Installed, measured, still running.
Some of these sites have been reporting every night for a decade. What follows is what we can publish: the size, the conditions and the numbers we measured — without the names, which we give on request.
Customer and site names are not published on this site. References for a comparable project are available on request.
Three adaptive installations with published figures.
All three were measured against the same LED luminaires running on an astronomical clock — not against the old discharge lamps. The results were published in Sensors in 2024 with the University of Bologna and CNIT.
−51 %
A main urban road measured over sixteen months of continuous operation — the longest of the three measurement periods.
Read the case study → Main roads · Vietnam−48 %
878 luminaires over six months, against a baseline of 693 kWh a day on the same LEDs.
Read the case study → Residential · Italy−24.6 %
2,957 luminaires on M4/M5 residential roads — the honest figure where the class is already low.
Read the case study →Where the devices are.
Most references stay anonymous — these are the families of site.
Municipal networks
Public tenders won in the Rhineland and in Hesse: city-wide lighting control with adaptive stretches on the main roads, and cabinets monitored from the same platform.
Provincial roads and new districts
Large single-stretch installations where the adaptive saving is measured against the newly installed LEDs themselves, and reported monthly.
Cities, campuses and large sites
Residential networks, industrial yards, car parks and airport landside areas, with feeder pillars under control alongside the luminaires.
Installations we can name.
We normally keep customer names out of our material. These are already public — published by us or by the partner who delivered them — so they can be written here, with what is actually in them.
Pero, Melzo, Cantello
Three municipal networks, more than 5,000 luminaires between them, with traffic and weather deciding the lighting class on the roads that carry the traffic — six adaptive areas in Pero alone. Reported at −73 % to −76 % against the installations they replaced.
Tan An
More than 3,000 luminaires on one network. The adaptive stretch of it — 878 luminaires on the main roads — is the installation measured at −48 % against the same LEDs over six months and published in Sensors.
The case study →Essen
Solar lighting in a public park: poles with their own photovoltaic column and battery, carrying Wi-Fi access points and sensors. No grid connection, no trench — and the lighting still under control.
Fulda
A whole city running adaptively, in sixteen projects. Measured over four months in 2025: −58 % against full-power LED, −37 % against a fixed 40 % night profile, −79 % against the lamps that were there before. Platform operated by us, hosted in Germany, data owned by the city.
The case study →Johor Bahru
More than 4,000 luminaires, and feeder pillars monitored on the same network: certified meters, phase and protection alarms, and the section affected by a failure identified without a van.
Around a hundred more installations — Italian municipalities, German cities, Malaysian and Vietnamese towns, industrial parks, campuses and campsites — stay unnamed. We will put you in touch with one comparable to your own.
Ask for a reference →
Johor Bahru — feeder pillars under control, in the platform
The lighting network carries the cabinets too.
In Johor Bahru the same field network that controls the luminaires also reads the feeder pillars: energy per line from certified meters, voltage and current per phase, and an alarm when a protection trips — which turns “a district is dark” into “this cabinet, this line, at this time”.
Cabinet & feeder-pillar control →Two baselines, one story.
You will see two families of figure in our material: around −75 % in some projects, and −51 %, −48 %, −24.6 % in the published ones. They do not contradict each other — one contains the other.
Replacing the old lamps: about half
Swapping discharge lamps for LED roughly halves the consumption. That part is not our technology — any competent LED retrofit does it, with or without a control system.
Running them adaptively: about half of what is left
This is the part that is ours, and it is the one we measured and published: against the same LED luminaires on an astronomical clock, −51 %, −48 % and −24.6 % on three real installations.
Together: 70 to 80 % less than before
Half, and then half again, is a quarter of the original bill. A project reported at −75 % against the old installation and one reported at −51 % against LED can be the same project, counted from two different starting points.
When we quote a figure we say which baseline it uses. If a supplier shows you a percentage without one, the number tells you nothing — it may be measuring a lamp change you would have made anyway.
The published method →A saving nobody can check is not a saving.
Every adaptive project runs with a declared baseline, a measurement period and a method that a third party can repeat. When we cannot measure something, we say it is not measured.
The published method →Comparing adaptive operation with the discharge lamps that were there before mixes two different projects into one number.
Sixteen months, or six, or twelve — never a good week in October.
A saving only means something next to the lighting class the road is designed for.
We can put you in touch with a comparable site.
Tell us the size and the type of road, and we will arrange a reference with the operator who runs it.