脱硫废水旋转喷雾干燥塔烟气分布器优化设计  

Optimization design of flue gas distributor for rotary spray drying tower of desulfurization wastewater

作　　者：陈沾兴 詹凌霄 陈海杰 杨林军[1] CHEN Zhanxing;ZHAN Lingxiao;CHEN Haijie;YANG Linjun(Key Laboratory of Energy Thermal Conversion and Control of Ministry of Education,Southeast University,Nanjing 210096,China;Datang Environment Industry Group Co.Ltd.,Beijing 100097,China)

机构地区：[1]东南大学能源热转换及其过程测控教育部重点实验室,江苏南京210096 [2]大唐环境产业集团股份有限公司,北京100097

出　　处：《中南大学学报(自然科学版)》2025年第3期868-880,共13页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(52076046)。

摘　　要：为了实现脱硫废水旋转喷雾干燥塔烟气分布器的优化设计,降低塔体阻力并提升蒸发性能,本文提出了一种基于CFD和LG-GPR-CIAL模型的优化方法。首先,利用CFD技术模拟干燥塔内流场分布,构建了稳态气液耦合模型获取样本;其次,引入LG-GPR-CIAL模型,利用高斯过程回归(GPR)构建代理模型,并借助遗传算法(GA)最小化留一法均方误差(LOO-MSE)以优化GPR模型超参数,主动学习策略采用置信度与重要性边界相结合的采样策略,每次仅选取置信区间最大且目标函数最优的一个点进行采样,实现样本量最小化与模型精度提升的双重目标;第三,通过调整内外导流板的几何参数,实现了对烟气分布器的优化设计,通过主动学习策略不断引入新的样本点,提高模型精度,最终确定最优解;最后,通过1个600MW燃煤机组脱硫废水旋转喷雾蒸发系统的实例,验证了该方法的有效性和准确性。研究结果表明:优化后的烟气分布器增加了出口气流的均匀性,降低了气流湍流强度,在保证脱硫废水旋转喷雾蒸发性能的同时,显著降低了10.05%的干燥塔的阻力。To optimize the design of flue gas distributors in desulfurization wastewater rotary spray drying towers,reduce tower resistance,and enhance evaporation performance,an optimization method integrating CFD and LG-GPR-CIAL models was proposed.Firstly,CFD technology was employed to simulate flow field distribution within the drying tower.A steady-state gas-liquid coupling model was established to generate training samples.Secondly,the LG-GPR-CIAL framework was introduced for optimization.A Gaussian process regression(GPR)surrogate model was constructed.Genetic algorithm(GA)was applied to minimize leave-one-out mean squared error(LOO-MSE)for optimizing GPR hyperparameters.An active learning strategy combining confidence intervals and importance boundary sampling was implemented,where only one sample with maximum confidence interval and optimal objective function value was selected per iteration for model retraining,ensuring minimal sample size and improved accuracy.Thirdly,geometric parameters of internal/external guide vanes were adjusted to optimize the flue gas distributor.New sample points were iteratively added through the active learning strategy to refine model precision,ultimately determining the optimal solution.Finally,the method was validated by using a 600 MW coal-fired unit's desulfurization wastewater rotary spray evaporation system.The results show that the optimized distributor achieves enhancing uniformity of outlet airflow,optimizes swirl number,and reduces turbulence intensity.The design reduces drying tower resistance by 10.05% and maintains evaporation performance.

关 键 词：旋转喷雾干燥塔 烟气分布器 数值模拟 优化设计 主动学习 

分 类 号：TK173[动力工程及工程热物理—热能工程]