RICERCA ITALIANA     

Research program

Innovative techniques for the enhancement of forced convection

University Co-ordinator

Politecnico di MILANO - ENERGETICA - ()

Research Unit Leader

Alfonso Giuseppe Vincenzo Niro

Description

PROJECT DESCRIPTION
We will investigate, experimentally and numerically, local and average heat transfer characteristics and pressure drop in forced convection inside wavy or ribbed channels. Channels have a rectangular cross-section with the upper and lower walls both wavy and plane with square cross-section ribs variously arranged with respect to the streamwise direction. Channels are operated under fixed temperature conditions and they are sufficiently long to allow attainment of fully developed conditions of velocity and temperature. Air flow rate can be varied so that the Reynolds number can range between 500 and 20000.
Experimental analysis will be carried out by evaluating the mean heat transfer characteristics of all investigated geometries by means of conventional measurements of temperatures and heat flux. The local heat transfer coefficients will be also evaluated but only for two-dimensional flows, i.e., at low Reynolds numbers and for surfaces with waviness or ribs orthogonal to the flow direction, because these measurements are performed by holographic interferometry or digital speckle photography (these techniques permit to measure the temperature- and temperature-gradient- field, respectively, inside the air flow over the heated surface but phenomena occurring along the propagation direction of light, that is in the channel width, are averaged).
However, it is very important to know the influence of ribs on fluid dynamics specially for three-dimensional geometries and turbulent flows; hence, local heat transfer characteristics will be analyzed as well by infrared thermography, even though in slightly different thermal-fluid dynamic conditions. Thermographic techniques, particularly infrared thermography, allow the measure of the whole temperature field over a surface, whatever its geometry, with good precision and spatial resolution. Thus, if the heat flux distribution is known, the heat transfer coefficient may be determined at each location. In order to grab thermographic images, either the channel has to be equipped with a IR-transparent window on the wall lying in front of the ribbed surface or operation have to take place under external convection conditions. Moreover, electrical heating is needed to produce a known distribution of heat flux over the tested surface (however, between the surface and the heater, a sufficiently thick layer will confine in few degrees the temperature differences over the surface itself). It is clear that this arrangement partially modifies operating conditions, but the resulting information is in any case very useful to understand thermal and fluid dynamic phenomena as well as to optimize geometrical features. Eventually, during the experimental program, attention will be devoted to improve the filtering techniques needed by image digital processing and required, in particular, for heat conduction calculations in the layer between the surface and the heater (this layer being thick, diffusive components change significantly the heat flux distribution).
In parallel with the experimental approach, the research line related to CFD will be developed by using customized commercial code. Computational Fluid Dynamic has been more and more used within the last decade since it has been able to improve its accuracy thanks to higher order numeric, to increase the number of available models, to be faster and cheaper. Nevertheless, the fundamental research is still required in the field mainly when dealing with the requirement of modeling turbulence. For industrial purposes, turbulent models , natively implemented in the commercial code such as those of the k-epsilon family, have been widely used thanks to their stability and lightness. However they cannot reliably reproduce all kind of phenomenon occurring in industrial flow, since they employ the Boussinesque hypothesis. This hypothesis assumes a linear relation between the deviatoric components of the Reynolds stress tensor and the strain rate tensor, and that the normal component is proportional only to the scalar turbulent kinetic energy.
Different lines in term of turbulence modeling have been opened within the research group and will be implemented in the current project. By using the approach known as EARSM (Explicit Algebraic Reynolds Stress Model), a non linear eddy viscosity model seems to fit better with the complex flow requirements (e.g., Pope in [19], Craft et al. [20], Baglietto [21]), since it adds higher order terms that bring more sensitivity of the Reynolds stresses to flow motion. The researcher have been working since 2004 on the implementation of different formulations for two equation turbulent models by using a non linear formulation of the Reynolds stress tensor (NLEVM): a second order k-epsilon model based on the studies of Shih, Zhu and Lumley (1993), respecting realizability and with a low Reynolds formulation, with the introduction of a proper damping function on the turbulent viscosity, has been implemented.
The use of a Large Eddy Simulation (LES) approach, moving away from RANS limitations, is the second domain of investigation. Indeed this method lies between RANS turbulence models and Direct Numerical Simulation: LES filters turbulence dimensional scales and operates a cut on the scales directly simulated. Solving directly mainly of the 80% of turbulent kinetic energy involved in the flow, LES requires fine spatial and temporal discretization and a three-dimensional work domain and it is therefore more demanding in term of CPU time and resources allocation than NLEVM. Both the approaches will be followed in the current research in order to reach the best compromise between accuracy of prediction and time constraint.



