About

Dr. Elia Merzari is a professor at the Department of Nuclear Engineering at The Pennsylvania State University, located in the Hallowell Building. His research focuses on high-fidelity multiphysics simulations, thermal-hydraulics, and neutronics, particularly in the context of nuclear industry applications. His work involves advanced computational modeling of complex nuclear systems, contributing to the understanding and optimization of nuclear reactor performance and safety. Dr. Merzari's expertise encompasses the development and application of simulation software to address challenges in nuclear engineering, supporting the advancement of nuclear technology through detailed and accurate modeling efforts.

Research topics

• Computer Science
• Computational science
• Parallel computing
• Physics
• Programming language
• Systems engineering
• Engineering
• Mechanics
• Aerospace engineering
• Distributed computing
• Operating system
• Nuclear physics

Selected publications

• Validation of NekRS pebble bed modeling for low flow scenarios

  Nuclear Engineering and Design · 2026-01-02

  articleOpen access
  PublisherDOI

• Development of MOSCATO: A CFD-Level Electrochemistry and Corrosion Simulator for Molten Salt Systems

  Nuclear Science and Engineering · 2026-04-23

  article
  PublisherDOI

• High-Fidelity Modelling of the Molten Salt Fast Reactor

  ArXiv.org · 2025-07-05

  articleOpen access

  The Molten Salt Fast Reactor (MSFR) is one of the six GEN-IV reactor designs. In the MSFR, the liquid fuel is the coolant, which moves throughout the primary circuit. This complex phenomenology requires multiphysics modeling. In the present paper, a model of the MSFR is developed in the multiphysics code Cardinal, considering neutronic-thermal hydraulic feedback and the transport of delayed neutron precursors (DNPs) and decay heat precursors (DHPs). OpenMC is used to solve neutronic equations, and NekRS is used to solve mass, momentum, energy, DNPs, and DHPs distribution. A RANS k-t turbulence model is used in NekRS. DNPs and DHPs are modeled using a convective-diffusion equation with modified source terms considering radioactive decay. Cardinal results showed a reasonable behavior for temperature, heat source, velocity, DNPs, and DHPs. However, the current limitations in OpenMC do not allow the modification of delayed neutron source locations. Ongoing efforts look to include this feature in future work to introduce DNP feedback in OpenMC.

  PublisherOA PDF

• NEAMS IRP challenge problem 2: Thermal striping of reactor Internals

  Nuclear Engineering and Design · 2025-02-03 · 1 citations

  articleOpen accessSenior author

  • The NEAMS IRP effort has assessed the accuracy and applicability of hybrid RANS models in support of thermal fatigue assessment. • Experimental data from the RCCS and DESTROJER facilities support rigorous assessment of CFD simulations. • High-fidelity LES simulation with NekRS provide upscaling of experimental results for turbulence modeling validation. • The STRUCT hybrid approach demonstrates LES like accuracy on URANS type meshes. • A two-level machine learning approach was demonstrated for reduced order modeling of thermal striping. Oscillatory mixing of non-isothermal liquid coolant streams in advanced reactors can lead to thermal fatigue damage to fuel and reactor components. The NEAMS IRP CP2 has been seeking the development of an accurate, yet computationally affordable, turbulence modeling option for thermal-striping predictions. Efficient mixing of coolant streams in upper internal structures, lower plena, and heat exchangers significantly impacts the design of these systems, as well as their operation, and maintenance. This challenge problem generalizes specific needs related to the TerraPower and General Atomics designs by developing a set of benchmarks to advance and quantify the accuracy of the thermal striping modeling approach. The focus of the activities is to advance and demonstrate a modeling practice capable of accurately representing the performance of the structural reactor components under the influence of thermal striping. Assessment against adiabatic quasi-2D jets striping has demonstrated great promise for the proposed turbulence approaches, with 2 orders of magnitude acceleration from the reference LES solution. More recent efforts have extended the quasi-2D validation to diabatic conditions, leveraging the demonstrated accuracy of the highly resolved LES methods, and with a new set of experimental data for heated parallel round jets. Further upscaling through the use of reduced order models is being evaluated, and a two-step machine learning approach has been demonstrated on thermal striping for 3 parallel jets.

  PublisherDOI

• Large Eddy Simulation of a 67-Pebble Bed Experiment

  Journal of Fluids Engineering · 2025-04-25

  article
  PublisherDOI

• Development and Application of Reduced-Order Models for Thermal-Fluid Dynamics in Molten Salt Reactors

  Nuclear Technology · 2025-09-30

  articleCorresponding
  PublisherDOI

• Temperature Field Reconstruction of Surfaces Heated Through Radiative Heat Transfer Using Convolutional Neural Networks

