Results for “AI/ML/DT”

DOE OSTI2025

Bipolar plate flow channel designs for vanadium redox flow battery: a review

Vanadium redox flow battery is one of the preferred systems for grid scale energy storage due to long service life (>20000 cycles), higher efficiency (>85 %), deep discharge capability (>95 % DoD), and inherent scalability and safety. Among system components, flow field is most critical as it governs electrolyte distribution, mass transport and hydraulic performance. Here, this review examines emerging flow channel geometries, highlighting the impact of channel width (0.66 to 1.5 mm), depth (1 to 1.5 mm) and land width (0.5 to 1.5 mm) can reduce pressure drop to <10 kPa while enabling power densities above 600 mW.cm -2 . A key finding is that low channel width to depth ratio (<1) enhances voltage and energy efficiencies by improving under rib convection. The review also provides an overview of progress and perspective in bipolar plate materials, manufacturability, and shunt current mitigation strategies for stack scaling up. Cost analysis emphasizes the influence of flow field designs on the levelized cost of storage and pathways towards the US DOE's $\$$0.05 per kWh energy generation cost target. In addition, opportunities for AI/ML/DT tools assisted design and data driven optimization strategies are also outlined to accelerate next generation flow field development.

Efficiency optimization↗

DOE OSTI2023

Dynamic Model Agnostic Reliability Evaluation of Machine-Learning Models Integrated in Instrumentation & Control Systems

In recent years, the field of data-driven neural network-based machine learning (ML) algorithms has grown significantly and spurred research in its applicability to instrumentation and control systems. While they are promising in operational contexts, the trustworthiness of such algorithms is not adequately assessed. Failures of ML-integrated systems are poorly understood; the lack of comprehensive risk modeling can degrade the trustworthiness of these systems. In recent reports by the National Institute for Standards and Technology, trustworthiness in ML is a critical barrier to adoption and will play a vital role in intelligent systems' safe and accountable operation. Thus, in this work, we demonstrate a real-time model-agnostic method to evaluate the relative reliability of ML predictions by incorporating out-of-distribution detection on the training dataset. It is well documented that ML algorithms excel at interpolation (or near-interpolation) tasks but significantly degrade at extrapolation. This occurs when new samples are "far" from training samples. The method, referred to as the Laplacian distributed decay for reliability (LADDR), determines the difference between the operational and training datasets, which is used to calculate a prediction's relative reliability. LADDR is demonstrated on a feedforward neural network-based model used to predict safety significant factors during different loss-of-flow transients. LADDR is intended as a "data supervisor" and determines the appropriateness of well-trained ML models in the context of operational conditions. Ultimately, LADDR illustrates how training data can be used as evidence to support the trustworthiness of ML predictions when utilized for conventional interpolation tasks.

97 MATHEMATICS AND COMPUTING↗