add minimal power for deferrable loads

[scruysberghs]

This would be handy for EV charging. For most cars and charging stations 6A is the minimal power you can charge your car at. If it is a 1fase (230V) system this means 1440W, if it is 3 fase (3x230V) it will be about 4200W. This is an industry standard to keep people from charging their car with too low a power settings that will cause a lot of inverter (AC/DC) losses.

Now you can only set a nominal_power_of_deferrable_load and a number of operating_hours . To plan car charging it would be nice to set the minimal power required to get the car to start charging.
I know a can look at the deferrable load sensor that emhass calculates and start charging if the power is above the minimal limet, but that way emhass prediction's can be off as I am not able to follow them to the W.

What I want this for is a form of opportunistic car charging plan I'm settings up. I do not need a certain amount of kwh's fed to the car but the car is sitting idle on the driveway connected to a charger and there are moments with solar surplus and/or very low dynamic prices.

I now use a template to calculate this opportistic charging potential and then set that a deferrable load once every 24 hrs or when the car is home and there is no specific charging goal.

Looks something like this

{%- set price_threshold = 0.15 -%}
{%- set grid_limit = 3000 -%}   {# Maximale grid-afname in W per halfuurblok #}
{%- set half_hour = 0.5 -%}     {# Elk blok duurt 0,5 uur #}
{%- set min_power = 1440 -%}    {# Minimale laadvermogen in W #}

{# Haal de forecast-gegevens op uit de 4 sensoren #}
{%- set grid_forecasts = state_attr('sensor.mpc_p_grid_forecast','forecasts') -%}
{%- set load_cost_forecasts = state_attr('sensor.mpc_unit_load_cost','unit_load_cost_forecasts') -%}
{%- set prod_price_forecasts = state_attr('sensor.mpc_unit_prod_price','unit_prod_price_forecasts') -%}
{%- set pv_curtail_forecasts = state_attr('sensor.mpc_p_pv_curtailment','forecasts') -%}

{# Bouw een lijst (ns.rows) waarin we per halfuurblok de data opslaan #}
{%- set ns = namespace(rows=[]) -%}
{%- set n = grid_forecasts | length -%}

{%- for i in range(n) -%}
  {%- set grid_entry = grid_forecasts[i] -%}
  {%- set dt = as_datetime(grid_entry.date) -%}
  {%- set hour = dt.strftime('%Y-%m-%d %H:00') -%}
  {%- set p_grid = grid_entry.mpc_p_grid_forecast | float -%}
  {%- set load_cost = load_cost_forecasts[i].mpc_unit_load_cost | float -%}
  {%- set prod_price = prod_price_forecasts[i].mpc_unit_prod_price | float -%}
  {%- set pv_curt = pv_curtail_forecasts[i].mpc_p_pv_curtailment | float -%}
  
  {# Bepaal de beschikbare power op basis van de voorwaarden: #}
  {%- set available_power = 0 -%}
  {%- if p_grid >= 0 and load_cost < price_threshold -%}
    {# Afname: beschikbare capaciteit is grid_limit minus de voorspelde afname #}
    {%- set available_power = grid_limit - p_grid -%}
  {%- elif p_grid < 0 and prod_price < price_threshold -%}
    {# Injectie: beschikbare capaciteit is de absolute waarde van p_grid #}
    {%- set available_power = p_grid | abs -%}
  {%- endif -%}
  
  {# Voeg evt. pv-curtailment toe als extra beschikbare energie #}
  {%- if pv_curt > 0 -%}
    {%- set available_power = available_power + pv_curt -%}
  {%- endif -%}
  
  {# Zorg dat negatieve waarden niet voorkomen #}
  {%- if available_power < 0 -%}
    {%- set available_power = 0 -%}
  {%- endif -%}
  
  {# Als het berekende available_power minder is dan het minimale laadvermogen, dan tellen we dat blok niet mee (0 Wh) #}
  {%- if available_power < min_power -%}
    {%- set energy_wh = 0 -%}
  {%- else -%}
    {%- set energy_wh = available_power * half_hour -%}
  {%- endif -%}
  
  {# Bepaal de kosten: als het blok via afname (p_grid >= 0) wordt gerekend, dan is de kostenfactor load_cost; 
     als injectie (p_grid < 0), dan wordt de prod_price gebruikt (dit kan een negatieve waarde geven, oftewel opbrengst) #}
  {%- if energy_wh > 0 -%}
    {%- if p_grid >= 0 and load_cost < price_threshold -%}
      {%- set cost = (energy_wh / 1000) * load_cost -%}
    {%- elif p_grid < 0 and prod_price < price_threshold -%}
      {%- set cost = (energy_wh / 1000) * prod_price -%}
    {%- else -%}
      {%- set cost = 0 -%}
    {%- endif -%}
  {%- else -%}
    {%- set cost = 0 -%}
  {%- endif -%}
  
