One example, start to finish

See what we do

Below is a real study, the way we actually run one — the problem, the model, the results, the fix. Then tell us about yours.

Skip to the form
Example study

A 200 MW data center
meets the grid

In July 2024, roughly 1,500 megawatts of data-center load dropped off the Eastern Interconnection at once. A larger event followed in early 2025. Regulators investigated, and the finding surprised people: the grid did its job. A routine sequence of faults cleared normally. The data centers tripped themselves — their UPS systems transferred to backup on each voltage sag, and count-based logic locked the facilities off the grid until someone manually reset them.

Our example study reproduces that failure mode end to end for a 200 MW hyperscale facility interconnecting at 230 kV, sweeps the conditions that matter, and then demonstrates the fix — which turns out to be a settings change, not new iron.

The model, in one line each
load · 200 MW hyperscale data center
interconnection · 230 kV point of interconnection
transformer · 240 MVA, 230/34.5 kV
grid strength · swept SCR 2 → 10
protection logic · UPS transfer at 0.88 pu, 3-count lockout
pass/fail · demand-side ride-through envelope

How we run a study

01

Every assumption in one place

One configuration file holds every modeling choice — load, grid strength, transformer, thresholds, sweep ranges. Change it, re-run, everything updates.

02

The model is code

We build the simulation model programmatically — every component placed and wired in reviewable code, not clicked together in a GUI. Reproducible by anyone, defensible line by line.

03

Sweep what matters

Fault type, clearing time, reclose sequences, grid strength, load steps — the answer is a map of where you are safe and where you are not, never a single point.

04

We attack our own results

Independent hand calculations and an adversarial self-audit go after every claim before you see it. What survives is what we stand behind — and what a PE stamps.

The results that tell the story

Representative waveforms, re-simulated in our open Python pipeline from the study’s documented parameters.

Multi-shot reclose sequence: voltage sags and grid draw, with UPS lockout on the third shot
The failure mode, reproduced. Three fault shots, each cleared normally by the grid. The facility rides the first two — transferring to backup and returning — but the third strike trips the disturbance counter, and 200 MW locks off the grid until a manual reset.
Baseline versus reset-on-recovery mitigation: grid draw through the same three-shot sequence
The fix costs nothing. Same three-shot sequence, one settings change: the disturbance counter resets after each clean recovery. The as-found configuration loses the facility on shot three; the corrected one rides the whole sequence.
Fault ride-through voltage trajectories for 4, 5, and 6-cycle clearing against the demand-side ride-through envelope
Judged against the envelope. Every run is graded against a demand-side ride-through boundary — depth and duration, not vibes. Here, the clearing-time sweep shows how long the facility can afford protection to take.
POI voltage response to a 40 percent synchronized load ramp at SCR 2, 3, 5 and 10
AI compute moves in lockstep. A synchronized 40% ramp is nothing on a strong grid and a real event on a weak one. Sweeping grid strength shows exactly where the interconnection stops being boring.
P-V nose curves by grid strength with the 200 MW operating point marked
And the hard limit. Static voltage-stability nose curves by grid strength. At SCR 2 the nose sits below the 200 MW operating point — this facility does not belong on that feeder without support, and the study says so plainly.

The grid did its job

Every fault in the sequence cleared normally.

The load ended the story

Count-based lockout, exactly as documented in the real events.

The fix was a setting

Reset-on-recovery logic. No new equipment, no new iron.

Now, about your project

This goes straight to the whole team. We reply with an honest read — even if the honest read is that you do not need us.

monegrid ~ % start a study

or email founder@monegrid.com