{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,9]],"date-time":"2026-03-09T17:47:47Z","timestamp":1773078467274,"version":"3.50.1"},"reference-count":53,"publisher":"IOP Publishing","issue":"3","license":[{"start":{"date-parts":[[2024,10,1]],"date-time":"2024-10-01T00:00:00Z","timestamp":1727740800000},"content-version":"vor","delay-in-days":30,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2024,10,1]],"date-time":"2024-10-01T00:00:00Z","timestamp":1727740800000},"content-version":"tdm","delay-in-days":30,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"crossref","award":["92152301"],"award-info":[{"award-number":["92152301"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2024,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>In recent years, the synergy between artificial intelligence and turbulence big data has given rise to a new data-driven paradigm in turbulence research. Data-driven turbulence modeling has emerged as one of the forefront directions in fluid mechanics. Most existing studies focus on feature construction, selection, and the development of modeling frameworks, often overlooking the practical deployment and application of trained models. This paper examines the entire process from model construction to real-world deployment, using data-driven turbulence modeling for high Reynolds number flows over complex three-dimensional configurations as a case study. Key stages include data generation, input-output feature construction, model training, model compilation and optimization, deployment, and validation. We successfully implemented the entire workflow in a heterogeneous supercomputing environment and, through mixed programming techniques, integrated the resulting turbulence model into the Platform for Hybrid Engineering Simulation of Flows (PHengLEI) open-source software framework. This allowed for mixed-precision simulations, with the main equations solved in double precision and the turbulence model in half precision. The new computational framework was validated through large-scale parallel numerical simulations on grids with tens of millions of elements for three-dimensional complex configurations. The results highlight the efficiency of our model deployment, with overall computational efficiency improving by 13.35% and the turbulence model\u2019s solution speed increasing by approximately 3.9 times. The accuracy of the computations was also confirmed, with the average relative error in the lift and drag coefficients calculated by the data-driven turbulence model within 3%. Across various computing nodes, the relative error in the computed aerodynamic coefficients remained within 1%, demonstrating the framework\u2019s scalability. Notably, our contributions have been incorporated as a case study in the latest PHengLEI open-source project<jats:sup>5<\/jats:sup>\n                  <jats:fn id=\"mlstad7d60fn2\">\n                     <jats:label>5<\/jats:label>\n                     <jats:p>\n                        <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/forge.osredm.com\/PHengLEI\/PHengLEI-TestCases\/tree\/master\/Y02_ThreeD_M6_Unstruct_Branch_Ascend\">https:\/\/forge.osredm.com\/PHengLEI\/PHengLEI-TestCases\/tree\/master\/Y02_ThreeD_M6_Unstruct_Branch_Ascend<\/jats:ext-link>.<\/jats:p>\n                  <\/jats:fn>.<\/jats:p>","DOI":"10.1088\/2632-2153\/ad7d60","type":"journal-article","created":{"date-parts":[[2024,9,19]],"date-time":"2024-09-19T22:58:21Z","timestamp":1726786701000},"page":"035085","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":5,"title":["Development and deployment of data-driven turbulence model for three-dimensional complex configurations"],"prefix":"10.1088","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-5197-4978","authenticated-orcid":true,"given":"Xuxiang","family":"Sun","sequence":"first","affiliation":[]},{"given":"Yilang","family":"Liu","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7799-833X","authenticated-orcid":true,"given":"Weiwei","family":"Zhang","sequence":"additional","affiliation":[]},{"given":"Yongzhong","family":"Wang","sequence":"additional","affiliation":[]},{"given":"Jingyuan","family":"Zou","sequence":"additional","affiliation":[]},{"given":"Zhengrong","family":"Han","sequence":"additional","affiliation":[]},{"given":"Yun","family":"Su","sequence":"additional","affiliation":[]}],"member":"266","published-online":{"date-parts":[[2024,10,1]]},"reference":[{"key":"mlstad7d60bib1","author":"Pope","year":"2000"},{"key":"mlstad7d60bib2","author":"Slotnick","year":"2014"},{"key":"mlstad7d60bib3","doi-asserted-by":"publisher","first-page":"77","DOI":"10.1146\/annurev-fluid-122316-045020","article-title":"Some recent developments in turbulence closure modeling","volume":"50","author":"Durbin","year":"2018","journal-title":"Annu. Rev. Fluid Mech."