BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Memento EPFL// BEGIN:VEVENT SUMMARY:MEchanics GAthering –MEGA- Seminar: Towards reduced-order models for metal-fueled chemical reactors in a carbonless circular energy econom y / A probabilistic framework for describing reactive multiphase flows DTSTART:20260604T124000 DTEND:20260604T134000 DTSTAMP:20260603T102830Z UID:543604b6cc92159e9ecd55969cbee8e3217114fdf613263cb6210494 CATEGORIES:Conferences - Seminars DESCRIPTION:Lisanne Gossel & Davide Mappelli (Dep of MEchanical Engineerin g\, TU Darmstadt)\nAbstract Lisanne: In search for an energy storage mate rial\, which enables future renewable energy supply on a global\nscale\, m etal powders recently gain growing interest in energy science and engineer ing as a recyclable chemical energy carrier. While being by far not as pre sent in the scientific and public discourse as other alternatives like hyd rogen and its derivatives\, metal powders are a promising alternative due to several advantageous properties. Iron is one of the mostly discussed ca ndidates to be used as an energy carrier\, since it has a high volumetric energy density\, is cheap and abundant and already established as a tradin g good. \nMetal powder combustion and reduction in chemical reactors is a t the core of the novel technology. The talk will focus on a physics-based reduced-order model for describing chemical reactors\, namely Chemical Re actor Networks (CRNs). CRNs are compartment-based reactor models employing strongly simplifying assumptions on the flow physics\, while enabling fas t and resource-efficient reactor simulations including detailed reaction m echanisms with thousands of species and reactions. We will focus on two re search aspects relevant for the development of the metal fuel technology.  \nFirst\, we will discuss recent achievements in the development of algo rithms for automatically deriving CRN models from primary data obtained fr om Computational Fluid Dynamics (CFD) simulations. Algorithmic CRN constru ction does not only help in tapping the potential of time efficiency assoc iated with CRNs\, but is essentially a way of finding valid and robust CRN models. We will discuss a novel modeling approach\, which puts a special emphasis on properly capturing the flow physics and mixing within the reac tor\, and yields CRN models that perform very good with respect to model c onsistency and robustness. \nA second important aspect of CRN models in t he context of metal energy carriers pertains to using CRNs in multi-query applications like process optimization and uncertainty quantification. Whe reas CRNs are in principle very well-suited for these tasks due to their h igh efficiency\, we will discuss how we can actually explore a design spac e using CRNs outside reference data points. \n\nAbstract Davide: In the context of increasing efforts toward environmental sustainability\, there is a growing need to improve the design and optimisation of industrial pro cesses through predictive modelling tools. This is particularly relevant f or high-temperature industrial processes\, which account for a large porti on of the energy consumed by the industrial sector. In the present contrib ution\, we specifically consider industrial processes in which a granular assembly of particles is thermochemically treated by a hot gas. Examples i nclude the calcination of limestone\, the reduction of iron ore\, the comb ustion of carbon particles and the pyrolysis of biomass. In view of indust rial-scale applications\, reactive granular assemblies with interstitial f lows are often described using Discrete Element Methods (DEM) for the soli d phase and unresolved Computational Fluid Dynamics (CFD) for the flow. A peculiarity of unresolved CFD methods is that the small-scale variability is eliminated upon the application of a filter. To close the resulting fil tered equations\, it is commonly assumed that the local gas composition co incides with its filtered value. However\, in the presence of intricate sm all-scale flow structures\, gas-particle mass and heat exchanges or highly non-linear chemical reactions\, the composition may vary on smaller scale s than the filter width\, rendering the small-scale homogeneity assumption questionable. In this talk\, we present a novel statistical approach\, in spired by methods developed for turbulent flows\, that addresses these lim itations by promoting the usual thermochemical fluid descriptors to random variables. Their behaviour is modelled by the temporal evolution of a one -point probability density function (PDF) associated with the spatial dist ribution of the gas-phase composition inside the local filter volume. An e volution equation for the PDF is derived and reduced to the case of large- scale homogeneity. By recasting the reduced PDF equation in terms of a sto chastic process\, the solution is obtained by applying a Monte Carlo schem e. The model is applied to the case of a packed bed of carbon particles he ated by interstitial methane combustion and the influence of the mixing li mitations on predictions of global observables is quantified. These findin gs demonstrate that PDF methods provide a promising framework for predicti ve simulations of reactive granular systems.\n\nBio Lisanne: I obtained a Bachelor and Master’s degree in physics from Technical University of Da rmstadt in Germany\, and spent one semester at Grenoble INP\, France durin g my Master’s. Being strongly interested in studying the energy transiti on\, I then joined the cluster project Clean Circles at TU Darmstadt\, whe re the use of iron powder as sustainable energy carrier has been studied i n a strongly interdisciplinary research environment. During my doctoral st udies\, which I conducted with the Institute for Mathematical Modeling and Analysis at TU Darmstadt\, I developed physics-based reduced-order models for iron-fueled chemical reactors. I completed my PhD in Mathematics in O ctober 2025 before joining the Department of Mechanical Engineering at TU Darmstadt and starting a PostDoc in the Institute for Numerical Methods in Fluid Mechanics\, where I am using statistical and population balance met hods to study turbulent aluminum dust flames. \n\nBio Davide: Davide Map elli is a PhD candidate at the Technical University of Darmstadt\, within the Institute for Numerical Methods in Fluid Mechanics\, under the supervi sion of Professor Fabian Sewerin. His research focuses on the development of probabilistic approaches for incorporating small-scale gas–particle t hermochemistry into models of granular assemblies. He received his B.Sc. i n Physics from the University of Milan in 2021\, followed by an M.Sc. from Heidelberg University in 2024. His academic background spans probabilisti c methods\, stochastic dynamics\, statistical physics\, and fluid dynamics . LOCATION:ME B1 10 https://plan.epfl.ch/?room==ME%20B1%2010 https://epfl.zo om.us/j/68096021948?pwd=DiE8amDIwYw1u3VtW2xWNGKtioKL2y.1 STATUS:CONFIRMED END:VEVENT END:VCALENDAR