- November 19, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
The first cleaning felt like a fix. Back-pressure dropped, the alarm cleared and the DG set went back to work. Then, a few months later, the same alarm came back. A second cleaning followed, and the gap before the next alarm was shorter. By the third visit someone in the maintenance team asked the question that this article exists to answer: why does the RECD keep clogging?
The uncomfortable answer is that a filter which keeps clogging is almost never the root problem. A retrofit emission control device (RECD) is a collector. It fills up when more material arrives than the device can remove, and it fills up faster when something upstream changes what arrives. Cleaning resets the level, but it does not touch whatever is feeding it. If the feed is unchanged, the filter will refill on schedule.
This guide treats a recurring blockage as a case file. We set out the balance that governs how fast a filter clogs, introduce a simple number called the rebuild rate that shows whether your problem is getting worse, line up seven suspects with the evidence that points to each, and finish with the order in which to investigate so that you spend the least money and time. The approach is general. Your device’s manual and your supplier’s limits always take priority over any figure here.
The clogging balance: why a filter fills at all
Before the suspects, a mental picture. Think of the filter as a bath with the tap on and a plug that works only some of the time.
- The tap is everything that arrives in the filter: soot from combustion, oily residue, unburnt fuel droplets, ash from lubricating oil and fuel additives, and occasionally debris from upstream.
- The plug is everything that removes material: passive regeneration (soot burning at ordinary exhaust temperature), assisted regeneration (heater, burner or loaded run) and cleaning.
- The water level is differential pressure (ΔP) across the filter, read at a comparable load.
Two facts matter. First, soot can be burned away by regeneration, but ash cannot. Ash only leaves when the filter is cleaned, so it accumulates for the life of the device, slowly shrinking the usable filter volume. Second, the plug does not work at all temperatures. Soot burns when the exhaust is hot enough for long enough, and a DG set that runs lightly loaded may never reach that point. Our guide to soot versus ash in RECD regeneration explains the difference in detail.
So a filter can clog repeatedly for three structural reasons, and sometimes for a fourth that is not real at all.
- The tap is too strong. The engine is making more soot, oil residue or ash than the filter was sized for.
- The plug is not working. Regeneration is failing to happen or to finish, because temperature, time, power or controls are short.
- The bath is too small. The device is undersized for the exhaust flow or the duty, so even a normal feed fills it quickly.
- The gauge is wrong. The filter is not actually as full as the reading claims.
Every one of the seven suspects below belongs to one of these four groups. Keep that grouping in mind, because it tells you which tests to run first.
The rebuild rate: one number that tells you where you stand
Most sites track whether a filter is blocked. Few track how long it takes to become blocked again. That interval is the single most useful number in this whole investigation.
Rebuild rate here means the run hours between the end of one cleaning (or the last verified clean state) and the point at which ΔP at a reference load again reaches the alarm or action level. Use run hours, not calendar days, because a standby set might sit idle for weeks.
Record it after every cleaning. A pattern appears quickly. The table below is an invented example for a 320 kVA set, to show how to read the numbers; your own figures will differ.
| Cleaning episode | Run hours until ΔP at 50% load reached action level | Change from previous |
|---|---|---|
| After install (baseline) | 640 | none |
| After 1st cleaning | 410 | down 36% |
| After 2nd cleaning | 260 | down 37% |
| After 3rd cleaning | 150 | down 42% |
| After 4th cleaning | 90 | down 40% |
What does a shrinking interval like this tell you? Several things, and the shape of the decline carries information.
- A steady fall of 35 to 40% each time suggests an engine-side feed that is worsening, or a residue that is not being removed by cleaning and is building up in the filter. Both are cumulative.
- A sudden fall after a particular date points to an event: an engine repair, an oil change to a different grade, a fuel supplier change, a new duty pattern.
- A flat interval that is simply short from the beginning points to sizing or duty. The filter never had the margin it needed.
- An interval that varies wildly often means the gauge, the load at reading or the cleaning quality is inconsistent, so the numbers cannot be compared.
If a falling rebuild rate is your pattern, stop booking cleanings until you have investigated. Each cleaning costs money and some risk to the filter, and it will not change the trend. Our article on the RECD maintenance guide lists the routine checks that make the investigation easier.
Suspect 1: The engine is making too much soot
The motive. A healthy diesel engine produces a modest amount of soot. An engine with worn or leaking injectors, a restricted air filter, a lazy turbocharger, over-fuelling or poor fuel will produce much more, and the filter dutifully collects it. Before filters existed, that engine would simply have smoked, and someone would have noticed. Now the filter hides the symptom until the pressure rises.
A fleet-maintenance expert quoted by Vehicle Service Pros makes the same point for trucks: engine condition is the most common reason for a filter plugging, and a restricted air filter changes the fuel-air balance and raises soot output. The advice there is to run an opacity test at each cleaning interval, so that the engine is checked before it feeds the freshly cleaned filter.
