Pith. sign in

REVIEW 3 major objections 5 minor 100 references

MTP-MD-driven finite-element modeling predicts that adding 2–5 mol% LaF3 or UF4 to FLiBe lowers heat-transfer efficiency by 8–11%, with UF4 having the stronger effect.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

A multiscale MTP-MD + finite-element pipeline predicts an 8–11% heat-transfer reduction for FLiBe with LaF3/UF4 additives, but the quantitative result depends on unvalidated, biased atomistic inputs.

T0 review reviewed 2026-08-01 challenge →

load-bearing objection A genuine end-to-end MTP-MD→FE pipeline with an unusually candid bias audit, but the headline 8–11% ternary prediction rests on an untested assumption that the MD biases cancel across compositions. the 3 major comments →

arxiv 2607.25803 v1 pith:4FPPXSLF submitted 2026-07-28 cond-mat.mtrl-sci

Integrating moment tensor potentials with finite-element modeling for heat transfer prediction in FLiBe-based molten salt systems

Mikhail Polovinkin , Ksenia Abramova , Oksana Rahmanova , Farit Valiev , Andrey Isakov , Andrey Goryachikh , Alexander Galashev , Yurii Zaikov
show 3 more authors
Dmitrii Maksimov Alexander Shapeev Nikita Rybin
This is my paper
classification cond-mat.mtrl-sci
keywords Molten saltsFLiBeMoment Tensor PotentialMolecular dynamicsFinite-element modelingHeat-transfer coefficientTransport propertiesMolten salt reactors
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that a fully computational pipeline — from machine-learned atomistic potentials through molecular dynamics to a finite-element model of an experimental heat exchanger — can predict how adding lanthanum fluoride or uranium fluoride changes the heat-transfer performance of the molten salt coolant FLiBe. The pipeline is validated against experiments for pure FLiBe: the finite-element model with literature transport properties reproduces measured heat transfer within 10% in laminar flow and 18% in transitional/turbulent flow, while using atomistic-derived properties overestimates the heat-transfer coefficient by a systematic 25–28%. Applied to FLiBe with 0–5 mol% LaF3 or UF4, the same model predicts a mean reduction in heat-transfer efficiency of 8–11% relative to pure FLiBe, with UF4 the stronger dopant. A sympathetic reader would care because this is the first reported end-to-end computational screening capability for molten-salt reactor coolant formulations, potentially reducing the need for difficult high-temperature experiments.

Core claim

The central claim is that doping FLiBe with UF4 or LaF3 degrades its heat-transfer performance in a predictable, computable way: 2–5 mol% of either additive lowers the heat-transfer coefficient by roughly 8–11%, and UF4 consistently has the stronger effect. The supporting discovery is that the MTP-MD-to-FE pipeline is quantitatively sound: with reference literature properties it reproduces the experimental thermal loop to within 10% (laminar) and 18% (transitional/turbulent), and the systematic 25–28% overestimation seen with MTP-derived properties is consistent with the known biases in predicted thermal conductivity (+20–30%) and viscosity (−20–25%). The authors state that the qualitative o

What carries the argument

The load-bearing machinery is the atomistic-to-continuum workflow: Moment Tensor Potentials (MTPs), which are machine-learned interatomic potentials fitted to density-functional-theory energies, forces, and stresses with active learning, are used in molecular dynamics to compute density, viscosity, thermal conductivity, and heat capacity. These properties are expressed as temperature- and composition-dependent analytical fits and mapped into a three-dimensional finite-element model of a Joule-heated pipe loop that solves coupled electric-current, Navier–Stokes, and energy equations without relying on empirical Nusselt correlations. The validation hinges on running the FE model twice for pure

Load-bearing premise

The 8–11% prediction holds only if the MTP's systematic errors on viscosity and thermal conductivity (about −20–25% and +20–30% for pure FLiBe) stay roughly the same when LaF3 or UF4 is added; the paper validates neither ternary transport property experimentally.

