In the realm of energy and fluid dynamics research, a team of scientists from various institutions has been exploring innovative ways to optimize the design of heated channels. The researchers, Piotr Kamiński and Karol Wawrzak from the Silesian University of Technology in Poland, Yiqing Li from the University of Chinese Academy of Sciences, Bernd R. Noack from the Technical University of Berlin, and Artur Tyliszczak from the Silesian University of Technology, have recently published their findings in the Journal of Fluid Mechanics.
The study focuses on the design of wavy walls in partially heated channels, aiming to minimize temperature variation while keeping pressure loss in check. The researchers used a combination of computational fluid dynamics (CFD) simulations and machine learning techniques, specifically Bayesian optimization (BO), to achieve this goal. The optimization process involved up to 1000 iterations, with the researchers defining a cost function that balanced the competing objectives of temperature variation and pressure loss.
Two shaping strategies were employed in the study. The first was a sinusoidal-type function defined by four parameters: two waviness amplitudes, wave count, and tilt. The second approach used a higher-dimensional method employing a Piecewise Cubic Hermite Interpolation Polynomial (PCHIP) with 19 control points. The results showed that the sinusoidal design reduced temperature variation by over 60 times but increased pressure loss fourfold. In contrast, the PCHIP shape offered only a 15-fold reduction in temperature variation but with a twofold increase in pressure loss.
The researchers also examined flow characteristics such as turbulent kinetic energy, pressure, temperature, and Nusselt number for both optimal and suboptimal shapes along the Pareto front. Based on these insights, they refined the BO result with human input, leading to a further 17.7% reduction in the cost function. This refinement involved replacing small-amplitude waviness periods with flat segments, which also significantly facilitated manufacturability.
The practical applications of this research for the energy sector are promising. Optimizing the design of heated channels can improve the efficiency of heat exchangers, which are crucial components in various energy systems, including power plants, HVAC systems, and industrial processes. By minimizing temperature variation and pressure loss, these optimized designs can enhance overall system performance and reduce energy consumption.
In conclusion, this study demonstrates the potential of combining CFD simulations with machine learning techniques to optimize the design of heated channels. The findings offer valuable insights for the energy industry, paving the way for more efficient and manufacturable heat exchanger designs. For further details, the research can be accessed in the Journal of Fluid Mechanics.
This article is based on research available at arXiv.