Jain Piyush Kumar, Lanjewar Atul, Jain Rahul, Rana Kunj Bihari
Department of Mechanical Engineering, Rajasthan Technical University, Kota, 324010, India.
Department of Mechanical Engineering, Maulana Azad National Institute of Technology, Bhopal, 462003, India.
Environ Sci Pollut Res Int. 2021 Feb 26. doi: 10.1007/s11356-021-12875-0.
Among all renewable energy sources, solar power is one of the major sources which contributes for pollution control and protection of environment. For a number of decades, technologies for utilizing the solar power have been the area of research and development. In the current research, thermal performance parameters of multi-gap V-roughness with staggered elements of a solar air heater (SAH) are experimentally investigated. The artificial neural network (ANN) is also utilized for predicting the thermal performance parameters of SAH. Experiments were executed in a rectangular channel with one roughened side at the top exposed to a uniform heat flux. A significant rise in thermal efficiency performance was reported under a predefined range of Reynolds number (Re) from 3000 to 14000 with an optimized value of relative roughness pitch ratio (P/e) and relative staggered rib length (w/g) as 12 and 1, respectively. The maximum thermal efficiency was attained in the range from 42.15 to 87.02% under considered Reynolds numbers for optimum value of P/e as 12 and w/g as 1. A multilayered perceptron (MLP) feed-forward ANN trained by the Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm was utilized to predict the thermal efficiency (η), friction (f), and Nusselt number (Nu). The thermal performance parameters such as P/e, w/g, Re, and temperature at the inlet, outlet, and plate were the critical input parameters/signals used in the ANN method. The optimum ANN arrangement/structure to predict the Nu, f, and η demonstrate higher accurateness in assessing the performance characteristics of SAH by attaining the root mean squared error (RMSE) in prediction and the Pearson coefficient of association (R) of 1.591 and 0.994; 0.0012 and 0.851; and 0.025 and 0.981, respectively. The prediction profile plots of the ANN demonstrate the influence of various input parameters on the thermal performance parameters.
在所有可再生能源中，太阳能是有助于污染控制和环境保护的主要能源之一。几十年来，太阳能利用技术一直是研究和开发的领域。在当前的研究中，对带有交错元件的多间隙V型粗糙度太阳能空气加热器（SAH）的热性能参数进行了实验研究。人工神经网络（ANN）也被用于预测SAH的热性能参数。实验在一个矩形通道中进行，通道顶部的一侧粗糙，暴露在均匀热流下。在雷诺数（Re）从3000到14000的预定义范围内，报告了热效率性能的显著提高，相对粗糙度间距比（P/e）和相对交错肋长度（w/g）的优化值分别为12和1。在考虑的雷诺数范围内，当P/e为12、w/g为1时，最大热效率在42.15%至87.02%之间。利用由布罗伊登-弗莱彻-戈德法布-香农（BFGS）算法训练的多层感知器（MLP）前馈神经网络来预测热效率（η）、摩擦系数（f）和努塞尔数（Nu）。诸如P/e、w/g、Re以及入口、出口和板处的温度等热性能参数是ANN方法中使用的关键输入参数/信号。预测Nu、f和η的最佳ANN排列/结构在评估SAH的性能特征方面表现出更高的准确性，预测中的均方根误差（RMSE）和皮尔逊关联系数（R）分别为1.591和0.994；0.0012和0.851；以及0.025和0.981。ANN的预测剖面图展示了各种输入参数对热性能参数的影响。
