Merge remote-tracking branch 'upstream/dev' into dev

pull/31/head
Jun He 2 years ago
commit f59914f265

@ -1,4 +1,4 @@
name: Tests
name: Build & Test
on:
push:
pull_request:
@ -6,19 +6,30 @@ on:
- cron: '45 10 * * *'
jobs:
test:
name: Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }}
runs-on: ${{ matrix.os }}
strategy:
matrix:
julia-version: ['1.6', '1.7']
julia-arch: [x64]
os: [ubuntu-latest, windows-latest, macOS-latest]
exclude:
- os: macOS-latest
julia-arch: x86
version: ['1.6', '1.7', '1.8', '1.9']
os:
- ubuntu-latest
arch:
- x64
steps:
- uses: actions/checkout@v2
- uses: julia-actions/setup-julia@latest
- uses: julia-actions/setup-julia@v1
with:
version: ${{ matrix.julia-version }}
- uses: julia-actions/julia-buildpkg@latest
- uses: julia-actions/julia-runtest@latest
version: ${{ matrix.version }}
arch: ${{ matrix.arch }}
- name: Run tests
shell: julia --color=yes --project=test {0}
run: |
using Pkg
Pkg.develop(path=".")
Pkg.update()
using UnitCommitmentT
try
runtests()
catch
exit(1)
end

@ -4,20 +4,8 @@
VERSION := 0.3
clean:
rm -rfv build Manifest.toml test/Manifest.toml deps/formatter/build deps/formatter/Manifest.toml
docs:
cd docs; julia --project=. make.jl; cd ..
rsync -avP --delete-after docs/build/ ../docs/$(VERSION)/
format:
cd deps/formatter; ../../juliaw format.jl
test: test/Manifest.toml
./juliaw test/runtests.jl
test/Manifest.toml: test/Project.toml
julia --project=test -e "using Pkg; Pkg.instantiate()"
.PHONY: docs test format install-deps
.PHONY: docs

@ -18,6 +18,8 @@ PackageCompiler = "9b87118b-4619-50d2-8e1e-99f35a4d4d9d"
Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
TimerOutputs = "a759f4b9-e2f1-59dc-863e-4aeb61b1ea8f"
MPI = "da04e1cc-30fd-572f-bb4f-1f8673147195"
[compat]
DataStructures = "0.18"
@ -26,5 +28,7 @@ GZip = "0.5"
JSON = "0.21"
JuMP = "1"
MathOptInterface = "1"
MPI = "0.20"
PackageCompiler = "1"
julia = "1"
TimerOutputs = "0.5"

@ -1,5 +0,0 @@
[deps]
JuliaFormatter = "98e50ef6-434e-11e9-1051-2b60c6c9e899"
[compat]
JuliaFormatter = "0.14.4"

@ -1,9 +0,0 @@
using JuliaFormatter
format(
[
"../../src",
"../../test",
"../../benchmark/run.jl",
],
verbose=true,
)

