Session H06: Convection and Buoyancy II

Physics-Informed Neural Networks for Forward and Inverse Multiphysics Heat Transfer Problems

Heat transfer modeling plays a key role in different scientific and engineering fields. However, often parameters are not fully known leading to ill-defined heat transfer problems, which cannot be easily tackled with traditional computational methods. Additionally, the coupling with nonlinear fluid flow phenomena further aggravates these issues. One approach for solving these kinds of problems is physics informed neural networks (PINN), which provides a hybrid data-driven and physics-based solution to ill-posed problems. In this talk, we present different applications of PINN in Multiphysics convective heat transfer problems. First, heat transfer in fins where conduction in solid is coupled with convection in fluid is considered. We quantify the base temperature and thermal conductivity of the solid using sparse temperature measurements in the fluid domain. We present the sensitivity of the results to sparse sensor placement strategies. Additionally, we study the challenges of getting a unique solution in inverse problems and propose remedies to constrain the solution space based on prior physical knowledge. Finally, we present forward and inverse modeling of convective heat transfer in a rotating fluid-saturated porous medium.   

Supercritical heat transfer deterioration and pseudo boiling

In many systems, from rocket engines to concentrated solar power, pressures are increased to increase efficiency, thus resulting in these systems operating at supercritical pressures. At subcritical pressures, heat transfer deterioration (HTD) is a sudden reduction of heat flux when certain temperature thresholds are exceeded. It is well known and linked to the liquid-vapor phase transition. However, a HTD is also observed at supercritical pressures, despite no classical phase transition. To explore this phenomenon, we performed simulations of heat transfer in laminar boundary layers and observed supercritical HTD (scHTD). We found that the onset and strength of scHTD are linked to pseudo boiling, a nonequilibrium higher-order phase transition between supercritical liquid-like and gas-like states. Results agree well with theoretical predictions of the onset temperature and the maximum pressure at which scHDT occurs. Thus, pseudo boiling is physically what causes scHDT. Pseudo boiling is coupled to drastic changes in fluid properties, and thus accurate modeling is essential; the popular Peng Robinson equation of state leads to erroneous results.   

Heat transfer enhancement by flapping of streamwise-consecutive foils in a channel

We have computationally considered enhancement of forced-convection heat transfer between parallel plates of infinite spanwise extent, using an identical pair of deformable foils whose motion is driven by aeroelastic instability, over a range of Reynolds numbers (Re) relevant to air-side cooling in heat exchangers.  Computations were performed using a spectral-element discretization of the flow and heat transfer, coupled to a finite-element discretization of the foil motion.  The two foils are cantilevered in the midplane of the channel, with the leading edge of one downstream of the trailing edge of the other.  In air at Re = 250 (based on channel width), the overall heat transfer enhancement (measured in terms of a Nusselt number) considerably exceeds twice the enhancement due to a single foil, and persists considerably farther downstream.  This is directly attributable to the fact that the motion of the second foil is driven by the finite-amplitude, unsteady, asymmetric flow in the wake of the upstream foil, whereas the motion of the upstream foil is due solely to linear instability of the motionless-foil configuration with respect to the nominally steady and symmetric flow upstream of its leading edge.     

Heat transfer enhancements on solar farms evaluated through an enstrophy balance

Reduction of solar photovoltaic (PV) module temperatures improves efficiencies of energy extraction and reduces fatigue and degradation of the panels. Knowledge of how the heat expelled and the modules themselves interact with the surrounding atmospheric boundary layer can contribute to improvements of cooling based on farm orientation and spacing, a method of non-invasive passive cooling of the modules.  Investigations into the mean and fluctuating enstrophy budget are performed to provide insight into the vortical structures that are created by the blunt body of the panel as well as from the heat exchange at the surfaces. Analysis is performed on solar farm arrangements with a variety of module row spacing using large-eddy simulations from the Uintah platform.   

How to cool in the desert: the fluid dynamics of water evaporation from our clothing in hot and dry conditions

About one-third of us live in "drylands" and could have an easier time coping with the more frequent and intensive heat waves by evaporating water from our clothing. In such dry and hot conditions, the near human body fluid dynamics that drive water evaporation are intricate. In particular, air temperature and humidity have a competing effect on air buoyancy and can cause flow stagnation and reversal. In addition, radiative heat from the environment also strongly influences the evaporation process. To understand the intricacies of personal evaporative cooling, I developed a comprehensive model that couples conductive, convective, evaporative, and radiative heat transfer with mass transport in natural or forced laminar flow. Using this model, I will show that the vast majority of water evaporated from our clothing is wasted due to solar radiation and excessive convective losses. I will also show that these issues can be resolved through systematic flow model-driven design of collapsible sun and wind "shade" layer [1].    

