Desuperheater not cooling: why steam temperature overshoots and what to check first
If steam downstream of the desuperheater is sitting above setpoint, the nozzle is the last thing to look at, not the first. Four causes cover most cases, and three of them cost nothing but a walk down the line: the unit is running below its turndown, the temperature element is in the wrong place, the spray water differential pressure has collapsed, or there is not enough straight pipe for the water to finish evaporating before it reaches the thermowell. Work through them in that order.
The load moved and the desuperheater did not
Check today's steam flow against the flow on the datasheet before anything else. Plants get de-rated, lines get re-purposed, one of two boilers gets retired — and the desuperheater sized for the original design case is now being asked to work at a third of it.
Below the nozzle's minimum controllable water flow, atomisation stops. Droplets get coarse. At the same time steam velocity has fallen, so the pipe is no longer carrying those droplets — they settle and run along the bottom of the line. The controller sees a hot reading, opens the spray valve further, and the extra water makes the atomisation worse, not better. That is the cycling you are watching on the trend.
The expensive wrong move here is ordering a larger unit because "it isn't cooling enough." A larger unit has a higher minimum flow. You would be buying more of the problem.
The thermowell decides what the controller believes
A temperature element set closer to the injection point than the datasheet allows does not read steam. It reads a wet mixture, and wet reads low. The controller then cuts spray on a false reading while the steam that reaches the superheater outlet, the turbine bypass or the process user is genuinely too hot. The reading looks controlled; the plant is not.
Too far in the other direction has its own failure. Deadtime scales inversely with flow, so a loop tuned at full load will hunt at part load — same valve, same nozzle, nothing broken. Before touching tuning constants, do a step test at the load you are actually running.
Two more things worth ten minutes: confirm the element feeding the controller is the one you think it is, and check insertion depth. A heavy well with an air gap between the element tip and the well bottom adds seconds of lag on its own.
Spray water: differential pressure, temperature, solids
Droplet size is set by the pressure differential across the nozzle, not by how far open the spray valve is. If the water is drawn off a feed pump discharge that gets throttled back with load, that differential falls exactly when the steam load falls. Two failures arrive together and it looks like the nozzle gave up.
Measure water pressure at the desuperheater flange and steam pressure at the same point, at the load where the problem shows. Not at design load. Not from the DCS faceplate for a transmitter three hundred metres away.
Then look at what the water carries. Spray water becomes steam; the solids in it stay behind. Deposits build at the orifice and the seat, capacity drops slowly over months, and the first symptom is usually overshoot at high load only — the opposite pattern from a turndown problem. Same for a strainer nobody has opened since commissioning.
Also worth confirming: no spray desuperheater brings steam down to saturation. Every design specifies a minimum approach above saturation temperature at the measuring point — commonly somewhere between about 5 and 15 K depending on type, though the figure on your datasheet is the only one that counts. If your setpoint sits inside that margin, nothing is faulty and no amount of water will get you there.
Reading the symptom before you order parts
| What you see | Most likely cause | How to confirm | What happens if you fix the other thing |
|---|---|---|---|
| Overshoot only at low load, temperature cycles | Water flow below the nozzle's controllable minimum | Compare actual water flow with the nozzle capacity curve at that point | A like-for-like nozzle changes nothing; a larger unit raises the minimum and makes it worse |
| Was fine last year, now overshoots at high load | Eroded or fouled nozzle, lost capacity | Strip and inspect the nozzle; get a spray water analysis | Re-tuning the loop chases a shrinking capacity and never catches it |
| Local reading normal, downstream equipment complains | Water still evaporating at the thermowell | Measure injection-to-well distance against the datasheet; compare with a second element further downstream | Adding spray on a false low reading drives liquid water into the line and cracks the wall |
| Temperature drops then swings hard | Deadtime longer than the tuning assumes | Step test at the present load, not at design load | Raising gain makes the swing bigger |
| Spray valve at 100%, temperature barely moves | Insufficient differential at the nozzle, or the valve is not stroking | Measure both pressures at the unit; verify actual stem travel against the positioner signal | Replacing the desuperheater when the pump or the positioner is the fault |
The straight run you actually have, not the one on the drawing
Evaporation length is stated in metres of straight pipe, and it is not interchangeable with a count of pipe diameters — it grows with droplet size and shrinks with steam velocity. Retrofits get this wrong constantly, because the spool piece was chosen to fit the space available, not the evaporation distance.
Put an elbow inside that distance and unevaporated water hits the outside of the bend. You get thermal fatigue cracking at the extrados and a wet, unrepresentative reading anywhere after it. Horizontal runs let droplets settle out; vertical-down flow is kinder to both. And if the liner or thermal sleeve downstream of the injection point was omitted during a repair, the pipe wall takes the quench directly — that shows up as cracking long before anyone connects it to a temperature problem.
Design type matters here too. Fixed-orifice nozzles hold atomisation over a modest water turndown; variable-area and steam-atomised designs stretch much further, and a quench-type desuperheater handles the geometry differently again. If the operating window has genuinely changed, the honest fix is a different type, not a bigger version of the one you have. The steam conditioning and control valve range covers the usual alternatives, and Kiekens DSH in the Netherlands builds specifically for this end of the problem. Where the unit is sound but the internals are worn, nozzle and trim refurbishment is usually faster than a replacement lead time.
The load point commissioning never tests
Start-up happens when the plant is being pushed. The boiler is run up, the station holds setpoint at or near full steam flow, the sheet is signed, and the loop is handed over. What almost never happens is a repeat of that test at the lowest load the plant will actually run at. Nothing looks wrong on day one, which is exactly why the check gets dropped — and why the complaint arrives six or twelve months later, usually after a seasonal swing or a changed production schedule parks the plant at part load for weeks at a time.
The mechanism is turndown, on two devices at once. At low steam flow the spray water demand is small, so the control valve works close to its seat, where its characteristic flattens and it may sit under the manufacturer's minimum controllable flow. At the same time the differential pressure across the nozzle falls. Below the minimum atomising differential, the spray stops breaking into fine droplets and leaves as a coarse jet. Water then wets the liner and the pipe wall instead of flashing into the steam, and reaches the sensor as intermittent wet slugs rather than as cooling. The reading dips, the controller closes the valve, and the real temperature climbs behind it — then overshoots hard when load comes back up.
The commissioning step that prevents all of this takes an afternoon: walk the load down in steps, and record the lowest steam flow at which the loop still holds setpoint without hunting. Compare that number with the plant's genuine minimum. If the plant will live below it, the answer is trim, a variable-area nozzle, or a second smaller spray valve for the low range — all far cheaper decided before the station is welded in than after.
If you are trying to work out whether an existing station covers your low-load range, a controller trend of steam temperature and spray valve position across one full low-load period is usually enough to start from — send that over and we can look at where the loop loses control.