Annual Report Department

2.3Thermal and fluid engineering

2.3.1Enhanced heat transfer in electronic equipment cooling

Air cooling of electronics is pushed to its limits due to increasing heat load and ongoing miniaturisation. More and more, critical components are liquid-cooled. At present a liquid cooling by means of a closed loop is investigated. A small heat exchanger is placed on thermally critical components and the heated fluid is directed to a series of air-cooled heat exchangers. In an experimental set up, the characteristics of an existing heat exchanger are measured. Based upon numerical modelling this heat exchanger is then geometrically optimised.

Furthermore, different heat transfer enhancement techniques will be investigated. Passive turbulence control with a surface modification technique is used to increase the heat transfer and to lower the pressure drop. This finally results in higher heat transfer at constant pumping power. For the air-cooling method on the other hand, the same principles are used to optimise heat sinks. Small geometrical features will enhance heat transfer and lower flow resistance. In this way, less cooling fins are needed which results in lower flow bypass around the heat sinks.

Publications and reports: 2003P01, 2003P02

Scientific staff: M. Baelmans, K. Nevelsteen

2.3.2Electrostatic actuation of liquid droplets for electronics cooling

The miniaturisation of electronic components and the rapid increase in power density of advanced microprocessors and electronic components have created a need for improved cooling technologies to achieve high heat dissipation rates. Some of future microprocessors and power-electronic components have been projected to dissipate over 1000 W/cm². Liquid cooling systems with micro channels enable to achieve relatively high heat fluxes. However, a large pressure drop is induced by the flow through these channels, leading to high requirements for the pumping facilities. Therefore, an alternative actuation technique, without a mechanical pump or valves is investigated: discrete liquid droplets are electrostatic actuated by electrodes. In this way, it is possible to create a flow of droplets. Moreover, the droplets trajectory can be directed towards specific hot spot locations of the electronic components.

Finite element modelling is used to calculate the shape of the droplet and the electrostatic actuation force acting on the droplet. Next, CFD modelling will be applied to model the motion of the droplet through the channel filled with air, the internal circulation flow in the droplet and the heat transfer between the droplet and the electronic component. The results of the modelling will be validated by an experimental set up. Using a test structure with a transparent substrate the flow can be visualised and the droplet speed can be measured. The heat transfer will be measured with a thermal test chip, containing resistors to dissipate a known amount of heat and diodes to measure the temperature. Finally, the results of the numerical models on one hand and the experimental results on the other, will be used to enhance both flow and heat transfer rate.

The research is undertaken in close collaboration with the division Microsystems, Components and Packaging of Imec.

Scientific staff: M. Baelmans, H. Oprins

2.3.3Design of a micro gas turbine for electrical power generation

This research aims at developing a micro gas turbine for the generation of portable electrical power. A standard Brayton gas turbine cycle with recuperation will be designed in particular. This study focuses mainly on the heat transfer phenomena in micro systems. Based on a detailed study of both heat transfer and pressure drop in small components optimal design rules for micro heat exchangers will be deduced. The thermal integration of the different micro gas turbine components (compressor, recuperator, combustion chamber and turbine) will also be investigated. Finally a micro gas turbine for electrical power generation will be built and tested.

Related projects: IWT SBO-powermems.
Publications and reports: 2003A17, 2003P35, 2003P36, 2003P37, 2003P38
Scientific staff: M. Baelmans, F. Verplaetsen, D. Reynaerts

2.3.4Large Eddy Simulations of turbulent flows using finite volume methods

Accurate simulations of turbulence still pose a major challenge in fluid mechanics. Indeed in this regime, the Navier Stokes (NS) equations show a highly chaotic behaviour and its solution contains various orders of length and time scales. In a Large Eddy Simulation (LES), these scales are filtered such that only the largest are retained. The smallest scales (referenced as SubGrid Scales (SGS)) give rise to a closure term in the corresponding filtered NS-equations. The ongoing research is directed towards the further development of LES in finite volume methods. Emphasis is on three different topics. First of all, theoretical and numerical work is performed on the interaction between subgrid models and the numerical discretisation scheme. The aim of this study is to formulate a priori methods which can be used to set up LES in the most optimal way, acounting for modeling and discretization errors. Second, the effect of different possible constructions of inlet boundary conditions on the solution of an LES or DNS is studied. Here, based on DNS simulations of a temporal mixing layer, a comprehensive sensitivity analysis is performed, with main emphasis on the determination of those statistical parameters which most determine the downstream evolution. In order to link this work to actual flow development in physical situations, an experimental program is executed simultaneously, where, in a wind tunnel, the development and similarity solution of a mixing layer is investigated for different turbulent inlet boundary conditions. Finally, the formulation of LES for (mainly non-premixed) combustion is studied. Here, the interaction between chemical reaction and the turbulent fluid flow, is investigated based on the mixed-is-burnt approach and on flamelet modelling. Work on this topic is mainly in an initial coding phase. Actual LES cases currently in use are homogeneous isotropic turbulence, the mixing layer, the plane jet flow, and combustion in a plane jet flow.