The evidence.
- Opacity or smoke readings upstream of the filter (or on a bypass test) higher than the engine’s earlier records.
- Black or grey smoke at the stack during load steps. See why a generator emits black smoke.
- A dirty air filter element, a boost pressure lower than the service manual expects, or an exhaust temperature higher than normal at the same load (a sign of over-fuelling).
- Fuel consumption per kWh creeping up.
The alibi. If the engine’s opacity is within the manufacturer’s limit at a loaded steady state, and the air filter and boost are healthy, this suspect is probably not the main cause. It may still be a contributor, so note it and move on.
The cheap test. Inspect the air filter and intake path, read boost at a stable load and arrange an opacity or smoke-number check before the filter is refitted after the next cleaning. A supplier or engine specialist can run it in under an hour.
Suspect 2: The exhaust never gets hot enough
The motive. Passive regeneration needs temperature. If the set spends most of its life at light load, the exhaust never reaches the range where soot burns, and the filter becomes a one-way store. The SAE paper on DPF regeneration systems for generators notes that generator engines are often run for long periods at light load with low exhaust temperature, which makes regeneration difficult. A US generator dealer’s maintenance note for standby sets gives a similar rule of thumb: keep the set above roughly 30 to 50% load and load-bank it periodically so that accumulated soot burns off. Those figures come from specific products and regions, so treat them as a pointer and ask your supplier for the number that applies to your device.
The evidence.
- A load history showing a high share of run hours below 20 or 30% of rating. Our guide to using DG load logs to assess RECD suitability shows how to build that table.
- Exhaust temperature at the filter inlet rarely or never reaching the supplier’s regeneration range.
- ΔP at a reference load rising steadily between loaded runs and flattening or falling after them.
- Weekly “test runs” with no load transfer.
The alibi. A set that regularly runs for hours above the supplier’s minimum load, with inlet temperature in range, is not starved of heat.
The cheap test. Plot one month of load against time. If most hours sit in the bottom band, you have your answer. See also choosing an RECD for a generator running at low load.
Suspect 3: Wet, oily soot is packing the filter
The motive. Light-load running does two things at once. It stops regeneration, and it makes the soot wetter. An engine that is run cool passes unburnt fuel and some lubricating oil into the exhaust, producing a sticky mixture that packs densely into the filter walls. This material raises ΔP faster per gram than dry soot, and it does not burn cleanly. This is the wet-stacking pattern, covered in depth in our article on RECD regeneration after wet stacking.
The evidence.
- Black, oily residue at the exhaust outlet, drains or silencer.
- A filter that comes back from cleaning with oily, dark deposits rather than dry grey soot.
- Strong smell and smoke during regeneration, with a slow ΔP recovery.
- Engine oil analysis showing fuel dilution.
The alibi. Dry, powdery soot and a clean outlet suggest the problem is quantity, not wetness.
The cheap test. Look and smell. Inspect drain points and the outlet at the next shutdown, and ask whoever cleans the filter to describe the deposit. Photographs help.
Suspect 4: Ash is quietly stealing capacity
The motive. Ash comes from the metallic additives in lubricating oil and from some fuel contaminants. It does not burn. Every litre of oil the engine consumes leaves a little ash behind, and over months the ash layer thickens at the closed ends of the filter channels, reducing the volume available for soot. The filter then fills faster, and each regeneration clears only the soot, not the ash.
The evidence.
- ΔP right after a good regeneration (or cleaning) creeping higher each time, even when the cycle itself looks normal.
- High engine oil consumption, or oil top-ups at short intervals. See engine oil consumption and RECD suitability.
- A rebuild rate that falls steadily even though the engine and duty have not changed.
- Cleaning reports (if any) describing heavy ash.
The alibi. A stable post-regeneration baseline over many cycles argues against significant ash build-up.
The cheap test. Compare ΔP at a reference load immediately after each regeneration across the last six months. If the baseline floor is rising, ash or residual deposit is the likely reason. Ask the supplier about an ash load estimate at the next service. Our article on handling soot and ash removed during RECD servicing explains what the cleaner should record.
Suspect 5: The regeneration hardware is not doing its job
The motive. If the RECD relies on a heater, burner or controller-driven cycle, any weakness in that chain turns the plug partly off. A heater circuit with a loose terminal, a breaker that tripped, a supply that dips on load, a sensor that reports the wrong temperature or an inhibit signal that blocks the cycle will each leave soot behind.
The evidence.
- Cycles that end by timer with low peak temperature. See how to interpret an RECD regeneration completion message.
- Heater current well below the commissioning value, or zero.
- Regeneration requests that never start. See RECD regeneration inhibit signals.
