Wrong data record and uniform value in the netCDF files

[reconjohn]

I tried to plot some of the nc files on the date Jan. 01, 2019. For example, below figures are for the meters LOWER RANGE and SRI ANTENNA. I found some of files recorded wrong and some files has all the same value in them. There seems to be peak demand of (kW), a 15 minute average of one minute windows (in figures below refer to name Speak and Sliding window. But when I plotted, I'm not sure if they are. The first row is for the Lower range and the 2nd row, Sri antenna.

We need to address first, the clear understanding of the variables (definitions if any), what variable (delivered energy, net energy, active power, voltage, and so on) we are going to use, and what resolution (minute, hour, day, week, month).

[reconjohn]

The file names as below for the plots above.

• "PokerFlatResearchRange-PokerFlat-PQube3EDelTot@2019-01-01T090006Z@PT23H@PT152F.nc"
• "PokerFlatResearchRange-PokerFlat-PQube3PTot@2019-01-01T090006Z@PT23H@PT152F.nc"
• "PokerFlatResearchRange-PokerFlat-PQube3SPeak@2019-01-01T090006Z@PT23H@PT152F.nc"
• "PokerFlatResearchRange-PokerFlat-PQube3ENetTot@2019-01-01T090006Z@PT23H@PT152F.nc"
• "PokerFlatResearchRange-PokerFlat-PQube3PPeak@2019-01-01T090006Z@PT23H@PT152F.nc",
• "PokerFlatResearchRange-PokerFlat-PkFltM1AntEDelTot@2019-01-01T091006Z@PT23H@PT140F.nc"
• "PokerFlatResearchRange-PokerFlat-PkFltM1AntPTot@2019-01-01T091006Z@PT23H@PT140F.nc"
• "PokerFlatResearchRange-PokerFlat-PkFltM1AntSlidingWindowRealPowerDemand3Ph@2019-01-01T091006Z@PT23H@PT140F.nc"
• "PokerFlatResearchRange-PokerFlat-PkFltM1AntENetTot@2019-01-01T091006Z@PT23H@PT140F.nc"

[aidowu]

Is Speak in the first row same as the 15 minute sliding window?

[reconjohn]

Yes as per the description in the channel XL file.

SlidingWindowRealPowerDemand3Ph	(kW) is a 15 minute average of one minute windows.  See SW Sub-Interval Length (60s) and SW Sub-Interval Count (15).	SlidingWindowRealPowerDemand3Ph@2017-10-29T081007Z@PT23H@PT142F.nc	SlidingWindowRealPowerDemand3Ph@2017-10-29T082007Z@PT23H@PT141F.nc	SlidingWindowRealPowerDemand3Ph@2017-10-29T083011Z@PT23H@PT141F.nc	SPeak@2017-10-29T080006Z@PT23H@PT151F.nc

[aidowu]

Why is the Speak already multiplied by time? Shouldn't it be in KW like the SlidingWindow... in the plot below?

[reconjohn]

Questions I have for the meeting today, Apr. 26, 2019.

In order of Ant, Tel, Sci and PQube. Something seem wrong with PQube: DelTot, NetTot, PPeak, PTot and SPeak.

• why negative and uniform value compared to the other 3 meters in DelTot?
• why the value so small compared to the other 3 meters in NetTot?
• what is PPeak? It seems to be wrong indexed. In the index description, PPeak is aligned with PfTot which is Power Factor Total in the description.
• more clear explanation for SlidingWindowRealPowerDemand3Ph
• what is SPeak? It is in kVA. According to the description, it is along with SlidingWindowRealPowerDemand3Ph, but its trend shows different.

Comparison of the 4 meters (Ant, Tel, Sci and PQube)

Comparison of different dates of the PQube (2018-90-01 vs. 2019-01-01)

[reconjohn]

Monthly Plot

Monthly plots for the 4 meters in order of PQ, Ant, Tel, and Sci Jan. 2019

Feb. 2019 image.png

Mar. 2019 image.png

Apr. 2019 image.png

Regarding the questions just for the reminder from the previous email,

• Why negative and uniform value in PQ compared to the other 3 meters for EDelTot? I assume it is recorded wrong.

• PTot in PQ is way higher than other 3 meters. (i.e., 1,000,000 kW vs. 250 kW or less). Is it recorded wrong?

• How PPeak in PQ could be interpreted? In the index description, PPeak is aligned with PfTot which is Power Factor Total but the trend of this one looks very different. PPeak may represent the one peak point during the month I guess.

• More clear explanation for SlidingWindowRealPowerDemand3Ph such that how it is calculated in relation to PTot (Active Power Total).

• What is SPeak? It is in kVA. According to the description, it is along with SlidingWindowRealPowerDemand3Ph, but its trend shows different. In general, SlidingWindowRealPowerDemand3Ph this variable would be very useful for our model.

• I tried to use timeseries models but I found the data behavior is a bit hard to use for the models. The variation is quite larger than expected. For example, In this plot, the active power total has values between 50 kW to 300 kW, we may end up with predicting between 50 to 300 meaning the quality of prediction with this data would not be helpful. Most of plots are showing the similar way. Understanding more the relationship among the variables and clear definition of how they are measured would be useful. Once we found how to simplify the data by reducing variations (which I assume some parts are from measuring errors), it would be easier to use predicting models.

At this point, I need further clarification as above to try analyses for model fitting, strategies for any rooms to reduce cost and any economical options.

Daily plots for the 4 meters in order of Ant, Tel, Sci, and PQ on 2018-90-01

Monthly plots for the 4 meters in order of PQ, Ant, Tel, and Sci

Jan. 2019

Feb. 2019 image.png

Mar. 2019 image.png

Apr. 2019 image.png

[chintanp]

@reconjohn what is the final status of this issue? Please post updates to this.

[reconjohn]

It was found that there were some error entries in the data reading and we raised the issues to ACEP. ACEP acknowledged the issues. In the mean time, PTot is figured as the variable worthwhile for the further analyses of demand charge regardless of the error entries. Along this line, energy consumption was calculated from PTot rather than the energy consumption data which is questionable. ACEP needs to continue to work on this issue.