Filippo Menghini

CFD and Neutron codes coupling on a computational platform

In this work we investigate the thermal-hydraulics behavior of a PWR nuclear reactor core, evaluating the power generation distribution taking into account the local temperature field. The temperature field, evaluated using a self-developed CFD module, is exchanged with a neutron code, DONJON-DRAGON, which updates the macroscopic cross sections and evaluates the new neutron flux. From the updated neutron flux the new peak factor is evaluated and the new temperature field is computed. The exchange of data between the two codes is obtained thanks to their inclusion into the computational platform SALOME, an open-source tools developed by the collaborative project NURESAFE. The numerical libraries MEDmem, included into the SALOME platform, are used in this work, for the projection of computational fields from one problem to another. The two problems are driven by a common supervisor that can access to the computational fields of both systems, in every time step, the temperature field, is extracted from the CFD problem and set into the neutron problem. After this iteration the new power peak factor is projected back into the CFD problem and the new time step can be computed. Several computational examples, where both neutron and thermal-hydraulics quantities are parametrized, are finally reported in this work.

Optimal control problems for the Navier-Stokes system coupled with the k - ω turbulence model

Optimal control of fluid-dynamics systems has gained attention in the last several years from the scientific community because of its potential use in design of new engineering devices and optimization of existing ones. Many research works have extensively studied the optimal control for the system of Navier-Stokes but the problem of turbulence in these works is usually not taken into account because of the many difficulties arising from the numerical implementation and solution of the optimality system. In this work turbulence is considered by coupling the k - ω two-equation turbulence model with the averaged Navier-Stokes system. The complete optimality system is derived and the existence of a weak solution proven. Some numerical examples are reported.

A Patient-Specific Computational Fluid Dynamics Model of the Left Atrium in Atrial Fibrillation: Development and Initial Evaluation

Atrial fibrillation (AF) is associated to a five-fold increase in the risk of stroke and AF strokes are especially severe. Stroke risk is connected to several AF related morphological and functional remodeling mechanisms which favor blood stasis and clot formation inside the left atrium. The goal of this study was therefore to develop a patient-specific computational fluid dynamics model of the left atrium which could quantify the hemodynamic implications of atrial fibrillation on a patient-specific basis. Hereto, dynamic patient-specific CT imaging was used to derive the 3D anatomical model of the left atrium by applying a specifically designed image segmentation algorithm. The computational model consisted in a fluid governed by the incompressible Navier-Stokes equations written in the Arbitrary Lagrangian Eulerian (ALE) frame of reference. In this paper, we present the developed model as well as its application to two AF patients. These initial results confirmed that morphological and functional remodeling processes associated to AF effectively reduce blood washout in the left atrium, thereby increasing the risk of clot formation. Our analysis is a step forward towards improved patient-specific stroke risk stratification and therapy planning.

Adjoint optimal control problems for the RANS system

Adjoint optimal control in computational fluid dynamics has become increasingly popular recently because of its use in several engineering and research studies. However the optimal control of turbulent flows without the use of Direct Numerical Simulation is still an open problem and various methods have been proposed based on different approaches. In this work we study optimal control problems for a turbulent flow modeled with a Reynolds-Averaged Navier-Stokes system. The adjoint system is obtained through the use of a Lagrangian multiplier method by setting as objective of the control a velocity matching profile or an increase or decrease in the turbulent kinetic energy. The optimality system is solved with an in-house finite element code and numerical results are reported in order to show the validity of this approach.

Fluid structure interaction solver coupled with volume of fluid method for two-phase flow simulations

In this work we propose to study the behavior of a solid elastic object that interacts with a multiphase flow. Fluid structure interaction and multiphase problems are of great interest in engineering and science because of many potential applications. The study of this interaction by coupling a fluid structure interaction (FSI) solver with a multiphase problem could open a large range of possibilities in the investigation of realistic problems. We use a FSI solver based on a monolithic approach, while the two-phase interface advection and reconstruction is computed in the framework of a Volume of Fluid method which is one of the more popular algorithms for two-phase flow problems. The coupling between the FSI and VOF algorithm is efficiently handled with the use of MEDMEM libraries implemented in the computational platform Salome. The numerical results of a dam break problem over a deformable solid are reported in order to show the robustness and stability of this numerical approach.

