Your Feeder Didn't Sign Up for This

Six months ago, that feeder was carrying a strip mall, two schools, and a church parking lot. Today it's got a 40-megawatt data hall going through commissioning, a cold-storage facility that just doubled its compressor bank, and a permitting office that approved both without asking anyone in operations. Commercial load growth doesn't wait for the next planning cycle. It shows up, energizes, and starts drawing current – and every megawatt of unplanned load is a megawatt your protection scheme wasn't coordinated for.

The Metric That Doesn't Care About Your Excuses

SAIDI and SAIFI don't grade on a curve for growth. Regulators don't discount your interruption minutes because the load materialized faster than your five-year capital plan anticipated. GridLab reports U.S. data center demand alone is projected to grow by nearly 10% annually, roughly doubling from 17 gigawatts in 2022 to 35 gigawatts by 2030, and increasingly that growth arrives concentrated on individual feeders rather than spread evenly across a service territory. Meanwhile, the U.S. distribution system's average SAIDI, even excluding major event days, climbed to 125.7 minutes per customer in the most recent EIA reporting year – the highest figure recorded in the past decade as reported in POWER Magazine. Reliability is degrading at the exact moment load is getting harder to forecast. That's not a coincidence you can plan around with a spreadsheet.

Speed Is the Variable Nobody Budgeted For

Traditional commercial load growth gave utilities years of lead time – permits, construction schedules, phased occupancy. Large modern loads don't play by that timeline according to Ascend Analytics. Data centers differ from traditional commercial or industrial load in three critical ways: speed, scale, and shape – load can materialize within 18 to 36 months instead of the 5 to 10 years utilities historically planned around. Compress the timeline and you compress the margin your protection engineers had to re-coordinate relays, resize CTs, and re-run fault studies before the load went live. Every shortcut taken under that compression shows up later as a SAIFI event.

The Instrumentation Problem Hiding Inside the Growth Problem

Here's the part that doesn't make the trade press headlines: rapid load growth is fundamentally a measurement problem before it's a capacity problem. You can't protect what you can't accurately see, and conventional current transformers were sized for yesterday's load profile. Push them past their design envelope and they saturate – the CT's magnetic core stops faithfully reproducing the primary current, protection relays see a distorted secondary waveform, and your fault detection either trips on phantom current or, worse, fails to trip on a real fault. Under normal load, saturation is a rare edge case. Under rapid commercial load growth – especially loads with harmonic-rich characteristics like variable-frequency drives, UPS systems, and switch-mode power supplies – it becomes a routine operating condition. A protection scheme that mis-times or mis-trips because its measurement layer lied to it is a self-inflicted SAIDI event.

What Actually Protects the Indices Under Pressure

Optical current and voltage sensing removes the saturation failure mode entirely – there's no magnetic core to drive into nonlinearity, because current is measured via the Faraday Effect and voltage via the Pockels Electro-Optical Effect, both light-based rather than electromagnetic. MICATU's Advanced Sensor Platform holds ±0.3–0.5% accuracy across the 4kV–72kV class regardless of how fast or how far the load ramps, sampling at 15,000 cycles per second with current harmonics resolved to the 11th and voltage harmonics to the 50th – visibility fine-grained enough to catch the harmonic signature of a newly energized commercial load before it becomes a coordination problem. Because the sensors are modular and mount directly on the conductor, they retrofit onto the feeder you already have – overhead, underground, or substation – without a system redesign, which matters when the load arrived faster than your capital plan did. Data flows via DNP3 into your ADMS, OMS, and FLISR systems, so the same instrumentation that protects the feeder also feeds the fault-location and restoration automation that keeps SAIDI minutes from stacking up when something does trip.

Coordination Windows Are Shrinking. Sensing Shouldn't Be the Bottleneck.

Reliability engineers already know the theory: resize the CT, re-run the coordination study, re-time the relays. What's changed is how much runway they get to do it. Ascend Analytics reports in ERCOT alone, 198 gigawatts of large load applied for interconnection in the first quarter of 2026, with 86 gigawatts under active review – a volume roughly equal to the region's entire current peak load. When the queue moves that fast, "we'll re-instrument next capital cycle" isn't a plan. It's a SAIFI event with a delivery date. 

 Rapid load growth is not going to wait for your next planning cycle. Neither should your sensing strategy – talk to a MICATU expert today.