- November 15, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
The differential pressure reading across the RECD is high, and someone asks the obvious question: do we need a regeneration or a cleaning? The two words get used interchangeably on many sites, and the confusion is expensive. Regenerate a filter that is full of ash, and nothing improves. Clean a filter that only needs a good loaded run, and you pay for downtime, handling and wear you did not need.
The difference is simple to state and harder to apply. Regeneration burns the soot, which is carbon, so that it leaves as gas. Cleaning physically removes what does not burn, mainly ash, plus other residues that regeneration cannot shift. DieselNet’s overview of diesel particulate filter regeneration describes how soot oxidation works, and every trade source agrees that ash stays behind. A trucking article drawing on lubricant specialists notes that upwards of 90% of the ash entering a filter comes from the engine oil (CCJ).
So the practical question is: how much of today’s excess pressure is soot, and how much is not? This article gives you a way to answer it with two readings and a subtraction. It then supplies a table linking common symptoms to the right service, a ladder of decision steps, a worked four-quarter example and short scenarios showing how the answer changes by duty.
Scope note. Designs differ: passive, heater-assisted, fuel-assisted and operator-initiated regeneration all exist. Whatever your device, the logic below holds, but the way you trigger regeneration depends on your model. Follow your supplier’s procedure.
What each service can and cannot do
Start with a map. The columns say whether regeneration, cleaning or neither can help with each cause of high pressure.
| Cause of high pressure | Regeneration helps? | Cleaning helps? | Notes |
|---|---|---|---|
| Soot loading (carbon) | Yes | Partly, but wasteful | Soot is exactly what regeneration is designed to burn |
| Oily or volatile soot fraction | Partly | Yes, often with a thermal step | Wet stacking leaves sticky deposits; see RECD regeneration after wet stacking |
| Ash from lubricating oil and fuel | No | Yes | Incombustible; stays until physically removed |
| Ash fused or glazed by repeated hot cycles | No | Sometimes | May resist ordinary cleaning |
| Dust and wear metals entering with intake air | No | Yes | Behaves like ash |
| Weak or ageing catalyst | Needs more heat or load | No | Cleaning may restore flow but not catalyst activity; see catalyst ageing |
| Cracked or melted element | No | No | Repair or replacement; see filter core vs complete assembly replacement |
| Blocked or faulty pressure sensor or line | No | No (clean the line, recalibrate) | See separating a sensor fault from a performance problem |
| Restriction in the exhaust pipework | No | No | Check bends, silencers and joints; see how exhaust backpressure affects a DG set |
| Reading taken at the wrong load | Irrelevant | Irrelevant | Re-read at the reference load |
Three of the rows deserve underlining. Ash cannot be burned. A cracked element cannot be cleaned into health. And a surprising share of “high pressure” complaints are really sensor, load or pipework problems, in which case neither service is the answer.
The two-reading test: finding the residual gap
You can split the excess pressure into a part regeneration removes and a part it does not, using only the pressure-at-reference-load reading you should already be logging. Our article on comparing RECD pressure readings at different generator loads explains why a reference load matters. The test needs three numbers, all read at the same load band.
- Baseline (C): pressure at the reference load when the filter was new or freshly cleaned, such as the commissioning value.
- Before (A): pressure at the reference load just before a regeneration.
- After (B): pressure at the same reference load after a regeneration that you have verified was successful.
Then:
- Excess pressure = A − C. This is how far above the clean baseline the filter has drifted.
- Soot part = A − B. This is what regeneration removed.
- Residual gap = B − C. This is what regeneration left behind.
The residual gap is the useful number. If it is small, regeneration is doing its job and the filter is not accumulating a permanent load. If it is large, or if it grows from cycle to cycle, something that regeneration cannot remove is building up.
What counts as a “verified successful” regeneration?
The test is only meaningful if the regeneration really worked. Before trusting reading B, check that the cycle reached its normal temperature, the heater or heat source behaved as usual, nothing interrupted the cycle and the load was comparable. If the cycle was partial, a high B reading tells you about the cycle, not the filter. For a failed or incomplete cycle, start with RECD regeneration not completing, and keep in mind the cautions in what to record when testing during regeneration.
