ReEDS-Model · patrickbrown4 · Apr 15, 2026 · May 21, 2026 · May 21, 2026 · May 21, 2026
diff --git a/cases.csv b/cases.csv
@@ -179,6 +179,7 @@ GSw_HourlyClusterRegionLevel,Indicate the region hierarchy level at which to agg
 GSw_HourlyClusterWeights,/-delimited relative weights for load and RE profiles used in clustering,^load_\d*.?\d+\/upv_\d*.?\d+\/wind-ons_\d*.?\d+\/wind-ofs_\d*.?\d+$,load_1/upv_1/wind-ons_1/wind-ofs_0,
 GSw_HourlyClusterYear,Year to use when clustering EFS and/or climate-modified data,N/A,2035,
 GSw_HourlyClusterTimestep,Indicate the time resolution to use in clustering algorithm (period=day or wek depending on GSw_HourlyType),hour; period,period,
+GSw_HourlyClusterMapMethod,How to map actual periods to representative periods: 'milp' minimizes sum of absolute errors but is slow; 'bestfirst' is orders of magnitude faster when using many weather years in GSw_HourlyWeatherYears,^(milp|bestfirst)$,milp,
 GSw_HourlyMinRElevel,Indicate the region hierarchy level at which to include minimum wind and solar capacity factor days (or False to ignore min-wind/solar CF days),false; False; FALSE; r; nercr; transreg; transgrp; cendiv; st; interconnect; country; usda_region; ccreg,interconnect,
 GSw_HourlyNormProfiles,Indicate whether to normalize load/wind/solar profiles before identification of representative periods,0; 1,0,
 GSw_HourlyNumClusters,Number of clusters to create (i.e. number of representative days or weks),int,33,

diff --git a/cases_test.csv b/cases_test.csv
@@ -1,4 +1,4 @@
-,Default Value,Pacific,USA_defaults,Mid_Case,USA_decarb,github_Pacific,github_Everything,github_MA_county_CC,Pacific_CC,Pacific_weks,Pacific_full_year,Interday_storage,Pacific_2020,Pacific_rep7,WY_county,WECC_county,PJM_county_CC,NYVT_mixed,OR_water,MonteCarlo_Random,MonteCarlo_LHS,Everything,Simple,USA_fast,USA_faster,Pacific_DR,Pacific_MGA,Pacific_LoadSite95,MARICTNYNJPAOH_Offshore,R2P
+,Default Value,Pacific,USA_defaults,Mid_Case,USA_decarb,github_Pacific,github_Everything,github_MA_county_CC,Pacific_CC,Pacific_weks,Pacific_full_year,Interday_storage,Pacific_2020,Pacific_rep15,WY_county,WECC_county,PJM_county_CC,NYVT_mixed,OR_water,MonteCarlo_Random,MonteCarlo_LHS,Everything,Simple,USA_fast,USA_faster,Pacific_DR,Pacific_MGA,Pacific_LoadSite95,MARICTNYNJPAOH_Offshore,R2P
 ignore,1,0,,,,,,,,,,,,,,,,,,,,,,,,,,,,
 GSw_Region,cendiv/Pacific,,country/USA,country/USA,country/USA,,st/ID.WY.NE.IA.IL,st/MA,,,,,,,st/WY,interconnect/western,transreg/PJM,st/NY.VT,st/OR,st/NE.NY.PA,st/NE.NY.PA,st/ID.WY.NE.IA.IL,st/KS,country/USA,country/USA,,,,st/MA.RI.CT.NY.NJ.PA.OH,
 endyear,2032,,2050,2050,2050,2029,2060,2026,,,,,,,,,,,2035,2030,2030,2060,2035,2050,2050,,,,,
@@ -21,7 +21,8 @@ GSw_PRM_CapCredit,,,,,,,,1,1,,,,,,,,1,,,,,,,,,,,,,
 GSw_PRM_scenario,,,,,,,,,static,,,,,,,,static,,,,,,,,,,,,,
 GSw_HourlyType,,,,,,,,,,wek,year,,,,,,,,,,,,,,,,,,,
 GSw_InterDayLinkage,,,,,,,,,,,,1,,,,,,,,,,,,,,,,,,
-GSw_HourlyWeatherYears,,,,,,,2012_2013,,,,,,2020,2007_2008_2009_2010_2011_2012_2013,,,,,,,,2012_2013,,,,2018,,,,
+GSw_HourlyWeatherYears,,,,,,,2012_2013,,,,,,2020,2007_2008_2009_2010_2011_2012_2013_2016_2017_2018_2019_2020_2021_2022_2023,,,,,,,,2012_2013,,,,2018,,,,
+GSw_HourlyClusterMapMethod,,,,,,,,,,,,,,bestfirst,,,,,,,,,,,,,,,,
 GSw_WaterCapacity,,,,,,,,,,,,,,,,,,,1,,,,,,,,,,,
 GSw_WaterMain,,,,,,,,,,,,,,,,,,,1,,,,,,,,,,,
 GSw_WaterUse,,,,,,,,,,,,,,,,,,,1,,,,,,,,,,,

diff --git a/docs/source/user_guide.md b/docs/source/user_guide.md
@@ -75,7 +75,7 @@ Here is partial list of remotely hosted files used by ReEDS:
       affinity='euclidean', linkage='ward')
   ```
 
-  - If set to 'optimized', then a two-step custom optimization is performed using the `hourly_repperiods.optimize_period_weights()` and `hourly_repperiods.assign_representative_days()` functions to minimize the deviation in regional load and PV/wind CF between the weighted representative periods and the full year.
