Why Does One of Two Similar RECD Installations Clog Faster?

Two sets stand side by side in the plant room. Same make, same kVA rating, same year, same exhaust layout and the same RECD model fitted by the same crew. After eight months, one RECD is on its second cleaning and the other has not needed one. The facility manager’s first thought is a bad filter. The second is a bad batch. Both are tempting and both are usually wrong.

When two apparently identical installations behave differently, that is not a mystery to fear. It is an opportunity. Most fault-finding is hard because you have no comparison: you cannot tell whether a number is normal for this set. With a twin you do have one. The healthy set is a free control, and every difference between the two is a candidate explanation. The art is to list the differences honestly, rank them and test them without changing too many things at once.

This guide gives you that method. We start by confirming that the difference is real, then introduce a twin sheet that lists every place two installations can differ, show how to rank the differences by plausibility, describe the five reasons that explain most cases and set out swap and split tests that tell you which difference matters. The advice is general. Your supplier’s data and the engine maker’s manual come first.

Why “identical” is a hypothesis, not a fact

Two sets are never truly identical. Even if they leave the factory a week apart with consecutive serial numbers, they diverge as soon as they are installed. One is nearer the wall, one sits under a vent. One is lead and starts first on every outage. One has had a new injector set, the other has not. One was topped up from a different fuel tank. One has a longer sensing tube. One has had its air filter changed more recently.

Calling them “identical” encourages the lazy conclusion that the difference in behaviour must come from the RECD, the one component that was bought as a pair. A better habit is to treat “identical” as a claim to be tested. For each element, ask whether the two are really the same, and how you know.

The test is easiest to see in the other direction. If two RECDs on matching engines clog at very different rates, and you swapped their filter cores and the pattern stayed with the set and not with the core, then the difference is in the engine, the duty, the fuel or the instruments, not in the filter. That one observation removes a whole family of suspects.

Step one: confirm that the difference is real

Before you list differences, check that the two sets are being measured the same way. Many twin mysteries disappear at this step.

Compare like with like. ΔP depends strongly on flow, so compare readings at the same load band. Our guide on comparing RECD pressure readings at different generator loads explains how.

Compare by run hours, not calendar time. A set that runs 400 hours in eight months and one that runs 90 will naturally differ. Compute a rebuild rate: run hours between a verified clean state and the point where ΔP at a reference load reaches an action level. Two sets with the same rebuild rate in hours are behaving the same, even if one needs attention sooner on the calendar. Our article on why an RECD keeps clogging explains the rebuild rate in detail.

Compare the gauges. Put one calibrated portable gauge on both sets, at the same taps’ positions, at the same load. If the controller on one set reads 20% higher than the gauge, you may be chasing an instrument fault. See separating a sensor fault from an RECD performance problem.

Compare the starting point. Was each filter in the same state after its last cleaning? Cleaning quality varies, and a filter returned with residual deposit starts the next interval at a disadvantage.

Compare the alarm thresholds and settings. Two controllers can have different limits, delays or learned values, so one set may “need cleaning” earlier on paper. Export both settings and compare.

If, after these checks, one set still loads at a clearly higher rate, the difference is real and the hunt begins.

The twin sheet

The twin sheet is a table with two columns of facts, one per set, and a third for notes. Fill it in from records and from a walk-round, and mark each row as same, different or unknown. The unknowns are as important as the differences, because an unknown is a place where evidence is missing.

GroupItemSet ASet BSame / different / unknown
EngineSerial number and year of build
EngineRun hours
EngineInjector, turbocharger or major repairs and dates
EngineBoost pressure at 75% load
EngineSmoke or opacity at a loaded steady state
LubricantOil brand, grade and sulphated-ash level
LubricantOil consumption (litres per 100 run hours)
LubricantLast oil analysis: fuel dilution, wear metals
AirAir filter condition and change date
FuelTank, supplier and delivery dates
DutyShare of run hours in each load band
DutyNumber of starts per month and average run length
DutyRole: lead, lag or standby; rotation schedule
DutyLoad transfers and step sizes
EnvironmentAmbient temperature range, dust, humidity, ventilation
Exhaust pathLength, bends, silencer, insulation
Exhaust pathTotal back-pressure at rated load
RECDSerial number, build batch, age
RECDCleaning history and ash report
RECDRegeneration count, duration and completion rate
ControlsSensor type, calibration date, tube length and routing
ControlsController firmware and settings
MaintenanceWho services it and how often
ChangesModifications, relocations, repairs since installation

Spend an hour on this. It is common to find, within the first twenty rows, two or three differences nobody knew about.

Ranking differences: size, direction, timing and dose

A long list of differences is not an answer. You need to rank them. Four questions do most of the work for each difference.