What would settle it

Measure the viscosity and thermal conductivity of FLiBe with 2–5 mol% UF4 and LaF3 at 900–1200 K. If the additive-induced changes disagree with the MTP-MD predictions — for example, if thermal conductivity does not drop as predicted, or viscosity rises less than predicted — then the predicted 8–11% heat-transfer reduction and possibly the ranking would not hold. Alternatively, run the finite-element model using experimental ternary transport properties and compare against the FE-MD predictions.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If correct, molten-salt reactor coolant formulations can be screened purely computationally: candidate additives can be ranked by their effect on heat-transfer coefficient before building experimental loops.
  • The predicted 8–11% heat-transfer penalty means that for a fixed pumping power and geometry, FLiBe–UF4 and FLiBe–LaF3 coolants require either larger temperature differences or larger heat exchangers to maintain the same thermal output.
  • The uniform 25–28% offset in the FE-MD configuration suggests the pipeline can be used with a systematic-bias correction for relative comparisons, not just qualitative ordering.
  • The stronger degradation caused by UF4 relative to LaF3 is tied to the atomistic structure: U–F complexes are more stable and disrupt the Be–F network more severely, which provides a physics-based explanation for the ranking.
  • The 74–26 LiF–BeF2 composition is characterized at the atomistic level, and its thermodynamic and transport data can be used as inputs for future finite-element validation of that composition.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The paper implicitly assumes the MTP biases on viscosity and thermal conductivity (−20–25% and +20–30%) are roughly composition-independent, so they cancel when comparing ternary to pure FLiBe; this assumption is untested, because no experimental viscosity or thermal-conductivity data exist for these ternary melts.
  • The framework could be extended to predict pressure drop and pumping-power changes, since the same viscosity increase that reduces heat transfer would also raise friction losses in the loop.
  • The observed deviation from standard correlations in the transitional regime (Re 2300–5000) suggests part of the model error is due to the turbulence closure rather than the atomistic potentials; refining the turbulence model would tighten the validation.
  • The same pipeline could be applied to other candidate additives (e.g., ThF4, CeF3) or used to optimize additive concentration within solubility limits, potentially mapping a composition-performance trade-off space.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper presents an integrated multiscale framework that couples actively trained Moment Tensor Potentials (MTPs) with molecular dynamics (MD) to compute density, viscosity, thermal conductivity, and heat capacity for pure FLiBe and FLiBe–LaF3/UF4 melts, and feeds these temperature- and composition-dependent properties into a 3D finite-element (FE) model of an experimental thermal loop. For pure FLiBe, the FE model using literature properties reproduces measured heat-transfer coefficients within 10% (laminar) and 18% (transitional/turbulent), while the same model using MTP-MD properties overestimates the heat-transfer coefficient by 25–28%, an offset the authors attribute to the MTP-MD biases on thermal conductivity (+20–30%) and viscosity (−20–25%). For ternary systems, the MTP-MD-driven FE model predicts an 8–11% reduction in heat-transfer efficiency relative to pure FLiBe, with UF4 having the larger effect. The paper explicitly states that the qualitative ordering is more robust than the absolute 8–11% value.

Significance. If the central quantitative claim (8–11% heat-transfer reduction for FLiBe–LaF3/UF4) were established with appropriate uncertainty, this would provide a useful physics-based screening capability for MSR coolant formulations. The paper has tangible strengths: it quantifies MTP-MD systematic biases against experiment for pure FLiBe; it gives explicit fitting forms and coefficients for density, viscosity, and thermal conductivity; and it honestly distinguishes the FE-ref validation from the FE-MD offset. The FE-ref validation is a credible test of the CFD model in a molten-salt loop, provided the circularity noted below is addressed. However, the 8–11% ternary prediction rests on a plausible but unvalidated composition-transferability assumption, and the effect size is smaller than the validation error. As it stands, the paper is a useful integrated-framework demonstration whose quantitative ternary conclusion needs either direct validation or a clear downgrade to a qualitative prediction.

major comments (3)
  1. [Sect. 2.3 (Eq. 7), Appendix B] The validation is partly circular. The experimental data reduction uses the same reference Cp, rho, mu, and k (Table B.7) that FE-ref uses in the simulation. The inlet–outlet enthalpy balance (Eq. 7) and the Nu/Re/Pr definitions (Eqs. 5a–5c) inherit these properties, so the reported 10%/18% agreement partly reflects consistency between the CFD solver and the data-reduction formula rather than an independent test of the physical property model. This weakens the statement that the end-to-end pipeline is 'validated' for pure FLiBe.
  2. [Sect. 3.3, Tables 4–6] The 8–11% reduction is a relative prediction that assumes the MTP-MD biases (eta −20–25%, kappa +20–30%), quantified only for pure FLiBe, are approximately composition-independent. No experimental viscosity or thermal-conductivity data for FLiBe–LaF3/UF4 are provided, and the kappa decrease with additives—a key driver of the predicted reduction—is inferred solely from MTP-MD with qualitative literature support. Table 1 reports force RMSEs of 74–93 meV/A for the ternary systems, notably larger than typical MTP accuracy, which gives additional reason to question whether the composition trend in kappa is quantitatively reliable. A shift of the bias with composition could easily change the magnitude or even the ordering of the 8–11% effect.
  3. [Sect. 3.3, Fig. 11] No uncertainty is propagated from the MD transport properties to the FE heat-transfer predictions. The MD standard errors mentioned in Section 2.1.2 are not carried through Eqs. 8–10 into the FE outputs, and the 8–11% figure is presented without confidence bounds. Since the effect size is of the same order as the FE-ref validation error (10–18%) and much smaller than the 25–28% FE-MD offset, the relative claim cannot be quantitatively assessed without such propagation.
minor comments (5)
  1. [Fig. 4 caption] Typo: 'experimantal' should be 'experimental'.
  2. [Highlights, Abstract] The highlight line 'FLiBe-LaF3/UF4 4 mol %' is unclear; the abstract reports 0–5 mol% window. The intended composition should be stated consistently.
  3. [Eq. (12)] The attribution of Eq. (12) to Hausen appears questionable; the commonly cited Hausen turbulent correction includes a (D/L)^(2/3) factor that is absent here. Please verify the source and notation.
  4. [Appendix C] Typo: 'root mean sqaure error' should be 'root mean square error'.
  5. [Data Availability] The statement that MTPs and FE files are 'available from the corresponding author upon reasonable request' is weak for a computational methods paper. Depositing the trained potentials and FE setup in a permanent repository would substantially improve reproducibility.

