You already control your heat pump. Now let it think ahead.
Asgard gives you full local control. ODIN adds the brain: an on-device Model Predictive Control (MPC) engine that rebuilds an optimized 48-hour plan every hour, combining day-ahead electricity prices, solar & weather forecasts, and a physics model of your house — learned from your own measurements.
REAL HOUSE PHYSICS
ODIN learns your home's heat loss, thermal mass and passive solar gain from real measured data — a physical model of your house, not a timer or a generic profile.
SET & FORGET
Set your comfort band once (e.g. 20.5–22.5 °C). ODIN plans every hour of heating, cooling and hot water within it — automatically, day after day. No daily fiddling.
100% LOCAL
All computation runs on the device. No cloud, no account, no subscription — your data never leaves your home.
Measure. Learn. Forecast. Plan.
Measure
Asgard reads your heat pump's real consumption, output, runtime and temperatures — every day.
Learn
ODIN derives your house's thermal time constant, heat loss and passive solar gain from that data. The model self-corrects across seasons.
Forecast
Day-ahead prices and a 48-hour weather & irradiance forecast are fetched (or pushed locally from Home Assistant).
Plan
Every hour, the solver rebuilds a cost-optimized plan for heating, cooling and hot water — always within your comfort band.
Grounded in peer-reviewed research
A four-month field study published in Applied Energy (2023) tested the same class of Model Predictive Control on a real, occupied family home with a heat pump. Shifting energy demand away from expensive evening price peaks reduced heating costs by up to 17%[1] — under winter-only conditions and with significant hardware limitations.
Where the study stopped
- Heavy control delays: the setup could not start the heat pump on demand — 60–120 minutes of physical lag.
- No hot-water integration: the expensive auxiliary immersion heater fired at random moments during price spikes.
- Winter only: tested November–March, missing all solar gains of the shoulder seasons.
Where ODIN goes further
- Instant compressor execution via Asgard — ODIN can act on any hourly price window.
- Integrated DHW scheduling locks hot water into the cheapest hour first, avoiding aux-heater fallbacks.
- Year-round solar optimisation steers runtime into forecast irradiance windows in every season.
Demand shifts away from evening price peaks using the building's thermal mass. Demonstrated on a real occupied home with floor heating over four months — the range depends on the comfort level chosen, despite the control delays and missing DHW integration described above.[1]
What load shifting can achieve when the controller acts instantly: for a heat pump with water-based floor heating and price-optimized MPC, Halvgaard et al.'s simulation reports "the optimized operating strategy saves 25–33% of the electricity cost" versus constant-price operation.[2]
In six 120-hour test-rig trials, Kuboth et al. measured an average 34.0% reduction in heat pump operating costs (up to 51.8% in the best trial), with PV self-consumption raised by an average 234.8%.[3]
Including the options that are cheaper than us
To be clear: none of these offer the same thing. EMHASS comes closest in spirit (local optimization, but generic), and commercial cloud subscriptions optimize to a certain extent (but off-site, by subscription). A BMS interface doesn't optimize at all. The table shows what each route realistically costs to get some form of smart control.
| Option | Per year | 5-year total |
|---|---|---|
| ODIN — add-on for Asgard The complete optimizer on top of the Asgard interface you already need for local control. Set your comfort band — the system handles the rest. 1-yr hardware warranty. | €0 | €199 |
| Cloud subscription services Commercial smart-control subscriptions — they optimise to a certain extent, but in the cloud, with an account, and costs that never stop. No learned house-physics model running in your own home. | €50–80 | €350–650 |
| Commercial BMS route The official Modbus/BMS interface (~€215) is an interface only — no optimizer. Add a BMS/PLC plus integration hours on top. | — | €650+ |
| Full DIY (HA + EMHASS) The closest alternative in spirit: free, local optimization. But EMHASS is a generic load scheduler with a simplified thermal model — no learned house physics, no direct DHW/compressor control — and a lot of hours of setup and tuning. | your time | €25 + hours |
Do the math with your own numbers
We won't promise you a saving — it depends on your contract, climate and house. So pick your own assumption and see what €199 means:
…so the €199 pays for itself in ≈ 0.4–1.7 years
What affects your savings
- Meaningful savings require a dynamic price contract; on a flat tariff ODIN mainly optimises solar self-consumption and comfort.
- ODIN is an add-on: it requires Asgard hardware and cannot run standalone.
- The physics model needs ±1 week to learn your house — the first days are not optimal.
- Savings scale with your comfort band: a wide band (e.g. ±1.5 °C) gives the solver room to shift load; a very tight band leaves little to optimize.
- Solar and weather forecasts are estimates; ODIN mitigates with safety margins, not magic.
- Home use only — no commercial deployments.
- All percentages are from published studies (field, simulation and lab test rigs) — not a guarantee for your house.
ODIN — €199, once.
Introduction price. No subscription, no cloud, no daily fiddling: set your comfort band and let the solver work.
Order ODIN — €199Requires Asgard. Questions first? Ask in the GitHub discussions.