RECD Planning for Synchronized Generators with Load Sharing

Once two or more diesel generator sets are synchronised onto a common bus, they stop behaving like separate machines. A power management system (PMS) or synchronising panel decides which sets run, how much each one carries and when a set is stopped to save fuel. Those decisions shape the exhaust conditions inside every retrofit emission control device on the site, so planning an RECD for synchronized generators is different from planning one for a lone standby set.

The good news is that the issues are predictable. This guide works through them one at a time: how load sharing actually divides the load, what that does to exhaust temperature and soot, whether identical sets need identical RECDs, what back-pressure differences really affect, why each set should keep its own exhaust, how RECD alarms should reach the sync panel, how to sequence the retrofit while keeping paralleled capacity, and how to monitor each set. A worked example with three sets sharing a 900 kW load sits in the middle, because the numbers make the rest of the guide easier to follow.

Issue 1: How paralleled sets actually share load

kW sharing on a common bus

When sets are paralleled, their output frequency is locked together. Each engine’s governor (or the load-sharing controller acting on it) adjusts fuel so that every set carries its agreed share of the real power, measured in kW. In the most common arrangement, called proportional or isochronous load sharing, each set carries the same percentage of its own rating. Three identical 500 kW sets on a 900 kW load will each carry close to 300 kW.

Reactive power (kVAr) is shared through the alternator voltage regulators and has little effect on the engine or the exhaust. For RECD planning, kW per set is the number to watch.

Some sites use base loading instead: one set holds a fixed output while others swing with the load, so the base-loaded set runs hot and steady. Confirm which mode your panel uses.

Load-demand start and stop

Most modern sync panels include load-demand control, sometimes called load-dependent start/stop. The PMS watches total load and starts an extra set when the running sets climb above a start threshold. When load falls far enough that fewer sets could carry it, the PMS stops a set after a time delay. The aim is fuel saving and engine health: fewer sets running at a healthier load rather than many sets idling.

Load-demand control is usually good news for an RECD, because it pushes the running sets towards higher load. The side effects are that the last set in the priority list may cycle several times a day, and conservative thresholds can keep extra sets running at a low share for long periods.

What to collect before you plan

Gather a few weeks of panel data first: bus kW, kW per set, running status and timestamped start and stop events. The method in using DG load logs to assess RECD suitability applies, but per set. Also record the start and stop thresholds, delays, minimum run timers and priority order.

Issue 2: Per-set load and exhaust temperature

Why share percentage drives the RECD

An RECD’s filter collects soot all the time. Whether that soot burns off in normal running (passive regeneration) depends heavily on exhaust temperature, and exhaust temperature depends mainly on engine load. A set at 60–75% load produces much hotter exhaust than the same set at 20–30%. DieselNet’s overview of DPF regeneration explains the chemistry; the practical point is that a lightly loaded engine produces more soot per hour relative to its ability to burn it off.

In a paralleled system, the per-set load is not the bus load divided by the installed capacity of the whole site. It is the bus load divided by the capacity of the sets that are actually running. That distinction is the whole reason the PMS settings matter to the RECD.

The low-load trap

If load-demand control is disabled, or its stop threshold is very cautious, every set may run through the night at 15–25% load. Every RECD then sees cool exhaust for hours, soot accumulates faster than it burns, and back-pressure rises. The same load on one or two sets would put those engines in a far better temperature band. This is why the guide on choosing an RECD for a generator running at low load applies to paralleled sites more often than people expect: the installed capacity looks large and the per-set load ends up small.

The cycling trap

The opposite problem appears when start and stop thresholds are close together and delays are short. The last set in the priority list then cycles on and off as load wanders around the threshold, often stopping before its exhaust has warmed fully. The patterns described in RECD selection for DG sets with frequent short start-stop cycles then apply to that one set even though the others run steadily.

The fix is usually in the PMS, not the RECD: widen the gap between start and stop thresholds, lengthen the stop delay and add a minimum run timer. These changes belong to the panel integrator and DG maintainer, but the RECD supplier should see the proposed settings.

Worked example (illustrative): three sets sharing a 900 kW load

All figures here are illustrative, chosen to show the method. Your own ratings, thresholds and loads will differ.

