AC Optimal Power Flow in Various Nonlinear Optimization Frameworks

ChrisRackauckas · September 13, 2023, 8:40pm

Update: the benchmark is now running on the SciMLBenchmarks! You can find it here:

https://docs.sciml.ai/SciMLBenchmarksOutput/dev/OptimizationFrameworks/optimal_powerflow/

Anyone can trigger the benchmark to run just by changing the code here:

SciML/SciMLBenchmarks.jl/blob/master/benchmarks/OptimizationFrameworks/optimal_powerflow.jmd

---
title: Optimal Powerflow Nonlinear Optimization Benchmark
author: Chris Rackauckas
---

## Data Load and Setup Code

This is generic setup code usable for all solver setups. Basically removing some unnecessary untyped dictionaries
before getting to the benchmarks.

```julia
PRINT_LEVEL = 0

# This is a soft upper limit to the time of each optimization.
# If times go above this, they will halt early
MAX_CPU_TIME = 100.0

# Maximum number of variables in an optimization problem for the benchmark
# Anything with more variables is rejected and not run
# This is for testing. 100 is a good size for running a test of changes

This file has been truncated. show original

It makes a cute table at the end with the current results:

Things to note:

I know the table formatting can be improved. If anyone knows how to make documenter show the whole thing I’d be happy to have the help in formatting it.
There is a parameter SIZE_LIMIT which tells the benchmark the maximum number of variables to use in the benchmark files. This is set to 100 right now just to play around with formatting, but I’ll bump it up and do a bigger run. I did do a bigger run a bit back so I know it works, just when fixing table formatting it’s annoying to wait a day haha.
Optimization.jl code was horrendously allocating and type-unstable. A common issue in the code was using d["$i"] on a dictionary, instead of just using an integer. The dictionaries are still there, so there’s still some indirection that should be removed and the performance should increase, but at least it’s type-stable and non-allocating now. I’ll actually optimize the code in the near future, but the code now is at least a lot simpler than the original code and won’t interfere with AD tooling.
Optimization.jl is now using a sparse reverse mode, but it’s not optimized yet. This case is a case we’re using to optimize things. See:

which are low hanging fruit identified from this.

I added a lot of validation cases so if you change the code around, assertions will throw failures if you break something.
I couldn’t get the CASADI code to run, I’d like some help figuring that out.