Energy Usage – It’s All Math

This is a quick little post where I make sure that I can still update this blog, and that I’m not missing any packages.

library(tidyverse)
library(xml2)

usage_data_xml <- read_xml('energy_usage.xml')

xml_preview <- readLines('energy_usage.xml', n = 100)
cat(xml_preview, sep = '\n')

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<ns3:entry xmlns:espi="http://naesb.org/espi" xmlns:ns3="http://www.w3.org/2005/Atom">
    <ns3:link href="https://prod.apigee.duke-energy.app/meter-usage/v1/usage" rel="self"/>
    <ns3:content>
        <ns3:id>urn:uuid:9de9a99e-cd41-5ff6-91ad-5725d03da0af</ns3:id>
        <espi:IntervalBlock>
            <espi:interval>
                <espi:servicePointId>6007045176</espi:servicePointId>
                <espi:Meter>
                    <espi:meterSerialNumber>108284252</espi:meterSerialNumber>
                    <espi:meterInstallDate>2016-08-16T00:00</espi:meterInstallDate>
                    <espi:meterTerminationDate></espi:meterTerminationDate>
                </espi:Meter>
                <espi:serviceType>ELECTRIC</espi:serviceType>
                <espi:unitOfMeasure>kWH</espi:unitOfMeasure>
                <espi:secondsPerInterval>900</espi:secondsPerInterval>
                <espi:duration>65746800</espi:duration>
                <espi:start>1733893200</espi:start>
            </espi:interval>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733893200</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.04</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733894100</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.02</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733895000</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.01</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733895900</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.02</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733896800</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.01</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733897700</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.02</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733898600</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.01</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733899500</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.02</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733900400</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.02</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733901300</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.01</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733902200</espi:start>
                </espi:timePeriod>
                <espi:readingQuality>ACTUAL</espi:readingQuality>
                <espi:value>0.02</espi:value>
            </espi:IntervalReading>
            <espi:IntervalReading>
                <espi:timePeriod>
                    <espi:start>1733903100</espi:start>
                </espi:timePeriod>

Okay now that I have a little sample of how the file is laid out I can try to read in the whole file, despite it being quite big, but I’m on the clock so who cares.

doc <- read_xml('energy_usage.xml')

ns <- xml_ns(doc)

# Metadata from the interval block
interval <- xml_find_first(doc, ".//espi:interval", ns)

meter_info <- tibble(
  service_point_id = xml_text(xml_find_first(interval, ".//espi:servicePointId", ns)),
  meter_serial_number = xml_text(xml_find_first(interval, ".//espi:meterSerialNumber", ns)),
  meter_install_date = xml_text(xml_find_first(interval, ".//espi:meterInstallDate", ns)),
  service_type = xml_text(xml_find_first(interval, ".//espi:serviceType", ns)),
  unit_of_measure = xml_text(xml_find_first(interval, ".//espi:unitOfMeasure", ns)),
  seconds_per_interval = as.integer(xml_text(xml_find_first(interval, ".//espi:secondsPerInterval", ns))),
  interval_start_unix = as.numeric(xml_text(xml_find_first(interval, ".//espi:start", ns)))
)

# Repeated meter readings
readings <- xml_find_all(doc, ".//espi:IntervalReading", ns)

usage_data <- map_dfr(readings, \(x) {
  tibble(
    reading_start_unix = as.numeric(xml_text(xml_find_first(x, ".//espi:timePeriod/espi:start", ns))),
    reading_quality = xml_text(xml_find_first(x, ".//espi:readingQuality", ns)),
    value = as.numeric(xml_text(xml_find_first(x, ".//espi:value", ns)))
  )
}) |>
  mutate(
    reading_datetime = as_datetime(reading_start_unix)
  )

usage_data <- bind_cols(
  usage_data,
  meter_info[rep(1, nrow(usage_data)), ]
)

reading_start_unix is the unix UTC date and time.
reading_quality is a junk value, this column will be dropped.
value is kilowatt used in that 15 minute interval.
reading_datetime is the date and time the 15 minute interval reading starts.

Something that immediately pops into my head, is a simple histogram of the data to see what times of day I use the most electricity. I am going to convert the data to being in my timezone, and drop the junk column.

usage_data_adjusted <- usage_data |>
    mutate(reading_time = with_tz(reading_start_unix, "America/New_York")) |>
    select(reading_time, value)

Now I am going to make a histogram of usage by hour.

usage_hourly <- usage_data_adjusted |>
    mutate(hour = hour(reading_time))

ggplot(usage_hourly, aes(x = hour, weight = value)) +
    geom_histogram(binwidth = 1) +
    scale_x_continuous(breaks = 0:23) +
    labs(
        x = "Hour of Day",
        y = "Total Electricity Usage (kWh)"
    ) +
    theme_minimal()

Well this isn’t that interesting since it seems like my usage is fairly spread out throughout the day, with usage dipping from 06:00 to 12:00 which is interesting because I generally wake up around 08:00 and start work at 10:00.