The site

A commercial campus runs three identical 625 kVA sets, each rated 500 kW at 0.8 power factor, synchronised on one bus. Total installed capacity is 3 × 500 = 1,500 kW. The PMS uses proportional kW sharing and load-demand control with these illustrative settings:

• Start threshold: start the next set when running sets exceed 85% of their combined rating.
• Stop threshold: stop a set only if the remaining sets would carry the load at 75% or less.
• Stop delay: 10 minutes below the stop condition before acting.
• Minimum run timer: 30 minutes after any start.

Daytime: 900 kW across three sets

At 900 kW with all three sets running, each carries 900 ÷ 3 = 300 kW. That is 300 ÷ 500 = 60% of rating. Would the PMS stop a set? With two sets, each would carry 900 ÷ 2 = 450 kW, which is 90% of rating. That is above the 75% stop limit (and above the 85% start limit), so all three keep running. At 60% load each, all three RECDs see a healthy exhaust temperature for passive regeneration.

Evening: load falls to 700 kW

With three sets, each carries 700 ÷ 3 = 233.3 kW, or 46.7%. If two sets carried it, each would take 700 ÷ 2 = 350 kW, or 70%. That is within the 75% stop limit, so after the 10-minute delay the PMS stops the third set in the priority order. The two remaining sets jump from 46.7% to 70% load. Their exhaust gets hotter; the stopped set’s RECD cools down with whatever soot it collected.

Night: load falls to 320 kW

With two sets, each carries 320 ÷ 2 = 160 kW, or 32%. One set alone would carry 320 kW, or 320 ÷ 500 = 64%, which is within 75%, so the PMS stops the second set. One set now runs at 64% through the night.

What happens without load-demand control

If the panel kept all three sets on the bus all night, each would carry 320 ÷ 3 = 106.7 kW, or 21.3%. That is the low-load trap described above, for all three RECDs at once.

Summary table

Bus load	Sets running (load-demand on)	kW per set	% of 500 kW	Sets running (load-demand off)	kW per set	% of 500 kW
900 kW	3	300.0	60.0%	3	300.0	60.0%
700 kW	2	350.0	70.0%	3	233.3	46.7%
320 kW	1	320.0	64.0%	3	106.7	21.3%

Checking the start side

Suppose the night load climbs back to 450 kW while one set is running. That set would be at 450 ÷ 500 = 90%, above the 85% start threshold, so the PMS starts a second set. Once synchronised, the two share 225 kW each, or 45%. The gap between the 85% start point and the 75% stop point, plus the delays, stops the second set from cycling straight back off. A gap of only 5 points would invite cycling.

What the example tells the RECD planner

• The lead set runs the most hours, mostly at 60–70% load. Its RECD collects the most ash but regenerates well.
• The last set runs only by day. Its RECD sees fewer hours but more cold starts per running hour.
• Disabling load-demand control would turn a healthy site into a low-load site at night, so a proposal to “keep all sets on for resilience” should consider the RECD consequences.

Issue 3: Identical RECDs on identical sets, or not?

The default: match them

If the sets are the same make, model and rating and share load proportionally, fit the same RECD model to each, installed the same way. Matched devices mean one spares list, one set of alarm setpoints and comparable readings. When one set’s differential pressure drifts from the others at the same load, the cause lies with that set, not the hardware.

When different RECDs can make sense

There are legitimate reasons to vary the specification:

• Different duty in practice. A base-loaded set, or one that cycles often at the end of the priority list, may justify a design better suited to its profile.
• Different exhaust geometry. A longer run or tighter space leaves less back-pressure budget, so a different size or layout may be needed.
• Mixed fleets. A 500 kVA and a 750 kVA set on the same bus each need an RECD sized to their own engine and exhaust flow.

Avoid unplanned variety, such as different models simply because different contractors quoted at different times. Decide the specification for the whole bus at the start, even if installation is phased.

Issue 4: Back-pressure differences between paralleled sets

What back-pressure does affect

Every RECD adds some exhaust restriction, and that restriction grows as soot collects. An engine working against higher back-pressure needs slightly more fuel for the same output and runs with a higher exhaust temperature. If back-pressure exceeds the engine maker’s limit, the engine can lose performance and suffer heat-related stress. The article on how exhaust back-pressure affects a DG set covers this in depth.

