{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T06:28:10Z","timestamp":1772692090229,"version":"3.50.1"},"reference-count":39,"publisher":"Oxford University Press (OUP)","issue":"10","license":[{"start":{"date-parts":[[2025,9,29]],"date-time":"2025-09-29T00:00:00Z","timestamp":1759104000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","award":["RS-2024-00355857"],"award-info":[{"award-number":["RS-2024-00355857"]}],"id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","award":["RS-2024-00410810"],"award-info":[{"award-number":["RS-2024-00410810"]}],"id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","award":["2022-0-00969"],"award-info":[{"award-number":["2022-0-00969"]}],"id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","award":["2022-0-00986"],"award-info":[{"award-number":["2022-0-00986"]}],"id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","award":["GTL24031-000"],"award-info":[{"award-number":["GTL24031-000"]}],"id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","award":["RS-2024-00410810"],"award-info":[{"award-number":["RS-2024-00410810"]}],"id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100014188","name":"Ministry of Science and ICT, South Korea","doi-asserted-by":"publisher","award":["RS-2025-02317327"],"award-info":[{"award-number":["RS-2025-02317327"]}],"id":[{"id":"10.13039\/501100014188","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100008783","name":"National Research Council of Science & Technology","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100008783","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100010663","name":"ERC","doi-asserted-by":"publisher","award":["2021-CoG-101043998"],"award-info":[{"award-number":["2021-CoG-101043998"]}],"id":[{"id":"10.13039\/100010663","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,11]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>The integration of predictive AI into computational design and engineering workflows is often hindered by the challenge of long-term forecasting performance. This study presents a robust framework for AI-accelerated flow simulation, specifically addressing the critical issue of error accumulation in auto-regressive (AR) surrogate models, which is a key bottleneck for their practical use in the design cycle. We introduce the first implementation of the two-step Adams\u2013Bashforth method specifically tailored for data-driven AR prediction, leveraging historical derivative information to enhance numerical stability without additional computational overhead. To validate our approach, we systematically evaluate time integration schemes across canonical two-dimensional partial differential equations (advection, heat, and Burgers\u2019 equations) before extending to complex Navier\u2013Stokes cylinder vortex shedding dynamics. Additionally, we develop three novel adaptive weighting strategies that dynamically adjust the importance of different future time-steps during multi-step rollout training. Our analysis reveals that as physical complexity increases, such sophisticated rollout techniques become essential, with the Adams\u2013Bashforth scheme demonstrating consistent robustness across investigated systems and our best adaptive approach delivering an 89% improvement over conventional fixed-weight methods while maintaining similar computational costs. For the complex Navier\u2013Stokes vortex shedding problem, despite using an extremely lightweight graph neural network with just 1177 trainable parameters and training on only 50 snapshots, our framework accurately predicts 350 future time-steps\u2014a 7:1 prediction-to-training ratio\u2014reducing mean squared error from 0.125 (single-step direct prediction) to 0.002 (Adams\u2013Bashforth with proposed multi-step rollout). Our integrated methodology demonstrates an 83% improvement over standard noise injection techniques and maintains robustness under severe spatial constraints; specifically, when trained on only a partial spatial domain, it still achieves 58% and 27% improvements over direct prediction and forward Euler methods, respectively. Our framework\u2019s model-agnostic design, operating at the fundamental level of AR prediction mechanics, enables direct integration with any neural network architecture without requiring model-specific modifications, introducing a versatile solution for robust long-term spatio-temporal predictions across various engineering disciplines.