Sizing solar for a diesel-powered site: why bigger is not always better

Start with the load profile

Solar can only displace diesel while the sun shines. The first figure to establish is the share of generator energy used in daylight hours. It should come from interval data covering weekdays, weekends and seasons.

A site that runs night shifts, or peaks in the evening, will save less than its annual consumption suggests. A 2015 Frankfurt School-UNEP Centre study looked at hybrid mini-grids with evening peaks. PV supplied 31 to 40% of total electricity, even though it was sized to carry the whole load at midday.

Timing matters as much as volume. ESMAP found that adding daytime business demand to a solar-hybrid mini grid cut its cost of electricity by 25%. It also noted that adding load alone generally means burning more diesel.

The generator sets a floor

Diesel generators should not run lightly loaded for long. Cummins recommends a minimum of 30% of nameplate rating and never less than 10% for extended periods. Caterpillar links extended running below 30% to wet stacking and deposits, and recommends periodic higher loading to clear them.

While a generator runs, solar can only supply the load above that floor. Take a 500 kW generator with a 30% floor serving a steady 300 kW daytime load. The generator must produce at least 150 kW, so solar can usefully supply at most 150 kW at any moment.

A hybrid controller enforces this limit. SMA’s controller, for example, cuts PV output within five to seven seconds when generator load would fall below the set minimum. In critical cases, it shuts the inverters down in under 2.5 seconds to prevent reverse power into the generator.

Spinning reserve

Running generators must also keep spare capacity for a sudden load step or a cloud over the array. HOMER, a widely used hybrid design tool, uses 10% of load as its default reserve and describes 25% of PV output as a standard addition. More solar therefore raises the reserve the generators must hold.

PV inverters follow the voltage and frequency set by the generators or the grid, as SMA’s guidance notes. Unless a grid-forming battery inverter provides that reference, at least one generator must keep running while solar produces. On multi-generator sites, reserve rules decide whether a unit can be switched off.

SMA’s guidance describes the impact of PV as very low up to 10 to 15% of generator rating, and calls for special controls above 20%. With its controller, it rates stable operation up to 60%, measured against the generators running in parallel.

Curtailment and diminishing returns

Once midday solar output exceeds the headroom above the generator floor, extra panels add energy only in the early morning and late afternoon. The rest is curtailed. Each additional kWp therefore saves less fuel than the one before.

A simplified clear-day illustration shows the effect, using the 500 kW example above. It assumes a solar profile peaking at 0.7 kW per kWp, about 5.3 kWh per kWp per day, and HOMER’s example fuel curve:

• 200 kWp: about 1,070 kWh a day, none curtailed, saving about 292 litres.
• 300 kWp: about 1,600 kWh produced, 15% curtailed, saving about 374 litres.
• 400 kWp: about 2,140 kWh produced, 31% curtailed, saving about 405 litres.
• 500 kWp: about 2,670 kWh produced, 42% curtailed, saving about 423 litres.

Doubling the array from 200 to 400 kWp raises fuel savings by less than 40%. Real results depend on weather, load variation and controls, so an hourly or sub-hourly simulation is essential.

The fuel savings arithmetic

A generator’s fuel use is close to a straight line: a fixed no-load amount per hour plus a marginal amount per kWh. In HOMER’s documented example, a 50 kW set burns 8.48 L/h at 25 kW and 15.3 L/h at 50 kW. That gives a marginal rate of 0.273 L/kWh and a no-load consumption of about 1.66 L/h.

At half load, that set averages 0.339 L/kWh. While it keeps running, however, solar only saves the marginal 0.273 L. Multiplying solar output by average consumption would overstate savings by about a quarter in this example.

A simple estimate follows. Litres saved equal the solar kWh actually absorbed times the marginal rate, plus generator hours avoided times the no-load rate. Multiply by the delivered fuel price and test a range of prices.

IFC’s 2019 study of backup generators put the fuel-only cost of their output at USD 0.20 to 0.50 per kWh across the countries modelled. Fewer running hours can also reduce maintenance, which should be costed separately.

When a battery changes the answer

A battery changes the constraints, not just the timing. With a grid-forming inverter and enough capacity, it can hold the reserve and let generators switch off around midday. Each hour off saves the no-load fuel as well as the marginal fuel, and cuts running hours.

It can also store midday energy that would otherwise be curtailed and release it in the evening. Whether that pays depends on the curtailed volume, evening load, battery cost, conversion losses and cycle life.

A battery sized only to cover interruptions of up to 15 minutes, as assumed in the 2015 study, does not shift energy into the evening. Without storage, the array that gives the best return is often smaller than the roof allows.

What to ask a supplier

• What interval data was used, over what period, and what share of generator energy falls in daylight hours?
• What minimum load was assumed for each generator, and does it match the manufacturer’s guidance?
• What spinning reserve rule was modelled, and how many generators must run at midday?
• How much solar energy is curtailed each year, in kWh and as a share of output?
• Are fuel savings based on the marginal fuel rate or an average L/kWh, and which fuel curve was used?
• Which controller will manage minimum load, reverse power and ramp rates, and is it compatible with the existing generator controls?
• If a battery is proposed, will it let the generators switch off, and what usable capacity, cycling and warranty apply?

The answers usually show whether a larger array adds value or only adds curtailed energy.