Deliverables
Aiming at working synergistically and reaching critical mass we will collaborate and organise joint activities with complementary ECfunded initiatives in particular other projects funded within the DTNMBP092018 call the EMMC Working Groups and the key players of the cluster of 51 ie the European Multiscale Materials Modelling Cluster of 5 ECfunded projects on multiscale modelling for nanomaterials and systems by design and the ICMEg CSA network We will aim to arrange yearly workshops with them focused on standardisation and interoperability demonstrations of applications plugfests moving data and applications across marketplaces exploitation strategies and thematic CECAM workshops on Machine Learning and HighPerformance Computing SURFsaraSupport for the EMMC will be a prominent part of this effort with Fraunhofer and SCM dedicating personmonths to contribute to relevant activities of the EMMCCSA which will run until 2019 including its commercial software deployment work package Furthermore SCM will strive to contribute with an active role in the EMMC Software Owner SWO group
Conference organised by ReaxProThe ReaxPro consortium will organise at least one main conference to involve the target groups in academia and industry.
Journal reviewsTowards the end of the project we will publish two reviews to disseminate the approach and our catalysisrelated results to members of the scientific community at large
Conference abstracts and presentations, peer-reviewed journal articlesAll partners will publish on the scientific results and technical advancements obtained during this project in peerreviewed journals and will present their activities at international conferences and professional meetingsThe theoretical and algorithmic achievements underpinning the technologies implemented in the proposed multiscale modelling platform will be published in journals encompassing the fields of Computer Science Materials and Process Modelling whereas results from case studies will be more appropriate for journals specialising in IndustrialApplied Chemistry Catalysis and Chemical Engineering Examples of journals that would be potential candidates for publishing our findings include Journal of Multiscale Modelling World Scientific Multiscale Modeling and Simulation Society for Industrial and Applied Mathematics Journal of Catalysis Journal of Computational Physics Computers Chemical Engineering Elsevier Journal of Chemical Physics American Institute of Physics ACS Catalysis Journal of Chemical Theory and Computation American Chemical Society Nature Chemistry Nature Materials Nature Catalysis Springer Nature Faraday Discussions Catalysis Science Technology Royal Society of ChemistryExamples of events that we are planning to attend include International Symposium on Chemical Reaction Engineering ISCRE World Conference on Computers in Education WCCE European Congress of Chemical Engineering ECCE European Congress on Catalysis EuropaCat European Symposium on ComputerAided Process Engineering ESCAPE North American Meeting NAM of the North American Catalysis Society NACS as well as meetings by the EMMC CECAM the European Federation of Chemical Engineering EFCE the Institution of Chemical Engineers IChemE the Royal Society of Chemistry RSC eg Faraday Discussions the American Institute of Chemical Engineers AIChE and the American Chemical Society ACS In addition we will attend more specialised events eg International Conference on Numerical Methods in Multiphase Flows International Conference on Computational Experimental Methods in Multiphase Complex Flow and workshops on OpenFOAM We will also disseminate our computational approaches to conferences pertinent to High Performance Computing including meetings by the Association for Computing Machinery ACM and the Institute of Electrical and Electronics Engineers IEEE Industrial partners will deliver presentations on business events such as the Society of Chemical Industry SCI Process Development Symposium SCM will also disseminate ReaxPros developments during its site visits to manufacturing companies
Brief reports on outreach activitiesWe will further pursue targeted outreach activities to the general public by leveraging our connections with local societies charities and institutions Eg university Open Days will be an ideal way to approach the public and all academic partners have organised such events in the past
An extensible ontology will be defined for ReaxPro adopting and enhancing the EMMO in collaboration with the pertinent EMMC Working Groups on Interoperability and Open Simulation Platform building on previous efforts such as the SimPhoNy and MoDeNa projects and the development of the Allotrope Data Format The outcome of these activities will be final ontology specifications extending the EMMO in Web Ontology Language OWL delivered by Fraunhofer IWM
ReaxPro meta-repository for storing and exchanging multiscale modelling dataA meta-repository will be delivered for multiscale models materials relations and pertinent data. This will build on Task (and corresponding deliverable) 2.3, investigating the possibility to integrate the data produced by the project with existing repositories where relevant (e.g. NOMAD, and potentially services by MARKETPLACE and VIMMP).
