Time to investigate the usage differences between the calculator and my bill during a few select periods. As boring as it is, I’m feeding my PG&E hourly data into my R calculator to see how the usage compares per TOU period and per billing period. It should be nearly perfect, rounding error excluded. As near as I can tell, PG&E exports power information with 2 decimal places of kWh accuracy, while they keep around 8 decimal places internally and in their calculations.
When I do the comparison, though, three billing periods have significant differences and the rest are near perfect. What’s more, is that even for the months with the big deviation between calculated and bill usage in some periods, the total usage still remains accurate. That means that the usage is misdistributed within those 3 billing months, but accurate for the monthly total.
Looking closely at the calendar and those three billing months, 2/21-3/21, 3/22-4/22 and 10/22-11/19, I can see a pattern. Daylight saving time 2019 began on 3/10 and ended on 11/3, right in the middle of two of the months with accuracy issues. The official start of the summer schedule for the EVA rate plan per the table is 5/1 and the official start of winter is 11/1, both again in the middle of billing periods that have issues.
The weekday rate grid from the OpenEI database - rows are months, columns are hours and the numbers correspond to TOU periods.
Guess I’m going to have to poke deeper into those billing months and the rate table…
Edit: OK - one quick answer. A clause buried deep in the tariff document for EVA and EVB rate schedules points out a non-standard time adjustment…
“DAYLIGHT SAVING TIME ADJUSTMENT: The time periods shown above will begin and end one hour later for the period between the second Sunday in March and the first Sunday in April, and for the period between the last Sunday in October and the first Sunday in November.”
So for those two periods following the advent of DST, and prefacing the change back to standard time, PG&E is really using a different table.
Arrrrgh ! I have just been reminded why TOU pricing is so hard.