J. Castro

Evaluation of a Small Capacity, Hot Water Driven, Air-Cooled H2O-LiBr Absorption Machine

prototype of an air-cooled absorption chiller of about 2 kW for air conditioning using H2O-LiBr has been developed. The unit has been conceived as a laboratory experimental test device with removable components to facilitate modifications of the initial design. Several tests have been carried out under different conditions. The experimental results have been compared with the theoretical ones based on global mass and energy balances over the different components of the system. Detailed simulation models for each heat exchanger have been developed and implemented in the numerical codes to calculate the overall heat transfer coefficients and subcooling values for the whole system simulation. The conclusions reported will lead to future design revisions and improvements to achieve better performance and reliability.

Multidimensional and Unsteady Simulation of Fin-and-Tube Heat Exchangers

This article presents a model for the analysis of fin-and-tube heat exchangers, focusing on the heat conduction processes within the finned tube bundle. A cutting cell discretization has been proposed for the fins to adapt to the tubes shape, while the tubes have been discretized in axial and angular directions to consider complex heat transfer coefficient variations. A set of results is given on fin efficiency and transient response comparing with well-established methods. A full-scale condenser is also analyzed as an illustrative result, detecting important thermal bridges through the fins.

Detailed analysis of turbulent flows in air curtains

In order to prevent entrainment, an air curtain should provide a jet with low turbulence level, and enough momentum to counteract pressure differences across the opening. Consequently, the analysis of the discharge plenum should be taken into consideration. Hence, the main object of this paper is to study the discharge chamber geometry and the presence of blades for ?ow orientation. This analysis is carried out in order to understand their in?uence on the jet produced. Studies presented are based on detailed numerical simulations and on experimental measurements. The in?uence of the turbulence model, boundary conditions and computational domain is investigated.

DNS of the wall effect on the motion of bubble swarms

Abstract This paper presents a numerical study of the gravity-driven motion of single bubbles and bubble swarms through a vertical channel, using High-Performance Computing (HPC) and Direct Numerical Simulation (DNS) of the Navier-Stokes equations. A systematic study of the wall effect on the motion of single deformable bubbles is carried out for confinement ratios CR = {2,4,6} in both circular and square channels, for a broad range of flow conditions. Then, the rising motion of a swarm of deformable bubbles in a vertical channel is researched, for void fractions α = {8.3%, 10.4%, 12.5%} and CR = {4, 6}. These simulations are carried out in the framework of a novel multiple marker interface capturing approach, where a conservative level-set function is used to represent each bubble. This method avoids the numerical and potentially unphysical coalescence of the bubbles, allowing for the collision of the fluid particles as well as long time simulations of bubbly flows. Present simulations are performed in a periodic vertical domain discretized by 2 × 106 control volumes (CVs) up to 16.6 × 106 CVs, distributed in 128 up to 2048 processors. The collective and individual behavior of the bubbles are analyzed in detail.

DNS of the Rising Motion of a Swarm of Bubbles in a Confined Vertical Channel

The motion of bubbles and droplets is ubiquitous in a variety of natural processes and technological applications, such as boiling heat transfer, steam generators of nuclear power plants, unit operations of the chemical engineering (e.g. distillation, absorption columns, bubble reactors), micro-devices, among others [8].

DNS OF FALLING DROPLETS IN A VERTICAL CHANNEL

This paper presents Direct Numerical Simulation (DNS) of the falling motion of single and multiple deformable drops in a vertical channel. A systematic study of the wall effect on the motion of single drop is performed for Eötvös number (0.5≤Eo≤5), Morton number (10−3≤M≤10-8), and confinement ratio CR = 2. Second, the gravity-driven motion of multiple drops and their interactions are studied in a periodic vertical channel for CR = 4. These simulations are performed using a multiple marker level-set methodology, integrated in a finite-volume framework on a collocated unstructured grid. Each droplet is described by a level-set function, which allows capturing multiple interfaces in the same control volume, avoiding the numerical merging of the droplets. Numerical algorithms for fluid motion and interface capturing have been developed in the context of the finite-volume and level-set methodology, surface tension is modeled by means of the continuous surface force approach, and the pressure-velocity coupling is solved using a fractional-step projection method. DNS of single drop shows that they migrate to the symmetry axis of the channel when the Reynolds number is low, following a monotonic approach or damped oscillations according to the dimensionless parameters. If Eötvös number increases, stronger oscillations around the symmetry axis are observed. Simulations of multiple drops show that the collision of two drops follows the drafting-kissing tumbling (DKT) phenomenon. Deformable drops do not collide with the wall, whereas DKT phenomenon in the droplet swarm leads to the formation of groups which move through the center of the channel.

On the validity of the Oberbeck-Boussinesq approximation in a tall differentially heated cavity with water

In the present work, the validity of the Oberbeck-Boussinesq approximation for the resolution of fluid flow and heat transfer phenomena inside a differentially heated cavity filled with water ( Pr = 3.44) is submitted to investigation. The cavity models, the integrated solar collector-storage element installed on an advanced facade. According to Gray and Giorgini (1976), the use of the Boussinesq approximation can be considered valid for variations of thermosphysical properties up to 10% with respect to the mean value. In the configuration under this study, there is a variation of about 17% in the dynamic viscosity and 15% in the thermal expansion coefficient. The significance of the Oberbeck-Boussinesq effects is studied comparatively by means of detailed direct numerical simulations (DNS). Numerical results reveal that for the conditions under study, important variations from the Oberbeck-Boussinesq conditions occur. These effects are observed as dissimilarities between boundary layers in hot and cold walls, and as a consequence on the local Nusselt number. In addition, there is a loss of symmetry in the flow within the cavity which is reflected in the temperature and velocity profiles.

New Implementations of Surface Tension Forces for PLIC-VOF Methods

A comparative study has been carried out between different models of implementing surface tension for PLIC-VOF methods: i) the well known Continuum Surface Force (CSF); ii) Staggered Grid Interface Pressure (SGIP); iii) Modified Meier’s. As main difference with respect the CSF model, the two last use information of the interface location for the calculation of the surface tension forces, therefore, they need a model of interface reconstruction. The models will be tested under static and dynamic conditions, for the case of a drop and bubble between water and air. In order to reduce the ‘parasitic’ currents that is characteristic of this type of models, two different strategies are used: i) a new method for the calculation of the interface curvature based on the use of polar coordinates; ii) the use of kernels for smoothing the jump of the colour function and the interface. Additionally, the influence of the order of accuracy of the reconstruction algorithms employed that affect the curvature estimation and the surface tension force calculation is also checked. After testing the different cases, it can be concluded that SGIP and modified Meier perform in a comparable way, and much better than CSF model without kernel, that is only competitive with the use of the kernels. However, modified Meier is preferred respect SGIP due to its straitghforward implementation on unstructured meshes.Copyright