DHW control parameters and logs and script and dataset
Various stats are gathered from the eddi (via the Internet-based API), including:
Heat-battery target and eddi state:
2024-11: days=1.5, heat battery mean kWh/d=3.5 (divert/d=0.0, boost/d=3.5).
2024-10: days=31.0, heat battery mean kWh/d=3.8 (divert/d=0.2, boost/d=3.6).
Recent month's daily inputs to the heat battery by divert from PV, and boost from grid:
While PV is producing, and at other times when the Enphase ACB is carrying load and balancing grid flow to close to zero, the eddi reports up to ~90W import. Standing watching the LED indicators on the import and export supply meters at ~9am today, they indicated slight export (or nothing) while the eddi display was showing intermittent imports of up to ~200W. So far I trust trust the accuracy of the eddi least out of it, the Enphase and the supply meters. Which is not to say that the eddi is horrible!
The eddi does not claim super-accuracy, and this is not a criticism of it! (Claimed measurement accuracy is ±1.5%.)
There was a small natural experiment while were were away for a fortnight. The calibrated utility and generation meters were read just before we left (morning of the 9th) and just after we returned (afternoon of the 24th).
(The eddi reported diversion to the Thermino of ~1kWh/d over this time.) Imports according to the supply meter for the entire period were ~0.1kWh. But the eddi claims imports of 6.9kWh from 10th to 24th inclusive (~0.5kWh/d).
The Enphase agrees with the supply meter at ~0.1kWh total for the core 14 days (112Wh):
% awk < net_energy_august_2022.csv -F, '/^2022-08-(10|11|12|13|14|15|16|17|18|19|20|21|22|23)/ {sum+=$5} END{print sum}' 112
Also, having just upped the Export Margin to 400W, the eddi seems to be regulating so as to allow >460W to be spilled .
Occasional observations during diversion (eg morning of 2022-09-03) reinforce the eddi claiming ~400W against the Enphase reporting ~450W+.
So there are some significant discrepancies to resolve...
YYYYMMDD.log
and frequency-response log records named YYYYMMDD.freqResp.log
YYYYMM.log.gz
, batched frequency response files named YYYYMM.freqResp.log.gz
, monthly by-day summaries of eddi energy sent to the heat battery by diversion and boost named YYYYMM.daily.csv