Skip to content

Load modeling

The Load device composes a Node (power sink only) with an implicit Connection to model power consumption based on forecast data. The connection includes a power limit segment.

Model Elements Created

graph LR
    subgraph "Device"
        SS["Node<br/>(is_source=false, is_sink=true)"]
        Conn["Connection<br/>{name}:connection<br/>(power_limit)"]
    end

    Node[Connection Target]


    Conn -->|linked via| SS
    Node -->|connects to| Conn
Model Element Name Parameters From Configuration
Node {name} is_source=false, is_sink=true
Connection {name}:connection power-limit segment values

Devices Created

Load creates 1 device in Home Assistant:

Device Name Created When Purpose
Primary {name} Always Load consumption tracking

Parameter mapping

The adapter transforms user configuration into connection segments:

User Configuration Segment Segment Field Notes
forecast PowerLimitSegment max_power_target_source Maximum consumption at each time
curtailment (shedding) PowerLimitSegment fixed True when curtailment is disabled (fixed demand)
connection Connection source Node to connect from
— PowerLimitSegment max_power_source_target Set to zero to prevent reverse flow
— Node is_source=false Load cannot provide power
— Node is_sink=true Load consumes power

Sensors Created

Load Device

Sensor Unit Update Description
power kW Real-time Power consumed by load
power_possible kW Real-time Maximum possible load (forecast)
forecast_limit_price $/kWh Real-time Marginal cost of serving this load

See Load Configuration for detailed sensor and configuration documentation.

Configuration Examples

Variable Load (Forecast)

Field Value
Name House Load
Forecast sensor.home_power_forecast
Connection Home Bus

Constant Load

Field Value
Name Base Load
Forecast 2.5
Connection Home Bus

Typical Use Cases

Whole-House Consumption: Use historical data or forecasting services to predict total home power consumption. Enables optimizer to time battery discharge and grid import optimally.

Constant Base Load: Model always-on consumption (refrigerator, networking equipment) with a fixed power value.

Scheduled Loads: Model predictable loads like pool pumps, HVAC, or EV charging with time-varying forecasts.

Physical Interpretation

Load represents power consumption that the system can choose to satisfy up to a forecast limit. When curtailment (shedding) is disabled, the forecast is enforced exactly. When enabled, the optimizer may shed the load if that reduces total system cost.

The model represents average power within each optimization period. This means reduced power can be interpreted as partial operation in whatever way fits the physical device (duty cycle, throttling, staging, etc.).

Configuration Guidelines

  • Forecast Accuracy: Critical for optimization quality. Underestimating causes real system to import more than planned. Overestimating may cause infeasibility. See Forecasts and Sensors.
  • Constant vs Variable: Use constant values for stable always-on loads. Use forecast sensors for time-varying consumption patterns.
  • Multiple Loads: Create separate Load elements for different consumption categories (base load, HVAC, EV charging) to track them independently.
  • Fixed Power: When curtailment is disabled, consumption equals the forecast exactly. When enabled, consumption may be reduced below the forecast based on economics and constraints.

Next steps