@ -4,7 +4,7 @@ Data Format
Input Data Format
-----------------
Instances are specified by JSON files containing the following main sections:
An instance of the stochastic security-constrained unit commitment (SCUC) problem is composed multiple scenarios. Each scenario should be described in an individual JSON file containing the main section belows. For deterministic instances, a single scenario file, following the same format below, may also be provided. Fields that are allowed to differ among scenarios are marked as "uncertain". Fields that are allowed to be time-dependent are marked as "time series".
* [Parameters](#Parameters)
* [Buses](#Buses)
@ -20,12 +20,14 @@ Each section is described in detail below. See [case118/2017-01-01.json.gz](http
This section describes system-wide parameters, such as power balance penalty, and optimization parameters, such as the length of the planning horizon and the time.
| Key | Description | Default | Time series?
| :----------------------------- | :------------------------------------------------ | :------: | :------------:
| `Version` | Version of UnitCommitment.jl this file was written for. Required to ensure that the file remains readable in future versions of the package. If you are following this page to construct the file, this field should equal `0.3`. | Required | N
| `Time horizon (h)` | Length of the planning horizon (in hours). | Required | N
| `Time step (min)` | Length of each time step (in minutes). Must be a divisor of 60 (e.g. 60, 30, 20, 15, etc). | `60` | N
| `Power balance penalty ($/MW)` | Penalty for system-wide shortage or surplus in production (in $/MW). This is charged per time step. For example, if there is a shortage of 1 MW for three time steps, three times this amount will be charged. | `1000.0` | Y
| Key | Description | Default | Time series? | Uncertain?
| :----------------------------- | :------------------------------------------------ | :------: | :------------:| :----------:
| `Version` | Version of UnitCommitment.jl this file was written for. Required to ensure that the file remains readable in future versions of the package. If you are following this page to construct the file, this field should equal `0.3`. | Required | No | No
| `Time horizon (h)` | Length of the planning horizon (in hours). | Required | No | No
| `Time step (min)` | Length of each time step (in minutes). Must be a divisor of 60 (e.g. 60, 30, 20, 15, etc). | `60` | No | No
| `Power balance penalty ($/MW)` | Penalty for system-wide shortage or surplus in production (in $/MW). This is charged per time step. For example, if there is a shortage of 1 MW for three time steps, three times this amount will be charged. | `1000.0` | No | Yes
| `Scenario name` | Name of the scenario. | `"s1"` | No | ---
| `Scenario weight` | Weight of the scenario. The scenario weight can be any positive real number, that is, it does not have to be between zero and one. The package normalizes the weights to ensure that the probability of all scenarios sum up to one. | 1.0 | No | ---
#### Example
@ -34,7 +36,9 @@ This section describes system-wide parameters, such as power balance penalty, an
"Parameters": {
"Version": "0.3",
"Time horizon (h)": 4,
"Power balance penalty ($/MW)": 1000.0
"Power balance penalty ($/MW)": 1000.0,
"Scenario name": "s1",
"Scenario weight": 0.5
}
}
```
@ -43,9 +47,9 @@ This section describes system-wide parameters, such as power balance penalty, an
This section describes the characteristics of each bus in the system.
| Key | Description | Default | Time series?
| :----------------- | :------------------------------------------------------------ | ------- | :-------------:
| `Load (MW)` | Fixed load connected to the bus (in MW). | Required | Y
| Key | Description | Default | Time series? | Uncertain?
| :----------------- | :------------------------------------------------------------ | ------- | :-----------: | :---:
| `Load (MW)` | Fixed load connected to the bus (in MW). | Required | Yes | Yes
#### Example
@ -77,33 +81,33 @@ This section describes all generators in the system. Two types of units can be s
#### Thermal Units
| Key | Description | Default | Time series?
| :------------------------ | :------------------------------------------------| ------- | :-----------:
| `Bus` | Identifier of the bus where this generator is located (string). | Required | N
| `Type` | Type of the generator (string). For thermal generators, this must be `Thermal`. | Required | N
| `Production cost curve (MW)` and `Production cost curve ($)` | Parameters describing the piecewise-linear production costs. See below for more details. | Required | Y
| `Startup costs ($)` and `Startup delays (h)` | Parameters describing how much it costs to start the generator after it has been shut down for a certain amount of time. If `Startup costs ($)` and `Startup delays (h)` are set to `[300.0, 400.0]` and `[1, 4]`, for example, and the generator is shut down at time `00:00` (h:min), then it costs \$300 to start up the generator at any time between `01:00` and `03:59`, and \$400 to start the generator at time `04:00` or any time after that. The number of startup cost points is unlimited, and may be different for each generator. Startup delays must be strictly increasing and the first entry must equal `Minimum downtime (h)`. | `[0.0]` and `[1]` | N
| `Minimum uptime (h)` | Minimum amount of time the generator must stay operational after starting up (in hours). For example, if the generator starts up at time `00:00` (h:min) and `Minimum uptime (h)` is set to 4, then the generator can only shut down at time `04:00`. | `1` | N
| `Minimum downtime (h)` | Minimum amount of time the generator must stay offline after shutting down (in hours). For example, if the generator shuts down at time `00:00` (h:min) and `Minimum downtime (h)` is set to 4, then the generator can only start producing power again at time `04:00`. | `1` | N
| `Ramp up limit (MW)` | Maximum increase in production from one time step to the next (in MW). For example, if the generator is producing 100 MW at time step 1 and if this parameter is set to 40 MW, then the generator will produce at most 140 MW at time step 2. | `+inf` | N
| `Ramp down limit (MW)` | Maximum decrease in production from one time step to the next (in MW). For example, if the generator is producing 100 MW at time step 1 and this parameter is set to 40 MW, then the generator will produce at least 60 MW at time step 2. | `+inf` | N
| `Startup limit (MW)` | Maximum amount of power a generator can produce immediately after starting up (in MW). For example, if `Startup limit (MW)` is set to 100 MW and the unit is off at time step 1, then it may produce at most 100 MW at time step 2.| `+inf` | N
| `Shutdown limit (MW)` | Maximum amount of power a generator can produce immediately before shutting down (in MW). Specifically, the generator can only shut down at time step `t+1` if its production at time step `t` is below this limit. | `+inf` | N
| `Initial status (h)` | If set to a positive number, indicates the amount of time (in hours) the generator has been on at the beginning of the simulation, and if set to a negative number, the amount of time the generator has been off. For example, if `Initial status (h)` is `-2`, this means that the generator was off since `-02:00` (h:min). The simulation starts at time `00:00`. If `Initial status (h)` is `3`, this means that the generator was on since `-03:00`. A value of zero is not acceptable. | Required | N
| `Initial power (MW)` | Amount of power the generator at time step `-1`, immediately before the planning horizon starts. | Required | N
| `Must run?` | If `true`, the generator should be committed, even if that is not economical (Boolean). | `false` | Y
| `Reserve eligibility` | List of reserve products this generator is eligibe to provide. By default, the generator is not eligible to provide any reserves. | `[]` | N
| `Commitment status` | List of commitment status over the time horizon. At time `t`, if `true`, the generator must be commited at that time period; if `false`, the generator must not be commited at that time period. If `null` at time `t`, the generator's commitment status is then decided by the model. By default, the status is a list of `null` values. | `null` | Y
| Key | Description | Default | Time series? | Uncertain?
| :------------------------ | :------------------------------------------------| ------- | :-----------: | :---:
| `Bus` | Identifier of the bus where this generator is located (string). | Required | No | Yes
| `Type` | Type of the generator (string). For thermal generators, this must be `Thermal`. | Required | No | No
| `Production cost curve (MW)` and `Production cost curve ($)` | Parameters describing the piecewise-linear production costs. See below for more details. | Required | Yes | Yes
| `Startup costs ($)` and `Startup delays (h)` | Parameters describing how much it costs to start the generator after it has been shut down for a certain amount of time. If `Startup costs ($)` and `Startup delays (h)` are set to `[300.0, 400.0]` and `[1, 4]`, for example, and the generator is shut down at time `00:00` (h:min), then it costs \$300 to start up the generator at any time between `01:00` and `03:59`, and \$400 to start the generator at time `04:00` or any time after that. The number of startup cost points is unlimited, and may be different for each generator. Startup delays must be strictly increasing and the first entry must equal `Minimum downtime (h)`. | `[0.0]` and `[1]` | No | Yes
| `Minimum uptime (h)` | Minimum amount of time the generator must stay operational after starting up (in hours). For example, if the generator starts up at time `00:00` (h:min) and `Minimum uptime (h)` is set to 4, then the generator can only shut down at time `04:00`. | `1` | No | Yes
| `Minimum downtime (h)` | Minimum amount of time the generator must stay offline after shutting down (in hours). For example, if the generator shuts down at time `00:00` (h:min) and `Minimum downtime (h)` is set to 4, then the generator can only start producing power again at time `04:00`. | `1` | No | Yes
| `Ramp up limit (MW)` | Maximum increase in production from one time step to the next (in MW). For example, if the generator is producing 100 MW at time step 1 and if this parameter is set to 40 MW, then the generator will produce at most 140 MW at time step 2. | `+inf` | No | Yes
| `Ramp down limit (MW)` | Maximum decrease in production from one time step to the next (in MW). For example, if the generator is producing 100 MW at time step 1 and this parameter is set to 40 MW, then the generator will produce at least 60 MW at time step 2. | `+inf` | No | Yes
| `Startup limit (MW)` | Maximum amount of power a generator can produce immediately after starting up (in MW). For example, if `Startup limit (MW)` is set to 100 MW and the unit is off at time step 1, then it may produce at most 100 MW at time step 2.| `+inf` | No | Yes
| `Shutdown limit (MW)` | Maximum amount of power a generator can produce immediately before shutting down (in MW). Specifically, the generator can only shut down at time step `t+1` if its production at time step `t` is below this limit. | `+inf` | No | Yes
| `Initial status (h)` | If set to a positive number, indicates the amount of time (in hours) the generator has been on at the beginning of the simulation, and if set to a negative number, the amount of time the generator has been off. For example, if `Initial status (h)` is `-2`, this means that the generator was off since `-02:00` (h:min). The simulation starts at time `00:00`. If `Initial status (h)` is `3`, this means that the generator was on since `-03:00`. A value of zero is not acceptable. | Required | No | No
| `Initial power (MW)` | Amount of power the generator at time step `-1`, immediately before the planning horizon starts. | Required | No | No
| `Must run?` | If `true`, the generator should be committed, even if that is not economical (Boolean). | `false` | Yes | Yes
| `Reserve eligibility` | List of reserve products this generator is eligibe to provide. By default, the generator is not eligible to provide any reserves. | `[]` | No | Yes
| `Commitment status` | List of commitment status over the time horizon. At time `t`, if `true`, the generator must be commited at that time period; if `false`, the generator must not be commited at that time period. If `null` at time `t`, the generator's commitment status is then decided by the model. By default, the status is a list of `null` values. | `null` | Yes | Yes
#### Profiled Units
| Key | Description | Default | Time series?
| :---------------- | :------------------------------------------------ | :------: | :------------:
| `Bus` | Identifier of the bus where this generator is located (string). | Required | N
| `Type` | Type of the generator (string). For profiled generators, this must be `Profiled`. | Required | N
| `Cost ($/MW)` | Cost incurred for serving each MW of power by this generator. | Required | Y
| `Minimum power (MW)` | Minimum amount of power this generator may supply. | `0.0` | Y
| `Maximum power (MW)` | Maximum amount of power this generator may supply. | Required | Y
| Key | Description | Default | Time series? | Uncertain?
| :---------------- | :------------------------------------------------ | :------: | :------------: | :---:
| `Bus` | Identifier of the bus where this generator is located (string). | Required | No | Yes
| `Type` | Type of the generator (string). For profiled generators, this must be `Profiled`. | Required | No | No
| `Cost ($/MW)` | Cost incurred for serving each MW of power by this generator. | Required | Yes | Yes
| `Minimum power (MW)` | Minimum amount of power this generator may supply. | `0.0` | Yes | Yes
| `Maximum power (MW)` | Maximum amount of power this generator may supply. | Required | Yes | Yes
#### Production costs and limits
@ -173,11 +177,11 @@ Note that this curve also specifies the production limits. Specifically, the fir
This section describes components in the system which may increase or reduce their energy consumption according to the energy prices. Fixed loads (as described in the `buses` section) are always served, regardless of the price, unless there is significant congestion in the system or insufficient production capacity. Price-sensitive loads, on the other hand, are only served if it is economical to do so.
| Key | Description | Default | Time series?
| :---------------- | :------------------------------------------------ | :------: | :------------:
| `Bus` | Bus where the load is located. Multiple price-sensitive loads may be placed at the same bus. | Required | N
| `Revenue ($/MW)` | Revenue obtained for serving each MW of power to this load. | Required | Y
| `Demand (MW)` | Maximum amount of power required by this load. Any amount lower than this may be served. | Required | Y
| Key | Description | Default | Time series? | Uncertain?
| :---------------- | :------------------------------------------------ | :------: | :------------: | :----:
| `Bus` | Bus where the load is located. Multiple price-sensitive loads may be placed at the same bus. | Required | No | Yes
| `Revenue ($/MW)` | Revenue obtained for serving each MW of power to this load. | Required | Yes | Yes
| `Demand (MW)` | Maximum amount of power required by this load. Any amount lower than this may be served. | Required | Yes | Yes
#### Example
@ -197,15 +201,15 @@ This section describes components in the system which may increase or reduce the
This section describes the characteristics of transmission system, such as its topology and the susceptance of each transmission line.
| Key | Description | Default | Time series?
| :--------------------- | :----------------------------------------------- | ------- | :------------:
| `Source bus` | Identifier of the bus where the transmission line originates. | Required | N
| `Target bus` | Identifier of the bus where the transmission line reaches. | Required | N
| `Reactance (ohms)` | Reactance of the transmission line (in ohms). | Required | N
| `Susceptance (S)` | Susceptance of the transmission line (in siemens). | Required | N
| `Normal flow limit (MW)` | Maximum amount of power (in MW) allowed to flow through the line when the system is in its regular, fully-operational state. | `+inf` | Y
| `Emergency flow limit (MW)` | Maximum amount of power (in MW) allowed to flow through the line when the system is in degraded state (for example, after the failure of another transmission line). | `+inf` | Y
| `Flow limit penalty ($/MW)` | Penalty for violating the flow limits of the transmission line (in $/MW). This is charged per time step. For example, if there is a thermal violation of 1 MW for three time steps, then three times this amount will be charged. | `5000.0` | Y
| Key | Description | Default | Time series? | Uncertain?
| :--------------------- | :----------------------------------------------- | ------- | :------------: | :---:
| `Source bus` | Identifier of the bus where the transmission line originates. | Required | No | Yes
| `Target bus` | Identifier of the bus where the transmission line reaches. | Required | No | Yes
| `Reactance (ohms)` | Reactance of the transmission line (in ohms). | Required | No | Yes
| `Susceptance (S)` | Susceptance of the transmission line (in siemens). | Required | No | Yes
| `Normal flow limit (MW)` | Maximum amount of power (in MW) allowed to flow through the line when the system is in its regular, fully-operational state. | `+inf` | Yes | Yes
| `Emergency flow limit (MW)` | Maximum amount of power (in MW) allowed to flow through the line when the system is in degraded state (for example, after the failure of another transmission line). | `+inf` | Y | Yes
| `Flow limit penalty ($/MW)` | Penalty for violating the flow limits of the transmission line (in $/MW). This is charged per time step. For example, if there is a thermal violation of 1 MW for three time steps, then three times this amount will be charged. | `5000.0` | Yes | Yes
#### Example
@ -231,11 +235,11 @@ This section describes the characteristics of transmission system, such as its t
This section describes the hourly amount of reserves required.
| Key | Description | Default | Time series?
| :-------------------- | :------------------------------------------------- | --------- | :----:
| `Type` | Type of reserve product. Must be either "spinning" or "flexiramp". | Required | N
| `Amount (MW)` | Amount of reserves required. | Required | Y
| `Shortfall penalty ($/MW)` | Penalty for shortage in meeting the reserve requirements (in $/MW). This is charged per time step. Negative value implies reserve constraints must always be satisfied. | `-1` | Y
| Key | Description | Default | Time series? | Uncertain?
| :-------------------- | :------------------------------------------------- | --------- | :----: | :---:
| `Type` | Type of reserve product. Must be either "spinning" or "flexiramp". | Required | No | No
| `Amount (MW)` | Amount of reserves required. | Required | Yes | Yes
| `Shortfall penalty ($/MW)` | Penalty for shortage in meeting the reserve requirements (in $/MW). This is charged per time step. Negative value implies reserve constraints must always be satisfied. | `-1` | Yes | Yes
#### Example 1
@ -269,10 +273,10 @@ This section describes the hourly amount of reserves required.
This section describes credible contingency scenarios in the optimization, such as the loss of a transmission line or generator.
| Key | Description | Default
| :-------------------- | :----------------------------------------------- | ----------
| `Affected generators` | List of generators affected by this contingency. May be omitted if no generators are affected. | `[]`
| `Affected lines` | List of transmission lines affected by this contingency. May be omitted if no lines are affected. | `[]`
| Key | Description | Default | Uncertain?
| :-------------------- | :----------------------------------------------- | :--------: | :---:
| `Affected generators` | List of generators affected by this contingency. May be omitted if no generators are affected. | `[]` | Yes
| `Affected lines` | List of transmission lines affected by this contingency. May be omitted if no lines are affected. | `[]` | Yes
#### Example
@ -323,8 +327,8 @@ The output data format is also JSON-based, but it is not currently documented si
Current limitations
-------------------
* Network topology remains the same for all time periods
* Network topology must remain the same for all time periods.
* Only N-1 transmission contingencies are supported. Generator contingencies are not currently supported.
* Time-varying minimum production amounts are not currently compatible with ramp/startup/shutdown limits.
* Flexible ramping products can only be acquired under the `WanHob2016` formulation, which does not support spinning reserves.
* The set of generators must be the same in all scenarios.

@ -1,12 +1,12 @@
# UnitCommitment.jl
**UnitCommitment.jl** (UC.jl) is a Julia/JuMP optimization package for the Security-Constrained Unit Commitment Problem (SCUC), a fundamental optimization problem in power systems used, for example, to clear the day-ahead electricity markets. The package provides benchmark instances for the problem and Julia/JuMP implementations of state-of-the-art mixed-integer programming formulations.
**UnitCommitment.jl** (UC.jl) is an optimization package for the Security-Constrained Unit Commitment Problem (SCUC), a fundamental optimization problem in power systems used, for example, to clear the day-ahead electricity markets. Both deterministic and two-stage stochastic versions of the problem are supported. The package provides benchmark instances for the problem, a flexible and well-documented data format for the problem, as well as Julia/JuMP implementations of state-of-the-art mixed-integer programming formulations and solution methods.
## Package Components
* **Data Format:** The package proposes an extensible and fully-documented JSON-based data specification format for SCUC, developed in collaboration with Independent System Operators (ISOs), which describes the most important aspects of the problem. The format supports all the most common generator characteristics (including ramping, piecewise-linear production cost curves and time-dependent startup costs), as well as operating reserves, price-sensitive loads, transmission networks and contingencies.
* **Benchmark Instances:** The package provides a diverse collection of large-scale benchmark instances collected from the literature, converted into a common data format, and extended using data-driven methods to make them more challenging and realistic.
* **Model Implementation**: The package provides a Julia/JuMP implementations of state-of-the-art formulations and solution methods for SCUC, including multiple ramping formulations ([ArrCon2000](https://doi.org/10.1109/59.871739), [MorLatRam2013](https://doi.org/10.1109/TPWRS.2013.2251373), [DamKucRajAta2016](https://doi.org/10.1007/s10107-015-0919-9), [PanGua2016](https://doi.org/10.1287/opre.2016.1520)), multiple piecewise-linear costs formulations ([Gar1962](https://doi.org/10.1109/AIEEPAS.1962.4501405), [CarArr2006](https://doi.org/10.1109/TPWRS.2006.876672), [KnuOstWat2018](https://doi.org/10.1109/TPWRS.2017.2783850)) and contingency screening methods ([XavQiuWanThi2019](https://doi.org/10.1109/TPWRS.2019.2892620)). Our goal is to keep these implementations up-to-date as new methods are proposed in the literature.
* **Model Implementation**: The package provides a Julia/JuMP implementations of state-of-the-art formulations and solution methods for the deterministic and stochastic SCUC, including multiple ramping formulations ([ArrCon2000](https://doi.org/10.1109/59.871739), [MorLatRam2013](https://doi.org/10.1109/TPWRS.2013.2251373), [DamKucRajAta2016](https://doi.org/10.1007/s10107-015-0919-9), [PanGua2016](https://doi.org/10.1287/opre.2016.1520)), multiple piecewise-linear costs formulations ([Gar1962](https://doi.org/10.1109/AIEEPAS.1962.4501405), [CarArr2006](https://doi.org/10.1109/TPWRS.2006.876672), [KnuOstWat2018](https://doi.org/10.1109/TPWRS.2017.2783850)) and contingency screening methods ([XavQiuWanThi2019](https://doi.org/10.1109/TPWRS.2019.2892620)). Our goal is to keep these implementations up-to-date as new methods are proposed in the literature.
* **Benchmark Tools:** The package provides automated benchmark scripts to accurately evaluate the performance impact of proposed code changes.
## Table of Contents
@ -35,7 +35,7 @@ Depth = 3
If you use UnitCommitment.jl in your research (instances, models or algorithms), we kindly request that you cite the package as follows:
* **Alinson S. Xavier, Aleksandr M. Kazachkov, Ogün Yurdakul, Feng Qiu**, "UnitCommitment.jl: A Julia/JuMP Optimization Package for Security-Constrained Unit Commitment (Version 0.3)". Zenodo (2022). [DOI: 10.5281/zenodo.4269874](https://doi.org/10.5281/zenodo.4269874).
* **Alinson S. Xavier, Aleksandr M. Kazachkov, Ogün Yurdakul, Feng Qiu**, "UnitCommitment.jl: A Julia/JuMP Optimization Package for Security-Constrained Unit Commitment (Version 0.4)". Zenodo (2022). [DOI: 10.5281/zenodo.4269874](https://doi.org/10.5281/zenodo.4269874).
If you use the instances, we additionally request that you cite the original sources, as described in the [instances page](instances.md).
@ -43,7 +43,7 @@ If you use the instances, we additionally request that you cite the original sou
```text
UnitCommitment.jl: A Julia/JuMP Optimization Package for Security-Constrained Unit Commitment
Copyright © 2020-2022, UChicago Argonne, LLC. All Rights Reserved.
Copyright © 2020-2023, UChicago Argonne, LLC. All Rights Reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:

