陳進(jìn),孫振業(yè),謝翌,馬金成
(重慶大學(xué)機(jī)械傳動(dòng)國家重點(diǎn)實(shí)驗(yàn)室,重慶400044)
γ?Reθ轉(zhuǎn)捩模型在風(fēng)力機(jī)翼型數(shù)值計(jì)算中的應(yīng)用
陳進(jìn),孫振業(yè),謝翌,馬金成
(重慶大學(xué)機(jī)械傳動(dòng)國家重點(diǎn)實(shí)驗(yàn)室,重慶400044)
很多湍流模型忽略了層流區(qū)域的存在,但實(shí)際流動(dòng)在翼型某位置處開始轉(zhuǎn)捩,此時(shí)模型顯然偏離實(shí)質(zhì),計(jì)算結(jié)果精度較低。因此加入γ?Reθ轉(zhuǎn)捩模型,將轉(zhuǎn)捩動(dòng)量厚度雷諾數(shù)Reθ作為經(jīng)驗(yàn)關(guān)聯(lián)函數(shù)來控制邊界層內(nèi)間歇因子γ的生成,再通過間歇因子來控制湍動(dòng)能產(chǎn)生項(xiàng),使湍流模型在層流區(qū)域失效。首先為了驗(yàn)證數(shù)值計(jì)算的準(zhǔn)確性,采用上述方法針對風(fēng)力機(jī)翼型A2121,在高雷諾數(shù)4×106下對幾種典型攻角的氣動(dòng)性能進(jìn)行計(jì)算,對比普通全湍流模型、湍流轉(zhuǎn)捩模型和風(fēng)洞試驗(yàn)的計(jì)算結(jié)果,發(fā)現(xiàn)湍流轉(zhuǎn)捩模型結(jié)果更精確。之后在更大攻角范圍-10.14°~25.09°內(nèi),采用此轉(zhuǎn)捩模型數(shù)值方法進(jìn)行氣動(dòng)仿真,發(fā)現(xiàn)其總體計(jì)算結(jié)果與風(fēng)洞試驗(yàn)實(shí)驗(yàn)數(shù)據(jù)較吻合,驗(yàn)證了此數(shù)值方法的正確性和有效性。
γ?Reθ轉(zhuǎn)捩模型;轉(zhuǎn)捩動(dòng)量厚度雷諾數(shù);間歇因子;風(fēng)力機(jī);翼型
翼型通常在失速分離開始時(shí)有較好的性能,故分離現(xiàn)象預(yù)測[1]尤為重要。直接數(shù)值模擬(DNS)和大渦模(LES)具有模擬轉(zhuǎn)捩過程的能力[2],但消耗計(jì)算資源大。平均Navier?Stokes方程(RANS)應(yīng)用廣泛,但無法準(zhǔn)確預(yù)測轉(zhuǎn)捩過程[3]。早期的研究通過經(jīng)驗(yàn)或半經(jīng)驗(yàn)方法來確定轉(zhuǎn)捩,如經(jīng)驗(yàn)關(guān)聯(lián)方法、En方法和基于間歇因子的預(yù)測方法?;诋?dāng)?shù)氐妮斶\(yùn)模型避開前述復(fù)雜過程,求解基本湍流模型和轉(zhuǎn)捩特征參數(shù)微分方程[4]。如Menter等提出當(dāng)?shù)仃P(guān)聯(lián)的γ-Reθ轉(zhuǎn)捩模型[5]。把經(jīng)驗(yàn)關(guān)聯(lián)和間歇因子結(jié)合,將轉(zhuǎn)捩動(dòng)量厚度雷諾數(shù)Reθ作為關(guān)聯(lián)函數(shù)來控制邊界層間歇因子γ生成,通過γ控制湍流生成,回避動(dòng)量厚度的計(jì)算,通過R輸運(yùn)方程實(shí)現(xiàn)計(jì)算當(dāng)?shù)鼗T撃P筒荒M邊界層轉(zhuǎn)捩的物理過程,可融入現(xiàn)代CFD框架。采用RANS和轉(zhuǎn)捩模擬結(jié)合的方法,對翼型A2121仿真。
1.1 控制方程
采用二維連續(xù)性方程和二維不可壓縮N?S方程,方程不再贅述。重點(diǎn)介紹湍流模型和轉(zhuǎn)捩模型控制方程。S?A、RNG k?ε、SST k?ω的控制方程,詳見相關(guān)參考文獻(xiàn)。Transition SST是在湍流兩方程k、ω模型的基礎(chǔ)上加入γ?Reθ兩方程轉(zhuǎn)捩模型而來。轉(zhuǎn)捩模型通過求解間歇函數(shù)γ(0?γ?1)來觸發(fā)轉(zhuǎn)捩。且需要與湍流模型聯(lián)合來控制轉(zhuǎn)捩發(fā)生,具體聯(lián)合方式就是使用間歇函數(shù)來修正k方程的生成項(xiàng)、破壞項(xiàng)和混合函數(shù)。
1)γ輸運(yùn)方程
參數(shù)Pγ和Eγ的表達(dá)式為
參數(shù)Ftub和Fonset的表達(dá)式為
參數(shù)Foneset1、Foneset2和Foneset3的表達(dá)式為
2) 輸運(yùn)方程
參數(shù)Pθt的表達(dá)式為
參數(shù)Fθt、δ、δBL和θBL的表達(dá)式為
