R. Hilbert - Academia.edu (original) (raw)
Papers by R. Hilbert
Proceedings of the Combustion Institute, 2002
Autoignition of turbulent non-premixed flames is encountered in practical systems such as diesel ... more Autoignition of turbulent non-premixed flames is encountered in practical systems such as diesel internal combustion engines, but remains a challenge for modelers. The simplest models are generally based on an ad hoc coupling between Arrhenius laws and the Magnussen eddy dissipation concept. Mastorakos et al. have shown, using direct numerical simulations with a single-step irreversible chemistry, that autoignition occurs for a given most reactive mixture fraction level and low scalar dissipation rates. They devised a model based on a conditional moment closure formulation. In this paper, an extended (by integrating over all the possible mixture fraction values) flame surface density concept is introduced and tested against two-dimensional direct numerical simulations including detailed chemistry and multicomponent transport phenomena, including thermal diffusion. The mean reaction rate to be modeled is split into two contributions: a flame generalized surface density, including a turbulence contribution, and mean reaction rate per unit of flame area that could be described from one-dimensional laminar flame computations. First results appear to be very promising.
Proceedings of the Combustion Institute, 2002
Direct numerical simulations (DNS) are ideally suited to investigate in detail turbulent reacting... more Direct numerical simulations (DNS) are ideally suited to investigate in detail turbulent reacting flows in simple geometries. When considering such problems as pollutant emission or stability limits, detailed models must generally be employed to describe the chemical processes with sufficient accuracy. Due to the huge cost of such simulations, they have been mostly restricted to two-dimensional configurations up to now, leading to unsolved questions concerning the generality of the obtained results. We have recently developed a three-dimensional DNS code leading to reasonable computing times, thanks to the low-Machnumber approximation and to a new chemistry reduction technique. This code is used here to investigate the evolution of premixed methane/air flame kernels placed in a homogeneous isotropic turbulence field. This situation typifies the initial flame development after spark ignition in a gas turbine or an internal combustion engine. Laminar reference computations are carried out in one and two dimensions and are compared with turbulent results obtained in two and three dimensions. Evolution of flame surface area, stretch rate, and flame front curvature are in particular presented and show considerable differences between two-dimensional and three-dimensional computations. The interest of repeating the computations to increase the statistical validity of the results is demonstrated in two dimensions, but is not sufficient to explain the discrepancy obtained with the three-dimensional computation. Further three-dimensional simulations are nevertheless needed to quantify more precisely the observed changes (slower increase of the equivalent radius, higher stretch rate, and curvature shifted toward positive values).
Turbulent combustion of fossil fuels is nowadays still by far the most important process to cover... more Turbulent combustion of fossil fuels is nowadays still by far the most important process to cover the worldwide energy needs. Furthermore, turbulent flames are widely used in a wealth of practical applications. Nevertheless, fundamental aspects of turbulent combustion are still poorly understood due to the complexity of the underlying, individual physical processes and due to their complex nonlinear coupling. In order to improve combustion processes, two complementary means are classically used: experimental investigations and numerical simulations. Both lead nowadays to a huge quantity of raw data. In order to extract all useful information from these data, a library containing essential postprocessing methods has been developed: it contains numerous tools to analyze and visualize 2D and 3D flames and flow fields, to investigate geometry and structure of flames both locally and globally, to quantify the interaction between flow fields and flames, and to determine statistics and correlations of all variables that are essential for model development. This library has been coded using scripts implemented into the MatLab platform. Available tools and illustrative examples are described in this paper in order to demonstrate the interest of this approach.
International Journal of Heat and Mass Transfer, 2006
We perform in this paper a multi-objective design optimization concerning the blade shape of a he... more We perform in this paper a multi-objective design optimization concerning the blade shape of a heat exchanger, considering the coupled solution of the flow/heat transfer processes. For this, a genetic algorithm is used. The aim of the procedure is to find the geometry most favorable to simultaneously maximize heat exchange while obtaining a minimum pressure loss. An in-house computer package, called OPAL, performs the optimization process in a fully automatic manner. It calls the pre-processor to generate the computational geometry as well as the mesh, it then performs the numerical simulation of the coupled fluid flow/heat transfer problem using Fluent, calculates the output parameters, and iterates the procedure. The genetic algorithm relies on a relatively large number of simulations, which may result in a considerable computational effort, depending on the configuration. The procedure can thus be performed in parallel on a Linux PC cluster to reduce user waiting time. A nearly optimal speed-up is obtained for the present configuration.
