A 17-year-old student from Montvale, New Jersey, has won a $50,000 scholarship for developing an artificial intelligence model that dramatically improves the accuracy of a critical nuclear reactor safety calculation.
Praadhyumn Indaana, a 2026 Davidson Fellow, reduced the average prediction error for critical heat flux (CHF) from approximately 63% with conventional formulas to just 5.5%, according to the Davidson Institute, which announced the award. The achievement has implications for nuclear reactor engineering, where the ability to predict thermal limits more precisely could eventually reduce the conservatism built into reactor design and operation.
The technical challenge Indaana tackled centers on the point at which boiling inside a nuclear reactor can no longer effectively remove heat. This threshold, known as critical heat flux, occurs when steam begins to displace liquid water from the surface of a fuel rod, causing cooling to deteriorate rapidly and the rod temperature to rise sharply. Because exceeding this limit carries potential safety consequences, reactor operators and engineers rely on mathematical formulas to predict where it occurs. However, Indaana identified that conventional equations are often developed and tuned around particular experimental conditions, and their accuracy can decline when applied outside those conditions.
His solution was a physics-regularized neural network — an AI model that incorporates constraints based on physical theory rather than learning solely from experimental data. The model was trained using data from more than 10,000 real experiments, and Indaana also incorporated bounds derived from hydrodynamic instability theory to prevent the model from producing predictions that violate known physical limits. This distinction proved important: early versions of his models performed well in data-rich regions but could generate physically unreasonable predictions when extrapolating into less represented high-quality-flow regions. He addressed this by adjusting the physics-based regularization according to how closely different empirical correlations agreed for individual samples, and by adding a penalty when predictions exceeded a known physical upper bound.
The resulting improvement was substantial. On a fixed testing dataset, the model achieved an R² value of 0.986, a statistical measure indicating how closely predictions matched observed data. Indaana compared its performance with widely used correlations, which he said produced errors an order of magnitude larger.
The project originated during a class on fusion energy that Indaana took through the Columbia University Science Honors Program. He asked why reactors that are already well characterized still need to operate conservatively, and the answer led him to critical heat flux and the uncertainty surrounding its prediction across different operating regimes. He subsequently decided to investigate whether machine learning could improve the situation while remaining constrained by established physical knowledge.
Indaana completed the project over roughly eight months, working independently through literature review, data collection, feature engineering, model development, statistical evaluation, and paper writing. He taught himself much of the thermohydraulics, physics-informed machine learning, and statistics required for the work, and carried out model development and experiments on his own computer, using cloud resources such as Kaggle for training. He also used statistical methods including bootstrap resampling and paired nonparametric tests rather than relying solely on a reduction in prediction error.
His interests extend beyond conventional nuclear reactors. Indaana told the Davidson Institute that he is interested in the intersection of materials science, quantum physics, computing and nuclear fusion, and hopes to study physics in college. His project also explores how physics-informed machine learning can be applied to engineering problems where knowledge is often expressed through empirical correlations and physical bounds rather than complete governing equations.
The Davidson Fellows Scholarship, administered by the Davidson Institute, recognizes young scholars for significant contributions to science, technology, engineering, and mathematics. The $50,000 award places Indaana among a select group of students honored for work that demonstrates both technical sophistication and practical relevance. While the achievement does not mean that nuclear reactors can simply be operated closer to the heat limit — reactor safety requirements remain stringent — more accurate prediction could help engineers understand the boundary more precisely and potentially reduce some of the conservatism built into reactor design and operation.
