In short. A load study records what a facility actually draws, when, and how fast it changes, so that a generator is sized on evidence rather than on nameplates. It logs kW, kVA, power factor, current and voltage per phase at a resolution fine enough to catch motor starts and load steps, over a period that includes the demanding conditions. The result is a load profile and a load model, not a single number.
Connected load is not demand
A plant may have 1,000 kW of equipment installed and never draw more than 450 kW, because not everything runs together, motors run below nameplate, and some equipment is seasonal. Sizing a generator on the installed total buys a set twice the size needed, which then runs lightly loaded, wet-stacks and costs more per kWh. Sizing on a remembered figure from the electrician risks the opposite. The load study replaces both with measurement, and it tells you something nameplates cannot: how the load behaves in the seconds after a step change, which is what a generator alternator actually struggles with.
What to measure and why
| Quantity | What it tells you | Resolution needed |
|---|---|---|
| Real power, kW, per phase and total | Engine loading and fuel consumption | 1 second averages for profile, faster for steps |
| Apparent power, kVA | Alternator loading and heating | Same as kW |
| Power factor | Relationship between the two and alternator margin | Derived, per interval |
| Current per phase | Imbalance, cable and breaker loading | 1 second, with waveform capture on events |
| Voltage per phase | Dips on starts, supply quality, imbalance | Half-cycle RMS during events |
| Harmonic distortion | UPS, drives and LED load content affecting the alternator | Snapshot at representative times |
| Event records | Size and duration of load steps and motor starts | Triggered capture at 10 ms or better |
A 15 minute demand reading from a utility meter averages away every event that matters. It shows a 300 kW plant, not the 180 kW step when the compressor and chiller restart together after an outage. Instrument resolution has to suit the question, which is why a proper study uses a three-phase power quality logger with event triggering rather than a clamp meter and a notebook.
kW, kVA and power factor
The engine supplies real power and the alternator supplies apparent power, and the two are related by power factor.
kVA = kW / PF
Here kW is the real power drawn, PF is the power factor of the load, and kVA is the apparent power the alternator must supply. Worked example (illustrative): a measured peak of 400 kW at a power factor of 0.8 corresponds to 500 kVA. The same 400 kW at a power factor of 0.92, after capacitor correction, is 435 kVA. Neither figure says anything yet about transient capability; it is the starting point of sizing, not the end. Generators are rated at 0.8 lagging, so a load with a leading power factor from lightly loaded UPS filters or long cable capacitance needs separate attention because alternator AVRs can lose control under leading reactive load.
Measure the sequence, not just the total
Two moments decide the generator: the highest sustained running demand, and the most demanding restart event. They rarely coincide. The running peak may come on a hot afternoon with all chillers at full load; the worst restart comes at the instant of utility failure when every motor contactor drops out and, when the generator closes, everything that is not sequenced tries to start at once. The study records which loads start together, their individual starting currents, and the order and delays of any existing restart logic. Motor starting kVA can be five to seven times running kVA for a direct-on-line start, and the alternator’s voltage dip during that period, together with the engine’s frequency dip, decides whether the set rides through or the AVR collapses the voltage and contactors drop out again. That transient analysis is covered in motor starting and motor starting and voltage-dip studies.
How long to log and under what conditions
One quiet shift is not a load study. The logging period should cover the production peak, the seasonal extreme and at least one deliberate restart test if operations allow it: a week in the hot season for a commercial building, a full campaign for a process plant. Note which equipment was running, which was down, and what the measurement did not see, such as a new line due next year, because a profile without its conditions is just a chart.
Phase imbalance and single-phase loads
Indian commercial buildings often carry heavy single-phase lighting and plug loads unevenly across the three phases. A generator is limited by its most loaded phase, so a set that appears to be at 70 percent on total kVA may be at 95 percent on one phase. The study reports per-phase loading and recommends rebalancing before sizing, which is cheaper than a bigger set.
From profile to load model
The measured profile is compared with the equipment inventory and the operator’s plans, and three cases are written down: the normal running case, the demanding restart case and the future case with planned additions. Each value carries a confidence state, measured, estimated or allowance, so that the sizing engineer can see what is known. The output is a sizing band with its transient requirement, for example a class of set that must hold voltage within an agreed dip when a stated motor starts against a stated existing bus load. This is handed to generator sizing and, for complex plants, to transient performance studies. It also becomes the basis for load-shedding priorities and for the restart sequence programmed into the AMF or PLC.
Using a load study on an existing set
When an installed generator trips on load, a load study is the fastest way to separate an undersized set from a faulty one. If the recorded step at the moment of trip is within the set’s published transient capability, the fault is in the engine, AVR, fuel or cooling; if it is beyond it, the fix is sequencing or capacity. Either way the argument ends with data. The study also quantifies how lightly a set is loaded, which explains wet stacking and high litres per kWh and can justify a smaller or a second set.
Common mistakes
- Using 15 or 30 minute utility demand data as the load basis; every motor start and load step is averaged away and the set is undersized for transients.
- Logging during a maintenance week or a low season; the study misses the peak and the result looks reassuring until summer.
- Reporting total kVA only; one phase is near its limit while the total looks comfortable and the set trips on phase overcurrent.
- Treating one measured value as the specification; growth, restart sequence and confidence state are lost and the design inherits hidden risk.
- Guessing motor starting current from a universal multiplier; soft-starter, star-delta and direct-on-line machines behave very differently and the alternator is sized wrongly.
What to send us
Send the equipment list with nameplate ratings and starting methods, the single-line diagram, recent utility bills, the operating schedule including seasonal peaks, any existing logger data, and photos of the main switchboard and metering where a logger would be connected. Tell us whether a controlled restart test is possible. Use the application engineering enquiry; the study usually follows an application assessment and precedes a site survey.
Questions engineers ask
How long should a load study run?
Long enough to include the demanding conditions, which for most commercial sites is one to two weeks in the hot season and for process plants a full production cycle. The period should include at least one restart event if operations allow, because that is the event the generator is sized for.
Can I size a generator from my electricity bill?
Only roughly. The bill gives maximum demand averaged over 15 or 30 minutes and energy per month, which bounds the running load but says nothing about load steps, motor starts, power factor at the moment of restart or phase imbalance. Use it as a cross-check, not as the basis.
What instrument is needed for a load study?
A three-phase power and power quality logger with current transformers on each phase and neutral, capable of one-second averaging for the profile and triggered event capture at 10 milliseconds or better for starts and steps. It should record kW, kVA, PF, voltage, current, harmonics and events together.
Why does my generator trip when the load is below its rating?
Usually because a single step, such as a large motor or several motors restarting together, exceeds the alternator’s transient capability even though the steady load is small, or because one phase is overloaded while the total looks fine. A load study with event capture shows which it is.
Technical content reviewed by Srinivas Reddy, Managing Partner, Reddy Generators. Last updated 19 October 2026.