Publications

Numerical simulations using large eddy simulation (LES) and Unsteady Reynolds Averaged Navier–Stokes (URANS) are carried out to identify the underlying mechanisms that govern the early soot evolution process in an n-dodecane spray flame at 21% O2 by molar concentration. A two-equation phenomenological soot model is used here to simulate soot formation and oxidation. Both ignition delay time (IDT) and lift-off length (LOL) are found to agree with experimental measurements. The transient evolution of soot mass, in particularly the soot spike phenomenon, is captured in the present LES cases, but not in the URANS cases. Hence, a comparison of numerical results from LES and URANS simulations is conducted to provide a better insight of this phenomenon. LES is able to predict the rapid increasing soot mass during the early stage of soot formation due to having a large favorable region of equivalence ratio (ϕ > 1.5) and temperature (T > 1800 K) for soot formation. This favorable region increases and then decreases to reach a quasi-steady state in the LES case, while it continues to increase in the URANS simulation during the early time. In addition, the soot spike is a consequence of the competition between soot formation and oxidation rates. The time instance when the total soot mass reaches peak value coincides with the time instance when the total mass of soot precursor reaches a plateau. The soot spike is formed due to the continuous increase of oxidizing species in the LES case which leads to a more dominant oxidation process than the formation process.

Railway ballast affected by heavy cyclic loading degrades and spreads resulting in an uncomfortable transportation caused by undesirable vibrations. Restoring a well sorted track ballast can be expensive. This paper analyzes track ballast deformation using the Discrete Element Method (DEM). The simulations are performed using the STAR-CCM+ software in a three-dimensional domain. Four track ballast models are studied. The first two models describe the ballast as spheres with and without rolling resistance, respectively. The third model uses a clump model that allows breaking of the ballast, whereas the fourth model describes the ballast as composite particles generated from 3D-scanned ballast stones. The sleepers and rails are modelled as DEM particles. As a supplement to the study of different ballast models, the influence of variation in the loading profile is investigated. The largest obtained deformation is observed in the ballast modelled as spheres and the smallest deformation in the ballast modelled from the 3D scanned ballast stones. The results highlight the importance of describing the ballast as non-spherical geometries.

Computer-Aided Engineering (CAE) has supported the industry in its transition from trial-and-error towards physics-based modelling, but our ways of treating and exploiting the simulation results have changed little during this period. Indeed, the business model of CAE centers almost exclusively around delivering base-case simulation results with a few additional operational conditions. In this contribution, we introduce a new paradigm for the exploitation of computational physics data, consisting in using machine learning to enlarge the simulation databases in order to cover a wider spectrum of operational conditions and provide quick response directly on field. The resulting product from this hybrid physics-informed and data-driven modelling is referred to as Simulation Digital Twin (SDT). While the paradigm can be equally used in different CAE applications, in this paper we address its implementation in the context of Computational Fluid Dynamics (CFD). We show that the generation of Simulation Digital Twins can be efficiently accomplished with the combination of the CFD tool TransAT and the data analytics platform eDAP.

Numerical simulations are performed to investigate the effects of ambient density (ρam) and nozzle diameter (Dnoz) on the ignition characteristic of diesel spray combustion under engine-like conditions. A total of nine cases which consist of different ρam of 14.8, 30.0, and 58.5 kg/m3 and different Dnoz of 100, 180, and 363μm are considered. The results show that the predicted ignition delay times are in good agreement with measurements. The current results show that the mixture at the spray central region becomes more fuel-rich as Dnoz increases. This leads to a shift in the high-temperature ignition location from the spray tip towards the spray periphery as Dnoz increases at ρam of 14.8 kg/m3 . At higher ρam of 30.0 and 58.5 kg/m3 , the ignition locations for all Dnoz cases occur at the spray periphery due to shorter ignition timing and the overly fuel-rich spray central region. The numerical results show that the first ignition location during the high-temperature ignition occurs at the fuel-rich region at ρam⩽30.0 kg/m3 across different Dnoz. At ρam = 58.5 kg/m3 , the ignition occurs at the fuel-lean region for the 100 and 180μm cases, but at the fuel-rich region for the 363μm nozzle case. This distinctive difference in the result at 58.5 kg/m3 is likely due to the relatively longer ignition delay time in the 363μm nozzle case. Furthermore, the longer ignition delay time as Dnoz increases can be related to the higher local scalar dissipation rate in the large nozzle case.

