Hilo HA integration consumption consistently higher than Hilo app & Hydro Quebec website/app

[mshamash]

Hello,

I've been running the Hilo HA integration for several weeks now, collecting good data on my electricity usage.
I noticed that the Hilo consumption sensor sensor.hilo_energy_total_low (I haven't reached 'medium' usage yet, so cannot comment on this one...) is consistently higher than what is reported in the Hilo app & Hydro Quebec website/app for all days.
The Hilo app reports the same value I see with HQ app/website (rounded up usually).

See some examples below:

Date	sensor.hilo_energy_total_low	Hilo app	HQ app/website
Oct 19	11.58 kWh	10.3 kWh	10.33 kWh
Oct 18	7.82 kWh	6.4 kWh	6.36 kWh
Oct 16	7.76 kWh	6.5 kWh	6.46 kWh

Is this a known issue? Or any suggestions on possible fixes? I already checked that my HA timezone is correct, and set to Eastern time for my user & the entire HA setup.

Thanks in advance.

[ic-dev21]

This is the behaviour you should be seeing from the sensor: a reset at midnight.

Energy calculation in HA is done by Riemann Sum integration so it is approximate. This is how HA works and I have no power over that. I’ve argued the math with them once and it took them over a year to fix a part of it.

No issues here considering the approximation.

[mshamash]

Thanks for the information. Your values are a lot closer to the HQ app than mine, I'm usually off by over 1kWh. That's odd... In your case it seems to be <100Wh, not such a big deal.

Yes the sensor resets at midnight, see photo below.

[mshamash]

Could my greater inaccuracies also be due to different scan intervals? I am using the default of 60s, but I see in the documentation that I can go down to 30s, is that what you're using? I assume I will get more real-time data this way.

[ic-dev21]

I am using 60s. The meter doesn’t refresh any faster than that.

[JeanNoumsi]

Did you find why Hilo and HQ show the same consumption?

[ic-dev21]

Did you find why Hilo and HQ show the same consumption?

They should be pretty close, the Hilo component being an approximation.