Two Gensets Installed Doesn't Mean Two Gensets Running

Walk into most Indian commercial sites with two DG sets — a hotel, a mid-sized hospital, a mall, a two-shift factory — and ask the operator when the second set runs. The answer is often some version of "both start together." Not because anyone decided that, but because the panel was commissioned that way, the electrician has always done it that way, and nothing has visibly gone wrong.

That's an installed-capacity answer to an operating question, and the two aren't the same thing.

Demand moves; installed capacity doesn't

The reason two sets exist is almost always a peak — a summer afternoon with every chiller loaded, a shift when every machine on the floor runs at once. That peak is real, and sizing for it is correct.

But the peak isn't the day. A hotel at 3 AM is lifts, corridor lighting, a handful of room ACs and the kitchen cold rooms. A hospital overnight is critical care, imaging on standby and lighting — a fraction of what the same building draws during OT hours. A single-shift factory after 6 PM is security lighting and a compressor. In every one of those hours, the connected load fits inside one set, often with room to spare.

Installed capacity is a fixed number chosen once. Demand is a curve that moves every hour. Running both sets means matching the fixed number instead of the curve, and paying the difference in diesel.

The second set isn't free just because it's already bought

The capital cost is sunk, which is exactly why the running cost gets overlooked. Four things accrue every hour a genset runs that doesn't need to be running:

Fuel, disproportionately. A genset's efficiency isn't flat across its load range — it peaks somewhere around 75-80% of rated load and gets meaningfully worse below that. Split a load that would have put one set at 70% across two sets, and both now sit near 35%, in the part of the curve where each litre of diesel buys less useful work. We've covered why efficiency moves with load and size in more detail; the short version is that two lightly-loaded sets burn more fuel to deliver the same kWh than one properly-loaded set.

Engine hours, at double rate. Service intervals and most AMC schedules are tied to running hours. Two sets running together means the site accumulates billable service events at twice the pace, for the same delivered energy. Over a monsoon season of frequent outages, that difference is not marginal.

Light-load wear. Manufacturers commonly advise against sustained operation below roughly 30% of rated load. Below that, cylinder temperatures are low enough that combustion is incomplete, and unburnt fuel and lube oil accumulate in the exhaust system — wet stacking. It's a maintenance problem you create slowly and pay for suddenly.

After-treatment, on newer machines. This one is specific to India right now. The CPCB IV+ fleet installed since July 2023 carries after-treatment hardware that depends on exhaust temperature to function and regenerate. Chronic light loading is exactly the condition that keeps exhaust temperatures down. The old habit of running both sets "to share the load" was tolerable on a CPCB II machine; it's a worse habit on the sets replacing them.

Redundancy is about availability, not simultaneity

The obvious objection is redundancy — and it's a fair one, but it's often argued in the wrong place. A second set exists so that a failure of the first doesn't take the site down. That's an availability requirement, and availability is satisfied by a set that is healthy, fuelled, and able to start on demand. It is not satisfied any better by a set that happens to already be spinning at 30% load.

This is the same logic behind N+1 and N+2 redundancy in data centres: the redundant unit's job is to be there. In a properly configured plant, a running set that fails is picked up by the standby set starting and synchronising — the sequence the panel exists to execute.

Where simultaneity genuinely earns its place is step load. If a single large motor, chiller compressor or fire pump can slam a block of load onto the bus that one set can't absorb without an unacceptable frequency or voltage dip, then the second set online is doing real electrical work by providing rotating reserve. That's an engineering judgement about your specific largest step load, made against the set's stated block-load acceptance — not a general rule that both should always run.

What automates the decision

The mechanism is load-demand start/stop, standard in most synchronising and power-management panels sold in India. The logic is straightforward: watch total bus load, and when it crosses an upper threshold — typically a percentage of the online capacity, held for a defined delay so a momentary spike doesn't trigger it — start, synchronise and load the next set. When load falls below a lower threshold for long enough, unload that set, run it through its cooldown period, and stop it.

Two configuration details matter more than they look:

• Hysteresis and delays. The start and stop thresholds must be far enough apart, with long enough timers, that a load sitting near the boundary doesn't cause the second set to start and stop repeatedly. Hunting like that is worse than simply leaving it running.
• Duty rotation. Configure the panel to alternate which set leads — weekly, per start, or by lower hour-meter reading. Without rotation, one set does nearly all the work and reaches overhaul years ahead of its twin, while the other develops the separate problems of a machine that almost never runs.

What to check on your own site

If you already have two sets, four questions are worth asking this week:

1. Does the panel support load-demand start/stop, and is it enabled? Many sites have the function and have never had it commissioned.
2. What are the current hour-meter readings on both sets? Nearly identical readings on a site whose load profile varies is the signature of "both always run."
3. What is your largest single step load, and can one set take it? This is the question that decides whether simultaneity is a real requirement or an inherited habit.
4. What does your AMC actually bill against — calendar visits or running hours? It determines how much the second set's idle hours are costing you.

If you're specifying a new two-set installation rather than reviewing an existing one, put load-demand start/stop and duty rotation in the panel scope explicitly, and size against a real load profile rather than a padded figure — our load calculator is a reasonable starting point, and the comparison tool lets you check published fuel-consumption figures across brands before you commit to a pair.

Having capacity available and having capacity running are two different decisions. Most sites only ever make the first one.

Confirm operating thresholds and load-sharing configuration with your genset manufacturer or panel supplier before changing a commissioned setup.