Article
Graph Modeling for Efficient Retrieval of Power Network
Model Change History
Ivana Dalˇcekovi´c
1,
* , Aleksandar Erdeljan
2
, Nikola Dalˇcekovi´c
1
and Jelena Marjanovi´c
1
Citation: Dalˇcekovi´c, I.; Erdeljan, A.;
Dalˇcekovi´c, N.; Marjanovi´c, J. Graph
Modeling for Efficient Retrieval of
Power Network Model Change
History. Energies 2021, 14, 8351.
https://doi.org/10.3390/en14248351
Academic Editors: Pierluigi Siano,
Hassan Haes Alhelou and
Amer Al-Hinai
Received: 30 October 2021
Accepted: 8 December 2021
Published: 11 December 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
Faculty of Technical Sciences, Department of Power, Electronic and Telecommunication Engineering,
University of Novi Sad, Trg D. Obradovi´ca 6, Novi Sad 21000, Serbia; nikola.dalcekovic@uns.ac.rs (N.D.);
jelena.stankovski@uns.ac.rs (J.M.)
2
Faculty of Technical Sciences, Department of Computing and Control Engineering, University of Novi Sad,
Trg D. Obradovi´ca 6, Novi Sad 21000, Serbia; ftn_erdeljan@uns.ac.rs
* Correspondence: ivana.kovacevic@uns.ac.rs
Abstract:
Power grids are constantly evolving, and data changes are increasing. Operational tech-
nology (OT) is controlled by IT technologies in smart grids, where changes in the physical world
impose changes in the software data model, as well as the continuous generation of data points,
resulting in time series datasets. The increased need for processing large amounts of data combined
with requirements to maintain and increase overall performances has created a significant challenge
for traditional database solutions and relational database models. The main idea of this paper was to
find and propose a graph model that will allow the retrieval of historical connectivity in a reduced
time complexity. Furthermore, the research question was addressed by evaluating three different
approaches where the results provide a foundation for the proposed design guidelines related to
optimizing graph-based databases for a modern smart grid system. The results of the experiments
demonstrated reduced time complexities from 3 to 5 times depending on the typical industry usage
patterns and the selected graph model. This suggests that the design decision may severely affect the
outcome for given smart grid use cases when using historical features in OT technologies. Therefore,
the main contribution of the research is the proposed guidelines on how to design an optimal graph
model that satisfies the described smart grid requirements.
Keywords: graph database; history; smart grids
1. Introduction
Power network grids (the grids herein) were designed in the previous century, but
the requirements and context of modern cities have forced the grid to evolve, rendering
current grids obsolete. Connectivity in the grids can vary as a consequence of: (1) changes
in the state of its elements such as switching the equipment on and off, or (2) physical
changes to the grid’s topology such as extending the feeders to new parts of the city or
replacing existing equipment. We focused on the first scenario, while the second scenario
introduces a much lower rate of changes in the equivalent period. If the grid is in an area
often exposed to hazards or climate disaster—connectivity will be more affected. In usual
scenarios, there are about a hundred changes during a day, and during storms (e.g., storm
mode), there are about several thousand. Distribution System Operators (DSO) make
various analyses as they must have insight into the connectivity of the whole network,
both in real-time and to keep a history for training purposes and post-accident analyses.
As operational technology (OT) is controlled by IT technologies in smart grids, changes in
the physical world impose changes in the software data model. Processing large amounts
of data, while maintaining performances for a real-time decision-making software system,
creates a problem for traditional database solutions and relational database models.
One of the main advantages of graph databases in smart grids over relational databases
and NoSQL stores is performance, as they are optimized for the graph data models. Graph
Energies 2021, 14, 8351. https://doi.org/10.3390/en14248351 https://www.mdpi.com/journal/energies