Circularity Check

1 steps flagged

Central 8-11% ternary prediction is model-derived and not fitted to experiment; the main partial circularity is that the FE-ref 'experimental validation' shares the same reference transport properties used to reduce the experimental heat-transfer data.

specific steps
  1. other [Section 2.3, Eq. (7); Appendix B, Table B.7]
    "The heat transfer parameters were determined based on the results of thermal rig tests using the reference temperature-dependent properties of the FLiBe base coolant. These parameters were subsequently used in the development of the finite-element validation model."

    Experimental heat-transfer parameters are reduced with the same reference rho, mu, Cp, and k (Appendix B/Table B.7) that FE-ref then takes as inputs; Eq. (7) even cross-checks the flow rate with the same Cp and rho. Both sides of the 'validation' therefore share the same transport-property closure, so the 10% (laminar) and 18% (transition/turbulent) agreement partly verifies that the FE discretization reproduces the input-property set, rather than independently validating the MTP-MD properties that are the novel part of the pipeline. The agreement is not a full identity because FE-ref still solves the coupled PDEs, so this is partial circularity only.

full rationale

No full circularity was found in the central 8-11% prediction: the FE-MD result is obtained by solving the FE transport equations with MTP-MD-derived property tables, and the target quantities (alpha_loc, Nu-Re) are not among the fitted property coefficients. The paper itself disclaims the 8-11% as an experimental fact and states that the qualitative ordering is the more robust result, which is consistent with a non-circular derivation. The main partial circularity is the FE-ref validation: experimental heat-transfer parameters are reduced using the same reference properties that FE-ref later consumes, making the 10-18% agreement partly a consistency check between two calculations sharing closure properties. This weakens the wording 'validating the end-to-end pipeline,' but it does not make the ternary prediction circular. The self-citations (Rybin 2024, Polovinkin 2026) support methodological choices such as D3 correction, MTP level 16, and the 5 A cutoff; they are not the load-bearing evidence for the composition trend. The lack of direct ternary transport validation and the 74-26 FE validation gap are robustness limitations, not circular steps.

Axiom & Free-Parameter Ledger

5 free parameters · 7 axioms · 0 invented entities

The headline 8–11% prediction is the end of a long chain of fitted and assumed quantities: 608 MTP weights, an unreported Hubbard U, property correlations fitted to MTP-MD output, and an assumed 5 Å screening length. The most fragile link is the composition-independence of the 20–30% property biases; since the predicted effect (8–11%) is smaller than the known biases, this assumption is load-bearing and untested.

