农业工程学报
農業工程學報
농업공정학보
Transactions of the Chinese Society of Agricultural Engineering
2015年
21期
125-131
,共7页
赵伟国%盛建萍%杨军虎%宋启策
趙偉國%盛建萍%楊軍虎%宋啟策
조위국%성건평%양군호%송계책
离心泵%优化%算法%叶轮%效率%参数化%神经网络
離心泵%優化%算法%葉輪%效率%參數化%神經網絡
리심빙%우화%산법%협륜%효솔%삼수화%신경망락
centrifugal pumps%optimization%algorithms%impellers%efficiency%parameterization%artificial neural network
为了提高离心泵的效率,以叶轮效率最大为优化目标进行优化设计。对叶轮进行参数化设计,以实现叶轮几何形状的自动控制以及为优化计算提供优化变量。选择控制叶片积叠线周向定位的2个参数作为优化变量,以?3°~3°作为优化变量的约束范围。利用人工神经网络的学习功能,建立了目标函数与优化变量之间的映射关系。采用遗传算法寻找目标函数的最优值,得到优化变量约束范围内的最优叶轮模型。数值计算结果表明:在设计流量点1200 m3/h时,优化后叶轮的效率较优化前提高了4.02个百分点,离心泵的效率提高了4.41个百分点,扬程提升了2.63 m。针对非设计工况点性能改善不明显这一问题,对原始蜗壳进行重新设计并与优化叶轮组合进行数值计算。在设计工况点效率提高了1.59%,在1.2倍设计工况点处效率提升了9.93%,在1.4倍设计工况点处效率提升了8.83%;较原始叶轮与原始蜗壳的组合,在设计工况点泵的效率提高了6%,在1.2倍设计工况点点效率提高了9.2%,在1.4倍设计工况点点效率提高了8.59%。优化拓宽了水泵运行的高效区,增强了泵的运行稳定性,离心泵的性能得到了优化,叶轮与蜗壳之间的匹配更合理。该研究对离心泵的优化设计提供了参考。
為瞭提高離心泵的效率,以葉輪效率最大為優化目標進行優化設計。對葉輪進行參數化設計,以實現葉輪幾何形狀的自動控製以及為優化計算提供優化變量。選擇控製葉片積疊線週嚮定位的2箇參數作為優化變量,以?3°~3°作為優化變量的約束範圍。利用人工神經網絡的學習功能,建立瞭目標函數與優化變量之間的映射關繫。採用遺傳算法尋找目標函數的最優值,得到優化變量約束範圍內的最優葉輪模型。數值計算結果錶明:在設計流量點1200 m3/h時,優化後葉輪的效率較優化前提高瞭4.02箇百分點,離心泵的效率提高瞭4.41箇百分點,颺程提升瞭2.63 m。針對非設計工況點性能改善不明顯這一問題,對原始蝸殼進行重新設計併與優化葉輪組閤進行數值計算。在設計工況點效率提高瞭1.59%,在1.2倍設計工況點處效率提升瞭9.93%,在1.4倍設計工況點處效率提升瞭8.83%;較原始葉輪與原始蝸殼的組閤,在設計工況點泵的效率提高瞭6%,在1.2倍設計工況點點效率提高瞭9.2%,在1.4倍設計工況點點效率提高瞭8.59%。優化拓寬瞭水泵運行的高效區,增彊瞭泵的運行穩定性,離心泵的性能得到瞭優化,葉輪與蝸殼之間的匹配更閤理。該研究對離心泵的優化設計提供瞭參攷。
위료제고리심빙적효솔,이협륜효솔최대위우화목표진행우화설계。대협륜진행삼수화설계,이실현협륜궤하형상적자동공제이급위우화계산제공우화변량。선택공제협편적첩선주향정위적2개삼수작위우화변량,이?3°~3°작위우화변량적약속범위。이용인공신경망락적학습공능,건립료목표함수여우화변량지간적영사관계。채용유전산법심조목표함수적최우치,득도우화변량약속범위내적최우협륜모형。수치계산결과표명:재설계류량점1200 m3/h시,우화후협륜적효솔교우화전제고료4.02개백분점,리심빙적효솔제고료4.41개백분점,양정제승료2.63 m。침대비설계공황점성능개선불명현저일문제,대원시와각진행중신설계병여우화협륜조합진행수치계산。재설계공황점효솔제고료1.59%,재1.2배설계공황점처효솔제승료9.93%,재1.4배설계공황점처효솔제승료8.83%;교원시협륜여원시와각적조합,재설계공황점빙적효솔제고료6%,재1.2배설계공황점점효솔제고료9.2%,재1.4배설계공황점점효솔제고료8.59%。우화탁관료수빙운행적고효구,증강료빙적운행은정성,리심빙적성능득도료우화,협륜여와각지간적필배경합리。해연구대리심빙적우화설계제공료삼고。
The centrifugal pump is one of the most widely used fluid machinery. However, 3 problems i.e. lower efficiency, unsteady flow and bad cavitations performance are perplexing the development of centrifugal pump. For a single centrifugal pump, the impeller is one of the most important flow components, so it is selected as the optimum objective. Parametric fitting is a prerequisite in impeller optimization design. This process provides optimization variables and controls impeller automatically for the optimization design. Bezier curve and B-spline curve are used to reconstruct the impeller to obtain the profile of the blade and the meridional surface. The stacking point is reference point which defines the position of the two-dimensional (2D) blade section on a stream surface. This point is first defined on the 2D blade section, and then positioned on the corresponding stream surface in the meridional and tangential directions. Trailing edge is selected as stacking