This study used self-luminous high-speed photography to visualize quasi-detonation propagation and deflagration-to-detonation transition (DDT) in a transparent round tube equipped with repeating orifice plates. Experiments were conducted in a combustion channel consisting of a 3.16 m square channel with a 7.6 cm by 7.6 cm cross-section connected to a 1.55 m cylindrical channel filled with orifice plates. Rectangular 'fence-type' obstacles were installed on the top and bottom of the square channel with a 3.8 cm opening between them. Two sets of orifice plates with different diameters, d, representing different blockage ratios (BR) were tested (d=5.33 cm for 50% BR and 3.81 cm for 75% BR orifice plates). Stoichiometric hydrogen-oxygen mixtures at initial pressures of 4-60 kPa were ignited at one end of the combustion channel. Average propagation velocities were derived from shock-time-of-arrival measurements using pressure transducers in the square channel and high-speed video filmed through the round tube. First and foremost, I'd like to thank my supervisor, Dr. Gaby Ciccarelli, for his guidance, support and teaching over the past two years. This degree has been a wonderful opportunity and it has been a real pleasure to work on this project with him.
The authors present generalized finite-volume-based discretized loss functions integrated into pressure-linked algorithms for physics-based unsupervised training of neural networks (NNs). In contrast to automatic differentiation-based counterparts, discretized loss functions leverage well-developed numerical schemes of computational fluid dynamics (CFD) for tailoring NN training specific to the flow problems. For validation, neural network-based solvers (NN-solvers) are trained by posing equations such as the Poisson equation, energy equation, and Spalart-Allmaras model as loss functions. The predictions from the trained NNs agree well with the solutions from CFD solvers while also providing solution time speed-ups of up to seven times. Another application of unsupervised learning is the novel hybrid loss functions presented in this study. Hybrid learning combines the information from sparse or partial observations with a physics-based loss to train the NNs accurately and provides training speed-ups of up to five times compared to a fully unsupervised method. Also, to properly utilize the potential of discretized loss functions, they are formulated in a machine learning (ML) framework (TensorFlow) integrated with a CFD solver (OpenFOAM). The ML-CFD framework created here infuses versatility into the training by giving loss functions the access to the different numerical schemes of the OpenFOAM. In addition, this integration allows for offloading the CFD programming to OpenFOAM, circumventing bottlenecks from manually coding new flow conditions in a solely ML-based framework like TensorFlow.
Current study examines the effect on coefficient of drag (Cd) of convoy of two reference car bodies (Ahmed body) by employing underbody diffuser on lead body. CFD analysis of convoy is done using Shear-Stress-Transport model under moving ground conditions. The lead body’s diffuser length is taken as 222m with diffuser angle of 0° (no diffuser), 3°, 5, 7°, 9°, 15°, 20°, 25°and 30° each at inter-vehicular 0.25 and 0.75 body length. Each configuration resulting was analyzed with lead body backlite angle of 25° (pre-critical) and 35° (post-critical) with follow body backlite angle remaining 25°. To understand the flow features developed on Ahmed body due to an underbody diffuser a preliminary CFD analysis is done on an isolated body with 25° and 35° backlite angles by applying each diffuser angle in current study. CFD analyses are conducted after performing two validation analyses from previous studies. The drag on lead and follow vehicles was found to also depend on the axial vortices due to diffuser in addition to those from backlite surface of lead body. Average drag on cases with diffuser is found to be lesser than the no diffuser cases up to a certain diffuser angle. Thus applying diffuser has resulted in potential for reducing the overall drag on convoy by deciding optimum configuration.
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.