- October 22, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
On a site with two or more DG sets, one set is usually the “lead”: the first to start when the grid fails. The others are “lag” sets that start only when load demands it or when the lead fails. Which set takes the lead role, and for how long, is a setting in the controller or power management system. Most people think of it as an engine wear question. It is also an emission control question, because the rotation scheme decides how many hours, how many starts and what kind of load each retrofit emission control device (RECD) sees over a year.
This article compares the four common rotation schemes side by side, then looks at the RECD consequences one factor at a time. A 12-month worked example with three sets shows the difference in numbers, and the final sections give practical settings for rotation interval, lag-set exercise and maintenance timing.
The four rotation schemes in brief
| Scheme | How the lead is chosen | Typical changeover | What it tries to balance |
|---|---|---|---|
| Fixed lead | One set is always lead; others are permanent lag | Never, except during faults or maintenance | Nothing; it keeps operation simple |
| Time-based rotation | Lead passes to the next set on a calendar | Weekly, fortnightly or monthly at a set time | Calendar time in the lead role |
| Hours-based balancing | Set with the fewest run hours becomes lead | When the hours gap exceeds a threshold | Cumulative run hours |
| Start-count based | Set with the fewest starts becomes lead | When the start-count gap exceeds a threshold | Number of starts |
Some sites combine schemes, for example monthly rotation with an override if the hours gap grows too large. If your site has never reviewed its setting, it may be running a factory default, often fixed lead.
What each scheme does to a set’s operating profile
Each scheme changes the four things that matter most to an RECD: run hours, starts, typical load and run length.
| Operating factor | Fixed lead | Time-based rotation | Hours-based balancing | Start-count based |
|---|---|---|---|---|
| Run hours per set | Lead very high, lag very low | Roughly even over a year, uneven month to month | Even, by design | Can drift apart if outages vary in length |
| Starts per set | Lead very high, lag few | Roughly even over a year | Roughly even, but not guaranteed | Even, by design |
| Typical load when running | Lead carries outage load alone; lag mostly shares peaks or runs tests | All sets see both lead and lag duty | All sets see both duties | All sets see both duties |
| Typical run length | Lead: outage length; lag: short peaks and test runs | Mixed for all sets | Mixed for all sets | Mixed for all sets |
| Outage-season exposure | Lead takes every summer peak | Depends on who leads in peak months | Self-corrects across seasons | Partly self-corrects |
Any rotating scheme gives every set regular loaded running. But a time-based scheme is only fair on average: whichever set leads in June, when outages peak, picks up extra hours. Hours-based balancing corrects that automatically.
A note on load when the lag set runs
What the lag set’s load looks like depends on how the sets are connected. On a synchronised bus with proportional load sharing, a lag set that joins at a peak takes an equal share with the lead, so its load is reasonable while it runs, but its runs are short. The article on RECD planning for synchronized generators with load sharing covers that arrangement in detail. On sites without paralleling, the lag set may only ever run during tests or lead failures, and its tests are often done off-load. That is where the RECD problems concentrate.
RECD consequences, factor by factor
Soot loading
The filter in an RECD collects soot whenever the engine runs. How fast it fills relative to how fast it clears depends on load and run length. Short, light runs produce soot without giving the exhaust time to get hot, so soot accumulates. Long, loaded runs clear it.
- Fixed lead: The lead set’s RECD sees heavy hours but mostly loaded running, so its soot balance is usually healthy. The lag sets see mostly short, light or off-load runs, so their soot balance is poor relative to their few hours.
- Rotating schemes: Every set gets stretches of lead duty, which act as natural clean-up periods for soot collected during lag duty.
Regeneration opportunity
Passive regeneration needs sustained exhaust temperature. DieselNet’s reference on DPF regeneration explains why temperature and time at temperature both matter. A 20-minute off-load test may never reach the temperature the filter needs. For active designs, regeneration may be triggered by the controller, but even then a set that runs only briefly may not complete a cycle.
- Fixed lead: The lag set may go months without a real regeneration opportunity.
- Rotating schemes: Each lead period gives a regeneration window. Shorter rotation intervals mean shorter gaps between windows for each set.
The guide on RECD selection for DG sets with frequent short start-stop cycles is relevant if a lag set’s profile is dominated by brief runs.
