Publications

2017
U. Rasthofer and V. Gravemeier, “Recent developments in variational multiscale methods for large-eddy simulation of turbulent flow ,” Archives of Computational Methods in Engineering, vol. 25, pp. 647–690, 2017. Publisher's VersionAbstract

The variational multiscale method is reviewed as a framework for developing computational methods for large-eddy simulation of turbulent flow. In contrast to other articles reviewing this topic, which focused on large-eddy simulation of turbulent incompressible flow, this study covers further aspects of numerically simulating turbulent flow as well as applications beyond incompressible single-phase flow. The various concepts for subgrid-scale modeling within the variational multiscale method for large-eddy simulation proposed by researchers in this field to date are illustrated. These conceptions comprise (i) implicit large-eddy simulation, represented by residual-based and stabilized methods, (ii) functional subgrid-scale modeling via small-scale subgrid-viscosity models and (iii) structural subgrid-scale modeling via the introduction of multifractal subgrid scales. An overview on exemplary numerical test cases to which the reviewed methods have been applied in the past years is provided, including explicit computational results obtained from turbulent channel flow. Wall-layer modeling, passive and active scalar transport as well as developments for large-eddy simulation of turbulent two-phase flow and combustion are discussed to complete this exposition.

J. J. Olsen, C. S. Hemmingsen, L. Bergmann, K. K. Nielsen, S. L. Glimberg, and J. H. Walther, “Characterization and erosion modeling of a nozzle-based inflow-control device,” Society of Petroleum Engineers Drill & Completion, vol. 32, no. 4, pp. 224-233, 2017. Publisher's VersionAbstract

In the petroleum industry, water-and-gas breakthrough in hydrocarbon reservoirs is a common issue that eventually leads to uneconomic production. To extend the economic production lifetime, inflow-control devices (ICDs) are designed to delay the water-and-gas breakthrough. Because the lifetime of a hydrocarbon reservoir commonly exceeds 20 years and it is a harsh environment, the reliability of the ICDs is vital.

With computational fluid dynamics (CFD), an inclined nozzle-based ICD is characterized in terms of the Reynolds number, discharge coefficient, and geometric variations. The analysis shows that especially the nozzle edges affect the ICD flow characteristics. To apply the results, an equation for the discharge coefficient is proposed.

The Lagrangian particle approach is used to further investigate the ICD. This allows for erosion modeling by injecting sand particles into the system. By altering the geometry and modeling several scenarios while analyzing the erosion in the nozzles and at the nozzle edges, an optimized design for incompressible media is found. With a filleted design and an erosion-resistant material, the mean erosion rate in the nozzles may be reduced by a factor of more than 2,500.

A. Andersen, T. Bohr, T. Schnipper, and J. H. Walther, “Wake structure and thrust generation of a flapping foil in two-dimensional flow,” Journal of Fluid Mechanics, vol. 812, 2017. Publisher's VersionAbstract

We present a combined numerical (particle vortex method) and experimental (soap film tunnel) study of a symmetric foil undergoing prescribed oscillations in a two-dimensional free stream. We explore pure pitching and pure heaving, and contrast these two generic types of kinematics. We compare measurements and simulations when the foil is forced with pitching oscillations, and we find a close correspondence between flow visualisations using thickness variations in the soap film and the numerically determined vortex structures. Numerically, we determine wake maps spanned by oscillation frequency and amplitude, and we find qualitatively similar maps for pitching and heaving. We determine the drag–thrust transition for both pitching and heaving numerically, and we discuss it in relation to changes in wake structure. For heaving with low oscillation frequency and high amplitude, we find that the drag–thrust transition occurs in a parameter region with wakes in which two vortex pairs are formed per oscillation period, in contrast to the common transition scenario in regions with inverted von Kármán wakes.

