Figure 1 - uploaded by Mengnan Li
Content may be subject to copyright.
Source publication
The development and validation of 3D multiphase computational fluid dynamics (M-CFD) model and physics-informed data-driven modeling require data of high-quality and high-resolution. Considering the difficulties in acquiring the corresponding experimental data in prototypic conditions, two-phase boiling simulations by Interface tracking method (ITM...
Contexts in source publication
Context 1
... the estimation of average temperature gradient for individual bubbles, the evaporation and condensation model is coupled with the Bubble Tracking Algorithm (BTA) [31]. BTA introduced a marker/ID field in PHASTA to identify and track each individual bubble in the domain as shown in Figure 1. The nodes within the region of interest are colored by the corresponding bubble ID while the rest of the domain is marked by zero ID value. ...
Context 2
... the estimation of average temperature gradient for individual bubbles, the evaporation and condensation model is coupled with the Bubble Tracking Algorithm (BTA) [31]. BTA introduced a marker/ID field in PHASTA to identify and track each individual bubble in the domain as shown in Figure 1. The nodes within the region of interest are colored by the corresponding bubble ID while the rest of the domain is marked by zero ID value. ...
Similar publications
It is well aligned with the general consensus that the quality of boiling heat transfer is heavily dependent upon bubbles characteristics. Moreover, the accuracy of various numerical models developed to simulate flow boiling on a heated wall heavily depends on empirical data on bubbles dynamic behavior. In this paper, the influence of radial pressu...
Citations
... Here, q is the volume-averaged heat flux, and h fg is the latent heat of evaporation. This algorithm was verified against analytical bubble growth model [29] and experimental data for pool and flow boiling [30]. ...
High-fidelity flow boiling simulations are conducted in a vertical mini channel with offset strip fins (OSF) using R113 as a working fluid. Finite-element code PHASTA coupled with level set method for interface capturing is employed to model multiple sequential bubble nucleation using transient three-dimensional approach. The code performance is validated against experiments for a single nucleation site in a vertical rectangular channel. To assess code performance, a study on the bubble departure from the wall in a mini channel with OSF is carried out first. Contributions from the microlayer are not considered due to low heat flux values applied to the channel (1 kW/m2). The influence of surface characteristics, such as contact angle and liquid superheat on bubble dynamics, is also analyzed as well as the local two-phase heat transfer coefficient. For higher void fractions, two conical nucleation cavities are introduced in the same channel with OSF. Observed bubble characteristics (departure diameter, bubble departure frequency) are evaluated and bubble trajectories are presented and analyzed. The local heat transfer coefficient is then evaluated for each simulation. The results show approximately a 2.5 time increase in the local heat transfer coefficient when the individual bubbles approach the wall. With smaller bubble nucleation diameters, the heat transfer coefficient can increase by up to a factor of two. Thus, the current work demonstrates the flow modeling capability of the boiling phenomena in complex geometry with OSF.
... This heat flux value is later being used to provide volume source term to model evaporation / condensation process. This model has been verified against analytical bubble growth [19] and validated against experimental data [20]. ...
With the growth of computational power, two-phase direct numerical simulations (DNS) have become more affordable. However, to conduct engineering-scale simulations (e.g., flow boiling in a heat exchanger) coarse-grid (CG) modeling is still required which inevitably leads to the increase in prediction errors. At the same time, over the last few years machine-learning (ML) technics have advanced to physics-informed data-driven (DD) modeling and neural operators that allow to learn the dependencies between function spaces. This study presents a ML-based model development workflow that allows to conduct CG two-phase simulations with improved accuracy. Finite-element code PHASTA coupled with level-set interface-tracking method is used to conduct DNS and CG simulations. The underlaying idea is to predict interface location for CG simulations instead of solving the hyperbolic PDE for the level-set function. Two neural networks are needed to be trained on DNS data to predict high-resolution velocity field (velocity reconstruction, step 1) and high-resolution level-set function (PDE surrogate, step 2). Once high-resolution level-set function at the current time step is available, it has to be mapped to the existing coarse grid (CG mapping, step 3). As the level-set solution is overwritten in the CG simulation, the code proceeds to the next time step. The full paper will describe the implementation and step-by-step evaluation of the proposed workflow on a set of increasingly complex test cases. We will also discuss technical challenges and lessons learned from the study, and recommendations for the CG methodology, the choice of ML algorithms, and the data needs.
