[rrd-users] Handling solar inverter data in rrdtool ?

Joel HATSCH j-ml at joel-hatsch.net
Fri Jul 6 00:50:48 CEST 2012

On Monday 02 July 2012 08:20:16 Simon Hobson wrote:
> Joel HATSCH wrote:
> >My inverter (SMA) records every 5 minutes the instantaneous power and the
> >overall yield (both in kW (resp kWh), float with 3 decimal places).
> I'd say use the counter. I assume the instantaneous reading is
> actually for the instant read rather than an average over the
> previous 5 minutes and so may well be significantly different.Long
> term it will probably average out and be fairly accurate, but short
> term may be way out.
> Eg, if a cloud passes over the panels just as you take a reading,
> then you may get a low figure when output is actually high.
> Similarly, if a break in the clouds passes just as you take a
> reading, you'd get an artificially high figure.
> Since you aren't likely to want anything other than kWH in your
> output, I'd suggest just storing that. Internally it will be stored
> as KWH/s (all things are stored as rates) and so on output you'll
> need to multiply by your time span to get a total amount of energy.
> Eg, if doing daily figures, multiply by 86400 and you'll kWH/day.
> Also, since your inverter is keeping the counter, you could afford to
> increase the heartbeat considerably. If you fail to collect data for
> a while, then your next collection will result in a big chunk of
> lecky being counted but it will be spread evenly through the
> intervening data slots. You'll lose the detail of when it was
> collected (eg a couple of days would show as a flat line at whatever
> the average rate was) but your totals would be correct and after
> consolidation you'd probably not notice it on the "one period/day"
> long term graphs.

Thanks to you and Steve for the hints.

I knew I was missing something very obvious... data is stored as rates, 
therefore you must multiply it by the period of integration/time. Was still 
thinking in discrete data series inside my spreadsheet.

I gave it a try, works as expected. Now I just need to do some fine-tuning 
with the graphing parameters to get my reports.


More information about the rrd-users mailing list