Template sensors to calculate inverter / battery losses?

[WyndStryke]

There are various losses in the system such as DC/AC conversions, round-trip battery efficiency, & also keeping the inverter itself running. Can these be calculated from the sensors that Modbus & integration provides?

Roughly speaking I notice around 80-120W usage seems to be unaccounted for, but this changes according to what the inverter seems to be doing (~70W in self-use mode, ~120W in force-charge mode even when at max SoC & therefore idle). This is on a KH7 with an ECS4300-H4 battery, and a small amount of winter PV, and very recent firmware.

For the instantaneous figure, if I sum pv1_power, pv2_power, pv3_power, pv4_power, grid_consumption, battery_discharge, and then subtract battery_charge, feed_in_power, and load_power. would that be accurate or am I forgetting anything? From the chart I can see that it sometimes flicks into negative loss for a moment so I'm not sure that I have it right (or perhaps that's just a timing artefact). Also I haven't taken changes in the battery SoC into account at all (since battery charging & discharging isn't 100% efficient).

For the daily figure, would it be better to use an integration on the above, or to calculate it from the _today sensors instead?

Or perhaps it might just be better to calculate it based on the datasheet efficiencies & throughputs (i.e., theoretical rather than measured)?

My end-goal is something along the lines of this Sankey chart ... it doesn't quite add up yet (not all CT clamps in place yet).

[FozzieUK]

There are various losses in the system such as DC/AC conversions, round-trip battery efficiency, & also keeping the inverter itself running. Can these be calculated from the sensors that Modbus & integration provides?

Roughly speaking I notice around 80-120W usage seems to be unaccounted for, but this changes according to what the inverter seems to be doing (~70W in self-use mode, ~120W in force-charge mode even when at max SoC & therefore idle). This is on a KH7 with an ECS4300-H4 battery, and a small amount of winter PV, and very recent firmware.

For the instantaneous figure, if I sum pv1_power, pv2_power, pv3_power, pv4_power, grid_consumption, battery_discharge, and then subtract battery_charge, feed_in_power, and load_power. would that be accurate or am I forgetting anything? From the chart I can see that it sometimes flicks into negative loss for a moment so I'm not sure that I have it right (or perhaps that's just a timing artefact). Also I haven't taken changes in the battery SoC into account at all (since battery charging & discharging isn't 100% efficient).

For the daily figure, would it be better to use an integration on the above, or to calculate it from the _today sensors instead?

FWIW this is what I use which (I only have 2 strings) but similar logic as you have used

      - name: "System Losses"
        device_class: "power"
        unit_of_measurement: "kW"
        state: >
          {% set sl = ((states('sensor.pv1_power') | float(default=0)
                      + states('sensor.pv2_power') | float(default=0)
                      + states('sensor.grid_consumption') | float(default=0)
                      + states('sensor.battery_discharge') | float(default=0)
                      - states('sensor.battery_charge') | float(default=0)
                      - states('sensor.feed_in') | float(default=0)
                      - (states('sensor.house_load_w')|int(default=0)/1000) )) | round(3) %}
          {% if sl > 0 %}
            {{ sl }}
          {% else %}
            {{ 0 | int }}
          {% endif %}

I have the negative check at the end or as you say the sampling artifacts can show a negative

I then integrate it and use a utility meter for the daily totals

[WyndStryke]

That's great, thank you :-)

I might also put in an upper bound to mirror the negative check, since I suspect those spikes might be artefacts too.