free parameters (5)
  • MTP parameters (level 16) = 608 parameters
    Fitted to DFT PBE+U-D3 energies/forces (Sect. 2.1.1); training-set force RMSE 74–93 meV/Å, no test-set errors.
  • Hubbard U for f-electron systems = not reported
    DFT+U applied to FLiBe-UF4 (Sect. 2.1.1) but U value omitted; affects U-F bonding and transport.
  • Density fit coefficients (Table 4) = ρ0 2.33–2.38, α 8.17–15.24, b −4.44e-4..−4.15e-4, β −3.58e-3..−1.16e-3
    Eq. 8 fitted to MTP-MD densities; used to set FE density field.
  • Viscosity fit coefficients (Table 5) = η0 6.99e-5..1.34e-4, α −1.82e-4..4.85e-4, E0 3118–4007 K, β 1177–16062 K
    Eq. 9 fitted to MTP-MD viscosities; carries the main −20–25% bias.
  • Thermal conductivity fit coefficients (Table 6) = κ0 1.42–1.67, α −2.22..−4.16, β −3.7e-4..−1.8e-4
    Eq. 10 fitted to MTP-MD conductivities; carries the +20–30% bias.
axioms (7)
  • domain assumption MTP with 5 Å cutoff and DFT-D3 correction adequately describes FLiBe and FLiBe-LaF3/UF4 interactions; electrostatics beyond 5 Å are negligible due to screening.
    Invoked in Sect. 2.1.1 to justify omitting explicit long-range electrostatics from the MTP; errors in density up to 10% without D3 were previously reported by the same group (Rybin 2024).
  • domain assumption PBE+U DFT with a plane-wave cutoff of 600 eV and Gamma-point sampling provides sufficiently accurate energies/forces for training.
    Sect. 2.1.1; no convergence tests vs k-points or U value; the Hubbard U is not specified.
  • domain assumption Active learning (D-optimality + MaxVol) guarantees robust extrapolation in production MD.
    Sect. 2.1.1; no extrapolation grades reported for the final production trajectories.
  • domain assumption Green-Kubo viscosity (25 ps integration) and Müller-Plathe conductivity (250 ps runs, 3136–3392 atoms) are converged.
    Sect. 2.1.2 and Appendix D; convergence shown for viscosity at one composition.
  • domain assumption The standard k-ε model with wall functions (30<y+<300) captures heat transfer in transitional/turbulent regimes.
    Appendix A; the paper itself attributes the FE-ref deviation (up to 18%) primarily to this turbulence model.
  • domain assumption The literature reference properties (Table B.7) are the correct benchmark for FLiBe.
    Sect. 2.3 and Table B.7; experimental data reduction uses the same property set.
  • ad hoc to paper MTP-MD systematic biases (μ: −20–25%, κ: +20–30%) are composition-independent.
    Implicit in Sect. 3.3 when the 8–11% relative reduction is computed without bias correction; no evidence provided.

reviewed 2026-08-01 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Integrating moment tensor potentials with finite-element modeling for heat transfer prediction in FLiBe-based molten salt systems." pith.science (2026). https://pith.science/paper/4FPPXSLF

@misc{pith2026260725803,
  author       = {Pith},
  title        = {Pith review of: Integrating moment tensor potentials with finite-element modeling for heat transfer prediction in FLiBe-based molten salt systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4FPPXSLF}},
  note         = {Machine review of arXiv:2607.25803}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Molten fluoride salts are promising heat-transfer media for advanced molten salt reactors (MSRs), where reliable thermophysical property determination is critical for component design and safety. We present an integrated multiscale framework that couples machine-learning-driven atomistic simulations with finite-element (FE) modeling to predict the heat-transfer performance of FLiBe-based salts in a linear heat exchanger. At the atomistic scale, Moment Tensor Potentials (MTPs), actively trained on ab initio data, are developed for pure FLiBe (66-34 and 74-26 LiF-BeF2 mol%), FLiBe-LaF3, and FLiBe-UF4. These potentials are used in molecular dynamics simulations to obtain temperature- and composition-dependent transport properties (density, viscosity, thermal conductivity, and isobaric heat capacity), which are mapped as inputs to a three-dimensional FE model of the experimental thermal loop. The FE model with literature transport properties reproduces the experimental heat-transfer behavior of pure FLiBe to within 10% in the laminar regime and 18% in the transitional and turbulent regimes, validating the end-to-end pipeline for this composition. The same model with MTP-MD-derived transport properties systematically overestimates the heat-transfer coefficient by 25-28%, an offset consistent with the MTP-MD biases on thermal conductivity and viscosity. Applied to the ternary systems FLiBe-LaF3 and FLiBe-UF4 over 0-5 mol%, the MTP-MD-driven FE model predicts a mean reduction in heat-transfer efficiency of 8-11% relative to pure FLiBe, with UF4 exhibiting the strongest effect. The qualitative ordering of the three systems is the more robust result; the absolute value of the 8-11% figure is contingent on the MTP accuracy. The framework is complementary to high-temperature experiments and provides a physics-based pathway for the rapid screening of MSR coolant formulations.

Figures

Figures reproduced from arXiv: 2607.25803 by Alexander Galashev, Alexander Shapeev, Andrey Goryachikh, Andrey Isakov, Dmitrii Maksimov, Farit Valiev, Ksenia Abramova, Mikhail Polovinkin, Nikita Rybin, Oksana Rahmanova, Yurii Zaikov.