This paper is associated with a video winner of a 2019 American Physical Society's Division of Fluid Dynamics (DFD) Gallery of Fluid Motion Award for work presented at the DFD Gallery of Fluid Motion. The original video is available online at the Gallery of Fluid Motion, https://doi.org/10.1103/APS.DFD.2019.GFM.V0018.

We present the high performance implementation of a new algorithm for simulating multiphase flows with bubbles and drops that do not coalesce. The algorithm is more efficient than the standard multi-marker volume-of-fluid method since the number of required fields does not depend on the number of bubbles. The capabilities of our methods are demonstrated on simulations of a foaming waterfall where we analyze the effects of coalescence prevention on the bubble size distribution and show how rising bubbles cluster up as foam on the water surface. Our open-source implementation enables high throughput simulations of multiphase flow, supports distributed as well as hybrid execution modes and scales efficiently on large compute systems.

The systematic identification of infected individuals is critical for the containment of the COVID-19 pandemic. Currently, the spread of the disease is mostly quantified by the reported numbers of infections, hospitalisations, recoveries and deaths; these quantities inform epidemiology models that provide forecasts for the spread of the epidemic and guide policy making. The veracity of these forecasts depends on the discrepancy between the numbers of reported, and unreported yet infectious, individuals. We combine Bayesian experimental design with an epidemiology model and propose a methodology for the optimal allocation of limited testing resources in space and time, which maximises the information gain for such unreported infections. The proposed approach is applicable at the onset and spread of the epidemic and can forewarn of a possible recurrence of the disease after relaxation of interventions. We examine its application in Switzerland; the open source software is, however, readily adaptable to countries around the world. We find that following the proposed methodology can lead to vastly less uncertain predictions for the spread of the disease, thus improving estimates of the effective reproduction number and the future number of unreported infections. This information can provide timely and systematic guidance for the effective identification of infectious individuals and for decision-making regarding lockdown measures and the distribution of vaccines.

We examine the efficiency of Recurrent Neural Networks in forecasting the spatiotemporal dynamics of high dimensional and reduced order complex systems using Reservoir Computing (RC) and Backpropagation through time (BPTT) for gated network architectures. We highlight advantages and limitations of each method and discuss their implementation for parallel computing architectures. We quantify the relative prediction accuracy of these algorithms for the long-term forecasting of chaotic systems using as benchmarks the Lorenz-96 and the Kuramoto–Sivashinsky (KS) equations. We find that, when the full state dynamics are available for training, RC outperforms BPTT approaches in terms of predictive performance and in capturing of the long-term statistics, while at the same time requiring much less training time. However, in the case of reduced order data, large scale RC models can be unstable and more likely than the BPTT algorithms to diverge. In contrast, RNNs trained via BPTT show superior forecasting abilities and capture well the dynamics of reduced order systems. Furthermore, the present study quantifies for the first time the Lyapunov Spectrum of the KS equation with BPTT, achieving similar accuracy as RC. This study establishes that RNNs are a potent computational framework for the learning and forecasting of complex spatiotemporal systems.

We present data driven kinematic models for the motion of bubbles in high-Re turbulent fluid flows based on recurrent neural networks with long-short term memory enhancements. The models extend empirical relations, such as Maxey-Riley (MR) and its variants, whose applicability is limited when either the bubble size is large or the flow is very complex. The recurrent neural networks are trained on the trajectories of bubbles obtained by Direct Numerical Simulations (DNS) of the Navier Stokes equations for a two-component incompressible flow model. Long short term memory components exploit the time history of the flow field that the bubbles have encountered along their trajectories and the networks are further augmented by imposing rotational invariance to their structure. We first train and validate the formulated model using DNS data for a turbulent Taylor-Green vortex. Then we examine the model pre- dictive capabilities and its generalization to Reynolds numbers that are different from those of the train- ing data on benchmark problems, including a steady (Hill’s spherical vortex) and an unsteady (Gaussian vortex ring) flow field. We find that the predictions of the developed model are significantly improved compared with those obtained by the MR equation. Our results indicate that data-driven models with history terms are well suited in capturing the trajectories of bubbles in turbulent flows.

