Originally Contributed by: Clayton Barrows and Sourabh Dalvi
The [Operations Problems example]](https://nbviewer.jupyter.org/github/NREL-SIIP/SIIPExamples.jl/blob/master/notebook/3_PowerSimulations_examples/02_sequential_simulations.ipynb) shows the basic building blocks of building optimization problems with PowerSimulations.jl. This example shows how to customize the enforcement of branch flow constraints as is common when trying to build large scale simulations.
using PowerSystems
using PowerSimulations
using PowerSystemCaseBuilder
For this simple example, we can use the HiGHS solver with a relatively relaxed tolerance.
using HiGHS # mip solver
solver = optimizer_with_attributes(HiGHS.Optimizer, "mip_rel_gap" => 0.5)
System from RTS-GMLC data¶sys = build_system(PSITestSystems, "modified_RTS_GMLC_DA_sys")
Since PowerSimulations will apply constraints by component type (e.g. Line), we need to change the component type of the lines on which we want to enforce flow limits. So, let's change the device type of certain branches from Line to MonitoredLine differentiate treatment when we build the model. Here, we can select inter-regional lines, or lines above a voltage threshold.
for line in get_components(Line, sys)
if (get_base_voltage(get_from(get_arc(line))) >= 230.0) &&
(get_base_voltage(get_to(get_arc(line))) >= 230.0)
#if get_area(get_from(get_arc(line))) != get_area(get_to(get_arc(line)))
@info "Changing $(get_name(line)) to MonitoredLine"
convert_component!(MonitoredLine, line, sys)
end
end
Let's start with a standard unit commitment template using the PTDFPowerModel network
formulation which only constructs the admittance matrix rows corresponding to "bounded" lines:
template = template_unit_commitment(network = PTDFPowerModel)
Notice that there is no entry for MonitoredLine, so we can add one:
set_device_model!(template, MonitoredLine, StaticBranch)
We can also relax the formulation applied to the Line components to an unbounded flow formulation.
This formulation still enforces Kirchoff's laws, but does not apply flow constraints.
set_device_model!(template, Line, StaticBranchUnbounded)
OperationsProblem¶uc_prob = DecisionModel(template, sys, horizon = 24, optimizer = solver)
build!(uc_prob, output_dir = mktempdir())
Solve the relaxed problem
solve!(uc_prob)
This notebook was generated using Literate.jl.