Figure 1
Figure 1. Figure 1: The schematic diagrams of (a) the experimental working section: 1 [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Distribution of clusters of UF in FLiBe-UF [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 2
Figure 2. Figure 2: Radial distribution functions in FLiBe-UF [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Temperature dependence of the density of molten salts, obtained with MTP-MD. (a) FLiBe (66 mol % – 34 mol %)-UF [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Temperature dependence of the viscosity of molten salts, obtained with MTP-MD. (a) FLiBe (66 mol % – 34 mol %)-UF [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Temperature dependence of diffusion coefficients of species in FLiBe (66 mol % – 34 mol %) with 5 mol % additions of (a) UF4 or (b) LaF3. Thermal conductivity is described using a linear equation: κ(T, m) = (κ0 + α · m) + β · T, (10) where T is the temperature, m is the molar fraction of the ad￾ditive (LaF3 or UF4), κ0 is the empirical zero-temperature in￾tercept (a fitting parameter with no physical meani… view at source ↗
Figure 7
Figure 7. Figure 7: Temperature dependence of the thermal conductivity of molten salts, obtained with MTP-MD. (a) FLiBe (66 mol % – 34 mol %)-UF [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Wall temperature along the pipe length for regimes R2 [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
Figure 11
Figure 11. Figure 11: Local heat transfer coefficient αloc along the pipe length for regimes R4 (3300<Re<6000) and R5 (5000<Re<8000), operating conditions provided in Tab. 3, computed for various coolants. Finite-element model predictions using MTP-MD-derived thermophysical properties are shown by solid lines for pure FLiBe 66 mol % – 34 mol %, FLiBe - 2 mol % LaF3, and FLiBe - 2 mol % UF4. Re < 2300 (laminar); the Hausen corr… view at source ↗
Figure 12
Figure 12. Figure 12: Heat transfer functions for the coolants: FLiBe and FLiBe con [PITH_FULL_IMAGE:figures/full_fig_p016_12.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

100 extracted references · 1 linked inside Pith

  1. [1]

    Russian Journal of Non-Ferrous Metals , volume=

    The viscosity of molten salts based on the LiF--BeF2 system , author=. Russian Journal of Non-Ferrous Metals , volume=. 2022 , publisher=

    2022

  2. [2]

    Journal of Fluorine Chemistry , volume=

    Thermodynamic properties and phase diagrams of fluoride salts for nuclear applications , author=. Journal of Fluorine Chemistry , volume=. 2009 , publisher=

    2009

  3. [3]

    Journal of Molecular Liquids , volume=

    Density and heat capacity of some molten mixtures in system LiF-BeF2-UF4 , author=. Journal of Molecular Liquids , volume=. 2021 , publisher=

    2021

  4. [4]

    Russian Metallurgy (Metally) , volume=

    Thermophysical properties of several molten mixtures of the LiF--BeF2--UF4 system , author=. Russian Metallurgy (Metally) , volume=. 2023 , publisher=

    2023

  5. [5]

    Fusion Engineering and Design , volume=

    Pre-conceptual design of a liquid metal to molten salt heat exchanger with a leak detection system , author=. Fusion Engineering and Design , volume=. 2022 , publisher=

    2022

  6. [6]

    International Journal of Heat and Mass Transfer , volume=

    Numerical and experimental study on heat transfer and flow features of representative molten salts for energy applications in turbulent tube flow , author=. International Journal of Heat and Mass Transfer , volume=. 2019 , publisher=

    2019

  7. [7]

    Thermal Science , volume=

    Heat transfer characteristics of molten salt flowing in steam generator , author=. Thermal Science , volume=

  8. [8]

    Annals of Nuclear Energy , volume=

    Fluoride salt coolant properties for nuclear reactor applications: A review , author=. Annals of Nuclear Energy , volume=. 2017 , publisher=

    2017

  9. [9]

    Nuclear Engineering and design , volume=

    Molten salt reactors: A new beginning for an old idea , author=. Nuclear Engineering and design , volume=. 2010 , publisher=

    2010

  10. [10]

    Annals of Nuclear Energy , volume=

    Molten salt actinide recycler and transforming system without and with Th--U support: Fuel cycle flexibility and key material properties , author=. Annals of Nuclear Energy , volume=. 2014 , publisher=

    2014

  11. [11]

    Nuclear Engineering and Design , volume=

    Consequences of molten salt solidification in a natural circulation flow visualization loop due to heater failure , author=. Nuclear Engineering and Design , volume=. 2024 , publisher=

    2024

  12. [12]

    Annals of Nuclear Energy , volume=

    Flow and heat transfer characteristics in molten salt reactor core channels under the coupled influence of wall heat flux and volumetric internal heat generation , author=. Annals of Nuclear Energy , volume=. 2026 , publisher=

