When a differential-pressure transmitter measures level in a closed tank, one of its two pressure connections goes to the vapor space at the top of the vessel. That connection is the reference leg, and its whole job is to feed the transmitter the pressure sitting above the liquid so the instrument can subtract it and see the liquid head alone. The reference leg comes in two flavors. A dry leg is an empty tube full of gas. A wet leg is a tube deliberately filled with liquid. Choosing between them is not a preference; it depends entirely on whether the vapor in your vessel will condense. Get it wrong, or let the leg change on you, and the level reading drifts without the level itself moving at all.
Wet Leg vs Dry Leg Reference in one line: The wet leg and dry leg are the two ways to run the reference (low-side) impulse line on a closed-tank DP level transmitter. A dry leg is an empty, gas-filled line used with non-condensing vapor, while a wet leg is filled with liquid and used when the vapor condenses, so a known constant head sits on the reference. The transmitter's calibration must account for the wet leg's head, and a lost or frozen leg drives the reading off scale.
A DP transmitter on a closed tank has a high side connected near the bottom, feeling vapor pressure plus liquid head, and a low side connected to the vapor space, feeling vapor pressure alone. The transmitter outputs the difference, so the vapor pressure cancels and the reading reflects only the liquid column. That low-side line to the vapor space is the reference leg, and the reading is only as trustworthy as the pressure that leg delivers. If the reference does not accurately represent the vapor space, the subtraction is wrong and the level is wrong with it.
A dry leg assumes the reference line stays full of gas at the same pressure as the vapor space, adding no head of its own. That holds when the vapor will not condense, and it keeps the installation simple with nothing to maintain in the line. The moment condensable vapor is present, though, the assumption breaks. Steam, wet hydrocarbon gas, and many process vapors slowly condense inside a nominally dry line, and the accumulating liquid adds an unknown, growing head to the low side that pulls the reading off. A leg that is supposed to be dry but is quietly filling itself is one of the more insidious level faults, because it drifts rather than fails outright.
A wet leg sidesteps that by filling the reference line on purpose, usually with the process condensate or a chosen fill fluid, so the leg carries a known and constant liquid head. That head is a fixed offset, and calibration handles it with zero suppression so the vessel still reads correctly across its range. The wet leg trades the simplicity of an empty line for predictability: instead of hoping no liquid collects, you guarantee the line is full and account for exactly how much head that puts on the reference.
A wet leg shifts the transmitter's zero, and getting the calibration right means measuring or knowing the height and density of the fill column and suppressing that head so an empty vessel still reads zero. This is zero suppression, and it is why a wet-leg DP transmitter reads a large negative differential when the tank is empty: the filled reference outweighs the near-empty high side. Set the suppression wrong and the whole span is offset. Because the calibration bakes in the leg's head, anything that changes that head, evaporation, a partial drain, or a density change in the fill fluid, moves the reading directly.
Keeping a wet leg healthy is largely about keeping it full and keeping its fluid consistent. Fill fluid can slowly evaporate or get pushed out during pressure upsets, and if the level in the leg drops, the reference head drops and the indicated tank level rises even though nothing was added to the tank. In cold climates, an unheated or unprotected wet leg can freeze, at which point the reference is no longer a liquid column at all and the reading goes wild. Heat tracing, proper fill fluids, and periodic verification of the leg are the routine defenses.
Dry legs have their own maintenance reality: they must actually stay dry. If a supposedly dry leg is on a service that condenses even occasionally, condensate pools in low points of the tubing and introduces intermittent head that shows up as an unexplained, wandering level. Sloping the impulse line so any condensate drains back, adding a knockout or drain pot, or simply converting to a wet leg on a genuinely wet service are the usual fixes. When neither leg can be kept honest, that is often the point where remote diaphragm seals replace the impulse lines entirely.
A failing reference leg rarely announces itself; it drifts, and drift is precisely what a trending system is good at exposing. When a DP level value is polled into a cloud SCADA platform like Merobix and stored with a timestamp, a slow, steady climb in indicated level while the tank is known to be static reads as a wet leg losing fill fluid. A reading that jumps and then behaves erratically on a cold night reads as a frozen leg. Seeing those signatures on a historical trend turns a mysterious level problem into a diagnosable one, without anyone standing at the vessel guessing.
The point of watching the leg remotely is that the transmitter itself has no idea its reference is wrong. It faithfully reports the differential pressure it feels, and if that differential is corrupted by a half-empty wet leg, the output is confidently incorrect. A monitoring layer that cross-checks level against known fills and draws, or against a second measurement on the same vessel, can flag a value that no longer makes physical sense long before it causes a bad control decision or a false inventory figure.
On remote and unmanned oil and gas sites, this early warning is the whole value, because a reference leg can freeze or drain over days with no one present to notice. Alarming on a level that drifts against a stable process, on impossible rates of change, or on temperature conditions likely to freeze a wet leg lets a cloud platform put a maintenance visit on the calendar for the right reason. The instrument measures the head; the monitoring system watches for the moment the reference stops meaning what the calibration assumed.
Use a wet leg whenever the vapor above the liquid will condense, such as steam service or wet hydrocarbon gas, because a nominally dry line would slowly fill with condensate and corrupt the reading. Filling the leg on purpose with a known fluid gives a constant, predictable head that calibration can compensate. Reserve a dry leg for genuinely non-condensing vapor where the line will reliably stay empty.
A common cause is a wet leg losing fill fluid through evaporation or a pressure upset. As the reference column shrinks, it puts less head on the transmitter's low side, and the instrument interprets the smaller differential as more liquid in the tank. Refilling and verifying the wet leg, then confirming the zero suppression, usually restores the correct reading.
A frozen wet leg is no longer a continuous liquid column, so the reference pressure it delivers becomes wrong or unstable and the level reading swings erratically or goes off scale. This is a classic cold-climate failure that heat tracing and proper fill-fluid selection are meant to prevent. If freezing is a recurring risk, remote diaphragm seals that eliminate the impulse line altogether are often the better solution.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.