Skip to main content

AI energy manager

Updated
⚡ Reduxi Configurator · User manual

AI energy manager

Let the Reduxi AI agent plan the battery, PV and EV charging from electricity prices.

⏱ approx. 10 minutes
📍 Strategies → Energy manager → Optimization modes
👤 Installer
← Negative price optimization☰ Manual contentsBattery, EVSE and HVAC optimization →
📋 In this article
  1. Overview
  2. Before you start
  3. Add the AI energy manager
  4. Parameters
  5. How it works with other modes
  6. Check the result
  7. Checklist

→ Overview

The AI energy manager hands the price optimization of a strategy to the Reduxi AI agent. The agent plans ahead from the electricity prices and sends the controller a schedule of time slots. For each slot the schedule can set:

  • the battery power;
  • the PV power, so the site does not export or produce at negative prices;
  • the EV charging power (beta, one EV charger per site);
  • the strategy's Goal, Import limit and Export limit.

Between slots, or when no schedule is available, the strategy runs on its normal optimization modes.

⚠️

Reduxi advanced feature

The AI energy manager is marked with * in the Configurator: "Reduxi advanced feature. Subscription may be required." The licensing terms are at https://support.reduxi.eu/hc/en-us/articles/50787492273681-Reduxi-licencing. Check the subscription before you promise savings to the site owner.

→ Before you start

  • Cloud connection — The AI schedule is calculated in the Reduxi Cloud and sent to the controller. The controller must be connected to the cloud. See Cloud connection.
  • Prices — The site needs a utility rate plan with buy and sell prices. The AI energy manager row is struck through when there are no prices for today. See Utility rate plan (electricity prices).
  • No specific price optimizations — Remove Negative price, Battery optimization, EVSE optimization and HVAC optimization from the strategy first. They cannot be used together with the AI energy manager.
Add dialog with the AI energy manager card greyed out and a message box reading: The AI Energy Manager and specific optimizations (Negative Price, Battery, EVSE, and HVAC) are mutually exclusive. You can only use one optimization mode at a time. A Close button closes the message; the advanced feature note with the Reduxi licencing link is shown at the bottom
Add dialog after Battery optimization was added: the AI energy manager card is greyed out and a message explains that the AI Energy Manager and specific optimizations are mutually exclusive

1 Add the AI energy manager

1
Open the strategy
In the left menu, click Strategies, then the Energy manager strategy. Open the Optimization modes tab.
2
Start a new mode
Click + below the table, then click Create next to AI energy manager. The editor opens on the Battery configuration tab.
3
Set up the battery
Keep Enable AI battery optimisation on Yes. Choose whether the battery may trade with the grid (Earning optimization (battery arbitrage)) or only charge from the grid or only discharge to the grid. Set the reserves under Advanced settings.
4
Check PV
Open PV configuration. PV optimization is always on; read the note there.
5
Set up EV charging (optional)
Open EVSE configuration and set Enable AI EVSE cost optimisation to Yes if the AI agent should plan EV charging.
6
Confirm
Click Confirm. The AI energy manager * row appears in the table.
7
Save
Click Save below the table. The strategy restarts with the new settings within a few seconds.
Edit dialog of the AI energy manager on the Battery configuration tab with the tabs PV configuration, EVSE configuration and HVAC configuration, Enable AI battery optimisation set to Yes with its explanation, Earning optimization (battery arbitrage) Enable set to Yes, the greyed Charging battery from grid and Discharge battery to grid cards, and Advanced settings with Min saving/earning per kWh 0.02 EUR/kWh, Capacity reserved to cover load 10 %, Capacity reserved for surplus (PV) production 10 % and the Target SOC range slider from 10 to 90
AI energy manager editor, Battery configuration tab: AI battery optimisation and earning optimization on, grid charging and discharging greyed out, min saving 0.02 EUR/kWh, 10 % reserved for load, 10 % reserved for PV, target SOC 10–90 %

⚠️

The AI agent overrides the goal and grid limits

While a time slot of the AI schedule is active, its goal, import limit and export limit replace those of the mode in effect, including Schedule entries and Voltage regulation. The main fuses, Grid connection fallback entries and a grid operator limit received over EEBUS still apply on top. The Apply strategy settings field on the Last measurements tab shows when this happens.

2 Parameters

Battery configuration

Parameter What it does Unit / default
Enable AI battery optimisation * Yes lets the AI agent set the battery power to save and earn money from the price differences. No leaves the battery to the strategy's normal modes and hides the battery settings below. Yes / No · Yes
Earning optimization (battery arbitrage) → Enable Yes: the battery charges at low prices and discharges at high prices, from and to the grid, to maximize earnings. While this is Yes, the two cards below are greyed out and both count as Yes. Yes / No · Yes
Charging battery from grid → Enable Only when arbitrage is No. Yes – charge at low prices from the grid and PV. No – charge from PV only. Yes / No · Yes
Discharge battery to grid → Enable Only when arbitrage is No. Yes – discharge to the grid at high prices. No – discharge only to cover the site's own load. Yes / No · No
Min saving/earning per kWh Minimum saving or earning per kWh of charged energy before the battery is used. It stands for the battery wear caused by each cycle; below this value the AI agent leaves the battery idle. Example from the tooltip: with 0.02 per kWh, a 100 kWh battery must bring at least 2 per full cycle. All prices of the plan, fees included, are taken into account and 8 % battery losses are assumed. The currency comes from the utility rate plan. currency/kWh · 0.02 for EUR, 0.2 for SEK, otherwise 0
Capacity reserved to cover load Part of the battery the AI agent does not use, kept for covering the site's load. Same value as the lower end of Target SOC range. % · 10 · 0–99
Capacity reserved for surplus (PV) production Part of the battery the AI agent leaves empty for surplus PV. Same value as 100 % minus the upper end of Target SOC range. % · 10 · 1–100
Target SOC range The state-of-charge band the AI agent works in. Drag the two handles; the fields above follow. Tooltip: "AI agent will keep battery SOC in desired range and allow for other modules to use the remaining battery capacity." % · 10–90

