You are spot on…the challenge is that the generation value can vary enormously by the second if it is a cloudy day making the time-stamping a real challenge. On a sunny day, generation does not vary so the Power Used curve ends up being much smoother.
yes this seems like the real problem - the intervals are not necessarily the same for the net consumption data and the solar generation, and then the solar output can and does vary wildly between two of its own intervals. i guess in the end the power averages end up correct, so its ok, but the instantaneous energy data can look weird.
.