Homec4science
Diffusion hybrid_potential (Romain_branch) Browse

hybridPOT/libs/demand_mapping.py
ddea7b1c1968Romain_branch

Last Change
Raw File

demand_mapping.py
View Options

import numpy as np
import pandas as pd
import xarray as xr
import os
import time
## ===== GLOBAL CONSTANTS ====== ##
SECTOR_MAP = {1211 : 'HOTELS',
1212 : 'HOTELS',
1220 : 'COMMERCIAL OFFICES',
1230 : 'COMMERCIAL OFFICES',
1231 : 'RESTAURANTS',
1261 : 'MEAN',
1262 : 'MEAN',
1263 : 'SCHOOLS',
1264 : 'HOSPITALS',
1265 : 'MEAN',}
## ====== ANNUAL SUM ======== ##
def get_yearly_sum( ds, var = None ):
# create an xarray datasets that contains, for each timestamp in "ds", the number of repetitions of that hour in the month
# and sums all values over a year
days_of_each_month = [31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31]
month = range(1,13)
days_of_month = pd.DataFrame( data = [ month, days_of_each_month ], index = ['month', 'n_days'] ).transpose()
timestamps = ds.timestamp.to_dataframe()
timestamps['month'] = timestamps.timestamp.dt.month
days_of_month = timestamps.merge( days_of_month,
on = 'month', how = 'left' ).drop( columns = 'month' ).set_index('timestamp').to_xarray()
if var is not None:
annual_sum = (ds[var] * days_of_month.n_days).sum( dim = 'timestamp' )
else:
annual_sum = (ds * days_of_month.n_days).sum( dim = 'timestamp' )
week_weighted = annual_sum.sel( daytype = 'weekday' ) * 5/7 + annual_sum.sel( daytype = 'weekend' ) * 2/7
return week_weighted
## ======= DEMAND MAPPING FUNCTIONS ====== ##
def get_cluster_demand_RES( BLD_df, demand, grouping_col = None, uniqueID_col = 'EGID', dwell_nb = 'N_DWELL',
SFH_demand = 'maison', MFH_demand = 'appart',
communal_demand = 'communal_app', add_communal_demand = True):
SFH = BLD_df[BLD_df[dwell_nb] == 1]
MFH = BLD_df[BLD_df[dwell_nb] > 1]
if grouping_col is None:
building_counts = { 'N_SFH' : len(SFH),
'N_DWELL' : MFH[dwell_nb].sum() }
else:
building_counts = xr.Dataset({ 'N_SFH' : SFH.groupby( grouping_col )[ uniqueID_col ].count(),
'N_DWELL' : MFH.groupby( grouping_col )[ dwell_nb ].sum() }
).fillna(0)
cluster_demand = ( building_counts['N_SFH'] * demand[SFH_demand]
+ building_counts['N_DWELL'] * demand[MFH_demand] )
if add_communal_demand:
cluster_demand = cluster_demand + ( building_counts['N_DWELL'] * demand[communal_demand] )
return [cluster_demand, building_counts]
def get_cluster_demand_SER( BLD_df, demand_norm, demand_annual_TWh = 15.6, # (from Swiss electricity statistics 2018)
grouping_col = None, service_area = 'SER_AREA', sector_map = SECTOR_MAP, map_col = 'GKLAS' ):
# Assumes that demand_norm sums to 1 for each month and weekend/weekday
# --> weekends, weekdays & months are weighted equally
# Create a copy of the input dataframe with relevant columns only
if grouping_col is None:
df = BLD_df[[service_area, map_col]].copy()
df['dummy'] = 1
grouping_vars = ['service_type', 'dummy']
else:
df = BLD_df[[service_area, map_col, grouping_col]].copy()
grouping_vars = ['service_type', grouping_col]
# Compute daily demand
df['area_weight'] = df[service_area] / df[service_area].sum() # use GW for scaling
df['daily_demand'] = df['area_weight'] * demand_annual_TWh * 1e12 / 365
# Map service types
df['service_type'] = df[ map_col ].map( sector_map )
# Group data according to service type and grouping_col (if applicable)
daily_demand_cluster = ( df.groupby( grouping_vars ).daily_demand.sum().unstack(level = 0)
.to_xarray().fillna(0) )
# Remove dummy (if applicable)
if grouping_col is None:
daily_demand_cluster = daily_demand_cluster.drop('dummy').squeeze()
hourly_demand_per_service = daily_demand_cluster * demand_norm
hourly_demand_all_services = hourly_demand_per_service.to_dataframe().sum(axis = 1).to_xarray()
# return the hourly demand
return hourly_demand_all_services, hourly_demand_per_service
c4science · Help