<\/jats:p>","DOI":"10.1093\/jcde\/qwaf099","type":"journal-article","created":{"date-parts":[[2025,9,29]],"date-time":"2025-09-29T12:59:45Z","timestamp":1759150785000},"page":"133-153","source":"Crossref","is-referenced-by-count":2,"title":["Model-agnostic AI framework with explicit time integration for long-term fluid dynamics prediction"],"prefix":"10.1093","volume":"12","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-2065-9139","authenticated-orcid":false,"given":"Sunwoong","family":"Yang","sequence":"first","affiliation":[{"name":"Cho Chun Shik Graduate School of Mobility, Korea Advanced Institute of Science and Technology , Daejeon 34051 ,","place":["Republic of Korea"]}]},{"given":"Ricardo","family":"Vinuesa","sequence":"additional","affiliation":[{"name":"University of Michigan Department of Aerospace Engineering, , Ann Arbor MI 48109 ,","place":["United States"]},{"name":"FLOW, Engineering Mechanics, KTH Royal Institute of Technology , Stockholm SE-100 44 ,","place":["Sweden"]}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3475-7477","authenticated-orcid":false,"given":"Namwoo","family":"Kang","sequence":"additional","affiliation":[{"name":"Cho Chun Shik Graduate School of Mobility, Korea Advanced Institute of Science and Technology , Daejeon 34051 ,","place":["Republic of Korea"]},{"name":"Narnia Labs , Daejeon 34051 ,","place":["Republic of Korea"]}]}],"member":"286","published-online":{"date-parts":[[2025,9,29]]},"reference":[{"key":"2025102112264084500_bib1","doi-asserted-by":"crossref","first-page":"qwaf055","DOI":"10.1093\/jcde\/qwaf055","article-title":"Nonlinear acoustic multicavitation with external ultrasound field in complex fluids: Numerical investigation","volume":"12","author":"AbuShaeer","year":"2025","journal-title":"Journal of Computational Design and Engineering"},{"key":"2025102112264084500_bib2","doi-asserted-by":"publisher","first-page":"689","DOI":"10.1016\/j.apr.2018.11.006","article-title":"Statistical models for multi-step-ahead forecasting of fine particulate matter in urban areas","volume":"10","author":"Ahani","year":"2019","journal-title":"Atmospheric Pollution Research"},{"key":"2025102112264084500_bib3","doi-asserted-by":"publisher","first-page":"115902","DOI":"10.1016\/j.cma.2023.115902","article-title":"Physics-integrated neural differentiable (pindiff) model for composites manufacturing","volume":"406","author":"Akhare","year":"2023","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"2025102112264084500_bib4","doi-asserted-by":"publisher","first-page":"5332","DOI":"10.1016\/j.eswa.2011.11.002","article-title":"A new hybrid for improvement of auto-regressive integrated moving average models applying particle swarm optimization","volume":"39","author":"Asadi","year":"2012","journal-title":"Expert Systems with Applications"},{"key":"2025102112264084500_bib5","doi-asserted-by":"publisher","first-page":"925","DOI":"10.1016\/j.eswa.2006.10.034","article-title":"Forecast approach using neural network adaptation to support vector regression grey model and generalized auto-regressive conditional heteroscedasticity","volume":"34","author":"Chang","year":"2008","journal-title":"Expert Systems with Applications"},{"key":"2025102112264084500_bib6","doi-asserted-by":"publisher","first-page":"330","DOI":"10.1016\/j.jcde.2016.06.001","article-title":"Heat and mass transfer of a second grade magnetohydrodynamic fluid over a convectively heated stretching sheet","volume":"3","author":"Das","year":"2016","journal-title":"Journal of Computational Design and Engineering"},{"key":"2025102112264084500_bib7","doi-asserted-by":"publisher","first-page":"71","DOI":"10.1016\/0010-0277(93)90058-4","article-title":"Learning and development in neural networks: The importance of starting small","volume":"48","author":"Elman","year":"1993","journal-title":"Cognition"},{"key":"2025102112264084500_bib8","first-page":"2083","article-title":"Graph u-nets","volume-title":"International Conference on Machine Learning","author":"Gao","year":"2019"},{"key":"2025102112264084500_bib9","doi-asserted-by":"publisher","first-page":"117023","DOI":"10.1016\/j.cma.2024.117023","article-title":"Bayesian conditional diffusion models for versatile spatiotemporal turbulence generation","volume":"427","author":"Gao","year":"2024","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"2025102112264084500_bib10","doi-asserted-by":"publisher","first-page":"065501","DOI":"10.1088\/1873-7005\/abb91d","article-title":"Cnn-lstm based reduced order modeling of two-dimensional unsteady flows around a circular cylinder at different reynolds numbers","volume":"52","author":"Hasegawa","year":"2020","journal-title":"Fluid Dynamics Research"},{"key":"2025102112264084500_bib11","doi-asserted-by":"publisher","first-page":"62","DOI":"10.1093\/jcde\/qwaf041","article-title":"Integrating