Skip to content

Commit

Permalink
Merge pull request #12 from ElaadNL/docstrings
Browse files Browse the repository at this point in the history 
Docstrings
  • Loading branch information
@thomasboselaad
thomasboselaad authored Nov 16, 2023
2 parents 644196c + bd61a8a commit f5473f7
Showing 1 changed file with 67 additions and 44 deletions.
111 changes: 67 additions & 44 deletions model.R
View file
Original file line number Diff line number Diff line change
Expand Up @@ -104,7 +104,7 @@ sample_annual_mileage <- function(n_runs, year) {
mileage_density$y <- mileage_density$y[mileage_density$x >= 0]
mileage_density$x <- mileage_density$x[mileage_density$x >= 0]
mileage_cdf <- cumsum(mileage_density$y) * diff(mileage_density$x[1:2])

# Sample annual mileages from the CDF, for the number of simulation runs (n_runs)
uniform_samples <- runif(n_runs, min = min(mileage_cdf), max = max(mileage_cdf))
mileage_samples <- approxfun(mileage_cdf, mileage_density$x)(uniform_samples)
Expand Down Expand Up @@ -165,7 +165,7 @@ sample_annual_endemand_cs <- function(sessions, n_runs) {
energy_per_month <- sessions %>%
arrange(desc(start_datetime)) %>%
# We only take the most recent whole year
filter(start_datetime > start_datetime[1] - years(1)) %>%
filter(start_datetime > start_datetime[1] - 3600 * 24 * 365) %>%
mutate(month=month(start_datetime)) %>%
group_by(month, cs_id) %>%
summarise(energy_sum=sum(energy)) %>%
Expand All @@ -180,7 +180,7 @@ sample_annual_endemand_cs <- function(sessions, n_runs) {
# We sample for each run a single month from all monthly energy sums
uniform_samples <- runif(n_runs, min = min(cdf), max = max(cdf))
samples <- approxfun(cdf, density$x)(uniform_samples)

# Multiply by 12 to get yearly demand
return (samples * 12)
}
Expand All @@ -194,6 +194,7 @@ sample_annual_endemand_cs <- function(sessions, n_runs) {
# After that we normalized the weekly energy demand relative to the annual energy demand.
#
# Args
# season_dist (dataframe): dataframe containing for each week the Q1 and Q3 values for power demand
# annual_energy_demand (double[]): vector containing annual energy demand samples
# n_runs (integer): number of simulation runs
#
Expand Down Expand Up @@ -243,6 +244,9 @@ sample_seasonality <- function(season_dist, annual_energy_demand, n_runs) {
# Args
# sessions (dataframe): dataframe containing session data
# sessions_week (dataframe): dataframe containing session data aggregated on a weekly level
# season_sample (dataframe): dataframe containing for `n` runs and for each week the power demand
# profile_type (string): The type of profile to simulate ("Charging Station" or "Electric Vehicle")
# n_runs (integer): The number of runs (profiles) to simulate
#
# Returns
# sample (dataframe): dataframe containing sampled sessions
Expand Down Expand Up @@ -357,8 +361,12 @@ convert_samples <- function(samples, kW) {
) %>%
dplyr::mutate(
session_id = row_number(),
start_datetime = floor_date(start_datetime, "15 mins"),
end_datetime = ceiling_date(pmin(end_datetime, start_datetime + 3600*24), "15 mins")
start_datetime = round_date(start_datetime, "15 mins"),
end_datetime = round_date(end_datetime, "15 mins"),
# Limit end times to 24 hours
end_datetime = pmin(end_datetime, start_datetime + 3600*24),
# Ensure that the sessions spans at least one interval
end_datetime = pmax(end_datetime, start_datetime + 60*15),
) %>%
filter(end_datetime > start_datetime)

Expand Down Expand Up @@ -487,10 +495,12 @@ adjust_overlapping_sessions <- function(samples) {
# Args
# samples (dataframe): dataframe containing sampled sessions
# profile_type (string): profile type (Electrical Vehicle or Charging Station)
# kW (double): The maximum power of a connection
# n_charging_points (integer): The number of charging points per charging station
#
# Returns
# samples (dataframe): dataframe containing sampled sessions
combine_simultaneous_sessions <- function(samples, profile_type, kW) {
combine_simultaneous_sessions <- function(samples, profile_type, kW, n_charging_points) {
if (profile_type == "Electric Vehicle") {
samples <- samples %>%
select(
Expand All @@ -513,10 +523,10 @@ combine_simultaneous_sessions <- function(samples, profile_type, kW) {
) %>%
dplyr::summarise(
power = sum(power),
n = n()
n = pmin(n(), n_charging_points)
)

samples[samples$power > 2 * kW,]$power <- 2 * kW
samples[samples$power > n_charging_points * kW,]$power <- n_charging_points * kW
}

