added upscale_droop folder (#21)

* added upscale_droop folder * Addressed comments --------- Co-authored-by: Phoebe <[email protected]>
UNSW-CEEM · Mar 21, 2024 · 7456c39 · 7456c39
1 parent a98cb9e
commit 7456c39
Show file tree

Hide file tree

Showing 3 changed files with 291 additions and 0 deletions.
diff --git a/upscale_droop/Upscaling DPV droop response documentation.docx b/upscale_droop/Upscaling DPV droop response documentation.docx
diff --git a/upscale_droop/upscale_droop_functions.R b/upscale_droop/upscale_droop_functions.R
@@ -0,0 +1,80 @@
+
+impose_sample_size_threshold_compliance <- function(Proportions, sample_threshold){
+
+  # Add OEM's to 'Other' if no corresponding capacity in CER
+  Proportions <- Proportions %>% ungroup(manufacturer) %>% mutate(sample_threshold)
+
+  Proportions <- mutate(Proportions,manufacturer = ifelse(is.na(Count), 
+                                                          'Other', manufacturer))
+  # Add manufacturers with a small sample size to Other group.
+  Proportions <- mutate(Proportions, Sample_size = ifelse(is.na(Sample_size), 0, Sample_size))
+
+  Proportions <- mutate(Proportions, manufacturer = ifelse(is.na(capacity), 
+                                                           'Other', manufacturer))
+  Proportions <- mutate(Proportions, 
+                        manufacturer = ifelse(Sample_size < sample_threshold | 
+                                                manufacturer == "Unknown" | 
+                                                manufacturer == "Multiple" |
+                                                manufacturer == "Mixed" | 
+                                                manufacturer == "" |
+                                                is.na(manufacturer), 
+                                              "Other", manufacturer)
+  )
+  # Recalculate disconnection count and sample size. 
+  Proportions <- group_by(Proportions, Standard_Version, manufacturer, StdComplianceCombined)
+  Proportions <- summarise(Proportions, 
+                           Count = sum(Count, na.rm = TRUE),
+                           Sample_size = sum(Sample_size, na.rm = TRUE),
+                           capacity = sum(capacity, na.rm = TRUE))
+
+  # overwrite "Other capacity"
+  cer_install_data <- filter(cer_install_data, state == region)
+  total_install_on_event_date <- sum(filter(cer_install_data, date == min(event_date, last(cer_install_data$date)))$capacity)
+  total_install_2005_std <- sum(filter(cer_install_data, date == "2016-10-16")$capacity)
+  total_install_2015_std <- sum(filter(cer_install_data, date == "2022-01-01")$capacity) - total_install_2005_std
+  total_install_2020_std <- total_install_on_event_date - total_install_2015_std - total_install_2005_std
+
+  Standard_Capacity <- c(total_install_2015_std, total_install_2020_std)
+  Standard_Version <- c("AS4777.2:2015", "AS4777.2:2020")
+
+  Standard_cap <- data.frame(Standard_Version, Standard_Capacity)
+  Standard_cap$Standard_Version <- as.character(Standard_cap$Standard_Version)
+
+  # 
+  # Set 'other' cap to be difference of total install of standard and the OEMS with > 30 samples'
+  Proportions <- mutate(Proportions, capacity = ifelse(manufacturer == "Other", 0, capacity))
+
+  Capacity_OEM <- group_by(Proportions, Standard_Version, manufacturer, capacity) %>% summarise()
+  Capacity_OEM <- group_by(Capacity_OEM, Standard_Version) %>% summarise(capacity_OEM = sum(capacity))
+
+  Capacities <- left_join(Standard_cap, Capacity_OEM, by = "Standard_Version")
+  Other_capacity <- mutate(Capacities, other_capacity = Standard_Capacity - capacity_OEM)
+  # 
+  Proportions <- left_join(Proportions, Other_capacity[c("Standard_Version", "other_capacity")], by = "Standard_Version")
+  Proportions <- mutate(Proportions, capacity = ifelse(manufacturer == "Other", other_capacity, capacity))
+
+  Proportions <- mutate(Proportions, proportion = Count / Sample_size)
+  Proportions <- data.frame(Proportions)
+  return(Proportions)
+}
+
+
+combine_erroneous_OEMs <- function(df) {
+
+
+  # manufacturers to 'other'
+  df <- mutate(df, manufacturer = ifelse(is.na(proportion_capacity),
+                                         'Other', manufacturer))
+  # Also rename any manufacturers that may have been missed as 'other' (probably redundant after the step above??)
