Performance and Robustness in Smart Predictive Digital Twins: A Review with Emphasis on Water Supply Systems

TEchMA 2025 New Frontiers in Mechanical Engineering recuperarportugal.gov.pt PIML[10] Wavelet Feature Extraction[8] Performance and Robustness in Smart Predictive Digital Twins: A Review with Emphasis on Water Supply Systems Abstract Water Supply Systems (WSS) play a critical role in ensuring a reliable and sufficient water supply to residential, agricultural and industrial sectors. In Europe, WSS consumes 1,13% of all energy, of which at least 60% is used by pumping stations [1]. Efficient WSS management is essential but complex due to factors like pumps’ variablespeed drivers, dynamic energy tariffs, local energy production and staff shortages, especially as experienced personnel retire [2]. Digital Twins (DT) offer a promising solution by replicating real systems and supporting operational decisionmaking [3]. More mature DT - Smart Predictive Digital Twins (SPDT) - integrate cloud computing, predictive engines, and optimisation modules, functioning as decision support tools to reduce operational costs. Despite their potential, SPDT remain underused in real-time applications, reflecting their immaturity and limited trust, particularly in critical systems such as WSS. This review investigates how SPDT performance and robustness are currently evaluated, with a focus on critical systems as WSS. A targeted literature review highlights the fragmented nature of performance assessment and the reactive implementation of robustness strategies. Few comprehensive frameworks exist, and approaches vary widely across studies and domains. This work synthesizes promising practices and underscores the need for systematic evaluation methods to build more trustworthy SPDT. Strengthening performance and robustness is essential for broader adoption and sustainable, efficient WSS operation. Motivation Water Supply Systems (WSS) are essential to life, yet they face increasing challenges in terms of management [1, 2]. To solve these issues, there is a need to implement smarter and more adaptative tools for supporting real-time decision making. Conclusions This review support the development of a structured evaluation framework, which can increase operators’ trust. By organizing existing robustness techniques along the SPDT development process, it shows how these can be integrated early improving both interoperability and system resilience. The adoption of trustworthy SPDT will enable data-driven decision-making and contribute to more sustainable and efficient WSS operations. Mariana Alão, Ana Luísa Reis, António Andrade-Campos Acknowledgments This work is supported by the doctoral grant (Ref. 2024.04917.BDANA) financed by the Portuguese Foundation for Science and Technology (FCT), by the FEDER and Regional Operational Program of the Center Region (CENTRO2030) within project I-ReTiS-LeaksD&Op nº 17304 (CENTRO2030-FEDER-01177300) and through the Portuguese Foundation for Science and Technology (FCT), supported by the Recovery and Resilience Plan (PRR), within project I-ReTiS-Leaks (2024.07270.IACDC). This work is funded by national funds through FCT –Fundação para a Ciência e a Tecnologia, I.P., under the project/support UID/00481 –Centre for Mechanical Technology and Automation (TEMA). Insights Evaluation practices and robustness techniques were organised according to their timing in the SPDT development process, demonstrating how they are addressed across different phases. References [1] - International Energy Agency, Energy, water and the Sustainable Development Goals: Excerpt from World Energy Outlook 2018, IEA, Paris, France, 2018. [Online]. Available: https://www.iea.org/reports/energy-water-and-the-sustainabledevelopment-goals. [Accessed: 28 May, 2025]. [2] - C. Kalyani, K. Pradnya, C. Samkit, and V. Patil, ‘Regular and Equal Water Supply System’, SAMRIDDHI : A Journal of Physical Sciences, Engineering and Technology, vol. 15, no. 01, pp. 34–37, Jan. 2023, doi: 10.18090/samriddhi.v15i01.05. [3] –M. Segovia and J. Garcia-Alfaro, “Design, modeling and implementation of digital twins,” Sensors, vol. 22, no. 14, p. 5396, 2022. doi: 10.3390/s22145396 [4] – A. Chong, Y. Gu, and H. Jia, ‘Calibrating building energy simulation models: A review of the basics to guide future work’, Dec. 15, 2021, Elsevier Ltd. doi: 