The department is researching the following topics:

Numerical simulation of novel combustion technologies
Tabulation of Dynamic Adaptive Chemistry (TDAC)
Reduced-order combustion models for turbulent reacting flows
Simulation of internal combustion engines
Uncertainty quanti cation for kinetic mechanism
Numerical simulation of indoor and outdoor fi re propagation

Numerical simulation of novel combustion technologies

MILD combustion represents a very attractive solution for combustion systems as it can provide high combustion efficiency with low NOx and soot emissions. The increasing interest in MILD combustion is also motivated by its large fuel exibility, representing a valuable technology for low-calori c value fuels, high-calori c industrial wastes and hydrogen-based fuels. Moreover, coupling MILD combustion with oxy-fuel technology is extremely ap- pealing for the steel industry, to increase the fuel exibility and to reduce fuel consumption and pollutant emissions.

Flameless combustion still appears worthy of investigations, in particular, the fundamental mechanism of the interaction between turbulent mixing and chemical kinetics needs to be elucidated. With respect to conventional ames, the turbulence levels infl ameless combustion are enhanced, due to flue gases recirculation, thus reducing mixing timescales. On the other hand, chemical timescales are increased, due to the dilution of the reactants. Hence, the turbulent and chemical timescales are of the same order of magnitude, thus leading to a very strong coupling. Subsequently, combustion models should take into account both mixing and chemical kinetics, leading to a very challenging task (see Figure 1).


Figure 1: Impact of the mechanism on the temperature distribution in a burner fed with CH4/H2: (a) ED/FR with global chemistry, (b) EDC with global chemistry, (c) EDC with DRM-19 and (d) EDC with GRI-3.0. DRM- 19 and GRI-3.0 are two detailed kinetic mechanisms.

Tabulation of Dynamic Adaptive Chemistry (TDAC)

To predict various phenomena in reactive systems, multi-dimensional models are generally employed to provide reasonable results and spatial resolution. Computational uid dynamics (CFD) simulations are very ecient tools to guide the design and the development of traditional or advanced combus- tion modes but also to get insight into the combustion phenomena. The fuel oxidation is modeled by solving the system of sti nonlinear ordinary di eren- tial equations (ODE) describing the species evolution in each computational cell for each time-step. In CFD simulations, solving this system dominates the computational cost and becomes prohibitive when several tens of species are included (around 90% of the computational time when 50 species are included). Therefore, when complex fuel kinetic mechanisms are required, the application of CFD is strongly limited, since hundreds of species and thousands of reactions are involved.

In this context, we develop the TDAC method to accommodate large kinetic mechanisms with CFD. It consists of two intrinsically coupled layers: tabulation and mechanism reduction (see Figure 2).


Figure 2: The tabulation layer tries to retrieve the solution of the ODE system at the query composition by linear approximation. For the cells where this fails, the mechanism reduction layer reduces the mechanism at runtime. It then only provides the active species to the ODE solver, which computes the mapping. ISAT grows or adds a point and extend this mapping to the full composition space.

To picture the complexity of a mechanism, we can represent it through a network of species connected to each others by reactions (see Figure 3). The reduction layer in TDAC dynamically reduces this mechanism by selecting the most important species at runtime (see animation of Figure 3).


Figure 3: Network of the connections between species through reactions for iso-octane. Nodes represent species and edges represent reactions.

Reduced-order combustion models for turbulent reacting flows

The current literature acknowledges two main classes of combustion models, amelet-like and PDF-like models. Flamelet-like models generally make a priori assumptions about the chemical state-space, which is described by a limited number of variables parameterizing the low-dimensional manifold.On the other hand, PDF methods make no assumption to restrict the species to a low-dimensional manifold and treat chemistry exactly. While PDF- like combustion models have demonstrated ability to accurately model non- premixed combustion systems, their application remains limited for large applications, due to their intrinsic CPU-intensive nature.

Within this framework, it appears clear that combustion models based on the automatic identifi cation of Low Dimensional Manifolds (LDM) would o er the great advantage of determining the dimensionality of the sub-space which adequately approximate the compositions occurring in diff erent regimes of turbulent combustion, thus limiting the computational e ort where it is needed (see Figure 4). The identifi cation of LDMs would be extremely benefi cial for both amelet-like and PDF models: for the former, it would lead to the de finition of appropriate parameters to tabulate the chemical state- space; for the latter, it would allow signifi cantly reducing the computational cost due to an optimal reduction of the problem size. In particular, LDMs methods able to identify the limiting chemical time-scales are extremely ap- pealing for Large Eddy Simulation (LES), to explicitly resolve the relevant scales on the LES grid, thus reducing the dependence of sub-grid models.

This model is employed for predictions, on more complex systems than the ones used to generate the model, as indicated by the animation in Figure 5, which shows the simulation of a syngas jet ame using a two-parameter eLDM model.



Figure 4: Combustion model generation based on empirical Low Dimensional Manifolds (eLDMs).

Figure 5: Syngas jet-ame simulation using a two-parameter eLDM model.


Simulation of internal combustion engines

To remain one of the major technology in future commercial powertrains, the internal combustion engine should address several key challenges such as the achievement of lower fuel consumption and pollutant emissions. Predicting the pollutants or the performance of new generation fuels or new engine concepts is a very important task.

In this context, we apply diff erent techniques to model the various combustion phenomena in the combustion chamber. In particular, in a recent study, we investigated the e ect of nitric monoxide injection in a homogeneous charge compression ignition engine (see Figure 6). The result of this study has also an impact on spark ignition engines where irregular phenomena called superknock might be observed.

This research project currently focus on the ignition and propagation of turbulent ames in direct injection spark ignition engine operated with natural gas (see animation of Figure 7). It also focuses on reacting sprays such as in diesel engines (see animation of Figure 8).



Figure 6: Network of the connections between species through reactions for iso-octane. Nodes represent species and edges represent reactions. 

 Figure 7: Gradient of density at the tip of a natural gas injector (courtesy of Dr. Catherine Duynslaegher)


Figure 8: The Sandia bomb is a constant volume vessel where combusting sprays are measured in detailed to validate numerical models.

Uncertainty quanti cation for kinetic mechanism

The development of reaction mechanisms leads to more and more complex systems as described above. There are trends or mechanism size that grow exponentially with the size of the fuel being modeled (see Figure 9). However, signifi cant uncertainties are linked to the data in those mechanisms. Increasing the size also hides the link between the precision on the parameters of the reactions and the end results.

In that framework, di fferent strategies exist to analyze the results and give an interval of con fidence around it.


Figure 9: The complexity of the mechanism increases with the size of the molecule being modeled.



Figure 10: Uncertainty on axial temperature predictions, associated with the kinetic mechanism. EDC combustion model.

Numerical simulation of indoor and outdoor fi re propagation

In the recent past, a series of catastrophic accidents involving tunnel res attracted the public attention: Nowadays, di fferent CFD models can be considered as well-established tools for the assessment of the consequences of res and for the design of related safety systems. The objective of the proposed research is that of validating numerical cools such as FDS (Fire Dynamics Simulator) and FireFOAM to model re propagation in indoor and outdoor con figurations .For indoor res (see Figure 11), the available propagation models will be tested to predict temperature distribution inside rooms, in diff erent con figuration, forced convection, no ventilation, and presence of energy sources. The results will be compared to literature data and benchmark to already validated codes. For outdoor con figurations (see Figure 12), the objective will be that of evaluating the impact and the hazard related to the radiation from res to exposed objects

Figure 11.



Figure 12

top of the page