+  - If set to 'optimized', then a two-step custom optimization is performed using the `hourly_repperiods.optimize_period_weights()` and `hourly_repperiods.match_act2rep_milp()` functions to minimize the deviation in regional load and PV/wind CF between the weighted representative periods and the full year.
   - If set to a string containing the substring 'user', then instead of optimizing the choice of representative periods for this run, the model reads a user-supplied file at `inputs/temporal/period_szn_{GSw_HourlyClusterAlgorithm}.csv`.
       - So if you want to use the example period:szn map, set `GSw_HourlyClusterAlgorithm=user` and provide `inputs/temporal/period_szn_user.csv`.
       - If you want to specify a different period:szn map, then create a file with your label in the filename and set `GSw_HourlyClusterAlgorithm` to that same label (which must contain the substring 'user'). For example, for `GSw_HourlyClusterAlgorithm=user_myname_20230130`, provide `inputs/temporal/period_szn_user_myname_20230130.csv`.

diff --git a/reeds/core/setup/b_inputs.gms b/reeds/core/setup/b_inputs.gms
@@ -6029,7 +6029,7 @@ Set
     starting_hour_nowrap(allh)             "Flag for whether allh is the first chronological hour by day type"
     final_hour(allh)                       "Flag for whether allh is the last chronological hour in a day type" 
     final_hour_nowrap(allh)                "Flag for whether allh is the last chronological hour in a day type"
-    nextszn(allszn,allszn)                 "Mapping between one actual period (allszn) and the next"
+    nextszn(actualszn,actualszn)           "Mapping between one actual period (actualszn) and the next"
     nextpartition(allszn,allszn)           "Mapping between one partition (allszn) and the next"
 * Peak demand
     maxload_szn(r,allh,t,allszn)           "hour with highest load within each szn"

diff --git a/reeds/core/solve/2_temporal_params.gms b/reeds/core/solve/2_temporal_params.gms
@@ -134,6 +134,16 @@ $include inputs_case%ds%stress%stress_year%%ds%nexth.csv
 $offdelim
 $onlisting
 /
+
+nextszn(actualszn,actualszn) "Mapping between one actual period (actualszn) and the next"
+/
+$offlisting
+$ondelim
+$include inputs_case%ds%nextszn.csv
+$offdelim
+$onlisting
+/
+
 $ONEMPTY
 nextpartition(allszn,allszn) "Mapping between one partition (allszn) and the next"
 /
@@ -181,13 +191,6 @@ starting_hour_nowrap(h)$[sum{szn, h_szn_start(szn,h) }$(not Sw_HourlyWrap)] = ye
 final_hour_nowrap(allh) = no ;
 final_hour_nowrap(h)$[sum{szn, h_szn_end(szn,h) }$(not Sw_HourlyWrap)] = yes ;
 
-* Get the order of actual periods
-nextszn(actualszn,actualsznn)$[(ord(actualsznn) = ord(actualszn) + 1)] = yes ;
-nextszn(actualszn,actualsznn)
-    $[(ord(actualszn) = smax(actualsznnn, ord(actualsznnn)))
-    $(ord(actualsznn) = smin(actualsznnn, ord(actualsznnn)))]
-    = yes ;
-
 $onOrder
 
 
@@ -894,10 +897,14 @@ szn_adj_gas(h)$frac_h_quarter_weights(h,"wint") =
 avail(i,r,h)$avail(i,r,h) = round(avail(i,r,h),3) ;
 can_imports_szn(r,szn,t)$can_imports_szn(r,szn,t) = round(can_imports_szn(r,szn,t),3) ;
 can_exports_h(r,h,t)$can_exports_h(r,h,t) = round(can_exports_h(r,h,t),3) ;
+frac_h_ccseason_weights(h,ccseason)$frac_h_ccseason_weights(h,ccseason) = round(frac_h_ccseason_weights(h,ccseason),3) ;
+gasadder_cd(cendiv,t,h)$gasadder_cd(cendiv,t,h) = round(gasadder_cd(cendiv,t,h),3) ;
 h_weight_csapr(h)$h_weight_csapr(h) = round(h_weight_csapr(h),3) ;