Parallel implementations of CatalyticFOAM modulesCatalyticFOAM already implements a spatial decomposition scheme that accelerates the computation of the source (reaction) terms over the domain of the CFD calculation. However, depending on the distribution of catalyst versus bulk fluid on the domain and the difference in the reactions in these two phases, it may happen that the computational load is unbalanced; some processors may need more time in calculating the source terms, resulting in other processors remaining idle during that time. We will address this issue with the development of a decomposition scheme that takes into account the computational load of each processor while performing the reactive computations. Moreover, a periodic redistribution of the computational cells across the domains according to the updated chemical load is expected to increase the efficiency of the methodology.
Beta version of platform architecture and open APIs for multiscale materials modellingEarly version of implementation of modular platform architecture with the development of workflows open application programming interfaces APIs and wrappers enabling users to interchange components eg different DFT codes
Turbulence modules on CatalyticFOAMCatalytic reactors investigated in the case studies of WP5 exhibit turbulent conditions. Thus, the CatalyticFOAM framework will be extended from laminar to turbulent conditions, in the context of the Reynolds-Averaged Navier-Stokes (RANS) approach, by considering the turbulence closure models already available in OpenFOAM. In particular, the attention will be focused on the k-epsilon and k-omega models. Also, efforts will be dedicated to exploit the possibility of applying the Reynolds Stress Model (RSM). Moreover, a critical assessment and validation of the closure models will be carried out in the context of catalytic reactors with particular attention to industrially-relevant conditions and regimes.
EMMO based data model specifications and implementation prototypesIn consultation with SCM and the academic partners, NLeSC will define a formal, machine interpretable definition of data types derived from the ontology, interrelations, simulation tasks and interfaces within the ReaxPro system. They will encode the data generated within ReaxPro into this data framework, together with SCM, UCL, UI and PoliMi. It will be a priority to ensure that data and tools in ReaxPro are findable, accessible, interoperable and reusable (FAIR), leveraging on NLeSC’s expertise on semantic web technologies and FAIR data. These activities will culminate with a data model specification and implementation, spearheaded by NLeSC.
ReaxPro integration apps on MARKETPLACE and VIMMP, use-case studies on EMMC MarketplaceA set of ReaxPro apps on the Marketplaces will be delivered as well as case studies published on the EMMC Marketplace The integration on the MARKETPLACE and VIMMP marketplaces will be spearheaded by Fraunhofer IWM and Fraunhofer IFAM respectively
Machine learning modules for linking or coupling of scalesTo further improve the efficiency of linking or coupling models, we will employ metamodeling approaches in the context of machine learning. The adoption of the general and flexible machine learning framework of T3.4 is expected to provide significant benefits in WP4. For instance, the connection of the electronic and atomistic scales will require a force-field representation, whereby the input is the set of atomic coordinates, and the output is the energy, a real-valued quantity. Connecting the mesoscopic scale (captured in KMC modules) to the continuum scale will require variables such as temperature, pressure and molar fractions of gas species as input, whereas the overall reaction rates (a real-valued vector, possibly with error margins) will be the output. CatalyticFOAM implements the in situ adaptive tabulation scheme (ISAT), which has successfully improved the efficiency of CFD calculations. The scheme, currently implemented as a serial code, essentially builds a meta-model for the reaction terms in a flexible way, allowing for “branches” of the model to be deleted if they are not used anymore. These cases will be implemented in the generic machine learning framework of T3.4, resulting in highly efficient metamodeling tools that will progressively accelerate runs of coupled models. SURFsara and NLeSC will research and develop (possibly neural network-based) approaches to efficiently and accurately identify potential energy surfaces (in collaboration with UI), optimise parameters for DFT, DFTB and ReaxFF methods (together with SCM), and couple KMC models with continuum reactor-scale simulations. We will also experiment with state-of-the-art Deep Tensor Neural Networks for predicting atomic energies and local chemical potentials in molecules, and reliable isomer energies. We will aim at a machine learning-based continuous representation of the chemical compound space that can be combined with the workflow approaches of T3.1 to enable the nearly autonomous study of multistep reactions.
