Download as pdf or txt
Download as pdf or txt
You are on page 1of 31

2000年12月1日10:29年度審查AR117-14

Annu。 Rev.流體機械。 2001. 33:415–43版權所有c 2001 by Annual Review。保留所有權利。

變​結 ˚FLOWS

羅傑·辛普森大號 ​航天系與海洋工程,弗吉尼亞理工學院和州立大學
,布萊克斯堡,弗吉尼亞州24061-0203;電子郵件:simpson@aoe.vt.edu

關鍵詞 ​三維,層流,湍流,邊界層,分離

■ ​摘要 ​當邊界層遇到附著在同一表面上的障礙物時,就會發生交匯。討論了
對於層流和湍流接近邊界層觀察到的鈍性和流線型障礙物的物理現象。障礙物
周圍的壓力梯度與繞障礙物的馬蹄形漩渦產生三維分離。除了極低的雷諾數層
流外,這些渦流非常不穩定,並導致高湍流強度,高表面壓力波動和傳熱速率
以及障礙物鼻部的沖刷。還對計算方法進行了綜述。應當使用捕獲大規模混沌
渦旋運動的方法進行計算。還回顧了控制,修改或消除此類渦流的一些工作。

引言

發生在一些實際的空氣動力學和流體動力學的情況下,包括外部氣動,渦
輪機械,潛艇,電子元件的冷卻,和河/橋接流量結流動。在所有這些情況
下,複雜的干擾流場和三維分離都是由表面上的上游邊界層產生的,該邊
界層遇到附著在該表面上的障礙物,例如機翼,渦輪葉片,帆或冷凝塔,
電子芯片或橋墩。在這裡,我們討論了對於層流和湍流接近邊界層交匯處
都已觀察到的可能的物理現象。還審查了有關修改或控制此類流的一些工
作。
首先,需要討論可能影響所有類型交匯處流動行為的物理因素,以建立一個了解其影響
的框架。圖1顯示了3:2橢圓形機頭和NACA 0020尾缸,它們在最大厚度
位置連接在一起,形成了“ Rood”機翼,其厚度(T)與弦長(C)的比為
0.24(以其設計者EP Rood命名) 。對於第一近似,障礙物周圍的壓力梯
度由障礙物的形狀,大小,對流道的阻塞以及附近的其他障礙物確定。順
流不利的壓力梯度引起的方法邊界層分離和形成多個馬蹄旋渦

,0066-4189/ 01 / 0115-0415 $ 14.00 ​415


2000年11月13日18時32年度綜述AR117-14
416 ​SIMPSON

圖1 ​圍繞馬蹄形渦流系統零迎角時的流線型障礙物(頂翼)。實線分離線穿過平均
流量鞍點S。虛線表示後沿分離。虛線表示湍流情況下的低平均剪切線。
2000年11月13日18:32年度評估AR117-14交界
流 ​417,

而翼展方向的壓力梯度導致近壁流繞障礙物移動。主要的馬蹄形渦旋與進
近邊界層渦旋具有相同的旋轉,而較小的次要渦旋具有相反的旋轉以保持
流線型拓撲。停滯點出現在表面上。這一點將下游/上游流動區域和繞過障
礙物側面的流動分開。分隔線穿過該停滯點並繞過障礙物的每一側。障礙
物兩側的流動使馬蹄形渦流向下游方向伸展。顯然,障礙物的高度與寬度
之比(H / T)對通過障礙物頂部而不是障礙物上方的流量有一定的影響,
特別是對於H / T ​< ​1的人。障礙物的鼻子還影響撞擊在障礙物上的水流中
,有多少流向下流向鼻子,流向上游邊界層表面並返回上游。因此,人們
會期望鼻子鈍鈍會影響停滯點和分隔線的位置。可能影響接合點流動的其
他因素的雷諾數和位移厚度(​δ∗​​ 包括進近邊界層),自由流湍流水平,表
面粗糙度以及障礙物周圍的邊界層,分離和渦旋。
如我們所見,在障礙物前面脫落的多個分離的馬蹄形渦旋就其位置,大小或環流而言很
可能不穩定。高雷諾數渦旋的行為和無粘性渦旋理論表明,這些馬蹄形渦
旋及其圖像誘發了自身和鄰近渦旋上的速度分量。渦流擴散到相鄰的質量
塊中,因此儘管其中一些渦流在某些條件下可能彼此“越過”,但它們也可
能合併成一個新的組合渦流。渦流的伸展似乎也是一個因素。因為已知在
不利壓力梯度的存在下,渦旋絲更容易產生不穩定的曲折,所以人們可能
會懷疑這發生在障礙物下游部分周圍的這些馬蹄形渦旋上。這種可能的結
果之一就是這些細絲幾乎纏結在一起,形成了一個“爆發”渦流結構,該結
構看起來像是一個更大的渦流。儘管這裡的討論是針對不可壓縮的情況,
但Lakshmanan&Tiwari(1994)和Batten等人(1999)討論了一些有關超音
速流動的工作,其中,機翼前部的衝擊會引起分離,並且產生與上述相同
的渦旋現象。
由於這些影響接合點流動的現象,在實際情況下會發生許多不良影響。由於主要的馬蹄
形渦流具有接近邊界層的旋轉感,因此它們沿障礙物夾帶了較高速度的自
由流流體,並增加了接合區域的阻力和熱傳遞。馬蹄形渦流的通常不穩定
的特性還可以增加鼻部區域中的熱傳遞和表面壓力波動水平。橋墩周圍的
馬蹄渦流結構會沖走土壤和岩石,並削弱橋基礎。控製或修改這些接合點
流旨在減少這些不利影響。
2000年11月13日18:32年度評估AR117-14

418 ​SIMPSON

其餘部分介紹了一些結果,這些結果描述了圍繞特定障礙物的層流和湍流的某些現
象。對於圓柱體,流線型機翼和安裝在光滑平面上的塊,已經做了很多工
作。但是,很少有實驗可以定量地檢查所有現象,而很少有計算可以捕獲
流的所有物理性質。由於篇幅所限,這裡不討論其他作品,但會在此處討
論的文章的參考文獻中引用這些其他作品。

LAMINAR JUNCTION流
過。

該機制僅在低接近邊界層位移厚度雷諾數重新進行​δ* ​= U∞δ*​/ ​小於 ​〜1000ν


,​其中 ​U∞​是接近自由流速度(Ballio等人,1998)。在一個調查文章,
Ballio等人分析可獲得的信息,包括其自己的結果, 的位置個​×S 在直徑T的

圓筒的前對於時間平均滯流點(貝克1985)雖然X​S​/​T趨於隨著增加,​δ∗​​ /​T
的稀疏數據量的散佈無法確定​Reδ​∗ 依賴性,甚至不能將T作為主要縮放參

數。
現有證據表明,隨著雷諾數的增加,圓柱狀障礙物的三維分離拓撲從單個穩定的初級
馬蹄形渦旋變為穩定的多個渦旋,如圖2所示,並具有復雜的T​/δ​∗ ​< ​22和
Re​T​<​7000(Baker 1979:圖6)。托馬斯(1987)中觀察到的穩定的渦
struc- TURE用於重新 Ť<1000
​ ​Kaul等(1985)來計算周圍重新的圓柱障礙
​ ​1000,發現一個主渦流和X​S​/​T具有一致實驗數據。為 ​δ/​T ​=
物穩定層流​T =
0.1,其中 ​δ ​是計算單個主渦旋重新接近邊界層厚度​T =
​ ​500中,兩個穩定主
渦用於重新​T =​ ​1500,和用於重新穩定主旋渦​Ť3 ,Visbal(1991)=
​ ​2600.
Visbal報告說,最上游的鞍點不在水面,而附著線上的鞍點在水面上游,
而不是分離線,這一點已由Chen&Hung( 1992年)和Coon&Tobak(1995
年)的實驗。正如Visbal指出的那樣,圖2所示的對稱平面中的多個渦旋拓
撲和Visbal的拓撲滿足了鞍形拓撲點/節點的約束(Hunt等,1978)。對於
穩態,Baker(1991)提出流動穩定性可以解釋為什麼渦旋數隨雷諾數增加
而增加。他推測,在高於臨界渦旋核心雷諾數的情況下,一次渦旋是不穩
定的,這導致其分解為其他具有較低渦旋核心雷諾數的渦旋。貝克提到,
三個主要渦旋及其反向旋轉夥伴的極限可能是由缺乏額外渦旋的空間所決
定的。必須提到的是Visbal只考慮了氣缸/板接合處流場的對稱半在他的數
值計算,其可以具有
2000年11月13日,18時32年度綜述AR117-14

JUNCTION FLOWS ​419


圖2 ​簡化模式​:(1)6​-渦流系統; (​b)
​ 四渦旋系統; (​三) ​ 雙渦旋系統。 ​S,
​ 分
隔線; ​A​,附件線; ​SP​,停滯點。 (在Baker 1979年之後。)
2000年11月13日18:32年度評估AR117-14