@ -1,57 +1,69 @@
JuMP Model
==========
In this page, we describe the JuMP optimization model produced by the function `UnitCommitment.build_model`. A detailed understanding of this model is not necessary if you are just interested in using the package to solve some standard unit commitment cases, but it may be useful, for example, if you need to solve a slightly different problem, with additional variables and constraints. The notation in this page generally follows [KnOsWa20].
In this page, we describe the JuMP optimization model produced by the function `build_model`. A detailed understanding of this model is not necessary if you are just interested in using the package to solve some standard unit commitment cases, but it may be useful, for example, if you need to solve a slightly different problem, with additional variables and constraints. The notation in this page generally follows [KnOsWa20].
Decision variables
------------------
UC.jl models the security-constrained unit commitment problem as a two-stage stochastic program. In this approach, some of the decision variables are *first-stage decisions*, which are taken before the uncertainty is realized and must therefore be the same across all scenarios, while the remaining variables are *second-stage decisions*, which can attain a different values in each scenario. In the current version of the package, all binary variables (which model commitment decisions of thermal units) are first-stage decisions and all continuous variables are second-stage decisions.
!!! note
UC.jl treats deterministic SCUC instances as a special case of the stochastic problem in which there is only one scenario, named `"s1"` by default. To access second-stage decisions, therefore, you must provide this scenario name as the value for `sn`. For example, `model[:prod_above]["s1", g, t]`.
### Generators
In this section, we describe the decision variables associated with the generators, which include both thermal units (e.g., natural gas-fired power plant) and profiled units (e.g., wind turbine).
#### Thermal Units
Name | Symbol | Description | Unit
:-----|:--------:|:-------------|:------:
`is_on[g,t]` | $u_{g}(t)$ | True if generator `g` is on at time `t`. | Binary
`switch_on[g,t]` | $v_{g}(t)$ | True is generator `g` switches on at time `t`. | Binary
`switch_off[g,t]` | $w_{g}(t)$ | True if generator `g` switches off at time `t`. | Binary
`prod_above[g,t]` |$p'_{g}(t)$ | Amount of power produced by generator `g` above its minimum power output at time `t`. For example, if the minimum power of generator `g` is 100 MW and `g` is producing 115 MW of power at time `t`, then `prod_above[g,t]` equals `15.0`. | MW
`segprod[g,t,k]` | $p^k_g(t)$ | Amount of power from piecewise linear segment `k` produced by generator `g` at time `t`. For example, if cost curve for generator `g` is defined by the points `(100, 1400)`, `(110, 1600)`, `(130, 2200)` and `(135, 2400)`, and if the generator is producing 115 MW of power at time `t`, then `segprod[g,t,:]` equals `[10.0, 5.0, 0.0]`.| MW
`reserve[r,g,t]` | $r_g(t)$ | Amount of reserve `r` provided by unit `g` at time `t`. | MW
`startup[g,t,s]` | $\delta^s_g(t)$ | True if generator `g` switches on at time `t` incurring start-up costs from start-up category `s`. | Binary
Name | Description | Unit | Stage
:-----|:-------------|:------: | :------:
`is_on[g,t]` | True if generator `g` is on at time `t`. | Binary | 1
`switch_on[g,t]` | True is generator `g` switches on at time `t`. | Binary| 1
`switch_off[g,t]` | True if generator `g` switches off at time `t`. | Binary| 1
`startup[g,t,s]` | True if generator `g` switches on at time `t` incurring start-up costs from start-up category `s`. | Binary| 1
`prod_above[sn,g,t]` | Amount of power produced by generator `g` above its minimum power output at time `t` in scenario `sn`. For example, if the minimum power of generator `g` is 100 MW and `g` is producing 115 MW of power at time `t` in scenario `sn`, then `prod_above[sn,g,t]` equals `15.0`. | MW | 2
`segprod[sn,g,t,k]` | Amount of power from piecewise linear segment `k` produced by generator `g` at time `t` in scenario `sn`. For example, if cost curve for generator `g` is defined by the points `(100, 1400)`, `(110, 1600)`, `(130, 2200)` and `(135, 2400)`, and if the generator is producing 115 MW of power at time `t` in scenario `sn`, then `segprod[sn,g,t,:]` equals `[10.0, 5.0, 0.0]`.| MW | 2
`reserve[sn,r,g,t]` | Amount of reserve `r` provided by unit `g` at time `t` in scenario `sn`. | MW | 2
!!! warning
The first-stage decision variables of the JuMP model are `is_on[g,t]`, `switch_on[g,t]`, `switch_off[g,t]`, and `startup[g,t,s]`. As such, the dictionaries corresponding to these variables do not include the scenario index in their keys. In contrast, all other variables of the created JuMP model are allowed to obtain a different value in each scenario and are thus modeled as second-stage decision variables. Accordingly, the dictionaries of all second-stage decision variables have the scenario index in their keys. This is true even if the model is created to solve the deterministic SCUC, in which case the default scenario index `s1` is included in the dictionary key.
#### Profiled Units
Name | Symbol | Description | Unit
:-----|:------:|:-------------|:------:
`prod_profiled[s,t]` | $p^{\dagger}_{g}(t)$ | Amount of power produced by profiled unit `g` at time `t`. | MW
Name | Description | Unit | Stage
:-----|:-------------|:------: | :------:
`prod_profiled[s,t]` | Amount of power produced by profiled unit `g` at time `t`. | MW | 2
### Buses
Name | Symbol | Description | Unit
:-----|:------:|:-------------|:------:
`net_injection[b,t]` | $n_b(t)$ | Net injection at bus `b` at time `t`. | MW
`curtail[b,t]` | $s^+_b(t)$ | Amount of load curtailed at bus `b` at time `t` | MW
Name | Description | Unit | Stage
:-----|:-------------|:------:| :------:
`net_injection[sn,b,t]` | Net injection at bus `b` at time `t` in scenario `sn`. | MW | 2
`curtail[sn,b,t]` | Amount of load curtailed at bus `b` at time `t` in scenario `sn`. | MW | 2
### Price-sensitive loads
Name | Symbol | Description | Unit
:-----|:------:|:-------------|:------:
`loads[s,t]` | $d_{s}(t)$ | Amount of power served to price-sensitive load `s` at time `t`. | MW
Name | Description | Unit | Stage
:-----|:-------------|:------:| :------:
`loads[sn,s,t]` | Amount of power served to price-sensitive load `s` at time `t` in scenario `sn`. | MW | 2
### Transmission lines
Name | Symbol | Description | Unit
:-----|:------:|:-------------|:------:
`flow[l,t]` | $f_l(t)$ | Power flow on line `l` at time `t`. | MW
`overflow[l,t]` | $f^+_l(t)$ | Amount of flow above the limit for line `l` at time `t`. | MW
Name | Description | Unit | Stage
:-----|:-------------|:------:| :------:
`flow[sn,l,t]` | Power flow on line `l` at time `t` in scenario `sn`. | MW | 2
`overflow[sn,l,t]` | Amount of flow above the limit for line `l` at time `t` in scenario `sn`. | MW | 2
!!! warning
Since transmission and N-1 security constraints are enforced in a lazy way, most of the `flow[l,t]` variables are never added to the model. Accessing `model[:flow][l,t]` without first checking that the variable exists will likely generate an error.
Since transmission and N-1 security constraints are enforced in a lazy way, most of the `flow[l,t]` variables are never added to the model. Accessing `model[:flow][sn,l,t]` without first checking that the variable exists will likely generate an error.
Objective function
------------------
@ -106,7 +118,12 @@ end
### Fixing variables, modifying objective function and adding constraints
Since we now have a direct reference to the JuMP decision variables, it is possible to fix variables, change the coefficients in the objective function, or even add new constraints to the model before solving it. The script below shows how can this be accomplished. For more information on modifying an existing model, [see the JuMP documentation](https://jump.dev/JuMP.jl/stable/manual/variables/).
Since we now have a direct reference to the JuMP decision variables, it is possible to fix variables, change the coefficients in the objective function, or even add new constraints to the model before solving it.
!!! warning
It is important to take into account the stage of the decision variables in modifying the optimization model. In changing a deterministic SCUC model, modifying the second-stage decision variables requires adding the term `s1`, which is the default scenario name assigned to the second-stage decision variables in the SCUC model. For an SUC model, the package permits the modification of the second-stage decision variables individually for each scenario.
The script below shows how the JuMP model can be modified after it is created. For more information on modifying an existing model, [see the JuMP documentation](https://jump.dev/JuMP.jl/stable/manual/variables/).
```julia
using Cbc
@ -122,13 +139,29 @@ model = UnitCommitment.build_model(
optimizer=Cbc.Optimizer,
)
# Fix a decision variable to 1.0
# Fix the commitment status of the generator "g1" in time period 1 to 1.0
JuMP.fix(
model[:is_on]["g1",1],
1.0,
force=true,
)
# Fix the production level of the generator "g1" above its minimum level in time period 1 and
# in scenario "s1" to 20.0 MW. Observe that the three-tuple dictionary key involves the scenario
# index "s1", as production above minimum is a second-stage decision variable.
JuMP.fix(
model[:prod_above]["s1", "g1", 1],
20.0,
force=true,
)
# Enforce the curtailment of 20.0 MW of load at bus "b2" in time period 4 in scenario "s1".
JuMP.fix(
curtail["s1", "b2", 4] =
20.0,
force=true,
)
# Change the objective function
JuMP.set_objective_coefficient(
model,
@ -178,10 +211,10 @@ for t in 1:T
# In this example, we assume a cost of $5/MW.
set_objective_coefficient(model, x[t], 5.0)
# Attach the new component to bus b1, by modifying the
# Attach the new component to bus b1 in scenario s1, by modifying the
# constraint `eq_net_injection`.
set_normalized_coefficient(
model[:eq_net_injection]["b1", t],
model[:eq_net_injection]["s1", "b1", t],
x[t],
1.0,
)

@ -4,28 +4,24 @@ Usage
Installation
------------
UnitCommitment.jl was tested and developed with [Julia 1.7](https://julialang.org/). To install Julia, please follow the [installation guide on the official Julia website](https://julialang.org/downloads/). To install UnitCommitment.jl, run the Julia interpreter, type `]` to open the package manager, then type:
UnitCommitment.jl was tested and developed with [Julia 1.9](https://julialang.org/). To install Julia, please follow the [installation guide on the official Julia website](https://julialang.org/downloads/). To install UnitCommitment.jl, run the Julia interpreter, type `]` to open the package manager, then type:
```text
pkg> add UnitCommitment@0.3
pkg> add UnitCommitment@0.4
```
To test that the package has been correctly installed, run:
```text
pkg> test UnitCommitment
```
If all tests pass, the package should now be ready to be used by any Julia script on the machine.
To solve the optimization models, a mixed-integer linear programming (MILP) solver is also required. Please see the [JuMP installation guide](https://jump.dev/JuMP.jl/stable/installation/) for more instructions on installing a solver. Typical open-source choices are [Cbc](https://github.com/JuliaOpt/Cbc.jl) and [GLPK](https://github.com/JuliaOpt/GLPK.jl). In the instructions below, Cbc will be used, but any other MILP solver listed in JuMP installation guide should also be compatible.
To solve the optimization models, a mixed-integer linear programming (MILP) solver is also required. Please see the [JuMP installation guide](https://jump.dev/JuMP.jl/stable/installation/) for more instructions on installing a solver. Typical open-source choices are [HiGHS](https://github.com/jump-dev/HiGHS.jl), [Cbc](https://github.com/JuliaOpt/Cbc.jl) and [GLPK](https://github.com/JuliaOpt/GLPK.jl). In the instructions below, Cbc will be used, but any other MILP solver listed in JuMP installation guide should also be compatible.
Typical Usage
-------------
### Solving user-provided instances
The first step to use UC.jl is to construct a JSON file describing your unit commitment instance. See [Data Format](format.md) for a complete description of the data format UC.jl expects. The next steps, as shown below, are to: (1) read the instance from file; (2) construct the optimization model; (3) run the optimization; and (4) extract the optimal solution.
The first step to use UC.jl is to construct JSON files that describe each scenario of your stochastic unit commitment instance. See [Data Format](format.md) for a complete description of the data format UC.jl expects. The next steps, as shown below, are to: (1) read the scenario files; (2) build the optimization model; (3) run the optimization; and (4) extract the optimal solution.
!!! note
> By default, UC.jl uses the extensive form to solve the problem. For a more advanced solution method, see below.
```julia
using Cbc
@ -33,7 +29,7 @@ using JSON
using UnitCommitment
# 1. Read instance
instance = UnitCommitment.read("/path/to/input.json")
instance = UnitCommitment.read(["/path/to/s1.json", "/path/to/s2.json"])
# 2. Construct optimization model
model = UnitCommitment.build_model(
@ -49,12 +45,24 @@ solution = UnitCommitment.solution(model)
UnitCommitment.write("/path/to/output.json", solution)
```
To read all files in a given folder, the [Glob](https://github.com/vtjnash/Glob.jl) package can be used:
```julia
using Glob
instance = UnitCommitment.read(glob("*.json", "/path/to/scenarios/"))
```
To solve deterministic instances, a single scenario file may be provided.
```julia
instance = UnitCommitment.read("/path/to/s1.json")
```
### Solving benchmark instances
UnitCommitment.jl contains a large number of benchmark instances collected from the literature and converted into a common data format. To solve one of these instances individually, instead of constructing your own, the function `read_benchmark` can be used, as shown below. See [Instances](instances.md) for the complete list of available instances.
UnitCommitment.jl contains a large number of deterministic benchmark instances collected from the literature and converted into a common data format. To solve one of these instances individually, instead of constructing your own, the function `read_benchmark` can be used, as shown below. See [Instances](instances.md) for the complete list of available instances.
```julia
using UnitCommitment
instance = UnitCommitment.read_benchmark("matpower/case3375wp/2017-02-01")
```
@ -137,6 +145,56 @@ solution = JSON.parsefile("solution.json")
UnitCommitment.validate(instance, solution)
```
## Progressive Hedging
By default, UC.jl uses the Extensive Form (EF) when solving stochastic instances. This approach involves constructing a single JuMP model that contains data and decision variables for all scenarios. Although EF has optimality guarantees and performs well with small test cases, it can become computationally intractable for large instances or substantial number of scenarios.
Progressive Hedging (PH) is an alternative (heuristic) solution method provided by UC.jl in which the problem is decomposed into smaller scenario-based subproblems, which are then solved in parallel in separate Julia processes, potentially across multiple machines. Quadratic penalty terms are used to enforce convergence of first-stage decision variables. The method is closely related to the Alternative Direction Method of Multipliers (ADMM) and can handle larger instances, although it is not guaranteed to converge to the optimal solution. Our implementation of PH relies on Message Passing Interface (MPI) for communication. We refer to [MPI.jl Documentation](https://github.com/JuliaParallel/MPI.jl) for more details on installing MPI.
The following example shows how to solve SCUC instances using progressive hedging. The script should be saved in a file, say `ph.jl`, and executed using `mpiexec -n <num-scenarios> julia ph.jl`.
```julia
using Cbc
using MPI
using UnitCommitment
using Glob
# 1. Initialize MPI
MPI.Init()
# 2. Configure progressive hedging method
ph = UnitCommitment.ProgressiveHedging()
# 3. Read problem instance
instance = UnitCommitment.read(["s1.json", "s2.json"], ph)
# 4. Build JuMP model
model = UnitCommitment.build_model(
instance = instance,
optimizer = Cbc.Optimizer,
)
# 5. Run the decentralized optimization algorithm
UnitCommitment.optimize!(model, ph)
# 6. Fetch the solution
solution = UnitCommitment.solution(model, ph)
# 7. Close MPI
MPI.Finalize()
```
When using PH, the model can be customized as usual, with a different formulations or additional user-provided constraints. Note that `read`, in this case, takes `ph` as an argument. This allows each Julia process to read only the instance files that are relevant to it. Similarly, the `solution` function gathers the optimal solution of each processes and returns a combined dictionary.
Each process solves a sub-problem with $\frac{s}{p}$ scenarios, where $s$ is the total number of scenarios and $p$ is the number of MPI processes. For instance, if we have 15 scenario files and 5 processes, then each process will solve a JuMP model that contains data for 3 scenarios. If the total number of scenarios is not divisible by the number of processes, then an error will be thrown.
!!! warning
Currently, PH can handle only equiprobable scenarios. Further, `solution(model, ph)` can only handle cases where only one scenario is modeled in each process.
## Computing Locational Marginal Prices
Locational marginal prices (LMPs) refer to the cost of supplying electricity at a particular location of the network. Multiple methods for computing LMPs have been proposed in the literature. UnitCommitment.jl implements two commonly-used methods: conventional LMPs and Approximated Extended LMPs (AELMPs). To compute LMPs for a given unit commitment instance, the `compute_lmp` function can be used, as shown in the examples below. The function accepts three arguments -- a solved SCUC model, an LMP method, and a linear optimizer -- and it returns a dictionary mapping `(bus_name, time)` to the marginal price.