PROJECT SCHEDULE
The project is for a two-years term and it is divided in two one-year stages. Goals and research activities are scheduled during the two stages as in following.

FIRST YEAR.
Goals. Investigation of the average thermal performances of wavy/ribbed channels; analysis of the local heat transfer characteristics for two-dimensional geometries; setting up of a second order non-linear numerical model.
Main actions. Setting up of test sections (rectangular channels with the upper and lower surfaces wavy or plane with square cross-section ribs variously arranged with respect to the flow direction); setting up of the experimental apparatus and instrumentation; carrying out experimental tests with conventional measurements (temperature, volume flow rate, heat transfer rate, heat flux, pressure drop); measurement of the temperature- and temperature-gradient- fields; comparison with conventional-technique results. Regarding numerical analysis, implementation of a non linear second order model for turbulent flow to be validated on the basis of literature bench marks.
Verification criteria. Reports and conference contributions on the obtained results.

SECOND YEAR.
Goals. Experimental analysis of the local heat transfer characteristics for three-dimensional geometries; refinement of the numerical model and analysis of three-dimensional cases; proposal of heat transfer correlations.
Main actions. Refinements of the current filtering techniques to process thermographic images; running of experimental tests for local measurements over three-dimensional geometries in internal/external forced convection. Improvement of the numerical model by Low-Reynolds approach and third-order version of the Boussinesque approximation; results validation by both literature and available experimental data. Concerning LES, we are not going to implement new models, but comparative studies to understand applicability ranges, use opportunities, and possible future developments. Comprehensive analysis of the experimental and numerical collected data in order to propose heat transfer correlations.


SINERGIES WITH THE OTHER RESEARCH UNITS
This project will be developed in close cooperation with the Research Units operating at Parma, Udine and Ancona. In fact, the Parma RU will experimentally study heat transfer over extended fins by means of the IR thermography; with a technique similar to that we will use to investigate three dimensional geometries; consequently, research activities of these two RU will be carried out in strong coordination by comparing experimental results while work is progressing. Strong cooperation is also expected with the Udine RU on computational activities because we share investigation of heat transfer inside rectangular channels with structured surfaces but curried out with methods and strategies fully independent; special interest arises from comparing the numerical results as the two Research Units used. Cooperation with the Ancona Research-Unit will essentially regard investigation methodologies, since they have a very relevant experience in optical techniques and a long tradition of cooperation with our group in this field.
Finally, we should supply the Bologna RU with experimental data on heat transfer in transient regime in addition to those in steady state; these data would be most useful for modeling activities performed by Bologna RU because they provide an insight view of local transfer mechanisms. However, we are aware that experimental analysis in transient regime is much more difficult than in steady-state. In turn, at the end of the first activity year, the Bologna RU could give us useful suggestions on how to reshape the actual wavy/ribbed channel for thermally-improved performances.


CONCLUSIONS
We would remark this research program is essentially aimed to setting up high-efficient extended surfaces that could conspicuously enhance air-side heat transfer in compact heat exchangers - this should help Italian manufacturers to remain competitive in this market - as well as in high-tech systems or largely used devices. The main results of this program will be a large data base, correlations and numerical codes for designing these high-efficient extended surfaces. As further fall-out, there will be the training of two young scientists and the implementation/improvement of two novel techniques, i.e., the IR thermography for local heat transfer measurements and non linear viscosity models within CFD codes.