  ASME Journal of Heat and Mass Transfer · 2025-05-22

  article

  Abstract Microreactors could play a crucial role in decarbonizing our energy portfolio. However, their development and implementation come with specific challenges, particularly regarding cost. Due to their compact size and the harsh operational environment, collecting real-time data on reactor operation can be challenging. Many probe designs are unable to withstand extreme conditions (e.g., temperature, radiation) in the reactor. In this context, using convolutional neural networks (CNNs) can pave the way for developing a nonintrusive approach that relies solely on ex-core sensors. A well-trained physics-informed CNN can reconstruct the distribution of a given physical quantity over a domain using only a few sensors, allowing us to reconstruct the desired field distribution even in a limited space or complex geometries where a large array of sensors is impractical. In this work, we present the initial steps toward developing a real-time tool for monitoring the thermal behavior of nuclear reactor pressure vessels. Based on an experimental setup, a computational model using the Multiphysics Object-Oriented Simulation Environment (moose) framework was built, where the Ray Tracing and Heat Conduction modules were used to evaluate the temperature distribution over a convex metal surface heated through radiative heat transfer. This metal surface represents a section of a heated nuclear reactor vessel wall. The model also accounts for solid mechanics physics through the moose Solid Mechanics module. In situ experimental data, acquired from a Texas A&M facility, were used to validate the computational model. Part of the data generated by the moose model was used to train the convolutional neural network to reconstruct the vessel wall's outer surface temperature. The CNN generalization was then compared against the experimental and computational data.

  PublisherDOI

• CFD simulation of interassembly bypass flow in Sodium Fast Reactors

  Nuclear Engineering and Design · 2025-05-06

  articleOpen accessSenior author

  Interassembly flow in Sodium Fast Reactors (SFRs) represents a bypass flow path exterior to the fuel assembly ducts. Heat transferred across this thin gap is an important component of core radial expansion, where the coupling between thermal-fluids, neutronics, and solid mechanics results in time-dependent duct bowing. These geometry changes can constitute a significant portion of the total reactivity response in transients, but are difficult to model in high-fidelity. Interassembly flow is also an important heat transfer mode during natural convection cooling. To improve our understanding of interassembly flow, this paper provides NekRS Reynolds Averaged Navier–Stokes (RANS) and Large Eddy Simulations (LES) of the interassembly flow in a 19-bundle fast reactor core. Time-averaged LES compares reasonably well with a k - τ RANS model, though RANS is not able to capture a crossflow which occurs at a large change in flow area between the duct–duct gaps and the open peripheral region. We predict velocity distributions and illustrate a multiscale postprocessing system that can be used to generate coarse-mesh closures for subchannel and porous media tools, and provide a dataset with average velocity for comparison with coarse-mesh tools. • LES and RANS simulations are performed of the interassembly flow in a small-size sodium fast reactor. • The flow is laminar in the duct–duct gaps and turbulent in the peripheral region. • Spatial averaging of the CFD simulations provides data for code-to-code comparison.

  PublisherDOI

• A Study of the Spatially Developing Laminar-Turbulent Transition in Rod Bundles

  Nuclear Science and Engineering · 2025-05-28

  articleSenior author
  PublisherDOI

• Toward Thermal Striping Predictions Using Large Eddy Simulation-Proper Orthogonal Decomposition Data in Reduced-Order Models

  Journal of Nuclear Engineering and Radiation Science · 2025-08-21

  article

  Abstract The study of turbulent mixing in nonisothermal coolant streams is crucial for understanding thermal striping, which can lead to thermal fatigue and degradation of internal components in advanced nuclear reactors. Thermal striping is closely linked to the fine 3D scales of turbulence. Historically, large eddy simulation (LES) or direct numerical simulation has been required to resolve these fine scales. Our investigation focuses on the mixing dynamics within the reactor cavity cooling system (RCCS) separate-effects test facility, where we examine the interaction of two parallel plane jets within a confined plenum. To begin, LES simulations were conducted to generate velocity statistics, along with time series data, for comparison with reduced order model (ROM) approaches. Power spectrum density (PSD) analysis of the velocity time series reveals a distinct low-frequency mixing mode, which is indicative of thermal striping. Next, we applied proper orthogonal decomposition (POD) to extract the dominant flow structures from high-fidelity instantaneous velocity snapshots. As a first step, POD modes were used to construct 2D ROMs that attempt to replicate the low-frequency mode associated with thermal striping. Sensitivity studies demonstrated that increasing the number of snapshots and POD modes improves 2D ROM accuracy, while also increasing the computational cost. To address this, we incorporated closure models and found that the constrained optimization ROM performed best across Reynolds numbers ranging from 100 to 10,000. By combining LES data with ROM techniques, we show that this approach offers a promising method for modeling the low-frequency modes linked to thermal striping.

  PublisherDOI

Frequent coauthors

• Paul Fischer

  86 shared

• Yassin A. Hassan

  Texas A&M University

  76 shared

• Dillon Shaver

  Argonne National Laboratory

  40 shared

• Ananias Tomboulides

  39 shared

• Hisashi Ninokata

  34 shared

• Victor Coppo Leite

  Idaho National Laboratory

  32 shared

• A. Obabko

  32 shared

• Adam Kraus

  Pennsylvania State University

  31 shared

Labs

• Elia Merzari LabPI

Awards & honors

• ANS Landis Young Member Engineering Achievement Award
• George Westinghouse Silver Medal from ASME
• R&D100 Award from R&D World Magazine
• ANS Fellow, American Nuclear Society (September 2023)
• NURETH Fellow, American Nuclear Society (August 2023)