@ -1,75 +0,0 @@
#!/bin/bash
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020-2021, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
if [ ! -e Project.toml ]; then
echo "juliaw: Project.toml not found"
exit 1
fi
if [ ! -e Manifest.toml ]; then
julia --project=. -e 'using Pkg; Pkg.instantiate()' || exit 1
fi
if [ ! -e build/sysimage.so -o Project.toml -nt build/sysimage.so ]; then
echo "juliaw: rebuilding system image..."
# Generate temporary project folder
rm -rf $HOME/.juliaw
mkdir -p $HOME/.juliaw/src
cp Project.toml Manifest.toml $HOME/.juliaw
NAME=$(julia -e 'using TOML; toml = TOML.parsefile("Project.toml"); "name" in keys(toml) && print(toml["name"])')
if [ ! -z $NAME ]; then
cat > $HOME/.juliaw/src/$NAME.jl << EOF
module $NAME
end
EOF
fi
# Add PackageCompiler dependencies to temporary project
julia --project=$HOME/.juliaw -e 'using Pkg; Pkg.add(["PackageCompiler", "TOML", "Logging"])'
# Generate system image scripts
cat > $HOME/.juliaw/sysimage.jl << EOF
using PackageCompiler
using TOML
using Logging
Logging.disable_logging(Logging.Info)
mkpath("$PWD/build")
println("juliaw: generating precompilation statements...")
run(\`julia --project="$PWD" --trace-compile="$PWD"/build/precompile.jl \$(ARGS)\`)
println("juliaw: finding dependencies...")
project = TOML.parsefile("Project.toml")
manifest = TOML.parsefile("Manifest.toml")
deps = Symbol[]
for dep in keys(project["deps"])
if dep in keys(manifest)
# Up to Julia 1.6
dep_entry = manifest[dep][1]
else
# Julia 1.7+
dep_entry = manifest["deps"][dep][1]
end
if "path" in keys(dep_entry)
println(" - \$(dep) [skip]")
else
println(" - \$(dep)")
push!(deps, Symbol(dep))
end
end
println("juliaw: building system image...")
create_sysimage(
deps,
precompile_statements_file = "$PWD/build/precompile.jl",
sysimage_path = "$PWD/build/sysimage.so",
)
EOF
julia --project=$HOME/.juliaw $HOME/.juliaw/sysimage.jl $*
else
julia --project=. --sysimage build/sysimage.so $*
fi

@ -19,6 +19,7 @@ include("model/formulations/KnuOstWat2018/structs.jl")
include("model/formulations/MorLatRam2013/structs.jl")
include("model/formulations/PanGua2016/structs.jl")
include("solution/methods/XavQiuWanThi2019/structs.jl")
include("solution/methods/ProgressiveHedging/structs.jl")
include("model/formulations/WanHob2016/structs.jl")
include("solution/methods/TimeDecomposition/structs.jl")
@ -51,6 +52,9 @@ include("solution/methods/XavQiuWanThi2019/filter.jl")
include("solution/methods/XavQiuWanThi2019/find.jl")
include("solution/methods/XavQiuWanThi2019/optimize.jl")
include("solution/methods/TimeDecomposition/optimize.jl")
include("solution/methods/ProgressiveHedging/optimize.jl")
include("solution/methods/ProgressiveHedging/read.jl")
include("solution/methods/ProgressiveHedging/solution.jl")
include("solution/optimize.jl")
include("solution/solution.jl")
include("solution/warmstart.jl")

@ -0,0 +1,230 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using MPI, Printf
using TimerOutputs
import JuMP
const to = TimerOutput()
function optimize!(model::JuMP.Model, method::ProgressiveHedging)::Nothing
mpi = MpiInfo(MPI.COMM_WORLD)
iterations = PHIterationInfo[]
consensus_vars = [var for var in all_variables(model) if is_binary(var)]
nvars = length(consensus_vars)
weights = ones(nvars)
if method.initial_weights !== nothing
weights = copy(method.initial_weights)
end
target = zeros(nvars)
if method.initial_target !== nothing
target = copy(method.initial_target)
end
params = PHSubProblemParams(
ρ = method.ρ,
λ = [method.λ for _ in 1:nvars],
target = target,
)
sp = PHSubProblem(model, model[:obj], consensus_vars, weights)
while true
iteration_time = @elapsed begin
solution = solve_subproblem(sp, params, method.inner_method)
MPI.Barrier(mpi.comm)
global_obj = compute_global_objective(mpi, solution)
target = compute_target(mpi, solution)
update_λ_and_residuals!(solution, params, target)
global_infeas = compute_global_infeasibility(solution, mpi)
global_residual = compute_global_residual(mpi, solution)
if has_numerical_issues(target)
break
end
end
total_elapsed_time =
compute_total_elapsed_time(iteration_time, iterations)
current_iteration = PHIterationInfo(
global_infeas = global_infeas,
global_obj = global_obj,
global_residual = global_residual,
iteration_number = length(iterations) + 1,
iteration_time = iteration_time,
sp_vals = solution.vals,
sp_obj = solution.obj,
target = target,
total_elapsed_time = total_elapsed_time,
)
push!(iterations, current_iteration)
print_progress(mpi, current_iteration, method.print_interval)
if should_stop(mpi, iterations, method.termination)
break
end
end
return
end
function compute_total_elapsed_time(
iteration_time::Float64,
iterations::Array{PHIterationInfo,1},
)::Float64
length(iterations) > 0 ?
current_total_time = last(iterations).total_elapsed_time :
current_total_time = 0
return current_total_time + iteration_time
end
function compute_global_objective(
mpi::MpiInfo,
s::PhSubProblemSolution,
)::Float64
global_obj = MPI.Allreduce(s.obj, MPI.SUM, mpi.comm)
global_obj /= mpi.nprocs
return global_obj
end
function compute_target(mpi::MpiInfo, s::PhSubProblemSolution)::Array{Float64,1}
sp_vals = s.vals
target = MPI.Allreduce(sp_vals, MPI.SUM, mpi.comm)
target = target / mpi.nprocs
return target
end
function compute_global_residual(mpi::MpiInfo, s::PhSubProblemSolution)::Float64
n_vars = length(s.vals)
local_residual_sum = abs.(s.residuals)
global_residual_sum = MPI.Allreduce(local_residual_sum, MPI.SUM, mpi.comm)
return sum(global_residual_sum) / n_vars
end
function compute_global_infeasibility(
solution::PhSubProblemSolution,
mpi::MpiInfo,
)::Float64
local_infeasibility = norm(solution.residuals)
global_infeas = MPI.Allreduce(local_infeasibility, MPI.SUM, mpi.comm)
return global_infeas
end
function solve_subproblem(
sp::PHSubProblem,
params::PHSubProblemParams,
method::SolutionMethod,
)::PhSubProblemSolution
G = length(sp.consensus_vars)
if norm(params.λ) < 1e-3
@objective(sp.mip, Min, sp.obj)
else
@objective(
sp.mip,
Min,
sp.obj +
sum(
sp.weights[g] *
params.λ[g] *
(sp.consensus_vars[g] - params.target[g]) for g in 1:G
) +
(params.ρ / 2) * sum(
sp.weights[g] * (sp.consensus_vars[g] - params.target[g])^2 for
g in 1:G
)
)
end
optimize!(sp.mip, method)
obj = objective_value(sp.mip)
sp_vals = value.(sp.consensus_vars)
return PhSubProblemSolution(obj = obj, vals = sp_vals, residuals = zeros(G))
end
function update_λ_and_residuals!(
solution::PhSubProblemSolution,
params::PHSubProblemParams,
target::Array{Float64,1},
)::Nothing
n_vars = length(solution.vals)
params.target = target
for n in 1:n_vars
solution.residuals[n] = solution.vals[n] - params.target[n]
params.λ[n] += params.ρ * solution.residuals[n]
end
end
function print_header(mpi::MpiInfo)::Nothing
if !mpi.root
return
end
@info "Solving via Progressive Hedging:"
@info @sprintf(
"%8s %20s %20s %14s %8s %8s",
"iter",
"obj",
"infeas",
"consensus",
"time-it",
"time"
)
end
function print_progress(
mpi::MpiInfo,
iteration::PHIterationInfo,
print_interval,
)::Nothing
if !mpi.root
return
end
if iteration.iteration_number % print_interval != 0
return
end
@info @sprintf(
"%8d %20.6e %20.6e %12.2f %% %8.2f %8.2f",
iteration.iteration_number,
iteration.global_obj,
iteration.global_infeas,
iteration.global_residual * 100,
iteration.iteration_time,
iteration.total_elapsed_time
)
end
function has_numerical_issues(target::Array{Float64,1})::Bool
if target == NaN
@warn "Numerical issues detected. Stopping."
return true
end
return false
end
function should_stop(
mpi::MpiInfo,
iterations::Array{PHIterationInfo,1},
termination::PHTermination,
)::Bool
if length(iterations) >= termination.max_iterations
if mpi.root
@info "Iteration limit reached. Stopping."
end
return true
end
if length(iterations) < termination.min_iterations
return false
end
if last(iterations).total_elapsed_time > termination.max_time
if mpi.root
@info "Time limit reached. Stopping."
end
return true
end
curr_it = last(iterations)
prev_it = iterations[length(iterations)-1]
if curr_it.global_infeas < termination.min_feasibility
obj_change = abs(prev_it.global_obj - curr_it.global_obj)
if obj_change < termination.min_improvement
if mpi.root
@info "Feasibility limit reached. Stopping."
end
return true
end
end
return false
end

@ -0,0 +1,18 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
function read(
paths::Vector{String},
::ProgressiveHedging,
)::UnitCommitmentInstance
comm = MPI.COMM_WORLD
mpi = MpiInfo(comm)
(length(paths) % mpi.nprocs == 0) || error(
"Number of processes $(mpi.nprocs) is not a divisor of $(length(paths))",
)
bundled_scenarios = length(paths) ÷ mpi.nprocs
sc_num_start = (mpi.rank - 1) * bundled_scenarios + 1
sc_num_end = mpi.rank * bundled_scenarios
return read(paths[sc_num_start:sc_num_end])
end

@ -0,0 +1,83 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using MPI, DataStructures
const FIRST_STAGE_VARS = ["Is on", "Switch on", "Switch off"]
function solution(model::JuMP.Model, method::ProgressiveHedging)::OrderedDict
comm = MPI.COMM_WORLD
mpi = MpiInfo(comm)
sp_solution = UnitCommitment.solution(model)
gather_solution = OrderedDict()
for (solution_key, dict) in sp_solution
if solution_key !== "Spinning reserve (MW)" &&
solution_key FIRST_STAGE_VARS
push!(gather_solution, solution_key => OrderedDict())
for (gen_bus_key, values) in dict
global T = length(values)
receive_values =
MPI.UBuffer(Vector{Float64}(undef, T * mpi.nprocs), T)
MPI.Gather!(float.(values), receive_values, comm)
if mpi.root
push!(
gather_solution[solution_key],
gen_bus_key => receive_values.data,
)
end
end
end
end
push!(gather_solution, "Spinning reserve (MW)" => OrderedDict())
for (reserve_type, dict) in sp_solution["Spinning reserve (MW)"]
push!(
gather_solution["Spinning reserve (MW)"],
reserve_type => OrderedDict(),
)
for (gen_key, values) in dict
receive_values =
MPI.UBuffer(Vector{Float64}(undef, T * mpi.nprocs), T)
MPI.Gather!(float.(values), receive_values, comm)
if mpi.root
push!(
gather_solution["Spinning reserve (MW)"][reserve_type],
gen_key => receive_values.data,
)
end
end
end
aggregate_solution = OrderedDict()
if mpi.root
for first_stage_var in FIRST_STAGE_VARS
aggregate_solution[first_stage_var] = OrderedDict()
for gen_key in keys(sp_solution[first_stage_var])
aggregate_solution[first_stage_var][gen_key] =
sp_solution[first_stage_var][gen_key]
end
end
for i in 1:mpi.nprocs
push!(aggregate_solution, "s$i" => OrderedDict())
for (solution_key, solution_dict) in gather_solution
push!(aggregate_solution["s$i"], solution_key => OrderedDict())
if solution_key !== "Spinning reserve (MW)"
for (gen_bus_key, values) in solution_dict
aggregate_solution["s$i"][solution_key][gen_bus_key] =
gather_solution[solution_key][gen_bus_key][(i-1)*T+1:i*T]
end
else
for (reserve_name, reserve_dict) in solution_dict
push!(
aggregate_solution["s$i"][solution_key],
reserve_name => OrderedDict(),
)
for (gen_key, values) in reserve_dict
aggregate_solution["s$i"][solution_key][reserve_name][gen_key] =
gather_solution[solution_key][reserve_name][gen_key][(i-1)*T+1:i*T]
end
end
end
end
end
end
return aggregate_solution
end

@ -0,0 +1,73 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using JuMP, MPI, TimerOutputs
Base.@kwdef mutable struct PHTermination
max_iterations::Int = 1000
max_time::Float64 = 14400.0
min_feasibility::Float64 = 1e-3
min_improvement::Float64 = 1e-3
min_iterations::Int = 2
end
Base.@kwdef mutable struct PHIterationInfo
global_infeas::Float64
global_obj::Float64
global_residual::Float64
iteration_number::Int
iteration_time::Float64
sp_vals::Array{Float64,1}
sp_obj::Float64
target::Array{Float64,1}
total_elapsed_time::Float64
end
Base.@kwdef mutable struct ProgressiveHedging <: SolutionMethod
initial_weights::Union{Vector{Float64},Nothing} = nothing
initial_target::Union{Vector{Float64},Nothing} = nothing
ρ::Float64 = 1.0
λ::Float64 = 0.0
print_interval::Int = 1
termination::PHTermination = PHTermination()
inner_method::SolutionMethod = XavQiuWanThi2019.Method()
end
struct SpResult
obj::Float64
vals::Array{Float64,1}
end
Base.@kwdef mutable struct PHSubProblem
mip::JuMP.Model
obj::AffExpr
consensus_vars::Array{VariableRef,1}
weights::Array{Float64,1}
end
Base.@kwdef struct PhSubProblemSolution
obj::Float64
vals::Array{Float64,1}
residuals::Array{Float64,1}
end
Base.@kwdef mutable struct PHSubProblemParams
ρ::Float64
λ::Array{Float64,1}
target::Array{Float64,1}
end
struct MpiInfo
comm::Any
rank::Int
root::Bool
nprocs::Int
function MpiInfo(comm)
rank = MPI.Comm_rank(comm) + 1
is_root = (rank == 1)
nprocs = MPI.Comm_size(comm)
return new(comm, rank, is_root, nprocs)
end
end

