The main objective of AI strategies is to maximize long-term savings and earnings.
The AI strategies use state-of-the-art machine learning and AI methods to control the devices connected to the Reduxi controller by considering electrical energy prices. For example:
- they control batteries (BESS) to charge in times of low prices and discharge in times of high prices
- they control electrical vehicle chargers (EVSE) to charge with high power when the prices are low and with low power or not at all when the prices are high
- they control the heat pumps and increase the heat production in times of low prices
- the strategies always provide peak shaving in order to keep the load or production within limits
In order to maximize the long-term profit of the user, an AI strategy must consider many specifics of energy devices and the grid. It considers:
- import and export power limits from/to the grid (to achieve peak shaving)
- degradation of the battery
- loss of battery charge/discharge cycle
- charging specifics of electrical vehicles
- delays in heat pump response
- etc.
The AI strategies are based on energy prices including the supplier and network fees. More information about price lists configuration can be found here: Reduxi price lists.
All the details on how to set up an AI optimization strategy can be found in this article: Strategies for Reduxi controller
AI strategy for battery management
The main specific of the AI battery strategy is to charge the battery in periods of low prices and discharge in periods of high prices. At the same time, it provides peak shaving which assures that the import and export limits are within limits (fuse limits, power limits, or any other power on the grid).
The most common implementation is to use a battery with dynamic (day-ahead) prices. The AI strategy makes a profit from the volatility (i.e. the spread) of the day-ahead prices. More than 90% of days allow us to make earnings just based on the daily price differences. Even more, many days provide two or even three chances to earn from the price difference. On average one can earn around 100 EUR per day for a 1 MWh battery.
An example of a charge/discharge cycle is shown below. The first charging cycle indicated in blue corresponds to low prices. The second one corresponds to the high prices. The earnings in this cycle were (TBD). As we can see the charging/discharging limits was respected at 200 kW.
Other AI strategies
- AI strategy for EVSE
- AI strategy for HVAC
- AI strategy with weather forecast
The description other AI strategies is yet to come. Stay tuned.