Working with Multi-Network Data

import Pkg
using PowerModels

case3file = Pkg.dir(dirname(@__DIR__), "test", "data", "matpower", "case3.m")

There are occasions when it is desirable to co-optimize multiple networks, these networks might encode time for dynamic network optimization, or scenarios for stochastic optimization.

To distinguish between network data (see The Network Data Dictionary) that correspond to a single network or to multiple networks, PowerModels.jl provides the function ismultinetwork(). For example, we can do the following:

println(case3file)
network_data = PowerModels.parse_file(case3file)
pm = instantiate_model(network_data, ACPPowerModel, PowerModels.build_opf)

PowerModels.ismultinetwork(pm)

PowerModels.jl would generally not read in network data as multi-networks. To generate multiple networks from the same network data, we use the following method

For example, we can make three replicates by calling

network_data3 = PowerModels.replicate(network_data, 3)

Observe that the structure of network_data3 is different from that of network_data, since it is a multi-network. The user can then modify each replicate of the network to vary in the corresponding parameter of interest. See test/common.jl for examples on setting up valid Multi-Network data.

To build a PowerModel from a multinetwork data dictionary (see Building PowerModels from Network Data Dictionaries), we supply multinetwork=true during the call to build_generic_model and replace build_opf with build_mn_opf,

pm3 = PowerModels.instantiate_model(network_data3, ACPPowerModel, PowerModels.build_mn_opf, multinetwork=true)

PowerModels.ismultinetwork(pm3)

data33 = PowerModels.replicate(data3, 3)

Because this is a common pattern of usage, we provide corresponding calls to solve_mn_opf (which behaves analogously to solve_opf, but for multinetwork data).