@ -1,3 +1,8 @@
name = "UnitCommitmentT"
uuid = "a3b7a17a-ab64-45e4-a924-cd5ae7dc644e"
authors = ["Alinson S. Xavier <git@axavier.org>"]
version = "0.1.0"
[deps]
Cbc = "9961bab8-2fa3-5c5a-9d89-47fab24efd76"
DataStructures = "864edb3b-99cc-5e75-8d2d-829cb0a9cfe8"
@ -6,21 +11,11 @@ GZip = "92fee26a-97fe-5a0c-ad85-20a5f3185b63"
HiGHS = "87dc4568-4c63-4d18-b0c0-bb2238e4078b"
JSON = "682c06a0-de6a-54ab-a142-c8b1cf79cde6"
JuMP = "4076af6c-e467-56ae-b986-b466b2749572"
JuliaFormatter = "98e50ef6-434e-11e9-1051-2b60c6c9e899"
LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
Logging = "56ddb016-857b-54e1-b83d-db4d58db5568"
MPI = "da04e1cc-30fd-572f-bb4f-1f8673147195"
MathOptInterface = "b8f27783-ece8-5eb3-8dc8-9495eed66fee"
PackageCompiler = "9b87118b-4619-50d2-8e1e-99f35a4d4d9d"
Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
Revise = "295af30f-e4ad-537b-8983-00126c2a3abe"
Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
[compat]
DataStructures = "0.18"
Distributions = "0.25"
GZip = "0.5"
JSON = "0.21"
JuMP = "1"
MathOptInterface = "1"
PackageCompiler = "1"
julia = "1"
UnitCommitment = "64606440-39ea-11e9-0f29-3303a1d3d877"

@ -1,21 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment
@testset "read_egret_solution" begin
solution =
UnitCommitment.read_egret_solution("$FIXTURES/egret_output.json.gz")
for attr in
["Is on", "Thermal production (MW)", "Thermal production cost (\$)"]
@test attr in keys(solution)
@test "115_STEAM_1" in keys(solution[attr])
@test length(solution[attr]["115_STEAM_1"]) == 48
end
@test solution["Thermal production cost (\$)"]["315_CT_6"][15:20] ==
[0.0, 0.0, 884.44, 1470.71, 1470.71, 884.44]
@test solution["Startup cost (\$)"]["315_CT_6"][15:20] ==
[0.0, 0.0, 5665.23, 0.0, 0.0, 0.0]
@test length(keys(solution["Is on"])) == 154
end

@ -1,22 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, LinearAlgebra, Cbc, JuMP, JSON, GZip
@testset "read v0.2" begin
instance = UnitCommitment.read("$FIXTURES/ucjl-0.2.json.gz")
@test length(instance.scenarios) == 1
sc = instance.scenarios[1]
@test length(sc.reserves_by_name["r1"].amount) == 4
@test sc.thermal_units_by_name["g2"].reserves[1].name == "r1"
end
@testset "read v0.3" begin
instance = UnitCommitment.read("$FIXTURES/ucjl-0.3.json.gz")
@test length(instance.scenarios) == 1
sc = instance.scenarios[1]
@test length(sc.thermal_units) == 6
@test length(sc.buses) == 14
@test length(sc.lines) == 20
end

@ -1,35 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Cbc, HiGHS, JuMP
import UnitCommitment: AELMP
@testset "aelmp" begin
path = "$FIXTURES/aelmp_simple.json.gz"
# model has to be solved first
instance = UnitCommitment.read(path)
model = UnitCommitment.build_model(
instance = instance,
optimizer = Cbc.Optimizer,
variable_names = true,
)
JuMP.set_silent(model)
UnitCommitment.optimize!(model)
# policy 1: allow offlines; consider startups
aelmp_1 =
UnitCommitment.compute_lmp(model, AELMP(), optimizer = HiGHS.Optimizer)
@test aelmp_1["s1", "B1", 1] 231.7 atol = 0.1
# policy 2: do not allow offlines; but consider startups
aelmp_2 = UnitCommitment.compute_lmp(
model,
AELMP(
allow_offline_participation = false,
consider_startup_costs = true,
),
optimizer = HiGHS.Optimizer,
)
@test aelmp_2["s1", "B1", 1] 274.3 atol = 0.1
end

@ -1,51 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Cbc, HiGHS, JuMP
import UnitCommitment: ConventionalLMP
function solve_conventional_testcase(path::String)
instance = UnitCommitment.read(path)
model = UnitCommitment.build_model(
instance = instance,
optimizer = Cbc.Optimizer,
variable_names = true,
)
JuMP.set_silent(model)
UnitCommitment.optimize!(model)
lmp = UnitCommitment.compute_lmp(
model,
ConventionalLMP(),
optimizer = HiGHS.Optimizer,
)
return lmp
end
@testset "conventional" begin
# instance 1
path = "$FIXTURES/lmp_simple_test_1.json.gz"
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 50.0
# instance 2
path = "$FIXTURES/lmp_simple_test_2.json.gz"
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 60.0
# instance 3
path = "$FIXTURES/lmp_simple_test_3.json.gz"
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 70.0
@test lmp["s1", "C", 1] == 100.0
# instance 4
path = "$FIXTURES/lmp_simple_test_4.json.gz"
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 70.0
@test lmp["s1", "C", 1] == 90.0
end

@ -1,84 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment
using JuMP
using Cbc
using JSON
import UnitCommitment:
ArrCon2000,
CarArr2006,
DamKucRajAta2016,
Formulation,
Gar1962,
KnuOstWat2018,
MorLatRam2013,
PanGua2016,
XavQiuWanThi2019,
WanHob2016
function _test(
formulation::Formulation;
instances = ["case14"],
dump::Bool = false,
)::Nothing
for instance_name in instances
instance = UnitCommitment.read("$(FIXTURES)/$(instance_name).json.gz")
model = UnitCommitment.build_model(
instance = instance,
formulation = formulation,
optimizer = Cbc.Optimizer,
variable_names = true,
)
set_silent(model)
UnitCommitment.optimize!(model)
solution = UnitCommitment.solution(model)
if dump
open("/tmp/ucjl.json", "w") do f
return write(f, JSON.json(solution, 2))
end
write_to_file(model, "/tmp/ucjl.lp")
end
@test UnitCommitment.validate(instance, solution)
end
return
end
@testset "formulations" begin
@testset "default" begin
_test(Formulation())
end
@testset "ArrCon2000" begin
_test(Formulation(ramping = ArrCon2000.Ramping()))
end
@testset "DamKucRajAta2016" begin
_test(Formulation(ramping = DamKucRajAta2016.Ramping()))
end
@testset "MorLatRam2013" begin
_test(
Formulation(
ramping = MorLatRam2013.Ramping(),
startup_costs = MorLatRam2013.StartupCosts(),
),
)
end
@testset "PanGua2016" begin
_test(Formulation(ramping = PanGua2016.Ramping()))
end
@testset "Gar1962" begin
_test(Formulation(pwl_costs = Gar1962.PwlCosts()))
end
@testset "CarArr2006" begin
_test(Formulation(pwl_costs = CarArr2006.PwlCosts()))
end
@testset "KnuOstWat2018" begin
_test(Formulation(pwl_costs = KnuOstWat2018.PwlCosts()))
end
@testset "WanHob2016" begin
_test(
Formulation(ramping = WanHob2016.Ramping()),
instances = ["case14-flex"],
)
end
end

@ -1,83 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Test, LinearAlgebra
import UnitCommitment: _Violation, _offer, _query
@testset "_ViolationFilter" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
sc = instance.scenarios[1]
filter = UnitCommitment._ViolationFilter(max_per_line = 1, max_total = 2)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = nothing,
amount = 100.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[1],
amount = 300.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[5],
amount = 500.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[4],
amount = 400.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[2],
outage_line = sc.lines[1],
amount = 200.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[2],
outage_line = sc.lines[8],
amount = 100.0,
),
)
actual = _query(filter)
expected = [
_Violation(
time = 1,
monitored_line = sc.lines[2],
outage_line = sc.lines[1],
amount = 200.0,
),
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[5],
amount = 500.0,
),
]
@test actual == expected
end

@ -1,37 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Test, LinearAlgebra
import UnitCommitment: _Violation, _offer, _query
@testset "find_violations" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
sc = instance.scenarios[1]
for line in sc.lines, t in 1:instance.time
line.normal_flow_limit[t] = 1.0
line.emergency_flow_limit[t] = 1.0
end
isf = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
lodf = UnitCommitment._line_outage_factors(
lines = sc.lines,
buses = sc.buses,
isf = isf,
)
inj = [1000.0 for b in 1:13, t in 1:instance.time]
overflow = [0.0 for l in sc.lines, t in 1:instance.time]
violations = UnitCommitment._find_violations(
instance = instance,
sc = sc,
net_injections = inj,
overflow = overflow,
isf = isf,
lodf = lodf,
max_per_line = 1,
max_per_period = 5,
)
@test length(violations) == 20
end

@ -1,147 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Test, LinearAlgebra
@testset "_susceptance_matrix" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
sc = instance.scenarios[1]
actual = UnitCommitment._susceptance_matrix(sc.lines)
@test size(actual) == (20, 20)
expected = Diagonal([
29.5,
7.83,
8.82,
9.9,
10.04,
10.2,
41.45,
8.35,
3.14,
6.93,
8.77,
6.82,
13.4,
9.91,
15.87,
20.65,
6.46,
9.09,
8.73,
5.02,
])
@test round.(actual, digits = 2) == expected
end
@testset "_reduced_incidence_matrix" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
sc = instance.scenarios[1]
actual = UnitCommitment._reduced_incidence_matrix(
lines = sc.lines,
buses = sc.buses,
)
@test size(actual) == (20, 13)
@test actual[1, 1] == -1.0
@test actual[3, 1] == 1.0
@test actual[4, 1] == 1.0
@test actual[5, 1] == 1.0
@test actual[3, 2] == -1.0
@test actual[6, 2] == 1.0
@test actual[4, 3] == -1.0
@test actual[6, 3] == -1.0
@test actual[7, 3] == 1.0
@test actual[8, 3] == 1.0
@test actual[9, 3] == 1.0
@test actual[2, 4] == -1.0
@test actual[5, 4] == -1.0
@test actual[7, 4] == -1.0
@test actual[10, 4] == 1.0
@test actual[10, 5] == -1.0
@test actual[11, 5] == 1.0
@test actual[12, 5] == 1.0
@test actual[13, 5] == 1.0
@test actual[8, 6] == -1.0
@test actual[14, 6] == 1.0
@test actual[15, 6] == 1.0
@test actual[14, 7] == -1.0
@test actual[9, 8] == -1.0
@test actual[15, 8] == -1.0
@test actual[16, 8] == 1.0
@test actual[17, 8] == 1.0
@test actual[16, 9] == -1.0
@test actual[18, 9] == 1.0
@test actual[11, 10] == -1.0
@test actual[18, 10] == -1.0
@test actual[12, 11] == -1.0
@test actual[19, 11] == 1.0
@test actual[13, 12] == -1.0
@test actual[19, 12] == -1.0
@test actual[20, 12] == 1.0
@test actual[17, 13] == -1.0
@test actual[20, 13] == -1.0
end
@testset "_injection_shift_factors" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
sc = instance.scenarios[1]
actual = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
@test size(actual) == (20, 13)
@test round.(actual, digits = 2) == [
-0.84 -0.75 -0.67 -0.61 -0.63 -0.66 -0.66 -0.65 -0.65 -0.64 -0.63 -0.63 -0.64
-0.16 -0.25 -0.33 -0.39 -0.37 -0.34 -0.34 -0.35 -0.35 -0.36 -0.37 -0.37 -0.36
0.03 -0.53 -0.15 -0.1 -0.12 -0.14 -0.14 -0.14 -0.13 -0.13 -0.12 -0.12 -0.13
0.06 -0.14 -0.32 -0.22 -0.25 -0.3 -0.3 -0.29 -0.28 -0.27 -0.25 -0.26 -0.27
0.08 -0.07 -0.2 -0.29 -0.26 -0.22 -0.22 -0.22 -0.23 -0.25 -0.26 -0.26 -0.24
0.03 0.47 -0.15 -0.1 -0.12 -0.14 -0.14 -0.14 -0.13 -0.13 -0.12 -0.12 -0.13
0.08 0.31 0.5 -0.3 -0.03 0.36 0.36 0.28 0.23 0.1 -0.0 0.02 0.17
0.0 0.01 0.02 -0.01 -0.22 -0.63 -0.63 -0.45 -0.41 -0.32 -0.24 -0.25 -0.36
0.0 0.01 0.01 -0.01 -0.12 -0.17 -0.17 -0.26 -0.24 -0.18 -0.14 -0.14 -0.21
-0.0 -0.02 -0.03 0.02 -0.66 -0.2 -0.2 -0.29 -0.36 -0.5 -0.63 -0.61 -0.43
-0.0 -0.01 -0.02 0.01 0.21 -0.12 -0.12 -0.17 -0.28 -0.53 0.18 0.15 -0.03
-0.0 -0.0 -0.0 0.0 0.03 -0.02 -0.02 -0.03 -0.02 0.01 -0.52 -0.17 -0.09
-0.0 -0.01 -0.01 0.01 0.11 -0.06 -0.06 -0.09 -0.05 0.02 -0.28 -0.59 -0.31
-0.0 -0.0 -0.0 -0.0 -0.0 -0.0 -1.0 -0.0 -0.0 -0.0 -0.0 -0.0 0.0
0.0 0.01 0.02 -0.01 -0.22 0.37 0.37 -0.45 -0.41 -0.32 -0.24 -0.25 -0.36
0.0 0.01 0.02 -0.01 -0.21 0.12 0.12 0.17 -0.72 -0.47 -0.18 -0.15 0.03
0.0 0.01 0.01 -0.01 -0.14 0.08 0.08 0.12 0.07 -0.03 -0.2 -0.24 -0.6
0.0 0.01 0.02 -0.01 -0.21 0.12 0.12 0.17 0.28 -0.47 -0.18 -0.15 0.03
-0.0 -0.0 -0.0 0.0 0.03 -0.02 -0.02 -0.03 -0.02 0.01 0.48 -0.17 -0.09
-0.0 -0.01 -0.01 0.01 0.14 -0.08 -0.08 -0.12 -0.07 0.03 0.2 0.24 -0.4
]
end
@testset "_line_outage_factors" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
sc = instance.scenarios[1]
isf_before = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
lodf = UnitCommitment._line_outage_factors(
lines = sc.lines,
buses = sc.buses,
isf = isf_before,
)
for contingency in sc.contingencies
for lc in contingency.lines
prev_susceptance = lc.susceptance
lc.susceptance = 0.0
isf_after = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
lc.susceptance = prev_susceptance
for lm in sc.lines
expected = isf_after[lm.offset, :]
actual =
isf_before[lm.offset, :] +
lodf[lm.offset, lc.offset] * isf_before[lc.offset, :]
@test norm(expected - actual) < 1e-6
end
end
end
end