420 ​SIMPSON
抑制了在較低Re處的不穩定脫落​T​,因為不存在不對稱不穩定。
托馬斯(1987)報告了三不穩定主渦流比範圍1000​< ​再 Ť<13000
​ ​和2.4 ​<T​/δ* ​<13.6,
與無量綱脫落頻率fT​/ ​U∞〜3×10​-​5(​重​筆 -​ ​1000),其中f是脫落頻率。
Visbal的計算還顯示出循環過程中的三個主要渦旋(圖3)與Thomas觀察到
的一致,並且表明不穩定過程的頻率隨雷諾數增加。 F T的計算Strouhal數​/
U∞≈0.21 ​重新​T =
​ ​5400是與近似協議的fT​/ U∞=
​ ​0.17由Thomas對於不同的H
/ T和T報導​/δ*​。 Thomas報告沒有對T一致的影響​/δ​∗ ​對其結果產生,並且兩
項研究均使用H / T ​> ​2。如圖3(根據Visbal的計算)所示,馬蹄渦旋3非常
靠近圓柱,最大渦旋極大地移動了。隨著渦旋逐漸向圓柱體移動,由於渦
旋拉伸而增加。渦流橫截面尺寸減小,渦度梯度增加,這導致粘性耗散增
加。托馬斯(Thomas)表明,該渦旋收縮為兩個對流,對流的下游。由於
渦旋3的感應速度場,相反符號的渦旋遠離平坦表面並圍繞馬蹄渦旋繞軌
道旋轉,並最終一路環繞渦旋3以減少其整體循環。無論哪種情況,無論
渦旋3收縮還是消散,渦旋2然後都向前移動以代替舊渦旋3,並與它的任
何殘餘物合併(圖3​f​,​g​)。 Seal等人(1997年)也從他們的實驗中得出了
相似的結論,並指出在附著鞍點下游的衝擊剪切層中的不穩定性可能決定
了分離或脫落的頻率。
Baker(1991)討論了關於對流不穩定馬蹄渦流性質的兩個假設:(​a)​ 由整個圓柱體
上游的整個系統(​b​的振動引起的;)由初級渦旋的渦流核心的振動引起
的結構體。貝克使用除托馬斯以外的數據,建議在T高值下​/δ​∗的中​,隨著雷

諾數Re​T的 增
​ 加,二次振盪首先出現,這是因為,渦旋的不穩定性。 為核

Reδ​∗ > ​在頻率150時的​˚Fδ*​/ U∞≈0.01。

​ ​這些振盪似乎能引發更強的主振盪為

的值​Ü∞(Tδ*)1/2​/ν> ​1000。 這些更強的振盪發生在頻率​


​ fδ​*​/ U∞≈0.08,
​ ​並持續

最多至​U∞(​Ťδ*)1/2​/ν≈1600, 在
​ 該流動變得完全湍流。在較低值 的T下​
​ /δ​∗​,
似乎振盪僅在單個頻率下發生,因為次級和初級振蕩的頻率彼此接近。從
所有這些研究中可以清楚地看出,不穩定過程是由馬蹄形渦流陣列的自然
不穩定性引起的,而不是由圓柱體後的卡曼渦旋脫落引起的。

-------------------------------------------------- -------→ ​圖3 ​的序列(​一-​ ​小時​在對稱平

面中計算流線圖案在的前面的)。 一個圓形圓柱體為運動的一個週期中,Re​
​ T =

5400流由右至左初級渦旋標記為1-4。 (摘自Visbal1991。)
2000年11月13日18:32年度回顧AR117-14交匯

流 ​421
2000年11月13日18:32年度回顧AR117-14
422 ​SIMPSON

Khan等人(1995年)報告了動量厚度雷諾茲的可視化結果數值​Reθ <
​ ​100的圍繞Rood機
翼流動,該機翼不如圓柱鈍。對於Re​T​< ​2500,由外部流向三個初級渦旋和
兩個反向旋渦的穩定渦旋系統供電。對於2500 ​< ​再 Ť<3500,
​ ​該渦流系統
振盪具有渦旋合併和包裹彼此圍繞。對於3900 ​< ​Re​T ​< ​6000,系統的下游
馬蹄渦流會周期性散發和分裂,這是托馬斯(Thomas,1987)報導的一個
圓柱,但在上游產生第四渦流之前沒有發生。該無量綱脫落頻率f T​/ U∞〜

0.28 ​是大約兩倍的由托馬斯在相同的重報告​標準 範圍。

Baker(1991)定義了等效的“直徑” T​e​為非圓形障礙物,因為該障礙物前方的分離距離
X​s 乘以的直徑與分離距離之比或T​
​ /​X​s 圓柱體。他用這個等效直徑形成了與

矩形棱柱體和圓柱體相似的脫落渦旋頻率相關性。 Seal等人(1995年)也
報導了在矩形鈍體周圍類似的不穩定的渦流行為。當在層流​​分離的上游施
加表面吸力時,Seal&Smith(1999a)報告說,不穩定的棚棚共同旋轉的
馬蹄形渦流的流向支路會纏繞在一起。

湍流流

交匯大多數實際的交匯流都是在湍流進場邊界層發生的。大量研究表明,
障礙物/身體交叉點周圍的時間平均地表油流。並非所有這些較早的研究都
在這裡進行了綜述。亨特等人(1978年),貝克(1980年),Dargahi(
̈01̧cmen&
1989),皮爾斯和鑫(1992),和 辛普森(1994)舉出很多較
早引用。 Sieverding(1985)顯示了渦輪機械端壁周圍的馬蹄形渦流模
式。 Ballio等人(1998年)回顧了有關分離鞍點的位置和對稱平面中平均
渦旋位置的數據。圖4顯示了在許多實驗中使用的Rood機翼前部的三維分
離。時間平均油流中顯示了三維停滯點(奇點)(圖4);從該點開始,
一條“閉合”的時間平均分離線位於機翼兩側。在這個停滯點和鼻子之間存
在時間平均的回流。低平均剪切線也位於該閉合間隔和機翼鼻之間,並且
在模型側面與該閉合間隔合併。
Hunt等人(1978年)和貝克(1980年)提出了一種類似於圖的平均流四渦流系統​2b 基 ​
於表面油流可視化和運動學考慮,針對鈍體的。一些計算和激光多普勒測
速(LDV)測量(Pierce&Tree 1990)報告了圓柱淚滴模型前的平均流兩
個渦旋系統。圖5表示Rood機翼前方的平均雙渦旋系統。但是,這種平均
渦流模式不能代表高度不穩定的瞬時流。
2000年11月13日18時32年度綜述AR117-14

JUNCTION流動 ​423
圖 ​4(​上​周圍的3三維流)表面油流動的可視化:2橢圓鼻/ NACA 0020尾翼垂直安裝到所
̈01̧cmen&
述表面上。白點是的MEA- surement位置 辛普森(1995年)。實線表示在圖7
̈01̧cmen&
所示的激光多普勒速度測量,通過製成的平面 ​
辛普森(1997年b)。 (​下)
測量和計算的輪廓(二維無粘性)平均靜壓係數(C​p​在測試牆上的)。符號表示的的測
̈01̧cmen&
量位置 ​ 辛普森(1995)。

流線形周圍的不穩定現像在
此強調作者和他的同事的工作,因為馬蹄渦旋“系統”和Rood機翼的不穩定
“雙峰”性質已成為許多研究的主題。在進近動量厚度處圍繞Rood機翼使用
三速度分量LDV測量Reynolds
2000年11月13日18:32年度評估AR117-14
424 ​SIMPSON

圖5 ​在前對稱平面上的時間平均速度矢量魯德翼在重新 ​θ= ​6700。該油流動的可視化,圖


4中,示出了在分離的x​/​T ​= -​0.47,並在X一個線低剪切的​/​T ​=-​0.28。 x處的x分量速度波
動的直方圖​/​T ​= −​0.204,y​/​T ​= ​0.0046插圖。 U速度分量的雙峰流動區由實線包圍。 (從
Devenport&辛普森1989年b。)

號​Reθ6300 的,Devenport&辛普森(1987A,B)是第一個報告雙峰速度概

率現象,顯示為圖5中的插圖;這種現像是造成該地區較高的表面壓力波動
水平和較高的熱傳遞率的原因。該圖還顯示了時間平均渦旋的橢圓形。
Devenport&Simpson(1988,1989a,b,1990a)的進一步研究表明,該區
域的流量在一定的時間間隔內從一種模式非週期性地切換到另一種模式,
其時間密度函數很可能是馬爾可夫函數,如Martinuzzi等人(1992)所暗示
的那樣,其工作將在後面討論。由於這種低頻,自感應的混沌轉換,在該
區域中發現了湍流能量產生和湍流應力,其數量級高於上游邊界層。
Devenport&辛普森(1987年b)發現,正常雷諾應力的生產方面, ​-​U

2∂U/∂x ​和 ​-​V 2∂V/∂Y,​如同樣朝向湍流動能的生產速率是重要的(等於k)
作為通常的剪切應力的生產, ​-​ ​U,v,∂U/​ ∂​年。 ​
流動可視化提供了對產生雙峰速度的渦旋系統的一些理解。 Dargahi(1989)顯示了不
穩定渦旋模式的後果。的不穩定渦流
2000年11月13日18:32年度審查AR117-14交界依次發生
流 ​425