  1. Size. Is the difference large enough to matter? A 2% difference in run hours is noise. A fourfold difference in time spent below 20% load is not.
  2. Direction. Does the difference point the right way? If set B clogs faster and set B runs lightly loaded, light-load running is a candidate. If set B runs at higher load, it is a poor one.
  3. Timing. Did the divergence begin when the difference began? A filter that clogged faster from the first interval cannot be explained by a repair done last month.
  4. Dose. If the difference is the cause, a larger difference should give a larger effect. Does the gap in clogging rate scale with the gap in oil consumption, for example, across the two sets and across time?

Score each difference roughly: strong, possible or unlikely. Do not pretend to precision. The aim is to decide the order of testing, with the most plausible and cheapest tests first.

DifferenceSizeDirectionTimingDoseVerdict
Set B has 30% of run hours below 20% load, A has 8%LargeRight waySince installationSeems to scaleStrong
Set B has a different fuel tankModerateUnknownAlwaysUnknownPossible
Set B has a 1.5 m longer sensing tubeSmallCould affect readingAlwaysN/APossible (instrument)
Set B was painted a different colourNoneNoneNoneNoneUnlikely

The five reasons that explain most twin differences

Research and field experience point to a handful of causes that recur. Knowing them keeps the investigation efficient.

1. Duty: who runs, at what load, for how long

The most common reason, by a wide margin, is that the two sets do not do the same job. The lead set starts on every outage and runs the lighter loads. The lag set only runs when demand is high. Or the reverse. A set that spends its life at light load has cooler exhaust, wetter soot and little passive regeneration. A set that runs hot and loaded has the opposite. See lead-lag rotation and RECD operating conditions, RECDs with synchronised generators and load sharing and using DG load logs to assess RECD suitability.

2. Oil consumption and oil quality

Ash comes from lubricant additives, and a set that burns more oil loads its filter with more ash. A SAE study of filters returned from the field found that lubricating-oil consumption rate and the oil’s sulphated-ash, phosphorus and sulphur levels were the main drivers of ash accumulation (SAE 2016-01-0928). Two engines of the same model can consume very different amounts of oil depending on wear, rings, valve seals or turbocharger condition. See engine oil consumption and RECD suitability and soot versus ash.

3. Engine condition and combustion

Injectors, boost, air filter and fuel all change soot output. A fleet article on root-cause work for a repeatedly failing aftertreatment unit recounts how a filter was replaced and the problem continued, until the engine-side cause was found (Fleet Maintenance). Compare boost, smoke and fuel consumption per kWh across the pair.

4. Temperature and regeneration behaviour

Not all regeneration is equal. A university-led study of ash in particulate filters reports that exhaust temperature and regeneration conditions have a profound effect on how ash packs and where it settles, while filter design (porosity, pore size, material) has a relatively small effect (MIT Sloan Automotive Laboratory presentation). Two sets with similar duty can still differ if one has a weaker heater, a regeneration inhibit that is firing more often, a poorly placed temperature sensor or different insulation. See RECD regeneration inhibit signals and rising RECD regeneration duration.

5. Instruments and the way ΔP is read

An instrument fault can make one set look worse. A longer tube, a sag that holds water, a drifting transmitter or a different alarm limit can all produce a difference that exists only on the screen. See pressure sensors versus pressure switches.

Beyond these five, remember fuel (separate tanks or deliveries), layout (a longer pipe loses more heat and creates different back-pressure), site conditions (dust, humidity) and differences in who maintains each set. The twin sheet covers them all.

Tests that separate the suspects

Once the differences are ranked, test them with the least intrusive method first. The principle is to change one thing and watch.

Test A: the duty swap

If set B is the lead and clogs faster, swap the roles for a defined period and watch whether the faster loading moves to set A. If it does, duty is the cause. If it stays with set B, the cause lies in the set itself. This test needs operational agreement and takes weeks, but it is clean, and it often costs nothing. Our article on sequencing RECD installations with N+1 generators shows how rotation plans work.

Test B: the instrument swap

Swap the transmitters or move a portable gauge between the two sets. If the high reading follows the instrument, it is an instrument problem. This is quick and cheap. Do it with the supplier’s agreement and with the system cold and isolated.

Test C: the settings comparison

Export both controllers’ settings and compare line by line: limits, delays, regeneration triggers, learned values, firmware version. Any difference is a candidate. See RECD controller settings and access control.

Test D: the oil and engine checks

Send an oil sample from each set for analysis and compare fuel dilution, wear metals and ash-forming elements. Measure oil consumption on each set over the same number of hours. Run an opacity check and compare boost. These tests separate engine condition from everything else.

Test E: the fuel split

If the sets use different tanks or deliveries, fill both from the same source for a period, or test samples from both. See RECD faults after a fuel delivery.

Test F: the filter core swap

Only as a last resort, and only with the supplier’s written agreement, swap filter cores between the two sets. Handling a core carries a risk of damage, so exhaust every cheaper test first. If the loading pattern stays with the set, the filter is exonerated. If it follows the core, the core is the cause, which is rare. See filter core versus complete assembly replacement and handling soot and ash removed during RECD servicing.

