ResourceAdequacy

Note: This package is still very much a work in progress and is subject to change. Email Gord for the latest status.

The Probabilistic Resource Adequacy Suite (PRAS) provides a modular collection of data processing and system simulation tools to assess power system reliability.

To use this functionality for capacity valuation, see CapacityCredit.jl. To import systems from PLEXOS, see PLEXOS2PRAS.jl.

Getting Started

Unleash your CPU cores

First, know that PRAS uses multi-threading, so be sure to set the environment variable controlling the number of threads available to Julia (36 in this Bash example, which is a good choice for Eagle nodes - on a laptop you would probably only want 4) before running scripts or launching the REPL:

export JULIA_NUM_THREADS=36

Architecture Overview

PRAS functionality is distributed across a range of different types of modules that can be mixed, matched, extended, or replaced to support the needs of a particular analysis. When assessing reliability or capacity value, one can define the modules to be used while passing along any associated parameters or options.

The categories of modules are:

Extractions: How should VG or load data be extracted from historical time-series data to create probability distributions at each timestep? Options are Backcast or REPRA.

Simulations: How should power system operations be simulated? Options are NonSequentialCopperplate or NonSequentialNetworkFlow.

Results: What level of detail should be saved out during simulations? Options are Minimal, Temporal, Spatial, SpatioTemporal, and Network.

Running an analysis

Analysis centers around the assess method with different arguments passed depending on the desired analysis to run. For example, to run a convolution-based reliability assessment (NonSequentialCopperplate) with VG distributions derived from simple backcasts (Backcast) and aggregate LOLE and EUE reporting (Minimal), one would run:

assess(Backcast(), NonSequentialCopperplate(), Minimal(), mysystemmodel)

To run a network flow simulation instead with 100,000 Monte Carlo samples, the method call becomes:

assess(Backcast(), NonSequentialNetworkFlow(100_000), Minimal(), mysystemmodel)

To use REPRA-style windowing (with a +/- 1-hour, +/- 10-day window) to generate VG distributions, the call becomes:

assess(REPRA(1, 10), NonSequentialNetworkFlow(100_000), Minimal(), mysystemmodel)

To save regional results in a multi-area system, change Minimal to Spatial:

assess(REPRA(1, 10), NonSequentialNetworkFlow(100_000), Spatial(), mysystemmodel)

To save regional results for each simulation period, use the SpatioTemporal result specification instead:

assess(REPRA(1, 10), NonSequentialNetworkFlow(100_000), SpatioTemporal(), mysystemmodel)

Querying Results

After running an analysis, metrics of interest can be obtained by calling the appropriate metric's constructor with the result object.

For example, to obtain the system-wide LOLE over the simulation period:

result = assess(Backcast(), NonSequentialNetworkFlow(100_000), SpatioTemporal(), mysystemmodel)
lole = LOLE(result)

Single-period metrics such as LOLP can also be extracted if the appropriate information was saved (i.e. if Temporal or SpatioTemporal result specifications were used). For example, to get system-wide LOLP for April 27th, 2024 at 1pm:

lolp = LOLP(result, DateTime(2024, 4, 27, 13))

Similarly, if per-region information was saved (i.e. if Spatial or SpatioTemporal result specifications were used), region-specific metrics can be extracted. For example, to obtain the EUE of Region A across the entire simulation period:

eue_a = EUE(result, "Region A")

If the results specification supports it (i.e. SpatioTemporal or Network), metrics can be obtained for both a specific region and time:

eue_a = EUE(result, "Region A", DateTime(2024, 4, 27, 13))

Finally, if using the Network result spec, information about interface flows and utilization factors can be obtained as well:

# Average flow from Region A to Region B during the hour of interest
flow_ab = ExpectedInterfaceFlow(
    result, "Region A", "Region B", DateTime(2024, 4, 27, 13))
    
# Same magnitude as above, but different sign
flow_ba = ExpectedInterfaceFlow(
    result, "Region B", "Region A", DateTime(2024, 4, 27, 13))
    
# Average utilization (average ratio of absolute value of actual flow vs maximum possible after outages)
utilization_ab = ExpectedInterfaceUtilization(
    result, "Region A", "Region B", DateTime(2024, 4, 27, 13))