    2026

  13. [13]

    Applied thermal engineering , volume=

    Heat transfer and pressure drop characteristics of molten fluoride salt in circular pipe , author=. Applied thermal engineering , volume=. 2013 , publisher=

    2013

  14. [14]

    Applied Thermal Engineering , volume=

    CFD modeling of natural circulation in LiCl-KCl molten salt closed loop , author=. Applied Thermal Engineering , volume=. 2026 , publisher=

    2026

  15. [15]

    Experimental thermal and fluid science , volume=

    Convective heat transfer of molten salt in circular tube with nonuniform heat flux , author=. Experimental thermal and fluid science , volume=. 2014 , publisher=

    2014

  16. [16]

    Materials , volume=

    Dynamic viscosity of the NaF-KF-NdF3 molten system , author=. Materials , volume=. 2022 , publisher=

    2022

  17. [17]

    Chemical Engineering and Processing: Process Intensification , volume=

    Molten salts database for energy applications , author=. Chemical Engineering and Processing: Process Intensification , volume=. 2013 , publisher=

    2013

  18. [18]

    Nuclear Engineering and Design , volume=

    Radiative heat transfer in FLiBe molten salt participating medium in a vertical heated tube under forced and mixed convection laminar flows , author=. Nuclear Engineering and Design , volume=. 2020 , publisher=

    2020

  19. [19]

    ACS applied materials & interfaces , volume=

    Modeling LiF and FLiBe molten salts with robust neural network interatomic potential , author=. ACS applied materials & interfaces , volume=. 2021 , publisher=

    2021

  20. [20]

    Physics of Fluids , volume=

    Heat transport and ionic dynamics in pure and mixed molten salts , author=. Physics of Fluids , volume=. 2025 , publisher=

    2025

  21. [21]

    Journal of Molecular Liquids , volume=

    Thermophysical properties of FLiBe using moment tensor potentials , author=. Journal of Molecular Liquids , volume=. 2022 , publisher=

    2022

  22. [22]

    Physical Chemistry Chemical Physics , volume=

    Compositional transferability of deep potential in molten LiF--BeF _2 and LaF _3 mixtures: prediction of density, viscosity, and local structure , author=. Physical Chemistry Chemical Physics , volume=. 2024 , publisher=

    2024

  23. [23]

    The Journal of Physical Chemistry B , volume=

    Comparative studies of the structural and transport properties of molten salt FLiNaK using the machine-learned neural network and reparametrized classical forcefields , author=. The Journal of Physical Chemistry B , volume=. 2021 , publisher=

    2021

  24. [24]

    Network formation in LiF- BeF2 , author=

    A first-principles description of liquid BeF2 and its mixtures with LiF: 2. Network formation in LiF- BeF2 , author=. The Journal of Physical Chemistry B , volume=. 2006 , publisher=

    2006

  25. [25]

    Progress in Nuclear Energy , volume=

    Thermal neutron scattering cross section of liquid FLiBe , author=. Progress in Nuclear Energy , volume=. 2017 , publisher=

    2017

  26. [26]

    Journal of Molecular Liquids , pages=

    Addressing cost-effective pair models for simulation of FLiBe: performance and limitations , author=. Journal of Molecular Liquids , pages=. 2025 , publisher=

    2025

  27. [27]

    Neue Gleichungen fur die Warmeubertragung bei freier oder erzwungener Stromung , author=. Allg. Waermetech , volume=

  28. [28]

    Teploenergetika , volume=

    To the question of heat transfer in turbulent pipe flow of liquids in tubes , author=. Teploenergetika , volume=

  29. [29]

    Industrial & Engineering Chemistry , volume=

    Heat transfer and pressure drop of liquids in tubes , author=. Industrial & Engineering Chemistry , volume=. 1936 , publisher=

    1936

  30. [30]

    Teploenergetika , volume=

    Experimental study of heat transfer of molten salt in a vertical circular channel under transient and turbulent flow conditions , author=. Teploenergetika , volume=

  31. [31]

    1995 , publisher=

    Navier--Stokes equations and nonlinear functional analysis , author=. 1995 , publisher=

    1995

  32. [32]

    Fundamentals of Thermal and Nuclear Power Generation , pages=

    Fundamentals for power engineering , author=. Fundamentals of Thermal and Nuclear Power Generation , pages=. 2021 , publisher=

    2021

  33. [33]

    Boca Raton, FL , year=

    Process Heat Transfer CRC Press , author=. Boca Raton, FL , year=

  34. [34]

    Journal of Physics: Conference Series , volume=

    Adjustment of the k- SST turbulence model for prediction of airfoil characteristics near stall , author=. Journal of Physics: Conference Series , volume=. 2016 , organization=