On a paralleled bus, this means a set whose RECD is more loaded than its neighbours will burn a little more fuel per kWh and run a little hotter at the same kW share.

What back-pressure does not directly affect

Back-pressure differences do not directly change how kW is shared. The governors and load-sharing controller decide each set’s kW, and they will keep the agreed share by giving the more restricted engine more fuel. The set does not “lose” its share simply because its RECD is dirtier. This is a common misunderstanding: operators sometimes see an uneven kW reading and blame the RECD, when the real cause is a load-sharing calibration, governor or controller issue.

The exception is back-pressure so high that an engine cannot produce its share near full load. That is an engine-protection matter, well beyond normal RECD operation, and a correctly sized and maintained RECD should never let it develop.

Practical steps

• Record clean-filter back-pressure on each set at commissioning, at a stated load, so you have a baseline per set.
• Compare sets at the same kW share, not at random moments. The method in comparing RECD pressure readings at different generator loads helps here.
• If one set’s back-pressure rises faster, look first at its operating profile (does it cycle more, or run lighter?) before suspecting the hardware.
• If kW sharing is uneven, ask the panel integrator to check the load-sharing system before anyone touches the RECD.

Issue 5: Separate exhausts versus a common header

Why separate exhausts are the rule

Some older installations route several engines into one shared header, and a single large RECD on it can look attractive. In practice, a stopped engine can see backflow from running engines unless dampers work perfectly, each engine’s back-pressure depends on how many others are running, and you cannot tell which engine produces the soot. Type approval is also normally tied to a device on a specific engine. These issues are covered in detail in can two diesel generators share one RECD.

For load-sharing sites, the strong recommendation is one RECD per engine, on that engine’s own exhaust line, before any shared stack or header.

If a header already exists

If a common header exists, place each RECD upstream of it on that engine’s branch. Where a branch lacks length or space, the DG maintainer and installer should plan the exhaust modification together.

Issue 6: Integrating RECD alarms with the sync panel and PMS

Decide what the PMS needs to know

Adding RECD status to the PMS gives operators one place to see everything, if the signals are chosen carefully. Typical candidates, if your RECD controller provides them, are:

• RECD healthy or fault (a summary signal).
• High back-pressure warning.
• High-high back-pressure, agreed as an engine-protection limit.
• Regeneration in progress, for active designs.
• Sensor or controller fault.

Signals can travel as hardwired volt-free contacts or over a protocol such as Modbus, depending on what the RECD controller and panel support. The Modbus Organization specifications and the guide to RECD Modbus data, units, scaling and register mapping are useful if you go the protocol route.

Alarm versus action

Most RECD alarms should inform, not act. A high back-pressure warning during a grid outage should prompt cleaning plans, not shed the set, because removing a set can overload the others. Where a PMS action makes sense, agree it in writing with the DG maintainer and, where relevant, the engine maker:

• Change priority: if one set’s RECD shows a persistent warning, the PMS could move it down the priority order so it runs less until cleaned, provided capacity allows.
• Prevent start for testing: a set with an active RECD fault could be blocked from routine test runs (not from emergency duty) until checked.
• Engine-protection limit: back-pressure above the engine maker’s maximum is the one case where a controlled stop may be justified. If a set must stop, the PMS should start a replacement first if one is available.

The article on RECD alarm versus generator trip sets out a full framework for these decisions, and the ANSI overview of ISA-18.2 alarm management is a good reference for keeping alarms meaningful. Do not let any single party, including the RECD supplier, decide engine trips alone.

Label signals per set

Tag every RECD signal with its set number, matching the canopy and PMS numbering. A “High DP” alarm that does not say which set is close to useless.

Issue 7: Retrofit sequencing while keeping paralleled capacity

The constraint

The number of sets needed for peak load is fixed by the load, not the project. If peak needs two of three sets, only one can be out at a time, and an N+1 site drops to N while it is, so keep the window short.