Ash accumulation
Ash is the non-combustible residue, mostly from lubricating oil additives, that stays in the filter after soot burns away. It builds up roughly in step with the oil the engine burns, and so roughly with run hours. It does not regenerate; it has to be removed by cleaning. DieselNet’s page on DPF ash covers the mechanism.
- Fixed lead: The lead set’s RECD reaches its ash cleaning interval much sooner. The lag sets may take years.
- Rotating schemes: All sets accumulate ash at a similar rate, which means they all reach cleaning at about the same time unless you plan otherwise (more on this below).
Wear and component stress
Every start puts the RECD’s sensors, gaskets, insulation and mountings through a thermal cycle, and the engine wears more per start than per steady running hour.
- Fixed lead: One set takes almost all the thermal cycles and hours. Its RECD hardware ages fastest, while the lag sets’ hardware sits idle and may suffer from moisture and corrosion instead.
- Start-count based: Spreads thermal cycles evenly, which suits sites with many short outages.
- Hours-based: Spreads hours evenly, which suits sites with fewer, longer outages.
Monitoring baselines
RECD monitoring relies on comparing today’s back-pressure and temperature with a known clean baseline at a similar load. The method in comparing RECD pressure readings at different generator loads only works if each set runs at a comparable load often enough to produce fresh readings.
- Fixed lead: The lead set has plenty of data. The lag set may have so few loaded hours that you cannot tell whether its RECD is healthy until it is called on during a long outage.
- Rotating schemes: Every set produces regular loaded data, and sets can be compared with each other at matched load bands.
Summary of consequences
| RECD factor | Fixed lead | Time-based | Hours-based | Start-count based |
|---|---|---|---|---|
| Soot balance on lag duty | Poor | Good overall | Good overall | Good overall |
| Regeneration windows | Lead plenty, lag rare | Regular for all | Regular for all | Regular for all |
| Ash cleaning timing | Lead early, lag late | Similar for all | Nearly identical for all | Similar for all |
| Thermal cycle spread | Concentrated on lead | Fairly even | Fairly even | Even |
| Fresh monitoring data | Lead only | All sets | All sets | All sets |
| Planning simplicity | Very simple | Simple | Needs hours data | Needs start data |
Worked example (illustrative): three sets over 12 months
All numbers in this example are illustrative, chosen to show the method. They are not measurements from a real site.
The site
An office complex in the NCR has three identical 500 kVA sets (400 kW each) on a synchronised bus. One set handles a typical outage on its own; a second set joins if load exceeds 85% of one set’s rating (340 kW), mostly on summer afternoons. The third set is the N+1 reserve.
Over the year there are 178 outage events totalling 445 hours, an average of 445 ÷ 178 = 2.5 hours per outage. On 34 of those events the lag set joins, for a total of 65 lag hours.
| Month | Outage events | Outage hours | Lag events | Lag hours |
|---|---|---|---|---|
| Jan | 8 | 20 | 0 | 0 |
| Feb | 6 | 15 | 0 | 0 |
| Mar | 10 | 25 | 0 | 0 |
| Apr | 16 | 40 | 3 | 5 |
| May | 24 | 60 | 8 | 15 |
| Jun | 28 | 70 | 10 | 20 |
| Jul | 22 | 55 | 6 | 12 |
| Aug | 18 | 45 | 4 | 8 |
| Sep | 16 | 40 | 3 | 5 |
| Oct | 12 | 30 | 0 | 0 |
| Nov | 10 | 25 | 0 | 0 |
| Dec | 8 | 20 | 0 | 0 |
| Total | 178 | 445 | 34 | 65 |
The two schemes compared
Scheme A, fixed lead. Set A is always lead, Set B is always lag, Set C is reserve. Sets B and C each get a 30-minute off-load test once a month.
Scheme B, monthly time-based rotation. The lead role moves each month in the order A, B, C. The next set in the order is lag that month, and the third is reserve. Each non-lead set gets one 1-hour test per month, run synchronised on building load, in addition to any lag duty.
