Three people, two systems, and a rule about numbers.
GreenTouch is a small cleantech research group at the University of Science and Technology of Mazandaran. We work on two problems a university building has at the same time: it draws more power than it needs at exactly the wrong moment, and it accumulates the air its occupants have already breathed. Neither system is deployed. This page says exactly how far along each one is.
Where this actually stands
Campus load model, forecast model and control logic are implemented and backtested against the simulation. Nothing is installed in a building.
Film deposited by sol-gel and characterised on our own apparatus. Room-scale figures are derived from a single reference run, not from a fielded trial.
Sensor nodes and the telemetry path exist as prototypes. The phase-change store, piezo tile and perovskite glazing are specified, not built.
Bench, modelled, design target, or published elsewhere. The tag sits next to the number, and the assumption behind it is printed beside it.
A building is measured constantly and asked almost nothing
Most campuses already meter their electricity, and many log CO₂ somewhere in the building management system. The data exists. What is missing is any point at which something acts on it while acting is still useful.
A demand charge is set by a single fifteen-minute interval in a whole month, and by the time it appears on an invoice the interval is weeks gone. Ventilation runs on a timetable written before the room was occupied. In both cases the measurement is downstream of the decision it should have informed, and the gap between the two is where the waste sits — not in the plant, which is usually doing what it was told.
GEST closes that gap by moving the decision forward two hours, which is roughly what a chiller and a thermal store need to respond. SABZPOOST closes it differently: instead of measuring, deciding and acting, it puts a surface in the room that reacts to whatever is in front of it, continuously, with no decision at all. One is a control problem and the other is a chemistry problem. They meet at the same building.
How we handle a number
The assumption travels with the number
A percentage with no baseline is not a result. Every quantity on this site carries the condition it was taken under — the room volume, the illuminance, the grid intensity, the modelled peak. If you disagree with an assumption you can replace it and recompute, which is the only useful kind of disagreement.
We publish what the systems do not do
SABZPOOST does nothing about particulate matter, and the page says so next to the figure for urban PM2.5. GEST removes about 8% of a modelled peak, not half of it. Stating the boundary is cheaper than having someone find it later, and it is the part that tends to be missing.
The residual is the interesting part
A forecast that is right is not information. The value of running a model against a building is the times it is wrong, and by how much, and whether that error keeps coming back in the same place — which is usually a fault rather than a modelling failure.
The three of us
Small enough that there is nothing between a figure on this site and the person who produced it. If something here looks wrong, one of us can go back to the run it came from — and would rather do that than defend it.
Mahan Imanian
Founder · systems and modelling
Builds the campus load model and the control logic that sits on top of it. Owns the question of whether a number on this site survives being checked.
Ali Panahandeh
Computer engineering · firmware and data
Sensor firmware, the MQTT telemetry path, and the training pipeline for the demand model. Keeps the clocks aligned so a 1 Hz series is actually 1 Hz.
Ashkan Amiri
Energy engineering · measurement
Thermal storage sizing and the bench work behind the photocatalysis numbers. Runs the flow-cell and writes down what it says, including when it disagrees with us.
| Field | GEST | SABZPOOST |
|---|---|---|
| Time-series forecasting | yes | no |
| Energy engineering | yes | no |
| Building services | yes | no |
| IoT instrumentationshared | yes | yes |
| Academic researchshared | yes | yes |
| Indoor air quality | no | yes |
| Materials science | no | yes |
| Photocatalysis | no | yes |
Fields the two systems sit across
- Photocatalysis
- Materials science
- Time-series forecasting
- IoT instrumentation
- Energy engineering
- Building services
- Indoor air quality
- Academic research
Neither system belongs to one discipline, which is the reason a group this size can work on both. The forecasting and the photocatalysis share almost no theory, but they share a building, a measurement culture, and the same three people.
Read the two in detail
GEST
Campus energy intelligenceGEST reads 312 metering and environmental points once a second, runs a demand forecast two hours ahead of the campus, and spends that lead time taking about 20 kW off a 257 kW peak — by easing lighting and cooling, and by discharging cooling banked overnight in a phase-change store.
Open GEST →SABZPOOST
Photocatalytic air chemistrySABZPOOST is a 20-nanometre doped anatase film applied by sol-gel. Under ordinary interior lighting it generates hydroxyl radicals at the surface and oxidises volatile organics where they touch it, taking a 300 m³ room from 500 to 300 ppb over four hours with no fan, no filter, and nothing to replace.
Open SABZPOOST →What would make this stronger is a real building
The gap between a simulated campus and a metered one is the whole remaining problem. If you run an estate, a lab, or a building services group and would consider a measurement-only pilot — no control authority, just instrumentation and a forecast we can be wrong in front of you — that is the collaboration we are looking for.