和兩個在圓柱體前面的次級渦流。流動可視化結果表明,在靠近壁的分離
流動區域中的平均流動方向主要是相反的方向。但是,沒有補充這些可視
化效果的定量數據。
氫的氣泡流的可視化的實驗和LDV測量由Kim等人(1991),誰在方法水隧道雷諾數重
新使用的魯德翼的2英尺弦清楚有機玻璃模型完成​θ​,基於所述momen-
TUM厚度(​θ​)為330和1100(T​/θ= ​56.6)。從視頻中可以看到,圖6顯示
了鼻子前自我誘發的不穩定現象的示意圖。在一系列流動事件(​a​)的開
始,鼻子前面存在一個大的馬蹄渦,這是由於高速自由流流體撞擊並向下
移動鼻子而產生的。由於這種流動的渦流線圍繞機翼伸展,因此渦流的橫
截面積會隨著時間的增加(而減小​b–g​)。同時,在分離物(下游形成次
級分離渦流​b​)的,並隨循環強度(增加而增加​c 和 ​ ​d)
​ 的。可能會形成其
他三次渦流(請參閱 ​c 和 ​ ​d)
​ 。有時,次級和第三級渦流合併在一起,與
馬蹄形渦流的前部合併,或者以“跳蛙式”的方式在馬蹄形渦流上移動,然
後再與馬蹄形渦流合併。由此產生的合併產生了一個更強大的大馬蹄渦流
,該渦流在機翼周圍伸展。在此非週期性序列的此階段中,前向流向機翼
​ ,並且該加速度會暫時穩定流。然後,流動變得不穩定(​f
(靠得很近​e)
),形成了新的大規模馬蹄渦,並且非週期性過程重新開始(​g) ​ 。在未
發布的某些視頻序列的再分析中,Devenport指出以下幾點:(​i​)在任何
時刻,似乎只有足夠的空間可容納一個大型馬蹄形渦旋;​(ⅱ)​第二和第
三渦流經常圍繞翼與所述馬蹄形渦流合併前掠; (​iii​)在第五階段,馬蹄渦
流似乎因粘性而消散或變得不連貫(見 ​e) ​ 。
​ 峰行為存在的主要因素。機翼的形狀是主要因素-因
如此處所示,進場流雷諾數 ​不是 雙
為它在很大程度上決定了機翼前面和周圍的壓力梯度,而壓力梯度又決定
了機翼周圍“馬蹄渦”的伸展率。當最靠近機翼鼻部的“馬蹄形渦流”被拉伸
時,在三維分離的下游形成了新的,更年輕的渦流。隨著時間的流逝,年
輕的漩渦會隨著時間的增長而增長,而最古老的漩渦則會通過拉伸而減小
大小。最終,這些新的渦流中的一個或多個與最舊的渦流合併,形成一個
最靠近鼻子的“舊”渦流。然後重複該過程,儘管該過程從來都不是周期性
的,並且似乎是馬爾可夫式的。 Agui&Andreopoulos(1992)和Agui等人
(1993)從圓柱前的煙流可視化報告了類似的情景,方法為​Reθ =
​ ​1840和
2.2 ​× ​10​4​< ​Re​T​< ​2.2 ​× ​10​5​。他們的數據顯示,大的負渦旋通量漂移之後總
是強烈的正渦旋通量,這表明蘑菇形蘑菇
2000年11月13日18:32年度評論AR117-14
2000年11月13日18:32年度評論AR117-14接合

流 ​427
次級渦流總是由初級渦流引起的。序列的頻率與下游的卡門脫落頻率無
關。
Hasan等人(1985年)測量了表面壓力波動,這表明由於結流現象,橢圓形鼻樑3:1前
面的低頻成分增加了。 Rife等人(1992年)研究了雙峰速度與Rood機翼/機
體連接處鼻子前部表面壓力波動之間的關係。速度autospectra顯示雙峰區
域由圍繞F T低頻率波動所支配​/ Ù∞=
​ 1/20,​其中,f是平均頻率,而T是最
大翼厚。在這些頻率下的速度和表面壓力波動之間發現了很強的統計一致
性,可以觀察到大規模的不穩定雙峰結構。
Shinpaugh(1994)在Devenport&Simpson的Rood機翼/機體連接風洞流的雙峰區域中,
以50 Hz的頻率快速掃描了垂直於機體表面的兩個速度分量的LDV。以雙峰
區域(X下的雙峰表面壓力水平​/​T ​= −​0.26)作為事件觸發點,所得的事件
閾值速度條件平均數支持單個主結旋流的流動模型,該模型改變了前部的
大小和位置。機翼。當表面壓力高於平均值時,結旋渦集中在鼻子附近。
當壓力信號低於平均值時,渦流離鼻子越遠,鼻子附近的回流就越小。這
些結果與Kim等(1991)的水隧道研究結果沒有衝突。由於上游分離渦流
的通過或與主渦流的合併與主馬蹄渦不同步,因此從壁壓水平觸發的集合
平均速度將使這些較小尺度的渦流速度平均。
̈01̧cmen
的Devenport&辛普森三速度分量LDV風洞數據(1990年a,B,1992)和數據
&​
辛普森(1997年b)表明,該雙峰結構存在速度分量垂直於馬蹄形渦流
核心周圍Rood機翼的前部,氣流在加速,橢圓形的渦旋被拉伸。圖7是基
剪切
於大量實驗測得的雷諾平均分佈圖;它示出了二次流流線和 ​-​U​V/​ ​U2​ ∞ ​
應力輪廓線的平面的的上游 ​
,其包括圖4的站5和垂直於翼面最大厚度。換
句話說,描述此次要流動的速度分量在此平面中。該平面上的機翼表面位
於S​/​T ​= ​0。 ​y ​與人體表面的法線距離以對數繪製,​10 比例以放大近壁細

節。大馬蹄渦旋位於該圖的上部中央。高速自由流流體在機翼附近被該渦
旋夾帶,並且該流體攜帶低湍流動能。較低速度的流體

←−−−−−−−−−−−− ------------ ​圖6 ​用於在的鼻部區域中的流事件序列描述模型 魯



德翼主體結。括號中的百分比表示事件的大致時間比例。 (從Kim等人
1991)2000年11月13日18時32年度評論AR117-14

428 ​SIMPSON
輪廓(粗線)表
圖7 ​平均數二次流(V,W)的流線(細線),和雷諾剪切應力 ​-​U​V/​ ​U​2∞ ​
示所示平面中Rood機翼-機體的接合流。 ​
圖4輪廓線1為0.0045;連續的2–9級,AF降低了

0.00033。 (來自 ​ ̈Oçcmen&Simpson 1997b。)

處於這種高速流動之下,並且也位於渦旋的另一側。更高的湍動能被這種
流體傳輸。伴隨雙峰結構,非常高的湍動能流體和剪切應力位於渦旋內部
和下方。
S附近的反義角渦​/​T ​= −​檢測到0.056。另一個大正常到的表面速度相對義渦流示 ​〜S​/​T
〜 ​-​0.56,其緊密地對應於通過在關閉分離滯流點經過表面油流表面摩擦線
的位置圖4.請注意,機翼和封閉式分離渦流之間沒有其他具有大法向壁速
度分量的渦流結構。因此,始於機翼機頭前部的低表面剪切應力線不得符
合此平面上的分離線要求。在最大厚度的下游,未觀察到雙峰結構(
Devenport&Simpson 1990b,1992),渦旋變得更圓(Fleming等,1991)
,並且由於不利的影響,鼻部區域的連貫渦旋結構可能“破裂”。壓力梯
度。
2000年11月13日18:32年度評論AR117-14交匯
流量 ​429
鼻形
Mehta(1984)和Rood(1984c)的影響表明,時間平均渦流的大小和強度隨著圓柱鼻的鈍度而增加。對於
那些形狀和其他形狀,Fleming等人(1991)開發了一種“鈍度因子”與機翼形狀對馬蹄渦的拉伸速度的影
響。 “鈍度因子”(BF)定義為
1​
BF ​= ​ 2
Ro​ [​ ​TX T​
S​
​ ​ T​XT
S​T + ​
]​, ​(1)
其中厚度 ​Ro​ 為
​ ​ 前沿 ​ST​ ​為距離半徑從 ​XT​ 開始, 是沿著翼型的最大弦到最大厚度的弦前緣位置。假設的非粘
​T, ​
性勢流二維模型,Fleming等人在在零迎角和14的形狀計算的平均從鼻子拉伸速率為28翼型形狀的最大厚度
12◦ ​
10攻擊的角​-​3 ​ ​<BF ​< ​1.07,包括以下討論的內容。平均渦旋拉伸速度(VS)從鼻子到的最大厚度(T)為
R​
​ ​ 0(V
VS =
s&∞S)​噸​; ​(2)它是適合以高的相關係數,並通過定
​ ​10​B(BF)​米​ ​,(3)
VS =
​ ​0.0142適用於12​度 ​攻角的
​ ​0.8816和 ​b ​= ​0.0482零角度的攻擊的情況下,且 ​m ​= ​0.8075和 ​01̧cmen̈B =
其中 ​m =

