3.1 Purpose

This report is intended to:

• load and test preliminary ‘black box’ (see Figure 3.1) EV monitoring data provided for assessment purposes by FlipTheFleet
• preliminary analysis of time of charging to understand impact on ‘peak’ electricity demand

Figure 3.1: The Black Box (Source: FlipTheFleet)

3.2 Requirements:

• a safe version of the test FlipTheFleet black box dataset stored on the University of Otago’s High-Capacity Central File Storage HCS at: /Volumes/hum-csafe/Research Projects/GREEN Grid/externalData/flipTheFleet/safe/testData/ftfSafeLatestAll.csv.gz
• original data made safe using ftf/dataProcessing/makeSafe.R

3.3 Code and report history

See:

• https://github.com/CfSOtago/evAnalysis/commits/master/ftf/analysis/testData/ftFBlackBoxTestDataChargeTime.Rmd

3.4 Support

This work was supported by:

• The New Zealand Ministry of Business, Innovation and Employment (MBIE) through the Renewable Energy and the Smart Grid (GREEN Grid) project;
• SPATIALEC - a Marie Skłodowska-Curie Global Fellowship based at the University of Otago’s Centre for Sustainability (2017-2019) & the University of Southampton’s Sustainable Energy Research Group (2019-2020).

3.5 Notes

This document was created using knitr in RStudio with R version 3.5.1 (2018-07-02) running on x86_64-apple-darwin15.6.0. Some of the R code has been included where used for information and reference purposes. Full code is available as noted above.

4.1 Load data

In this section we load and describe the pre-processed safe data from /Volumes/hum-csafe/Research Projects/GREEN Grid/externalData/flipTheFleet/safe/testData/ftfSafeLatestAll.csv.gz. Note that this data does not contain Latitude/Longitude or the vehicle registration number. This has been replaced by a unique hash code (per vehicle).

ftfSafeDT <- data.table::as.data.table(readr::read_csv(dFile))

## Parsed with column specification:
## cols(
##   .default = col_integer(),
##   `Date (GPS)` = col_character(),
##   Altitude = col_double(),
##   `Speed (GPS)` = col_double(),
##   `Speed (Speedometer)` = col_double(),
##   SOC = col_double(),
##   AHr = col_double(),
##   `Pack volts` = col_double(),
##   `Pack amps` = col_double(),
##   `Pack 1 temp (C)` = col_double(),
##   `Pack 2 temp (C)` = col_double(),
##   `Pack 3 temp (C)` = col_double(),
##   `Pack 4 temp (C)` = col_double(),
##   `12V battery (amps)` = col_double(),
##   Hx = col_double(),
##   VIN = col_character(),
##   `12V battery (volts)` = col_double(),
##   `ACC (V)` = col_double(),
##   SOH = col_double(),
##   `SOH (version 2)` = col_double(),
##   `Charger (amps)` = col_double()
##   # ... with 86 more columns
## )

## See spec(...) for full column specifications.

## Warning in rbind(names(probs), probs_f): number of columns of result is not
## a multiple of vector length (arg 1)

## Warning: 11793 parsing failures.
## row # A tibble: 5 x 5 col     row col      expected       actual    file                             expected   <int> <chr>    <chr>          <chr>     <chr>                            actual 1 12490 cabin_t… no trailing c… .6666666… '/Volumes/hum-csafe/Research Pr… file 2 12494 cabin_t… no trailing c… .6666666… '/Volumes/hum-csafe/Research Pr… row 3 12496 cabin_t… no trailing c… .6666666… '/Volumes/hum-csafe/Research Pr… col 4 12499 cabin_t… no trailing c… .2222222… '/Volumes/hum-csafe/Research Pr… expected 5 12501 cabin_t… no trailing c… .2222222… '/Volumes/hum-csafe/Research Pr…
## ... ................. ... .......................................................................... ........ .......................................................................... ...... .......................................................................... .... .......................................................................... ... .......................................................................... ... .......................................................................... ........ ..........................................................................
## See problems(...) for more details.

# re-create rDate & rTime
ftfSafeDT <- ftfSafeDT[, rTime := hms::as.hms(rDateTime)]
ftfSafeDT <- ftfSafeDT[, rDate := lubridate::date(rDateTime)]

# re-create rDow to stop it acting like a factor with alphabetic ordering
ftfSafeDT <- ftfSafeDT[, rDow := lubridate::wday(rDateTime, lab = TRUE)]

That loaded 46,741 observations from 2 vehicles. Table 4.1 summarises the available data for each EV.

4.2 Check data quality

Next we check for NA in dates and other key variables.

It looks like there are NAs in the GPS derived Date & Time variables.

Since we really need to know the date and time, we remove these NAs from the data although they comprise 17472 (37.38 %) observations.

GPS NA means no GPS signal. This could be when the vehicle is charging inside a garage or passes through a tunnel. As we need to know the date & time of charging for all observations, this will need to be fixed.

4.3 Check variables of interest for this analysis

Check charger related variables. These are:

• Pack amps
• Pack volts

DM: “To determine charger power the variables ‘Pack volts’ and ‘Pack amps’ give you everything you need (with some extra code to separate out regen, and it is also useful to do some deltas of voltage and SoC over time to deal with noise in the current sensors). Negative pack amps with speed also at zero is charging (speed greater than zero and negative pack amps is regen).”

Multiplying these two will give power in W. Figures 4.1 to 4.3 examine the distribution of amps, volts and the derived W.

Figure 4.1 shows a density plot for pack amps by EV and by whether or not the amps are -ve, +ve or 0. As we can see, positive amps (providing power to the vehicle) has a cluster of readings close to 0 and then a long positive tail. Negative amps (re-charging the battery) appears to have two distinct clusters and a similarly long negative tail. Only one vehicle reports a large number of 0 Amp readings.

Figure 4.1: Distribution of Pack amp readings by car

Next we check volts. Again we seperate -ve and +ve values. In this case it appears that we should probably filter out:

• -ve volts
• volts > 1000

Figure 4.2: Distribution of charger volt readings

Finally (Figure 4.3) shows the distirbution of the derived -ve, zero and +ve power values using the following filters:

• volts < 0 and volts > 1000 = “Volt error?”

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

## Warning: Removed 288 rows containing non-finite values (stat_bin).

Figure 4.3: Distribution of derived power demand using these filters

As noted above, battery charging will be occuring when power is negative. This will be from the grid when speed is zero.