Test G: the exhaust path review

Compare the pressure budget of the two exhaust paths from engine to stack. A longer run, an extra bend or a different silencer changes both back-pressure and temperature. See high back-pressure after RECD installation.

Reading the results

ResultMeaning
Faster loading moves to the other set when duty is swappedThe cause is duty
High reading follows the transmitterThe cause is the instrument
A settings difference explains an earlier alarmThe cause is configuration, not loading
Oil consumption is several times higher on the faster setAsh and oil-related loading
Boost or opacity is worse on the faster setEngine condition
Loading stays with the set after a duty and instrument swap, and oil and engine are equalLook at exhaust path, regeneration hardware and fuel
Loading follows the filter coreCore defect or damage (rare)

Keep a note of every result, including the ones that exonerate a suspect.

When twins are not twins

Sometimes the sets look identical on the nameplate and are not. Check for these.

  • Different engine ratings or tunes under the same kVA label.
  • Different build years or revisions of the same model, with different turbochargers or injectors.
  • Different RECD builds, such as a different filter volume, substrate or catalyst, even where the order sheets looked alike.
  • Different exhaust flow because of a different turbocharger or altitude.

Our guides to same-kVA DG sets needing different RECDs and engine displacement and RECD selection explain why kVA alone does not guarantee matching behaviour. If the sets are not twins, the comparison still helps, but you must adjust for the difference.

A worked illustration: the lead and the lag

The following is an illustrative scenario, not a real customer. A data centre has two 500 kVA sets, DG-1 and DG-2, with the same RECD model fitted on the same day. After eight months DG-1 has reached its action level for ΔP at 50% load twice, while DG-2 has not.

Step one. The team checks like with like. A portable gauge agrees with the controller on DG-2 but reads about 12% lower than the controller on DG-1 at 50% load. Both controllers have the same settings.

Step two: the twin sheet. The sheet shows many rows the same, and four different. DG-1 is the lead set, with 160 run hours against 55 for DG-2. DG-1 has 70% of its hours below 25% load because of weekly no-load tests. DG-1’s oil consumption is about twice DG-2’s. The sensing tube on DG-1 is longer and sags.

Step three: ranking. Duty is large, points the right way and dates from installation. Oil consumption is possible. The instrument difference is small but affects the reading. The sagging tube is cheap to fix.

Step four: tests. The sensing line is re-routed and the gauge difference falls to 2%. The reading on DG-1 drops, but its ΔP is still higher than DG-2’s at the same load. The team then swaps lead and lag for three months and changes weekly tests to loaded tests. DG-2 begins to show a somewhat faster rise, and DG-1’s rebuild rate lengthens. An oil analysis shows that DG-1 has fuel dilution, consistent with light-load running, and its oil top-ups are frequent.

Conclusion. Duty was the dominant cause, an instrument artefact exaggerated the gap and oil consumption was a secondary contributor. The RECDs were fine. The fix was a rotation plan and loaded tests, a repaired sensing line and an engine check on DG-1’s oil consumption.

Recording the comparison

A twin investigation produces knowledge that applies to the whole fleet, so keep it. A one-page twin report should include the sets compared, the twin sheet, the ranking, the tests run and their results, the conclusion and the action. File it with the asset records, and see the RECD asset register for a multi-site fleet and standardising RECD specifications across factory sites. When you buy the next pair, a comparison like this one will tell you which differences you need to control from day one: matched duty, matched oil, matched instruments.

Questions people ask

Does one set clogging faster mean the RECD is defective?

Rarely. A defective filter usually shows damage signs, such as a falling ΔP, smoke or rattling. A faster rebuild rate on one set usually reflects duty, engine condition, oil consumption or instruments.

Should I just swap the RECDs between the sets?

Not first. Swapping is the most intrusive test and carries handling risk. Do the cheap tests, and swap only with the supplier’s agreement.

What if I only have one set?

Compare it with itself over time (rebuild rates, before and after changes), with a similar set at another site, or with the supplier’s commissioning data. See reading an RECD case study’s results.

How many differences are too many?

If you find ten differences, rank them. Most will not matter. The few that survive the size, direction, timing and dose tests are where to spend your effort.

Can both sets have different problems?

Yes. A pair can differ for two reasons at once. Test one at a time and note partial effects, as the worked example shows.

Closing note

When two similar installations behave differently, the difference is usually in the duty, the oil, the engine, the regeneration or the instruments, not in the filter. Confirm the gap is real, fill in the twin sheet, rank the differences by size, direction, timing and dose, and test them with swaps and splits from the cheapest to the most intrusive. If you would like help building a twin sheet or comparing two sets’ logs, you can see our retrofit emission control devices, the FAQ or contact the Aceget team.

Related Reading

Sources



Leave a Reply