Tennet (Netherlands)

Abstract The integration of artificial intelligence (AI) into safety-critical systems, where human operators remain central to decision-making, introduces various challenges that existing AI frameworks struggle to address comprehensively. Key concerns involve designing a socio-technical system that balances AI transparency, trust, and explainability with the imperative for robust and reliable decision-making. Presently, while numerous sector-specific solutions exist, a holistic framework that effectively integrates human expertise with AI capabilities remains absent, leaving critical gaps in system design, deployment, and oversight. This chapter proposes a multidisciplinary conceptual framework to enhance human-AI collaboration in critical infrastructures such as power grids, railways, and air traffic management. The different design steps were guided by the requirements of these industrial domains. The framework combines key design principles that support human cognition, leveraging insights from decision theory, mathematics, and specialized engineering domains to optimize AI-assisted decision-making. Furthermore, it embeds trustworthiness and risk assessment methodologies, using tools such as the Assessment List for Trustworthy Artificial Intelligence (ALTAI) tool to ensure compliance with ethical and regulatory requirements.

The Dutch Electricity Grid Code defines requirements for the quality of the supply voltage. For HV (110kV/150kV) and EHV (220kV/380kV), every commercial connection must be equipped with a PQ meter, resulting in approx. 130 PQ meters. The paper reviews the measured trends between 2019 and 2023. It can be concluded that overall, compared to 2019, steady state voltages and voltage unbalance have increased, measured THD has decreased for most nodes and no major changes are noticeable with respect to voltage dips. TenneT is taking various measures to remain compliant with the Grid Code requirements. As most of the power quality phenomena propagate to lower voltage levels, this trend analysis is also important for the DNOs.

We present a massively parallel solver that accelerates DC loadflow computations for power grid topology optimization tasks. Our approach leverages low-rank updates of the Power Transfer Distribution Factors (PTDFs) to represent substation splits, line outages, and reconfigurations without ever refactorizing the system. Furthermore, we implement the core routines on Graphics Processing Units (GPUs), thereby exploiting their high-throughput architecture for linear algebra. A two-level decomposition separates changes in branch topology from changes in nodal injections, enabling additional speed-ups by an in-the-loop brute force search over injection variations at minimal additional cost. We demonstrate billion-loadflow-per-second performance on power grids of varying sizes in workload settings which are typical for gradient-free topology optimization such as Reinforcement Learning or Quality Diversity methods. While adopting the DC approximation sacrifices some accuracy and prohibits the computation of voltage magnitudes, we show that this sacrifice unlocks new scales of computational feasibility, offering a powerful tool for large-scale grid planning and operational topology optimization.

As roof space has become increasingly occupied, facades serve as an alternative for deploying PV, often in the form of Building-Integrated Photovoltaics (BIPV). In this study, a comprehensive analysis over multiple years of the value of integrating PV on facades with different configurations is provided. The comparison comprises financial, technical and environmental metrics and spans the Dutch and German electricity sector. As expected, due to the much higher yield of optimally oriented PV, it generates higher revenues and reduce more CO2 emissions than facade PV. Over the course of five years of this study, south, east, and west facades reduce 1725, 1492, 1335 kg of CO2 emission per kWp of PV installation, while this is 2434 for the optimally oriented PV. However, analyzing the value factor of PV-generated electricity shows east and west facade are increasingly generating more economic value compared to optimally-oriented PV: the value factor or capture rate of optimally-oriented PV decreased to 0.73 in 2023, whereas the value factor for east facade and west facade remained higher (0.87 and 0.84, respectively). From a technical viewpoint, facade PV results in much higher self-consumption ratios 42% for east facade, 46% for west facade, compared to only 30% for optimally-oriented (all 1 kWp installed per 1 MWh demand). Concluding, facade PV, while having a much lower yield than optimally oriented PV, does hold some significant advantages over optimally-oriented PV – mainly in technical terms and in economic terms. Therefore, facade PV can be seen as a very promising option for the PV sector in general, and the BIPV sector specifically – especially in a more and more congested electricity grid.