... Here ̄ is the volume-averaged heat flux (this procedure is needed to ensure the correct code behavior on unstructured meshes), ℎ is the latent heat of evaporation. This algorithm was verified against analytical bubble growth model [29] and experimental data for pool and flow boiling [30]. ...
High-fidelity flow boiling simulations are conducted in a vertical minichannel with offset strip fins (OSF) using R113 as a working fluid. Finite-element code PHASTA coupled with level set method for interface capturing is employed to model multiple sequential bubble nucleation using transient three-dimensional approach. The code performance is validated against experiments for a single nucleation site in a vertical rectangular channel. To test the code performance, the studies for a bubble departing from the wall in a minichannel with OSF are carried out first. Due to low heat flux values applied to the channel (1 kW/m2) contribution from the microlayer is not considered. The influence of surface characteristics such as contact angle and liquid superheat on bubble dynamics are analyzed. Local two-phase heat transfer coefficient is also investigated. To achieve higher void fractions, two conic nucleation cavities are introduced in the same channel with OSF. Observed bubble characteristics (departure diameter, bubble departure frequency) are evaluated and bubble trajectories are presented and analyzed. Local heat transfer coefficient is evaluated for each simulation case. The results show approximately 2.5 times increase in the local heat transfer coefficient when individual bubbles approach the wall. With a smaller bubble nucleation diameter, heat transfer coefficient increases by two times. The current work shows the capability of modeling flow boiling phenomena in such complex geometry as OSF as well as data processing advantages of high-resolution simulations.
... There has been significant progress in the application of PHASTA code for direct numerical simulations on boiling mechanics, including boiling capability development and application Li, 2019;Li et al., 2019;. Initially, the boiling capability was implemented with the contact angle sub-grid model and then verified and validated for both pool boiling and flow boiling cases . ...
... PHASTA uses the level-set method, coupled with the finite element-based Navier-Stokes flow solver, which can achieve accurate solutions in simple and complex geometries and phase conditions (Feng and Bolotnov, 2015). Under the Consortium for the Advanced Simulation of Light Water Reactors (CASL), a boiling model was implemented in PHASTA to study boiling phenomena with high-resolution interfaceresolved simulations Li, 2019;Li et al., 2019). Such highfidelity boiling simulations have become more affordable with the advancement of HPC. ...
Boiling has proved to be one of the most efficient means for heat transfer and is a very important phenomenon during severe accident scenarios in light water reactors. High-fidelity pool boiling simulations can provide a numerical database for improving mechanistic boiling models by allowing for specific evaluation of interactions among bubbles. Previously published pool boiling simulations investigated two nucleation sites in which bubble growth at one site suppressed nucleation at the other site. Based on previous study results, more complicated interface-capturing simulations on pool boiling were conducted using PHASTA code with locally refined unstructured mesh.
First, different boundary conditions (BCs) were assessed to support robustness and reproducibility of the boiling model. Then, a scale study was conducted at a larger domain with nine nucleation sites where either nine or four nucleation sites are activated. Involving more nucleation sites increased the complexity of bubble interactions from surrounding sites. Finally, bubble departure behavior influenced by wall heat flux was investigated. When heat flux was increased, the order of bubble departure changed, but diagonal bubbles always departed after one another. The departure time interval between the first and second bubble reduced as heat flux increased. The corresponding frequency was almost linearly proportional to the heat flux. In addition, bubble departure behavior was found to be greatly influenced by the nucleation site pattern. Multiple nucleation sites resulted in superimposed inhibitive effects from surrounding sites to each bubble, which extensively delayed the departure. This new observation was not discussed in previously published works.
The work presented here provides new insight on the fundamental understanding of boiling phenomena, contributes to the development of a 3D multiphase computational fluid dynamics (M-CFD) model, and provides a more comprehensive database for data-driven pool boiling studies.
