首页|New Findings from Stanford University in the Area of Machine Learning Reported ( A Physics-informed Machine Learning Model for the Prediction of Drop Breakup In Two-phase Flows)
New Findings from Stanford University in the Area of Machine Learning Reported ( A Physics-informed Machine Learning Model for the Prediction of Drop Breakup In Two-phase Flows)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning DailyNews Daily News-Data detailed on Machine Learning have been presented. According to news originatingfrom Stanford, California, by New sRx correspondents, research stated, "Predictive simulations of two-phaseflows are highly sought after because of their widespread applications in propulsion, energy, agriculture,and medicine. One crucial goal for many of these simulation s is the accurate and efficient prediction ofthe size distribution and number d ensity of atomized drops."Financial support for this research came from Advanced Simulation and Computing (ASC) programof the US Department of Energy's National Nuclear Security Adminis tration (NNSA) via the PSAAP-IIICenter at Stanford University.Our news journalists obtained a quote from the research from Stanford University , "The multi-scalenature of these flows makes it practically impossible to capt ure all scales within a single simulation. Inparticular, the breakup processes producing the smallest drops through secondary breakup often necessitateresolut ions far below the Kolmogorov scale. Consequently, models must be employed for s econdarybreakup. Existing physics- based and stochastic breakup models are not universal and fail to account forthe local and instantaneous flow field and dro p geometry. We present a physics-informed machine learningmodel for predicting the statistics of daughter drops generated during the breakup of under-resolved drops.By training on high-fidelity simulations, the model can predict breakup o utcomes from severely underresolvedinput fields. This is made possible by a ca reful choice of quantities of interest and by takinginspiration from the discre te nature of breakup events to encode the temporal evolution via a mixtureof si gmoid functions. We showcase proof-of-concept results from the canonical setting s of 3D Taylor-Green vortex flows and homogeneous isotropic turbulence. Compared to results generated by low-resolutionsimulations (i.e., without a model) and baseline state-of-the-art models, our approach achieves superioraccuracy in pre dicting drop size distribution and critical quantities of interest, such as surf ace area."
StanfordCaliforniaUnited StatesNor th and Central AmericaCyborgsEmerging TechnologiesMachine LearningStanfo rd University
NETL
NSTL