Related projects: FWO.-project G.0130.02: Large Eddy Simulation for multiphase flows and complex physics flows with emphasis on combustion applications.

Publications and reports: 2003P19, 2003P15, 2003A20, 2003P46

Scientific staff: M. Baelmans, M. Fathali, J. Meyers, W. Derkinderen, G. Rubio

2.3.5Computational Fluid Dynamics for aero-acoustic application

Aero-acoustics deals with the sound generated by fluid flow. The ability to predict and understand this flow-generated sound can optimise design with respect to this noise. In spite of several analytical and numerical solutions in this research, a lot of open questions in this field still remain.

Different research topics are pursued. First of all, research is directed at the development of stochastic noise generation and radiation (SNGR) methods, which can be used in combination with Reynolds-Averaged Navier-Stokes simulations (RANS) to provide the noise source terms. A circular jet with M=0.86 is studied. Emphasis is on the correct axisymetric formulation of the SNGR method, on the algortihms which provide the best physical time correlations and on the way the turbulent energy spectrum, needed in the SNGR, should be discretized. As an alternative to the RANS-SNGR methodology to produce the noice sources, large-eddy simulations are considered. To this end, an LES code, specifically designed for Computational aero-acoustics, is being developed. Finally, research is also directed at the development of a propagation code, which is needed to efficiently calculate the noise propagation from the source region to the far field.  The method being studied, is based on the solution of the linearized Euler  equations (LEE), which are discretized using a discontinuous Galerkin method.

Publications and reports: 2003P47

Scientific staff: M. Baelmans, W. Desmet, M. Mesbah, G. Rubio, W. De Roeck, Y. Reymen

2.3.6Numerical computation of air, moisture and heat transfer in chicory root cold store

To provide the market with high quality Belgian endive all year round, cooling is applied to store the chicory roots for an extended period of 9 months. Suboptimal air conditions (air flow rate, air temperature and relative humidity) may cause the chicory roots to suffer freezing injury, water loss due to evaporation, rotting due to a non-effective cooling. Besides, non-uniform cooling causes the roots to have a large variability in quality, which will affect the chicon quality afterwards. Therefore, storage operations need to be designed to minimise product losses. This research aimed to develop a model, which predicts the airflow, heat and mass transfer processes in the bulk of chicory roots and industrial cold stores. A two-phase model for heat and mass transfer inside bulk of chicory roots was developed. The model predictions were validated for the air and product temperature, the air moisture content and the product weight loss. A transient three-dimensional CFD model was developed to calculate the velocity and temperature distribution in an existing empty cold store and for a loaded one. The model was validated by means of velocity and temperature measurements. CFD was finally applied to study the effect of the air gaps in bins to the cooling process and to optimise the design of a new cold store. It was shown that models of different complexity can and should be used to study heat and mass transfer in cold stores at different scales.

This research is undertaken in close collaboration with prof. B. Nicolai and M.L. Hoang from the Laboratory of Postharvest Technology, Department of Agro-Engineering and Economics, K.U.Leuven.

Publications and reports: 2003P31, 2003P32, 2003D02, 2003P48

Scientific staff: M. Baelmans

2.3.7Comparison between database and mechanistic modelling of imperfectly mixed fluids

It is widely recognised that the internal dynamics of heat and moisture transfer in an imperfectly mixed ventilated airspace have a fairly complex and spatially heterogeneous nature. For many researchers concerned with HVAC it is a major challenge to control these heat and moisture transfer dynamics by using model based control theory. However, before this can be applied, it is first required to have an appropriate dynamic model of the process to be controlled.