},{"key":"mlstad7d60bib4","doi-asserted-by":"publisher","first-page":"357","DOI":"10.1146\/annurev-fluid-010518-040547","article-title":"Turbulence modeling in the age of data","volume":"51","author":"Duraisamy","year":"2019","journal-title":"Annu. Rev. Fluid Mech."},{"key":"mlstad7d60bib5","doi-asserted-by":"publisher","first-page":"444","DOI":"10.7638\/kqdlxxb-2019.0036","article-title":"Progresses in the application of machine learning in turbulence modeling","volume":"37","author":"Weiwei","year":"2019","journal-title":"Acta Aerodynamica Sinica"},{"key":"mlstad7d60bib6","doi-asserted-by":"publisher","DOI":"10.1016\/j.advengsoft.2023.103525","article-title":"A physics-informed neural network based simulation tool for reacting flow with multicomponent reactants","volume":"185","author":"Sun","year":"2023","journal-title":"Adv. Eng. Softw."},{"key":"mlstad7d60bib7","doi-asserted-by":"publisher","DOI":"10.1063\/5.0022561","article-title":"Feature selection and processing of turbulence modeling based on an artificial neural network","volume":"32","author":"Yin","year":"2020","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib8","doi-asserted-by":"publisher","first-page":"155","DOI":"10.1017\/jfm.2016.615","article-title":"Reynolds averaged turbulence modelling using deep neural networks with embedded invariance","volume":"807","author":"Ling","year":"2016","journal-title":"J. Fluid Mech."},{"key":"mlstad7d60bib9","doi-asserted-by":"publisher","first-page":"1153","DOI":"10.1007\/s42241-019-0089-y","article-title":"Recent progress in augmenting turbulence models with physics-informed machine learning","volume":"31","author":"Zhang","year":"2019","journal-title":"J. Hydrodyn."},{"key":"mlstad7d60bib10","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevFluids.8.084604","article-title":"Enhancing the shear-stress-transport turbulence model with symbolic regression: a generalizable and interpretable data-driven approach","volume":"8","author":"Wu","year":"2023","journal-title":"Phys. Rev. Fluids"},{"key":"mlstad7d60bib11","doi-asserted-by":"crossref","DOI":"10.2514\/6.2015-1284","article-title":"New approaches in turbulence and transition modeling using data-driven techniques","author":"Duraisamy","year":"2015"},{"key":"mlstad7d60bib12","doi-asserted-by":"crossref","DOI":"10.2514\/6.2017-0993","article-title":"Augmentation of turbulence models using field inversion and machine learning","author":"Singh","year":"2017"},{"key":"mlstad7d60bib13","doi-asserted-by":"publisher","DOI":"10.1063\/1.4947045","article-title":"Using field inversion to quantify functional errors in turbulence closures","volume":"28","author":"Singh","year":"2016","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib14","doi-asserted-by":"publisher","first-page":"2215","DOI":"10.2514\/1.J055595","article-title":"Machine-learning-augmented predictive modeling of turbulent separated flows over airfoils","volume":"55","author":"Singh","year":"2017","journal-title":"AIAA J."},{"key":"mlstad7d60bib15","doi-asserted-by":"publisher","first-page":"758","DOI":"10.1016\/j.jcp.2015.11.012","article-title":"A paradigm for data-driven predictive modeling using field inversion and machine learning","volume":"305","author":"Parish","year":"2016","journal-title":"J. Comput. Phys."},{"key":"mlstad7d60bib16","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijheatfluidflow.2022.109004","article-title":"Data-driven turbulence modeling in separated flows considering physical mechanism analysis","volume":"96","author":"Yan","year":"2022","journal-title":"Int. J. Heat Fluid Flow"},{"key":"mlstad7d60bib17","doi-asserted-by":"crossref","DOI":"10.2514\/6.2022-0937","article-title":"In search of data-driven improvements to RANS models applied to separated flows","author":"Rumsey","year":"2022"},{"key":"mlstad7d60bib18","doi-asserted-by":"publisher","DOI":"10.1063\/5.0097438","article-title":"Data augmented turbulence modeling for three-dimensional separation flows","volume":"34","author":"Yan","year":"2022","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib19","doi-asserted-by":"publisher","DOI":"10.1063\/5.0008493","article-title":"Improving