Meter Data Turns Out to Be the Cheapest Way to Audit a Grid Model

A conference recap published on 27 February 2026 collects results from utilities well into distribution automation, and one of them reframes what advanced metering infrastructure is for. AEP Texas, working with Sensewaves, processes 105 million daily AMI data points across 1.1 million endpoints. The analysis uncovered 1,480 missing and 699 incorrect GIS relationships in the utility's own network model.

That is the interesting result because of what it implies about sequencing. A distribution utility's geographic information system is supposed to describe which customer is fed from which transformer, from which feeder, through which switch. Every advanced function — fault location, isolation and service restoration, voltage optimisation, hosting-capacity analysis — assumes that description is right. When meter telemetry disagrees with the model, the model is usually the one that is wrong, because the meters are physically where they say they are and the records were maintained by hand across decades of changes.

Meter data therefore audits the asset model more cheaply than any survey could, and the binding constraint on distribution automation turns out to be model integrity rather than sensor coverage. That reframes grid-edge work as reconciliation between AMI, GIS and the distribution management system, which is a data problem rather than a deployment one.

Other results in the recap point the same direction. SDG&E, working with Toumetis, used power-quality data sampled at 61,440 samples per second and reported 2.4 SAIDI minutes saved in 2024. DTE Energy presented a five-year automation roadmap that does not reach full automation until 2029, and the recap observes that most advanced distribution functions remain in advisory mode in utility control rooms today. ERCOT reported 65.2 MW deployed in its aggregated DER pilot with a further 124.5 MW in the pipeline.

The programme-scale notes are worth reading alongside the technical ones: Eversource's AMI deployment involves more than 300 people and over 20 vendors, and Exelon found that testing its distribution management system separately across four operating companies creates integration bottlenecks. Where these programmes stall is coordination, not hardware.