D. Kim, C. Bowman, J. D. T. Bonis-O’Donnell, A. Matzavinos, and D. Stein, “Giant acceleration of DNA diffusion in an array of entropic barriers,” Physical Review Letters, vol. 118, no. 4, 2017. Publisher's VersionAbstract

We investigate with experiments and computer simulations the nonequilibrium dynamics of DNA polymers crossing arrays of entropic barriers in nanofluidic devices in a pressure-driven flow. With increasing driving pressure, the effective diffusivity of DNA rises and then peaks at a value that is many times higher than the equilibrium diffusivity. This is an entropic manifestation of “giant acceleration of diffusion.” The phenomenon is sensitive to the effective energy landscape; thus, it offers a unique probe of entropic barriers in a system driven away from equilibrium.

2016
L. Kulakova, P. Angelikopoulos, P. E. Hadjidoukas, C. Papadimitriou, and P. Koumoutsakos, “Approximate Bayesian Computation for Granular and Molecular Dynamics Simulations,” in Proceedings of the Platform for Advanced Scientific Computing - PASC '16, 2016. Publisher's Version
P. E. Hadjidoukas, et al., “High throughput simulations of two-phase flows on Blue Gene/Q,” in Parallel Computing: On the Road to Exascale – ParCo 2015, 2016, vol. 27, pp. 767–776. Publisher's VersionAbstract
CUBISM-MPCF is a high throughput software for two-phase flow simu- lations that has demonstrated unprecedented performance in terms of floating point operations, memory traffic and storage. The software has been optimized to take advantage of the features of the IBM Blue Gene/Q (BGQ) platform to simulate cav- itation collapse dynamics using up to 13 Trillion computational elements. The per- formance of the software has been shown to reach an unprecedented 14.4 PFLOP/s on 1.6 Million cores corresponding to 72% of the peak on the 20 PFLOP/s Se- quoia supercomputer. It is important to note that, to the best of our knowledge, no flow simulations have ever been reported exceeding 1 Trillion elements and reach- ing more than 1 PFLOP/s or more than 15% of peak. In this work, we first ex- tend CUBISM-MPCF with a more accurate numerical flux and then summarize and evaluate the most important software optimization techniques that allowed us to reach 72% of the theoretical peak performance on BGQ systems. Finally, we show recent simulation results from cloud cavitation comprising 50000 vapor bubbles.
S. Wu, P. Angelikopoulos, G. Tauriello, C. Papadimitriou, and P. Koumoutsakos, “Fusing heterogeneous data for the calibration of molecular dynamics force fields using hierarchical Bayesian models,” J. Chem. Phys. vol. 145, no. 24, pp. 244112, 2016. Publisher's VersionAbstract
We propose a hierarchical Bayesian framework to systematically integrate heterogeneous data for the calibration of force fields in Molecular Dynamics (MD) simulations. Our approach enables the fusion of diverse experimental data sets of the physico-chemical properties of a system at different thermodynamic conditions. We demonstrate the value of this framework for the robust calibration of MD force-fields for water using experimental data of its diffusivity, radial distribution function, and density. In order to address the high computational cost associated with the hierarchical Bayesian models, we develop a novel surrogate model based on the empirical interpolation method. Further computational savings are achieved by implementing a highly parallel transitional Markov chain Monte Carlo technique. The present method bypasses possible subjective weightings of the experimental data in identifying MD force-field parameters.
M. Gazzola, A. A. Tchieu, D. Alexeev, A. de Brauer, and P. Koumoutsakos, “Learning to school in the presence of hydrodynamic interactions,” J. Fluid Mech. vol. 789, pp. 726–749, 2016. Publisher's Version
J. Chen, J. H. Walther, and P. Koumoutsakos, “Ultrafast cooling by covalently bonded graphene-carbon nanotube hybrid immersed in water,” Nanotechnology, vol. 27, no. 46, pp. 465705, 2016. Publisher's VersionAbstract