@ -0,0 +1,60 @@
module UnitCommitmentT
using JuliaFormatter
using UnitCommitment
using Test
include("usage.jl")
include("import/egret_test.jl")
include("instance/read_test.jl")
include("instance/migrate_test.jl")
include("model/formulations_test.jl")
include("solution/methods/XavQiuWanThi19/filter_test.jl")
include("solution/methods/XavQiuWanThi19/find_test.jl")
include("solution/methods/XavQiuWanThi19/sensitivity_test.jl")
include("solution/methods/ProgressiveHedging/usage_test.jl")
include("transform/initcond_test.jl")
include("transform/slice_test.jl")
include("transform/randomize/XavQiuAhm2021_test.jl")
include("validation/repair_test.jl")
include("lmp/conventional_test.jl")
include("lmp/aelmp_test.jl")
basedir = dirname(@__FILE__)
function fixture(path::String)::String
return "$basedir/../fixtures/$path"
end
function runtests()
println("Running tests...")
UnitCommitment._setup_logger(level = Base.CoreLogging.Error)
@testset "UnitCommitment" begin
usage_test()
import_egret_test()
instance_read_test()
instance_migrate_test()
model_formulations_test()
solution_methods_XavQiuWanThi19_filter_test()
solution_methods_XavQiuWanThi19_find_test()
solution_methods_XavQiuWanThi19_sensitivity_test()
solution_methods_ProgressiveHedging_usage_test()
transform_initcond_test()
transform_slice_test()
transform_randomize_XavQiuAhm2021_test()
validation_repair_test()
lmp_conventional_test()
lmp_aelmp_test()
end
return
end
function format()
JuliaFormatter.format(basedir, verbose = true)
JuliaFormatter.format("$basedir/../../src", verbose = true)
return
end
export runtests, format
end # module UnitCommitmentT

@ -0,0 +1,23 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment
function import_egret_test()
@testset "read_egret_solution" begin
solution =
UnitCommitment.read_egret_solution(fixture("egret_output.json.gz"))
for attr in
["Is on", "Thermal production (MW)", "Thermal production cost (\$)"]
@test attr in keys(solution)
@test "115_STEAM_1" in keys(solution[attr])
@test length(solution[attr]["115_STEAM_1"]) == 48
end
@test solution["Thermal production cost (\$)"]["315_CT_6"][15:20] ==
[0.0, 0.0, 884.44, 1470.71, 1470.71, 884.44]
@test solution["Startup cost (\$)"]["315_CT_6"][15:20] ==
[0.0, 0.0, 5665.23, 0.0, 0.0, 0.0]
@test length(keys(solution["Is on"])) == 154
end
end

@ -0,0 +1,24 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, LinearAlgebra, Cbc, JuMP, JSON, GZip
function instance_migrate_test()
@testset "read v0.2" begin
instance = UnitCommitment.read(fixture("/ucjl-0.2.json.gz"))
@test length(instance.scenarios) == 1
sc = instance.scenarios[1]
@test length(sc.reserves_by_name["r1"].amount) == 4
@test sc.thermal_units_by_name["g2"].reserves[1].name == "r1"
end
@testset "read v0.3" begin
instance = UnitCommitment.read(fixture("/ucjl-0.3.json.gz"))
@test length(instance.scenarios) == 1
sc = instance.scenarios[1]
@test length(sc.thermal_units) == 6
@test length(sc.buses) == 14
@test length(sc.lines) == 20
end
end

@ -0,0 +1,168 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, LinearAlgebra, Cbc, JuMP, JSON, GZip
function instance_read_test()
@testset "read_benchmark" begin
instance = UnitCommitment.read(fixture("case14.json.gz"))
@test repr(instance) == (
"UnitCommitmentInstance(1 scenarios, 6 thermal units, 0 profiled units, 14 buses, " *
"20 lines, 19 contingencies, 1 price sensitive loads, 4 time steps)"
)
@test length(instance.scenarios) == 1
sc = instance.scenarios[1]
@test length(sc.lines) == 20
@test length(sc.buses) == 14
@test length(sc.thermal_units) == 6
@test length(sc.contingencies) == 19
@test length(sc.price_sensitive_loads) == 1
@test instance.time == 4
@test sc.lines[5].name == "l5"
@test sc.lines[5].source.name == "b2"
@test sc.lines[5].target.name == "b5"
@test sc.lines[5].reactance 0.17388
@test sc.lines[5].susceptance 10.037550333
@test sc.lines[5].normal_flow_limit == [1e8 for t in 1:4]
@test sc.lines[5].emergency_flow_limit == [1e8 for t in 1:4]
@test sc.lines[5].flow_limit_penalty == [5e3 for t in 1:4]
@test sc.lines_by_name["l5"].name == "l5"
@test sc.lines[1].name == "l1"
@test sc.lines[1].source.name == "b1"
@test sc.lines[1].target.name == "b2"
@test sc.lines[1].reactance 0.059170
@test sc.lines[1].susceptance 29.496860773945
@test sc.lines[1].normal_flow_limit == [300.0 for t in 1:4]
@test sc.lines[1].emergency_flow_limit == [400.0 for t in 1:4]
@test sc.lines[1].flow_limit_penalty == [1e3 for t in 1:4]
@test sc.buses[9].name == "b9"
@test sc.buses[9].load == [35.36638, 33.25495, 31.67138, 31.14353]
@test sc.buses_by_name["b9"].name == "b9"
@test sc.reserves[1].name == "r1"
@test sc.reserves[1].type == "spinning"
@test sc.reserves[1].amount == [100.0, 100.0, 100.0, 100.0]
@test sc.reserves_by_name["r1"].name == "r1"
unit = sc.thermal_units[1]
@test unit.name == "g1"
@test unit.bus.name == "b1"
@test unit.ramp_up_limit == 1e6
@test unit.ramp_down_limit == 1e6
@test unit.startup_limit == 1e6
@test unit.shutdown_limit == 1e6
@test unit.must_run == [false for t in 1:4]
@test unit.min_power_cost == [1400.0 for t in 1:4]
@test unit.min_uptime == 1
@test unit.min_downtime == 1
for t in 1:1
@test unit.cost_segments[1].mw[t] == 10.0
@test unit.cost_segments[2].mw[t] == 20.0
@test unit.cost_segments[3].mw[t] == 5.0
@test unit.cost_segments[1].cost[t] 20.0
@test unit.cost_segments[2].cost[t] 30.0
@test unit.cost_segments[3].cost[t] 40.0
end
@test length(unit.startup_categories) == 3
@test unit.startup_categories[1].delay == 1
@test unit.startup_categories[2].delay == 2
@test unit.startup_categories[3].delay == 3
@test unit.startup_categories[1].cost == 1000.0
@test unit.startup_categories[2].cost == 1500.0
@test unit.startup_categories[3].cost == 2000.0
@test length(unit.reserves) == 0
@test sc.thermal_units_by_name["g1"].name == "g1"
unit = sc.thermal_units[2]
@test unit.name == "g2"
@test unit.must_run == [false for t in 1:4]
@test length(unit.reserves) == 1
unit = sc.thermal_units[3]
@test unit.name == "g3"
@test unit.bus.name == "b3"
@test unit.ramp_up_limit == 70.0
@test unit.ramp_down_limit == 70.0
@test unit.startup_limit == 70.0
@test unit.shutdown_limit == 70.0
@test unit.must_run == [true for t in 1:4]
@test unit.min_power_cost == [0.0 for t in 1:4]
@test unit.min_uptime == 1
@test unit.min_downtime == 1
for t in 1:4
@test unit.cost_segments[1].mw[t] 33
@test unit.cost_segments[2].mw[t] 33
@test unit.cost_segments[3].mw[t] 34
@test unit.cost_segments[1].cost[t] 33.75
@test unit.cost_segments[2].cost[t] 38.04
@test unit.cost_segments[3].cost[t] 44.77853
end
@test length(unit.reserves) == 1
@test unit.reserves[1].name == "r1"
@test sc.contingencies[1].lines == [sc.lines[1]]
@test sc.contingencies[1].thermal_units == []
@test sc.contingencies[1].name == "c1"
@test sc.contingencies_by_name["c1"].name == "c1"
load = sc.price_sensitive_loads[1]
@test load.name == "ps1"
@test load.bus.name == "b3"
@test load.revenue == [100.0 for t in 1:4]
@test load.demand == [50.0 for t in 1:4]
@test sc.price_sensitive_loads_by_name["ps1"].name == "ps1"
end
@testset "read_benchmark sub-hourly" begin
instance = UnitCommitment.read(fixture("case14-sub-hourly.json.gz"))
@test instance.time == 4
unit = instance.scenarios[1].thermal_units[1]
@test unit.name == "g1"
@test unit.min_uptime == 2
@test unit.min_downtime == 2
@test length(unit.startup_categories) == 3
@test unit.startup_categories[1].delay == 2
@test unit.startup_categories[2].delay == 4
@test unit.startup_categories[3].delay == 6
@test unit.initial_status == -200
end
@testset "read_benchmark profiled-units" begin
instance = UnitCommitment.read(fixture("case14-profiled.json.gz"))
sc = instance.scenarios[1]
@test length(sc.profiled_units) == 2
first_pu = sc.profiled_units[1]
@test first_pu.name == "g7"
@test first_pu.bus.name == "b4"
@test first_pu.cost == [100.0 for t in 1:4]
@test first_pu.min_power == [60.0 for t in 1:4]
@test first_pu.max_power == [100.0 for t in 1:4]
@test sc.profiled_units_by_name["g7"].name == "g7"
second_pu = sc.profiled_units[2]
@test second_pu.name == "g8"
@test second_pu.bus.name == "b5"
@test second_pu.cost == [50.0 for t in 1:4]
@test second_pu.min_power == [0.0 for t in 1:4]
@test second_pu.max_power == [120.0 for t in 1:4]
@test sc.profiled_units_by_name["g8"].name == "g8"
end
@testset "read_benchmark commitmemt-status" begin
instance = UnitCommitment.read(fixture("case14-fixed-status.json.gz"))
sc = instance.scenarios[1]
@test sc.thermal_units[1].commitment_status == [nothing for t in 1:4]
@test sc.thermal_units[2].commitment_status == [true for t in 1:4]
@test sc.thermal_units[4].commitment_status == [false for t in 1:4]
@test sc.thermal_units[6].commitment_status ==
[false, nothing, true, nothing]
end
end

@ -0,0 +1,40 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Cbc, HiGHS, JuMP
import UnitCommitment: AELMP
function lmp_aelmp_test()
@testset "aelmp" begin
path = fixture("aelmp_simple.json.gz")
# model has to be solved first
instance = UnitCommitment.read(path)
model = UnitCommitment.build_model(
instance = instance,
optimizer = Cbc.Optimizer,
variable_names = true,
)
JuMP.set_silent(model)
UnitCommitment.optimize!(model)
# policy 1: allow offlines; consider startups
aelmp_1 = UnitCommitment.compute_lmp(
model,
AELMP(),
optimizer = HiGHS.Optimizer,
)
@test aelmp_1["s1", "B1", 1] 231.7 atol = 0.1
# policy 2: do not allow offlines; but consider startups
aelmp_2 = UnitCommitment.compute_lmp(
model,
AELMP(
allow_offline_participation = false,
consider_startup_costs = true,
),
optimizer = HiGHS.Optimizer,
)
@test aelmp_2["s1", "B1", 1] 274.3 atol = 0.1
end
end

@ -0,0 +1,53 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Cbc, HiGHS, JuMP
import UnitCommitment: ConventionalLMP
function solve_conventional_testcase(path::String)
instance = UnitCommitment.read(path)
model = UnitCommitment.build_model(
instance = instance,
optimizer = Cbc.Optimizer,
variable_names = true,
)
JuMP.set_silent(model)
UnitCommitment.optimize!(model)
lmp = UnitCommitment.compute_lmp(
model,
ConventionalLMP(),
optimizer = HiGHS.Optimizer,
)
return lmp
end
function lmp_conventional_test()
@testset "conventional" begin
# instance 1
path = fixture("lmp_simple_test_1.json.gz")
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 50.0
# instance 2
path = fixture("lmp_simple_test_2.json.gz")
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 60.0
# instance 3
path = fixture("lmp_simple_test_3.json.gz")
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 70.0
@test lmp["s1", "C", 1] == 100.0
# instance 4
path = fixture("lmp_simple_test_4.json.gz")
lmp = solve_conventional_testcase(path)
@test lmp["s1", "A", 1] == 50.0
@test lmp["s1", "B", 1] == 70.0
@test lmp["s1", "C", 1] == 90.0
end
end

