP, line static pressure, and controller output first. A partially plugged leg still moves, which is worse than a dead one: it looks plausible, the loop gets detuned around the extra lag, and it oscillates the day the plug clears.
| Panel symptom | First suspect | Confirm by |
|---|---|---|
| DP flat at zero with known throughput | Both legs blocked, or the manifold equalizer left open | Close the equalizer; crack each drain and watch for a live push of fluid |
| DP pinned upscale, insensitive to rate changes | Low-side leg plugged; it holds the pressure trapped when it sealed | Overlay DP against line static pressure. If DP tracks static, the low leg is dead |
| Reading lags process steps by seconds to minutes | Partial restriction; impulse-line time constant has climbed | Bump the equalizer and time the return against the transmitter damping setting |
| Slow zero walk over hours, worse overnight | Wax, hydrate or freezing in the leg; tracing failed | Check tracer supply, insulation integrity, and leg temperature at the low point |
| Spiky, noisy signal at steady flow | Gas in a liquid leg, or a liquid slug in a dry gas leg | Vent the high point; refill and re-zero |
| Negative DP at true zero flow | Unequal wet-leg fill heights | Drain both legs, refill to the same level with the specified fill fluid |
| Reading jumps, then settles after valve manipulation | Mobile partial plug | Schedule a rod-out. Re-zeroing hides it for a week |
Know Why the Taps Fill
An impulse line is a zero-velocity branch off a live process line. Everything the process carries into it settles, and nothing carries it back out. Upstream of a separator you have the worst combination: produced sand, corrosion product, emulsion, paraffin and free water, at the point in the plant with the least conditioning.
Three mechanisms account for nearly all of it:
- Gravity settling in the tubing. Horizontal or sagging runs collect solids at every low point. A liquid leg that does not slope continuously down to the transmitter, or a gas leg that does not slope continuously up, builds a trap.
- Thermal deposition. The impulse line runs colder than the process. Wax and hydrate form on the tubing wall first, then bridge. Cold-weather failures that clear by mid-morning are this.
- Phase inversion in the leg. Gas breaks out of a wet leg, or condensate collects in a dry leg. The head error shows as a zero shift long before the leg blocks outright.
Length makes all three worse. Every extra metre of small-bore tubing adds settling volume and hydraulic lag. Shortest possible run, correct slope, and a drain or rodding point at the low end are the design levers.
Clear the Leg Without Wrecking the Cell
- Trend and photograph the reading before you touch a valve. You lose the evidence once you bleed it.
- Prove the cell. Close both process blocks, open the equalizer. DP should walk to zero and hold there. If it does, the transmitter is good. That is not the fault; stop chasing it.
- Sequence the manifold correctly, every time: equalizer open before either block closes, blocks closed before any vent or drain opens. One-sided static pressure on the cell is how you deform a diaphragm and buy a permanent zero shift.
- Bleed each leg in turn at its drain. A clear leg gives an immediate, steady push. A blocked leg gives a burp and then nothing.
- Rod or flush toward the process with the transmitter isolated. Use the tap valve and rodding port if the design has one. Never blow debris toward the cell.
- Refill wet legs to identical height on both sides with the specified fill fluid, and vent trapped gas at the high point.
- Restore in reverse: blocks open first, equalizer closed last.
- Re-zero only at verified no-flow conditions.
If the same tap plugs twice in a year, the installation is wrong. Stop rodding and change the design.
Choose Purge, Seal, or a Different Primary Element
The impulse-line clause of API 551 (2.2.7) treats purging as a design answer to solids service, not a maintenance workaround. Typical purged applications in oil and gas treatment are separator inlet DP flow and interface level, produced-water and sand-laden streams, knockout drum and flare header level, waxy or asphaltenic crude, and amine or glycol circuits carrying filter debris. Rules that decide whether purge works:
- Purge both legs from a common header through matched, individually metered restrictions. Unequal purge rate is a permanent DP offset.
- Purge supply pressure must stay above maximum process pressure at the tap under every condition, including relief set pressure. Loss of margin means process into the purge header.
- Use a constant-flow arrangement (differential regulator plus restrictor, or rotameter with needle valve) rather than a bare needle valve. Line pressure swings modulate a bare needle and move your zero.
- One check valve per leg, injection point as close to the tap as the piping allows so the purge sweeps the whole dead volume.