The increasing power density and the decreasing dimensions of transistors present severe thermal challenges to the design of modern microprocessors. Furthermore, new technologies such as three-dimensional chip-stack architectures require novel cooling solutions for their thermal management. Here, we demonstrate, through transient heat-dissipation simulations, that a covalently bonded graphene-carbon nanotube (G-CNT) hybrid immersed in water is a promising solution for the ultrafast cooling of such high-temperature and high heat-flux surfaces. The G-CNT hybrid offers a unique platform to integrate the superior axial heat transfer capability of individual CNTs via their parallel arrangement. The immersion of the G-CNT in water enables an additional heat dissipation path via the solid–liquid interaction, allowing for the sustainable cooling of the hot surface under a constant power input of up to 10 000 W cm-2.
C. Wen, Y. Yang, J. H. Walther, K. M. Pang, and Y. Feng, “Effect of delta wing on the particle flow in a novel gas supersonic separator,” Powder Technology, vol. 304, pp. 261–267, 2016. Publisher's VersionAbstract
The present work presents numerical simulations of the complex particle motion in a supersonic separator with a delta wing located in the supersonic flow. The effect of the delta wing on the strong swirling flow is analysed using the Discrete Particle Method. The results show that the delta wings re-compress the upstream flow and the gas Mach number decreases correspondingly. However, the Mach number does not vary significantly from the small, medium and large delta wing configurations. The small delta wing generates a swirl near its surface, but has minor influences on the flow above it. On the contrary, the use of the large delta wing produces a strong swirling flow in the whole downstream region. For the large delta wing, the collection efficiency reaches 70% with 2 μm particles, indicating a good separation performance of the proposed supersonic separator.
C. S. Hemmingsen, K. M. Ingvorsen, S. Mayer, and J. H. Walther, “LES and URANS simulations of the swirling flow in a dynamic model of a uniflow-scavenged cylinder,” International Journal of Heat and Fluid Flow, vol. 62, pp. 213–223, 2016. Publisher's VersionAbstract
The turbulent swirling flow in a uniflow-scavenged two-stroke engine cylinder is investigated using computational fluid dynamics. The investigation is based on the flow in a scale model with a moving piston. Two numerical approaches are tested; a large eddy simulation (LES) approach with the wall-adaptive local eddy-viscosity (WALE) model and a Reynolds-Averaged Navier-Stokes approach using the k - w Shear-Stress Transport model. Combustion and compression are neglected. The simulations are verified by a sensitivity study and the performance of the turbulence models are evaluated by comparison with experimental results. Both turbulence models produce results in good agreement with experimental data. The agreement is particularly good for the LES, immediately after the piston passes the bottom dead center. Furthermore, in the piston standstill period, the LES predicts a tangential profile in agreement with the measurements, whereas the k - w SST model predicts a solid body rotation. Several instabilities are identified during the scavenging process. The formation of a vortex breakdown with multiple helical vortex structures are observed after the scavenge port opening, along with the shedding of vortex rings with superimposed swirl. The turbulence models predict several flow reversals in the vortex breakdown region through the scavenge process. Flow separations in the scavenge ports lead to a secondary axial flow, in the separated region. The secondary flow exits in the top of the scavenge ports, resulting in large velocity gradients near the cylinder liner above the scavenge ports.
M. M. Hejlesen and J. H. Walther, “A multiresolution method for solving the poisson equation using high order regularization,” Journal of Computational Physics, vol. 326, pp. 188–196, 2016. Publisher's VersionAbstract

