A meter shows negative active power when the current it measures is more than 90° out of phase with the voltage it pairs that current with. Export does this, and so does a reversed or misplaced current sensor. Other wiring faults leave the total positive and wrong, in the worst common case by two thirds. Find the cause before you change any data.
This guide is part of the power fundamentals series.
Export and sign convention
A site with solar exports when generation exceeds load, often around midday. A battery or a CHP unit can reverse the flow within minutes, so one 15-minute interval can hold both import and export. At the incomer of such a site, negative total power can be correct.
Most meters use the load (consumer) convention, in which import is positive. A meter on a generator or battery feeder is often set to the generator convention, so that its output reads positive. A four-quadrant meter keeps import and export in separate registers. IEC 62056-6-1 identifies them as OBIS 1.8.0 (active energy import) and 2.8.0 (active energy export). Net energy for an interval is import minus export. On a site that imports and exports in the same interval, net energy and gross import are different numbers, so record which one each report uses.
Three settings hide the sign:
- A destination that stores the absolute value of power hides real export.
- A system that adds the import and export registers gives a figure that matches neither.
- A meter set to absolute energy accumulation counts reverse flow as consumption. Schneider Electric documents the absolute and signed modes for its meters. On such a meter, a circuit with reversed CTs shows negative kW while the kWh register still rises at the correct rate, so a check on energy alone passes.
Test the convention with a known load. Switch on a resistive load, such as a heater, with no generation running. Active power must rise with a positive sign and a power factor near 1 on each phase that feeds it. Write the convention of the meter and of every system that stores its data into the point list.
Check each phase before the total
On a balanced load, one reversed CT gives +1, +1 and −1 per phase, so the meter reports about one third of the true total. That can pass as a quiet day. Read voltage, current, power factor and active power on each phase under a normal load, with no export:
| What you see on one phase | Likely cause |
|---|---|
| Negative power. Power factor the same size as the other phases, but negative | The CT faces the wrong way, or S1 and S2 are swapped at the meter |
| Power factor near zero or negative, with normal current | The CT is on another phase, so its current is paired with the wrong voltage |
| Positive power factor well away from the other phases, with normal current | The CT is on another phase and is also reversed |
| Current that does not match a clamp meter on the same conductor | Wrong CT ratio or secondary rating set in the meter, or the CT is on another conductor |
| Zero or very low current with load on the circuit | The CT is not connected, or it is round a whole multicore cable, where the phase currents cancel |
A CT on the wrong phase shifts the angle between its current and the paired voltage by 120°. A load with a true power factor of 0.9 (φ = 25.8°) then reads cos 145.8° = −0.83 or cos −94.2° = −0.07, depending on which phase the current comes from. A load at unity power factor reads −0.5. If the CT is also reversed, the shift becomes 60°, and the same 0.9 load reads +0.83 or +0.07. The +0.83 case is the hard one to see. Power on that phase is positive and about 8% low, and 0.83 looks like an ordinary power factor. On a balanced load the three power factors are close, so check any phase that stands apart. The three-phase power guide explains the pairing.
IEC 61869-2 marks the primary terminals P1 and P2 and the secondary terminals S1 and S2, with P1 and S1 at the same polarity. Mount each CT with its P1 side towards the supply. Connect S1 and S2 to the terminals that the meter instructions give for that phase. Some meters can invert a channel in software. That corrects the data but leaves the CT against its label, so mark the inversion on the drawing and in the point list.
Record the original channel mapping and the readings before the correction, so that the record shows what was wrong.
Two-element metering on a 3-wire supply
The phase checks above assume a 4-wire supply metered with three elements. A 3-wire supply can be metered with two elements, because Blondel's theorem shows that n conductors need n − 1 elements. Each element then measures a line-to-line voltage against a line current, so element power is not phase power. On a balanced load the two elements read V × I × cos(30° + φ) and V × I × cos(30° − φ), where V is the line-to-line voltage and I is the line current. When the power factor falls below 0.5 (φ > 60°), the first element reads negative. That reading is correct, and the sum of the two is still the true total. Do not reverse a CT to make it positive.