Combustion Theory and Modelling, 2003
Combustion and Flame, 2004
A turbulent nonpremixed H 2 /air flame is simulated using 2D direct numerical simulations coupled... more A turbulent nonpremixed H 2 /air flame is simulated using 2D direct numerical simulations coupled with a complete chemical scheme and a detailed transport model. The influence of differential diffusion is evidenced by comparing these results in terms of scatterplots and local flame structures with similar computations using a Lewis number unity hypothesis. The fast chemistry limit and thus the maximum flame temperatures are calculated using one-point equilibrium calculations. The fact that flame temperatures above the maximum flame temperature calculated with the Lewis number unity assumption can be observed with the detailed transport model is discussed and can be attributed to the fact that the mixture fraction and temperature do not have the same diffusion coefficient. A simplified model is proposed to account for this effect.
Combustion and Flame, 2002
The autoignition of a laminar non-premixed flame placed in a field of homogeneous isotropic turbu... more The autoignition of a laminar non-premixed flame placed in a field of homogeneous isotropic turbulence has been studied previously using single-step chemistry and/or simplified models for diffusion processes. The existence of a specific value of the mixture fraction, called "most-reactive," and the importance of the scalar dissipation rate to predict the ignition location were demonstrated. The effect of the turbulence intensity on the ignition time was found to be non-monotonic. In this work, we wish to assess the influence of more realistic chemistry and transport models on ignition location and time. To do so, direct simulations are carried out using a detailed reaction scheme, multicomponent diffusion velocities and accurate thermodynamic properties. We observe that the turbulent non-premixed flame ignites always faster than the laminar one, even for the highest Reynolds numbers investigated. The scalar dissipation rate can still be used to predict the ignition site, as was observed in simple chemistry simulations. But the most-reactive conditions must of course be determined using the detailed modeling, and cannot any more be analytically predicted. The interest of repeating the direct simulations to get rid of the influence of random initial conditions is also demonstrated.
Progress in Energy and Combustion Science, 2004
More and more publications can be found in recent years where detailed models are employed to des... more More and more publications can be found in recent years where detailed models are employed to describe the chemical and molecular transport processes controlling flame structure. Up to a recent past, such studies were restricted to simple zero-or one-dimensional laminar computations, like ignition in a fully premixed mode, freely propagating laminar premixed flames or counter-flow flames. Since such models are now often used to investigate turbulent flames in multi-dimensional computations, we feel it is useful to review the literature on this subject and give a synthesis of the obtained results. To be more specific, we consider only in this review publications where (1) chemical processes are modeled with a multi-step reaction scheme, taking at least an intermediate species into account; or (2) molecular diffusion processes of the individual species are represented by a more elaborate model than assuming unity Lewis numbers; and (3) the retained configuration leads to unsteady strain-rate and curvature (or stretch-rate) variations in the reaction zone. Over 200 recent publications have been found to respect these criteria. Summarizing the results, one can say that there appears to be a growing need for simulations relying on detailed models for chemistry and transport processes, probably due to the fact that restrictions concerning pollutant emissions motivate a request for more accurate, quantitative results. Progress must still be accomplished concerning the identification of chemical pathways, the accurate determination of rate constants, and the development of reliable but efficient chemistry reduction techniques. The impact of the retained molecular diffusion model is higher than expected at the beginning of this study. Even for turbulent configurations, the global impact of these models can be comparable to switching between two different detailed chemical schemes. Concerning local flame structure, the transport models play an essential role, in particular for high flame curvatures and far from stoichiometry. As a whole, the need for matching the accuracy level of the chosen chemical and transport models is emphasized, since describing a physical phenomenon in great detail while, at the same time, representing another phenomenon of comparable importance with a very rough model, prevents really quantitative (and even perhaps qualitative) predictions. Specific difficulties concerning validation are also identified. q
Proceedings of the Combustion Institute, 2002
Autoignition of turbulent non-premixed flames is encountered in practical systems such as diesel ... more Autoignition of turbulent non-premixed flames is encountered in practical systems such as diesel internal combustion engines, but remains a challenge for modelers. The simplest models are generally based on an ad hoc coupling between Arrhenius laws and the Magnussen eddy dissipation concept. Mastorakos et al. have shown, using direct numerical simulations with a single-step irreversible chemistry, that autoignition occurs for a given most reactive mixture fraction level and low scalar dissipation rates. They devised a model based on a conditional moment closure formulation. In this paper, an extended (by integrating over all the possible mixture fraction values) flame surface density concept is introduced and tested against two-dimensional direct numerical simulations including detailed chemistry and multicomponent transport phenomena, including thermal diffusion. The mean reaction rate to be modeled is split into two contributions: a flame generalized surface density, including a turbulence contribution, and mean reaction rate per unit of flame area that could be described from one-dimensional laminar flame computations. First results appear to be very promising.
