The quantification of the prediction accuracy in large eddy simulations (LES) is very challenging due to various interacting errors associated with this approach. When dealing with errors in LES using implicit filtering, numerical and modeling errors have drawn the interest of many researchers. Little attention has been paid to other sources of discrepancies between LES results and reference data, namely sampling errors, influence of the initial conditions, improper boundary conditions or uncertainties issuing from reference data. A framework of metrics that includes all these issues is addressed in the present paper to study subgrid-scale (SGS) models for LES and to quantify their prediction accuracy and computational costs. The method is applied to a simple wall-bounded turbulent flow at moderate Reynolds number. It turns out from the results obtained with six commonly used SGS models that wall-adapting models (WALE and SIGMA) and localized dynamic models reproduce the physics of the flow field more faithfully, reveal a superior prediction accuracy and have a similar computational cost than models using van Driest wall damping. Especially at the viscous wall region (r + < 50), wall-adapting and localized dynamic models are more accurate, reflecting the proper near wall behavior of such models. Relying on the analysis of sources of various errors, uncertainties in LES are estimated and systematically assessed, and their influence on simulation results is quantified. Finally, engineering estimations of the required averaging time to obtain basic estimates of statistical quantities with a predetermined degree of accuracy are suggested.

In this work, entropy generation analysis is applied to characterize and optimize a turbulent impinging jet on a heated solid surface. In particular, the influence of plate inclinations and Reynolds numbers on the turbulent heat and fluid flow properties and its impact on the thermodynamic performance of such flow arrangements are numerically investigated. For this purpose, novel model equations are derived in the frame of Large Eddy Simulation (LES) that allows calculation of local entropy generation rates in a post-processing phase including the effect of unresolved subgrid-scale irreversibilities. From this LES-based study, distinctive features of heat and flow dynamics of the impinging fluid are detected and optimal operating designs for jet impingement cooling are identified. It turned out that (1) the location of the stagnation point and that of the maximal Nusselt number differ in the case of plate inclination; (2) predominantly the impinged wall acts as a strong source of irreversibility; and (3) a flow arrangement with a jet impinging normally on the heated surface allows the most efficient use of energy which is associated with lowest exergy lost. Furthermore, it is found that increasing the Reynolds number intensifies the heat transfer and upgrades the second law efficiency of such thermal systems. Thereby, the thermal efficiency enhancement can overwhelm the frictional exergy loss.

In this work, near-wall thermal transport processes and entropy generation mechanisms in a turbulent jet impinging on a 45 ∘ -inclined heated surface are investigated using a direct numerical simulation (DNS). The objectives are to analyze the subtle mechanisms of heat transport in the vicinity of an inclined impinged wall, to determine the causes of irreversibilities that are responsible for the reduction of performance of impingement cooling applications and to provide a comprehensive dataset for model development and validation. Results for near-wall thermal characteristics including heat fluxes are analyzed. An entropy production map is provided from the second law analysis. The following main outcomes can be drawn from this study: (1) the location of peak heat transfer occurs not directly at the stagnation point; instead, it is slightly shifted towards the compression side of the jet, while at this region, the heat is transported counter to the temperature gradient; (2) turbulent thermal and fluid flow transport processes around the stagnation point are considerably different from those found in other near-wall-dominated flows and are strongly non-equilibrium in nature; (3) heat fluxes appear highly anisotropic especially in the vicinity of the impinged wall; (4) in particular, the heated wall acts as a strong source of irreversibility for both entropy production related to viscous dissipation and to heat conduction. All these findings imply that a careful design of the impinged plate is particularly important in order to use energy in such a thermal arrangement effectively. Finally, this study confirms that the estimation of the turbulent part of the entropy production based on turbulence dissipation rates in non-reacting, non-isothermal fluid flows represents a reliable approximate approach within the second law analysis, likewise in the context of computationally less expensive simulation techniques like RANS and/or LES.

No abstract

Dense matrix inversion is a basic procedure in many linear algebra algorithms. A computationally arduous step in most dense matrix inversion methods is the inversion of triangular matrices as produced by factorization methods such as LU decomposition. In this paper, we demonstrate how triangular matrix inversion (TMI) can be accelerated considerably by using commercial Graphics Processing Units (GPU) in a standard PC. Our implementation is based on a divide and conquer type recursive TMI algorithm, efficiently adapted to the GPU architecture. Our implementation obtains a speedup of 34x versus a CPU-based LAPACK reference routine, and runs at up to 54 gigaflops/s on a GTX 280 in double precision. Limitations of the algorithm are discussed, and strategies to cope with them are introduced. In addition, we show how inversion of an L-and U-matrix can be performed concurrently on a GTX 295 based dual-GPU system at up to 90 gigaflops/s.

Abstract:In the present paper, direct numerical simulation (DNS) and particle image velocimetry (PIV) have been applied complementarily in order to generate a database of near-wall turbulence properties of a highly turbulent jet impinging on a solid surface under different inclination angles. Thereby, the main focus is placed on an impingement angle of 45 • , since it represents a good generic benchmark test case for a wide range of technical fluid flow applications. This specific configuration features very complex flow properties including the presence of a stagnation point, development of the shear boundary layer and strong streamline curvature. In particular, this database includes near-wall turbulence statistics along with mean and rms velocities, budget terms in the turbulent kinetic energy equation, anisotropy invariant maps, turbulent length/time scales and near-wall shear stresses. These properties are useful for the validation of near-wall modeling approaches in the context of Reynolds-averaged Navier-Stokes (RANS) and large-eddy simulations (LES). From this study, in which further impingement angles (0 • , 90 • ) have been considered in the experiments only, it turns out that (1) the production of turbulent kinetic energy appears negative at the stagnation point for an impingement angle other than 0 • and is balanced predominantly by pressure-related diffusion, (2) quasi-coherent thin streaks with large characteristic time scales appear at the stagnation region, while the organization of the flow is predominantly toroidal further downstream, and (3) near-wall shear stresses are low at the stagnation region and intense in regions where the direction of the flow changes suddenly.

No abstract

A second law analysis in combustion systems is performed along with an exergy loss study by quantifying the entropy generation sources using, for the first time, three different approaches: a classical-thermodynamics-based approach, a novel turbulence-based method and a look-up-table-based approach, respectively. The numerical computation is based on a hybrid filtered Eulerian stochastic field (ESF) method coupled with tabulated detailed chemistry according to a Famelet-Generated Manifold (FGM)-based combustion model. In this work, the capability of the three approaches to capture the effect of the Re number on local exergy losses is especially appraised. For this purpose, Sandia flames D and E are selected as application cases. First, the validation of the computed flow and scalar fields is achieved by comparison to available experimental data. For both flames, the flow field results for eight stochastic fields and the associated scalar fields show an excellent agreement. The ESF method reproduces all major features of the flames at a lower numerical cost. Next, the second law analysis carried out with the different approaches for the entropy generation computation provides comparable quantitative results. Using flame D as a reference, for which some results with the thermodynamic-based approach exist in the literature, it turns out that, among the sources of exergy loss, the heat transfer and the chemical reaction emerge notably as the main culprits for entropy production, causing 50% and 35% of it, respectively. This fact-finding increases in Sandia flame E, which features a high Re number compared to Sandia flame D. The computational cost is less once the entropy generation analysis is carried out by using the Large Eddy Simulation (LES) hybrid ESF/FGM approach together with the look-up-table-based or turbulence-based approach.