A worked example: one cycle
An illustrative 400 kVA DG set has a commissioning baseline of 30 mbar at 50% load. The maintenance team reads these values at 50% load.
| Reading | Value (mbar) |
|---|---|
| Baseline C | 30 |
| Before regeneration A | 62 |
| After a verified regeneration B | 41 |
- Excess pressure = 62 − 30 = 32 mbar
- Soot part = 62 − 41 = 21 mbar, so regeneration removed 21 ÷ 32 = about 66% of the excess
- Residual gap = 41 − 30 = 11 mbar, about 34% of the excess, and about 37% above the baseline (11 ÷ 30)
Reading it: regeneration did most of the work, but about a third of the excess pressure is still there after a good cycle. One cycle alone does not prove an ash problem, because soot may partially bake and a different load band might shift the numbers. It does tell you to watch the residual gap.
A worked example: four quarters
The same set is measured at each quarterly regeneration, always at 50% load, with baseline 30 mbar.
| Quarter | Before (A) | After (B) | Soot part (A − B) | Residual gap (B − 30) | Residual as % of baseline |
|---|---|---|---|---|---|
| Q1 | 58 | 36 | 22 | 6 | 20% |
| Q2 | 60 | 39 | 21 | 9 | 30% |
| Q3 | 63 | 43 | 20 | 13 | 43% |
| Q4 | 66 | 48 | 18 | 18 | 60% |
The soot part barely changes, between 22 and 18 mbar, so the engine and duty are producing about the same amount of soot per quarter and regeneration is removing it. But the residual gap has tripled from 6 to 18 mbar. That is the signature of material that regeneration is not removing, most likely ash and possibly some residue. A rising residual gap with a steady soot part is the clearest signal that cleaning is due, even though each regeneration still “completes”.
The team might agree a trigger with the supplier, for example “schedule cleaning when the residual gap exceeds 50% of the baseline at the reference load”. In this example the trigger is crossed in Q4. The percentage is an illustration, not a standard, and your supplier may specify different limits or a different measure.
Cross-check with the oil
Ash comes mainly from engine oil, so oil consumption should tell a consistent story. Compare the residual gap trend with oil top-ups over the same period. If the gap rose faster than usual, check whether oil consumption rose too. See engine oil consumption and RECD suitability. If the engine is burning more oil, cleaning treats the symptom, and the engine team needs to treat the cause.
Symptom lookup: what you see, and the likely service
The two-reading test is the main tool. This lookup covers situations where you cannot yet run it.
| What you see | Most likely explanation | First service to consider |
|---|---|---|
| Pressure high before a cycle, falls to near baseline after a good cycle | Soot loading, regeneration working | Regeneration, or a change to duty so passive regeneration works |
| Pressure falls after a cycle but stays well above baseline, and the gap grows over months | Ash building up | Cleaning |
| Pressure rises again within hours or days of a good cycle | Oily soot, wet stacking, engine producing excess soot or sensor issue | Engine and duty review first; cleaning only after the cause is fixed |
| Cycles start but never complete | Heat, power or control problem | Regeneration diagnosis; see RECD regeneration not completing |
| Cycles are requested more and more often | Ash, catalyst ageing, duty change or engine soot | Compare residual gap and engine data before choosing |
| Pressure suddenly low for the load | Possible crack, bypass or seal failure | Inspection by the supplier, neither service |
| Pressure reads high with the set at rest or at no flow | Sensor offset or blocked line | Sensor check |
| Pressure high only at high load | Exhaust restriction or undersized path | Pipework review |
| Pressure normal but regeneration behaviour has worsened | Catalyst ageing or heater problem | Supplier diagnostic |
Notice how many rows lead away from both services. That is the point of listing them.
The decision ladder
Work through these steps in order. Stop when you reach an answer.
Step 0: Is the reading real? Check that the sensor and tapping lines are clear, that the reading was taken at the reference load after the set had settled for about half an hour and that the instrument is calibrated. If the reading cannot be trusted, fix that first.
Step 1: Is it a one-off or a trend? Look at the last several readings at the reference load. A single high reading after a long light-load run is different from a steady upward drift.
Step 2: Has regeneration been tried and verified? If not, and the filter is within the supplier’s limits, a supervised loaded run or regeneration may be the first step. Do not repeat failed attempts blindly; our article on repeated attempts and damage risk is can repeated failed regeneration damage an RECD filter.