We present a particle method for estimating the curvature of interfaces in volume-of-fluid simulations of multiphase flows. The method is well suited for under-resolved interfaces, and it is shown to be more accurate than the parabolic fitting that is employed in such cases. The curvature is computed from the equilibrium positions of particles constrained to circular arcs and attracted to the interface. The pro- posed particle method is combined with the method of height functions at higher resolutions, and it is shown to outperform the current combinations of height functions and parabolic fitting. The algorithm is conceptually simple and straightforward to implement on new and existing software frameworks for multiphase flow simulations thus enhancing their capabilities in challenging flow problems. We evaluate the proposed hybrid method on a number of two- and three-dimensional benchmark flow problems and illustrate its capabilities on simulations of flows involving bubble coalescence and turbulent multiphase flows.

Fish schooling implies an awareness of the swimmers for their companions. In flow mediated environments, in addition to visual cues, pressure and shear sensors on the fish body are critical for providing quantitative information that assists the quantification of proximity to other fish. Here we examine the distribution of sensors on the surface of an artificial swimmer so that it can optimally identify a leading group of swimmers. We employ Bayesian experimental design coupled with numerical simulations of the two-dimensional Navier Stokes equations for multiple self-propelled swimmers. The follower tracks the school using information from its own surface pressure and shear stress. We demonstrate that the optimal sensor distribution of the follower is qualitatively similar to the distribution of neuromasts on fish. Our results show that it is possible to identify accurately the center of mass and the number of the leading swimmers using surface only information.

Sponges are suspension feeders that filter vast amounts of water. Pumping is carried out by flagellated chambers that are connected to an inhalant and exhalant canal system. In ‘leucon’ sponges with relatively high-pressure resistance due to a complex and narrow canal system, pumping and filtering are only possible owing to the presence of a gasket-like structure (forming a canopy above the collar filters). Here, we combine numerical and experimental work and demonstrate how sponges that lack such sealing elements are able to efficiently pump and force the flagella-driven flow through their collar filter, thanks to the formation of a ‘hydrodynamic gasket’ above the collar. Our findings link the architecture of flagellated chambers to that of the canal system, and lend support to the current view that the sponge aquiferous system evolved from an open-type filtration system, and that the first metazoans were filter feeders.

The present study examines the quantity of surface wetting in a two-phase oil and water pipe flow. The study is performed by employing an Eulerian-Eulerian computational fluid dynamics model using the S-gamma droplet size distribution model within STAR-CCM+. In the North Sea, production of oil and gas, water-phase surface processes such as scale and corrosion account for 40-50% of operating expenses. The objective of the study is to investigate best practices for the prediction of phase distribution aimed at evaluating the degree of the wall in contact with the water phase (water-wetting). The model is validated by performing detailed numerical simulations corresponding to the experimental studies by Kumara, Halvorsen, and Melaaen (Meas. Sci. Technol., vol. 20, p. 114004, 2009). The comparison yields good agreement with the observed measurements with slight deviations in the predicted dispersion rate but accurate prediction of the liquid holdup. Comparison of droplet sizes to those observed in experiments by Elseth (PhD, Telemark University College, 2001) indicates that tuning of the S-gamma model is necessary to provide accurate droplet size predictions. The surface wetting is then evaluated with its interdependence with liquid holdup and dispersion rate. Increase in the dispersion with a decrease in the Richardson number is observed in agreement with stability analysis of the Kelvin-Helmholtz instability.

