**1. Introduction**

Digital transformation is an essential element for urban governance, enabling the efficient coordination and cooperation of multiple stakeholders involved in the urban environment. In the last decade, the concept of the smart city has gained tremendous importance and public/private institutions have started thinking about innovative ways to develop solutions tackling the complexity of the various phenomena [1]. At their core, these solutions rely on IoT and ICT to realize the perception, control and intelligent services aimed to optimize resource usage, bettering the services for the citizens [1,2]. To this aim, real-time access to accurate and open data is central to unlocking the economic value of the smart city potential, opening a rich ecosystem to suppliers for the development of new applications and services [3].

Digital Twin (DT) technology has revived the interest in the smart city concept, understood as a digital replica of an artefact, process or service, sufficient to be the basis for decision making [4]. This digital replica and the physical counterpart are often connected by streams of data, feeding and continuously updating the digital model, used as a descriptive or predictive tool for planning and operational purposes. While the concept of Digital Twin is by no means new, recent advances in 5G connectivity, AI, the democratization of sensing technology, etc., provide a solid technological basis and a new framework for

**Citation:** Bujari, A.; Calvio, A.; Foschini, L.; Sabbioni, A.; Corradi, A. A Digital Twin Decision Support System for the Urban Facility Management Process. *Sensors* **2021**, *21*, 8460. https://doi.org/10.3390/ s21248460

Academic Editor: Antonio Puliafito

Received: 13 November 2021 Accepted: 17 December 2021 Published: 18 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**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/).

investment. As an example, ABI Research predicts that the cost benefits deriving from the adoption of urban DTs alone could be worth USD 280 billion by 2030 [5].

Recognizing these benefits, different public and private institutions have already started investing in the technology, and this fact is attested by several urban DT initiatives, aimed at addressing different and complex problems in the smart city context. In the H2020 DUET project, the Flanders region aims at building a multi-purpose DT platform addressing the impact of mobility on the environment, to reduce the impact on human health [6]. In the same project, the city of Pilsen is building a proof-of-concept DT focusing on the interrelation between transport and noise pollution. The pilot aims to demonstrate the concept of the technology across transport and mobility, urban planning and the environment and well-being limits [7].

Recently, Bentley Systems and Microsoft announced a strategic partnership to advance DT for city planning and citizen engagemen<sup>t</sup> [8]. This partnership materialized in 2020 in a collaboration with the city of Dublin, pursuing the development of a large-scale urban DT used as a support tool for citizens to engage, from the safety of their own homes, in new development projects in their local communities.

It is evident that the success of an urban Digital Twin is directly impacted by the quality and quantity of the data sources that it relies on to model the physical counterpart. The modeling component, being either visual, e.g., based on a dashboard showing a consolidated view of the data, or data-driven, e.g., AI or statistical, is another core element of the framework. In this work, we discuss a concrete DT solution for the Urban Facility Management (UFM) process. The process under scrutiny comprises multiple stakeholders acting on a shared environment, e.g., different companies involved in the maintenance of various urban assets, each having a local view of the overall maintenance process. Their view on the activities is periodically consolidated in a joint conference called by the municipality, where mid-to-long-term operational details are discussed and a global, coarse-grained view of the process is established.

The scope and aim of the project is to showcase the use of DT technology as a decision support system, guiding UFM operators in their activity through the use of a rich set of correlation tools depicting the activities inside an area of interest. The system is equipped with a scheduling functionality, consulted to find feasible schedules for maintenance interventions, while minimizing some predetermined indexes, e.g., disturbance on mobility, interferences with other planned city events, etc. From a technological viewpoint, the Interactive Planning Platform for City District Adaptive Maintenance Operations (IPPODAMO [9]) is a proof-of-concept DT consisting of a (distributed) multi-layer geographical system, fed with heterogeneous data sources originating from different urban data providers. The data are initially staged at ingestion points, dedicated ingress machines, where they undergo syntactic and semantic transformations, and are successively forwarded to a big data processing framework for further refinements. The data are subject to different algorithmic processes, aimed at building a coherent view of the dynamics inside an area of interest, exploited by the UFM operated to make informed decisions on potential future maintenance operations. This work builds on a prior work [10], extending the study with a detailed discussion of some core system components, showcasing the advanced capabilities of the decision support system.

The article is organized as follows. Section 2 provides a concise background on the concepts and technological ecosystem adopted in this work. Section 3 briefly describes some data sources considered in the project, followed by a high-level functional overview of the multi-layer geographical system. Section 4 presents two distinct big data processing pipelines, providing some insights into their implementation and performance trends. While preserving the general aspect of our study and without loss of generality, in Section 5, we present the use cases and functionalities currently targeted by the platform. Finally, Section 6 draws the conclusions, delineating some future work.