1.2 幾何模型和計(jì)算網(wǎng)格
本次實(shí)驗(yàn)翼型A2121為型線形狀優(yōu)化翼型[6?8],相對厚度為21%。計(jì)算域長度為30倍弦長,寬度為25倍弦長[9?11]。采用結(jié)構(gòu)化網(wǎng)格,對流場參數(shù)變化梯度大的區(qū)域加密。無關(guān)性分析表明50萬網(wǎng)格滿足計(jì)算要求。圖1為局部放大圖。
圖1 仿真使用的網(wǎng)格Fig.1 Mesh grid used in simulation
第1層網(wǎng)格高度為0.000 1 mm,壁面處Y+<0.1。邊界條件選用速度進(jìn)口和壓力出口,翼型為標(biāo)準(zhǔn)壁面無滑移邊界條件。
1.3 數(shù)學(xué)模型及離散方法
分別采用了γ-Reθ、S?A、RNG k?ε、SST k?ω模型進(jìn)行計(jì)算,其中對S?A、SST k?ω進(jìn)行低Re數(shù)修正,RNG k?ε采用標(biāo)準(zhǔn)壁面函數(shù)。采用雙精度定常分離式求解器求解穩(wěn)態(tài)N?S方程??刂品匠虒α黜?xiàng)采用二階迎風(fēng)格式離散,壓力項(xiàng)采用二階格式離散,其他項(xiàng)均采用QUICK格式離散,速度和壓力耦合采用SIMPLE算法,收斂精度標(biāo)準(zhǔn)為10-6。
試驗(yàn)在NF-3二元試驗(yàn)段進(jìn)行,翼型迎角變化范圍-10.14°~25.09°,試驗(yàn)雷諾數(shù)4×106。
2.1 不同湍流模型的計(jì)算結(jié)果對比
圖2為A2121翼型在3種典型攻角下,采用4種湍流模型計(jì)算出的表面壓力系數(shù)分布及峰值的局部放大圖。圖2(a)α=-8.11°時(shí),除了RNG k?ε計(jì)算的壓力系數(shù)偏離較大外,其他3種模型結(jié)果相近。圖2(b)取吸力面橫坐標(biāo)0~100 mm范圍內(nèi)的壓力系數(shù)分布。在吸力面最大負(fù)壓位置X=8 mm處,γ-Reθ、SST k?ω、S?A及RNG k?ε模型與實(shí)驗(yàn)結(jié)果的相對誤差分別為12.103%、13.700%、15.914%、21.842%。如圖2(c)、(d),α=0°時(shí)4種模型的計(jì)算結(jié)果接近。在X=160 mm處γ-Reθ、SST k?ω、S?A及RNG k?ε與實(shí)驗(yàn)結(jié)果的相對誤差分別為4.677%、7.280%、5.750%、7.381%。如圖2(e)、(f),α=25°時(shí)壓力系數(shù)分布差異較大。RNG k?ε在計(jì)算分離渦附近壓力分布時(shí)偏差最大,在橫坐標(biāo)范圍為0~240 mm的吸力面壓力系數(shù)遠(yuǎn)偏離實(shí)驗(yàn)值,平均誤差為93.968%。SST k?ω在吸力面橫坐標(biāo)0~100 mm范圍的平均誤差為38.543%。S?A在橫坐標(biāo)為200~760 mm范圍內(nèi),吸力面壓力系數(shù)平均誤差為88.943%。由于γ?Reθ能夠準(zhǔn)確模擬轉(zhuǎn)捩現(xiàn)象,其對分離開始發(fā)生位置及深度分離區(qū)的壓力系數(shù)計(jì)算較準(zhǔn)確。在全弦長范圍內(nèi)γ-Reθ、SST k?ω、S?A及RNG k?ε實(shí)驗(yàn)結(jié)果的平均相對誤差分別為36.810%,53.158%,52.014%,120.302%。
圖2 翼型表面壓力系數(shù)分布Fig.2 Pressure coefficient along with the chord
圖3 為分別采用4種模型得出的翼型A2121升力系數(shù)和阻力系數(shù),并與實(shí)驗(yàn)數(shù)據(jù)對比。
圖3 翼型A2121的升力系數(shù)和阻力系數(shù)Fig.3 Lift and drag coefficient of airfoil A2121