Release version of platform architecture and open APIs for multiscale materials modellingRelease version of implementation of modular platform architecture with the development of workflows, open APIs and wrappers, enabling users to interchange components.
API for accessing repositories on external resourcesPartners Fraunhofer IWM and IFAM will collaborate on providing the access to the marketplace repositories with common API NLeSC with Fraunhofer IWM will develop support for H5CUDS while with IFAM the support for ADF The outcome of these efforts will be an API for accessing repositories on external resources
Zacros version incorporating KMC- acceleration method based on optimisationTime-scale separation is a common issue in KMC simulation, whereby certain frequent events dominate the computational effort. Yet, after rapidly reaching quasi-equilibrium, their frequent execution is unnecessary, as they are anyway limited by the slow events. To improve simulation efficiency we can rescale the frequency of these events, but we need to minimise the loss of accuracy. We will thus formulate an optimisation problem with the kinetic constants of the events as the decision variables. The objective function will contain two contributions: (i) computational expense, quantified by the number of KMC steps performed; (ii) accuracy, estimated by assessing the error in the numbers of molecules of surface/gas species after running low- and high-accuracy simulation instances. The algorithm will minimise both conflicting objectives, also enabling the user to favour speed over accuracy or vice versa.
NEGF-enhanced DFT(B) implementationThe capabilities of the ADF modelling suite to treat efficiently models of catalytic reactions on extended surfaces will be improved using techniques originating from the nonequilibrium Greens function NEGF method for the representation of the semiinfinite bulk material underneath the catalyst
Automated force field parameter fitting implementationA systematic approach will be developed for the automatic generation of highly accurate and reliable Reactive Force-Field parameterisations from the results of DFT calculations, to reduce the computational effort in evaluating the atomic interactions and the system energy.
Computationally efficient and portable implementation of mesoscopic-level homogenisation approachesEfficient kinetic models using high-accuracy homogenisation approaches will be further developed, at the mesoscopic scale. These approaches capture the effect of adsorbate-adsorbate lateral interactions on activation barriers at a much lower computational cost than KMC. This is achieved by treating a small cluster of lattice sites explicitly (e.g. sites up to 2nd nearest-neighbours of the reaction sites), while handling longer-range correlations in a mean-field sense (homogeneous model). Preliminary results on this method have shown exceptional promise, with reductions up to 5 orders of magnitude in computational time compared to KMC. The approach entails the computation of weighted sums (partition functions and average rate constants), involving a large number of terms. The existing computational code will thus be improved by implementing OpenMP parallelisation, for which we expect negligible parallel slowdown due to the nature of the problem, involving sums of independently computed terms.
EON version implementing Gaussian Process regressionThe Gaussian Process regression a machinelearning approach has been used recently to dramatically reduce the computational effort in evaluations of the atomic interactions and the total energy of a given system which require identifying the mechanism of the various possible elementary transitions that can occur in the simulated system and estimating their rate at given conditions such as temperature This approach will be further developed so as to make optimal use of each quantum chemistry or DFT calculation including the Hessian of the energy at the initial state and final states which are also needed for evaluating the transition rate in harmonic transition state theory In some cases more than one minimum energy path connects a given initial and final state In such cases global optimisation of paths can be applied
Modular refactoring of ADF codesDifferent components (including ADF, BAND, DFTB and ReaxFF) will be further modularised with the aim to significantly increase the efficiency of massively parallel runs, as well as to allow for a very flexible usage of individual algorithmic components from within the employed scripting environments.
