- November 21, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
The alarm history looks harmless until someone sorts it by date. Then a pattern jumps out. The retrofit emission control device (RECD) on a DG set behaves for weeks, and then in the second week of the monsoon it throws a pressure alarm, a communication fault or a heater trip. A dry fortnight follows and nothing happens. The next heavy shower brings the alarm back.
When a fault follows the weather, it is tempting to say “it’s the rain” and move on. That is only half an answer, because rain is not one thing. Wet weather brings humid air, cold surfaces that collect condensation, water running down a stack, damp cabinets, wet insulation, more grid outages (and so more generator starts) and sometimes fuel tanks that take in water through their vents. Each of these reaches the RECD by a different route and needs a different fix. Spending money on a new filter when the real culprit is a wet sensing tube is an expensive way to learn the difference.
This guide is built around following the water. We start with a simple weather ledger that tells you whether the correlation is real. Then we trace five routes by which moisture can reach an RECD installation, show how to separate genuine moisture effects from the duty changes that come with wet weather, and give you a rain-day test protocol that produces evidence in a single afternoon. The advice is general. Your supplier’s manual and ingress ratings for your device come first.
Why “only in wet weather” is such a useful clue
Most faults give you very little to work with. A fault that appears only in wet weather gives you a condition to test, which is a gift. It says that the fault depends on something that changes with the weather and that the same equipment behaves normally without it. That narrows the field in three ways.
First, it makes a permanent mechanical defect less likely. A cracked filter core or a collapsed baffle does not care whether it is raining. If the symptom vanishes on dry days with no other change, the hardware is probably not the root cause, although a marginal part may be tipped over the edge by moisture.
Second, it points to something that moisture can change: an electrical resistance, a pressure signal, a surface temperature, the weight of a soot deposit or the position of a damper that sticks when wet.
Third, it gives you a way to prove your answer. If you think a wet sensing tube is responsible, you can dry the tube and see whether the fault disappears in the next rain. A hypothesis that can be tested is better than one that can only be argued.
A caution belongs here. A truck-maker service bulletin published through the US safety regulator describes the filter differential pressure sensor on a heavy-duty engine as susceptible to moisture intrusion, which can set false fault codes and lead to unnecessary filter replacement (NHTSA-filed bulletin). That is a vehicle engine, not a stationary RECD, but it shows the pattern: a wet measurement chain can cause an alarm that looks like a blocked filter. Treat it as an illustration of a failure mode, not as proof that your sensor behaves the same way.
Step one: build a weather ledger
Before testing anything, check that the pattern is real. Memory is a poor witness. People remember the alarm on the rainy night and forget the quiet rainy weeks. A ledger removes that bias.
Take the alarm and event log from the RECD controller (or the building management system) for the last six to twelve months and add weather columns. The table below shows a layout. Rainfall and humidity can come from a local weather service or a cheap logger in the plant room.
| Date and time of event | Event (alarm, trip, comm loss) | Genset running? Run hours that day | Rain in the previous 24 h (mm) | Relative humidity (%) | Ambient temperature (°C) | Plant room condition | Load band during event |
|---|---|---|---|---|---|---|---|
Add one more row type that most sites forget: the control days. For each event, note one or two comparable days with no event, when the set ran at a similar load for a similar time. If the set ran on ten rainy days and alarmed on seven, that is strong. If it ran on ten rainy days and alarmed on one, the link is weak. And if the set ran on forty dry days with the same alarm on six of them, the weather is probably a red herring.
Three patterns are worth looking for. Lag: some effects appear during rain, others a day later when soaked insulation steams on start-up. Dose: a fault that scales with rainfall points to water entering, while one that follows humidity points to condensation. Time of day: faults at the first start after a damp night suggest condensation, and faults mid-run in a downpour suggest water entering from outside.
Our guide to the RECD event log suggests what else to capture, and the article on detecting missing data in RECD monitoring records will help if your log has gaps on exactly the days that matter.
Route 1: moisture in the pressure-sensing chain
This is the most common culprit for faults that appear and disappear. A differential pressure reading depends on two small tubes, a transmitter and a connector. Each has ways to go wrong in damp conditions.
Water in the tubes. The tubes run from taps on a hot exhaust to a cooler transmitter. Moisture in the exhaust condenses in the tube wherever it is cold, and monsoon air makes the tubes colder and wetter. A slug of water in a tube changes the reading, damps it or freezes it at a value. Tubes with a sag or a low point hold the water. Tubes that slope away from the tap let it drain back into the exhaust, where it is harmless.
Water at the transmitter. Some transmitters have vent holes or breather membranes. Water drawn in through them, or condensation inside a poorly sealed housing, can cause drift or a false reading.
Water at the connector and cable. A connector that is not fully sealed lets damp air in. Corrosion at the pins raises resistance, so the signal wanders. Cable glands that were not tightened or that point upward collect water.
