Water Heater Connection Leak: The Tank Is Often Innocent

Run a dry hand down a water heater's shell, from the top rim to the base, and you can find every inch dry, while a wet ring keeps spreading on the floor beside it. The two facts contradict each other, and the tank gets blamed anyway. The other version is a heater replaced eight months ago and a wet floor that came back regardless. A new appliance does not fix a fitting. The valves and connectors feeding a heater are their own set of parts, and several are usually older than the tank.
Every Heating Cycle Works the Joints Loose
Each draw of hot water triggers a heating cycle. The steel expands and contracts, and the shell is rigid enough that it does not flex to accommodate the change, so the strain ends up in the threaded connections at either end. Thermal expansion loads a threaded joint rather than pulling it open, so the damage is cumulative: the same small load applied thousands of times can work a connection loose or wear a sealing face flat. Hard water speeds that along, because scale on brass seats and gasket faces stops one face compressing flat against another.
Scale also leaves the best evidence available here: a joint that has wept on and off for months grows a crusted white or pale green ring at the point of escape, and a steel nipple leaves a rust-colored streak below its threads. A concealed leak has to be located before anything is opened, which is why a dedicated detection specialist brings the equipment and the plumbing contractor performs the repair. Nothing at a water heater is concealed, and that is why these joints go first.
The Flex Connectors and Their Nuts Get Ruled Out First
Two corrugated stainless steel connectors usually form the final link between the house piping and the tank, one on the cold inlet and one on the hot outlet. Each end carries a union nut that spins free on the connector and threads onto a nipple at the tank. The threads seal nothing: the seal is a gasket squeezed flat between the connector's flange face and the machined face of the nipple, flat against flat rather than a metal flare seating into a cone.
The union nut at each end: The gasket inside takes a compression set over years of heat cycling and stops springing back. A nut cranked down hard at installation deforms it rather than seating it.
The corrugated body: A connector bent tight to bridge an awkward gap work-hardens at the fold, and the split shows as a bead in a single trough of the corrugation, well away from either nut.
The nipple it threads onto: A separate joint into the tank, which can weep at its own threads while the nut above it stays dry.
That nipple is also where two metals meet. Tank tappings are steel, and the pipe beyond is often copper or brass, and in direct contact with water, the steel gives ground: it corrodes while the copper is protected.
[NOTE: A dielectric nipple separates the steel tank from copper piping with an insulating liner. Once that liner degrades or scale bridges it, corrosion restarts at that exact joint, which is why a rust-stained weep tends to appear at the nipple.]
The Cold Inlet Stop Fails Differently Than the Connectors
The shutoff valve on the cold line feeding the heater leaks in a way none of the other fittings do. An older multi-turn gate or globe stop seals its stem with packing under a packing nut, which has to seal against a shaft designed to move. A valve that sat in one position for a decade, then was closed and reopened once, is the classic setup for a stem weep that was not there the week before. The tell is timing: a drip appearing a day or two after work at the heater points at the stem, not at the connectors somebody just replaced.
The Relief Valve Is Reporting, Not Malfunctioning
The temperature and pressure relief valve threads into a tapping on the tank, and a discharge line runs from it toward the floor. Two different problems get called a leaking relief valve. Water at the threaded joint where the valve screws into the tank is a connection leak: mineral crust on those threads with the tube dry inside says the seal into the tank is the issue, not the valve.
Water leaving the open end of the discharge line is the other case, and it is not a connection leak at all. That valve is a safety device, so water at the mouth of the tube is the valve reporting a condition inside the tank rather than a fitting failing. What is driving that discharge is its own diagnosis with its own set of causes, covered separately from the connections in this list.
Water reaching the floor from a heater connection can be near tank temperature and will scald. A relief valve discharging steam or very hot water means the system is venting pressure, so keep clear of the discharge and call the same day.
The Drain Valve and the Expansion Tank Tee Sit Below the Eye Line
Near the bottom of the tank sits the drain valve, plastic-bodied on a great many heaters. The plastic seat distorts slowly under years of tank heat, and the resulting failure carries a timing signature none of the other fittings here has: it turns up in the days after a flush, when a valve that had sat closed and undisturbed for years was opened, run, and then asked to seat again on a face that no longer holds the shape it was molded to. That timing is worth more than it sounds, because a drain valve is a replaceable part and the thing it gets mistaken for, a tank giving out through its own body, is not.
Where an expansion tank is fitted, it hangs on a tee in the cold line and brings two additional joints with it: the threaded connection to the tee and the air valve under the cap at its end. Inside is a rubber diaphragm holding water on one side and air on the other. When it splits, water crosses to the air side, and water at that air valve is one of the few unambiguous signals in this list.
A Dry Set of Connections Moves the Question Under the Floor
Suppose every joint comes back dry, and the shell with them. At that point the water at the heater's feet may not have started there at all.
In slab-on-grade construction, the hot line leaves a garage-sited heater and runs under the concrete to the rest of the house, so a hot-side slab leak surfaces as warm water on the garage floor, which from a few feet away looks identical to a leaking hot connection. Warm water near a water heater does not rule out a slab leak, and it does not rule one in. The connections get checked first because they are visible, reachable, and quick to eliminate, not because they are the likelier answer. Once they are all dry, a leak under the slab has to be ruled out before the heater is blamed, and that is its own diagnosis, handled as slab leak repair rather than as water heater work.
Frequently Asked Questions
No, and it can make the leak worse. Sealant goes onto threads that were never holding water back, and tape shavings that migrate onto the machined sealing face will hold the joint open, so a nut that needed one gasket now leaks for a second reason. Replace the gasket, or replace the connector.
Heat trap nipples are the short fittings threaded into the inlet and outlet tappings on many tanks, each holding a small plastic ball or flapper that blocks the convection loop that would otherwise carry heat out of the tank into the piping. They are direction-specific and usually color-coded, blue for cold and red for hot, and a nipple can weep at its own threads while the connector nut above it stays dry.
Support is what gets missed. An expansion tank should be supported on a bracket or strap rather than hanging the entire load off the threaded tee, because a tank that has taken on water due to a failed diaphragm is far heavier than it was on installation day. A tank left hanging on the tee also puts a bending load on that threaded connection, which is why the tee is the joint that tends to weep first in an unsupported installation.
It does. A recirculating setup brings a pump, a check valve, and at least two more threaded unions into the area, each cycling with the same heat the connectors do. What it does not do is raise metered water use, because the pump moves the same water around a closed loop rather than pulling fresh water through the meter.
Yes, because hoses get left attached after a flush and forgotten. A weeping valve then drips into the hose instead of onto the floor and discharges wherever the far end sits, or soaks into a coiled hose and evaporates, so the floor stays dry while the loss continues. A far end lower than the valve can also siphon.
Three things, usually. A new connector stretched or squeezed to bridge a spacing it was never sized for, leaving a permanent load on both nuts. An old nipple left in place whose sealing face is pitted or out of round, so a fresh gasket cannot seat flat. And a connector was replaced only on the side that was dripping, even though both are the same age and have undergone the same number of heat cycles.
Get the connections eliminated before the heater takes the blame — a visible joint is the fastest way to rule out a leak. American Discount Plumbing serves Phoenix and the Valley. ROC #150707. Call (602) 883-2787.