We present a novel high order multiresolution Poisson solver based on regularized Green's function solutions to obtain exact free-space boundary conditions while using fast Fourier transforms for computational efficiency. Multiresolution is a achieved through local refinement patches and regularized Green's functions corresponding to the difference in the spatial resolution between the patches. The full solution is obtained utilizing the linearity of the Poisson equation enabling super-position of solutions. We show that the multiresolution Poisson solver produces convergence rates that correspond to the regularization order of the derived Green's functions.

E. Hovad, J. Spangenberg, P. Larsen, J. H. Walther, J. Thorborg, and J. H. Hattel, “Simulating the DISAMATIC process using the discrete element method — a dynamical study of granular flow,” Powder Technology, vol. 303, pp. 228–240, 2016. Publisher's VersionAbstract
The discrete element method (DEM) is applied to simulate the dynamics of the flow of green sand while filling a mould using the DISAMATIC process. The focus is to identify relevant physical experiments that can be used to characterize the material properties of green sand in the numerical model. The DEM parameters describing the static friction coefficients are obtained using a ring shear tester and the rolling resistance and cohesion value is subsequently calibrated with a sand pile experiment. The calibrated DEM model is used to model the sand shot in the DISAMATIC process for three different sand particle flow rates as captured on the corresponding video footage of the interior of the chamber. A mould chamber with three ribs mounted on the fixed pattern plate forming four cavities is chosen as a reference geometry to investigate the conditions found in the real moulding process. The geometry of the cast part and the casting system can make the moulding process complicated due to obstacles such as ribs that deflect the sand flow causing “shadows effects” around the cavities of the mould. These dynamic effects are investigated by the qualitative flow dynamics and quantitative mould filling times captured in the video footage and simulated by the calibrated DEM model. Both two- and three-dimensional DEM models are considered and found to produce results in good agreements with the video footage of the DISAMATIC process.
N. K. Karna, E. Oyarzua, J. H. Walther, and H. A. Zambrano, “Effect of the meniscus contact angle during early regimes of spontaneous imbibition in nanochannels,” Physical Chemistry Chemical Physics, vol. 18, no. 47, pp. 31997–32001, 2016. Publisher's VersionAbstract

Nanoscale capillarity has been extensively investigated; nevertheless, many fundamental questions remain open. In spontaneous imbibition, the classical Lucas–Washburn equation predicts a singularity as the fluid enters the channel consisting of an anomalous infinite velocity of the capillary meniscus. Bosanquet's equation overcomes this problem by taking into account fluid inertia predicting an initial imbibition regime with constant velocity. Nevertheless, the initial constant velocity as predicted by Bosanquet's equation is much greater than those observed experimentally. In the present study, large scale atomistic simulations are conducted to investigate capillary imbibition of water in slit silica nanochannels with heights between 4 and 18 nm. We find that the meniscus contact angle remains constant during the inertial regime and its value depends on the height of the channel. We also find that the meniscus velocity computed at the channel entrance is related to the particular value of the meniscus contact angle. Moreover, during the subsequent visco-inertial regime, as the influence of viscosity increases, the meniscus contact angle is found to be time dependent for all the channels under study. Furthermore, we propose an expression for the time evolution of the dynamic contact angle in nanochannels which, when incorporated into Bosanquet's equation, satisfactorily explains the initial capillary rise.

