Physics-informed machine-learning (PIML) enables the integration of domain knowledge with machine learning (ML) algorithms, which results in higher data efficiency and more stable predictions. This provides opportunities for augmenting—and even replacing—high-fidelity numerical simulations of complex turbulent flows, which are often expensive due to the requirement of high temporal and spatial resolution. In this review, we (i) provide an introduction and historical perspective of ML methods, in particular neural networks (NN), (ii) examine existing PIML applications to fluid mechanics problems, especially in complex high Reynolds number flows, (iii) demonstrate the utility of PIML techniques through a case study, and (iv) discuss the challenges and opportunities of developing PIML for fluid mechanics.
High-pressure conditions in diesel engines can often surpass the thermodynamic critical limit of the working fluid. Consequently, the injection of fuel at these conditions can lead to complex behaviors that remain only incompletely understood. This study is concerned with investigating the application of a diffuse-interface method in conjunction with a finite-rate chemistry model in large-eddy simulations of diesel spray injection and ignition in a supercritical ambient environment. The presented numerical approach offers the capability of simulating these complex conditions without the need for parameter tuning that is commonly employed in spray-breakup models. Numerical simulations of inert and reacting n-dodecane sprays — under the Engine Combustion Network Spray A and Spray D configurations — are studied, and results are compared with experimental data for liquid/vapor penetration lengths and ignition timing. In addition, parametric studies are performed to identify flow sensitivities arising from the variation in nozzle diameters between both injectors, along with the impact of low-temperature oxidation on ignition in Spray D simulations. Spray A simulations are found to be insensitive to turbulence, and predictions for penetration length and ignition behavior are in good agreement with experiments. In contrast, Spray D predictions for penetration length and ignition delay demonstrated significant sensitivities to in-nozzle turbulence, introducing uncertainty to the predicted results and stipulating the need for quantitative measurements for model evaluation.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.