return (samples)
Expand Down Expand Up @@ -586,12 +596,15 @@ create_profile <- function(samples, n_runs) {
# max_capacity (double): The maximum capacity of the charging station
# base_capacity (double): The base capacity which is always available to connected vehicles
# allowed_capacity_fractions (dataframe): Dataframe containing the allowed capacity fractions per timestamp
#
# Returns
# df_cps (dataframe): df_cps with for each interval the capacity added
create_capacities_from_fractions <- function(
df_cps,
kW,
max_capacity,
base_capacity,
allowed_capacity_fractions
df_cps,
kW,
max_capacity,
base_capacity,
allowed_capacity_fractions
) {
# Map the capacity fractions to the right date-time index
df_cps <- df_cps %>%
Expand Down Expand Up @@ -669,40 +682,50 @@ distribute_overcapacity <- function(df_cps) {
# This function is the main function of the model. It contains the simulation pipeline.
#
# Args
# sessions = individual session data
# sessions_week = session data aggregated on weekly level
# profile_type = level on which we calculated the charging profile: "Electric Vehicle" or "Charging Station"
# charging_location = type of charging location: "public", "work" or "home"
# n_runs = number of simulation runs, default 100
# year = year for which we calculate the charging profile, default 2023
# kW = session power rate, default 11
# ev_cs_ratio = EV/CS ratio when calculating the charging profile on CS-level, default 3
# regular_profile = whether we calculate a regular charging profile or 'netbewust' charging profile
# dashboard = whether the simulation pipeline is being called in dashboard mode or not
# sessions (dataframe): individual session data
# sessions_week (dataframe): session data aggregated on weekly level
# profile_type (string): level on which we calculated the charging profile: "Electric Vehicle" or "Charging Station"
# charging_location (string): type of charging location: "public", "work" or "home"
# n_runs (integer): number of simulation runs, default 100
# year (integer): year for which we calculate the charging profile, default 2023
# kW (double): session power rate in kW, default 11
# ev_cs_ratio (integer): EV/CS ratio when calculating the charging profile on CS-level, default 3
# regular_profile (boolean): whether we calculate a regular charging profile or 'netbewust' charging profile
# base_capacity (double): The base capacity of a charging point when Smart Charging
# max_capacity (double): The maximum capacity of a charging station as a whole
# n_charging_points (integer): The number of charging points per charging station
# times (list[list]): The times during which smart charging is enabled. It should be given as a list of lists.
# capacity_fractions_path (optional, string): Path to CSV containing for each interval the additional capacity fraction
# season_dist_path (string): Path to an RDS containing a dataframe with a seasonality distribution
#
# Returns
# A list containing the individual charging profiles and aggregated charging profile
simulate <- function(
sessions,
sessions_week,
profile_type = "Electric Vehicle",
charging_location = "public",
n_runs = 100,
year = 2023,
by = "15 mins",
# 11kW is the assumed maximum charging rate of an EV using three-phase AC charging
kW = 11.0,
ev_cs_ratio = 3,
regular_profile = TRUE,
# 4kW is the base capacity per CP as defined by "Slim laden voor iedereen 2022 - 2025"
base_capacity = 4,
# P = U * I, and divide by 1000 to get kW
# In general it is assumed that CS' are connected by three-phase 25A cables
max_capacity = (3 * 25 * 230) / 1000,
times=list(list(floor_start=17, floor_end=23, pre_slope=NULL, post_slope=7)),
capacity_fractions_path = NULL,
season_dist_path = "data/Input/seasonality_distribution.rds"
sessions,
sessions_week,
profile_type = "Electric Vehicle",
charging_location = "public",
n_runs = 100,
year = 2023,
by = "15 mins",
# 11kW is the assumed maximum charging rate of an EV using three-phase AC charging
kW = 11.0,
ev_cs_ratio = 3,
regular_profile = TRUE,
# 4kW is the base capacity per CP as defined by "Slim laden voor iedereen 2022 - 2025"
base_capacity = 4,
# P = U * I, and divide by 1000 to get kW
# In general it is assumed that CS' are connected by three-phase 25A cables
max_capacity = (3 * 25 * 230) / 1000,
# Number of charging points per charging station (2 by default, should be set to 1 for home charging)
n_charging_points = 2,
times=list(list(floor_start=17, floor_end=23, pre_slope=NULL, post_slope=7)),
capacity_fractions_path = NULL,
season_dist_path = "data/Input/seasonality_distribution.rds"
) {
# Currently changing the interval size is not supported
by = "15 mins"

# Convert the week into ISO week to ensure week sampling is always done from Monday to Sunday
sessions <- sessions %>% mutate(actual_week=isoweek(start_datetime))

Expand All @@ -726,7 +749,7 @@ simulate <- function(
session_sample <- flatten_samples(session_sample)
session_sample <- calculate_power(session_sample, kW)
session_sample <- adjust_overlapping_sessions(session_sample)
session_sample <- combine_simultaneous_sessions(session_sample, profile_type, kW)
session_sample <- combine_simultaneous_sessions(session_sample, profile_type, kW, n_charging_points)

df_cps <- create_profile(session_sample, n_runs)

Expand All @@ -742,7 +765,7 @@ simulate <- function(
}

df_cps <- distribute_overcapacity(df_cps)

# Remove padding
df_cps <- df_cps[df_cps$date_time >= "2022-01-01 00:00:00",]
df_cps <- df_cps[df_cps$date_time <= "2022-12-31 23:59:59",]
Expand Down

0 comments on commit f5473f7

Please sign in to comment.