ℹ️

How the charge and discharge switches combine

Without arbitrage you choose one direction of grid trading:

  • Charging battery from grid Yes: the battery takes cheap grid energy and discharges only to the site's load.
  • Discharge battery to grid Yes: the battery charges from PV only and sells to the grid at high prices.

Exactly one of the two is Yes: switching one of them changes the other. Switching arbitrage off sets Charging battery from grid to Yes and resets Capacity reserved to cover load to 10 %.

PV configuration

PV configuration tab of the AI energy manager editor with the note that AI PV cost optimisation is always enabled, a greyed Enable AI PV cost optimisation switch set to Yes, and the explanation of how PV is limited at negative prices
AI energy manager editor, PV configuration tab: AI PV cost optimisation is always enabled
Parameter What it does Unit / default
Enable AI PV cost optimisation Always Yes and cannot be changed. The AI agent keeps PV at full production while the price is positive. If the buy price is positive but the sell price is negative, PV is limited to the site's own load so nothing is exported at a loss. If the buy price is also negative, PV is stopped. Some PV inverters cannot be controlled externally. Yes (fixed)

💡

PV that must never be limited

The note on the tab says: if negative price limitation is not required, set the PV power plant to manual mode with maximum production. Do that on the PV device's control card. See Control type and manual setpoints.

EVSE configuration

EVSE configuration tab of the AI energy manager editor with the Enable AI EVSE cost optimisation switch set to No, the bold note that this is a BETA premium feature supporting only one EV charger per site, the explanation of cost optimised charging, and the advanced feature note with the Reduxi licencing link
AI energy manager editor, EVSE configuration tab: Enable AI EVSE cost optimisation set to No, with the beta note
Parameter What it does Unit / default
Enable AI EVSE cost optimisation * Yes: the AI agent picks the cheapest time to charge the connected vehicle. No: the vehicle charges at the highest available power as soon as it is plugged in, unless it is limited elsewhere (vehicle settings, other strategies). Beta feature; supports only one EV charger per site; the mobile app is recommended for the best experience. Yes / No · No

HVAC configuration

The HVAC configuration tab shows only Basic settings with the text "Coming soon ...". The AI energy manager does not control heat pumps yet; they follow the strategy's normal HVAC settings.

HVAC configuration tab of the AI energy manager editor showing a Basic settings card with the text Coming soon
AI energy manager editor, HVAC configuration tab: Coming soon
Parameter What it does Unit / default
Confirm Closes the editor and adds or updates the AI energy manager * row. Nothing is saved yet. —

AI energy manager row in the table

The row shows the name AI energy manager * and an edit and a bin icon. Below the table the note "* Reduxi advanced feature. Subscription may be required." appears with the licensing link. The name is struck through, with an info icon, when the utility rate plan has no prices for today.

3 How it works with other modes

  • The strategy first picks the mode in effect as usual: Voltage regulation, then the first matching Schedule entry, then the Default mode. An active AI time slot then replaces the goal and the grid limits of that mode, and sets the battery, PV and EV charging power.
  • If the AI schedule has no slot for the current time, for example because the controller has been offline from the cloud for a long time, the mode in effect runs unchanged.
  • External EMS setpoints win over AI setpoints for the same device type, and a Flexibility activation wins over both for the battery. See Flexibility and External EMS.
  • Negative price, Battery optimization, EVSE optimization and HVAC optimization cannot be used at the same time. The Add dialog greys out the conflicting cards and shows a message when you click Create.

4 Check the result

On the strategy's Last measurements tab, the AI agent card shows:

  • Enable — ON when an AI mode is configured;
  • Battery active power setpoint, EVSE active power setpoint, HVAC state setpoint — the values of the current slot;
  • Apply strategy settings — whether the current slot also replaces the goal and the grid limits.

5 Checklist

  • The subscription for the AI energy manager is in place.
  • The controller is connected to the Reduxi Cloud.
  • The utility rate plan has prices; the AI energy manager * row is not struck through.
  • Target SOC range leaves enough battery for the site's load and for surplus PV.
  • Min saving/earning per kWh matches the battery's wear cost.
  • Enable AI EVSE cost optimisation is Yes only if there is exactly one EV charger.
  • After saving, the AI agent card on Last measurements shows setpoints.
← Negative price optimization☰ Manual contentsBattery, EVSE and HVAC optimization →