Step 3: Compute the residual gap. Use the two-reading test. Compare it with the trend and with the oil record.
Step 4: Decide.
| Residual gap behaviour | Decision |
|---|---|
| Small and stable | Regeneration is adequate; keep monitoring |
| Growing steadily | Plan cleaning before the supplier’s limit; schedule it |
| Large and sudden | Investigate for oil contamination, fused deposits, damage or a sensor change before cleaning |
| Not measurable (no verified regeneration possible) | Inspect the element and consult the supplier |
Step 5: After any service, re-baseline. Record pressure at the reference load, and compare it with the commissioning figure. See why pressure can stay high after a cleaning service for what to do if it does not return close to baseline.
Why choosing wrongly costs more than it seems
The cost of cleaning when regeneration was enough
Cleaning means isolating the set, removing the element, handling a fragile ceramic part, possible transport, waste handling and refitting. Each handling event carries a small risk of damage, and each thermal cycle during a cleaning adds a little wear. Frequent unnecessary cleaning is also an expensive habit, as you can see in a service-cost log built following tracking RECD service cost per operating hour.
The cost of regenerating when cleaning was needed
Repeated regeneration of an ash-loaded filter wastes heater energy or fuel, extends the time the set is carrying extra load and may stress the element. In some designs it also pushes the controller into failed cycles because the filter cannot reach its recovery target, which may then be reported as a regeneration fault when the real problem is ash. Our article on soot versus ash in RECD regeneration explains why regeneration cannot fix ash.
The cost of delay
Ash loading is gradual. Leaving it until pressure alarms sound means the engine works against higher back-pressure, and the filter’s soot storage capacity has been reduced by the ash, so cycles become more frequent. A planned cleaning is cheaper than an unplanned one.
How the test changes with the regeneration method
The two-reading test works for every design, but the way you obtain reading B differs. Follow your supplier’s procedure for your own model, and treat the descriptions below as a general guide.
| Design | How “after” is obtained | Points to watch |
|---|---|---|
| Passive (catalyst-assisted) | There is no button. Reading B comes from a supervised loaded run at a load the supplier approves, long enough for exhaust temperature to reach its working range | Confirm temperature actually reached the range; a light-load run proves nothing. Planning such runs is covered in choosing an RECD for a generator running at low load |
| Electric-heater-assisted | The controller starts a cycle, or an operator requests one | Check heater current, bed temperature and any interruption before trusting the result; see RECD regeneration not completing |
| Fuel-assisted | The controller doses fuel or starts a burner | Verify that ignition and flame confirmation occurred, and follow the supplier’s safety rules |
| Operator-initiated at load | The operator raises load for a defined period | Keep the cycle protected from stops and load drops; see RECD regeneration while supplying building loads |
In every case, wait until the set has settled at the reference load after the cycle before reading B. A reading taken while the filter is still hot and the flow is changing is not comparable with the baseline.
What if you cannot trigger a regeneration at all?
Some sites cannot run the set at the load the supplier approves, perhaps because the building load is small or the set is committed to supplying a critical load. In that case you cannot obtain a clean reading B, and the honest conclusion is “unknown”. Do not guess. Options include arranging a load-bank run, using a planned outage, asking the supplier for an assessment on site or inspecting the element. A weight and flow test after removal gives the missing evidence directly.
A one-page log for the two-reading test
Keep the numbers in one place so that trends are visible. One row per verified regeneration is enough.
| Date | Run hours | Load band | Baseline C | Before A | After B | Soot part A − B | Residual B − C | Residual % of baseline | Cycle verified? | Notes |
|---|---|---|---|---|---|---|---|---|---|---|
| Yes / No |
Three habits make the log trustworthy. First, fill the “cycle verified?” column honestly, and leave unverified cycles out of the trend. Second, always record the load band, because a reading with no load is a reading with no meaning. Third, review the last four rows every quarter, looking at the direction of the residual column more than at any single value. If the residual is growing and the soot part is steady, you are looking at the pattern that points to cleaning. If both are growing together, the engine may be producing more soot, which is a different conversation. Our article on what to record in an RECD event log lists the surrounding events worth noting, such as repairs, fuel changes and long idle spells.
Three short scenarios
These are illustrative, not real sites.