What it looks like in the data. The pressure trace is noisy, jumps without a change in load, sticks at one value, or disagrees with an independent gauge. The effect is worst after rain and clears when the sensor dries. The alarm may be a high differential pressure alarm, a low-pressure alarm or a sensor-fault message.
How to test it. Compare the controller reading with a calibrated portable gauge at a steady load during a damp spell. Then dry or blow out the sensing lines with the system cold and isolated, as the supplier’s procedure allows, and compare again. Check that the tubes slope continuously back towards the tap and that nothing sits below the transmitter inlet. Compare the symptoms with our article on separating a sensor fault from an RECD performance problem, and see pressure sensors versus pressure switches for why instrument types behave differently when wet.
Route 2: condensation inside the exhaust path
An exhaust that has been standing cold in humid air is a condensation machine. Combustion produces a lot of water vapour. On a cold start, that vapour meets cold pipe walls, a cold silencer and a cold canister, and turns into liquid. Short runs make this worse, because the system never gets hot enough for long enough to drive the water out.
Where the water goes. It collects at low points: silencer drains, the bottom of horizontal runs, the lower part of the canister and any dip in the pipework. It soaks into insulation, wets soot deposits and can run into sensing taps.
What it looks like. The fault appears on the first start of the day or after a long idle, in damp or cool weather. White vapour may show at the outlet. Pressure readings may be high at first and then settle as the system warms. Drains show water or a rusty, sooty liquid.
Why it matters for the RECD. Wet soot is heavier and denser than dry soot and raises differential pressure faster per gram. Condensate also corrodes parts over time. A set that spends the monsoon running short, lightly loaded test runs is exactly the situation in which this happens. Our article on RECD regeneration after wet stacking explains the soot side, and choosing an RECD for a generator running at low load covers duty.
How to test it. Check drains and low points for water. Compare readings from a cold, damp-morning start with a hot restart in the afternoon. If the early reading is higher and falls as the system warms, moisture is likely. A loaded run of adequate length, agreed with the supplier, should clear it. If the problem is limited to the first few minutes after a cold start, see our companion article on RECD pressure spikes at startup.
Route 3: rain entering from the outside
Water can enter an exhaust system from the top, from the side or through the walls of a plant room.
The stack and rain cap. A badly designed or corroded rain cap lets rain run down the stack, particularly in wind. Water reaching a hot canister can produce steam, local cooling and wet deposits. Water standing in a silencer or lower pipe adds weight and corrosion.
Insulation and heat shields. Soaked insulation holds water for days. When the system heats, the steam and smell appear, and insulation that has lost its insulating value changes the temperature behaviour of the canister and the readings of nearby sensors. Our guide on RECD insulation and heat shields explains the arrangement and what to do about weatherproofing.
Seams, doors and penetrations. Door gaskets, welded seams and conduit entries on a canister or cabinet can let rain in if the sealing has aged.
What it looks like. The fault appears during or just after rain and is stronger in wind-driven rain or after a heavy downpour. Staining or rust marks form below joints and seams. Insulation is damp or stained. Water drips from drains at the bottom of the silencer after a storm.
How to test it. Inspect the cap, seams, door gaskets, wall sleeves and insulation after a heavy shower, from a safe position and with the system isolated and cool. A hose test, done carefully and never on a hot system, can reproduce the leak path. Our article on protecting an RECD against rainwater and condensation covers the design measures, and the guide to RECD installation in a flood-prone plant room addresses the situation where water arrives at floor level. For coastal sites, where humidity and salt combine, see choosing an RECD for coastal, salt-laden environments.
Route 4: damp air at the intake and in the plant room
The engine breathes the air around it. If that air is wet, or the intake path is wet, the effects reach the exhaust.
A wet or restricted air filter. An air filter element that has soaked up moisture restricts flow and can raise soot output, which loads the RECD faster. Check the element, the housing seals and the intake duct for water. Our generator air filter maintenance guide explains what a healthy element looks like.
Louvres and ducts. Rain-driven water entering through intake louvres, or a blocked drain on an intake plenum, can wet the filter and the room.
Recirculation. In humid still air, hot exhaust and radiator air can recirculate into the intake. Our guide on RECD installation near fresh air intakes covers the layout.
What it looks like and how to test it. A rise in smoke or soot loading during monsoon weeks, with a damp or choked air filter and a humid plant room. This route is rarely the sole cause, but it adds to the others. Inspect the filter and intake path during a wet spell, compare the air-filter restriction with the engine maker’s limit and look for wet patches below the louvres.
Route 5: electrical and control ingress
The RECD is not only a pipe. It has a controller, heater circuits, sensors, cables and sometimes a communication link. Damp air is hard on all of them.