10.1016/j.enbuild.2021.111533. [5] –F. Psarommatis and G. May, ‘A standardized approach for measuring the performance and flexibility of digital twins’, Int J Prod Res, vol. 61, no. 20, pp. 6923–6938, 2023, doi: 10.1080/00207543.2022.2139005. [6] – J. Zeng et al., ‘Machine learning enabled fusion of CAE data and test data for vehicle crashworthiness performance evaluation by analysis’, Structural and Multidisciplinary Optimization, vol. 66, no. 4, Apr. 2023, doi: 10.1007/s00158-02303553-5. [7] – L. Hui, M. Wang, L. Zhang, L. Lu, and Y. Cui, ‘Digital Twin for Networking: A Data-Driven Performance Modeling Perspective’, IEEE Netw, vol. 37, no. 3, pp. 202–209, May 2023, doi: 10.1109/MNET.119.2200080. [8] – X. Zhang and Y. Lu, ‘Robustness of wavelet energy features for damage detection with neural network and digital twin’, Proceedings of the Institution of Civil Engineers: Engineering and Computational Mechanics, Sep. 2024, doi: 10.1680/jencm.24.00008. [9] –A. Di Meglio, N. Massarotti, and P. Nithiarasu, ‘A physics-driven and machine learning-based digital twinning approach to transient thermal systems’, Int J Numer Methods Heat Fluid Flow, vol. 34, no. 6, pp. 2229–2256, Jul. 2024, doi: 10.1108/HFF-10-2023-0616. [10] – P. Laplante, ‘Trusting Digital Twins’, Computer (Long Beach Calif), vol. 55, no. 7, pp. 73–77, Jul. 2022, doi: 10.1109/MC.2022.3149448. [11] - J. Jeon and G. Theotokatos, ‘A Framework to Assure the Trustworthiness of Physical Model-Based Digital Twins for Marine Engines’, J Mar Sci Eng, vol. 12, no. 4, Apr. 2024, doi: 10.3390/jmse12040595. [12] –R. Vrabič, J. A. Erkoyuncu, M. Farsi, and D. Ariansyah, ‘An intelligent agent-based architecture for resilient digital twins in manufacturing’, CIRP Annals, vol. 70, no. 1, pp. 349–352, Jan. 2021, doi: 10.1016/j.cirp.2021.04.049. [13] - M. Lezoche and H. Panetto, ‘Cyber-Physical Systems, a new formal paradigm to model redundancy and resiliency’, Enterp Inf Syst, vol. 14, no. 8, pp. 1150–1171, Sep. 2020, doi: 10.1080/17517575.2018.1536807. Objective Through a targeted literature search, this work synthesises how the performance and robustness of SPDT are currently addressed, especially in their application in critical systems such as WSS. Analytical Performance Metrics[4] Component Development Before Deployment In Real Time Time Related Metrics[5] Context Specific Metrics[6] DT Flex Approach[5] Efficiency[7] Flexibility[7] When to evaluate DT Model Redundancy[9] MCUQ[11] Real-time Updating[12] Quality Assurance[13] How to increase DT robustness Smart Predictive Digital Twins as Decision Support Systems Digital Twins (DT) are dynamic digital replicas that can mimic the physical with the virtual world in real-time, making them indistinguishable [3]. They are composed by 3 core components: →Physical asset,including the sensors and actuators connected to them; →Virtual asset, that reproduces the physical asset with high fidelity; →Connection between them allowing the communication and synchronization [3]. Smart Predictive Digital Twins (SPDT) are a more mature form of DT that have other components that allow more intensive calculus and providing them predictive and optimising capabilities. Growing awareness for using evaluation frameworks The need for these frameworks is frequently mentioned, when specific methods are not. Evaluation methods are uncoordinated Approaches vary widely depending on the problem with no standard framework Robustness strategies are mostly reactive Usually applied as workarounds instead of systematically Lack of holistic evaluation framework Currently the long-term performance and resilience are not addressed Essential for residential, agricultural and industry Physical Asset Virtual Asset Predictive Engine Physical Asset Virtual Asset Optimiser They still face some issues, related with lack of robustness, real-time interoperability and model accuracy. Need for Decision Support System WSS provide water High energy consumers Complex Systems Retiring Operators At least 60% for pump operators Multiple Variables Loss valuable knowledge Fig 1 / Key challenges driving the shift in Water Supply System management. Fig 2 / Digital Twins and Smart Predictive Digital Twins composition. Fig 3 / Mapping evaluation metrics and robustness techniques to different stages of SPDT development. Real-time communication Predictive communication