Hydrogen instead of natural gas: Optimal burner geometry through CFD simulation

How can CFD simulation help me?

This is where Ansys Fluent comes into play for CFD simulation. CFD stands for Computational Fluid Dynamics and is a method of numerically calculating the flow of gases and liquids. CFD simulations allow you to virtually simulate your burner and vary various parameters. This allows you to see how the flow, combustion, and heat transfer in your burner behave and what effects this has on performance and emissions. You can also compare different burner designs and choose the best one.

Other benefits of CFD simulation of industrial burners include:

• Saving money and time: No need for real tests with expensive prototypes.
• Increase safety: Predict and avoid flashback and spontaneous combustion.
• Improve efficiency and environmental compatibility: Optimize fuel-to-air ratio and reduce NOx emissions.

Simulate combustion - Start with the grid

Once you have mastered the first steps and clearly outlined your task, it is worth focusing on grid generation. The grid consisting of individual cells has a significant influence on the accuracy and stability of the simulation. A high-quality grid should always have the following properties:

• Sufficient resolution to capture all relevant flow structures,
• smooth transition between regions to reduce the numerical error, and
• an alignment of the cell surfaces that is as orthogonal as possible to the streamlines.

So, what does a suitable grid for hydrogen combustion look like? Here are a few tips for optimal grid generation [1]. The table below provides an orientation of important grid settings for the burner itself including nozzles for the inflow of reactants, for the combustion chamber and its core as well as for the outer areas where no reaction takes place. In detail, every CFD simulation always requires a sensitivity study to minimize the influence of the grid resolution. If turbulent structures of the flow and the interaction of chemistry and flow are to be calculated in particular detail, a much finer mesh may also be required.

By the way: The MOSAIC Poly Hexcore technology patented in Ansys Fluent provides you with a high-quality structured mesh in the combustion chamber, which can also be calculated excellently on a cluster. And with the Fault-Tolerant Meshing Workflow, you can also mesh difficult and complicated CAD geometries, as the method provides automatic correction mechanisms. If you are not sure about the required mesh resolution or just want to resolve the flame front more precisely, you can use the adaptive meshing for polyhedron meshes (PUMA algorithm). A predefined criterion based on the so-called Flame Indicator refines the mesh only where necessary [2].

Which combustion model is the right one? EDC vs. FGM model

We now come to the selection of the appropriate combustion model. This should be chosen so that it sufficiently and accurately describes the chemical reactions and the turbulence-chemistry interactions. Common models are the EDC (Eddy Dissipation Concept) and the FGM (Flamelet Generated Manifold). The EDC model is based on the idea that chemical reactions take place in the finest turbulent eddies. The reaction rate is influenced by the vortex size and frequency, which in turn depend on the turbulence. Specifically, the EDC model utilizes an extensive reaction mechanism and solves the reaction equations using efficient tabulation algorithms such as the ISAT (In-Situ Adaptive Tabulation Algorithm).

In contrast, the FGM model is based on the assumption that the internal flame structure follows a one-dimensional representation. This is calculated and tabulated in advance using a laminar 1D flame solver such as Chemkin. The FGM model can, for example, be combined with the skeletal reaction mechanism, which contains 10 gas components (the so-called species) and 25 reactions. If the reaction scope is not sufficient, the UCSD mechanism (57 species, 268 reactions) is an alternative. The following table compares the EDC and FGM models. The table below provides an example setup for the FGM model for a partially premixed flame, as well as recommendations for numerical settings. Sufficiently large flow times ensure a meaningful transient averaging of the results.

When we talk about combustion, we also need to consider the turbulence model. For cases with stable flow fields, RANS models (k-ω SST, Realizable k-ϵ) are an efficient solution for many engineering problems. However, there is often a strong coupling between flow and reaction. For example, unstable flows lead to local extinction of the flame. In order to capture flame formation and pollutant emissions as accurately as possible, a Large Eddy Simulation (LES) is the best choice. However, LES is generally considered to be very complex. A good compromise between accuracy and effort is the Stress-Blended Eddy Simulation (SBES), because it can switch between RANS and LES models.