@ -0,0 +1,86 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment
using JuMP
using Cbc
using JSON
import UnitCommitment:
ArrCon2000,
CarArr2006,
DamKucRajAta2016,
Formulation,
Gar1962,
KnuOstWat2018,
MorLatRam2013,
PanGua2016,
XavQiuWanThi2019,
WanHob2016
function _test(
formulation::Formulation;
instances = ["case14"],
dump::Bool = false,
)::Nothing
for instance_name in instances
instance = UnitCommitment.read(fixture("$(instance_name).json.gz"))
model = UnitCommitment.build_model(
instance = instance,
formulation = formulation,
optimizer = Cbc.Optimizer,
variable_names = true,
)
set_silent(model)
UnitCommitment.optimize!(model)
solution = UnitCommitment.solution(model)
if dump
open("/tmp/ucjl.json", "w") do f
return write(f, JSON.json(solution, 2))
end
write_to_file(model, "/tmp/ucjl.lp")
end
@test UnitCommitment.validate(instance, solution)
end
return
end
function model_formulations_test()
@testset "formulations" begin
@testset "default" begin
_test(Formulation())
end
@testset "ArrCon2000" begin
_test(Formulation(ramping = ArrCon2000.Ramping()))
end
@testset "DamKucRajAta2016" begin
_test(Formulation(ramping = DamKucRajAta2016.Ramping()))
end
@testset "MorLatRam2013" begin
_test(
Formulation(
ramping = MorLatRam2013.Ramping(),
startup_costs = MorLatRam2013.StartupCosts(),
),
)
end
@testset "PanGua2016" begin
_test(Formulation(ramping = PanGua2016.Ramping()))
end
@testset "Gar1962" begin
_test(Formulation(pwl_costs = Gar1962.PwlCosts()))
end
@testset "CarArr2006" begin
_test(Formulation(pwl_costs = CarArr2006.PwlCosts()))
end
@testset "KnuOstWat2018" begin
_test(Formulation(pwl_costs = KnuOstWat2018.PwlCosts()))
end
@testset "WanHob2016" begin
_test(
Formulation(ramping = WanHob2016.Ramping()),
instances = ["case14-flex"],
)
end
end
end

@ -0,0 +1,40 @@
using HiGHS
using MPI
using JuMP
using UnitCommitment
UnitCommitment._setup_logger(level = Base.CoreLogging.Error)
function fixture(path::String)::String
basedir = dirname(@__FILE__)
return "$basedir/../../../../fixtures/$path"
end
# Initialize MPI
MPI.Init()
# Configure progressive hedging method
ph = UnitCommitment.ProgressiveHedging()
# Read problem instance
instance = UnitCommitment.read(
[fixture("case14.json.gz"), fixture("case14.json.gz")],
ph,
)
# Build JuMP model
model = UnitCommitment.build_model(
instance = instance,
optimizer = optimizer_with_attributes(
HiGHS.Optimizer,
MOI.Silent() => true,
),
)
# Run the decentralized optimization algorithm
UnitCommitment.optimize!(model, ph)
# Fetch the solution
solution = UnitCommitment.solution(model, ph)
# Close MPI
MPI.Finalize()

@ -0,0 +1,16 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using MPI
function solution_methods_ProgressiveHedging_usage_test()
basedir = dirname(@__FILE__)
@testset "ProgressiveHedging" begin
mpiexec() do exe
return run(
`$exe -n 2 $(Base.julia_cmd()) --project=test $basedir/ph.jl`,
)
end
end
end

@ -0,0 +1,86 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Test, LinearAlgebra
import UnitCommitment: _Violation, _offer, _query
function solution_methods_XavQiuWanThi19_filter_test()
@testset "_ViolationFilter" begin
instance = UnitCommitment.read(fixture("case14.json.gz"))
sc = instance.scenarios[1]
filter =
UnitCommitment._ViolationFilter(max_per_line = 1, max_total = 2)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = nothing,
amount = 100.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[1],
amount = 300.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[5],
amount = 500.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[4],
amount = 400.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[2],
outage_line = sc.lines[1],
amount = 200.0,
),
)
_offer(
filter,
_Violation(
time = 1,
monitored_line = sc.lines[2],
outage_line = sc.lines[8],
amount = 100.0,
),
)
actual = _query(filter)
expected = [
_Violation(
time = 1,
monitored_line = sc.lines[2],
outage_line = sc.lines[1],
amount = 200.0,
),
_Violation(
time = 1,
monitored_line = sc.lines[1],
outage_line = sc.lines[5],
amount = 500.0,
),
]
@test actual == expected
end
end

@ -0,0 +1,39 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Test, LinearAlgebra
import UnitCommitment: _Violation, _offer, _query
function solution_methods_XavQiuWanThi19_find_test()
@testset "find_violations" begin
instance = UnitCommitment.read(fixture("case14.json.gz"))
sc = instance.scenarios[1]
for line in sc.lines, t in 1:instance.time
line.normal_flow_limit[t] = 1.0
line.emergency_flow_limit[t] = 1.0
end
isf = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
lodf = UnitCommitment._line_outage_factors(
lines = sc.lines,
buses = sc.buses,
isf = isf,
)
inj = [1000.0 for b in 1:13, t in 1:instance.time]
overflow = [0.0 for l in sc.lines, t in 1:instance.time]
violations = UnitCommitment._find_violations(
instance = instance,
sc = sc,
net_injections = inj,
overflow = overflow,
isf = isf,
lodf = lodf,
max_per_line = 1,
max_per_period = 5,
)
@test length(violations) == 20
end
end

@ -0,0 +1,149 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Test, LinearAlgebra
function solution_methods_XavQiuWanThi19_sensitivity_test()
@testset "_susceptance_matrix" begin
instance = UnitCommitment.read(fixture("/case14.json.gz"))
sc = instance.scenarios[1]
actual = UnitCommitment._susceptance_matrix(sc.lines)
@test size(actual) == (20, 20)
expected = Diagonal([
29.5,
7.83,
8.82,
9.9,
10.04,
10.2,
41.45,
8.35,
3.14,
6.93,
8.77,
6.82,
13.4,
9.91,
15.87,
20.65,
6.46,
9.09,
8.73,
5.02,
])
@test round.(actual, digits = 2) == expected
end
@testset "_reduced_incidence_matrix" begin
instance = UnitCommitment.read(fixture("/case14.json.gz"))
sc = instance.scenarios[1]
actual = UnitCommitment._reduced_incidence_matrix(
lines = sc.lines,
buses = sc.buses,
)
@test size(actual) == (20, 13)
@test actual[1, 1] == -1.0
@test actual[3, 1] == 1.0
@test actual[4, 1] == 1.0
@test actual[5, 1] == 1.0
@test actual[3, 2] == -1.0
@test actual[6, 2] == 1.0
@test actual[4, 3] == -1.0
@test actual[6, 3] == -1.0
@test actual[7, 3] == 1.0
@test actual[8, 3] == 1.0
@test actual[9, 3] == 1.0
@test actual[2, 4] == -1.0
@test actual[5, 4] == -1.0
@test actual[7, 4] == -1.0
@test actual[10, 4] == 1.0
@test actual[10, 5] == -1.0
@test actual[11, 5] == 1.0
@test actual[12, 5] == 1.0
@test actual[13, 5] == 1.0
@test actual[8, 6] == -1.0
@test actual[14, 6] == 1.0
@test actual[15, 6] == 1.0
@test actual[14, 7] == -1.0
@test actual[9, 8] == -1.0
@test actual[15, 8] == -1.0
@test actual[16, 8] == 1.0
@test actual[17, 8] == 1.0
@test actual[16, 9] == -1.0
@test actual[18, 9] == 1.0
@test actual[11, 10] == -1.0
@test actual[18, 10] == -1.0
@test actual[12, 11] == -1.0
@test actual[19, 11] == 1.0
@test actual[13, 12] == -1.0
@test actual[19, 12] == -1.0
@test actual[20, 12] == 1.0
@test actual[17, 13] == -1.0
@test actual[20, 13] == -1.0
end
@testset "_injection_shift_factors" begin
instance = UnitCommitment.read(fixture("/case14.json.gz"))
sc = instance.scenarios[1]
actual = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
@test size(actual) == (20, 13)
@test round.(actual, digits = 2) == [
-0.84 -0.75 -0.67 -0.61 -0.63 -0.66 -0.66 -0.65 -0.65 -0.64 -0.63 -0.63 -0.64
-0.16 -0.25 -0.33 -0.39 -0.37 -0.34 -0.34 -0.35 -0.35 -0.36 -0.37 -0.37 -0.36
0.03 -0.53 -0.15 -0.1 -0.12 -0.14 -0.14 -0.14 -0.13 -0.13 -0.12 -0.12 -0.13
0.06 -0.14 -0.32 -0.22 -0.25 -0.3 -0.3 -0.29 -0.28 -0.27 -0.25 -0.26 -0.27
0.08 -0.07 -0.2 -0.29 -0.26 -0.22 -0.22 -0.22 -0.23 -0.25 -0.26 -0.26 -0.24
0.03 0.47 -0.15 -0.1 -0.12 -0.14 -0.14 -0.14 -0.13 -0.13 -0.12 -0.12 -0.13
0.08 0.31 0.5 -0.3 -0.03 0.36 0.36 0.28 0.23 0.1 -0.0 0.02 0.17
0.0 0.01 0.02 -0.01 -0.22 -0.63 -0.63 -0.45 -0.41 -0.32 -0.24 -0.25 -0.36
0.0 0.01 0.01 -0.01 -0.12 -0.17 -0.17 -0.26 -0.24 -0.18 -0.14 -0.14 -0.21
-0.0 -0.02 -0.03 0.02 -0.66 -0.2 -0.2 -0.29 -0.36 -0.5 -0.63 -0.61 -0.43
-0.0 -0.01 -0.02 0.01 0.21 -0.12 -0.12 -0.17 -0.28 -0.53 0.18 0.15 -0.03
-0.0 -0.0 -0.0 0.0 0.03 -0.02 -0.02 -0.03 -0.02 0.01 -0.52 -0.17 -0.09
-0.0 -0.01 -0.01 0.01 0.11 -0.06 -0.06 -0.09 -0.05 0.02 -0.28 -0.59 -0.31
-0.0 -0.0 -0.0 -0.0 -0.0 -0.0 -1.0 -0.0 -0.0 -0.0 -0.0 -0.0 0.0
0.0 0.01 0.02 -0.01 -0.22 0.37 0.37 -0.45 -0.41 -0.32 -0.24 -0.25 -0.36
0.0 0.01 0.02 -0.01 -0.21 0.12 0.12 0.17 -0.72 -0.47 -0.18 -0.15 0.03
0.0 0.01 0.01 -0.01 -0.14 0.08 0.08 0.12 0.07 -0.03 -0.2 -0.24 -0.6
0.0 0.01 0.02 -0.01 -0.21 0.12 0.12 0.17 0.28 -0.47 -0.18 -0.15 0.03
-0.0 -0.0 -0.0 0.0 0.03 -0.02 -0.02 -0.03 -0.02 0.01 0.48 -0.17 -0.09
-0.0 -0.01 -0.01 0.01 0.14 -0.08 -0.08 -0.12 -0.07 0.03 0.2 0.24 -0.4
]
end
@testset "_line_outage_factors" begin
instance = UnitCommitment.read(fixture("/case14.json.gz"))
sc = instance.scenarios[1]
isf_before = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
lodf = UnitCommitment._line_outage_factors(
lines = sc.lines,
buses = sc.buses,
isf = isf_before,
)
for contingency in sc.contingencies
for lc in contingency.lines
prev_susceptance = lc.susceptance
lc.susceptance = 0.0
isf_after = UnitCommitment._injection_shift_factors(
lines = sc.lines,
buses = sc.buses,
)
lc.susceptance = prev_susceptance
for lm in sc.lines
expected = isf_after[lm.offset, :]
actual =
isf_before[lm.offset, :] +
lodf[lm.offset, lc.offset] * isf_before[lc.offset, :]
@test norm(expected - actual) < 1e-6
end
end
end
end
end

@ -0,0 +1,30 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Cbc, JuMP
function transform_initcond_test()
@testset "generate_initial_conditions!" begin
# Load instance
instance = UnitCommitment.read(fixture("case118-initcond.json.gz"))
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
sc = instance.scenarios[1]
# All units should have unknown initial conditions
for g in sc.thermal_units
@test g.initial_power === nothing
@test g.initial_status === nothing
end
# Generate initial conditions
UnitCommitment.generate_initial_conditions!(sc, optimizer)
# All units should now have known initial conditions
for g in sc.thermal_units
@test g.initial_power !== nothing
@test g.initial_status !== nothing
end
# TODO: Check that initial conditions are feasible
end
end