+hours(h)$hours(h) = round(hours(h),3) ;
 load_exog(r,h,t)$load_exog(r,h,t) = round(load_exog(r,h,t),3) ;
 load_exog_static(r,h,t)$load_exog_static(r,h,t) = round(load_exog_static(r,h,t),3) ;
 minloadfrac(r,i,h)$minloadfrac(r,i,h) = round(minloadfrac(r,i,h),3) ;
+numdays(szn)$numdays(szn) = round(numdays(szn),3) ;
 szn_adj_gas(h)$szn_adj_gas(h) = round(szn_adj_gas(h), 3) ;
 cap_hyd_szn_adj(i,szn,r)$cap_hyd_szn_adj(i,szn,r) = round(cap_hyd_szn_adj(i,szn,r),3) ;
 peakdem_static_ccseason(r,ccseason,t)$peakdem_static_ccseason(r,ccseason,t) = round(peakdem_static_ccseason(r,ccseason,t),2) ;

diff --git a/reeds/input_processing/hourly_plots.py b/reeds/input_processing/hourly_plots.py
@@ -14,7 +14,6 @@
 from pathlib import Path
 sys.path.append(str(Path(__file__).parent.parent.parent))
 import reeds
-from reeds.input_processing import hourly_repperiods
 from reeds import plots
 plots.plotparams()
 
@@ -135,9 +134,9 @@ def plot_ldc(
     ### Get clustered load, repeating representative periods based on how many
     ### periods they represent
     numperiods = period_szn.value_counts().rename('numperiods').to_frame()
-    numperiods['yearperiod'] = numperiods.index.map(hourly_repperiods.szn2yearperiod).values
-    numperiods['year'] = numperiods.index.map(hourly_repperiods.szn2yearperiod).map(lambda x: x[0])
-    numperiods['yperiod'] = numperiods.index.map(hourly_repperiods.szn2period)
+    numperiods['yearperiod'] = numperiods.index.map(reeds.timeseries.szn2yearperiod).values
+    numperiods['year'] = numperiods.index.map(reeds.timeseries.szn2yearperiod).map(lambda x: x[0])
+    numperiods['yperiod'] = numperiods.index.map(reeds.timeseries.szn2period)
     periods = [[row.yearperiod] * row.numperiods for (i,row) in numperiods.iterrows()]
     periods = [item for sublist in periods for item in sublist]
 
@@ -150,11 +149,11 @@ def plot_ldc(
     hourly_out = hourly_in.unstack('h_of_period').loc[periods].stack('h_of_period')
 
     #### Daily
-    periodly_in = hourly_in.groupby('yperiod').mean()
+    periodly_in = hourly_in.groupby(['year','yperiod']).mean()
     ## Index doesn't matter; replace it so we can take daily mean
     periodly_out = hourly_out.copy()
     hourly_out.index = hourly_in.index.copy()
-    periodly_out = hourly_out.groupby('yperiod').mean()
+    periodly_out = hourly_out.groupby(['year','yperiod']).mean()
 
     ### Get axis coordinates: properties = rows, regions = columns
     properties = periodly_out.columns.get_level_values('property').unique().values
@@ -563,7 +562,7 @@ def get_profiles(regions, year):
         dforig = pd.concat(dforig, axis=1)
 
         ### Representative profiles
-        periodmap = period_szn.map(hourly_repperiods.szn2yearperiod).to_frame()
+        periodmap = period_szn.map(reeds.timeseries.szn2yearperiod).to_frame()
         periodmap['year'] = periodmap.szn.map(lambda x: x[0])
         periodmap['yperiod'] = periodmap.szn.map(lambda x: x[1])
         periodmap = periodmap.loc[periodmap.year==year].yperiod
@@ -657,7 +656,7 @@ def plot_load_days(profiles, rep_periods, period_szn, sw, reeds_path, figpath):
     """
     """
     ### Input processing
-    idx_reedsyr = period_szn.map(hourly_repperiods.szn2yearperiod)
+    idx_reedsyr = period_szn.map(reeds.timeseries.szn2yearperiod)
     medoid_profiles = profiles.loc[rep_periods]
     centroids = profiles.loc[rep_periods]
     centroid_profiles = centroids * profiles.stack('h_of_period').max()