Reconcile a main meter with its submeters
Draw the metering tree before you compare numbers. Show the main meter, each submeter below it, each circuit without a submeter, and every source of generation or storage. Mark any meter on the other side of a transformer. The US Federal Energy Management Program groups meters at building, distribution and end-use level, which is a useful way to lay out the tree.
Then compare energy over the same period:
| Meter | Energy over the period |
|---|---|
| Main meter | 1,000 kWh |
| Submeter A | 400 kWh |
| Submeter B | 350 kWh |
| Unallocated | 250 kWh |
The 250 kWh is unallocated energy until you know what it contains: circuits without submeters, shared services, transformer losses or a different measurement boundary. Do not report it as meter error. Label it in the point list and trace it circuit by circuit.
Set the tolerance before you compare
Agree the tolerance for the whole measurement chain before you compare. Add the accuracy limits of each meter and its CTs for a worst case.
These limits apply under reference conditions, and they widen at low current. IEC 61869-2 allows a class 1 CT ±3% ratio error at 5% of rated current, so a submeter that runs lightly loaded for most of the month carries a wider band. Adding the limits is conservative. A root-sum-square combination gives a tighter band, but it needs a justified uncertainty model. The high-accuracy metering guide explains the difference.
Align the time, units and counters
Check these points before you suspect a meter:
| Check | Common cause of a difference |
|---|---|
| Start and end time | Midnight in one system, the billing read time in another |
| Time zone | Local time against UTC, or a daylight-saving change inside the period |
| Interval label | One system labels an interval by its start, another by its end |
| Quantity | Average kW compared with a kWh increment |
| Scale | Wh read as kWh, or a CT ratio applied twice |
| Sign | Net energy compared with gross import, or an absolute-value register |
| Counter history | A counter reset, a rollover or a replaced meter |
| Completeness | Missing intervals, duplicates or stale values |
For cumulative counters, subtract readings taken at the same timestamps. The interval demand calculator turns a counter difference into average kW. Mark an interpolated boundary value as interpolated, so that nobody treats it as a measured reading.
A negative difference between two counter readings means a reset, a rollover, a replaced meter or a data fault. Mark the interval and find the event before you use the data. The commissioning checklist gives the rollover arithmetic for a 16-bit register.
Record the correction
After a correction, retest under a normal load and one other operating state, such as a machine on and off. Keep the failed result, the change and the passed retest in the commissioning record. If only part of a difference is explained, record the rest as open. Do not scale or correct historical data without a record of why.
Per-phase data with a ZEM
The ZEM electricity monitor ships with its current sensors connected and calibrated to the meter, so the sensor ratio and type are set before installation. The installer still places each sensor on its phase and in the right direction. The ZEM reports voltage, current, active power, power factor and energy for each phase, so the checks in the phase table work from the first reading. When the main meter and the submeters all report to the same Gateway, Edge stores their readings there on one time base, so the intervals line up for the comparison. The energy monitoring commissioning checklist lists the records to keep at handover.
Common questions
Why is my energy meter showing negative power?
Power flows out through the meter, the meter or the software uses the opposite sign convention, or a current sensor is reversed or paired with the voltage of another phase. On a 4-wire supply with no generation, one negative phase almost always means a wiring fault on that phase. On a 3-wire supply metered with two elements, one element reads negative whenever the power factor is below 0.5, and that reading is correct.
How do I know if a CT is installed backwards?
With a normal load and no export, active power on that phase is negative. Its power factor has the same size as the other phases but the opposite sign, and the current is normal. A qualified person turns the CT so that P1 faces the supply, or swaps S1 and S2 at the meter, with the circuit isolated or the CT secondary shorted as the maker requires.
Why do my submeters not add up to the main meter?
First check that the periods, units and time zones match. Then draw the metering tree. Circuits without submeters, shared services, transformer losses and generation behind a submeter all make a difference. Compare that difference with the combined accuracy limits of the meters. Anything larger is real energy that the tree does not yet explain.