Multi-objective design-for-control of water distribution networks with global bounds (OptE)

Published: 17-01-2021| Version 2 | DOI: 10.17632/ppnthgdgcp.2
Contributor:
Joy Ulusoy

Description

This dataset describes the bi-objective design-for-control problems and presents the obtained results for two network models (modified Net25 and pescara) studied in the publication "Multi-objective design-for-control of water distribution networks with global bounds", by Aly-Joy Ulusoy, Filippo Pecci and Ivan Stoivanov. The general design-for-control problem consists in simultaneously selecting elements to be added to the existing water distribution networks (WDN) from predefined sets of candidate valves (CNV) and pipes (CNP), and optimizing the controls of new and existing pressure control valves. The problem considers the conflicting objectives of AZP (or average zone pressure, a surrogate measure of pressure induced pipe stress and leakage) and I_r (or resilience index, a surrogate measure of network resilience). We refer to the main manuscript for a definition of AZP and I_r, problem formulation and interpretation of the corresponding results presented in this dataset (i.e. results of the optimal bi-objective design-for-control problems BOMINLP_{AZP,-I_r}(Q) and BOMINLP_{-I_r,AZP}(Q)). The information and results provided in this dataset aim to facilitate future comparison of our work against alternative solution methods. Accordingly, we provide below, for each network model: - the .inp network model, to access using EPANET or a text editor, with characteristics of network pipes (diameter, length, roughness coefficients) and nodes (elevation, demand). Please use the provided network models (25nodes_PRVs2.inp and pescara_ed.inp) for comparison purposes, as the original models they are based on have been modified for the purpose of this study. - a description of the design-for-control problem investigated by the authors of the main manuscript and corresponding to the presented results - the results of the optimal bi-objective design-for-control problems (BOMINLP_{AZP,-I_r} and BOMINLP_{-I_r,AZP}) obtained by the authors with the proposed solution method (candidate valves and pipes selected for installation in solutions in L^{NDS}, corresponding AZP and resilience index objective values, and lower bound set L^{LBS}). Solutions in L^{NDS}_{AZP,-I_r} and L^{NDS}_{-I_r,AZP} belonging to the final sets L^{NDS} (defined in Section 3.3 of the main manuscript, and represented in Figures 1c and 3c) are highlighted in dark blue. - the AZP and -I_r values corresponding to the optimal configuration solutions in L^{NDS}_{AZP,-I_r} and L^{NDS}_{-I_r,AZP}, computed using the H-W friction head loss formula Finally, we refer the reader to the main manuscript for illustrations of the results and a description of how to obtain, based on the results presented in the supplementary material, an outer approximation of the Pareto front of the bi-objective design-for-control problems (i.e. potentially non-dominated solutions with global non-dominance bounds).

Files