DNS Database at Half Resolution for MILD Combustion of Renewable Fuel Mixtures in a Temporally-Evolving Mixing Layer

ENCODING: Project description ENCODING is a Marie Curie Doctoral Network aimed at advancing sustainable combustion technologies using hybrid physics-based, data-driven modelling. The project is supported by 17 collaborators, including 2 national research centres, 5 universities, and 10 industrial partners from 7 countries, which provide a multidisciplinary environment for 10 Doctoral Candidates (DCs). The project addresses the challenges related to Renewable Synthetic Fuels (RSF) deployment in the industrial sector, focusing on the following aspects: fuel flexibility operation, low-emissions combustion technologies, and the creation of predictive models for complex systems. ENCODING integrates Machine Learning (ML) techniques with physics-based knowledge to develop digital twins for real-time monitoring and control of combustion systems. The multidisciplinary nature of ENCODING is reflected in the project structure, which comprises 8 Work Packages (WPs) and how these relate to each other. The WPs cover: data generation; simulation model development; feature identification; reduced-model generation; sensing strategies; and Digital Twins testing.For more info, visit our website: https://encoding.ulb.be and/or our LinkedIn page: https://www.linkedin.com/company/93121193/admin/dashboard/ Introduction Alternative carbon-free fuels, such as ammonia and hydrogen, are essential for a sustainable future energy system. During the transition period, when these fuels are not yet widely available, blending them with conventional fuels is often necessary. However, their unique combustion properties pose significant challenges to achieving fuel-flexible, stable, and efficient combustion with low emissions. In this context, MILD (Moderate or Intense Low oxygen Dilution) [1] combustion offers a promising solution. By recirculating hot combustion products into fresh reactants, MILD combustion exhibits lower oxygen concentration and higher unburned mixture temperature, resulting in reduced peak temperatures, lower emissions, and enhanced stability. In this study, Direct Numerical Simulations (DNS) of a temporally evolving mixing layer configuration [2] are designed and used to preliminarily explore autoignition and flame propagation in realistic non-premixed MILD burner conditions. DNS Dataset Configuration The simulation considers the mixing dynamics of three streams, including air, fuel, and hot combustion products, as shown in Fig. 1a. The domain consists of a box with periodic boundary conditions in the streamwise (x) and spanwise (z) directions are employed, while an outlet boundary condition is imposed in the crosswise (y) direction. For the computation of the DNS, the in-house code CIAO was employed. Chemical reactions are modeled with a reduced model derived from the NUIG kinetic mechanism [3], with 35 species and 489 reactions. For the initial parameter definition, two different Damköhler numbers (ratio of a characteristic flow time scale and a chemical time scale) are defined. The chemical time is taken as the ignition delay time of the mixture, which is determined from a series of homogeneous reactor calculations at different mixture fraction compositions using FlameMaster [4]. The auto-ignition delay time is defined as the lowest time at which the profile has a maximum temperature higher than that of the hot products (1250 K). This indicates that the mixture has ignited at 14 ms (Fig. 1b). A first Damköhler number is taken as the ratio of the air-hot product mixing time and the chemical time (DaHA), while the second Damköhler number (DaFA) considers the fuel-air mixing time as the flow timescale. Defining tign is crucial. Since the ignition is physically impossible before tign, because dissipation suppresses the ignition process [5], tign is a lower bound for the ignition delay time in the turbulent combustion scenario considered in the DNS. A summary of the physical and numerical parameters for the DNS dataset can be found in Tab 1. For the DNS, the DaHA is set to 0.2, resulting in a sufficiently high dilution level and reduced peak temperatures required for MILD combustion, i.e., the ignition is slower than the mixing process between air and hot products. A DaFA of 0.2 is similarly employed for the fuel-air mixing system. Temporal jet DNS parameters Fuel [mol%] 75% CH4 – 25% H2 Equivalence ratio 0.8 H_air[mm] 25 H_fuel [mm] 0.8 [Lx, Ly, Lz][mm] [750,750,375] [Nx, Ny, Nz] [133,83,66] Fuel-air shear velociy[m/s] 10 Hot products-air shear velocity [m/s] 40 Fuel-air Reynolds number 300 Hot products-air Reynolds number 10000 Fuel-air Damköhler number 0.2 Hot products-air Damköhler number 0.2 Chemical time [milliseconds] 14 Simulation time [milliseconds] 40 Max timestep [microseconds] 1.5 Table 1: Summary of numerical and physical simulation parameters for the considered DNS. Included Datasets This deliverable includes seven datasets, each representing a different time instant in the temporal evolution of the previously shown configuration. Specifically, the most representative time instants of the ignition process, activated by the mixing of reactants with combustion products, have been selected. The datasets are shown in Fig. 2(a). The temperature rise within the domain is low, with an average temperature in the reactive region of 1500 K. In Fig. 2(b), the distribution of the OH radical is shown on a logarithmic scale. The OH radical is well distributed in a larger reaction zone and not restricted to two sheets like canonical non-premixed flames. The dataset is provided as HDF5 (Hierarchical Data Format version 5), commonly using the .h5 file extension. HDF5 is an open-source file format and set of tools designed for storing and managing large, complex data collections. The dataset is structured as a list of tuples, where each tuple consists of an index and the corresponding variable name. This structure provides a clear and organized mapping of indices to their respective variables, facilitating efficient data access and analysis. The dataset includes some key chemical species in the kinetic mechanism, followed by additional variables: · Temperature; · Enthalpy; · Heat Release Rate; · Source term of H radical. The dataset includes the previously mentioned variables and the three components of velocity, density, and dynamic viscosity. The supplementary material provides a comprehensive list of all variables. Conclusion The chosen configuration successfully achieved MILD combustion conditions as indicated by the small temperature increase and the well-distributed OH field. All datasets shown in Figure 2 have been uploaded, each containing the whole species present in the kinetic mechanism and some turbulent and thermodynamic variables; in addition, a Python script that allows reading the variables together with a function to produce a 2D contour plot of a selected variable has been uploaded. These datasets can be used both to improve simulations that are less computationally expensive than DNS, such as LES and RANS, and to further our understanding of MILD combustion. In the future, additional simulations will be carried out using other alternative fuels, such as ammonia. Acknowledgment The authors gratefully acknowledge the computing time provided to them at the NHR Center NHR4CES at RWTH Aachen University (project number p0021021). This is funded by the Federal Ministry of Education and Research. This ENCODING project has received funding from the European Union's Horizon Europe research and innovation programme under the Marie Skłodowska-Curie grant agreement No 101072779. The results of this publication/presentation reflect only the author(s) view and do not necessarily reflect those of the European Union. The European Union can not be held responsible for them. Licensing Information This dataset is shared under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. You are free to share, adapt, and use the data for any purpose, provided that proper attribution is given to the authors of the dataset.For more details on the terms of the CC BY 4.0 license, please refer to the Creative Commons website.