the k-omega-gamma-A(r) transition model by the field inversion and machine learning framework","volume":"32","author":"Yang","year":"2020","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib20","doi-asserted-by":"publisher","DOI":"10.1063\/5.0136420","article-title":"A unified method of data assimilation and turbulence modeling for separated flows at high Reynolds numbers","volume":"35","author":"Wang","year":"2023","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib21","doi-asserted-by":"publisher","first-page":"3526","DOI":"10.2514\/1.J062664","article-title":"Learning neural-network-based turbulence models for external transonic flows using ensemble Kalman Method","volume":"61","author":"Liu","year":"2023","journal-title":"AIAA J."},{"key":"mlstad7d60bib22","doi-asserted-by":"publisher","DOI":"10.1016\/j.ast.2023.108632","article-title":"Physical interpretation of neural network-based nonlinear eddy viscosity models","volume":"142","author":"Zhang","year":"2023","journal-title":"Aerosp. Sci. Technol."},{"key":"mlstad7d60bib23","doi-asserted-by":"publisher","DOI":"10.1063\/1.5061693","article-title":"Machine learning methods for turbulence modeling in subsonic flows around airfoils","volume":"31","author":"Zhu","year":"2019","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib24","doi-asserted-by":"publisher","DOI":"10.1016\/j.ast.2020.106452","article-title":"Turbulence closure for high Reynolds number airfoil flows by deep neural networks","volume":"110","author":"Zhu","year":"2021","journal-title":"Aerosp. Sci. Technol."},{"key":"mlstad7d60bib25","doi-asserted-by":"publisher","DOI":"10.1016\/j.compfluid.2021.105298","article-title":"High Reynolds number airfoil turbulence modeling method based on machine learning technique","volume":"236","author":"Sun","year":"2022","journal-title":"Comput. Fluids"},{"key":"mlstad7d60bib26","doi-asserted-by":"publisher","DOI":"10.1063\/5.0076273","article-title":"Analysis on numerical stability and convergence of Reynolds averaged Navier\u2013Stokes simulations from the perspective of coupling modes","volume":"34","author":"Liu","year":"2022","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib27","doi-asserted-by":"publisher","first-page":"22","DOI":"10.1016\/j.jcp.2016.05.003","article-title":"Machine learning strategies for systems with invariance properties","volume":"318","author":"Ling","year":"2016","journal-title":"J. Comput. Phys."},{"key":"mlstad7d60bib28","doi-asserted-by":"crossref","DOI":"10.1109\/ICMLA.2015.38","article-title":"Using machine learning to understand and mitigate model form uncertainty in turbulence models","author":"Ling","year":"2015"},{"key":"mlstad7d60bib29","doi-asserted-by":"publisher","DOI":"10.1016\/j.jcp.2019.108910","article-title":"Deep neural networks for data-driven LES closure models","volume":"398","author":"Beck","year":"2019","journal-title":"J. Comput. Phys."},{"key":"mlstad7d60bib30","doi-asserted-by":"crossref","DOI":"10.1103\/PhysRevFluids.2.054604","article-title":"Searching for turbulence models by artificial neural network","volume":"2","author":"Gamahara","year":"2017","journal-title":"Phys. Rev. Fluids"},{"key":"mlstad7d60bib31","article-title":"Neural network-based modelling of unresolved stresses in a turbulent reacting flow with mean shear","author":"Nikolaou","year":"2019"},{"key":"mlstad7d60bib32","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1017\/jfm.2016.803","article-title":"Deep learning in fluid dynamics","volume":"814","author":"Kutz","year":"2017","journal-title":"J. Fluid Mech."},{"key":"mlstad7d60bib33","doi-asserted-by":"publisher","first-page":"51","DOI":"10.1109\/MCE.2022.3181759","article-title":"Lightweight deep learning: an overview","volume":"13","author":"Wang","year":"2022","journal-title":"IEEE Consum. Electron. Mag."