+  df <- mutate(df, 
+               manufacturer = ifelse(manufacturer == "Unknown" | 
+                                       manufacturer == "Multiple" |
+                                       manufacturer == "Mixed" | 
+                                       manufacturer == "" |
+                                       is.na(manufacturer), 
+                                     "Other", manufacturer))
+  return(df)
+}
+
+
+
diff --git a/upscale_droop/upscale_droop_script.R b/upscale_droop/upscale_droop_script.R
@@ -0,0 +1,211 @@
+########################################### PARAMATERS and External Functions #####################################
+
+# Make sure to highlight and run the two functions at the bottom of the script first before running! 
+
+source("load_tool_environment.R")
+source("upscale_droop/upscale_droop_functions.R")
+
+event_date <- "2022-11-12"
+region <- "SA"
+site_norm <- FALSE # If true, no external capacity factor used
+external_capacity_factor <- 0.28
+
+
+underlying_data_file <- "C:/Users/mtrollip/Local/GitHub/DER_disturbance_analysis/data/2022-11-12/phoebe_results_longer_window/20221112_underlying_35min_window.csv"
+CER_install_data_file <- "inbuilt_data/cer_cumulative_capacity_and_number.csv"
+CER_install_manufacturer_data_file <- "inbuilt_data/cer_cumulative_capacity_and_number_by_manufacturer.csv"
+
+# Where you want to store your outputted results
+output_directory <- "C:/Users/mtrollip/Local/GitHub/DER_disturbance_analysis/data/2022-11-12/phoebe_results_longer_window/droop_scale_response"
+
+########################################### Read in Data ###################################
+
+# Underlying data 
+UD_raw <- read.csv(file = underlying_data_file, header = TRUE, stringsAsFactors = FALSE)
+
+########################################### Process Data ###################################
+
+# Here we:
+#1. filter out bad sites
+#2. filter out standards that arent expected to provide droop
+#3. overwrite the 2015 VDRT and Transition 2020-21 standards as 2015 Standard 
+#4. Sub in 2020 droop rsponse column into 2015 droop response column for 2020 circuits (the 2015 droop response
+#   column becomes the combined droop response column)
+#5. bucket Compliant and Non-compliant responding together as one group 
+#6. Bucket together droop response and standard into a combined standard and compliance column
+
+# Filter out sites that arent expected to perform droop response
+
+standards_with_droop <- c('AS4777.2:2015', 'AS4777.2:2015 VDRT', 'AS4777.2:2020', 'Transition 2020-21')
+
+UD <- filter(UD_raw, Standard_Version %in% standards_with_droop)
+
+# # Filter out any sites with "bad data" 
+BadCategories <- c("Not enough data", "Undefined", "UFLS Dropout")
+# Select distinct site_id/response_category combos
+BadSiteIds <- group_by(UD_raw, site_id, compliance_status) %>% summarise()
+# Filter to get a list of site_ids to remove
+# Note that R reads in the NAs as actual NA values which is annoying. You do you R.
+BadSiteIds <- filter(BadSiteIds,compliance_status %in% BadCategories | is.na(compliance_status))
+# Remove bad site_ids. Note the ! negates the %in% operator to make in 'not in'
+UD <- filter(UD,!site_id %in% BadSiteIds$site_id)
+
+# Group 2015 VDRT and Transition 2020-21 into 2015 Standard
+UD <- mutate(UD, Standard_Version = 
+                            ifelse(Standard_Version %in% c("AS4777.2:2015 VDRT", "Transition 2020-21"),
+                                   "AS4777.2:2015", Standard_Version))
+
+# write in 2020 standard into the droop compliance column if standard is 2020
+UD <- mutate(UD, compliance_status = ifelse(Standard_Version == "AS4777.2:2020", compliance_status_2020, compliance_status))
+
+# Concatenate standards with droop response (for grouping)
+UD <- mutate(UD, compliance_status = ifelse(compliance_status %in% c("Non-compliant Responding", "Compliant"), "Responding",
+                                            ifelse(compliance_status == "Non-compliant", "Not Responding",  compliance_status)))
+
+UD <- mutate(UD, StdComplianceCombined = paste(Standard_Version, compliance_status))
+
+######################## Get proportion (by count) of each droop response per standard and OEM  ###############################
+
+Proportions <- group_by(UD, Standard_Version, c_id, compliance_status, manufacturer) %>% summarise()
+Proportions <- mutate(Proportions, StdComplianceCombined = paste(Standard_Version, compliance_status))
+TotalPerStandard_and_OEM <- group_by(Proportions, Standard_Version, manufacturer) %>% summarise(Sample_size=n())
+Proportions <- group_by(Proportions, Standard_Version, StdComplianceCombined, manufacturer) %>% summarise(Count=n())
+Proportions <- left_join(Proportions, TotalPerStandard_and_OEM, by = c("Standard_Version","manufacturer"))