artificial intelligence and machine learning with numerical simulation for enhanced thermal performance of ternary nanofluid","volume":"12","author":"Hussain","year":"2025","journal-title":"Journal of Computational Design and Engineering"},{"key":"2025102112264084500_bib12","doi-asserted-by":"publisher","first-page":"10770","DOI":"10.1002\/er.7879","article-title":"Finite volume method network for the acceleration of unsteady computational fluid dynamics: Non-reacting and reacting flows","volume":"46","author":"Jeon","year":"2022","journal-title":"International Journal of Energy Research"},{"key":"2025102112264084500_bib13","doi-asserted-by":"publisher","first-page":"124900","DOI":"10.1016\/j.ijheatmasstransfer.2023.124900","article-title":"Residual-based physics-informed transfer learning: A hybrid method for accelerating long-term cfd simulations via deep learning","volume":"220","author":"Jeon","year":"2024","journal-title":"International Journal of Heat and Mass Transfer"},{"key":"2025102112264084500_bib14","doi-asserted-by":"publisher","first-page":"126907","DOI":"10.1016\/j.eswa.2025.126907","article-title":"Physics-constrained graph neural networks for spatio-temporal prediction of drop impact on OLED display panels","volume":"274","author":"Kim","year":"2025","journal-title":"Expert Systems with Applications"},{"key":"2025102112264084500_bib15","doi-asserted-by":"publisher","first-page":"217","DOI":"10.1017\/jfm.2019.700","article-title":"Data-driven prediction of unsteady flow over a circular cylinder using deep learning","volume":"879","author":"Lee","year":"2019","journal-title":"Journal of Fluid Mechanics"},{"key":"2025102112264084500_bib16","article-title":"Fourier neural operator for parametric partial differential equations","author":"Li","year":"2020"},{"key":"2025102112264084500_bib17","doi-asserted-by":"publisher","first-page":"117441","DOI":"10.1016\/j.cma.2024.117441","article-title":"Differentiability in unrolled training of neural physics simulators on transient dynamics","volume":"433","author":"List","year":"2025","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"2025102112264084500_bib18","article-title":"A causality-deeponet for causal responses of linear dynamical systems","author":"Liu","year":"2022"},{"key":"2025102112264084500_bib19","doi-asserted-by":"publisher","first-page":"31","DOI":"10.1038\/s42005-024-01521-z","article-title":"Multi-resolution partial differential equations preserved learning framework for spatiotemporal dynamics","volume":"7","author":"Liu","year":"2024","journal-title":"Communications Physics"},{"key":"2025102112264084500_bib20","doi-asserted-by":"publisher","first-page":"218","DOI":"10.1038\/s42256-021-00302-5","article-title":"Learning nonlinear operators via deeponet based on the universal approximation theorem of operators","volume":"3","author":"Lu","year":"2021","journal-title":"Nature Machine Intelligence"},{"key":"2025102112264084500_bib21","article-title":"Transolver++: An accurate neural solver for pdes on million-scale geometries","author":"Luo","year":"2025"},{"key":"2025102112264084500_bib22","doi-asserted-by":"publisher","first-page":"1136","DOI":"10.1109\/SMC53992.2023.10394294","article-title":"Causal deep operator networks for data-driven modeling of dynamical systems","volume-title":"2023 IEEE International Conference on Systems, Man, and Cybernetics (SMC)","author":"Nghiem","year":"2023"},{"key":"2025102112264084500_bib23","doi-asserted-by":"publisher","first-page":"045020","DOI":"10.1088\/2632-2153\/ac1fc9","article-title":"Graph convolutional networks applied to unstructured flow field data","volume":"2","author":"Ogoke","year":"2021","journal-title":"Machine Learning: Science and Technology"},{"key":"2025102112264084500_bib24","article-title":"Learning mesh-based simulation with graph networks","author":"Pfaff","year":"2020"},{"key":"2025102112264084500_bib25","doi-asserted-by":"publisher","first-page":"686","DOI":"10.1016\/j.jcp.2018.10.045","article-title":"Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations","volume":"378","author":"Raissi","year":"2019","journal-title":"Journal of Computational physics"},{"key":"2025102112264084500_bib26","doi-asserted-by":"publisher","first-page":"109316","DOI":"10.1016\/j.asoc.2022.109316","article-title":"Multi-output spatio-temporal air pollution forecasting using neural network approach","volume":"126","author":"Samal","year":"2022","journal-title":"Applied Soft Computing"},{"key":"2025102112264084500_bib27","first-page":"8459","article-title":"Learning to simulate complex physics with graph