... where is the bubble radius, is the radius of temperature gradient collection shell [13], is the liquid phase thermal conductivity. This model has been verified against analytical bubble growth model [15] and validated against experimental data [16]. ...
Flow boiling in a minichannel with offset strip fins (OSF) is investigated in the current study. Finite element code PHASTA coupled with the level set method is used to model multiple sequential bubble nucleation from two conic cavities using transient three-dimensional approach. Refrigerant R113 fluid properties are used for the working fluid. Single-phase simulations are performed prior to two-phase modeling to achieve fully developed velocity and temperature profiles in the OSF geometry. These results are used as initial conditions for the flow boiling simulations. Since temperature distributions uncertainty may correspond to different locations in the overall heat exchanger, several studies for different wall superheat values (2, 4 and 6 K) are conducted. Observed bubble characteristics (departure diameter, bubble departure frequency) are evaluated and bubble trajectories are presented and analyzed. Local heat transfer coefficient is evaluated for each simulation case. The results show approximately 2.5 times increase in the local heat transfer coefficient when individual bubbles approach the wall. With a smaller bubble nucleation diameter, heat transfer coefficient increases by two times. The current work shows the capability of modeling flow boiling phenomena in such complex geometry as OSF as well as data processing advantages of high-resolution simulations.
... Here -̄ is the volume averaged heat flux (this procedure is needed to ensure the correct code behavior on unstructured mesh), ℎ is the latent heat of evaporation. This model performance was verified against analytical bubble growth model [23] and experimental data for pool and flow boiling [24]. ...
In the flow boiling the heat transfer rate is enhanced due to phase change as well as more effective convective heat transfer. With the growth of computational power, simulations may facilitate the design process by numerically evaluating the heat exchanger designs through robust boiling simulation capabilities. In this study, the direct numerical simulation (DNS) is employed to analyze the flow behavior and estimate two-phase heat transfer coefficients in a minichannel with offset strip fins (OSFs) for refrigerant R113. Finite-element flow solver PHASTA coupled with level-set method to capture the interfacial dynamics is used to conduct simulations. Bubble dynamics characteristics (bubble departure diameter, departure time) are investigated at different conditions, including the apparent contact angle variation. Since the boiling process is significantly influenced by the temperature distribution in the channel, sensitivity studies with different superheat liquid values are conducted. The local heat transfer enhancement is analyzed on the surfaces along the bubble path. The present research shows the possibility to model flow boiling in a complex geometry with unstructured mesh which assists in early stage design of advanced heat exchangers.
... An ITM-based boiling model has recently been developed and implemented in PHASTA to enable large-scale boiling simulations in complex geometries. 51 The phase-change process is governed by an evaporation-condensation model, 52 which couples a scalar equation that calculates the gas volume increase/decrease due to the local interface temperature gradient with the continuity and momentum equations. A saturation temperature is assumed for vapor bubbles. ...
This critical review paper outlines the recent progress in high-resolution numerical simulations of two-phase coolant flow in light water reactor–relevant geometries by resolving the water-vapor interface. Rapid development of capabilities in high-performance computing is creating exciting opportunities to study complex reactor thermal-hydraulic phenomena. Today’s advances in thermal-hydraulic analysis and interface-resolved simulations will help pave the way to the next level of understanding of two-phase flow behavior in complex geometries. This paper consists of two major parts: (1) a brief review of direct numerical simulation and interface tracking simulation and (2) several opportunities in the near future to apply cutting-edge simulation and analysis capabilities to address the nuclear-related multiphase flow challenges. The first part will discuss typical computational methods used for the simulations and provide some examples of the past work as well as computational cost estimates and affordability of such simulations for research and industrial applications. In the second part specific application examples are discussed, from adiabatic bubbly flow simulations in pressurized water reactor subchannel geometry to the modeling of nucleate boiling. The uniqueness of this study lies in the specific focus on applications with nuclear engineering interest as well as new generation modeling and analysis methodologies. Together with the ever-growing computing power, the related large-scale two-phase flow simulations will become indispensable for the improved scientific understanding of complex two-phase flow phenomena in nuclear reactors under normal operation and postulated accident conditions.
