wdvorkin / dp_cc_opf

Privacy-preserving optimal power flow problem for distribution grids

DP-CC-OPF: Differentially Private Chance-Constrained OPF

This repository contains supplementary materials for the paper Differentially Private Optimal Power Flow for Distribution Grids by V. Dvorkin, F. Fioretto, P. Van Hentenryck, J. Kazempour, and P. Pinson.

The proof of Theorem 2 is contained in Appendix.pdf

The optimization models were implemented in Julia (v.1.4) using JuMP modeling language for mathematical optimization embedded in Julia. The models require Mosek comercial optimization solver, which needs to be installed and licensed.

To activate the packages in Project.toml, clone the project, e.g., using git clone, and cd to the project directory and call

$ julia 
julia> ]
(@v1.4) pkg> activate .
(DP_CC_OPF) pkg> instantiate

where julia is an alias to Julia installation. To run the code, cd to the project directory and call

$ julia DP_CC_OPF.jl

By default, the program returns the solution of the "CC_OPF" mechanism and stores the results in ~/output/CC_OPF. To run the other mechanisms, parse "D_OPF", "ToV_CC_OPF", "TaV_CC_OPF" or "CVaR_CC_OPF" using option -m, e.g.

$ julia DP_CC_OPF.jl -m "CVaR_CC_OPF"

The results will be stored in ~/output/CVaR_CC_OPF.

dp_cc_opf's Issues

The matrix D_u/U_n/D_I/U_I/T definition

Hello, dear wdvorkin
I am studying your DP_CC_OPF paper, it is a great job.
I have some questions about D_u/U_n/D_I/U_I/T definition in the program. Are they related to the flow direction of nodes?
I am checking the Fig. 4. in A. Papavasiliou, “Analysis of distribution locational marginal prices,” IEEE Trans. Smart Grid, vol. 9, no. 5, pp. 4872–4882, 2017.，and can’t understand how these matrices represent this figure.
Can you explain or send me some explanatory materials？ [email protected]
Thank you very much.