Over the past decades, this has led to the development of sophisticated numerical CFD models which offer bright prospects as a design, and process optimisation tool, but are too complex to be used for model based control purposes. On the other hand, data-based or statistical methods allow to model the apparently complex nature of the heat and moisture transfer processes in a dynamically simplified manner. These models are simple in structure, inherently stochastic in form and very useful for model based control purposes. However their development is highly dependent on experimental data.

The purpose of this research work is primarily to compare the two modelling techniques using real time series data. In later stages, to use the CFD prediction instead of experimental data so that it is possible to build a data-based model, which subsequently can be applied for control purposes. In other word, the final goal is to build simplified models without performing experimental work, which can be used for controller development.

This research is undertaken in close collaboration with prof. D. Berckmans and T. Zerihun Desta Tadiwos from the Laboratory for agricultural buildings research, K.U.Leuven.

Publications and reports: 2003P20, 2003P28

Scientific staff: M. Baelmans, J. Meyers

2.3.8The physics of condensation phenomena

Although condensation of vapour seems to be a very common and simple phenomenon, it can become very compli­cated in the presence of non-condensable gasses and/or in combination with simultaneous momentum transport. An accurate description of condensation in these conditions, is important for different kinds of processes. This research aims at a better understanding of the different mass, energy and momentum transfer processes occurring at the interface between vapour and liquid.

Without applying any phenomenological closing equation, but starting from first principles, the thermal equilibrium between vapour and liquid is modelled, using statistical thermodynamics and direct computer simulation of the condensation process on a microscopic level. Therefore, rigorous analysis of the governing equations is undertaken and software for the direct simulation has been written.

Special care is taken of boundary conditions as these determine to a large extent the physics of the simulations. On-line statistical analysis and computation of thermodynamical properties (temperatures, pressure, internal energy, virial coefficient) provide the necessary tools for interpretation of the results of the simulations. Detailed results for the stress and pressure distribution, particle distribution and velocity distributions are generated.

Publications and reports: 2003P10

Scientific staff: W. D’haeseleer, G. Van den Branden

2.3.9Thermal modelling of oil-filled power transformers

An important parameter in the design of power transformers is the hotspot or highest winding temperature. Indeed, this temperature has a major influence on the ageing of a transformer. Both manufacturers and customers are there­fore interested in an accurate prediction of this temperature. Hence a detailed steady state model of the thermal behaviour of a power transformer is developed. In order to reduce computational efforts the governing equations, describing oil flow and oil temperature distribution are integrated over each oil channel. This leads to one-dimensional momentum and temperature equations for each channel. Addition of momentum and continuity equation in each channel connection completes the oil flow model. The oil temperature equations are combined with a lumped system approach for each thermal mass (e.g. winding layer, winding disc, radiator, core…). The obtained global model is reduced to two sets of coupled linear equations by using standard CFD techniques. Thus, a fast and adequate solution procedure is obtained. Once this has been done, the model is split up in different components so each part of the transformer can be calculated separately, allowing the design team to see the consequences of their changes without having to make the combined calculation of the total transformer. For this model, flow correlations in 4way-intersections are derived based upon detailed laminar flow computations with a CFD code. Evaluation of these correlations by means of comparison with experiments is currently undertaken.

Related projects: IWT-project with Pauwels Trafo Belgium: Modelling and investigation of new cooling techniques for power transformers.
Scientific staff: M. Baelmans, K. De Troch

2.3.10Active control of jets

In this research, the active control for jet vectoring is studied with respect to heating & ventilation air conditioning, automotive applications and burners.  An experimental set-up has been built to study the deflection of 2-D jets by acoustic actuation.  Parameters such as frequency and amplitude of oscillatory non-fluid control jets are optimised in terms of efficiency. Also the active control of annular, axi-symmetric jets with cross-flow injection of small, secondary control jets is studied.

Publications and reports: 2003P49

Scientific staff: E. Van den Bulck, M. Vanierschot

2.3.11Hysteris phenomena in annular swirling flows

This study investigates the hysteresis that is observed in the sudden expansion of annular swirling jets.  Experiments have indicated that, depending upon the particulars of the jet outlet geometry (conical expansion, stepwise expansion, …), a large hysteresis loop has been observed between a regular jet outflow and a coanda outflow.  This behaviour is simulated with the aid of a CFD programme, using axi-symmetric RANS and various turbulence models.  These simulations also indicate the hysteresis.

Publications and reports: 2003P09, 2003P27

Scientific staff: E. Van den Bulck, M. Vanierschot