Monthly run hours per set
| Month | Scheme A: Set A | Scheme A: Set B | Scheme A: Set C | Scheme B roles (lead / lag / reserve) | Scheme B: Set A | Scheme B: Set B | Scheme B: Set C |
|---|---|---|---|---|---|---|---|
| Jan | 20 | 0.5 | 0.5 | A / B / C | 20 | 1 | 1 |
| Feb | 15 | 0.5 | 0.5 | B / C / A | 1 | 15 | 1 |
| Mar | 25 | 0.5 | 0.5 | C / A / B | 1 | 1 | 25 |
| Apr | 40 | 5.5 | 0.5 | A / B / C | 40 | 6 | 1 |
| May | 60 | 15.5 | 0.5 | B / C / A | 1 | 60 | 16 |
| Jun | 70 | 20.5 | 0.5 | C / A / B | 21 | 1 | 70 |
| Jul | 55 | 12.5 | 0.5 | A / B / C | 55 | 13 | 1 |
| Aug | 45 | 8.5 | 0.5 | B / C / A | 1 | 45 | 9 |
| Sep | 40 | 5.5 | 0.5 | C / A / B | 6 | 1 | 40 |
| Oct | 30 | 0.5 | 0.5 | A / B / C | 30 | 1 | 1 |
| Nov | 25 | 0.5 | 0.5 | B / C / A | 1 | 25 | 1 |
| Dec | 20 | 0.5 | 0.5 | C / A / B | 1 | 1 | 20 |
| Total | 445 | 71 | 6 | 178 | 170 | 186 |
Check: Scheme A totals 445 + 71 + 6 = 522 hours, which is 445 outage hours, plus 65 lag hours, plus 12 test hours (two sets × 12 months × 0.5 h). Scheme B totals 178 + 170 + 186 = 534 hours, which is 445 + 65 + 24 test hours (two sets × 12 months × 1 h).
Annual starts and run length
Starts follow the same logic. In Scheme A, Set A starts for all 178 outages, Set B for 34 lag events plus 12 tests, and Set C for 12 tests. In Scheme B, each set starts for the outages in its lead months, the lag events in its lag months, and 8 tests (one in each of the 8 months it is not leading).
| Measure | Scheme A: Set A | Scheme A: Set B | Scheme A: Set C | Scheme B: Set A | Scheme B: Set B | Scheme B: Set C |
|---|---|---|---|---|---|---|
| Run hours | 445 | 71 | 6 | 178 | 170 | 186 |
| Starts | 178 | 46 | 12 | 79 | 75 | 82 |
| Average run per start (h) | 2.50 | 1.54 | 0.50 | 2.25 | 2.27 | 2.27 |
| Off-load starts | 0 | 12 | 12 | 0 | 0 | 0 |
How the Scheme B starts add up: Set A leads in Jan, Apr, Jul and Oct (8 + 16 + 22 + 12 = 58 starts), is lag in Mar, Jun, Sep and Dec (0 + 10 + 3 + 0 = 13 starts) and has 8 tests, giving 79. Set B leads in Feb, May, Aug and Nov (6 + 24 + 18 + 10 = 58), is lag in Jan, Apr, Jul and Oct (0 + 3 + 6 + 0 = 9) and has 8 tests, giving 75. Set C leads in Mar, Jun, Sep and Dec (10 + 28 + 16 + 8 = 62), is lag in Feb, May, Aug and Nov (0 + 8 + 4 + 0 = 12) and has 8 tests, giving 82. The total, 79 + 75 + 82 = 236, matches Scheme A’s 178 + 46 + 12 = 236.
What the numbers mean for the RECDs
- Set A under Scheme A runs 445 hours, all loaded. Its soot balance is healthy, but ash builds fastest. If the supplier’s ash cleaning interval were, say, 400 hours (an illustrative figure; use your supplier’s actual guidance), Set A’s cumulative hours would reach 400 at the end of October (20 + 15 + 25 + 40 + 60 + 70 + 55 + 45 + 40 + 30 = 400). That is exactly when the winter GRAP period begins and DG use is under the closest scrutiny.
- Set C under Scheme A runs 6 hours in 12 half-hour off-load starts. Its RECD barely warms up, never sees a meaningful regeneration window, and produces no useful monitoring data. If the lead fails during a long summer outage, nobody knows the condition of the device that is about to carry the building.
- Set B under Scheme A sits in between: mostly loaded summer peaks, but a quarter of its starts (12 of 46) are off-load tests.
- Under Scheme B, all three sets run between 170 and 186 hours, every run is loaded, and average run length is about 2.25 hours for all of them. Each set gets a four-month lead cycle with long runs that clear soot, and each produces regular data for comparison.