Heat transfer enhancement over passive motion inducing surfaces

Biofouling is a main cause of decreased performance in applications that rely on heat transfer. To reduce biofouling, coatings are often applied. However, these coatings typically have low thermal conductivity, resulting in reduced performance. Liquid Infused Surfaces (LIS) have been shown as an interesting option for such coatings, and have also been shown to reduce drag in both laminar and turbulent flows. LIS rely on internal motion within the material to create a mobile interface with the external flow. Here we study this internal flow to enhance convection within the material, which has the potential to yield anti-biofouling without loss of thermal performance. An experimental study is performed based on a numerical study which indicated that these surfaces can increase heat transfer significantly. Various surface modifications and designs are explored that mimic the features of the canonical surfaces used in the numerical study. Drag reduction and heat transfer measurements are performed in a small-scale water setup which allows both laminar and turbulent flows to be studied.    

Mixing it up: Vortex generation as a means to enhance solar panel convection

Lifespan and efficiency of solar photovoltaic (PV) cells plummet with increased cell temperature. Array and module variations such as height, spacing and inclination angle can promote convective cooling via increased turbulence. However, such large-scale renovations prove costly. Thus, more adaptable, inexpensive approaches to solar panel cooling are necessary. This study explores adoption of modular vortex generators (VGs) to enhance surface convection via turbulent boundary layer adhesion. Inspired by the aviation and electronics industries, variations in VG geometry, configuration, and location are examined with respect to PV module cooling. Wind tunnel experiments consider both single and arrayed inclined plates, with inflow velocities ranging from 3.5 to 15 m/s, encompassing a range of conditions from a gentle breeze to high winds. Thermocouple and cold-wire anemometry data capture both module-level cooling and thermal boundary layer behavior, while hotwire and particle image velocimetry data are used to examine momentum boundary layer changes and wake effects due to VG presence. Results from this work can inform an array of industrial applications, while simultaneously giving fundamental insight into VG cooling effects for inclined surfaces in external flows.   

Extreme events of out-of-plane vorticity in Rayleigh-B\’enard convection from PIV experiments

We present a study on extreme events of out-of-plane vorticity in Rayleigh-B\'{e}nard convection from stereoscopic PIV measurements. The range of Rayleigh numbers (Ra) studied goes from low to moderate values (Ra = 1.5\times10^{4}; 2\times10^{4}; 1\times10^{5}; 2\times10^{5}; 5\times10^{5}). The three largest Rayleigh numbers are obtained pressurising the whole set-up including cameras and objective lenses, up to 4.5 bars. The working fluid is air and therefore the Prandtl number of the experiments is 0.7. The measurements are performed in a Rayleigh B\'enard convection cell with aspect ratio 10, and vertical distance between the two horizontal plates of 3 cm.  At these experimental parameters, a transition from Gaussian to intermittent statistics of the velocity derivatives was observed from DNS data only. The main goal of this study is to reproduce DNS data that are acquired at the same Rayleigh numbers to study far-tail events of the out-of-plane vorticity component. We show that the experimental results are able to reproduce the statistics of DNS data of the same flow well, and present a detailed study of extreme events of the out-of-plane vorticity both from their spatial structure and time evolution.   

Controlling supergranule aggregation in convection by weak rotation

Thermal convection in a periodic, horizontally extended layer of aspect ratio $\Gamma = 60$ that is driven by a constant heat flux is analysed by means of high-resolution spectral element simulations of three-dimensional turbulent Rayleigh-B\'enard convection. In this case, a process of gradual aggregation results -- independently of the varying Rayleigh number in a range from Ra $\sim 10^{4}$ to Ra $\sim 10^{7}$ at fixed Prandtl number Pr $= 1$ -- in a domain-filling convection cell.

Here we report the extension of this previous study by the inclusion of (weak) rotation around the vertical axis, limiting the gradual process effectively. Although global statistical measures are mostly unaffected, the final pattern size is governed by the strength of rotation which is characterized by Rossby numbers in the range $\infty \geq$ Ro $\geq 1$. Several series of simulations at different Ra indicate a linear scaling law of the final pattern size with Ro. This implies that pattern formation can be influenced already by weak rotation, similar to the behaviour at the onset of rotating, flux-driven convection.

Our studies might have interesting implications for atmospheric and stellar convection processes where heat fluxes are typically prescribed at the boundaries of the convection zone.   

The influence of system-level design elements on convective cooling in solar farms

When the temperature of solar photovoltaic (PV) modules rises, efficiency drops and module degradation accelerates. Thus, the PV community aims to reduce module operating temperatures. Previous studies of solar farms have illustrated that incoming flow characteristics, turbulent mixing, and array geometry can strongly impact convective cooling, as measured by the convective heat transfer coefficient h. In the fields of heat transfer and vegetated canopy flow, previous work has shown that system-level design elements – e.g., flow diverters, barriers, or windbreaks – can passively alter the flow, enhance turbulent mixing, and influence convection. However, the PV community has not yet explored how such design elements may enhance convective cooling in solar farms. Here, high-resolution large-eddy simulations model the flow and heat transfer through solar farms with system-level design elements. A control volume analysis is then performed to evaluate the net heat flux and compute h, which quantifies the influence of system-level design elements on convective cooling, and thus, module temperature and power output.