  {%- set ns.rows = ns.rows + [ {
        'time': dt,
        'hour': hour,
        'energy': energy_wh,
        'available_power': available_power,
        'cost': cost
      } ] -%}
{%- endfor -%}

{# Toon per halfuurblok de berekende waarden #}
{%- for row in ns.rows -%}
  {{ row.time }}: available_power={{ row.available_power | round(2) }} W, energy={{ row.energy | round(2) }} Wh, cost={{ row.cost | round(2) }} €<br>
{%- endfor -%}

<hr>

{# Tel het totaal op over de komende 24 uur #}
{%- set ns_total = namespace(total_energy=0, total_cost=0) -%}
{%- for row in ns.rows -%}
  {%- set ns_total.total_energy = ns_total.total_energy + row.energy -%}
  {%- set ns_total.total_cost = ns_total.total_cost + row.cost -%}
{%- endfor -%}
total for next 24 hrs: {{ ns_total.total_energy | round(2) }} Wh, total cost: {{ ns_total.total_cost | round(2) }} €

resulting in

2025-03-05 20:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-05 21:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-05 21:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-05 22:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-05 22:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-05 23:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-05 23:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 00:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 00:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 01:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 01:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 02:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 02:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 03:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 03:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 04:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 04:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 05:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 05:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 06:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 06:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 07:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 07:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 08:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 08:30:00+01:00: available_power=269.84 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 09:00:00+01:00: available_power=579.16 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 09:30:00+01:00: available_power=985.67 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 10:00:00+01:00: available_power=1430.12 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 10:30:00+01:00: available_power=1944.32 W, energy=972.16 Wh, cost=0.04 €<br>2025-03-06 11:00:00+01:00: available_power=2511.79 W, energy=1255.89 Wh, cost=0.05 €<br>2025-03-06 11:30:00+01:00: available_power=3000.0 W, energy=1500.0 Wh, cost=0.14 €<br>2025-03-06 12:00:00+01:00: available_power=3000.0 W, energy=1500.0 Wh, cost=0.14 €<br>2025-03-06 12:30:00+01:00: available_power=3000.0 W, energy=1500.0 Wh, cost=0.14 €<br>2025-03-06 13:00:00+01:00: available_power=5192.96 W, energy=2596.48 Wh, cost=0.23 €<br>2025-03-06 13:30:00+01:00: available_power=6961.22 W, energy=3480.61 Wh, cost=0.31 €<br>2025-03-06 14:00:00+01:00: available_power=6634.09 W, energy=3317.05 Wh, cost=0.3 €<br>2025-03-06 14:30:00+01:00: available_power=6150.59 W, energy=3075.3 Wh, cost=0.34 €<br>2025-03-06 15:00:00+01:00: available_power=5599.6 W, energy=2799.8 Wh, cost=0.31 €<br>2025-03-06 15:30:00+01:00: available_power=1988.88 W, energy=994.44 Wh, cost=0.05 €<br>2025-03-06 16:00:00+01:00: available_power=1361.58 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 16:30:00+01:00: available_power=761.69 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 17:00:00+01:00: available_power=185.95 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 17:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 18:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 18:30:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 19:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br>2025-03-06 19:30:00+01:00: available_power=3804.63 W, energy=1902.32 Wh, cost=0.25 €<br>2025-03-06 20:00:00+01:00: available_power=0.0 W, energy=0.0 Wh, cost=0.0 €<br><hr>

total for next 24 hrs: 24894.04 Wh, total cost: 2.28 €

[scruysberghs]

what do you think @davidusb-geek ? Or do you see a way to do this with just the nominal power and running hours parameters for deferrable loads we have now?

[davidusb-geek]

Yes we will implement this. It has been an open feature request since 2023. Its #129. After the current PR on forecast adjustment I could start working on this.

[scruysberghs]

nice, let me know if I can help, also with the PV forecasting model. I have some sites with verry special shading paterns during the day that would be great to test this feature.

Was thinking something similar than minimal charging power for ev deferrable loads might actually be needed for p_batt_forecast. If you look at graphs showing efficiency (AC/DC conversion) it drops off dramaticly in the lower power ranges, anything below 15% of inverter max gives you losses of 15% (up to 40%) with most inverters.
Has battery efficiency depending on charging/discharging power (in percentage from max) ever been considered as a feature? Might not be so easy to implement as hybrid inverters don't have those losses when charging from solar (DC/DC).
Typical graph:

[davidusb-geek]

Variable efficiency will be indeed interesting. However not so easy to implement on MILP formulation, we would need to use piecewise linear functions. It's a possibility but not currently in the roadmap. We could countour this with just a minimum charging power value. The day we do this for deferrable the same approach can be used for battery power.

[anselmhuewe]

why not using the historical power generation of the PV system, map them against the historical prediction and adjust future predictions based on the historical error on an hourly bases? Thats what I do with www.shyft-power.com, and it works pretty well, especially for "complicated" PV systems.

[davidusb-geek]

That's exactly what it's been cooking here PR #476