A comparative assessment of machine-learning (ML) methods for active flow control is performed. The chosen benchmark problem is the drag reduction of a two-dimensional Kármán vortex street past a circular cylinder at a low Reynolds number ( Re = 100). The flow is manipulated with two blowing/suction actuators on the upper and lower side of a cylinder. The feedback employs several velocity sensors. Two probe configurations are evaluated: 5 and 11 velocity probes located at different points around the cylinder and in the wake. The control laws are optimized with Deep Reinforcement Learning (DRL) and Linear Genetic Programming Control (LGPC). By interacting with the unsteady wake, both methods successfully stabilize the vortex alley and effectively reduce drag while using small mass flow rates for the actuation. DRL has shown higher robustness with respect to different initial conditions and to noise contamination of the sensor data; on the other hand, LGPC is able to identify compact and interpretable control laws, which only use a subset of sensors, thus allowing for the reduction of the system complexity with reasonably good results. Our study points at directions of future machine-learning control combining desirable features of different approaches.
Purpose
Fused deposition modeling (FDM) is booming as a manufacturing technique in several industrial fields because of its ease of use, the simple-to-meet requirements for its machinery and the possibility to manufacture individual specimens cost-effectively. However, there are still large variations in the mechanical properties of the prints dependent on the process parameters, and there are many discrepancies in the literature as to which are the optimal parameters.
Design/methodology/approach
In this paper, thermal evolution of the printed specimens is set as the main focus and some phenomena that affect this evolution are explored to differentiate their effects on the mechanical properties in FDM. Interlayer waiting times, the thermal effects of the position of the extruder relative to the specimens and the printing layout are assessed. Thermal measurements are acquired during deposition and tensile tests are performed on the specimens, correlating the mechanical behavior with the thermal evolution during printing.
Findings
Additional waiting times do not present significant differences in the prints. Thermal stabilization of the material is observed to be faster than whole layer deposition. The layout is seen to affect the thermal gradients in the printed specimens and increase the fragility. Strain at breakage variations up to 64% are found depending on the layout.
Originality/value
This study opens new research and technological discussions on the optimal settings for the manufacturing of high-performance mechanical components with FDM through the study of the thermal gradients generated in the printed specimens.
This paper describes a unified approach to the identification of time-varying One-machine Infinite-Bus (OMIB) equivalents based on the use of finite-differences. The key to this approach is the local decomposition of the classical two-machine equivalent into a multi-machine time-varying system that captures the interplay between machines in each critical cluster.By addressing the interplay between subgroups of coherent generators the proposed procedure opens new avenues for both the evaluation and the control of the inter-area mode phenomenon. Analytical criteria are developed which predict the existence and stability character of the system response as a function of energy concepts. Energy-based techniques are then used to infer the energy transfer among the groups of coherent generators. In addition to giving physical insight into the problem of post-fault inter-area oscillations, the method allows for the inclusion of inter-machine behavior in a systematic manner and results in a computationally efficient algorithm.The practical use of the proposed techniques is tested on a realistic 45-machine power system model representing parts of the Mexican interconnected system.
The recent advancements in high-resolution turbulence-statistics computation from ensemble particle tracking velocimetry (EPTV) data are now opening new possibilities in turbulent-flow characterisation. Measurements of full-field boundary layer profiles with a fine resolution close to the wall and up to the freestream with one single imaging setup are now feasible, thus paving the way to direct characterisation of turbulent-boundary-layer (TBL) parameters with composite-profile formulations. In this work, we build a framework for the estimation of the uncertainty of EPTV in performing this task. The effect of systematic errors due to finite spatial resolution and of random error due to convergence are investigated under different window size. Then we introduce random errors to simulate the effects on convergence issues on the velocity profile and, consequently, on the estimation of turbulent-boundary-layer parameters. The statistical dispersion of the estimated parameters provides an estimation of the uncertainty range. We validate with experimental data this flexible tool to estimate a priori the expected uncertainty level of the most relevant turbulent-boundary-layer parameters in zero-pressure-gradient TBL, being the method based on existing profiles from high-fidelity simulation or from analytical composite-profile formulations when such data are not available.
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.