- Alarm loss of purge. A falling rotameter float is a plug forming.
- Confirm the purge fluid cannot upset the measurement it feeds or the product spec downstream — water purge into a dehydration train or an analyzer slipstream causes a bigger problem than the plug did.
Where no compatible purge fluid exists, go to remote diaphragm seals with capillaries and accept the temperature-induced zero shift and slower response, or add a flushing ring if the diaphragm face itself fouls. Where the transmitter can sit at the tap, direct mounting beats every other option because there is no impulse line to plug. Heat tracing solves freezing, wax and hydrate — it does nothing for sand.
The head-type section covers several primary elements beyond the concentric orifice, and the solids question usually decides which one:
| Primary element | Where it fits | Solids behaviour |
|---|---|---|
| Concentric orifice | Clean liquid, gas, steam; default choice, best-documented coefficients | Sediment banks up behind the plate in horizontal lines; taps blind |
| Eccentric / segmental orifice | Entrained solids or condensate; bore set low for solids, high for gas | Passes the second phase; higher uncertainty than concentric |
| Venturi | Tight permanent-pressure-loss budget, abrasive streams | Tolerant, but the pressure taps are still dead legs |
| Flow nozzle | High-velocity or erosive steam service, shorter than a venturi | Body tolerant; tap plugging unchanged |
| V-cone / wedge | Dirty, viscous, low-Reynolds and slurry duty with short straight run available | Best of the head types on slurry; wedge taps accept purge readily |
| Averaging pitot | Large lines, low loss, retractable under pressure | Do not use on solids or wax. The sensing ports are the first thing to block |
| Integral orifice | Small lines, direct-mounted | No impulse line, but a fine bore that blocks on any debris |
Verify the Fix, Then Watch the Repeats
- Step the process or bump the equalizer and time the settle. Both legs should return within the configured damping, not tens of seconds behind it.
- Cross-check the DP flow against an independent number: separator level rate-of-change, tank gauging, or a pump curve.
- On a multivariable transmitter, watch the static pressure output. A static that no longer moves with line pressure is a blocked high-side tap.
- Log purge rotameter position and supply pressure on the operator round. That number falls before the reading does.
- Re-check on the first cold night. Wax and hydrate faults hide in daylight.
Recurring traps on this class of equipment: re-zeroing to mask a wet-leg imbalance, so the offset returns in days; pressurising the cell one side at a time; tracing wax service hot enough to coke or cool enough to gel; purging a single leg; and long unsupported tubing runs whose sags each hold the wrong phase. Access is the quiet one — if the manifold, drains and rodding ports cannot be reached safely, the line stays plugged. The access clause that references rolling platforms (2.2.3) covers exactly the locations where fixed steel is impractical: tank-top instruments, pipe-rack transmitters, and column-mounted level bridles. Where a portable platform is the maintenance plan, walk the wheel route and the footprint during design review and confirm the drain and rodding points are reachable from it.
Stop when the blocker is clause wording rather than hardware. If a sentence in the impulse-line or access clauses admits two readings and a design decision turns on it, file a request for interpretation with API instead of picking one and defending it at audit. For purge fluid compatibility, seal fill selection, and capillary temperature error on a named service, take the fluid data to the transmitter manufacturer's application engineering group and get the answer in writing.
FAQ
Why does a DP flow transmitter read high when an impulse line plugs?
A blocked leg freezes at whatever pressure was trapped when it sealed. If the low-side leg plugs, the transmitter sees live high-side pressure against a static reference, so DP climbs with line pressure and pins upscale regardless of flow. Overlay DP against line static pressure on the trend to confirm it.
Why do purged impulse lines still drift in slurry service?
Almost always asymmetric purge: unequal flow between the two legs puts a fixed offset on the DP. Check that both legs run from the same header through matched, individually metered restrictors, that supply pressure stays above maximum process pressure at the tap, and that a constant-flow device rather than a bare needle valve sets the rate.
Why does API 551 reference rolling platforms for instrument access?
Some instrument locations — tank tops, pipe-rack runs, column-mounted bridles — cannot take fixed platforms without blocking access elsewhere, so a portable rolling platform becomes the maintenance route. Verify during design review that the manifold, drains, vents and rodding ports are all reachable from that platform and that the wheel path and footprint actually exist at grade.