In 07engineer/residential_loadshapes:

library(plyr)
library(tidyverse)
library(knitr)
library(lubridate)
library(stringr)
library(residential)
#options(digits=2)

opts_chunk$set(echo=FALSE, cache = FALSE, message = FALSE, warning = FALSE)

Calibration Plots

sim_path <- "L:/P/1631/Task 4 - Baseline Profiles/Residential Calibrated Models 09182017/FCZ1_FCZ11/FCZ2"
billing_data_path <- "L:/P/1631/Task 4 - Baseline Profiles/Residential Pre-Processor 091417/Data/Cleaned Data/engineering"

sim_file <- "FCZ2_SINGLEFAMILY_GAS_LOW-Run7Meter.csv"

zone <- 2

# Read in billing data file 
billing_data <- read_csv(str_c(billing_data_path, "/FCZ", zone, ".csv")) %>%
  mutate(date = ymd_hms(str_c(date, " ", hour, ":00:00")))

# Pick apart the simulation file name to get building subcategory for meter data filter
run_elements <- str_split(str_replace(sim_file, "Meter.csv", ""), "_")[[1]]
#zone <- as.integer(str_replace_all(run_elements[1], "[[:alpha:]]", "")) 
level <- str_replace(run_elements[4], "-Run", "")%>% 
      str_replace_all("[:digit:]", "")
type <- run_elements[2]
fuel <- run_elements[3]

sim <- read_EP_Output(str_c(sim_path, sim_file, sep = "/")) %>%
  rename(kW_sim = `ElectricityNet:Facility`)

results <- billing_data %>%
  filter(usage_level == level, 
         heat_fuel == fuel, 
         iou_building_type == type) %>%
  select(kW_bills = kW, TemperatureF) %>%
  bind_cols(select(sim, date, kW_sim)) %>%
  select(date, everything()) %>%
  mutate(kW_bills  = kW_bills / sum(kW_bills), 
         kW_sim = kW_sim / sum(kW_sim))
  # mutate(kW_bills_norm  = kW_bills / sum(kW_bills), 
  #        kW_sim_norm = kW_sim / sum(kW_sim))

results <- add_schedule_columns(results)

results_long <- results %>% 
  select(date, kW_sim, kW_bills, month, wday, hour, weekend, TemperatureF) %>% 
  gather(source, kW, kW_sim:kW_bills)

ggplot(results_long, aes(x = date, y = kW, color = source)) + geom_point() + ggtitle("Energy Over Time (hourly)")

# ggplot(results_long, aes(x = date, y = TemperatureF, color = source)) + geom_point() + ggtitle("Temperature Over Time")


# ggplot(results_long, aes(x = TemperatureF, y = kW, color = source)) + geom_point() + ggtitle("Power as a function of TempF")


ggplot(results_long, aes(x = date, y = kW, color = source)) + geom_line() + xlim(mdy_hm("1/15/2014 1:00"), mdy_hm("1/22/2014 1:00")) + 
  ggtitle("Sample winter Week")
ggplot(results_long, aes(x = date, y = kW, color = source)) + geom_line() + xlim(mdy_hm("7/20/2014 1:00"), mdy_hm("7/27/2014 1:00")) + 
  ggtitle("Sample Summer Week")
ggplot(results_long, aes(x = date, y = kW, color = source)) + geom_line() + xlim(mdy_hm("4/15/2014 1:00"), mdy_hm("4/22/2014 1:00")) + 
  ggtitle("Sample Spring Week")

# Daily Averages
daily <- results_long %>% 
  mutate(date = floor_date(date, unit = "day")) %>% 
  group_by(date, source) %>% 
  summarise(kW = mean(kW), TemperatureF = mean(TemperatureF))

ggplot(daily, aes(x = TemperatureF, y = kW, color = source)) + geom_point() + ggtitle("Daily Avg Power as a Function of TempF")

daily$Temp_bin <- cut(daily$TemperatureF, 15)

ggplot(daily, aes(x = Temp_bin, y = kW)) + geom_violin() + facet_wrap(~source) + ggtitle("Daily Avg TempF Bin Visual")
ggplot(daily, aes(x = Temp_bin, y = kW, color = source)) + geom_violin() + ggtitle("Daily Avg TempF Bin Visual")
ggplot(daily, aes(x = date, y = kW, color = source)) + geom_point() + ggtitle("Daily Avg Energy Over Time")