Proceedings of the Combustion Institute, 2002
Direct numerical simulations (DNS) are ideally suited to investigate in detail turbulent reacting... more Direct numerical simulations (DNS) are ideally suited to investigate in detail turbulent reacting flows in simple geometries. When considering such problems as pollutant emission or stability limits, detailed models must generally be employed to describe the chemical processes with sufficient accuracy. Due to the huge cost of such simulations, they have been mostly restricted to two-dimensional configurations up to now, leading to unsolved questions concerning the generality of the obtained results. We have recently developed a three-dimensional DNS code leading to reasonable computing times, thanks to the low-Machnumber approximation and to a new chemistry reduction technique. This code is used here to investigate the evolution of premixed methane/air flame kernels placed in a homogeneous isotropic turbulence field. This situation typifies the initial flame development after spark ignition in a gas turbine or an internal combustion engine. Laminar reference computations are carried out in one and two dimensions and are compared with turbulent results obtained in two and three dimensions. Evolution of flame surface area, stretch rate, and flame front curvature are in particular presented and show considerable differences between two-dimensional and three-dimensional computations. The interest of repeating the computations to increase the statistical validity of the results is demonstrated in two dimensions, but is not sufficient to explain the discrepancy obtained with the three-dimensional computation. Further three-dimensional simulations are nevertheless needed to quantify more precisely the observed changes (slower increase of the equivalent radius, higher stretch rate, and curvature shifted toward positive values).
Turbulent combustion of fossil fuels is nowadays still by far the most important process to cover... more Turbulent combustion of fossil fuels is nowadays still by far the most important process to cover the worldwide energy needs. Furthermore, turbulent flames are widely used in a wealth of practical applications. Nevertheless, fundamental aspects of turbulent combustion are still poorly understood due to the complexity of the underlying, individual physical processes and due to their complex nonlinear coupling. In order to improve combustion processes, two complementary means are classically used: experimental investigations and numerical simulations. Both lead nowadays to a huge quantity of raw data. In order to extract all useful information from these data, a library containing essential postprocessing methods has been developed: it contains numerous tools to analyze and visualize 2D and 3D flames and flow fields, to investigate geometry and structure of flames both locally and globally, to quantify the interaction between flow fields and flames, and to determine statistics and correlations of all variables that are essential for model development. This library has been coded using scripts implemented into the MatLab platform. Available tools and illustrative examples are described in this paper in order to demonstrate the interest of this approach.
International Journal of Heat and Mass Transfer, 2006
We perform in this paper a multi-objective design optimization concerning the blade shape of a he... more We perform in this paper a multi-objective design optimization concerning the blade shape of a heat exchanger, considering the coupled solution of the flow/heat transfer processes. For this, a genetic algorithm is used. The aim of the procedure is to find the geometry most favorable to simultaneously maximize heat exchange while obtaining a minimum pressure loss. An in-house computer package, called OPAL, performs the optimization process in a fully automatic manner. It calls the pre-processor to generate the computational geometry as well as the mesh, it then performs the numerical simulation of the coupled fluid flow/heat transfer problem using Fluent, calculates the output parameters, and iterates the procedure. The genetic algorithm relies on a relatively large number of simulations, which may result in a considerable computational effort, depending on the configuration. The procedure can thus be performed in parallel on a Linux PC cluster to reduce user waiting time. A nearly optimal speed-up is obtained for the present configuration.
