Modeling and Simulation Case Studies | JuliaHub

JuliaHub, JuMP for mathematical optimization | Case Study

Written by Los Alamos | Jan 2, 2023 9:53:54 PM

Los Alamos National Laboratory (LANL) is one of the oldest and largest laboratories in the US Department of Energy National Laboratory system. The LANL Advanced Network Science Initiative (ANSI) team uses Julia for critical infrastructure optimization including:

  • Predict and mitigate the impact of extreme events: helping utilities and government agencies plan for and mitigate potential outages from hurricanes, ice storms or earthquakes
  • Grid Optimization Challenge: develop a new paradigm for testing and evaluation of emerging grid optimization algorithms to accelerate adoption (ARPA-e benchmark code)
  • Modeling and preparing responses to cyberattacks and other threats

As LANL scientist Carleton Coffrin explains, our dependence on the power grid has increased dramatically. Financial markets, commerce, medicine, aviation safety, transportation, law enforcement and national security depend crucially on power that is always on – and they need more of it than ever before.

At the same time, power grids are subject to more dynamic events – increased use of wind, water and solar energy, more extreme weather events, and new heavy users of electricity such as electric cars, smart home devices and commodity high performance computing increase the variability of both supply and demand.

Coffrin asks,

How can we design and operate our critical infrastructure to improve its resilience in response to a more dynamic and disruptive world?

To solve these problems, the ANSI team uses JuMP for mathematical optimization in Julia together with more than 25 free and open source Julia packages created and contributed to by LANL scientists including InfrastructureModels.jl, PowerModels.jl, GasModels.jl, GasGridModels.jl, WaterModels.jl, Alpine.jl, Juniper.jl, PowerModelsDistribution.jl, PowerModelsProtection.jl, PowerModelsONM.jl, WaterModelsAnnex.jl, GasModelsMultiPipeline.jl, PowerModelsAnnex.jl, PowerModelsRestoration.jl, PowerModelsStability.jl, PowerModelsGMD.jl, PowerModelsAnalytics.jl, GasPowerModels.jl, GrapicalModelLearning.jl, PowerModelsSecurityConstrained .jl, MomentOpt.jl and PetroleumModels.jl.

The LANL Advanced Network Science Initiative team has made Julia and JuMP their core research platform to build commercial-grade power transmission optimization software.

Why Julia and JuMP?

1. Speed and scale for massive problems: Solving the Grid Optimization Challenge involves 1-5 billion decision variables and constraints and provides a solution within 5 minutes

2. Track record of success across a wide range of use cases: Transmission Network Storage OptimizationTransmission Network Vulnerability and ResilienceImpacts from Solar Storms

And the results?

Coffrin reports the following stats for PowerModels.jl:

Today PowerModels.jl is just one part of the much larger InfrastructureModels.jl ecosystem:

  • InfrastructureModels .jl: PetroleumModels.jl, PowerModelsDistribution.jl, PowerModels.jl, GasModels.jl, WaterModels.jl
  • 85,000+ lines of code
  • 1000s of labor hours
  • LANL Distinguished Performance Award 2019