- Rising cycle length. See what a rising regeneration duration tells you.
The alibi. A log of regular, verified cycles with normal heater behaviour and ΔP recovery clears this suspect.
The cheap test. Read the last ten cycles. Check heater current and supply voltage during one. Verify that the controller’s last firmware and settings match the commissioning record.
Suspect 6: The filter is too small for the job
The motive. A filter is chosen for an exhaust flow and a duty. If the engine’s real flow is higher than the sizing assumed (a turbocharged engine with a higher rating than the nameplate suggested, an engine that was re-tuned, a set that runs at higher altitude or with heavy loads), the soot arrives faster per unit of filter volume. The same filter that would be comfortable on one set fills quickly on another. Our guides to engine displacement and RECD selection and same-kVA DG sets needing different RECDs explain why kVA alone is not a safe basis.
The evidence.
- A short rebuild rate from the very first interval, with no worsening trend.
- ΔP at rated load much closer to the alarm level than the commissioning sheet suggested.
- Mismatch between the engine data the supplier used and the engine’s actual configuration, for instance a changed turbocharger or a rating uprate.
The alibi. A device specified from verified engine data, with a healthy first interval and a steady rebuild rate, is probably correctly sized.
The cheap test. Compare the engine model, rating, displacement and exhaust flow in your purchase documents against the engine’s current state. If you cannot find the original data, our guide on specifying an RECD from OEM exhaust flow data shows what to request.
Suspect 7: The reading is wrong
The motive. Sometimes the filter is not blocked. The gauge says it is. Pressure sensor lines can fill with condensate or soot, tubes can kink or leak, a transmitter can drift, a switch can be set too low, and a reading taken at full load can be compared with a limit set for half load. Any of these will produce a “clogged” alarm on a filter that is partly clean.
The evidence.
- ΔP that jumps or sticks with no change in load.
- Readings that disagree with a second gauge or a manual manometer.
- An alarm that appears at a different ΔP than the supplier’s limit at that load.
- A filter that comes back from cleaning with little deposit.
The alibi. Two independent readings that agree, taken at a comparable load, clear this suspect.
The cheap test. Blow through or inspect the sensing lines, check for condensate, compare against a calibrated portable gauge at a stable load and apply the load-matching rule from our guide on comparing RECD pressure readings at different generator loads. The article on separating a sensor fault from an RECD performance problem offers a fuller routine, and pressure sensors versus pressure switches explains why the type of instrument matters.
The investigation order: cheapest and most likely first
Do not test all seven at once. A sequence that works for most sites is below. It starts with the tests that cost almost nothing and can eliminate whole groups.
| Step | Question | Method | Cost and time | Groups it can clear |
|---|---|---|---|---|
| 1 | Is the reading real? | Compare gauge to a second reading at the same load; inspect sensing lines | Under an hour | The gauge |
| 2 | What is the rebuild-rate pattern? | Tabulate run hours between cleanings | An hour at a desk | Points to tap, plug or bath |
| 3 | What does the duty look like? | Load-band table for the last 3 to 6 months | A few hours | The plug (temperature) |
| 4 | What is the deposit? | Photographs and description from the cleaner; inspect outlet and drains | At next cleaning | Wet soot versus dry soot versus ash |
| 5 | Is the engine feeding too much? | Air filter, boost, opacity test, oil analysis | Half a day | The tap |
| 6 | Is the regeneration hardware healthy? | Cycle log, heater current, supply, settings | Half a day | The plug (hardware) |
| 7 | Is the device the right size? | Compare engine data to the original specification | Desk work plus supplier call | The bath |
At every step write down the result, even a clean one. A record of what has been ruled out is as valuable as a record of what was found.
Two lookalikes that waste money
“It must be a bad batch of filters.” Possible, but rare. A defective substrate usually shows as cracking, a sudden drop in ΔP or a failed efficiency test, not as a steady rebuild-rate decline.
“A fuel additive will stop it.” Additives are sold on the promise of lowering soot or the ignition temperature of soot, but they do not repair an injector, shorten a light-load duty or remove ash. If the underlying cause is mechanical or operational, an additive is a distraction. Our article on whether diesel fuel additives can replace an RECD explains the limits. Do not add anything to the fuel or oil that your RECD supplier and engine maker have not approved.
What to do with the answer
Once the investigation points to a cause, the fix usually falls into one of three kinds.
Fix the feed. Repair or tune the engine, change the air filter, correct the oil consumption, change the fuel supplier. This is the cheapest and most permanent fix when the tap is the problem. Coordinate with the RECD supplier so that the engine work and the filter service are sequenced sensibly; our guide to coordinating DG engine repairs with the RECD service team covers the order.