A prediction method, known as the Coupled Time Scale (CTS) method, is proposed in the current work to estimate the ignition delay time (IDT) of liquid spray combustion by only performing an inert spray simulation and a zero-dimensional (0-D) homogeneous reactor (HR) simulation. The method is built upon the assumption that if the majority of the vapor regions in a spray has a composition close to the most reactive mixture fraction, which can be obtained by performing 0-D HR calculations, these regions will then have a high probability to undergo high-temperature ignition in the spray. The proposed method is applied to estimate the high-temperature IDT of n-dodecane sprays. Two nozzle diameters (Dnoz) of 90 μm and 186 μm which correspond to Spray A and Spray D in the Engine Combustion Network [1] respectively, are considered. Both Dnoz are tested at three ambient temperatures (Tam) of 800 K, 900 K, and 1000 K. The fidelity of the proposed CTS method is verified by comparing the predicted IDT against CFD simulated IDT and measured IDT. Comparison of the estimated IDT from the CTS method to the measured IDT yields a maximum relative difference of 24%. Meanwhile, a maximum relative difference of 33% is found between the IDTs computed from the CTS method and the large eddy simulations of the associated reacting sprays across the different Tam,Dnoz, and chemical mechanisms considered in this study.

In the present study, large-eddy simulations (LES) are used to identify the underlying mechanism that governs the ignition phenomena of spray flames from different nozzle diameters when the ambient temperature (T_am) varies. Two nozzle sizes of 90µm and 186µm are chosen. They correspond to the nozzle sizes used by Spray A and Spray D, respectively, in the Engine Combustion Network. LES studies of both nozzles are performed at three T_am of 800K, 900K, and 1000K. The numerical models are validated using the experimental liquid and vapor penetration, mixture fraction (Z) distribution, as well as ignition delay time (IDT). The ignition characteristics of both Spray A and Spray D are well predicted, with a maximum relative difference of 14% as compared to the experiments. The simulations also predict the annular ignition sites for Spray D at T_am ⩾ 900K, which is consistent with the experimental observation. It is found that the mixture with Z ⩽ 0.2 at the spray periphery is more favorable for ignition to occur than the overly fuel-rich mixture of Z > 0.2 formed in the core of spray. This leads to the annular ignition sites at higher T_am. Significantly longer IDT for Spray D is obtained at T_am of 800K due to higher scalar dissipation rates (χ) during high temperature (HT) ignition. The maximum χ during HT ignition for Spray D is larger than that in Spray A by approximately a factor of 5. In contrast, at Tam=1000K, the χ values are similar between Spray A and Spray D. This elucidates the increase in the difference of IDT between Spray D and Spray A as T_am decreases. This may explain the contradicting findings on the effects of nozzle diameters on IDT from literature.

In the present study, large-eddy simulations (LES) are used to identify the underlying mechanism that governs the ignition phenomena of spray flames from different nozzle diameters when the ambient temperature (T_am) varies. Two nozzle sizes of 90µm and 186µm are chosen. They correspond to the nozzle sizes used by Spray A and Spray D, respectively, in the Engine Combustion Network. LES studies of both nozzles are performed at three T_am of 800K, 900K, and 1000K. The numerical models are validated using the experimental liquid and vapor penetration, mixture fraction (Z) distribution, as well as ignition delay time (IDT). The ignition characteristics of both Spray A and Spray D are well predicted, with a maximum relative difference of 14% as compared to the experiments. The simulations also predict the annular ignition sites for Spray D at T_am ⩾ 900K, which is consistent with the experimental observation. It is found that the mixture with Z ⩽ 0.2 at the spray periphery is more favorable for ignition to occur than the overly fuel-rich mixture of Z > 0.2 formed in the core of spray. This leads to the annular ignition sites at higher T_am. Significantly longer IDT for Spray D is obtained at T_am of 800K due to higher scalar dissipation rates (χ) during high temperature (HT) ignition. The maximum χ during HT ignition for Spray D is larger than that in Spray A by approximately a factor of 5. In contrast, at Tam=1000K, the χ values are similar between Spray A and Spray D. This elucidates the increase in the difference of IDT between Spray D and Spray A as T_am decreases. This may explain the contradicting findings on the effects of nozzle diameters on IDT from literature.