Reducing losses in 0.4 kV networks from smart-metering data

From the data a distribution company already collects, GRID-IN shows at which consumers energy is lost and why, and calculates the technical losses of the network. The result is a list of addresses for field inspection, ordered by expected return. No additional hardware is needed.

The problem

The metering system (AMI) shows the substation imbalance: how much was supplied at the input and how much was metered at consumers. It does not show where the difference is lost or what it is: physics, theft, a faulty meter, a wrong substation assignment. Inspecting every consumer is expensive; random inspections find little.

What GRID-IN finds

Non-technical losses

unmetered consumption — meter tampering, load bypassing the meter;
unauthorised consumption, including houses missing from the metering records;
cryptocurrency mining;
faulty meters and meters that stopped reporting;
“extra” meters — supplied from a different substation than recorded.

Technical losses

losses of the substation and of every overhead and cable section — for the period and by hour;
current load of sections relative to the rated current;
voltages at nodes and at consumers;
diagram vs. metering mismatches: a meter drawn on the wrong substation, a house on the map with no service connection on the diagram.

Inputs and outputs

Inputs	Outputs
load profiles of consumer meters and of the substation input meter — required; meter event logs, billing, consumer register; single-line diagram of the 0.4 kV network — for technical losses. Scale: individual substations or a whole district. Exports are accepted as the metering system produces them.	field target register: address, meter, type of irregularity, estimated unmetered energy, probability of confirmation; targets grouped by location and ordered by return — highest first; map and substation diagram with targets marked; estimated savings, kWh per year, per substation and per project; technical losses and network state by section.

Inputs

Outputs

load profiles of consumer meters and of the substation input meter — required;
meter event logs, billing, consumer register;
single-line diagram of the 0.4 kV network — for technical losses.

Scale: individual substations or a whole district. Exports are accepted as the metering system produces them.

field target register: address, meter, type of irregularity, estimated unmetered energy, probability of confirmation;
targets grouped by location and ordered by return — highest first;
map and substation diagram with targets marked;
estimated savings, kWh per year, per substation and per project;
technical losses and network state by section.

How it is built

PREPData preparation: cleaning, metering reconciliation, map and diagram

GRIDCOBalance groups: assembled or verified

GRIDJOBalance method without a network diagram

GRIDSOAnalysis of events, billing and profiles

GRIDTONetwork state calculation from the diagram

FUSIONOne list of findings with a score

The methods are independent: each works with whatever data is available. Without a diagram, the diagram-free methods work; without event logs, the others do. The more data, the more findings are confirmed by several methods at once. The balance method does not lose accuracy as the imbalance grows.

Results for the field crew

The register opens in a browser and on a phone: list, map, route to the point, substation diagram with a description of the anomaly and an on-site inspection procedure. The register exports to PDF. The visit outcome — “confirmed”, “not confirmed”, “no access” — is recorded in the same place and is visible to the manager.

Illustrative example. A 0.4 kV substation, 64 consumers, one month. Supplied at the input: 41,800 kWh; metered at consumers: 35,900; imbalance: 5,900 (14%), of which technical losses: 1,700. GRID-IN points to three addresses with estimates of ≈ 2,600, 1,100 and 500 kWh: three addresses to check instead of 64.

How to start

Pilot. A metering export for several substations over one period — files as they are.
Calculation and register. Targets and technical losses for each substation.
Field visits. Crews check the targets and record outcomes — the confirmed share is visible.
Roll-out. A whole district; regular uploads by hand or via the API.

GRID-IN runs as a cloud service or on the distribution company’s server. Exchange with other systems is via a REST API: data upload, calculation start, targets, visit outcomes.