I feel like I shouldn't need a template sensor for a static forecast...

[ando2040]

I use a forecast chart to look for the best upcoming nights to charge an EV. For example I will try to run the battery low for Sunday night/Monday morning (11th) - when demand drops nearest to the renewables, carbon load will be lowest and in many cases I'll also get a cheaper electricity rate because my tariff tracks the wholesale price every 30 minutes and renewables are cheap...

The UK electricity system operator doesn't provide a forecast for nuclear, so I've synthesised one in a template sensor using the current value. The standard UK practice is that nuclear provides the base generation, and doesn't change very often (it's been pretty static the last 7 days so static is the best assumption for the next period.)

            {% set a = state_attr('sensor.national_grid_wind_forecast_fourteen_day', 'forecast') -%}
            {% set b = states('sensor.national_grid_grid_generation_nuclear_mw') -%}
            {% set ns = namespace(items=[]) -%}
            {% for i in range(a | length) -%}
              {% set start_time = a[i].start_time.isoformat() -%}
              {% set generation_total = b | int -%}
              {% set ns.items = ns.items + [{'start_time': start_time, 'generation': generation_total}] %}
            {% endfor %}
            {{ ns.items | list }}

As you should see, the template just copies the timestamps from the wind forecast and substitutes the current state of nuclear generation for every future value.

This feels a bit kludgey and non-optimal though it works very well as you see in the screenshot. It feels like I ought to be able to do this simple forward projection in a plotly function but I've had a play and I can't find a way to project the current value forward - should I keep trying or am I already at the best solution?

Thanks in advance for any guidance!

type: custom:plotly-graph
title: Long Range UK Generation Forecast
time_offset: 300h
layout:
  yaxis:
    title: GW
    range:
      - 0
      - null
defaults:
  entity:
    unit_of_measurement: GW
    filters:
      - resample: 1m
      - multiply: 0.001
    line:
      width: 0
  yaxes:
    fixedrange: true
entities:
  - entity: sensor.national_grid_grid_demand_fourteen_day_forecast
    name: |
      $ex `Demand (now) ${(ys.at(-1) / 1000).toFixed(2)} GW`
    yaxis: y1
    line:
      color: red
      width: 1
  - entity: sensor.national_grid_grid_demand_fourteen_day_forecast
    name: Fcst
    yaxis: y1
    showlegend: false
    line:
      color: red
      width: 1
      dash: 2
    filters:
      - fn: |-
          ({ meta }) => ({
            xs: [...meta.forecast.map(({ start_time }) => new Date(start_time))],
            ys: [...meta.forecast.map(({ national_demand }) => national_demand)],
          }) 
  - entity: sensor.national_grid_grid_generation_nuclear_mw
    showlegend: true
    fillcolor: orange
    stackgroup: 1
    name: |
      $ex `Nuclear (now) ${(ys.at(-1) / 1000).toFixed(2)} GW`
  - entity: sensor.national_grid_grid_generation_solar_mw
    showlegend: true
    fillcolor: yellow
    stackgroup: 1
    name: |
      $ex 'Solar (now) ' + (ys.at(-1)/1000).toFixed(2) + ' GW'      
  - entity: sensor.national_grid_grid_generation_wind_mw
    line:
      color: green
      dash: dot
      fillcolor: none
    stackgroup: 1
    name: |
      $ex 'Wind (now) ' + (ys.at(-1)/1000).toFixed(2) + ' GW'
  - entity: sensor.national_grid_nuclear_14_day_synthetic_forecast
    name: Nuclear projected from generation_nuclear_mw
    showlegend: false
    fillcolor: gold
    stackgroup: 1
    filters:
      - fn: |-
          ({ meta }) => ({
            xs: [...meta.forecast.map(({ start_time }) => new Date(start_time))],
            ys: [...meta.forecast.map(({ generation }) => generation)],
          }) 
  - entity: sensor.national_grid_embedded_solar_forecast_fourteen_day
    showlegend: false
    fillcolor: lightgoldenrodyellow
    stackgroup: 1
    filters:
      - fn: |-
          ({ meta }) => ({
            xs: [...meta.forecast.map(({ start_time }) => new Date(start_time))],
            ys: [...meta.forecast.map(({ generation }) => generation)],
          }) 
  - entity: sensor.national_grid_wind_forecast_fourteen_day
    showlegend: false
    fillcolor: lightgreen
    stackgroup: 1
    filters:
      - fn: |-
          ({ meta }) => ({
            xs: [...meta.forecast.map(({ start_time }) => new Date(start_time))],
            ys: [...meta.forecast.map(({ generation }) => generation)],
          }) 
hours_to_show: 330
refresh_interval: 10

[dbuezas]

Cool plot and use case!

You can use the xaxis of wind and a fixed value for nuclear:

  - entity: sensor.national_grid_wind_forecast_fourteen_day # note this is wind
    name: Nuclear projected from generation_nuclear_mw # note the name is nuclear
    showlegend: false
    fillcolor: gold
    stackgroup: 1
    filters:
      - fn: |-
          ({ meta }) => ({
            xs: meta.forecast.map(({ start_time }) => new Date(start_time)),
            ys: meta.forecast.map(() => hass.states["sensor.national_grid_grid_generation_nuclear_mw"].state * 0.001), 
          }) 

[ando2040]

Thanks! I was struggling because I started from the other entity - so obvious when you see it!

[dbuezas]

:)

[ando2040]

This is not a criticism, you essentially gave me the answer but I did need to add hass to the function declaration which I haven't done before so thanks again for helping me learn! I also have multiply: 0.001 in my entity defaults as per a previous optimisation discussion we had, so I removed that from the individual filter too:

  - entity: sensor.national_grid_wind_forecast_fourteen_day # note this is wind
    name: Nuclear projected from generation_nuclear_mw # note the name is nuclear
    showlegend: false
    fillcolor: gold
    stackgroup: 1
    filters:
      - fn: |-
          ({ meta, hass }) => ({
            xs: meta.forecast.map(({ start_time }) => new Date(start_time)),
            ys: meta.forecast.map(() => hass.states["sensor.national_grid_grid_generation_nuclear_mw"].state), 
          }) 

[dbuezas]

Oh, yes, sorry. Got too used to $ex that auto adds all params

Btw, I added the multiplication because I saw you had the scaling in a filter inside the defaults, and those should get overridden if an entity also has filters, so do doble check that!

[ando2040]

Interesting. It definitely works as shown with MW in the sensor but GW on the scale, and didn't show up visibly on the plot until I removed the multiply 0.001 - I assumed that was due to the defaults.

I didn't post the new chart because it looks exactly the same as the original, but using the more elegant config.