對于升力系數(shù),當(dāng)攻角為α=-8.11°時(shí),γ?Reθ、SST k?ω、S?A及RNG k?ε模型計(jì)算所得值均小于實(shí)驗(yàn)結(jié)果-0.2681。與圖2(a)壓力系數(shù)曲線包圍面積小于實(shí)驗(yàn)吻合。RNG k?ε模型相對誤差最大,為27.005%。當(dāng)α=0°時(shí),不同模型的升力系數(shù)接近,相對誤差均小于5.4%,與圖2(c)中壓力分布相近吻合。當(dāng)α=25°時(shí),S?A模型的升力系數(shù)偏小,其相對誤差為34.973%。RNG k?ε和SST k?ω在吸力面前緣的壓力過低,預(yù)測升力系數(shù)大大超過實(shí)驗(yàn)值,相對誤差分別為74.518%、22.008%。γ?Reθ能捕捉轉(zhuǎn)捩,升力系數(shù)也最準(zhǔn)確,相對誤差為5.299%。對于阻力系數(shù),當(dāng)攻角為α=-8.11°和α=0°時(shí)采用γ?Reθ模型計(jì)算值與實(shí)驗(yàn)值最為接近,其相對誤差分別為24.5%,4.65%。在深度失速區(qū)α=25°表現(xiàn)略差,其相對誤差為20%。此時(shí)流動(dòng)處于強(qiáng)烈的非穩(wěn)態(tài)狀態(tài),采用穩(wěn)態(tài)算法會(huì)帶來一定的誤差。
2.2 A2121翼型在不同攻角下的氣動(dòng)性能
γ?Reθ模型對翼型升力、阻力計(jì)算精度高。采用RANS和γ?Reθ相結(jié)合的方法,在對翼型A2121在攻角-10.14°~25.09°范圍內(nèi)進(jìn)行計(jì)算,升力系數(shù)、阻力系數(shù)隨攻角的變化如圖4所示。
當(dāng)α<11°時(shí),升力及阻力系數(shù)與實(shí)驗(yàn)值很接近,說明采用平均雷諾與γ?Reθ轉(zhuǎn)捩模型相結(jié)合對于升力模擬計(jì)算很精確。當(dāng)α≥11°深失速時(shí)精度相對較差,升力系數(shù)精度比阻力系數(shù)略高。偏差較大可能原因有:1)大攻角下翼型對風(fēng)洞的阻塞度增加,同時(shí)深失速下吸力面漩渦尺度與翼型尺寸相當(dāng)或更大,洞壁干擾修正不足;2)大分離深失速狀態(tài)下的非定常流場產(chǎn)生了渦的周期性脫落,定常數(shù)值模擬只是實(shí)際流場的平均近似。阻力系數(shù)誤差較大的一個(gè)重要原因是摩擦阻力比升力小一個(gè)數(shù)量級。與風(fēng)洞數(shù)據(jù)的總體對比來看,由于對轉(zhuǎn)捩現(xiàn)象的捕捉能力較好,RANS和轉(zhuǎn)捩相結(jié)合的模擬方法取得了比較高的精度。
采用不同的湍流模型對風(fēng)力機(jī)翼型A2121進(jìn)行數(shù)值仿真,并同實(shí)驗(yàn)進(jìn)行對比,結(jié)論如下:1)通過間歇因子來控制湍動(dòng)能產(chǎn)生項(xiàng),使湍流模型在層流區(qū)域失效,γ?Reθ轉(zhuǎn)捩模型能夠更加準(zhǔn)確地捕捉物理實(shí)質(zhì)。2)γ?Reθ轉(zhuǎn)捩模型得到的翼型壓力系數(shù)分布、升力阻力系數(shù)的與實(shí)驗(yàn)結(jié)果相近,且變化趨勢相同,證實(shí)了此模型理論和使用價(jià)值。
[1]REZA T Z,MAHMOOD S,AMIR K.Prediction of boundary layer transition based on modeling of laminar fluctuations u?sing RANS approach[J].Chinese Journal of Aeronautics,2009,22(2):113?120.
[2]ROBERT S K,YARAS M I.Large?eddy simulation of transi?tion in a separation bubble[J].Journal of Fluids Engineer?ing,2006,128(2):232?238.
[3]GENC M S,KAYNAK ü,YAPICI H.Performance of tran?sition model for predicting low Re aerofoil flows without/with single and simultaneous blowing and suction[J].European Journal of Mechanics?B/Fluids,2011,30(2):218?235.
[4]WALTERS D K,LEYLEK J H.A new model for boundary layer transition using a single?point RANS approach[J].ASME J Turbomachinery,2004,126:193?202.
[5]MENTER F R,LANGTRY R B,LIKKI S R,et al.A corre?lation?based transition model using local variables—Part I:model formulation[J].Journal of turbomachinery,2006,128(3):413?422.
