Every breakthrough technology has an origin story, but few begin as a "contingency plan-Z." The I.C.E. (Integrated Cold Energy) Battery started that way — a theoretical pivot during a difficult period that turned into something real enough to get accepted into NUS Enterprise BLOCK71 at Singapore Science Park in June 2025.
The vision: affordable cold energy storage for Singapore's HDB public housing, where air conditioning accounts for a significant chunk of household energy costs.
From Heat to Cold
After our 2022 sand battery experiment and the subsequent US patent, a question kept nagging me: if we can store thermal energy as heat with 98% efficiency, why not store it as cold?
Most of us think about thermal energy in terms of direct consumption — burning fuel for heat or running air conditioners for cooling. But what if we could capture and store thermal energy at both ends of the spectrum?
The evolution from Sand Battery Prototype #0 to I.C.E Battery Prototype #0 — two sides of the thermal energy spectrum
Why Cooling Matters More Than You Think
Here's the urgency behind this project: the International Energy Agency projects that energy demand for space cooling could triple by 2050. We're already seeing it in Southeast Asia — as temperatures rise, more people buy air conditioners, which consumes more electricity (often from fossil fuels), which accelerates the warming that drives cooling demand. A vicious cycle.
For a city like Singapore, where it's 30°C+ year-round and HDB flats house 80% of the population, air conditioning isn't a luxury — it's a health necessity. And it's expensive.
The Contingency That Became a Company
In November 2024, as challenges with our original thermal storage work intensified, what I'd been treating as theoretical exploration became a practical lifeline. I bought the domain thermion.energy — partly strategic, partly an act of faith that thermal energy technology had legs beyond heat storage.
The technical breakthrough came from a fundamental insight: "storing cold" is really about removing heat and keeping it out. Unlike traditional cooling systems that consume energy continuously, the I.C.E. Battery charges during off-peak hours — using cheap, often renewable electricity to chill a storage medium to -22°C (-7.6°F) or lower — and then delivers cooling passively when needed.
No continuous compressor running. No peak-hour electricity bills. Just stored cold, released through heat exchange.
I.C.E Battery Prototype #1 — industrial scale pilot at Bien Hoa Farm, January 2025
The Bien Hoa Farm Pilot
Our first real-world test happened at a farm facility in Bien Hoa, outside Ho Chi Minh City. The application was agricultural cooling — maintaining consistent temperatures for temperature-sensitive crops and providing cold storage for harvested produce.
The monitoring infrastructure runs on the same HomeLab stack we use for everything else: Raspberry Pis, Node-RED, MQTT sensors, ZeroTier for cross-site networking, and Cloudflare Tunnel for remote access. The system processes over 10,000 data points daily — temperature, humidity, energy consumption, and system health metrics — all from a Pi sitting in the facility.
Where It Goes From Here
The I.C.E. Battery applications we're most interested in, beyond agriculture:
- Cold chain logistics: zero-emission refrigerated storage and transport, replacing diesel-powered units
- Off-grid cooling: solar-powered I.C.E. Battery systems for locations with unreliable grid access
- HDB housing: the original Singapore vision — affordable overnight-charged cooling for public housing units
The economics are compelling: charging during off-peak hours and delivering cooling during peak demand reduces electricity costs by 40-60%. The systems use water-based thermal storage instead of harmful refrigerants, and a 20+ year operational lifespan keeps replacement costs near zero.
For the full technical breakdown — the three-stage process, IoT architecture, and sensor specifications — see our companion technical post.
What started as plan-Z is now the plan. That's a lesson I keep relearning: the best ideas don't always announce themselves as such.