Scenario A: a standby set after a long monsoon
A 250 kVA standby set has run for many short, light runs. Pressure at the reference load is 52 mbar against a baseline of 33. A supervised loaded run of an hour brings it to 38 mbar and it holds at 37 to 38 over the next two runs. The residual gap is about 5 mbar, or 15% of baseline. Decision: regeneration or a loaded run was enough. Change the test routine to include loaded runs. See choosing an RECD for a generator running at low load.
Scenario B: a prime-power set with thousands of hours
A 500 kVA set in a mine runs for thousands of hours a year. Quarterly regeneration reads 61 before and 49 after, then 62 and 52, then 64 and 55, against a baseline of 34. The residual gap grows from 15 to 18 to 21 mbar. Decision: the soot part is steady and the residual gap keeps growing, so book a cleaning, preferably with an exchange element to protect production, and review oil consumption.
Scenario C: pressure high after a “successful” cycle that was actually partial
A telecom site logs pressure of 60 mbar before and 54 mbar after a cycle that the controller called complete. The cycle log shows the heater current was low and the peak temperature was well below normal, so the cycle was partial. Decision: neither service yet. The cycle is not a valid “after” reading. Investigate the heater supply and light-load behaviour first, then repeat the test.
What to do on the day of the decision
- Write down the three numbers (C, A, B), with load, date and run hours.
- Compute the soot part and the residual gap.
- Record the oil consumption since the last cleaning.
- Decide using the ladder, and write the decision and its reason in the maintenance log.
- If cleaning is chosen, use the questions in RECD filter cleaning: on-site vs off-site service compared, and take the “before” readings before the element leaves.
- If regeneration is chosen, follow the supplier’s procedure, protect the cycle from interruption and verify the result.
- After the service, read pressure at the reference load for the next few runs and compare it with the plan.
Mistakes to avoid
- Using pressure readings at different loads. Flow changes pressure, so a reading at 30% load cannot be compared with one at 60%.
- Trusting the controller’s “complete” message as proof that regeneration worked. Verify with values.
- Cleaning on a calendar alone when condition data say it is not needed, or ignoring the calendar when condition data say it is overdue. See RECD cleaning intervals: calendar time vs operating hours vs condition.
- Expecting cleaning to fix regeneration behaviour. Cleaning restores flow. It does not repair a weak heater or restore a poisoned catalyst.
- Skipping the baseline. Without it you cannot compute the residual gap.
- Treating one reading as a trend.
FAQ
Does regeneration remove ash?
No. Regeneration oxidises soot, which is carbon. Ash is the incombustible residue of oil additives, wear metals and dust, and it stays in the filter until it is physically removed. That is why cleaning is a separate service.
Can I just keep regenerating until pressure falls?
Not safely. If the pressure has not fallen after a verified cycle, more cycles will not remove the ash and may add thermal stress and wasted energy. Stop, compute the residual gap and decide.
How do I know my regeneration was successful enough to use as the “after” reading?
Check that the temperature reached its normal range, the heat source behaved normally, no interruption occurred, the load was comparable and the pressure fell. If any of these fail, do not use the reading.
What if I have no commissioning baseline?
Use the lowest pressure you have recorded at a reference load after a verified regeneration or a recent cleaning, and ask the supplier for the nominal clean figure for your element. Start logging now so that you have a baseline for the future.
How big should the residual gap be before cleaning?
There is no universal number. Agree a trigger with the supplier, based on your element, your pressure limits and your duty, and write it into the maintenance plan. The 50% of baseline used in this article is an example.
Is cleaning needed if the filter never shows a pressure alarm?
Possibly. Ash capacity and pressure limits differ, and a filter can be losing capacity long before an alarm. Trend the residual gap and check the supplier’s interval.
Bringing it together
Regeneration and cleaning address different materials, so the right choice depends on how much of today’s excess pressure is soot and how much is not. Take the three readings, compute the soot part and the residual gap, check the cycle was valid, cross-check against oil use and follow the decision ladder. If the gap is small, regenerate or adjust duty. If it is growing, plan a cleaning. If neither fits, look for a sensor, pipework or element problem.
If you would like help reading your own pressure and regeneration data, explore our retrofit emission control devices, read our DG set emission regulations overview or contact the Aceget team.
[…] Pressure at the reference load stays well above baseline after verified regenerations, and inspection suggests residue as well as ash. See RECD regeneration vs filter cleaning: which service is needed. […]