A practical sequence

1. Confirm capacity. List the peak bus load during the planned work period and check that the remaining sets can carry it within their rating. Use the PMS data, not nameplate assumptions.
2. Choose the first set. Usually the set lowest in the priority order, so the PMS keeps its normal lead sets available.
3. Change the PMS configuration. Remove the set from the priority list and block it from auto-start while it is isolated. Confirm the PMS will not try to start it on a load step.
4. Install, test and commission. Include a loaded run on the bus, checking that the set synchronises and takes its share normally with the RECD in place.
5. Return to service. Restore the set to the priority list and confirm alarms appear correctly on the panel.
6. Repeat for the next set, only after the previous one is back in service and stable.

The broader guidance in sequencing RECD installations on an N+1 generator system and preserving backup power during an RECD retrofit applies directly. Where the remaining capacity is tight, a temporary rental set can cover the gap; check how it will connect, because a rental set often cannot join the existing sync bus without extra work.

Timing and the mixed fleet

In Delhi NCR, plan the work away from peak summer outages and ideally before the winter GRAP period, when DG restrictions tighten. During the transition, keep retrofitted sets at the top of the priority order so they carry most run hours, and record which sets ran when.

Issue 8: Monitoring each set

Per-set, not per-bus

Bus-level kW and fuel totals do not tell you anything about an individual RECD. Monitoring should capture, for each set:

• kW share and running status, with timestamps from the PMS.
• RECD differential pressure or back-pressure, and inlet temperature.
• Regeneration events (for active designs) and alarm history.
• Run hours and start counts.

Aligning these streams matters. A pressure reading means little unless you know what load the set was carrying at that moment. The guide to aligning RECD sensor timestamps with generator runtime logs shows how to do this.

Use the fleet as its own reference

Identical sets with identical RECDs at the same share should trend alike. When one diverges, check its operating history first (more short cycles, more hours at low share?), then the sensor, then the device. Show all sets side by side at a matched load band, and review per-set load profiles and PMS thresholds yearly, because loads change.

Planning checklist for paralleled sites

Item	Who owns it	Done?
Per-set kW logs for at least two to four weeks	Site / panel integrator
PMS mode recorded (proportional or base load)	Panel integrator
Start and stop thresholds, delays and minimum run timers recorded	Panel integrator
Per-set exhaust layout checked, separate lines confirmed	DG maintainer / installer
RECD specification agreed for the whole bus	Site / RECD supplier
Back-pressure baseline per set at a stated load	Installer / RECD supplier
Alarm list and PMS actions agreed in writing	Site / DG maintainer / panel integrator
Signals tagged by set number on panel	Panel integrator
Retrofit sequence with capacity check for each step	Site / DG maintainer
Monitoring aligned to PMS timestamps	Site / BMS contractor

Frequently asked questions

Will an RECD on one set upset load sharing across the bus?

Not directly. Load sharing is controlled by the governors and the load-sharing controller, which adjust fuel to keep each set at its agreed kW share. A set with higher RECD back-pressure will use slightly more fuel and run slightly hotter for the same share, but it should still carry its share. If sharing becomes uneven, check the load-sharing system first.

Do we have to fit RECDs to all paralleled sets at the same time?

No. Most sites retrofit one set at a time to keep enough capacity on the bus. During the transition, keep retrofitted sets high in the priority order where possible so they carry most of the running hours.

Can one large RECD on our common exhaust header serve all three sets?

It is not recommended. Back-pressure on each engine then depends on how many others are running, stopped engines can see backflow, and you cannot tell which engine is causing a problem. One RECD per engine, on its own exhaust line, is the reliable approach.

Should the PMS stop a set when its RECD raises an alarm?

For most alarms, no. During an outage, removing a set can overload the others. Engine-protection limits, such as back-pressure above the engine maker’s maximum, are the exception, and the response should be agreed with the DG maintainer and engine maker. Moving a set down the priority order is often a gentler option.

Related Reading

• Can Two Diesel Generators Share One RECD?
• How Lead-Lag Generator Rotation Affects RECD Operating Conditions
• Sequencing RECD Installations on an N+1 Generator System
• RECD Alarm vs Generator Trip: How to Define the Control Response
• RECDs for Data Centres: Retrofit and Redundancy Testing
• How to Rank DG Sets for a Phased RECD Rollout

Plan Your Paralleled DG Retrofit with Aceget

Paralleled and load-sharing sites need an RECD plan that fits how the PMS actually runs each set. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA. Review our certifications, then contact our team with your per-set load data and panel settings so we can discuss the right approach for your bus.