- The time-based imbalance is visible too: Set C runs 186 hours because it happened to lead in June and September. Hours-based balancing would have pulled the three sets closer to the average of 534 ÷ 3 = 178 hours each.
Scheme B costs 12 more run hours a year (534 against 522) because its tests are longer and loaded. That is the price of keeping all three RECDs and engines in known, healthy condition.
Recommendations
Choosing a rotation interval
For most standby sites, a lead changeover every one to four weeks works well. Longer intervals make it likelier that one set absorbs a whole outage season. A monthly rotation with an hours-gap override (for example, switch early if one set is 25–50 hours ahead) captures most of the benefit. Where outages are many and very short, consider start-count balancing. Fix the changeover at a quiet hour in IST, away from known high-risk periods.
Give the lag set real exercise
Whatever scheme you use, a set that is not leading still needs periodic loaded running. Off-load tests prove that the engine starts, but they are poor for the engine and worse for the RECD. Practical options:
- Run the set synchronised on building load for 30–60 minutes once a month, if the panel allows.
- Use a load bank where paralleling is not possible.
- Agree a minimum load for test runs with your DG maintainer and the engine maker’s guidance; many sites target a substantial share of rating rather than idle.
The articles on whether an RECD suits a DG set that runs only occasionally and choosing an RECD for a generator running at low load give more background on why idle running hurts.
Align rotation with RECD maintenance
Balanced rotation has one side effect that catches sites out: all the RECDs reach their cleaning intervals together. If all three need filter cleaning in the same fortnight, you lose redundancy while the work is done. Plan around it:
- Stagger on purpose. Configure the hours-based override with a deliberate offset, or schedule the first cleaning of each set at different points, so due dates spread across the year.
- Clean at the end of a lead period. A set coming off lead duty has just finished its heaviest running; take it out of the rotation for cleaning then, and let the next set lead.
- Avoid peak seasons. Keep cleaning out of the summer outage peak and the winter GRAP months where possible. The RECD maintenance guide covers what cleaning involves.
- Lock out during work. Remove the set from the rotation and block auto-start while its RECD is being serviced, then restore it and confirm its alarms.
Keep the baselines fresh
Use the rotation to refresh monitoring. Each time a set completes a lead period, note its back-pressure and inlet temperature at a standard load band. Over a year this gives a clean trend for every set. Keep run hours and start counts in the same record, using consistent timestamps; the guide to the RECD event log and what to record suggests a format.
During a phased retrofit
If only some sets have RECDs yet, rotation needs care. During GRAP periods in Delhi NCR, restrictions on DG use make it sensible to keep RECD-fitted sets in the lead role and limit non-retrofitted sets to emergency backup, subject to the rules in force. The GRAP rules for DG sets in Delhi NCR and the CPCB website are the starting points for current requirements. The posts on ranking DG sets for a phased RECD rollout and sequencing RECD installations on an N+1 generator system help decide the order.
Frequently asked questions
Does a fixed-lead scheme damage the RECD on the lead set?
Not usually. The lead set’s RECD runs plenty of loaded hours, which suits regeneration. Its main consequence is faster ash build-up and earlier cleaning. The bigger risk sits with the lag sets, whose devices see mostly short or off-load runs.
How often should we rotate the lead set?
For most standby sites, every one to four weeks, with an override if run hours drift too far apart. Very short intervals add changeovers without much extra benefit.
Is a 15-minute off-load test enough for the lag set?
It proves the set starts, but it does little for the engine or the RECD. A monthly loaded run of 30–60 minutes, on building load or a load bank, is a much better exercise. Agree the test load with your DG maintainer.
Will balanced rotation make all our RECDs need cleaning at once?
It can. Stagger the cleaning on purpose, either by building an hours offset into the rotation or by scheduling each set’s first cleaning at a different time, so you never lose more than one set at a time.
Related Reading
- RECD Planning for Synchronized Generators with Load Sharing
- Sequencing RECD Installations on an N+1 Generator System
- Understanding Generator Load Factor and Duty Cycle
- RECD Maintenance Guide
- How to Rank DG Sets for a Phased RECD Rollout
Talk to Aceget About Your Multi-Set Site
The way your sets share duty shapes how every RECD on site performs. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA. See our certifications, then contact our team with your run-hour and rotation records so we can discuss what suits your sets.