C. Wen, A. Li, J. H. Walther, and Y. Yang, “Effect of swirling device on flow behavior in a supersonic separator for natural gas dehydration,” Separation and Purification Technology, vol. 168, pp. 68–73, 2016. Publisher's VersionAbstract
The supersonic separator is a revolutionary device to remove the condensable components from gas mixtures. One of the key issues for this novel technology is the complex supersonic swirling flow that is not well understood. A swirling device composed of an ellipsoid and several helical blades is designed for an annular supersonic separator. The supersonic swirling separation flow of natural gas is calculated using the Reynolds Stress model. The results show that the viscous heating and strong swirling flow cause the adverse pressure in the annular channel, which may negatively affect the separation performance. When the swirling flow passes through the annular nozzle, it will damage the expansion characteristics of the annular nozzle. The blade angles and numbers are both optimized by evaluating the swirling and expansion effects for the supersonic separation.
M. Zayernouri and A. Matzavinos, “Fractional adams–bashforth/moulton methods: an application to the fractional keller–segel chemotaxis system,” Journal of Computational Physics, vol. 317, pp. 1–14, 2016. Publisher's VersionAbstract
We first formulate a fractional class of explicit Adams–Bashforth (A-B) and implicit Adams–Moulton (A-M) methods of first- and second-order accuracy for the time-integration of Dtτ0Cu(x,t)=g(t;u), τ∈(0,1], where Dtτ0C denotes the fractional derivative in the Caputo sense. In this fractional setting and in contrast to the standard Adams methods, an extra history load term emerges and the associated weight coefficients are τ-dependent. However when τ=1, the developed schemes reduce to the well-known A-B and A-M methods with standard coefficients. Hence, in terms of scientific computing, our approach constitutes a minimal modification of the existing Adams libraries. Next, we develop an implicit–explicit (IMEX) splitting scheme for linear and nonlinear fractional PDEs of a general advection–reaction–diffusion type, and we apply our scheme to the time–space fractional Keller–Segel chemotaxis system. In this context, we evaluate the nonlinear advection term explicitly, employing the fractional A-B method in the prediction step, and we treat the corresponding diffusion term implicitly in the correction step using the fractional A-M scheme. Moreover, we perform the corresponding spatial discretization by employing an efficient and spectrally-accurate fractional spectral collocation method. Our numerical experiments exhibit the efficiency of the proposed IMEX scheme in solving nonlinear fractional PDEs.
F. Wermelinger, B. Hejazialhosseini, P. Hadjidoukas, D. Rossinelli, and P. Koumoutsakos, “An Efficient Compressible Multicomponent Flow Solver for Heterogeneous CPU/GPU Architectures,” in Proceedings of the Platform for Advanced Scientific Computing - PASC \textquotesingle16, 2016. Publisher's VersionAbstract
We present a solver for three-dimensional compressible multicomponent flow based on the compressible Euler equations. The solver is based on a finite volume scheme for structured grids and advances the solution using an explicit Runge-Kutta time stepper. The numerical scheme requires the computation of the flux divergence based on an approximate Riemann problem. The computation of the divergence quantity is the most expensive task in the algorithm. Our implementation takes advantage of the compute capabilities of heterogeneous CPU/GPU architectures. The computational problem is organized in subdomains small enough to be placed into the GPU memory. The compute intensive stencil scheme is offloaded to the GPU accelerator while advancing the solution in time on the CPU. Our method to implement the stencil scheme on the GPU is not limited to applications in fluid dynamics. The performance of our solver was assessed on Piz Daint, a XC30 supercomputer at CSCS. The GPU code is memory-bound and achieves a per-node performance of 462 Gflop/s, outperforming by 3.2× the multicore- based Gordon Bell winning cubism-mpcf solver [16] for the offloaded computation on the same platform. The focus of this work is on the per-node performance of the heterogeneous solver. In addition, we examine the performance of the solver across 4096 compute nodes. We present simulations for the shock-induced collapse of an aligned row of air bubbles submerged in water using 4 billion cells. Results show a final pressure amplification that is 100× stronger than the strength of the initial shock.
An Efficient Compressible Multicomponent Flow Solver for Heterogeneous CPU/GPU Architectures
S. Mishra, C. Schwab, and J. Šukys, “Multi-level Monte Carlo finite volume methods for uncertainty quantification of acoustic wave propagation in random heterogeneous layered medium,” Journal of Computational Physics, vol. 312, pp. 192–217, 2016. Publisher's VersionAbstract

We consider the very challenging problem of efficient uncertainty quantification for acoustic wave propagation in a highly heterogeneous, possibly layered, random medium, characterized by possibly anisotropic, piecewise log-exponentially distributed Gaussian random fields. A multi-level Monte Carlo finite volume method is proposed, along with a novel, bias-free upscaling technique that allows to represent the input random fields, generated using spectral FFT methods, efficiently. Combined together with a recently developed dynamic load balancing algorithm that scales to massively parallel computing architectures, the proposed method is able to robustly compute uncertainty for highly realistic random subsurface formations that can contain a very high number (millions) of sources of uncertainty. Numerical experiments, in both two and three space dimensions, illustrating the efficiency of the method are presented.