情況。 ​&Simpson(1994)了實驗性的表面壓力波動和 對六個不同形狀的機翼的機翼表面/機體連接處的機



頭的鼻子形狀(鈍度)對雙峰渦流行為的影響進行其他測量。更高的鼻子鈍度BF產生了更強的鼻子渦流系
統。桑迪亞1850與BF ​= ​0.0133和NACA 0012與BF ​= ​0.0287產生沒有明顯的雙峰旋渦結構或更低的壓力波動
,而NACA 0015與BF ​= ​0.0452和模型具有較高BF值一樣。 BF圓鼻淚滴模型 ​= ​1.07的產生的最大渦旋拉伸參
數為1.13。沃爾特和帕特爾(Walter&Patel(1996))報導,在一個薄的半橢圓形前面沒有馬蹄形渦流結構
,C / T ​= ​H / T ​= ​6(BF ​= ​0.0189);但是,前緣掃過減少了接合處的向下流動。在尾流中觀察到弱結渦的證
據。
Wake Flow and Drag
Fleming et al (1991, 1993) examined available data sets for Rood wing/body junction flows from several different
research groups over a range of approach momentum ​δ/​T. The product Reynolds of Renumbers ​T​and Re​θ​Re—or ​θ
and boundary layer-to-wing-thickness ratios Fleming's approach flow “momentum deficit
November 13, 2000 18:32 Annual Reviews AR117-14

430 ​SIMPSON

factor” (MDF)—correlated observed effects on the flow in the downstream


adverse pressure gradient region around the wing, the corner separated flow, and
in the wake. As MDF increases, streamwise velocity distortions decrease,
secondary flow patterns are more elliptic, the vortex core and the vorticity are
closer to the wall, the spanwise spacing of the legs of the vortices increase in the
wake, and the core vorticity and helicity levels increase when
nondimensionalized on ​δ ​and U​0​. Rood (1984a,b) and Rood & Keller (1984)
observed low-frequency antisymmetric structures on each side of a wing in the
wake, which they attributed to vortex shedding at the trailing edge.
Roach & Turner (1985) reviewed earlier strut drag data. Together with their own data, for
Reynolds numbers based on the strut chord C (Re​C​>​4 ​× ​10​5​), they determined
that the incremental increase in drag D caused by the juncture flow was given by
D ​= ​1.9q​δ∗​​ T, where q is the free-stream dynamic pressure. They noted that
neither the aspect ratio of height to chord (H/C) nor the cross-sectional shape of
the wing has much influence on the incremental drag increase. This dependence
on ​δ∗​​ T is closely related to Fleming's MDF (​∼θ​T) that correlated the wing wake
behavior for the Rood wing. Some ways to reduce the drag are discussed under
“Control of Junction Flows.”

Effect of Angle of
Attack
For a nonzero angle of attack, the approach flow encounters a more blunt nose,
and a stronger horseshoe vortex structure is formed. Shizawa et al (1996) inves-
tigated a constant-thickness cylindrical wing with a semicircular nose (C/T ​= ​6,
H/T ​= ​2, ​δ/​T ​= ​0.44). As the angle of attack increased ​<​15​◦​, the turbulent kinetic
energy and shearing stresses increased on the suction side; the downstream leg
of the horseshoe vortex structure remained at the same distance from the flat
surface, but it shifted away from the wing toward the free-stream direction. On
the pressure side, the downstream leg of the vortex moved away from the
juncture but at the same distance from the wing.
Wood & Westphal (1982) used a cylindrical NACA 0012 airfoil (H/C ​= ​20) at a 10​◦ ​angle of
attack with an approach boundary layer of Re​θ ​= ​4200. The pressure distribution
near the junction was slightly affected, with a 16% decrease in the section lift
coefficient. The suction-side horseshoe vortex leg was stronger; it distorted the
wing wake within the body boundary layer and resembled a sin- gle vortex
embedded in a boundary layer far downstream in the wake. Wood & Westphal
reported that the shearing stresses behave qualitatively like eddy viscosity
turbulence model results.
Rood & Anthony (1985) showed that the shape of the tail of the wing has an effect. For a
wing at a slight angle of attack, a tail that produces a greater adverse pressure
gradient increases the size of the vortex legs and the area over which their
̈ ̈
unsteadiness is felt. Ozcan & Ol ̧cmen (1988) observed the surface skin
friction topological nodes and saddle points downstream of a wing at an angle of
attack.
November 13, 2000 18:32 Annual Reviews AR117-14

JUNCTION FLOWS ​431

Effect of Sweep of a
Wing

Ahmed & Khan (1995) examined the effects of leading edge sweep angle –45​◦ ​<
< ​45​◦ ​on the juncture flow structure around a Rood wing on a flat surface at zero
angle of attack. For increasing backsweep, ​< ​0, the separation lines moved closer
to the wing as the vortex moved downstream and greater flow moved down
toward the flat surface; opposite trends occurred for forward sweep. The
maximum nose region mean reversed flow velocity next to the surface increased

slightly for increasing backsweep –30​◦ < < ​0◦​​ , but was lower for ​= ​–45​◦​; for
forward sweep, it decreased continuously to almost zero for ​= ​45​◦​. The strength
of the mean vortex seemed to follow these trends of mean reversed flow

velocity. For forward sweep of ​= ​20​◦ for the NACA 0015 wing at several angles
of attack, Bernstein & Hamid (1995) found from pressure distributions and wake
surveys that the section lift decreases and the pressure drag increases as the
junction is approached. Arnott et al (1996) found that for negative and small
angles of attack, the pressure drag coefficients away from the junction were
negative and, thus, the flow interaction with the junction was favorable. Walter
& Patel (1996) observed a downwash in the wake that was caused by the
backward trailing edge sweep from the top of their ellipsoidal wing; this
produced a momentum surplus rather than a deficit.

Effects on Heat Transfer Around Streamlined


Wings
Lewis et al (1994) measured the time- and spatially resolved surface heat flux in
the nose region of three different heated wing/body junctions that had showed
bimodal velocity and pressure phenomena: NACA 0015, Rood wing (BF ​=
0.32), and the circular-nosed teardrop model. No bimodal heat-flux fluctuation
histograms were observed on the flat surface. As in other works cited by Lewis
et al for other streamlined shapes and turbine blades, local mean heat flux in
front of the nose increased up to a factor of 3 over the heat flux of the approach
boundary layers. This enhanced heat flux scaled on the local maximum rms of
velocity fluctuations above the surface. The rms of the heat flux fluctuations
were ​≤​25% of the mean heat flux in the vortex-dominated nose region. Some
major findings from this study were the following: (​a)​ The enhancement in heat
flux, in the region downstream of the time-mean separation line and upstream of
the time-mean vortex center, is caused by the high levels of turbulent stresses
near the wall produced by the large-scale unsteadiness of the horseshoe vortex;
and (​b​) the maximum level of heat flux occurs in the immediate vicinity of the
wing/endwall junction because of the effects of the vortex on the mean
flowfield. The main vortex transports cold outer boundary layer fluid down the
leading edge of the wing, perpendicular to the endwall, where it impinges on the
endwall. The strength of this impinging jet is related to the size and location of
the primary vortex that is controlled by the bluntness of the wing nose.
Wroblewski & Eibeck (1992) and other cited coworkers examined the
heat-transfer mechanisms downstream of a circular-nosed teardrop model. This
November 13, 2000 18:32 Annual Reviews AR117-14

432 ​SIMPSON
highly blunt nose shape produced large-scale unsteadiness of the horseshoe
vortex structure that resulted in ​≤​50% increase in heat transfer downstream.