Combustion Theory and Modelling, 2003
Combustion and Flame, 2004
A turbulent nonpremixed H 2 /air flame is simulated using 2D direct numerical simulations coupled... more A turbulent nonpremixed H 2 /air flame is simulated using 2D direct numerical simulations coupled with a complete chemical scheme and a detailed transport model. The influence of differential diffusion is evidenced by comparing these results in terms of scatterplots and local flame structures with similar computations using a Lewis number unity hypothesis. The fast chemistry limit and thus the maximum flame temperatures are calculated using one-point equilibrium calculations. The fact that flame temperatures above the maximum flame temperature calculated with the Lewis number unity assumption can be observed with the detailed transport model is discussed and can be attributed to the fact that the mixture fraction and temperature do not have the same diffusion coefficient. A simplified model is proposed to account for this effect.
Combustion and Flame, 2002
The autoignition of a laminar non-premixed flame placed in a field of homogeneous isotropic turbu... more The autoignition of a laminar non-premixed flame placed in a field of homogeneous isotropic turbulence has been studied previously using single-step chemistry and/or simplified models for diffusion processes. The existence of a specific value of the mixture fraction, called "most-reactive," and the importance of the scalar dissipation rate to predict the ignition location were demonstrated. The effect of the turbulence intensity on the ignition time was found to be non-monotonic. In this work, we wish to assess the influence of more realistic chemistry and transport models on ignition location and time. To do so, direct simulations are carried out using a detailed reaction scheme, multicomponent diffusion velocities and accurate thermodynamic properties. We observe that the turbulent non-premixed flame ignites always faster than the laminar one, even for the highest Reynolds numbers investigated. The scalar dissipation rate can still be used to predict the ignition site, as was observed in simple chemistry simulations. But the most-reactive conditions must of course be determined using the detailed modeling, and cannot any more be analytically predicted. The interest of repeating the direct simulations to get rid of the influence of random initial conditions is also demonstrated.
Progress in Energy and Combustion Science, 2004
More and more publications can be found in recent years where detailed models are employed to des... more More and more publications can be found in recent years where detailed models are employed to describe the chemical and molecular transport processes controlling flame structure. Up to a recent past, such studies were restricted to simple zero-or one-dimensional laminar computations, like ignition in a fully premixed mode, freely propagating laminar premixed flames or counter-flow flames. Since such models are now often used to investigate turbulent flames in multi-dimensional computations, we feel it is useful to review the literature on this subject and give a synthesis of the obtained results. To be more specific, we consider only in this review publications where (1) chemical processes are modeled with a multi-step reaction scheme, taking at least an intermediate species into account; or (2) molecular diffusion processes of the individual species are represented by a more elaborate model than assuming unity Lewis numbers; and (3) the retained configuration leads to unsteady strain-rate and curvature (or stretch-rate) variations in the reaction zone. Over 200 recent publications have been found to respect these criteria. Summarizing the results, one can say that there appears to be a growing need for simulations relying on detailed models for chemistry and transport processes, probably due to the fact that restrictions concerning pollutant emissions motivate a request for more accurate, quantitative results. Progress must still be accomplished concerning the identification of chemical pathways, the accurate determination of rate constants, and the development of reliable but efficient chemistry reduction techniques. The impact of the retained molecular diffusion model is higher than expected at the beginning of this study. Even for turbulent configurations, the global impact of these models can be comparable to switching between two different detailed chemical schemes. Concerning local flame structure, the transport models play an essential role, in particular for high flame curvatures and far from stoichiometry. As a whole, the need for matching the accuracy level of the chosen chemical and transport models is emphasized, since describing a physical phenomenon in great detail while, at the same time, representing another phenomenon of comparable importance with a very rough model, prevents really quantitative (and even perhaps qualitative) predictions. Specific difficulties concerning validation are also identified. q