},{"key":"mlstad7d60bib34","doi-asserted-by":"crossref","DOI":"10.1109\/ICCV.2017.298","article-title":"Learning efficient convolutional networks through network slimming","author":"Liu","year":"2017"},{"key":"mlstad7d60bib35","article-title":"Tensor comprehensions: framework-agnostic high-performance machine learning abstractions","author":"Vasilache","year":"2018"},{"key":"mlstad7d60bib36","article-title":"TVM: end-to-end optimization stack for deep learning","author":"Chen","year":"2018"},{"key":"mlstad7d60bib37","doi-asserted-by":"crossref","DOI":"10.1145\/3079856.3080246","article-title":"In-datacenter performance analysis of a tensor processing unit","author":"Jouppi","year":"2017"},{"key":"mlstad7d60bib38","doi-asserted-by":"publisher","first-page":"217","DOI":"10.1016\/bs.adcom.2020.11.001","article-title":"Architecture of neural processing unit for deep neural networks","volume":"vol 122","author":"Lee","year":"2021"},{"key":"mlstad7d60bib39","doi-asserted-by":"publisher","first-page":"879","DOI":"10.1109\/JPROC.2008.917757","article-title":"GPU computing","volume":"96","author":"Owens","year":"2008","journal-title":"Proc. IEEE"},{"key":"mlstad7d60bib40","doi-asserted-by":"publisher","first-page":"1909","DOI":"10.1002\/fld.5125","article-title":"Research on grid\u2010dependence of neural network turbulence model","volume":"94","author":"Song","year":"2022","journal-title":"Int. J. Numer. Methods Fluids"},{"key":"mlstad7d60bib41","doi-asserted-by":"publisher","first-page":"137","DOI":"10.1007\/978-981-19-2879-6_5","article-title":"Huawei MindSpore AI development framework","author":"Huawei Technologies CO. L.","year":"2023"},{"key":"mlstad7d60bib42","doi-asserted-by":"publisher","first-page":"210","DOI":"10.3969\/j.issn.1007-130X.2020.02.004","article-title":"Design of general CFD software PHenglei","volume":"42","author":"Zhong","year":"2020","journal-title":"Comput. Eng. Sci."},{"key":"mlstad7d60bib43","doi-asserted-by":"publisher","first-page":"2368","DOI":"10.11897\/SP.J.1016.2019.02368.","article-title":"PHengLEI: a large scale parallel CFD framework for arbitrary grids","volume":"42","author":"Zhong","year":"2019","journal-title":"Chin. J. Comput."},{"key":"mlstad7d60bib44","doi-asserted-by":"crossref","DOI":"10.2514\/6.1992-439","article-title":"A One-equation Turbulence Model for Aerodynamic Flows","author":"Spalart","year":"1992"},{"key":"mlstad7d60bib45","author":"Schmitt","year":"1979"},{"key":"mlstad7d60bib46","doi-asserted-by":"crossref","DOI":"10.2514\/6.2007-254","article-title":"Statistical analysis of CFD solutions from the third AIAA drag prediction workshop","author":"Morrison","year":"2007"},{"key":"mlstad7d60bib47","doi-asserted-by":"publisher","first-page":"762","DOI":"10.2514\/1.30571","article-title":"CFL3D\/OVERFLOW results for DLR-F6 wing\/body and drag prediction workshop wing","volume":"45","author":"Sclafani","year":"2008","journal-title":"J. Aircr."},{"key":"mlstad7d60bib48","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s42774-021-00088-5","article-title":"Generalization enhancement of artificial neural network for turbulence closure by feature selection","volume":"4","author":"Zhu","year":"2022","journal-title":"Adv. Aerodyn."},{"key":"mlstad7d60bib49","doi-asserted-by":"publisher","DOI":"10.1063\/1.4927765","article-title":"Evaluation of machine learning algorithms for prediction of regions of high Reynolds averaged Navier Stokes uncertainty","volume":"27","author":"Ling","year":"2015","journal-title":"Phys. Fluids"},{"key":"mlstad7d60bib50","doi-asserted-by":"publisher","first-page":"3883","DOI":"10.2514\/1.J062711","article-title":"Turbulence modeling via data assimilation and machine learning for separated flows over airfoils","volume":"61","author":"Shan","year":"2023","journal-title":"AIAA J."},{"key":"mlstad7d60bib51","article-title":"Rezero is all you need: fast convergence at large depth","author":"Bachlechner","year":"2021"},{"key":"mlstad7d60bib52","author":"Chen","year":"2021"},{"key":"mlstad7d60bib53","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,10,1]],"date-time":"2024-10-01T08:14:23Z","timestamp":1727770463000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad7d60"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,9,1]]},"references-count":53,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2024,10,1]]},"published-print":{"date-parts":[[2024,9,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/ad7d60","relation":{},"ISSN":["2632-2153"],"issn-type":[{"value":"2632-2153","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,9,1]]},"assertion":[{"value":"Development and deployment of data-driven turbulence model for three-dimensional complex configurations","name":"article_title","label":"Article Title"},{"value":"Machine Learning: Science and Technology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2024 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2024-05-04","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-09-19","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-10-01","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}