This paper aims to provide an insight into the measured background changes of harmonics due to operational changes in a typical Dutch transmission grid. Multiple use cases on different locations throughout a meshed 150kV grid have been considered. The nodes that were studied had measured exceedances of planning levels or were indicated to be critical for the future in earlier studies. This study provides an insight into the measured response of harmonics with respect to different operational changes such as specific scheduled outages that occurred and the impact of capacitor banks. Per use-case, individual conclusions are reported. The analysis was conducted on data of power quality meters (PQM) and various other data sources provided by the Dutch TSO TenneT. Data-processing, visualization, and computations were performed using Python. These results are useful for model validation, planning purposes, and maintaining power quality.

Variable Renewable Energy (VRE) sources are characterized by production uncertainty that is largely due to weather forecast errors. This leads to greater volumes in the real-time balancing markets that drive Imbalance Market (IM) price higher, creating economic opportunities for flexibility providers. However, lack of information on foreseen IM prices and regulation states at the time of day-ahead market closure, reduces the flexibility providers' opportunity of optimizing their portfolio. The objective of this paper is to investigate to which extent there is a correlation between the IM prices and weather parameters in the Dutch Imbalance Market. A Deep Learning (DL) model based on Feed-Forward Deep Neural Network (FFDNN) is developed with the aim to support VRE asset owners and flexibility providers to predict IM price ranges and regulation states in the day-ahead horizon. The parameters considered are temperature, solar radiation, wind speed, relative humidity, and cloud cover. A benchmark model using Support Vector Machine (SVM) is used to compare with the DL's model performance. Both models are trained and tested using data from the weather prediction model provided by Royal Netherlands Meteorological Institute (KNMI), and historical IM prices from the Dutch Transmission System Operator (Tenne'T), for the years 2018–2020.

Elevated thermal stress-related aging is significant on the oil-impregnated paper (OIP) used as insulation in high-pressure gas cables (HPGCs). The aim of this article is to develop a cheap alternative for lab dielectric measuring and characterizing temperature-dependent parameters for OIP. First, this article derives the operating thermal conditions of the grid-aged cable based on IEC standards after analyzing the loading data using machine learning techniques to determine the elevated temperature levels for the experiments. Second, a novel lab-fabricated inexpensive electronics circuit is developed for polarization and depolarization current (PDC) measurements which can be adapted for such measurements over expensive commercial devices. From the measured parameters, an extended three-branch Debye model is optimized using a developed error function approach based on the Akaike information criterion (AIC) and goodness of fit. The model indicated a reduction in the branch resistance with temperature elevation and aging, whereas the branch capacitance revealed an increasing trend. The resultant relaxation time (RC) showed a decrease overall. Last, a short-duration frequency domain spectrum was analyzed and extrapolated to obtain parameters for a wide range of frequencies and fit in a Cole–Cole model, derived for oil-paper insulation. The time constants obtained from this model also confirmed a reducing trend across the temperature and aging variations and the model parameter, the alpha coefficient showed a decreasing trend. Last, the effect of the measured dielectric parameters is reflected with breakdown values to investigate the effect of temperature on the electrical life of insulation.

Operational asset management describes the asset management function which deals with the shortest time horizon, and relates to asset decision-making at its most granular level. Decisions revolve around prioritizing which devices to target with replacements, which maintenance activities to choose, and when is the best time to perform them. Operational asset management follows an ongoing process in which understanding failure modes, with their probabilities and consequences, is essential. Condition assessment is fundamental to deciding what mitigations to apply and when. A range of testing and data collection methods are necessary to determine which populations are in what condition. Failure modes and associated risks vary with asset type and severity and therefore require different types of remediation.Investment in gathering equipment condition and criticality data is necessary to provide a quantitative basis for prioritization of maintenance actions and investment. Operational asset management focuses on near-term asset investment planning, such as more versus less maintenance, time-based versus condition-based versus reliability centered maintenance (RCM) etc. and mixtures of these, asset repair versus end-of-life management, asset overloading vs. loss of life decision-making.Operational asset managers use asset registries and asset data management tools, asset condition assessment, asset monitoring data, and interpretation, health indices, criticality, to justify and rank near-term asset investment decisions, as well as spares and spare parts utilization/planning.