COUPLING OF FLUID STUCTURE INTRACTION SOLVER WITH A VOF METHOD FOR MULTIPHASE STRUCTURE INTERACTION

In this work we propose a preliminary model to study the deformation of solid structures induced by the interaction with a two-phase flow. The study of Fluid Structure Interaction and multiphase problems is of great interest because of many potential applications ranging from the biomedical field to the pressure tank design. We use a monolithic approach for the FSI problem while a Volume Of Fluid method (VOF) is considered for the reconstruction and the advection of the multiphase interface. An unstructured, time dependent computational grid and a fine Cartesian mesh are used for the FSI and the VOF problem, respectively. The interaction between the two different grids is obtained by projecting the velocity and the displacement field into the Cartesian grid and the color function into the unstructured mesh. This projection is performed with the MEDmem libraries included in the Salome platform. Concerning the VOF method, for an accurate reconstruction of the interface a huge number of computational elements are required and a multilevel algorithm coupled to an efficient compression-expansion technique is developed to reduce computational costs and memory requirements. After the mathematical description of the problems we test the proposed algorithm with different cases where the solid domain undergoes to both small and large deformation.

ADJOINT OPTIMAL CONTROL PROBLEMS FOR FLUID-STRUCTURE INTERACTION SYSTEMS

In the last years adjoint optimal control has been increasingly used for design and simulations in several research fields. Applications to Computational Fluid Dynamics problems dedicated to the study of transient-diffusion equations, shape optimization problems, fluid-solid conjugate heat transfer and turbulent flows can be found in literature. The study of FluidStructure Interaction problems gained popularity recently because of many interesting applications in engineering and biomedical fields. In this paper we study adjoint optimal control problems for Fluid-Structure Interaction systems in order to improve the advantages of using FSI simulations when designing engineering devices where fluid-dynamical interactions between a fluid and a solid play a significant role. We assess distributed optimal control problems with the purpose to control the fluid behavior by moving the solid region to obtain a desired fluid velocity in specific parts of the domain. The adjoint equations of the FSI monolithic system are derived and the optimality system solved for some simple cases with an in-house finite element code with mesh-moving capabilities for the study of large displacements in the solid. The approach presented in this work is general and can be used to assess different objectives and types of control in future works.

NUMERICAL VALIDATION OF A FOUR PARAMETER LOGARITHMIC TURBULENCE MODEL

Computational Fluid Dynamics codes are used in many industrial applications in order to evaluate interesting physical quantities, such as the heat transfer in turbulent flows. Commercial CFD codes use only turbulence models with an imposed constant turbulent Prandtl number Prt, which can give accurate results only for simulations when a strong similarity between the velocity field and the temperature field can be assumed. For fluids with a low Prandtl number, as for heavy liquid metals, a constant turbulent Prandtl number leads to an overestimation of the heat transfer, so experimental results and Direct Numerical Simulation cannot be reproduced. In this work we propose a new k-Ω-kθ-Ωθ turbulence model as an improvement of the k-ω-kθ-ωθ turbulence model, already validated by the authors, where Ω and Ωθ are calculated as the natural logarithm of the variables ω and ωθ. With this reformulation of the previous turbulence model we obtain some important advantages in numerical stability and robustness of the code. Results for the simulations of fully developed turbulent flows in two and three dimensional geometries are reported and compared with experimental correlations and DNS data, when available.

Numerical validation of a κ-ω-κ θ -ω θ heat transfer turbulence model for heavy liquid metals

The correct prediction of heat transfer in turbulent flows is relevant in almost all industrial applications but many of the heat transfer models available in literature are validated only for ordinary fluids with Pr 1. In commercial Computational Fluid Dynamics codes only turbulence models with a constant turbulent Prandtl number of 0.85 — 0.9 are usually implemented but in heavy liquid metals with low Prandtl numbers it is well known that these models fail to reproduce correlations based on experimental data. In these fluids heat transfer is mainly due to molecular diffusion and the time scales of temperature and velocity fields are rather different, so simple turbulence models based on similarity between temperature and velocity cannot reproduce experimental correlations. In order to reproduce experimental results and Direct Numerical Simulation data obtained for fluids with Pr 0.025 we introduce a κ-e-κθ-eθ turbulence model. This model, however, shows some numerical instabilities mainly due to the strong coupling between κ and e on the walls. In order to fix this problem we reformulate the model into a new four parameter κ-ω-κθ-ωθ where the dissipation rate on the wall is completely independent on the fluctuations. The model improves numerical stability and convergence. Numerical simulations in plane and channel geometries are reported and compared with experimental, Direct Numerical Simulation results and with results obtained with the κ-e formulation, in order to show the model capabilities and validate the improved κ-ω model.