Fix the plug. Change how the set is run (loaded test runs instead of idling, load-bank sessions, concentrating load on fewer sets), repair the heater circuit, correct inhibit settings. See parked versus in-service RECD regeneration for the options a regeneration plan can use.
Fix the bath. If the device is truly undersized, talk to the supplier about a larger or different design, or a filter core change. Our article on filter core versus complete assembly replacement sets out what to ask.
After any fix, reset the experiment. Record the new baseline, restart the rebuild-rate count and compare the next interval with the previous ones. A fix has worked only if the interval lengthens.
A worked illustration: reading the case file
The following is an illustrative scenario, not a real customer. A hospital’s 250 kVA DG set has had three cleanings in fourteen months. The intervals were 520, 310 and 180 run hours.
Step 1. A portable gauge at 50% load reads within 3 mbar of the controller. The reading is real.
Step 2. The rebuild rate has fallen about 40% each time, so the problem is cumulative.
Step 3. The load history shows 70% of run hours below 20% load, because weekly tests run without a load transfer. This explains why passive regeneration rarely happens, but it does not explain why the interval is shrinking each time.
Step 4. The cleaner’s photographs show oily black deposit at the inlet face and a visible ash layer at the channel ends.
Step 5. The oil analysis reveals fuel dilution, and the air filter is heavily clogged with dust after a construction project beside the plant room. Boost is slightly low.
Step 6. Heater current is normal and cycles complete, but many are partial because of the low load.
Diagnosis. Three causes are stacking. Light-load running (suspect 2) leaves the soot wet (suspect 3). A neglected air filter has raised soot output (suspect 1). Ash is accumulating more quickly than expected because the wet, fuel-diluted oil is being consumed faster (suspect 4). The filter is not defective.
Fix. Replace the air filter and investigate the boost, change the oil, move weekly tests to a building load of at least 40% for 45 minutes, add a loaded run after each outage and re-baseline. The next interval, tracked over the following quarter, is the proof. The point of the example is that no single test found the problem. The case file did.
Prevention: a quarterly routine that keeps the file short
A short routine every quarter prevents most recurring blockages.
- Record ΔP at the reference load and compare it with the baseline.
- Update the rebuild-rate table, including run hours since the last clean.
- Review the load-band table and count loaded runs.
- Check the air filter and intake path, and note boost at a steady load.
- Read the last cycles: peak temperature, heater current, interruptions.
- Look at the outlet and drains for oil.
- Check engine oil consumption and the last oil analysis.
- Sign and file the page with the asset record. Our guide to the RECD event log suggests a format.
Quick answers
How often should an RECD need cleaning?
There is no universal interval. It depends on the device, engine condition, duty and oil consumption. The useful measure is your own rebuild rate, compared with the interval your supplier expected at commissioning. A sudden or steady shortening is the warning sign.
Can a clogged RECD damage the engine?
A heavily loaded filter raises exhaust back-pressure, and sustained high back-pressure can reduce power and raise exhaust temperature. Our guide to how exhaust back-pressure affects a DG set explains the effects.
Is repeated cleaning bad for the filter?
Handled properly, cleaning is a normal service step, but each handling and cleaning event carries some risk, and the cleaning method matters. Ask the supplier which methods suit your filter, and avoid improvised methods.
Should I replace the filter after repeated clogging?
Not until you know why it clogs. A new filter on the same engine and duty will repeat the pattern. Replace when inspection shows damage or a loss of catalyst activity, not simply because it is full. See catalyst poisoning in an RECD.
Does regeneration remove ash?
No. Regeneration burns soot. Ash stays until the filter is cleaned.
When to call the supplier
Call the supplier, with your rebuild-rate table and load-band data in hand, if the interval has fallen by more than a third twice running, if ΔP after a regeneration does not return near baseline, if the heater or controller behaves oddly, or if you find oil, coolant or fuel in the exhaust system. For a broader view of what to record before the call, see our guide to the alarm response matrix for operators.
Bringing it together
An RECD that keeps clogging is telling you that something is feeding it faster than it can clear, that its clearing mechanism is not working, that it is too small, or that its gauge is wrong. Track the rebuild rate, work through the seven suspects in order of cost, record every result and fix the cause before booking another cleaning.
If you would like a second pair of eyes on your cleaning history and load data, you can explore our retrofit emission control devices, read the FAQ or contact the Aceget team with your ΔP and run-hour records.
Related Reading
- RECD Regeneration Not Completing: Causes and Checks
- Soot vs Ash in RECD Regeneration
- RECD Regeneration After Wet Stacking
- How to Use DG Load Logs to Assess RECD Suitability
- Engine Oil Consumption and RECD Suitability
[…] See why an RECD keeps clogging […]
[…] that deserves a test, not a celebration. Record the floor in your asset file, and see our guide to why an RECD keeps clogging for the rebuild rate and floor […]