Insulation leakage in heater circuits. Moisture in a terminal box or on an element’s terminals lowers insulation resistance. Earth-fault protection can then trip in damp weather and clear when things dry.
Condensation in the controller cabinet. Cabinets that breathe in humid air and then cool overnight form condensation on boards and terminals. Faults may be intermittent, with controller resets, false inputs or a communication loss. Cabinet heaters and breathers are there for a reason, and when they fail, the monsoon finds out.
Corroded connectors and terminals. Resistance changes create noisy sensor signals, false temperature readings or intermittent communication. The mechanism is the same as in Route 1 but located in the wiring.
Battery and supply issues. Damp or corroded battery terminals reduce cranking and control supply voltage. A dip at start-up can cause a controller reset, which is a different problem from moisture in the exhaust but is triggered by the same weather. See our article on RECD regeneration and battery-supported controls.
Communication links. Network cables, converters and junction boxes in damp conditions drop packets or lose connection. See RECD communication failure with the DG controller.
What it looks like. Alarms with no matching physical change, such as comm loss, sensor out of range, heater earth fault or controller restarts, clustered in damp periods and clearing when it dries. The exhaust itself behaves normally on an independent instrument.
How to test it. With isolation and permits in place, inspect terminal boxes and cabinets for moisture, rust, tracking marks and loose glands. Measure insulation resistance on heater circuits when the system is cold and damp, and again when dry, using the supplier’s method and limits. Check that cabinet heaters and breathers work.
Separating moisture from the duty change that comes with the rain
Here is the trap that catches many careful engineers. Wet weather changes how the generator is used. Monsoon storms bring grid outages. More outages mean more starts, more short runs and more load transfers. A set that runs perhaps ten hours a month in the dry season may run fifty in July. The alarms may be caused by the extra duty and not by the water.
The two explanations leave different fingerprints.
| If the cause is… | You would expect… |
|---|---|
| Moisture itself | The same fault on rainy days even at ordinary duty. A fault that appears on a weekly test run in rain but not in dry weather. Fault clears after the system dries. |
| Duty change | The fault tracks run hours, number of starts or time at light load, not rainfall. It appears after several outage-driven runs, whatever the weather on the day. |
| Both together | The fault appears only when a wet spell and heavy duty coincide, such as the third outage in a week with damp insulation. |
To tell them apart, put run hours and number of starts in your ledger, then compare rainy days at normal duty with dry days at heavy duty. Our guide on using DG load logs to assess RECD suitability shows how to summarise load, and the article on passive RECD regeneration and seasonal changes explains how season alters regeneration behaviour.
The rain-day test protocol
When the ledger suggests a link, run a controlled test. The aim is to see whether the fault can be made to appear or disappear by changing one variable at a time. Plan it with the supplier and the electrical lead, and keep to safe working practice.
Before the next wet spell
- Take a dry-weather baseline at three loads: differential pressure, temperatures, heater current, controller supply voltage and an independent gauge reading.
- Photograph the stack outlet, rain cap, canister seams, cabinet interiors and insulation.
- Fit a low-cost temperature and humidity logger in the plant room and in the controller cabinet.
- Note the insulation resistance of heater circuits when dry, if the supplier’s procedure includes this.
During or just after rain 5. Repeat the same readings at the same loads. Compare with the baseline. 6. Check the independent gauge against the controller. A difference that was absent when dry points to the sensing chain. 7. Inspect for water at drains, cabinet interiors, glands and insulation. 8. Record the first-start-of-day behaviour separately from the mid-day behaviour.
One change at a time 9. Dry or replace the sensing tubes and re-route any sag (with the system cold and isolated). 10. Re-seal and re-torque cable glands, and fit or repair cabinet heaters or breathers. 11. Cover or repair the rain cap, flashing or insulation jacket. 12. Run a loaded session of adequate length after a damp period to drive off condensate, with the supplier’s agreement.
After each change, wait for the next comparable weather event and record the result. If you change five things at once and the fault disappears, you learn that the combination worked, but not which part mattered. That may be good enough for a quick fix, and it is poor evidence for the next site.
Reading the pattern: a lookup table
| What you see | Likely route | First check |
|---|---|---|
| Noisy or stuck pressure signal in rain, independent gauge normal | 1 Sensing chain | Tubes, transmitter vent, connector |
| High reading on cold, damp mornings that settles on warming | 2 Condensation | Drains, low points, tube slope |
| Fault during heavy or wind-driven rain, staining below seams | 3 Rain entering | Rain cap, seams, flashing, insulation |
| Faults in monsoon weeks, wet air filter, humid plant room | 4 Damp intake air | Air filter, louvres, recirculation |
| Comm loss, resets or heater earth faults in damp spells | 5 Electrical ingress | Cabinets, glands, insulation resistance |
| Fault tracks outage count more than rainfall | Duty change | Run hours, starts, light-load share |
| Fault appears only when a wet spell and heavy duty coincide | Combination | Address both |
Fixes that last
A dry cloth works for a day. Fixes that survive the monsoon deal with the route.
