How 8,760 dispatch models handle generator part-load efficiency: an audit-ready breakdown

Key takeaways

• A positive fuel-curve intercept is what penalizes running an oversized generator lightly loaded.
• Minimum generator loading can force surplus energy; the model must send it to storage or record it as dumped output.
• Battery efficiency must be applied consistently to charging, stored energy, and later discharge—not as an annual shortcut.
• The strongest audit is an hourly balance around minimum SOC, generator starts, and periods of renewable surplus.

Why annual fuel-per-kWh shortcuts fail

A diesel generator does not burn fuel in direct proportion to electrical output all the way to zero. Pumps, cooling, friction, and engine operation create a no-load or intercept penalty whenever the unit is on. That is why a 100 kW set carrying 15 kW typically consumes more fuel per delivered kWh than the same set carrying 80 kW.

An annual model that multiplies total generator energy by one constant fuel rate erases the timing of starts, minimum loading, storage charging, and renewable surplus. Two systems with the same annual diesel kWh can therefore have materially different fuel use.

The implemented linear fuel curve

MicrogridModeler uses the common intercept-plus-slope form F = F0 × Prated + F1 × Pout whenever the generator is on. F0 has units of litres per hour per rated kW; F1 has units of litres per kWh of electrical output. If the generator is off, fuel is zero.

The default local-engine coefficients are F0 = 0.08 L/h per rated kW and F1 = 0.25 L/kWh, with a 15% minimum-load ratio. These are editable screening defaults, not a substitute for a manufacturer curve. HOMER Pro publishes the same mathematical form in its generator documentation, although project coefficients and dispatch behavior can differ.

Sources for this section: HOMER Pro generator fuel-curve slope

What happens inside one chronological hour

The off-grid loop first sends PV and wind to load. Renewable surplus charges storage within converter, charge-power, capacity, and SOC limits. Any remaining renewable output is curtailed. The remaining deficit is served by storage and the generator under the selected dispatch strategy. Any final deficit is unmet load.

The exact storage energy at the end of hour t becomes the starting point for hour t+1. That propagation is why an 8,760 model can reveal a three-day cloudy spell that an average-day model misses.

1. Compute load, PV, and optional wind available in hour t.
2. Serve load from renewables and charge storage with renewable surplus.
3. Determine storage deliverable power from SOC, discharge power, converter power, and efficiency.
4. Commit the generator if the real deficit cannot be covered under the dispatch rule.
5. Apply minimum output, route generator surplus to storage when allowed, and record the remainder as dumped output.
6. Calculate fuel from the actual generator setpoint and close the energy balance.

Load-following and cycle-charging expose different fuel trade-offs

In cycle-charging mode, a committed generator runs at rated power. This can improve its effective litres per kWh, but only if the extra output displaces future low-load runtime or charges storage usefully. If the battery is full or charge-power-limited, the unused output must be recorded as dumped energy.

MicrogridModeler’s load-following mode is deficit-driven but uses a 24-hour net-deficit look-ahead once the generator is committed. The unit can remain on and bank enough energy to coast through the near-term deficit, reducing rapid starts. It never starts only to charge. This is a deterministic rule-based dispatch, not a perfect-foresight operating-cost optimization.

The engine also exposes NPS-compatible diesel-first and maintain-maximum-SOC policies for controlled comparisons. The public methodology documents all four policies; this section focuses on the two modes where generator surplus is most important to the part-load fuel audit.

How to detect hidden energy dumping

Hidden dumping occurs when a model forces a generator above load but reports only the energy serving load or charging storage. Fuel is then charged for one output while the energy table accounts for less. The cure is not to forbid surplus; minimum loading and cycle charging can create legitimate surplus. The cure is to name and balance it.

For each hour, verify that generation plus discharge and imports equals load plus charging, curtailment, exports, conversion treatment, and unmet load. At annual scale, separately total renewable curtailment and generator-dumped energy. A model that cannot expose those series is difficult to audit.

• Check every generator-on hour below the minimum-load setpoint.
• Check hours when the battery is at maximum SOC or charge power.
• Compare generator electrical output with genset-to-load plus genset-to-battery plus dumped genset output.
• Compare hourly fuel with F0 × rated kW + F1 × actual generator output.
• Confirm that annual excess-energy metrics include dumped generator output as well as renewable curtailment.

From screening curve to decision-grade generator data

For concept screening, a transparent linear curve is often better than an opaque efficiency label. Before procurement, replace defaults with manufacturer fuel points at several loads, confirm fuel heating-value convention and ambient derating, and decide whether a piecewise curve or generator fleet is required.

Also test minimum up/down time, start fuel, maintenance intervals, spinning reserve, and N+1 policy when those constraints affect the decision. The local single-generator screen should hand those questions to detailed engineering rather than pretending they do not exist.

Keep exploring

Sources and review notes

This article is grounded in the cited technical sources and the stated modeling assumptions. Recheck project inputs, equipment data, and local requirements before using it for design.

FAQ