化学反应工程与工艺
化學反應工程與工藝
화학반응공정여공예
Chemical Reaction Engineering and Technology
2015年
4期
307-314
,共8页
黄雪锋%孙国刚%张玉明%朱喆%李涛
黃雪鋒%孫國剛%張玉明%硃喆%李濤
황설봉%손국강%장옥명%주철%리도
湍球塔%液相含率预测%偏最小二乘回归法%流体动力学
湍毬塔%液相含率預測%偏最小二乘迴歸法%流體動力學
단구탑%액상함솔예측%편최소이승회귀법%류체동역학
turbulent contact absorber%liquid holdup forecasting%partial least-squares regression%hydrodynamics
利用搭建的湍球塔实验装置,考察了操作气速、静床高度、喷淋液量、支承网开孔率和湍球直径等参数对湍球塔床层压降和液相含率的影响特性;运用因次分析π定理和偏最小二乘法,得到了液相含率的回归模型.引入前人Gel和V-Noakovic模型,并基于文献实验数据对各模型预测效果作了对比分析.结果表明,偏最小二乘法处理小容量液相含率样本和自变量强相关问题行之有效,用液相含率新模型预测两组文献实验数据的均方百分比误差分别为2.5%和3.1%,预测的精确度优于Gel和V-Noakovic模型,且新模型适用范围更大.偏最小二乘法用于湍球塔床层液相含率预测建模切实可行.
利用搭建的湍毬塔實驗裝置,攷察瞭操作氣速、靜床高度、噴淋液量、支承網開孔率和湍毬直徑等參數對湍毬塔床層壓降和液相含率的影響特性;運用因次分析π定理和偏最小二乘法,得到瞭液相含率的迴歸模型.引入前人Gel和V-Noakovic模型,併基于文獻實驗數據對各模型預測效果作瞭對比分析.結果錶明,偏最小二乘法處理小容量液相含率樣本和自變量彊相關問題行之有效,用液相含率新模型預測兩組文獻實驗數據的均方百分比誤差分彆為2.5%和3.1%,預測的精確度優于Gel和V-Noakovic模型,且新模型適用範圍更大.偏最小二乘法用于湍毬塔床層液相含率預測建模切實可行.
이용탑건적단구탑실험장치,고찰료조작기속、정상고도、분림액량、지승망개공솔화단구직경등삼수대단구탑상층압강화액상함솔적영향특성;운용인차분석π정리화편최소이승법,득도료액상함솔적회귀모형.인입전인Gel화V-Noakovic모형,병기우문헌실험수거대각모형예측효과작료대비분석.결과표명,편최소이승법처리소용량액상함솔양본화자변량강상관문제행지유효,용액상함솔신모형예측량조문헌실험수거적균방백분비오차분별위2.5%화3.1%,예측적정학도우우Gel화V-Noakovic모형,차신모형괄용범위경대.편최소이승법용우단구탑상층액상함솔예측건모절실가행.
Hydrodynamics experiments were carried out in a pilot-scale turbulent contact absorber (TCA), and the effects of experimental parameters, such as operation gas velocity, static bed height, free open area of grid, ball diameter on bed pressure drop and liquid holdup were investigated. The liquid holdup regression model was presented using Buchingham π-theorem and partial least squares regression(PLSR) algorithm. Furthermore, the 2 liquid holdup models respectively proposed by Gel perin and V-Noakovic were cited, to compare with the new liquid holdup model. Then the 3 models were evaluated based on literature experimental data. The results showed that PLSR algorithm was effective in handling the samples with small capacity and variables with strong correlation. The accuracy and precision of the liquid holdup model by PLSR was better than those of the models cited from literature. By validating with 2 groups of experimental data in literature, the mean square errors(MSPE) of the new liquid holdup model were 2.5% and 3.1%, respectively. The accuracy and precision of predicting liquid holdup was better than those of Gel perin's and V-Noakovic's models, which indicated that the partial least-squares regression algorithm to predict TCA bed liquid holdup was feasible.