Sharp-Edged Bluff Body


Flows
Larousse et al (1991) and Martinuzzi et al (1993) report bimodal velocity proba-
bility distribution functions in front of prismatic obstacles and show a horseshoe
vortex behavior that is similar to that for streamlined wings. The instantaneous
unsteady four-vortex model of Figure 2​b​is supported by these data.
Mode-averaged velocity vectors in the plane of symmetry show two modes that
look much like those of Devenport & Simpson. The stronger backflow mode has
a large ellipti- cally shaped vortex in front of the cube with the sense of the
spanwise vorticity of the on-coming boundary layer; a much weaker elliptical
vortical structure of the same sense is upstream, and a very small counterrotating
structure lies on each side of the large structure. Using particle-image
velocimeter data, Praisner et al (1997) reported a similar structure in front of an
H/T ​= ​1.32 block that was 0.255 T long. They also report local surface heat flux
of up to three times that for the approach boundary layer; for his block, Blair
(1985) reported twice the heat transfer rate in front and 2.6 times greater near the
front of the side. Similar results are reported by Meinders et al (1999).
Figure 8 shows the main features for small aspect ratio obstacles (Martinuzzi & Tropea
1993b). This schematic model was constructed from surface oil-flow
visualizations, surface pressure distributions, and detailed LDV measurements.
Martinuzzi & Tropea (1993a) show that the time between bimodal switches T
appears to be a Markovian process, that is, one whose probability density func-
tion is (1​/​B)exp(​− ​T​/​B), where B is a normalizing parameter. In the plane of
symmetry, Martinuzzi et al (1993) found that the largest turbulence kinetic
energy production rates coincide with areas of maximum vorticity in the shear
layer that

Figure 8 ​Schematic representation of the flow around a surface-mounted cube. (From


Martinuzzi & Tropea 1993b.)
November 13, 2000 18:32 Annual Reviews AR117-14
JUNCTION FLOWS ​433

envelops the obstacle and along the horseshoe vortex system. The production
rate for ​u2​ ​contributes most, with the production rates for ​v ​2 ​and ​w2​ ​nullifying
one another. Martinuzzi & Tropea (1993a) suggest that large-scale motions,
especially the large spanwise stretching motion ​∂​W​/∂​Z, are responsible for this
nullifying exchange. This stretching rate is related to the bluntness of the
obstacle. In the horseshoe vortex and the wake recirculation region, the
turbulence kinetic en- ergy production rate of the normal stresses is comparable
or larger than the shear stress production rate. In the wake and downstream
recovery region, the convected turbulence kinetic energy is dissipated locally.

Scouring Junction
Flows
When a river flow encounters a bridge pier, the horseshoe vortex phenomena
that are described above develop around the base of the pier, scouring the river
bed and possibly removing soil and rock. These phenomena have been known to
weaken bridge foundations and cause the collapse of bridges; a large number of
bridges are susceptible to this danger. Dargahi (1990) relates the chronological
scouring processes to the flow behavior. Much of the material removal occurs in
regions with strong velocities, such as around the sides of the pier, and regions
of strong fluctuating horseshoe vortices, such as the nose and wake regions. For
a circular cylindrical pier, the scouring first occurs in the wake because of the
wake vortices and the accelerated side flow. Material at the side is removed, and
then the horseshoe vortices at the nose begin their work in earnest. Dargahi
shows two primary vortices of the sense of the upstream boundary layer that dig
rippled depressions in the bed nose region over time. Evidently, the initial
vortices for the uneroded bed are unsteady, as portrayed by Kim et al (1991), but
become more spatially stabilized once some depressions have been formed.
After some time, the nearly final equilibrium position of the river bed looks much like that
shown in Figure 9, with a scour hole in front and on the sides of the pier without
rippled depressions and a deposition-dune located downstream. In this condition,
the velocities and unsteadiness have been reduced by the increased flow area
around the pier in the hole. The dune deposition occurs because of the
deceleration of material-laden fluid. Among others, Melville (1997) presents
correlations for the depth of the scour hole around piers and abutments.

Turbulent Junction Flow


Calculations
Algebraic isotropic eddy viscosity and mixing length models continue to be
used, even for complex three-dimensional cases. Simpson (1996) discussed the
limita- tions of these algebraic turbulence models for three-dimensional
turbulent boun- dary layer calculations. Such models produce acceptable mean
flow results for pressure-driven boundary layers away from adverse pressure
gradients and sep- aration, even though they do not account for anisotropy and
lags between the mean velocity gradients and shearing stresses. Under these
latter conditions, these
November 13, 2000 18:32 Annual Reviews AR117-14

434 ​SIMPSON

Figure 9 ​Side view of flow around a bridge pier in an erodible bed.

shortcomings and the overprediction of the eddy viscosity produce poorer


results. At least six different groups have used the Baldwin-Lomax turbulence
model to calculate Rood wing juncture flows for the various experiments
discussed by Fleming et al (1993). The results of Sung & Griffin (1991) and
Deng & Piquet (1992) are representative. The surface pressure distribution,
location of the saddle point of separation, location of the mean nose vortex, and
the trailing edge root separation were captured to some degree. Although the
mean flow is fairly well calculated away from the wing, the isovelocity contours
show lower-than-measured values near the wing toward the trailing edge and in
the wake.
The data of Devenport & Simpson (1992, 1990b) and Fleming et al (1993) for the AOE​/​VPI
Rood wing flow are included in the ERCOFTAC test case data bank at the
University of Surrey, which is accessed through http://fluindigo.mech.surrey.
ac.uk/. Bonnin et al (1996) and Rizzi & Vos (1998) reported the calculations of
six groups that used versions of the k-​ε​model and one group that used a
Reynolds stress equation (RSE) model. The general mean streamlines of the
flow are captured by the calculations, but the nose primary vortex is too thin and
also is too short in the RSE model results. The best k distribution is obtained by
the renormalization group version of the k-​ε ​model. This model and the RSE
also calculate the best k around the side of the wing. All of the models
underpredict the vortex strength, even downstream, as in the near wake. The
essential inability of eddy-viscosity closures to simulate anisotropic turbulence
is a reason for poor calculations of recirculation regions or intense vortices. The
RSE model predicted k best in the wake, whereas the renormalization group k-​ε
model produced a low k. This comparative work showed some difficulties in
using wall functions for this type of flow.
Devenport & Simpson (1992) tested the closure assumptions for six different models with
their detailed data and found that the Cebeci-Smith algebraic model and an
algebraic-stress model were best at calculating the magnitude of the shearing
̈
stress around the juncture; the k-​ε ​model performed poorly. Ol ̧cmen &
Simpson
November 13, 2000 18:32 Annual Reviews AR117-14

JUNCTION FLOWS ​435

(1996a,b, 1997a) also tested the closure assumptions of seven other second-order
turbulence models with their detailed data. Some models followed the trends of
the data in the outer region and away from the juncture vortex, but none captured
the high Reynolds-averaged double and triple fluctuation products around the
bimodal flow.​Parneix et al (1998) calculated the AOE/VPI Rood wing flow

using the V2F model.


​ This model uses the standard k-​ε ​model, modified for
near-wall turbulence anisotropy and nonlocal pressure-strain effects, and it
retains a linear eddy viscosity assumption. Calculations of the separation line are
similar to those revealed by the oil-flow visualizations, and the wake results are
fairly well represented. In the approach plane of symmetry, the damping of the
near-nose turbulent transport prevents the spurious production of turbulence near
the wing that occurs with the k-​ε​model. Deng & Visonneau (1998) used a
two-equation scalar turbulence model and a near-wall second-moment closure to
compute this flow. The Reynolds stress transport model simulated the
anisotropic behavior of the normal Reynolds stresses and amplification of
longitudinal vorticity. This anisotropy was also responsible for strong values of
shearing stress ​−​uw​ ​near the horizontal surface. It seems clear that such
improvements to Reynolds-averaged models are necessary to capture much of
the important phenomena for complex vortical flows. Naturally, because a
Reynolds-averaged model cannot simulate the chaotic large-scale vortex
structure in front of the nose, the high mean turbulent kinetic energy k around
the mean vortex location is not well predicted. Large-eddy simulations (LES)
should be used to capture the chaotic vortical motions.
Murikami et al (1993) showed that an LES of the Larousse et al block flow pro- duces close
agreement with experimental results, whereas calculations from dif- ferential
Reynolds-averaged second-moment closure models do not. Rodi (1997)
discusses k-​ε ​and LES calculations of this flow by several different groups. Even
with some model modifications, the k-​ε ​calculations failed to estimate
downstream reattachment lengths. This is because the bimodal unsteady velocity
fluctuations are produced by the sequence of unsteady vortex motions.
Reynolds-averaged closure models do not distinguish between large-scale
unsteady vortices and smaller-scale turbulence. The LES methods simulated
most of the complex fea- tures fairly well, even quantitatively. LES is clearly
more suited for calculating these complex vortical flows, but at a much higher
cost.