K. M. Pang, N. Karvounis, J. H. Walther, and J. Schramm, “Numerical investigation of soot formation and oxidation processes under large two-stroke marine diesel engine-like conditions using integrated CFD-chemical kinetics,” Applied Energy, vol. 169, pp. 874–887, 2016. Publisher's VersionAbstract
In this reported work, multi-dimensional computational fluid dynamics studies of diesel combustion and soot formation processes in a constant volume combustion chamber and a marine diesel engine are carried out. The key interest here is firstly to validate the coupling of a newly developed skeletal n-heptane mechanism and a revised multi-step soot model using laser extinction measurements of diesel soot obtained at different ambient pressure levels in an optical accessible, constant volume chamber experiment. It is revealed that ignition delay times and liftoff lengths generated using the new skeletal model are close to those produced by the larger and more comprehensive chemical mechanisms, apart from those at the low pressure condition. The current study also demonstrates that the variation of averaged soot volume fraction with respect to the change of combustion chamber pressure captured using the revised soot model agrees reasonably well with the measurements in terms of peak values. The numerical model is subsequently applied to investigate the flame development, soot/nitrogen monoxide formation and heat transfer in a two-stroke, low-speed uniflow-scavenged marine diesel engine operating at full load condition, where optical measurements are not available. Comparisons to the experimental data show that the simulated pressure rise starts 1.0 crank angle degree in advance and the calculated peak pressure is 1.7% lower. The associated flame liftoff length is negligible, yielding higher local equivalence ratio and soot volume fraction values as compared to those under similar test condition in the constant volume chamber. With the use of the revised model, the total heat transfer to the walls calculated when soot radiative heat loss is taken into account is approximately 30% higher compared to that when only convective heat loss is considered. The averaged nitrogen monoxide concentration is 7.7% lower when both convective and soot radiative heat losses are accounted for but the net soot mass production is less sensitive to soot radiation. A sensitivity study reveals that neither increasing nor decreasing the soot absorption coefficient by 30% from the baseline setup is influential to nitrogen monoxide formation, soot mass production and heat transfer. The findings here aid to gain insights and provide a better understanding of the combustion and soot processes in large, uniflow-scavenged marine engines. The numerical model developed in this work can also be applied to explore different phenomena in this combustion system.
B. O. Andersen, N. F. Nielsen, and J. H. Walther, “Numerical and experimental study of pulse-jet cleaning in fabric filters,” Powder Technology, vol. 291, pp. 284–298, 2016. Publisher's VersionAbstract
Pulse-jet cleaning and understanding of the complex physics are essential when designing fabric filters used for air pollution control. Today, low-pressure cleaning is of particular interest due to demand for reduced compressed air consumption. Pulse-jet cleaned fabric filters have been studied for many years by experimental investigation and to a limited extent by Computational Fluid Dynamics (CFD). The majority of the studies have focused on high-pressure cleaning systems, and the CFD models presented are so far two-dimensional (2D). In the work presented here, pulse-jet cleaning of low-pressure fabric filters (2 bar) is studied using a full three-dimensional (3D) CFD model. Experimental results obtained in a pilot-scale test filter with 28 bags, in length of 10 m and in general full-scale dimensions of the cleaning system are used to verify the reliability of the present CFD model. The validated CFD model reveals the strong compressible effects, a highly transient behaviour, the formation of compressible vortex rings and the shock cell phenomenon within the overexpanded supersonic jet. The cleaning nozzles and venturi design aid or oppose the pulse-pressure within the bags, and this plays an important role in the resulting efficiency of removing the dust layer from the bags. The CFD simulation shows that the traditional straight-bore nozzles provide substantial misalignment of the jet, and the add-on nozzle design offers only limited improvement. Further, the need for venturis in low-pressure filters and the importance of optimising the venturi design are demonstrated. The working principle of the venturi is to restrict backflow which is detrimental to the pressure rise in the bags. Reducing the venturi throat diameter is shown to reduce backflow and improve the pulse-pressure.

Pages