- Sensing lines. Short, continuously sloping, supported, insulated where the supplier advises, with drains or traps if the design allows. Replace brittle tubing.
- Condensation. Plan loaded runs instead of idle test runs, drain low points as part of routine checks, and keep insulation intact. See the RECD maintenance guide for routine checks.
- Rain entry. Repair rain caps and flashing, replace soaked insulation, reseal seams and doors with materials the supplier approves for the temperature.
- Intake air. Fit or repair drip trays and louvres, keep filter housings sealed, and replace wet elements.
- Electrical. Use correctly rated glands facing downward, drip loops on cables, working cabinet heaters, anti-corrosion treatment on terminals and clean battery terminals.
- Records. Keep the weather ledger going through the next season. The same ledger can later show whether the fix worked.
Avoid permanent fixes that hide the signal. Raising an alarm limit or masking a sensor input because “it only happens in the rain” can leave a real fault unprotected on the day when the rain and a genuine blockage coincide. Our article on reducing nuisance RECD alerts without hiding faults explains how to tune responsibly.
A worked illustration: the alarm that followed the monsoon
The following is an illustrative scenario, not a real customer. A hospital annexe in a coastal city has a 320 kVA DG set with an RECD fitted a year earlier. In July the controller raises a high differential pressure alarm on four mornings. The set had not been cleaned and ΔP at a reference load was normal at the last service.
The ledger. The alarms all fall on mornings after more than 20 mm of overnight rain. The same set ran at similar loads on dry mornings in April and May without any alarm. Run hours in July are triple those of May because of outages, but the alarm days do not line up with the highest-hours days.
The first test. On the next alarm morning, a portable gauge connected in parallel reads about 40% lower than the controller. By noon both agree. The pattern points to the sensing chain.
Inspection. Four metres of tube run horizontally from the high-pressure tap with a visible sag at the midpoint, close to a cool wall. A little water is found in the sag, and the connector at the transmitter has green corrosion on one pin.
The fix. The installer re-routes the tubing with a continuous slope back to the tap, shortens it, replaces the connector, fits a drip loop and seals the gland. The team adds a plant-room humidity logger and keeps the ledger running.
The result. Through August, three heavy-rain mornings produce no alarm and the independent gauge agrees with the controller on each. The team notes that cold-start moisture in the silencer is still present, so they replace the weekly no-load test with a loaded test of an hour. The pattern was a combination: a wet sensing chain triggering the alarm, and a duty pattern that kept the exhaust damp.
Questions people ask
Is it normal for an RECD to behave differently in monsoon weather?
Some change is expected. Cooler, damper conditions affect condensation, soot character and regeneration behaviour. A fault that sets off alarms or trips is not normal and deserves tracing.
Can rain damage the filter itself?
Water reaching a hot canister is a concern because of thermal stress and corrosion. How much risk a particular design carries is a question for the supplier. Prevent water from reaching the canister in the first place.
How do I know the fault is moisture and not a failing sensor?
A failing sensor is wrong in all weathers. A moisture-affected one is wrong when it is damp and right when dry. The weather ledger and an independent gauge are the quickest way to tell. If the sensor disagrees in dry weather too, see our article on separating a sensor fault from an RECD performance problem.
Does wet fuel cause the same pattern?
It can, because rain and humidity can bring water into fuel tanks. If your faults began after a fuel delivery in wet weather, read our companion guide on RECD faults after a fuel delivery.
Before the next monsoon
Drain the low points, slope and check the sensing tubes against an independent gauge, inspect the rain cap and insulation, test cabinet heaters and glands, check the air filter, plan loaded test runs and start the weather ledger on day one.
If you would like help reading a weather ledger against your controller log, or checking the layout of sensing lines on your installation, you can browse our retrofit emission control devices, read the FAQ or get in touch with the Aceget team.
Related Reading
- Protecting an RECD Against Rainwater and Condensation
- RECD Installation in a Flood-Prone Plant Room
- Choosing an RECD for Coastal, Salt-Laden Environments
- Separating a Sensor Fault from an RECD Performance Problem
- RECD Pressure Spikes Only at Startup: What Could Explain Them?
Sources
- NHTSA-filed service bulletin: DPF differential pressure sensor susceptible to moisture intrusion (heavy-duty engine)
- NHTSA-filed service bulletin: moisture intrusion and false pressure-sensor codes, inspection of sensor tubes
- Delphi: DPF differential pressure sensor, construction, failure causes and testing
[…] time of day or weather, and the alarm returns whenever the cycle comes round. See our article on RECD faults that appear only in wet weather for the weather […]