The day the lights went out — and why SAMST exists

Every company has a “why”. Ours has a date.

In June 2024 a regional blackout paralysed the Balkans. Power went out across multiple countries at once. And during the crisis, the region was essentially flying blind — nobody could say exactly how or where the failure had started until long after it was over.

That is the part worth sitting with. Not that a failure occurred; failures occur. But that a modern interconnected transmission system could fail, and the people responsible for it could not immediately answer the most basic question: where?

The blind spot has a geography

Montenegro’s transmission network runs through the Dinaric Alps. The terrain is spectacular and, from an infrastructure standpoint, punishing: remote, mountainous, frequently inaccessible in winter, and almost entirely without reliable cellular coverage.

Inspecting a tower there is a day’s work. Inspecting all of them is a programme. Doing it often enough to catch a fault in the window between “beginning” and “cascading” is not realistic with crews and vehicles, no matter how good the crews are.

So the towers sit unobserved, and systemic failures hide in precisely that gap.

What we decided to build

SAMST — Smart Audit & Monitoring System for Transmission Towers — started from a narrow question: what would it take for the tower itself to report its condition, from a location with no power and no signal, and keep doing it through a winter?

That constraint drove almost every design decision that followed:

• LoRaWAN, not cellular. Kilometre-scale range at milliwatt power budgets. A node survives on battery where there is no mains supply, and keeps reporting during the cellular outages that tend to accompany exactly the events you want visibility into.
• Retrofit, not rebuild. Sensors mount onto towers that are already standing. No structural modification and no outage window, because a solution requiring either would never reach the assets that need it.
• Baselines per asset, not global thresholds. A tower on an exposed ridge and a tower in a sheltered valley do not have the same normal. An alert is only useful if it means this asset has changed.

From “what happened” to “what is about to happen”

The point was never forensics. Knowing precisely how the 2024 blackout propagated would have been valuable, but the far more valuable thing is an early warning triggered while the failure is still a deviation rather than an event.

That is the shift we are trying to deliver: moving operators from what happened? to what is about to happen? — with enough lead time to do something about the answer.

We are not building a system. We are building the observability layer that national energy security turns out to depend on. By the time the next extreme weather event arrives, the grid should be able to talk back.