Control of Junction
Flows
To reduce the adverse effects of the unsteady horseshoe vortex flows,
modifications to the wing/body geometry or the approach flow have been
examined. In one other case, McGinley (1987) used vortex generators
downstream of a wing to produce counterrotating vortices that interact and partly
nullify the circulation in the wakes of the horseshoe vortices under some
conditions. A fairing around the entire base of a wing has been used for many
years on aircraft because, in some situations, it can substantially reduce the
interference drag of the junction (Hoerner
December 1, 2000 10:29 Annual Reviews AR117-14

436 ​SIMPSON

1965). Haines (1983) advises a fairing designer to do the following: (​a)​


eliminate flow separations, including those that lead to standoff vortices, (​b)​
reduce cross- flows in boundary layers, (​c​) merge different streams smoothly,
and (​d)​ avoid the development of thick boundary layers.
Devenport et al (1990) examine the effects of a constant-radius circular con- cave fillet (0.53
T) wrapped around the entire base of the Rood wing ​δ​/T of 0.26 and 0.5 and
angles attack​<​12​◦​. This modification had the effect of making the wing shape
more blunt and not preventing the leading edge separation. The bimodal
unsteadiness at the nose was not reduced, and the size and strength of the vortex
legs and the unsteadiness in the wake were not reduced. The fillet increased the
spectral level of surface-pressure fluctuations under the streamwise legs of the
horseshoe vortex structure. This work and earlier cited cases for smaller
radius-to- thickness ratio wrap-around concave fillets showed increases in
overall turbulence kinetic energy in the horseshoe vortex legs, as compared with
the unfilleted cases. With the possible exception of reduction of interference
drag, which appears from these latter investigations to be small at best, the
wrap-around fillet does not modify the juncture flow in a desirable way.
However, Devenport et al (1992) also showed that a leading-edge strake or fillet can reduce
adverse pressure gradients and eliminate the separation, the bimodal
unsteadiness, and the horseshoe vortex for zero angle of attack of the wing. The
magnitude of the surface pressure fluctuations, the cross-stream pressure gradi-
ents, and the nonuniformity of the wake were all reduced from the case without
the strake. Figure 10 shows a front perspective view and side view of surface
oil-flow visualizations of the unseparated flow on that strake in front of the
Rood body at zero angle of attack. At an angle of attack, the leading edge fillet
pro- duces a ​pressure ​side separation, as shown by Devenport et al (1992) for
angles ​<​12​◦​. The LDV measurements on the pressure side (Simpson 1996) show
that lower-velocity nearer-wall upstream boundary layer fluid is diverted away
from the wing by the “bluntness” of the fillet. This accumulation of lower-speed
fluid moves under the higher-speed upstream fluid and forms a separation. The
turbu- lence levels are elevated near the leading-edge fillet because of the
accumulation of upstream near-wall turbulent fluid and not because of
large-velocity-fluctuation fluid structures that produce the bimodal velocity
histograms. No bimodal his- tograms were observed at measurement locations.
Near the maximum thickness of the wing, the mean velocities and the turbulence
values show that higher-speed lower-turbulence free-stream fluid has moved
downward close to the wall near the wing, which indicates a streamwise vortical
motion. Pierce & Shin (1990) also found that flat-sided pyramid leading-edge
fillets on a Rood wing reduced the strength and size of the primary horseshoe
vortex at zero angle of attack, with the most effective fillet studied being one
boundary layer thickness high by two wing thicknesses long. Dickenson (1990)
recommends a constant radius leading- edge fillet that is twice the approach
boundary layer thickness. Bernstein & Hamid (1996) used an asymmetric curved
leading-edge strake on their 20​◦ ​swept-back NACA 0015 wing, which reduced
the horseshoe vortex intensity and spread the
November 13, 2000 18:32 Annual Reviews AR117-14

JUNCTION FLOWS ​437

turbulent activity over a greater region. They concluded that peak turbulence in-
tensities were reduced in the wake, but it was unclear if the total drag was
reduced. Using a water laminar approach boundary layer, LaFleur & Langston
(1993) formed a contoured ice surface around a circular cylinder mounted on a
cooled, initially flat, surface. Upstream of the cylinder, the resulting ice surface
was higher upstream of a formed depression of the order of T wide that was next
to the cylinder. The three large-diameter laminar vortices formed contours of the
depression. To the rear of the cylinder, a fairing ridge was formed. The
depression and the rear fairing ridge looked similar to the zones of scour and
deposition around a bridge pier, respectively (Figure 9). Wind tunnel tests of this
6​
contour shape and cylinder (H/T ​= ​2.5) at Re​T ∼
​ ​10​ and at zero angle of attack
indicated 18% reduction in drag for the same upstream turbulent boundary layer
conditions. It appears that the depression region permits lower velocities and
lower peak vorticity magnitudes around the cylinder because of the increased
flow area. This modification to the junction geometry does not prevent
separation or appreciably reduce the approach vorticity.
Others have removed a portion of the upstream boundary layer by suction, which reduces the
source of the vorticity from which the horseshoe vortex is generated. Philips et
al (1992) used a suction slot 1.07 T wide by 1.36 T long that was directly in
front of a circular-nosed cylinder of radius 0.36 T and 1.45 T height that flared
out to a constant-width, boat-tailed cylinder for the last half of the C ​= ​6.43
T-long model. For a suction volumetric flow rate Q twice that of the approach
boundary layer, the large-scale horseshoe vortex could no longer be found. The
streamwise edges of the suction hole created streamwise vorticity, which made it
impossible to eliminate the circulation. Johnson et al (1994) used a 1.32 T
diameter circular hole centered 1.32 T upstream of a 3:1
elliptic-nose/NACA0020 tail cylinder. This wing shape was selected to be less
blunt than the Rood wing. At the highest volumetric suction rate examined [Q ​=
20.4U​∞​(hole area)], the upstream vortex was eliminated, but a weak vortex was
formed on the solid surface downstream of the hole. Johnson et al (1994) also
examined the role of blowing through the hole. As the blowing rate increased,
the formation of the vortex was delayed, but the plume created by the blowing
was larger than the original vortex, and a structure of greater size and apparent
strength was created.
Barberis et al (1998) demonstrated that the location of the suction slot was criti- cal. They
located a 0.28 T wide rectangular suction slot between x​/​T ​= −​0.33 to ​−​0.10,
where x is the streamwise distance, along the centerline in front of a circular-
nosed/faired tail cylinder; they required only a suction flow-rate of 0.1U​∞​δ​∗ ​T to
eliminate the horseshoe vortex structure. With the suction opening centered
about the zero-suction saddle point upsteam (X​s​/​T ​= −​0.41), progressively
increasing suction caused the horseshoe vortex structure to move toward the
nose; a suction flow-rate of 1.05U​∞​δ​∗ ​T was required to remove the vortex. Seal
& Smith (1999b) applied only a small amount of suction (maximum Q ​= ​0.17 of
approach bound- ary layer volumetric flowrate) through a spanwise slot 0.73 T
long located 0.25 T upstream of a circular cylinder, for a low Reynolds number
turbulent flow. As
November 13, 2000 18:32 Annual Reviews AR117-14

438 ​SIMPSON

observed by Kim et al (1991), the horseshoe vortex system without suction is bi-
stable, and it moves from an upstream general location to a nearer nose location
where the vortex resides for a more limited time. The small amount of suction
weak- ened the instantaneous vortex-surface interactions, weakened the
time-averaged vortex, and reduced the Reynolds-averaged stresses of the
vortices. By apparently making the vortex structure more steady, surface
pressure fluctuations and heat- transfer would be reduced. All of the suction
experiments were conducted for zero angle of attack cases, so their effectiveness
is expected to be poorer at angles of attack at which some upstream vorticity
bypasses the suction slot and impinges on the wing.
Of these ideas for reducing the effects of the horseshoe vortex, the strake, or leading edge
fillet appears to be the best. Haines (1983), Maughmer et al (1989), and
Bertelrud et al (1988) suggest a blending of the wing-body region at the front of
the juncture for a modest reduction in drag. Hoerner (1965) shows that a fairing
downstream of the maximum thickness is important for reducing the drag on
intersecting struts. Such fairings should be designed to reduce the adverse
pressure gradients and eliminate the near-trailing-edge corner separation.
To try to reduce the scour around bridge piers, some have used collars around the pier to
reduce the downflow in front of the pier, even though these collars do not
prevent the formation of the horseshoe vortices. Dargahi (1990) points out that
the collars must be down in the near-bed region to have much effect. Gupta
(1987) used an inverted delta wing upstream to create shed vortices of opposite
sense to the original horseshoe vortices and to modify significantly the shed
horseshoe vortex structure for zero angle of attack. Because the flow direction
can change during floods, this scour countermeasure may not be as effective for
other flow directions.

CONCLUDING
REMARKS

Typical physical features of some practical junction flows around blunt and
stream- lined obstacles have been discussed. The pressure gradients around an
obstacle produce a three-dimensional separation with horseshoe vortices that
wrap around the obstacle. Only for the lowest Reynolds numbers do laminar
junction flows remain steady, limited to three primary horseshoe vortices. A
periodic shedding of the stretched laminar horseshoe vortex next to the obstacle
occurs at higher Reynolds numbers.
For approach turbulent boundary layers, bimodal aperiodic chaotic large-scale horseshoe
vortex phenomena accompany all obstacle/body junction flows with sufficient
“bluntness” and are responsible for observed high turbulence intensities, surface
pressure fluctuations, heat transfer rates, and scour in front of the obstacle.
Recent Reynolds-averaged turbulence models show improved agreement with
ex- periments in regions away from bimodal flow zones. Further improvements
are needed. LES appear to be more suited to the calculation of these flows,
which
November 13, 2000 18:32 Annual Reviews AR117-14

JUNCTION FLOWS ​439


suggests the need for additional length scale and spatial correlation information
that can be obtained from future experiments.
Fairings that reduce or eliminate adverse pressure gradients around an obstacle appear to be a
good way to reduce juncture flow drag. The juncture vortex structure can be
prevented under zero angle of attack conditions by using a properly designed
upstream strake that eliminates the separation. Well-placed upstream suction can
also remove the approach boundary layer and greatly reduce the horseshoe
vortex structure. However, it appears unlikely that one can eliminate the
separation for all angles of attack.