+Proportions <- mutate(Proportions, Proportion = Count/Sample_size)
+
+
+################################### Get CER installed capacity per group ##########################################
+
+#1. gets capacity per Standard and OEM
+#2. Assigns portions of that capacity to each group based on Proportions in previous section 
+
+# Get fleet capacity data
+cer_install_data <- read.csv(CER_install_data_file,
+                             header = TRUE, stringsAsFactors = FALSE)
+manufacturer_install_data <- read.csv(CER_install_manufacturer_data_file,
+                                      header = TRUE, stringsAsFactors = FALSE)
+
+# Get manufacturer installed capacity for each OEM and all Standards with droop compliance at the time of the event
+manufacturer_install_data <- calc_installed_capacity_by_standard_and_manufacturer(manufacturer_install_data)
+manufacturer_install_data <- get_manufacturer_capacitys(manufacturer_install_data, event_date, region)
+manufacturer_install_data <- filter(manufacturer_install_data, Standard_Version %in% standards_with_droop)
+# combined VDRT and Transition with 2015
+manufacturer_install_data <- mutate(manufacturer_install_data, Standard_Version = 
+                                      ifelse(Standard_Version %in% c("AS4777.2:2015 VDRT", "Transition 2020-21"),
+                                             "AS4777.2:2015", Standard_Version))
+manufacturer_install_data <- group_by(manufacturer_install_data, Standard_Version, s_state, manufacturer)
+manufacturer_install_data <- summarise(manufacturer_install_data, capacity = sum(capacity))
+
+Proportions <-  merge(Proportions, manufacturer_install_data, by = c('Standard_Version', 'manufacturer'), 
+                      all = TRUE)
+
+# Combined any OEM with less than 30 samples into 'Other' group. Correctly adjust 'Other' install capacity
+Proportions <- impose_sample_size_threshold_compliance(Proportions, 30)
+
+# Get capacity per droop response as a proportion of total isntalled capacity for that OEM and Standard. 
+Proportions <- mutate(Proportions, proportion_capacity = proportion*capacity)
+
+
+##################################### Upscale MW profile by OEM ###################################
+# Two options here
+# Option 1: Use a 'site performance factor'. The power profile of a site is divided by the sites max capacity. 
+#           The average site performance factor for given OEM, Standard and Droop response is used to represent the capacity factor per timestep for that class. 
+#           The average site performance factor is multiplied by the proportional installed capacity of that OEM, Standard and droop response (stored in Proportions)
+#           The result is summed over all OEMs for that given Standard and droop response to provide the upscaled MW profile per droop response class and Standard 
+
+# Option 2: Use 'external capacity factor'. The power profile is normalised to its output immediately before the event 
+#           (i.e this gives an output of 1 just before event). 
+#           The normalised power trace is multiplied by the proportional installed capacity for that OEM, Standard and droop response (store in Proportions)
+#           This would mean each class would at be at its maximum capacity output immediately before the event (as normalised value is 1 at pre event interval)
+#           the traces are then scaled by the external capacity factor, 0.X, such that the outputs is at X% its maximum capacity during the pre event interval   
+
+###################################################################################################################
+
+# Option 1
+if(site_norm) {
+  # Get the average site performance factor in underlyng data for each class per timestep
+   site_performance_factor <- group_by(UD, site_id, ts) %>%
+     summarise(site_performance_factor = first(site_performance_factor),
+               manufacturer = first(manufacturer), Standard_Version = first(Standard_Version),
+               StdComplianceCombined = first(StdComplianceCombined))
+
+   # frst join UD with proportions to identify which OEMs have been assigned to "other" (come up as NA in proportion_cap)
+   site_performance_factor <- left_join(site_performance_factor, Proportions[c("StdComplianceCombined", "proportion_capacity", "manufacturer")],
+                                        by = c("StdComplianceCombined","manufacturer"))
+
+   # combine any OEM's into 'Other' that are < 30 samples or have unknown, multiple / mixed. This apears as with a proportion capacity of 'na' following the join  
+   site_performance_factor <- combine_erroneous_OEMs(site_performance_factor)
+
+
+  # Combine and get average site performance capacity factor for each class with the OEMs > 30 samples. 