    2016

  35. [35]

    Procedia Computer Science , volume=

    Algebraic and k-ɛ Turbulence Model Comparison in the Problem of the Boundary Layer on the Walls of a Rotating Axisymmetric Diffuser , author=. Procedia Computer Science , volume=. 2015 , publisher=

    2015

  36. [36]

    Energy Procedia , volume=

    A CFD Comsol model for simulating complex urban flow , author=. Energy Procedia , volume=. 2017 , publisher=

    2017

  37. [37]

    Procedia Engineering , volume=

    Experimental validation of computer fluid dynamics simulation aimed on pressure distribution on gable roof of low-rise building , author=. Procedia Engineering , volume=. 2017 , publisher=

    2017

  38. [38]

    Proceedings of the international multiconference of engineers and computer scientists , volume=

    Wall Y strategy for dealing with wall-bounded turbulent flows , author=. Proceedings of the international multiconference of engineers and computer scientists , volume=

  39. [39]

    2012 , publisher=

    Dimensionless physical quantities in science and engineering , author=. 2012 , publisher=

    2012

  40. [40]

    Das aehnlichkeitsgesetz bei reibungsvorg

    Blasius, Heinrich , booktitle=. Das aehnlichkeitsgesetz bei reibungsvorg. 1913 , publisher=

    1913

  41. [41]

    Advances in heat transfer , volume=

    Heat transfer and friction in turbulent pipe flow with variable physical properties , author=. Advances in heat transfer , volume=. 1970 , publisher=

    1970

  42. [42]

    1998 , publisher=

    Verification and validation in computational science and engineering , author=. 1998 , publisher=

    1998

  43. [43]

    TRANSACTIONS of NNSTU n.a

    Minimum basis of tasks of the validation of methods of flow modeling with low Prandt numbers , author=. TRANSACTIONS of NNSTU n.a. R.Е. ALEKSEEV , volume=. 2018 (in Russian) , publisher=

    2018

  44. [44]

    Advances in neural information processing systems , volume=

    MACE: Higher order equivariant message passing neural networks for fast and accurate force fields , author=. Advances in neural information processing systems , volume=

  45. [45]

    Nature Machine Intelligence , volume=

    The design space of E (3)-equivariant atom-centred interatomic potentials , author=. Nature Machine Intelligence , volume=. 2025 , publisher=

    2025

  46. [46]

    Nature Communications , volume=

    SuperSalt: equivariant neural network force fields for multicomponent molten salts system , author=. Nature Communications , volume=. 2025 , publisher=

    2025

  47. [47]

    The Journal of Physical Chemistry B , volume=

    Physicochemical Properties and Structure of FLiBeTh Salts: Insights from Machine Learning Accelerated Molecular Dynamics Simulations , author=. The Journal of Physical Chemistry B , volume=. 2025 , publisher=

    2025

  48. [48]

    2017 , issn =

    Active learning of linearly parametrized interatomic potentials , journal =. 2017 , issn =

    2017

  49. [49]

    Physical review B , volume=

    Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set , author=. Physical review B , volume=. 1996 , publisher=

    1996

  50. [50]

    Physical review letters , volume=

    Generalized gradient approximation made simple , author=. Physical review letters , volume=. 1996 , publisher=

    1996

  51. [51]

    Physica Scripta , volume=

    Pseudopotentials plane waves--projector augmented waves: A primer , author=. Physica Scripta , volume=

  52. [52]

    Chemical reviews , volume=

    Dispersion-corrected mean-field electronic structure methods , author=. Chemical reviews , volume=. 2016 , publisher=

    2016

  53. [53]

    The Journal of chemical physics , volume=

    Hybrid functionals based on a screened Coulomb potential , author=. The Journal of chemical physics , volume=. 2003 , publisher=

    2003

  54. [54]

    Computational Materials Science , volume=

    Hubbard U parameters for transition metals: Correction of the Bloch state and atomic values , author=. Computational Materials Science , volume=. 2025 , publisher=

    2025

  55. [55]

    arXiv preprint arXiv:2405.04967 , year=

    Mattersim: A deep learning atomistic model across elements, temperatures and pressures , author=. arXiv preprint arXiv:2405.04967 , year=

    Pith/arXiv arXiv

  56. [56]

    The Journal of Physical Chemistry C , volume=

    Correction to Density Functional Theory Calculations and Machine Learning Interatomic Potentials for Molten Salts to Achieve Experimental Accuracy , author=. The Journal of Physical Chemistry C , volume=. 2024 , publisher=

    2024

  57. [57]

