HierarchicalNetwork

Solve the multi-stage stochastic linear programming for the hierarchical distribution network.

REQUIRE Julia v1.0 and Gurobi solver with JuMP package

1. Data Generation
2. Method
   1. Perfect Foresight
   2. Scenario-Based Robust MPC with Nested Dantzig-Wolfe

Data Generation

For the data generation (e.g. network topology, demand profile, lattice of PV etc..), please check the notebook test/data_generation.ipynb. The generated data is saved in data/Case1.

Method

Here we present the description of methods for managing uncertainty.

Perfect Foresight

run_parallel_perfect_foresight.jl implements parallelly the perfect foresight policy on the network to decide the interface flow values (p_in and p_out) which will be used for SDDP implementation. The result is saved as JLD2 format.

Implementation is done for one branch of the medium voltage network. I would recommend to set the number of samples, NSamples, greater than 500 in order to get 'plausible' data.

Scenario-Based Robust MPC with Nested Dantzig-Wolfe

test/run_dantzig_wolfe.jl implements the SBR-MPC with the nested DW to solve the problem. Again, we consider on one branch of the medium voltage network.