curve. Bezier-line-Bezier curve can be used to fit tangential location. The optimization variables are the angle between linear segment and vertical direction and the angle between the second Bezier curve and vertical direction with the span of 1, which 2 variables control the tangential position of stacking line on the 2D blade section. The range of -3°-3° is chosen as the constraint condition of optimization variables. Recently, CFD (computationalfluiddynamics) technology has been widely applied to numerical computation of the three-dimensional viscous flow inside turbomachinery, which has made much progress. Meanwhile, many excellent optimization algorithms have been proposed. Fortunately, the CFD technology isn’t confined to the research of centrifugal pump inner flow. Combining the CFD technology and optimization algorithm will play a very important role in the increase of pump efficiency, the decrease of flow loss and the extension of high-performance areas. An automatic optimization design platform for the centrifugal impellers is constructed by the genetic algorithm combined with the parameterization method and the commercial computational fluid dynamics software NUMECA. Based on the genetic algorithm and the artificial neural network, a new optimization method for the optimization of a centrifugal impeller is presented. Different from the traditional optimization method, the performance of centrifugal impeller is predicted with the CFD technology in the new developed method. The relationship between objective function and optimization variables is established by the learning function of artificial neural network. The results show that the efficiency of impeller achieves the maximum, when the angle between linear segment and vertical direction is-2.886° and the angle between the second Bezier curve and vertical direction with the span of 1 is1.31°. Compared with the original, the efficiency is improved by 4.02% for optimum impeller in the design point. The centrifugal pump efficiency is increased by 4.41%, and the head is increased by 2.63 m. Volute is one of important flow components and has a great effect on the single centrifugal pump. The loss in volute is very great with optimized impeller, or with original one, especially in the large flow area. The volute is redesigned and the numerical simulation of modified volute with optimum impeller is performed for the flow field analysis of the flow passage components. The efficiency is improved by 1.59% compared with the pump with optimum impeller and original volute in design point, and by 6% compared with the pump with original impeller and original volute in design point. The performance of centrifugal pump is optimized, and the purpose of energy saving is achieved. These findings confirm that the optimization design method is effective for the centrifugal impellers.