networks","volume-title":"International conference on machine learning","author":"Sanchez-Gonzalez","year":"2020"},{"key":"2025102112264084500_bib28","doi-asserted-by":"publisher","first-page":"1361","DOI":"10.1038\/s41467-024-45578-4","article-title":"$\\beta$-variational autoencoders and transformers for reduced-order modelling of fluid flows","volume":"15","author":"Solera-Rico","year":"2024","journal-title":"Nature Communications"},{"key":"2025102112264084500_bib29","doi-asserted-by":"publisher","first-page":"7067","DOI":"10.1016\/j.eswa.2012.01.039","article-title":"A review and comparison of strategies for multi-step ahead time series forecasting based on the nn5 forecasting competition","volume":"39","author":"Taieb","year":"2012","journal-title":"Expert Systems with Applications"},{"key":"2025102112264084500_bib30","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v29i1.9590","article-title":"Improving multi-step prediction of learned time series models","volume-title":"Proceedings of the AAAI Conference on Artificial Intelligence","author":"Venkatraman","year":"2015"},{"key":"2025102112264084500_bib31","doi-asserted-by":"publisher","first-page":"025013","DOI":"10.1088\/2632-2153\/abc327","article-title":"Automated multi-layer optical design via deep reinforcement learning","volume":"2","author":"Wang","year":"2021","journal-title":"Machine Learning: Science and Technology"},{"key":"2025102112264084500_bib32","doi-asserted-by":"publisher","first-page":"116813","DOI":"10.1016\/j.cma.2024.116813","article-title":"Respecting causality for training physics-informed neural networks","volume":"421","author":"Wang","year":"2024","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"2025102112264084500_bib33","doi-asserted-by":"crossref","first-page":"4184","DOI":"10.1145\/3534678.3539045","article-title":"Learning large-scale subsurface simulations with a hybrid graph network simulator","volume-title":"Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining","author":"Wu","year":"2022"},{"key":"2025102112264084500_bib34","doi-asserted-by":"publisher","first-page":"117075","DOI":"10.1016\/j.cma.2024.117075","article-title":"Data-driven physics-informed neural networks: A digital twin perspective","volume":"428","author":"Yang","year":"2024","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"2025102112264084500_bib35","article-title":"Enhancing graph u-nets for mesh-agnostic spatio-temporal flow prediction","author":"Yang","year":"2024"},{"key":"2025102112264084500_bib36","doi-asserted-by":"publisher","first-page":"118254","DOI":"10.1016\/j.cma.2025.118254","article-title":"Data-Efficient Deep Operator Network for Unsteady Flow: A Multi-Fidelity Approach with Physics-Guided Subsampling","volume-title":"Computer Methods in Applied Mechanics and Engineering","author":"Yang","year":"2025"},{"key":"2025102112264084500_bib37","doi-asserted-by":"publisher","first-page":"125715","DOI":"10.1016\/j.eswa.2024.125715","article-title":"Goal-driven long-term marine vessel trajectory prediction with a memory-enhanced network","volume":"263","author":"Zhang","year":"2025","journal-title":"Expert Systems with Applications"},{"key":"2025102112264084500_bib38","doi-asserted-by":"publisher","first-page":"117990","DOI":"10.1016\/j.cma.2025.117990","article-title":"Predicting change, not states: An alternate framework for neural PDE surrogates","volume":"441","author":"Zhou","year":"2025","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"2025102112264084500_bib39","article-title":"Strategies for pretraining neural operators","author":"Zhou","year":"2024"}],"container-title":["Journal of Computational Design and Engineering"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/jcde\/advance-article-pdf\/doi\/10.1093\/jcde\/qwaf099\/64427124\/qwaf099.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jcde\/advance-article-pdf\/doi\/10.1093\/jcde\/qwaf099\/64427124\/qwaf099.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jcde\/advance-article-pdf\/doi\/10.1093\/jcde\/qwaf099\/64427124\/qwaf099.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,21]],"date-time":"2025-10-21T16:26:49Z","timestamp":1761064009000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/jcde\/article\/12\/10\/133\/8267850"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9,29]]},"references-count":39,"journal-issue":{"issue":"10","published-print":{"date-parts":[[2025,10,11]]}},"URL":"https:\/\/doi.org\/10.1093\/jcde\/qwaf099","relation":{},"ISSN":["2288-5048"],"issn-type":[{"value":"2288-5048","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2025,10]]},"published":{"date-parts":[[2025,9,29]]}}}