    Communications Chemistry , volume=

    Computational methods to simulate molten salt thermophysical properties , author=. Communications Chemistry , volume=. 2022 , publisher=

    2022

  58. [58]

    1995 , author =

    Fast Parallel Algorithms for Short-Range Molecular Dynamics , journal =. 1995 , author =

    1995

  59. [59]

    Thompson and H

    Aidan P. Thompson and H. Metin Aktulga and Richard Berger and Dan S. Bolintineanu and W. Michael Brown and Paul S. Crozier and Pieter J. LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales , journal =. 2022 , issn =

    2022

  60. [60]

    Multiscale Modeling & Simulation , volume=

    Moment tensor potentials: A class of systematically improvable interatomic potentials , author=. Multiscale Modeling & Simulation , volume=. 2016 , publisher=

    2016

  61. [61]

    and Gubaev, Konstantin and Podryabinkin, Evgeny V

    Novikov, Ivan S. and Gubaev, Konstantin and Podryabinkin, Evgeny V. and Shapeev, Alexander V. , title =. 2020 , publisher =

    2020

  62. [62]

    The Journal of Chemical Physics , volume =

    Podryabinkin, Evgeny and Garifullin, Kamil and Shapeev, Alexander and Novikov, Ivan , title =. The Journal of Chemical Physics , volume =. 2023 , issn =

    2023

  63. [63]

    From ultrasoft pseudopotentials to the projector augmented-wave method , author =. Phys. Rev. B , volume =. 1999 , publisher =

    1999

  64. [64]

    Generalized Gradient Approximation Made Simple , author =. Phys. Rev. Lett. , volume =. 1996 , publisher =

    1996

  65. [65]

    The Journal of chemical physics , volume=

    A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu , author=. The Journal of chemical physics , volume=. 2010 , publisher=

    2010

  66. [66]

    Journal of computational chemistry , volume=

    Effect of the damping function in dispersion corrected density functional theory , author=. Journal of computational chemistry , volume=. 2011 , publisher=

    2011

  67. [67]

    2024 , issn =

    Thermophysical properties of Molten FLiNaK: A moment tensor potential approach , journal =. 2024 , issn =

    2024

  68. [68]

    Journal of Physics: Condensed Matter , volume=

    Electronic correlation effects in transition-metal sulfides , author=. Journal of Physics: Condensed Matter , volume=

  69. [69]

    Physical Review B , volume=

    Implementation of the projector augmented-wave LDA+ U method: Application to the electronic structure of NiO , author=. Physical Review B , volume=. 2000 , publisher=

    2000

  70. [70]

    Molecular Physics , volume =

    Nose, Shuichi , title =. Molecular Physics , volume =. 2002 , publisher =

    2002

  71. [71]

    Journal of the Physical Society of Japan , volume =

    Kubo ,Ryogo , title =. Journal of the Physical Society of Japan , volume =

  72. [72]

    , title =

    Green, Melville S. , title =. The Journal of Chemical Physics , volume =

  73. [73]

    The Journal of Chemical Physics , volume =

    Hess, Berk , title = ". The Journal of Chemical Physics , volume =

  74. [74]

    2017 , publisher=

    Computer Simulation of Liquids , author=. 2017 , publisher=

    2017

  75. [75]

    Best Practices for Computing Transport Properties 1

    Maginn, Edward and Messerly, Richard and Carlson, Daniel and Roe, Daniel and Elliott, Jarrell , year =. Best Practices for Computing Transport Properties 1. Self-Diffusivity and Viscosity from Equilibrium Molecular Dynamics [Article v1.0] , volume =

  76. [76]

    The Journal of Chemical Physics , volume =

    Müller-Plathe, Florian , title =. The Journal of Chemical Physics , volume =

  77. [77]

    SN Applied Sciences , volume=

    Heat loss along the pipeline and its control measures , author=. SN Applied Sciences , volume=. 2023 , publisher=

    2023

  78. [78]

    Nuclear Engineering and Technology , volume=

    Measurement of local wall temperature and heat flux using the two-thermocouple method for a heat transfer tube , author=. Nuclear Engineering and Technology , volume=. 2019 , publisher=

    2019

  79. [79]

    Journal of Nuclear Materials , volume = 537, pages = 152219, issn =

    The structure of molten FLiNaK , author =. Journal of Nuclear Materials , volume = 537, pages = 152219, issn =

  80. [80]

    Inorganic Chemistry , volume = 48, number = 23, pages =

    Measuring Self-Diffusion Coefficients up to 1500 K: A Powerful Tool to Investigate the Dynamics and the Local Structure of Inorganic Melts , author =. Inorganic Chemistry , volume = 48, number = 23, pages =

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 1, 2026.