ACKNOWLEDGMEN
TS

Portions of the referenced work at VPI&SU have been supported by Office of


Naval Research, Naval Sea Systems Command, Defense Advanced Research
Projects Agency, Air Force Office of Scientific Research, and the National
Aeronautics and Space Administration. The author gratefully acknowledges the
support of these agencies.

Visit the Annual Reviews home page at


www.AnnualReviews.org

LITERATURE
CITED
cylinders mounted normal to flat pla-
Agui JH, Andreopoulos J. 1992. Experimental tes. ​J. Wind Eng Ind. Aerodyn.
investigation of a three-dimensional bound- :263–74 Baker CJ. 1991. The oscillation of
ary layer flow in the vicinity of an up- right rseshoe vortex systems. ​J. Fluids Eng.
wall-mounted cylinder. ​J. Fluids Eng. 3:489–95 Ballio F, Bettoni C, Franzetti S.
114:566–76 Agui JH, Honkan A, 98. A survey of time-averaged
Andreopoulos J. 1993. In- stantaneous aracteristics of laminar and turbulent
structure of a three-dimensional boundary rseshoe vortices. ​J. Fluids Eng. 1​ 20:233–42
layer flow approaching an upright wall rberis D, Molton P, Malaterre T. 1998.
mounted cylinder. In ​Near-Wall Turbu- lent n- trol of 3D turbulent boundary layer
Flows,​ pp. 813–25. New York: Elsevier para- tion caused by a wing-body junction.
Ahmed A, Khan MJ. 1995. ​Effect of sweep on 出口。 Therm. Fluid Sci. 1​ 6:54–63 Batten
wing-body juncture flows.​ AIAA-95-0868. Craft TJ, Leschziner MA, Loyau H. 1999.
Presented at Am. Inst. Aeronaut. Astronaut. ynolds-stress-transport modeling for
Aerosp.科學。 Meet., 33rd, Reno Arnott AD, mpressible aerodynamics applications.
Bernstein L, Petty DG. 1996. A note on the AA J. ​37:785–97 Bernstein L, Hamid S.
pressure drag of a forward-swept-wing plate 95. On the effect of a swept-wing/plate
junction. ​Aerosp. J. ​100:281–84 Baker CJ. nction flow on the lift and drag. ​Aerosp. J.
1979. The laminar horseshoe vortex. :293–305 Bernstein L, Hamid S. 1996. On
J. Fluid Mech. 9​ 5:347–67 Baker CJ. 1980. The e effect of a strake-like junction fillet on the
turbulent horseshoe vor- and drag of a wing. ​Aerosp. J. 1​ 00:39–52
tex. ​J. Wind Eng. Ind. Aerodyn. ​6:9–23 Baker rtelrud A, Szodruch J, Olsson J. 1988. ​Flow
CJ. 1985. The position of points of max- operties Associated with Wing/ Body
imum and minimum shear stress upstream of nctionsin Wind Tunnel and Flight​.
November 13, 2000 18:32 Annual Reviews AR117-14

440 ​SIMPSON
enport WJ, Simpson RL. 1988. LDV mea-
ments in the flow past a wing-body
ICAS-88-4.3.3,​ Int. Congr. Aerosp. Sci., ction. In ​Proc.詮釋。 SYMP。申請Laser
Jerusalem, Israel Blair MF. 1985. Heat mometry Fluid Mech., 4th, Lisbon, Por-
transfer in the vicinity of a large-scale al,​ paper 2.3. Lisbon: Inst. Super. Tec的。
obstruction in a turbulent bound- ary layer. enport WJ, Simpson RL. 1989a. The
AIAA J. Propuls. Power ​1:158–60 Bonnin JC, u- lence structure near an appendage-body
Buchal T, Rodi W. 1996. ERCOFTAC c- tion. In ​Proc. SYMP。 Naval
workshop on data bases and testing of drodyn., 17th, The Hague, Netherlands,​ pp.
calculation methods for turbulent flows (3–7 –77. Wash- ington, DC: Natl.科學院。
April 1995). ​ERCOFTAC Bull. 2​ 8:48–54 enport WJ, Simpson RL. 1989b. Time-
Chen CH, Hung CM. 1992. Numerical study endent structure in wing-body junction
of ws. In​Turbulent Shear Flows​, ed. F Durst,
juncture flows. ​AIAA J. ​30:1800–7 Coon MD, Tobak aunder, F Schmidt, J Whitelaw, 6:232–
M. 1995. Experimental study of saddle point Heidelberg: Springer Devenport WJ,
of attachment in laminar junc- ture flow. ​AIAA pson RL. 1990a. Time- dependent and
J. ​33:2288–92 Dargahi B. 1989. The turbulent e-averaged turbulence structure near the
​ 出
flow field around a circular cylinder. 進 e of a wing-body junc- tion. ​J. Fluid Mech.
口。 Fluids 8​ :1– 12 Dargahi B. 1990. :23–55 Devenport WJ, Simpson RL.
Controlling mechanism of local scour. ​ASCE 0b. ​An Ex- perimental Investigation of the
J. Hydraul.工程。 1​ 16:1197–214 Deng GB, w Past an Idealized Wing-Body Junction.​
Piquet J. 1992. Navier-Stokes com- putations &SU Rep. VPI-AOE-172,​ Va. Polytech.
of horseshoe vortex flows. 詮 ​ 釋。 J. Numer. . State Univ., Blacksburg, Va. Devenport
Methods Fluids ​15:99–124 Deng GB, Simpson RL. 1992. Flow past a
Visonneau M. 1999. Computation of a g-body junction: experimental evalua- tion
wing-body junction flow with a new urbulence models.​AIAA J.3​ 0:873–81
Reynolds-stress transport model. In ​Proc. kenson SC. 1990. Appendage hull junction
SYMP。 Naval Hydro., 22nd, Washington, ws and their application to sailboat keels.
DC​, pp. 691–707. Washington, DC: Natl.科學 OC。詮釋。 SYMP。 Hydrodyn.
院。 Devenport WJ, Agarwal NK, Dewitz form. Enhanc. Marine Appl., 2nd, Univ.
MB, Simpson RL, Poddar K. 1990. Effects of Kingston,​ pp. 47–58 Fleming J, Simpson
a fillet on the flow past a wing-body junction. Devenport WJ. 1991. ​An Experimental
AIAA J. ​28:2017–24 Devenport WJ, Dewitz dy of a Turbulent Wing- Body Junction and
MB, Agarwal NK, Simpson RL. 1992. Effects ke Flow. VPI&SU Re- port VPI-AOE-179,​
of a leading-edge fillet on the flow past an Polytech. Inst. State Univ., Blacksburg,
appendage-body junction. ​AIAA J. Fleming J, Simpson RL, Devenport WJ.
30:2177–83 Devenport WJ, Simpson RL. 3. An experimental study of a turbulent
1987a. Some time-dependent features of g- body junction flow. 進 ​ 出口。 Fluids
turbulent appendage-body junction flows. In 366–78 Gupta AK. 1987. Hydrodynamic
Symp. Naval Hydro., 16th, Berkeley,​ ​CA​, pp. dification of the horseshoe vortex at
312–35. Washington, DC: Natl.科學院。 ical pier junc- tion with ground. ​物理學。
Devenport WJ, Simpson RL. 1987b. ​Turbu- ds 3​ 0:1213– 15 Haines AB. 1983.
lent structure near the nose of a wing-body odynamic interference: a general
junction​. ​AIAA-87-1310.​ Presented at AIAA rview. Special course on sub-
Fluid Dyn. Plasma Dyn. Lasers Conf., 19th, c/transonic aerodynamic interference
Honolulu
November 13, 2000 18:32 Annual Reviews AR117-14
JUNCTION FLOWS 4 ​ 41
wis DJ, Simpson RL, Diller TE. 1993.
me- resolved surface heat flux
for aircraft.​ ​AGARD-R-712, p​ p. 9.1–9.52. easurements in the wing/body junction
NATO-AGARD, Neuilly-sur-Seine, Fr. Hasan rtex. ​J. Thermo- phys. Heat Transf.
MAZ, Casarella MJ, Rood EP. 1985. An 656–63 Martinuzzi R, Melling A, Tropea C.
experimental study of the flow and wall pres- 93. ​Reynolds stress field for the turbulent
sure field around a wing-body junction. In w around a surface-mounted cube placed in
Shear-Flow-Structure Interaction Phenom- channel.​ Presented at Symp. Turbul. Shear
ena​, 1:89–95.上午。 SOC。機甲。工程。 ows, 9th, Kyoto, Japan, paper 13–4
Noise Control Acoust. Hoerner S. 1965. artinuzzi R, Tropea C. 1993a. Flow around
Fluid-Dynamic Drag,​ pp. 8.10–8.19. urface-mounted cube. In ​Proc. Laser Tech.
Bakersfield, CA: Hoerner Fluid Dyn. 400 pp. 請Fluid Mech., Lisbon, Portugal,​ pp.
Hunt JC, Abell CJ, Peterka JA, Woo H. 1978. 8–414. Heidelberg: Springer Martinuzzi R,
Kinematical studies of the flows around free opea C. 1993b. The flow around
and surface-mounted obstacles: applying rface-mounted, prismatic obstacles placed in
topology to flow visualization.​J. Fluid Mech. ully developed channel flow. ​J. Fluids Eng.
86:179–200 Johnson MJ, Ravindra K, Andres 5:85–92 Maughmer M, Hallman D,
R. 1994.​Com- parative study of the szkowski R, Chappel G, Waitz I. 1989.
elimination of the wing fuselage junction perimental in- vestigation of wing/fuselage
vortex by boundary layer suction​. egration ge- ometries. ​J. Aircr. ​26:705–11
AIAA-94-0293. P ​ resented at Am. Inst. cGinley CB. 1987. Note on vortex control in
Aeronaut. Astronaut. Aerosp.科學。 Meet. mulated sail-hull interaction. ​J. Ship Res.
Exhib., 32nd, Reno Kaul UK, Kwak D, :136–38 Mehta RD. 1984. Effect on a wing
Wagner G. 1985. ​A com- putational study of se shape on the flow in a wing/body
saddle point separation and horseshoe vortex nction. ​Aerosp. J. ​88:456–60 Meinders ER,
system. AIAA-85-0182. ​Presented at Am. Inst. njalic K, Martinuzzi R. 1999. Interaction
Astronaut. Aaeronaut. Aerosp.科學。 Meet., tween the flow field and the lo- cal
23rd, Reno Khan MJ, Ahmed A, Trosper JR. nvective heat transfer of a single wall-
1995. Dynam- ics of the juncture vortex. ounted cube in a turbulent channel flow. ​J.
AIAA J. ​33:1273– 78 Kim SA, Walker DA, at Transf. 1​ 21:564–73 Melville BW. 1997.
Simpson RL. 1991. ​Ob- servation and er and abutment scour: integrated approach.
Measurements of Flow Struc- tures in the CE J. Hydraul.工程。 1​ 23:125–36
Stagnation Region of a Wing- Body Junction. urikami S, Mochida A, Ooka R. 1993. ​Nu-
VPI&SU Rep. VPI-E-91-20, V ​ a. Polytech. rical simulation of flowfield over surface-
Inst. State Univ., Blacksburg LaFleur RS, ounted cube with various second-moment
Langston LS. 1993 Drag reduction of a osure models.​ Presented at Symp. Turbul.
cylinder/endwall junction using the ice-
formation method. ​J. Fluids Eng. ​115:26–32 ear Flows, 9th, Kyoto, Japan, paper 13–5 ​Ol
Lakshmanan B, Tiwari SN. 1994. Investi-
gation of three-dimensional separation at men ̈MS, Simpson RL. 1994. Influence of
ng shapes on surface pressure fluctuations
wing/body junctions in supersonic flows. ​J.
Aircr. 3​ 1:64–71 Larousse A, Martinuzzi R, wing-body junctions. ​AIAA J. 3​ 2:6–15 Ol

