Full Time Series Output

Overview

Directory: full_time_series/ (inside the results directory)

Condition: Only written when WriteFullTimeseries = true in case_settings.json and the system uses time-domain reduction (TDR).

When time-domain reduction (TDR) is used, the optimization runs over a small set of representative periods (e.g., 3 representative weeks) rather than the full year (8,760 hours). The full_time_series/ subdirectory contains the standard operational outputs expanded back to all TotalHoursModeled hours using the period map. This makes it easier to compute annual statistics, produce time series plots, or compare model outputs against historical data — without having to manually apply subperiod weights.

If TDR is not used (i.e., the model already runs over the full year), WriteFullTimeseries has no effect and the subdirectory is not created.

File sizes

Full time series files can be very large (up to 8,760 rows × many components). Long-format files are automatically written in compressed .csv.gz format to manage file sizes. Wide-format files use standard .csv.

Files Produced

The following files are written inside full_time_series/:

File	Format	Condition	Description
`flows.csv` or `flows.csv.gz`	Wide / Long (`.gz`)	Always (if enabled)	Expanded commodity flows for all edges
`non_served_demand.csv` or `non_served_demand.csv.gz`	Wide / Long (`.gz`)	Always (if enabled)	Expanded non-served demand for all NSD-enabled nodes
`storage_level.csv` or `storage_level.csv.gz`	Wide / Long (`.gz`)	Always (if enabled)	Expanded storage state of charge
`curtailment.csv` or `curtailment.csv.gz`	Wide / Long (`.gz`)	Always (if enabled)	Expanded VRE curtailment
`balance_duals.csv`	Wide	`DualExportsEnabled = true`	Expanded locational marginal prices

The file format (.csv vs .csv.gz) is determined automatically based on the OutputLayout setting:

Long format → compressed .csv.gz
Wide format → standard .csv

The column structure of each file is identical to the corresponding file in the parent results directory (e.g., flows.csv in full_time_series/ has the same columns as flows.csv in results/), except that the time column now runs from 1 to TotalHoursModeled instead of only the representative time steps.

Reconstruction Method

Full time series are reconstructed from representative-period results using the period map (SubPeriodMap) specified in the time_data.json input file.

The reconstruction proceeds as follows:

For each full-year sub-period, look up which representative sub-period it was assigned to (via the period map).
Copy the values for that representative sub-period's time steps into the corresponding positions for the full-year sub-period.
Repeat for all full-year sub-periods.

\[\text{full\_ts}[\text{full\_period\_hours}] = \text{rep\_ts}[\text{representative\_hours}]\]

Padding for incomplete period maps. If the period map covers fewer hours than TotalHoursModeled (e.g., 52 × 168 = 8,736 < 8,760), the remaining hours are filled by repeating the values of the representative sub-period corresponding to the last calendar sub-period in the map. A warning is logged when padding occurs.

Configuration

Setting	File	Default	Effect
`WriteFullTimeseries`	`case_settings.json`	`false`	Set to `true` to enable writing of the `full_time_series/` subdirectory. Has no effect if TDR is not used.
`OutputLayout` (or per-variable keys)	`macro_settings.json`	`"long"`	Controls whether files are written in long (`.csv.gz`) or wide (`.csv`) format.
`DualExportsEnabled`	`macro_settings.json`	`true`	When `true`, also writes `full_time_series/balance_duals.csv`.

Assumptions

Only for TDR runs. If the system does not use TDR (i.e., no period map is provided), WriteFullTimeseries = true has no effect and a warning is logged. The full year is already represented in the standard outputs.
Storage levels. The cyclic-within-period constraint means the storage level at the end of a representative sub-period equals the level at its start. This is faithfully reproduced in the expanded storage level: the same state-of-charge profile repeats for all full-year sub-periods assigned to the same representative sub-period.
Large files. At 8,760 rows × hundreds or thousands of components, long-format files can exceed hundreds of MB. The automatic .csv.gz compression typically reduces file sizes by 90%.
Period map required. The period map (SubPeriodMap) must be specified in time_data.json. See Time Data for how to configure TDR and the period map.

Examples

Enabling Full Time Series Output

In case_settings.json:

{
    "SolutionAlgorithm": "Monolithic",
    "WriteFullTimeseries": true
}

Reading Full Time Series in Julia

using CSV, DataFrames

# Read long-format flows
flows_full = CSV.read("results/full_time_series/flows.csv.gz", DataFrame)
# flows_full has 8760 × n_components rows (one per (component, hour) pair)

# Annual electricity generation per component
elec = filter(r -> r.commodity == "Electricity", flows_full)
annual_gen = combine(groupby(elec, :resource_id), :value => sum => :annual_MWh)

Directory Structure After Full Time Series Export

results/
├── capacity.csv
├── costs.csv
├── flows.csv
├── storage_level.csv
├── curtailment.csv
├── non_served_demand.csv
├── balance_duals.csv
├── co2_cap_duals.csv
├── time_weights.csv
└── full_time_series/
    ├── flows.csv.gz           ← 8760-hour flows (compressed long format)
    ├── storage_level.csv.gz
    ├── curtailment.csv.gz
    ├── non_served_demand.csv.gz
    └── balance_duals.csv      ← wide format, uncompressed