Short answer
An industrial electricity bill has components that behave nothing alike. Consumption is billed on kWh, demand is billed on the highest recorded interval of kW or kVA, and power factor penalties are billed on the relationship between real and apparent power. Demand and power factor are set by conditions lasting minutes, invisible until the invoice arrives weeks later. Feeder level metering is what converts both from a monthly surprise into a controllable variable.
Three bills wearing one invoice
Most plants manage electricity as a single quantity: total kWh, tracked monthly, compared against production. That framing hides the fact that the invoice is composed of components with entirely different causes and entirely different remedies.
| Component | What sets it | The remedy | Visibility without metering |
|---|---|---|---|
| Consumption | Total energy used across the period | Efficiency, scheduling, load reduction | Visible on the bill, but with no attribution |
| Demand charge | The single highest kW or kVA interval in the period | Load sequencing, peak shaving, on site generation timed to the peak | None. The event is over long before the invoice arrives |
| Power factor penalty | Ratio of real to apparent power across the billing period | Correction equipment, sized against measured data | None, and usually misdiagnosed when it is noticed |
| Transformer and distribution losses | Loading, ageing and layout of the distribution network | Load balancing, reconfiguration, replacement where justified | None. These losses are inside your own network |
Peak demand: a fifteen minute event, billed for months
A demand charge is levied on the highest rate of draw recorded in any interval, typically fifteen minutes, and in many tariffs it ratchets, persisting across subsequent billing periods.
In manufacturing, peaks are usually coincidence rather than requirement. A shift start where every motor starts together. A compressor loading at the same moment a chiller stages up and a furnace draws. Individually reasonable, collectively expensive, and nobody planned the collision because nobody could see it.
Power factor: the penalty that is nearly always fixable
Poor power factor means a facility draws more apparent power than the real power it consumes, usually because of inductive loads such as motors, transformers and lighting. Utilities penalise it because the network has to carry the additional current regardless.
It is close to a solved engineering problem. Correction equipment is mature and well understood. What goes wrong is sizing and placement, because plants install correction based on nameplate assumptions rather than on measured load profile, and then either under correct, which leaves the penalty in place, or over correct, which creates its own problems including leading power factor and resonance with harmonics.
Feeder level measurement solves this by showing where the poor power factor actually originates, so correction lands at the load rather than at the incomer where it is least effective.
What feeder level metering makes possible
- Attribution. Energy allocated to units, lines and processes rather than to the site, which is the precondition for any intensity metric per tonne or per litre of output.
- Peak diagnosis. The peak interval broken down by contributing feeder, which turns peak reduction from a guess into a sequencing decision.
- Correct power factor correction. Sized and located against measured data rather than nameplate assumptions.
- Loss estimation. Transformer and distribution losses estimated end to end, so losses inside your own network are visible rather than absorbed.
- Solar optimisation. On site generation is worth far more offsetting a peak than exporting at a low tariff, and that decision requires both sides measured live.
- Motor and asset health. Current signature changes on a feeder are an early indicator of mechanical problems, which turns an energy meter into a condition monitoring input.
What it takes to install
Electrical sub metering is among the least disruptive instrumentation on a plant. Split core current transformers clip around existing conductors without breaking the circuit, and voltage references are taken from existing points within the panel. Existing smart meters and power quality analysers are read over Modbus or an equivalent protocol and consumed directly, so previous investment is not wasted.
The practical constraint is usually panel access and safe working rather than technology, which is why the survey establishes the panel list and the switching requirements before anything is committed.
What is a power factor penalty and how do you avoid it?
It is a charge levied when the ratio of real power to apparent power falls below a threshold set by the utility, usually caused by inductive loads such as motors and transformers.
It is avoided with correction equipment, but the equipment must be sized and located against measured load data. Correction applied at the incomer based on nameplate assumptions frequently under performs, and over correction can create leading power factor and harmonic resonance problems of its own.
How do you reduce peak demand in a factory?
Identify what actually coincides at the peak interval, which requires feeder level measurement, then stagger it. Shift starts, compressor loading, chiller staging and furnace cycles can usually be separated by minutes with no production impact.
Then time any on site generation or storage to cover the remaining peak, and set alerts as demand approaches the period high so the control room has the option to act while the interval is still open.
Is feeder level metering disruptive to install?
It is among the least disruptive instrumentation available. Split core current transformers clip around existing conductors without breaking the circuit, and existing smart meters and analysers can be read over standard protocols rather than replaced.
The constraint is safe panel access rather than technology, and that is scoped during the survey.
Why not just use the utility smart meter data?
Utility interval data tells you what the site did in aggregate. It cannot tell you which load caused a peak, where poor power factor originates, or what any individual unit consumed, which means it supports neither diagnosis nor attribution.
It is a useful input, and Aqueduct consumes it, but it is a boundary measurement rather than a management one.
