Refine
Document Type
Way of publication
- Open Access (6)
Keywords
Institute
- FG Numerische Strömungs- und Gasdynamik (54)
- FG Mechanische Verfahrenstechnik (12)
- Energie-Innovationszentrum / Scientific Computing Lab (4)
- FG Numerische Mathematik und Wissenschaftliches Rechnen (2)
- FG Prozess- und Anlagentechnik (2)
- FG Flug-Triebwerksdesign (1)
- FG Thermodynamik / Thermische Verfahrenstechnik (1)
BTU
The detailed numerical representation of turbulent mixing processes is a standing challenge for non-premixed chemically reacting flows. The full range of relevant flow scales needs to be captured and it is also necessary to distinguish turbulent advective from molecular diffusive processes in order to represent Reynolds and Schmidt number effects. These requirements are addressed here by utilizing two different map-based stochastic turbulence modeling strategies. The one-dimensional turbulence (ODT) model utilizes event-based turbulence modeling, whereas the hierarchical parcel-swapping (HiPS) model is a fully event-based mixing model. ODT provides full-scale resolution at affordable costs by dimensional model reduction based on the boundary-layer approximation to shear flow. HiPS is far less costly than ODT but currently limited to locally homogeneous isotropic turbulence. The physics-compatible modeling capabilities with respect to phase-space representation of turbulent mixing are demonstrated for two canonical cases using standalone model formulations.
The One-Dimensional Turbulence (ODT) model is applied to reactive flows in open and closed systems represented by a lifted jet flame in a vitiated coflow, and a constant volume autoignition configuration, respectively. ODT is a one-dimensional model for turbulent flow simulations, which uses a stochastic formulation to represent the effcts of turbulent advection. Diffusion and reaction effcts along the ODT domain are considered by deterministic evolution equations. This work is an effort to verify the applicability and effiency of the model for open and closed systems. In the open system case, ODT results are compared against experimental results of a lifted methane/air jet flame detailed in the work of Cabra et al.
[1]. In the closed system case, a periodic, constant volume domain is used to investigate the sensitivity of the ignition evolution to initial temperature and composition inhomogeneities of a lean n-heptane/air mixture. In the latter context, ODT results are compared to DNS results from Luong et al. [2]. Results for the jet and constant volume configuration
show a reasonable match with the experimental and DNS data, considering the reduced order of the model and the underlying assumptions for each case. At the jet configuration, a dependence of the flame evolution on the turbulence intensity parameter can be seen. For the closed system, initial temperature and composition inhomogeneities allow a mitigation
of the undesirable rapid pressure rise.
Abstract Constant volume premixed lean n-Heptane/air autoignition at high pressure is investigated using the One-Dimensional Turbulence (ODT) model. The configuration consists of a 1D fixed volume domain with a prescribed velocity spectrum and temperature fluctuations superimposed on an initial uniformly elevated scalar field. The sensitivity of the heat release rate and pressure evolution to the initial temperature distribution is studied by imposing different initial temperature fields while holding the mean, RMS and integral length scale of the field constant. Three detailed chemical mechanisms are employed for the prediction of autoignition and heat release rate. To mitigate the high computational cost associated with the calculation of the chemical source terms in the stiff complex mechanisms, an approach based on the Strang-Splitting method is presented. Finally, a …
The one-dimensional turbulence (ODT) model resolves a full range of time and length scales and is computationally efficient. ODT has been applied to a wide range of complex multi-scale flows, such as turbulent combustion. Previous ODT comparisons to experimental data have focused mainly on planar flows. Applications to cylindrical flows, such as round jets, have been based on rough analogies, e.g., by exploiting the fortuitous consistency of the similarity scalings of temporally developing planar jets and spatially developing round jets. To obtain a more systematic treatment, a new formulation of the ODT model in cylindrical and spherical coordinates is presented here. The model is written in terms of a geometric factor so that planar, cylindrical, and spherical configurations are represented in the same way. Temporal and spatial versions of the model are presented. A Lagrangian finite-volume implementation is used with a dynamically adaptive mesh. The adaptive mesh facilitates the implementation of cylindrical and spherical versions of the triplet map, which is used to model turbulent advection (eddy events) in the one-dimensional flow coordinate. In cylindrical and spherical coordinates, geometric stretching of the three triplet map images occurs due to the radial dependence of volume, with the stretching being strongest near the centerline. Two triplet map variants, TMA and TMB, are presented. In TMA, the three map images have the same volume, but different radial segment lengths. In TMB, the three map images have the same radial segment lengths, but different segment volumes. Cylindrical results are presented for temporal pipe flow, a spatial nonreacting jet, and a spatial nonreacting jet flame. These results compare very well to direct numerical simulation for the pipe flow, and to experimental data for the jets. The nonreacting jet treatment overpredicts velocity fluctuations near the centerline, due to the geometric stretching of the triplet maps and its effect on the eddy event rate distribution. TMB performs better than TMA. A hybrid planar-TMB (PTMB) approach is also presented, which further improves the results. TMA, TMB, and PTMB are nearly identical in the pipe flow where the key dynamics occur near the wall away from the centerline. The jet flame illustrates effects of variable density and viscosity, including dilatational effects.