+   site_performance_factor <- group_by(site_performance_factor, ts,
+                                       manufacturer, Standard_Version, StdComplianceCombined) %>%
+      summarise(average_site_performance_factor = mean(site_performance_factor))
+
+
+   # add back the proportion capacities
+   upscale_MW_profile_OEM <- left_join(site_performance_factor, Proportions[c("StdComplianceCombined", "proportion_capacity", "manufacturer")],
+                                       by = c("StdComplianceCombined","manufacturer"))
+
+   # upscale per class 
+   upscale_MW_profile_OEM <- mutate(upscale_MW_profile_OEM, upscale_MW = average_site_performance_factor*proportion_capacity/1000)
+
+   # Combine the total from each OEM
+   upscale_MW_profile <- group_by(upscale_MW_profile_OEM, ts, StdComplianceCombined)
+   upscale_MW_profile <- summarise(upscale_MW_profile, upscale_MW = sum(upscale_MW))
+
+   pivot_upscaled_MW_profile <- pivot_wider(select(upscale_MW_profile, c("ts","StdComplianceCombined", "upscale_MW")), names_from = StdComplianceCombined,
+                                             values_from =upscale_MW)
+   write.csv(pivot_upscaled_MW_profile, paste(output_directory,"droop_compliance_upscale_by_OEM_site_normalisation.csv",sep=""), row.names = FALSE)
+
+
+} else{
+  c_id_norm_power <- UD[c('ts', 'c_id', 'c_id_norm_power', 'manufacturer', 'Standard_Version', 'StdComplianceCombined')]
+
+
+    # Add OEMs < 30 samples or that dont exist in CER into 'other'.  
+    c_id_norm_power <- left_join(c_id_norm_power, Proportions[c("StdComplianceCombined", "proportion_capacity", "manufacturer")], by = c("StdComplianceCombined","manufacturer"))
+
+    # For all OEMs in the c_id_norm_power df that are not present in the list of OEMs with n>30, set these 
+    # manufacturers to 'other'
+    c_id_norm_power <- combine_erroneous_OEMs(c_id_norm_power)
+
+    #Recombine and get an average normalised power profile for each class
+    average_c_id_norm_power <- group_by(c_id_norm_power, ts, manufacturer, StdComplianceCombined, Standard_Version) %>%
+      summarise(average_c_id_norm_power = mean(c_id_norm_power))
+
+    # add back the porportion capacities
+    upscale_MW_profile_OEM <- left_join(average_c_id_norm_power, 
+                                         Proportions[c("StdComplianceCombined", "proportion_capacity", "manufacturer")],
+                                         by = c("StdComplianceCombined","manufacturer"))
+
+
+    # upscale per class 
+    upscale_MW_profile_OEM <- mutate(upscale_MW_profile_OEM, upscale_MW = average_c_id_norm_power*proportion_capacity/1000)
+
+
+    # Combine the total per timestamp from each OEM for each Standard and response type
+    upscale_MW_profile <- group_by(upscale_MW_profile_OEM, ts, StdComplianceCombined)
+    upscale_MW_profile <- summarise(upscale_MW_profile, upscale_MW = sum(upscale_MW))
+
+
+    # filter out Off at t0 class as not applicable when dividing by pre-event interval output (will be dividing by 0)
+    upscale_MW_profile <- filter(upscale_MW_profile, !StdComplianceCombined %in% c("AS4777.2:2015 Off at t0", "AS4777.2:2020 Off at t0"))
+
+    # Get fleet capacity per Standard and droop response type
+    Prop_per_class <- group_by(Proportions, StdComplianceCombined) %>% summarise(total_capacity = sum(proportion_capacity)/1000)
+
+    # Add fleet capacity per Standard and droop response type to profile and divide to re-normalise the traces
+    upscale_MW_profile <- left_join(upscale_MW_profile, Prop_per_class, by = c("StdComplianceCombined"))
+    upscale_MW_profile <- mutate(upscale_MW_profile, normalised_power = upscale_MW/total_capacity)
+    upscale_MW_profile <- mutate(upscale_MW_profile, upscale_MW = upscale_MW * external_capacity_factor)
+
+
+    pivot_upscaled_MW_profile <- pivot_wider(select(upscale_MW_profile, c("ts","StdComplianceCombined", "upscale_MW")), names_from = StdComplianceCombined,
+                                            values_from =upscale_MW)
+
+    # Get the average c_id norm power per class based on 
+    write.csv(pivot_upscaled_MW_profile, paste(output_directory,"droop_compliance_upscale_by_OEM_external_cap_factor_norm.csv", sep=""), row.names = FALSE)
+
+
+}