ReaxPro will take part in the Open Research Data Pilot and the consortium will ensure that our results comply with the H2020 Open Science guidelines and in particular that data generated within the project follow FAIR principles It will deliver a first draft of the Data Management Plan DMP in month 6 based on the preliminary plan specified in Section 224 describing the data management life cycle for the data to be collected processed and generated within ReaxPro Such DMP will be updated over the course of the project reflecting significant changes such as the availability of new kinds of data or changes in the consortium policies or composition As a minimum the DMP will be updated for each periodic evaluation of the project and for the final review
A website will be launched and maintained by Fraunhofer, for the purposes of communicating the activities of ReaxPro throughout the duration of the project and beyond. Content will be contributed by all partners. We will develop a social media strategy (Facebook, LinkedIn, Research Gate, CORDIS, Twitter, YouTube) and press strategy (press releases, news articles, interviews).
ReaxPro MODA diagramsMODA workflows generated with EMMC online tool.
Report on existing platforms, tools and APIs for multiscale materials modellingThe state of the art in terms of platforms, tools and APIs will be analyzed, towards the development of a modular platform architecture for multiscale materials modelling.
Market analysis reportThe consortium will commission an analysis of licensing models and marketplace solutions for materials modelling across the whole value chain.
Publications
Author(s):
Andraž Pavlišič, Matej Huš, Anže Prašnikar, Blaž Likozar
Published in:
Journal of Cleaner Production, Issue 275, 2020, Page(s) 122958, ISSN 0959-6526
Publisher:
Elsevier BV
DOI:
10.1016/j.jclepro.2020.122958
Author(s):
Pomeroy, Brett; Grilc, Miha; Gyergyek, Sašo; Likozar, Blaž
Published in:
Chemical Engineering Journal, Issue 9, 2020, Page(s) publication number 127553, ISSN 1385-8947
Publisher:
Elsevier BV
DOI:
10.1016/j.cej.2020.127553
Author(s):
Ravipati, Srikanth; Savva, Giannis D.; Christidi, Ilektra-Athanasia; Guichard, Roland; Nielsen, Jens; Réocreux, Romain; Stamatakis, Michail
Published in:
Computer Physics Communications, Issue 6, 2021, Page(s) / publication number 108148, ISSN 0010-4655
Publisher:
Elsevier BV
DOI:
10.1016/j.cpc.2021.108148
Author(s):
Pomeroy, Brett; Grilc, Miha; Likozar, Blaž
Published in:
Renewable and Sustainable Energy Reviews, Issue 6, 2021, Page(s) publication number 111748, ISSN 1364-0321
Publisher:
Elsevier BV
DOI:
10.1016/j.rser.2021.111748
Author(s):
Savva, Giannis D.; Benson, Raz L.; Christidi, Ilektra-Athanasia; Stamatakis, Michail
Published in:
Physical Chemistry Chemical Physics, 2023, ISSN 2046-2069
Publisher:
Royal Society of Chemistry
DOI:
10.1039/D2CP04424B
Author(s):
Brett Pomeroy, Miha Grilc, Sašo Gyergyek, Blaž Likozar
Published in:
Applied Catalysis B: Environmental, Issue Volume 334, 2023, Page(s) 122868, ISSN 0926-3373
Publisher:
Elsevier BV
DOI:
10.1016/j.apcatb.2023.122868
Author(s):
Pomeroy, Brett; Grilc, Miha; Likozar, Blaž
Published in:
Green Chemistry, Issue 8, 2021, Page(s) 7996–8002, ISSN 1463-9262
Publisher:
Royal Society of Chemistry
DOI:
10.1039/d1gc02086b
Author(s):
Kojčinović, Aleksa; Kovačič, Žan; Huš, Matej; Likozar, Blaž; Grilc, Miha
Published in:
Applied Surface Science, Issue 3, 2020, Page(s) 1 to 7 /publication number 148836, ISSN 0169-4332
Publisher:
Elsevier BV
DOI:
10.1016/j.apsusc.2020.148836
Author(s):
Daniele Micale, Mauro Bracconi, and Matteo Maestri
Published in:
ACS Engineering Au, 2024, ISSN 2694-2488
Publisher:
American Chemical Society
DOI:
10.1021/acsengineeringau.3c00066
Author(s):
Micale, Daniele; Ferroni, Claudio; Uglietti, Riccardo; Bracconi, Mauro; Maestri, Matteo
Published in:
Chemie Ingenieur Technik, 2022, ISSN 0009-286X
Publisher:
Wiley - V C H Verlag GmbbH & Co.