@ -0,0 +1,106 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
import Random
import UnitCommitment: XavQiuAhm2021
using Distributions
using Random
using UnitCommitment, Cbc, JuMP
function get_scenario()
return UnitCommitment.read_benchmark(
"matpower/case118/2017-02-01",
).scenarios[1]
end
system_load(sc) = sum(b.load for b in sc.buses)
test_approx(x, y) = @test isapprox(x, y, atol = 1e-3)
function transform_randomize_XavQiuAhm2021_test()
@testset "XavQiuAhm2021" begin
@testset "cost and load share" begin
sc = get_scenario()
# Check original costs
unit = sc.thermal_units[10]
test_approx(unit.min_power_cost[1], 825.023)
test_approx(unit.cost_segments[1].cost[1], 36.659)
test_approx(unit.startup_categories[1].cost[1], 7570.42)
# Check original load share
bus = sc.buses[1]
prev_system_load = system_load(sc)
test_approx(bus.load[1] / prev_system_load[1], 0.012)
randomize!(
sc,
XavQiuAhm2021.Randomization(randomize_load_profile = false),
rng = MersenneTwister(42),
)
# Check randomized costs
test_approx(unit.min_power_cost[1], 831.977)
test_approx(unit.cost_segments[1].cost[1], 36.968)
test_approx(unit.startup_categories[1].cost[1], 7634.226)
# Check randomized load share
curr_system_load = system_load(sc)
test_approx(bus.load[1] / curr_system_load[1], 0.013)
# System load should not change
@test prev_system_load curr_system_load
end
@testset "load profile" begin
sc = get_scenario()
# Check original load profile
@test round.(system_load(sc), digits = 1)[1:8] [
3059.5,
2983.2,
2937.5,
2953.9,
3073.1,
3356.4,
4068.5,
4018.8,
]
randomize!(
sc,
XavQiuAhm2021.Randomization();
rng = MersenneTwister(42),
)
# Check randomized load profile
@test round.(system_load(sc), digits = 1)[1:8] [
4854.7,
4849.2,
4732.7,
4848.2,
4948.4,
5231.1,
5874.8,
5934.8,
]
end
@testset "profiled unit cost" begin
sc = UnitCommitment.read(
fixture("case14-profiled.json.gz"),
).scenarios[1]
# Check original costs
pu1 = sc.profiled_units[1]
pu2 = sc.profiled_units[2]
test_approx(pu1.cost[1], 100.0)
test_approx(pu2.cost[1], 50.0)
randomize!(
sc,
XavQiuAhm2021.Randomization(randomize_load_profile = false),
rng = MersenneTwister(42),
)
# Check randomized costs
test_approx(pu1.cost[1], 98.039)
test_approx(pu2.cost[1], 48.385)
end
end
end

@ -0,0 +1,68 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, LinearAlgebra, Cbc, JuMP, JSON, GZip
function transform_slice_test()
@testset "slice" begin
instance = UnitCommitment.read(fixture("case14.json.gz"))
modified = UnitCommitment.slice(instance, 1:2)
sc = modified.scenarios[1]
# Should update all time-dependent fields
@test modified.time == 2
@test length(sc.power_balance_penalty) == 2
@test length(sc.reserves_by_name["r1"].amount) == 2
for u in sc.thermal_units
@test length(u.max_power) == 2
@test length(u.min_power) == 2
@test length(u.must_run) == 2
@test length(u.min_power_cost) == 2
for s in u.cost_segments
@test length(s.mw) == 2
@test length(s.cost) == 2
end
end
for b in sc.buses
@test length(b.load) == 2
end
for l in sc.lines
@test length(l.normal_flow_limit) == 2
@test length(l.emergency_flow_limit) == 2
@test length(l.flow_limit_penalty) == 2
end
for ps in sc.price_sensitive_loads
@test length(ps.demand) == 2
@test length(ps.revenue) == 2
end
# Should be able to build model without errors
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = modified,
optimizer = optimizer,
variable_names = true,
)
end
@testset "slice profiled units" begin
instance = UnitCommitment.read(fixture("case14-profiled.json.gz"))
modified = UnitCommitment.slice(instance, 1:2)
sc = modified.scenarios[1]
# Should update all time-dependent fields
for pu in sc.profiled_units
@test length(pu.max_power) == 2
@test length(pu.min_power) == 2
end
# Should be able to build model without errors
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = modified,
optimizer = optimizer,
variable_names = true,
)
end
end

@ -0,0 +1,66 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, LinearAlgebra, Cbc, JuMP, JSON
function _set_flow_limits!(instance)
for sc in instance.scenarios
sc.power_balance_penalty = [100_000 for _ in 1:instance.time]
for line in sc.lines, t in 1:4
line.normal_flow_limit[t] = 10.0
end
end
end
function usage_test()
@testset "usage" begin
@testset "deterministic" begin
instance = UnitCommitment.read(fixture("case14.json.gz"))
_set_flow_limits!(instance)
optimizer =
optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = instance,
optimizer = optimizer,
variable_names = true,
)
@test name(model[:is_on]["g1", 1]) == "is_on[g1,1]"
# Optimize and retrieve solution
UnitCommitment.optimize!(model)
solution = UnitCommitment.solution(model)
# Write solution to a file
filename = tempname()
UnitCommitment.write(filename, solution)
loaded = JSON.parsefile(filename)
@test length(loaded["Is on"]) == 6
# Verify solution
@test UnitCommitment.validate(instance, solution)
# Reoptimize with fixed solution
UnitCommitment.fix!(model, solution)
UnitCommitment.optimize!(model)
@test UnitCommitment.validate(instance, solution)
end
@testset "stochastic" begin
instance = UnitCommitment.read([
fixture("case14.json.gz"),
fixture("case14.json.gz"),
])
_set_flow_limits!(instance)
@test length(instance.scenarios) == 2
optimizer =
optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = instance,
optimizer = optimizer,
)
UnitCommitment.optimize!(model)
solution = UnitCommitment.solution(model)
end
end
end

@ -0,0 +1,43 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, JSON, GZip, DataStructures
function parse_case14()
return JSON.parse(
GZip.gzopen(fixture("case14.json.gz")),
dicttype = () -> DefaultOrderedDict(nothing),
)
end
function validation_repair_test()
@testset "repair!" begin
@testset "Cost curve should be convex" begin
json = parse_case14()
json["Generators"]["g1"]["Production cost curve (MW)"] =
[100, 150, 200]
json["Generators"]["g1"]["Production cost curve (\$)"] =
[10, 25, 30]
sc = UnitCommitment._from_json(json, repair = false)
@test UnitCommitment.repair!(sc) == 4
end
@testset "Startup limit must be greater than Pmin" begin
json = parse_case14()
json["Generators"]["g1"]["Production cost curve (MW)"] = [100, 150]
json["Generators"]["g1"]["Production cost curve (\$)"] = [100, 150]
json["Generators"]["g1"]["Startup limit (MW)"] = 80
sc = UnitCommitment._from_json(json, repair = false)
@test UnitCommitment.repair!(sc) == 1
end
@testset "Startup costs and delays must be increasing" begin
json = parse_case14()
json["Generators"]["g1"]["Startup costs (\$)"] = [300, 200, 100]
json["Generators"]["g1"]["Startup delays (h)"] = [8, 4, 2]
sc = UnitCommitment._from_json(json, repair = false)
@test UnitCommitment.repair!(sc) == 4
end
end
end

@ -1,28 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, Cbc, JuMP
@testset "generate_initial_conditions!" begin
# Load instance
instance = UnitCommitment.read("$FIXTURES/case118-initcond.json.gz")
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
sc = instance.scenarios[1]
# All units should have unknown initial conditions
for g in sc.thermal_units
@test g.initial_power === nothing
@test g.initial_status === nothing
end
# Generate initial conditions
UnitCommitment.generate_initial_conditions!(sc, optimizer)
# All units should now have known initial conditions
for g in sc.thermal_units
@test g.initial_power !== nothing
@test g.initial_status !== nothing
end
# TODO: Check that initial conditions are feasible
end

@ -1,83 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
import Random
import UnitCommitment: XavQiuAhm2021
using Distributions
using Random
using UnitCommitment, Cbc, JuMP
function get_scenario()
return UnitCommitment.read_benchmark(
"matpower/case118/2017-02-01",
).scenarios[1]
end
system_load(sc) = sum(b.load for b in sc.buses)
test_approx(x, y) = @test isapprox(x, y, atol = 1e-3)
@testset "XavQiuAhm2021" begin
@testset "cost and load share" begin
sc = get_scenario()
# Check original costs
unit = sc.thermal_units[10]
test_approx(unit.min_power_cost[1], 825.023)
test_approx(unit.cost_segments[1].cost[1], 36.659)
test_approx(unit.startup_categories[1].cost[1], 7570.42)
# Check original load share
bus = sc.buses[1]
prev_system_load = system_load(sc)
test_approx(bus.load[1] / prev_system_load[1], 0.012)
randomize!(
sc,
XavQiuAhm2021.Randomization(randomize_load_profile = false),
rng = MersenneTwister(42),
)
# Check randomized costs
test_approx(unit.min_power_cost[1], 831.977)
test_approx(unit.cost_segments[1].cost[1], 36.968)
test_approx(unit.startup_categories[1].cost[1], 7634.226)
# Check randomized load share
curr_system_load = system_load(sc)
test_approx(bus.load[1] / curr_system_load[1], 0.013)
# System load should not change
@test prev_system_load curr_system_load
end
@testset "load profile" begin
sc = get_scenario()
# Check original load profile
@test round.(system_load(sc), digits = 1)[1:8]
[3059.5, 2983.2, 2937.5, 2953.9, 3073.1, 3356.4, 4068.5, 4018.8]
randomize!(sc, XavQiuAhm2021.Randomization(); rng = MersenneTwister(42))
# Check randomized load profile
@test round.(system_load(sc), digits = 1)[1:8]
[4854.7, 4849.2, 4732.7, 4848.2, 4948.4, 5231.1, 5874.8, 5934.8]
end
@testset "profiled unit cost" begin
sc =
UnitCommitment.read("$FIXTURES/case14-profiled.json.gz").scenarios[1]
# Check original costs
pu1 = sc.profiled_units[1]
pu2 = sc.profiled_units[2]
test_approx(pu1.cost[1], 100.0)
test_approx(pu2.cost[1], 50.0)
randomize!(
sc,
XavQiuAhm2021.Randomization(randomize_load_profile = false),
rng = MersenneTwister(42),
)
# Check randomized costs
test_approx(pu1.cost[1], 98.039)
test_approx(pu2.cost[1], 48.385)
end
end

@ -1,66 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, LinearAlgebra, Cbc, JuMP, JSON, GZip
@testset "slice" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
modified = UnitCommitment.slice(instance, 1:2)
sc = modified.scenarios[1]
# Should update all time-dependent fields
@test modified.time == 2
@test length(sc.power_balance_penalty) == 2
@test length(sc.reserves_by_name["r1"].amount) == 2
for u in sc.thermal_units
@test length(u.max_power) == 2
@test length(u.min_power) == 2
@test length(u.must_run) == 2
@test length(u.min_power_cost) == 2
for s in u.cost_segments
@test length(s.mw) == 2
@test length(s.cost) == 2
end
end
for b in sc.buses
@test length(b.load) == 2
end
for l in sc.lines
@test length(l.normal_flow_limit) == 2
@test length(l.emergency_flow_limit) == 2
@test length(l.flow_limit_penalty) == 2
end
for ps in sc.price_sensitive_loads
@test length(ps.demand) == 2
@test length(ps.revenue) == 2
end
# Should be able to build model without errors
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = modified,
optimizer = optimizer,
variable_names = true,
)
end
@testset "slice profiled units" begin
instance = UnitCommitment.read("$FIXTURES/case14-profiled.json.gz")
modified = UnitCommitment.slice(instance, 1:2)
sc = modified.scenarios[1]
# Should update all time-dependent fields
for pu in sc.profiled_units
@test length(pu.max_power) == 2
@test length(pu.min_power) == 2
end
# Should be able to build model without errors
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = modified,
optimizer = optimizer,
variable_names = true,
)
end

@ -1,62 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, LinearAlgebra, Cbc, JuMP, JSON
function _set_flow_limits!(instance)
for sc in instance.scenarios
sc.power_balance_penalty = [100_000 for _ in 1:instance.time]
for line in sc.lines, t in 1:4
line.normal_flow_limit[t] = 10.0
end
end
end
@testset "usage" begin
@testset "deterministic" begin
instance = UnitCommitment.read("$FIXTURES/case14.json.gz")
_set_flow_limits!(instance)
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = instance,
optimizer = optimizer,
variable_names = true,
)
@test name(model[:is_on]["g1", 1]) == "is_on[g1,1]"
# Optimize and retrieve solution
UnitCommitment.optimize!(model)
solution = UnitCommitment.solution(model)
# Write solution to a file
filename = tempname()
UnitCommitment.write(filename, solution)
loaded = JSON.parsefile(filename)
@test length(loaded["Is on"]) == 6
# Verify solution
@test UnitCommitment.validate(instance, solution)
# Reoptimize with fixed solution
UnitCommitment.fix!(model, solution)
UnitCommitment.optimize!(model)
@test UnitCommitment.validate(instance, solution)
end
@testset "stochastic" begin
instance = UnitCommitment.read([
"$FIXTURES/case14.json.gz",
"$FIXTURES/case14.json.gz",
])
_set_flow_limits!(instance)
@test length(instance.scenarios) == 2
optimizer = optimizer_with_attributes(Cbc.Optimizer, "logLevel" => 0)
model = UnitCommitment.build_model(
instance = instance,
optimizer = optimizer,
)
UnitCommitment.optimize!(model)
solution = UnitCommitment.solution(model)
end
end

@ -1,39 +0,0 @@
# UnitCommitment.jl: Optimization Package for Security-Constrained Unit Commitment
# Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
# Released under the modified BSD license. See COPYING.md for more details.
using UnitCommitment, JSON, GZip, DataStructures
function parse_case14()
return JSON.parse(
GZip.gzopen("$FIXTURES/case14.json.gz"),
dicttype = () -> DefaultOrderedDict(nothing),
)
end
@testset "repair!" begin
@testset "Cost curve should be convex" begin
json = parse_case14()
json["Generators"]["g1"]["Production cost curve (MW)"] = [100, 150, 200]
json["Generators"]["g1"]["Production cost curve (\$)"] = [10, 25, 30]
sc = UnitCommitment._from_json(json, repair = false)
@test UnitCommitment.repair!(sc) == 4
end
@testset "Startup limit must be greater than Pmin" begin
json = parse_case14()
json["Generators"]["g1"]["Production cost curve (MW)"] = [100, 150]
json["Generators"]["g1"]["Production cost curve (\$)"] = [100, 150]
json["Generators"]["g1"]["Startup limit (MW)"] = 80
sc = UnitCommitment._from_json(json, repair = false)
@test UnitCommitment.repair!(sc) == 1
end
@testset "Startup costs and delays must be increasing" begin
json = parse_case14()
json["Generators"]["g1"]["Startup costs (\$)"] = [300, 200, 100]
json["Generators"]["g1"]["Startup delays (h)"] = [8, 4, 2]
sc = UnitCommitment._from_json(json, repair = false)
@test UnitCommitment.repair!(sc) == 4
end
end
Loading…
Cancel
Save