# Monthly 
monthly <- results_long %>%
  group_by(source, month) %>%
  summarise(sum_kW = sum(kW, na.rm = TRUE)) 

ggplot(monthly, aes(x = month, y = sum_kW, fill = source)) + geom_bar(position = "dodge", stat = "identity") + ggtitle("Total kWh/ft^2 per month")

Read in the EnergyPlus meter output file and compare magnitudes

meter <- sim

names(meter) <- names(meter) %>%
  str_replace(":Electricity", "")

# Ditch names with a colon
meter <- meter[,!str_detect(names(meter), ":")]

meter_long <- meter %>% 
  gather(key = "enduse", value = "kW", -date)

# # Only keep top 5 contributors: 
# top5 <- meter_long %>% 
#   group_by(enduse) %>% 
#   summarise(mean_kW = mean(kW)) %>% 
#   arrange(desc(mean_kW))
# 
# meter_long <- meter_long %>% 
#   filter(enduse %in% top5$enduse[1:5])


ggplot(meter_long, aes(x = date, y = kW, color = enduse, linetype = enduse)) + geom_line() +
  xlim(mdy_hm("1/15/2014 1:00"), mdy_hm("1/22/2014 1:00")) + 
  ggtitle("Sample winter Week")

ggplot(meter_long, aes(x = date, y = kW, color = enduse, linetype = enduse)) + geom_line() +  
  xlim(mdy_hm("7/20/2014 1:00"), mdy_hm("7/27/2014 1:00")) + 
  ggtitle("Sample Summer Week")

Error Calculations

The percent error for several relevant time slices are plotted below.

##################################
#### Calculate Errors ############
##################################

percent_error <- function(bills, simulation) {
  round((mean(simulation) - mean(bills)) / mean(simulation) * 100)
}

results <- add_schedule_columns(results)

slices <- tibble()
slice <- results %>% 
  filter(weekend == "weekday", 
         hour > 13 & hour <= 18,
         month == "Jul" | month == "Aug") %>% 
  mutate(sliceName = "summer_weekday_afternoons") 

slices <- slice

slice <- results %>% 
  filter(weekend == "weekday", 
         hour > 1 & hour <= 4,
         month == "Mar" | month == "Apr") %>% 
  mutate(sliceName = "spring_nights") 

slices <- slices %>% bind_rows(slice)

slice <- results %>% 
  filter(weekend == "weekday", 
         hour > 7 & hour <= 11,
         month == "Jan" | month == "Feb") %>% 
  mutate(sliceName = "winter_mornings") 

slices <- slices %>% bind_rows(slice)

slice <- results %>% 
  filter(weekend == "weekend", 
         hour > 1 & hour <= 4,
         month == "Jun" | month == "Jul") %>% 
  mutate(sliceName = "weekend_daytime") 

slices <- slices %>% bind_rows(slice)

calibration_errors <- slices %>% 
  group_by(sliceName) %>% 
  summarise(percent_error = percent_error(kW_bills, kW_sim))

NMBE <- calc_NMBE(results$kW_sim, results$kW_bills) 
CVRMSE <- calc_CVRMSE(results$kW_sim, results$kW_bills)  

kable(calibration_errors)

ggplot(calibration_errors, aes(x = sliceName, y = percent_error)) + geom_col()


write.csv(daily, "daily.csv")

For the entire year,
The NMBE is r NMBE.
The CVRMSE is r CVRMSE.

Simulation Total Annual Energy: r sum(sim$kW_sim)
Billing Total Annual Energy: r sum(billing_data$kW)
Percent difference: r (sum(sim$kW_sim) - sum(billing_data$kW)) / sum(billing_data$kW) * 100 %

07engineer/residential_loadshapes documentation built on May 20, 2019, 5:45 p.m.