[6]陳進(jìn),程江濤,SHEN Wenzhong.基于表面曲率光滑連續(xù)性的風(fēng)力機(jī)翼型改進(jìn)設(shè)計(jì)研究[J].太陽能學(xué)報(bào),2013,34(1):547?553.CHEN Jin,CHENG Jiangtao,SHEN Wenzhong.Research on improved design of airfoil profiles based on the continuity of airfoil surface curvature of wind turbines[J].Acta Ener?glae Solaris Sinica,2013,34(1):547?553.
[7]CHEN Jin,WANG Quan,PANG Xiaoping.Improvement of airfoil design using smooth curvature technique[J].Renew?able Energy,2013,51:426?435.
[8]陳進(jìn),張石強(qiáng),王旭東.基于粗糙度敏感性研究的風(fēng)力機(jī)專用翼型設(shè)計(jì)[J].空氣動(dòng)力學(xué)學(xué)報(bào),2011,29(2):142?149.
CHEN Jin,ZHANG Shiqiang,WANG Xudong.Dedicated wind turbine airfoil design based on the roughness sensitivity considera?tions[J].Acta Aerodynamica Sinica,2011,29(2):142?149.
[9]李仁年,李銀然,王秀勇,等.翼型后緣厚度對氣動(dòng)性能影響的數(shù)值分析[J].空氣動(dòng)力學(xué)學(xué)報(bào),2011,29(2):205?210.
LI Rennian,LI Yinran,WANG Xiuyong et al.Numerical a?nalysis of aerodynamic performance of airfoil's trailing edge thickness[J].Acta Aerodynamica Sinica,2011,29(2):205?210.
[10]周大高,柳陽威,文曉慶.改進(jìn)SA模型對翼型分離流動(dòng)的數(shù)值模擬[J].北京航空航天大學(xué)學(xué)報(bào),2012,38(10):1384?1388.
ZHOU Dagao,LIU Yangwei,WEN Xiaoqing.Modification of SA model for predicting airfoil flow separation[J].Jour?nal of Beijing University of Aeronautics and Astronautics,2012,38(10):1384?1388.
[11]李銀然,李仁年,王秀勇,等.計(jì)算模型維數(shù)對風(fēng)力機(jī)翼型氣動(dòng)性能預(yù)測的影響[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2011,42(2):115?119.
LI Yinran,LI Rennian,WANG Xiuyong,et al.Effects of the calculation models with different dimension on the aero?dynamic performance prediction for wind turbine airfoil[J].Transactions of the Chinese Society of Agricultural Machin?ery,2011,42(2):115?119.
γ?Reθtransition model in numerical simulation of airfoil for wind turbine
CHEN Jin,SUN Zhenye,XIE Yi,MA Jincheng
(State Key Laboratory of Mechanical Transmission,Chongqing University,Chongqing,400044,China)
Most of the turbulence models ignored the existence of laminar flow.However,the reality is that transi?tion process will be triggered at some point on airfoil.Therefore,full turbulence models have deviated from the real flow phenomenon,and so the precision of the results attained from these methods are comparatively low.Conse?quently theγ?R eθtransition model is added into full turbulence models here,where transition momentum thickness Reynolds number is treated as empirical correlation function to control the production of intermittency factor γ.Then intermittency factor is used to control the production term of the turbulent kinetic energy,making the turbulence model unavailable in laminar flow area.Above all,in order to verify the accuracy of the numerical simulation,aforementioned methods are used to simulate the aerodynamic performance of the airfoil A2121 under some typical angles of attack,with the high Reynolds number of4×106.It is found that the transition model is of higher accura?cy through comparing the results gained from full turbulence models,transition model and wind tunnel experiment.Afterwards,this transitional turbulence model is adopted to get the aerodynamic data within the attack angle ranges from-10.14°to 25.09°.The experiment and computational fluid dynamics method are demonstrated and com?pared.The results from transitional turbulence model showed great agreement with the outcomes from wind tunnel experiment,which provide a validation of this numerical simulation method for airfoil.
γ?Reθtransition model;transition momentum thickness Reynolds number;intermittency;wind turbine;numerical simulation
10.3969/j.issn.1006?7043.201309013
http://www.cnki.net/kcms/doi/10.3969/j.issn.1006?7043.201309013.html
TK83;TH12
A
1006?7043(2015)02?0218?04
2013?09?13.網(wǎng)絡(luò)出版時(shí)間:
國家863計(jì)劃資助項(xiàng)目(2012AA051301);國家自然科學(xué)基金資助項(xiàng)目(51175526).
陳進(jìn)(1956?),男,教授,博士生導(dǎo)師;孫振業(yè)(1990?),男,博士研究生.
孫振業(yè),E?mail:dennysun@aliyun.com.