DOI:
10.1002/cite.202100196
Author(s):
Savva, Giannis D.; Benson, Raz L.; Christidi, Ilektra-Athanasia; Stamatakis, Michail
Published in:
Philosophical Transactions of the Royal Society A, 2023, ISSN 1364-503X
Publisher:
Royal Society of London
DOI:
10.1098/rsta.2022.0235
Author(s):
Daniele Micale; Matteo Maestri; Riccardo Uglietti; Mauro Bracconi
Published in:
Industrial & Engineering Chemistry Research, Issue 8, 2021, Page(s) 6687-6697, ISSN 0888-5885
Publisher:
American Chemical Society
DOI:
10.1021/acs.iecr.0c05845
Author(s):
Žan Kovačič; Blaž Likozar; Matej Huš
Published in:
ACS Catalysis, Issue 10 (24), 2020, Page(s) 14984-15007, ISSN 2155-5435
Publisher:
American Chemical Society
DOI:
10.1021/acscatal.0c02557
Author(s):
M. Žula, M. Grilc, A. Kostyniuk, G. Tofani, E. Jasiukaitytė-Grojzdek, T. Ročnik Kozmelj, R. Kumar Chowdari, Žan Lavrič, J. Teržan, B. Hočevar, A. Jakob, E. Rakić, B. Pomeroy, M. M. Cajnko, F. A. Vicente, D. Marinič, A. Oberlintner, U. Novak, D. Benedetto Tiz, M. Huš, B. Likozar
Published in:
Chimia, Issue 77, 2023, Page(s) 816, ISSN 0009-4293
Publisher:
Schweizerische Chemische Gedellschaft
DOI:
10.2533/chimia.2023.816
Author(s):
M. Pineda; M. Stamatakis
Published in:
Crossref, 2022, ISSN 0021-9606
Publisher:
American Institute of Physics
DOI:
10.1063/5.0083251
Author(s):
Srikanth Ravipati, Mayeul d’Avezac, Jens Nielsen, James Hetherington, Michail Stamatakis
Published in:
The Journal of Physical Chemistry A, Issue 124/35, 2020, Page(s) 7140-7154, ISSN 1089-5639
Publisher:
American Chemical Society
DOI:
10.1021/acs.jpca.0c03571
Author(s):
Peña-Torres, Alejandro; Ali, Abid; Stamatakis, Michail; Jónsson, Hannes
Published in:
Physical Review B, 2022, ISSN 1098-0121
Publisher:
American Physical Society
DOI:
10.1103/physrevb.105.205411
Author(s):
Žan Kovačič, Blaž Likozar, Matej Huš
Published in:
Fuel, Issue Volume 328, 2022, Page(s) 125322, ISSN 0016-2361
Publisher:
Elsevier BV
DOI:
10.1016/j.fuel.2022.125322
Author(s):
Schumann, J., Stamatakis, M., Michaelides, A. et al.
Published in:
Nature Chemistry, 2024, ISSN 1755-4330
Publisher:
Nature Publishing Group
DOI:
10.1038/s41557-023-01424-6
Author(s):
Yorick L. A. Schmerwitz, Vilhjálmur Ásgeirsson, and Hannes Jónsson
Published in:
Journal of Chemical Theory and Computation, Issue 20 (1), 2024, Page(s) 155-163, ISSN 2210-271X
Publisher:
Elsevier BV
DOI:
10.1021/acs.jctc.3c01111
Author(s):
Mollinga, Joris; Codreanu, Valeriu
Published in:
Lecture Notes in Networks and Systems/ Book: Intelligent Computing, Issue 3, 2021, Page(s) 102-117, ISSN 2367-3389
Publisher:
Springer
DOI:
10.1007/978-3-030-80119-9_4
Searching for OpenAIRE data...
There was an error trying to search data from OpenAIRE
No results available