Investigating dissipative roughness effects on turbulent drag using a stochastic turbulence model
(2023)
The joint modeling of flow hydrodynamics and electrokinetics is a relatively unexplored area of turbulent flow research. We address a lack of available models for electrohydrodynamic (EHD) turbulent flow utilizing a lower-order approach, the stochastic One-Dimensional Turbulence (ODT) model. ODT is constructed on the principles of the direct energy cascade of Navier–Stokes turbulence, with key emphasis on the accurate resolution of the small molecular transport scales within a notional line-of-sight. We investigate two canonical flow configurations to demonstrate the applicability of the model
in the simulation of EHD flows. First, we investigate EHD effects in zero-pressure-gradient turbulent boundary layers by two-way coupled model application to plane Couette flow of a dilute electrolyte. Second, we apply the one-way coupled model to EHD-enhanced gas flow through a vertical pipe with an inner concentric electrode, where electric fields are generated by means of a corona discharge and the corresponding effect of a continuum ionic charge density field.
The One-Dimensional Turbulence model is modified in this work for its application to a classical electrohydrodynamic (EHD) problem. Being the first study case, this work is focused on the influence of electrostatic fields and space charge on the velocity field inside a wire-plate Electrostatic Precipitator (ESP) with one-way-coupling dynamics. The study case is an attempt to replicate velocity profiles and Turbulent Kinetic Energy (TKE) budgets obtained in the Direct Numerical Simulation (DNS) carried out by Soldati and Banerjee (1998). Qualitative trends are confirmed in preliminary ODT results, thus showing the potential of the stochastic ODT modeling approach for other types of EHD flows.
The stochastic One-Dimensional Turbulence (ODT) model is used in combination with a Large Eddy Simulation (LES) approach in order to illustrate the potential of the fully coupled model (ODTLES) for highly turbulent flows. In this work, we use a new C++ implementation of the ODTLES code in order to analyze the computational performance in a classical incompressible turbulent channel flow problem. The parallelization potential of the model, as well as its physical and numerical consistency are evaluated and compared to Direct Numerical Simulations (DNSs). The numerical results show that the model is capable of reproducing a representative part of the DNS data at a cheaper computational cost. This advantage can be enhanced in the future by the implementation of a straightforward parallelization approach.
Electrohydrodynamic‐enhanced internal pipe flows from a One‐Dimensional Turbulence perspective
(2020)
The computational cost of Direct Numerical Simulations (DNS) that resolve all scales rises with the cube of the Reynolds numberand is currently not feasible for real world applications. Large Eddy Simulations (LES) overcome this limitation by only resolving the large scale effects and completely model the small scaleeffects. This results in a strong dependence of the accuracy onthe chosen subgrid-scale model. The model adaptivity concept discussed in [Schmidt et al., ICDERS, 2007] dynamically uses different Linear Eddy Model (LEM) types [Kerstein, LEM, 1988] for stochastic closure to model the turbulent flame speed of apremixed flame within LES. In the talk, we will first summarize the progress of our group on One Dimensional Turbulence (ODT) [Kerstein, ODT, 1999] and Hierarchical Parcel Swapping (HiPS) [Kerstein, HiPS, 2013] based on reactive stand-alone simulations, [Jozefik et al., Combust. Flame, 2015] [Jozefik et al., Combust. Flame, 2016] [Medina et al., Combust. Flame, 2018]. Second,we will sketch a possible combination of the model adaptivity concept [Schmidt et al., ICDERS, 2007] and recent efficient ODTLES implementations [Glawe et al., Z. Angew. Math. Mech.,2018] to dynamically use LEM, ODT and HIPS together inside ofLES solvers to simulate turbulent reactive flows.