Tropea C. 1991. Flow around
surface-mounted, three- dimensional men ̈MS, Simpson RL. 1995. An experi-
obstacles. In ​Proc. SYMP。 Turbul. Shear ental study of three-dimensional pressure-
Flow, 8th, Munich, Ger​. pp. 127–39. ven turbulent boundary layer. ​J. Fluid
Berlin/Springer-Verlag ech. 2​ 90:225–62
November 13, 2000 18:32 Annual Reviews AR117-14
442 ​SIMPSON
symmetry plane of a turbulent junction vortex.
J. Fluids Eng. 1​ 12:16–22 Praisner TJ, Seal
CV, Takmaz L, Smith CR. 1997.
Ol ̧cmen ̈MS, Simpson RL. 1996a.
Spatial-temporal turbulent flowfield
Experimen- tal
​ transport-rate budgets in heat transfer behavior in end-wall junc-
complex three- dimensional turbulent flows at s. ​詮釋。 J. Heat Fluid Flow. 1​ 8:42– 151
a wing/body junction​. AIAA-96-2035. e MC, Devenport WJ, Simpson RL.
Presented at Am. Inst. Aeronaut. Astronaut. 2.​An Experimental Study of the
ationship Be- tween Velocity and Pressure
​ ̧cmen
Fluid Dyn. Conf., 27th, New Orleans Ol
ctuations in a Wing-Body Junction.​
&SU Report VPI- AOE-188, V ​ a. Polytech.
̈MS, Simpson RL. 1996b. Theoretical ​and
. State Univ., Blacksburg, Va. Rizzi A,
experimental pressure-strain compari- son in
J. 1998. Toward establishing cred- ibility
a pressure-driven three-dimensional turbulent
omputational fluid dynamics sim- ulations.
boundary layer. ​AIAA-96-2141. Presented at
A J. ​36:668–75 Roach PE, Turner JT.
Am. Inst. Aeronaut. Astronaut. Theor. Fluid
5. Secondary loss gen- eration by gas
​ ̧cmen ̈MS,
Mech. Meet., 1st, New Orleans Ol ine support struts. 詮​ 釋。 J. Heat Fluid
w. 6​ :79–88 Rodi W. 1997. Comparison of
Simpson RL. 1997a.​Experimental ​evaluation S and RANS calculations of the flow
of turbulence diffusion models in complex 3-D und bluff bod- ies. ​J. Wind Eng. Ind.
flow near a wing/body junction​. odyn. ​69–71:55– 75 Rood EP. 1984a.
AIAA-97-0650. Presented at Am. Inst. Aero- erimental investigation of the turbulent
naut. Astronaut. Aerosp.科學。 Meet., 35th, ge scale temporal flow in the wing-body
Reno ​Ol ̧cmen ̈MS, Simpson RL. 1997b. ction. ​PhD diss. Cathol. Univ. Am.,
shington, DC Rood EP. 1984b. ​The
Some fea- tures
​ of a turbulent wing-body arate spatial extents of the trailing
junction vortical flow.​ AIAA-97-0651. seshoe root vortex legs from a wing and
Presented at Am. Inst. Aeronaut. Astronaut, e junction. AIAA-84-1526. ​Presented at
. Inst. Aeronaut. Astronaut. Fluid Dyn.
Aerosp.科學。 Meet., 35th, Reno ​Ozcan ̈O,
f., 14th Rood EP. 1984c. The governing
uence of the nose radius on the unsteady
Ol ̧cmen ̈MS. 1988. Measurements of
cts of large scale flow structure in the
turbulent flow behind a wing-body junction.
ulent wing and plate junction flow. In
AIAA J. ​26:494–96 Parneix S, Durbin PA,
ME Forum on Un- steady Flow​, FGD, ed.
Behnia M. 1998. Com- putation of 3d
Rothe, 15:7–9. New York: Am. SOC。機
turbulent boundary layers us- ing the v2-f
Eng., Fluids Eng. Div. Rood EP,
model. ​Flow Turbul. Combust. ​60:19–46
hony DG. 1985. Tail profile ef- fects on
Phillips DB, Cimbala JM, Treaster AL. 1992.
eady large scale flow structure in the wing
Suppression of the wing-body junction vor-
plate junction. In ​ASME Fo- rum on
tex by body surface suction. ​J. Aircr. ​29:118–
teady Flow​, FGD, ed. PH Rothe,
122 Pierce FJ, Shin J. 1990. An experimental
30–32. New York: Am. SOC。機甲。
inves- tigation of effects of leading-edge
., Fluids Eng. Div. Rood EP, Keller JE.
fillets on a turbulent junction vortex. In ​Proc.
4. Evidence of large scale time dependent
詮釋。 SYMP。 Hydrodyn. Perform. Enhanc.
w in the wing- wall interaction wake. In
Marine Appl., 2nd, Univ. RI, Kingston,​ pp.
teady Turbu- lent Boundary Layers and
61–71 Pierce FJ, Shin J. 1992. The
ction​, FGD, 12:39–44. New York: Am.
development of a turbulent vortex system. ​J.
C。機甲。 Eng., Fluids Eng. Div.
Fluids Eng. 1​ 14:559–65 Pierce FJ, Tree IK.
1990. The mean flow struc- ture on the
November 13, 2000 18:32 Annual Reviews AR117-14

JUNCTION FLOWS 4 ​ 43
derstanding of basic aspects of secondary
ws in turbine blade passages. ​J. Turbo-
Seal CV, Smith CR. 1999a. Visualization of a ach. 1​ 07:248–57 Simpson RL. 1996.
mechanism for three-dimensional interac- tion pects of turbulent bound- ary layer
and near-wall eruption. ​J. Fluid Mech. paration. ​PROG。 Aerosp.科學。
394:193–203 Seal CV, Smith CR. 1999b. The :457–521 Sung CH, Griffin MJ. 1991.
control of turbu- lent end-wall boundary provements in incompressible turbulent
layers using surface suction. ​進出口。 Fluids rseshoe vortex juncture calculations.​
27:484–96 Seal CV, Smith CR, Akin O, AA-91-0022. P ​ re- sented at Am. Inst.
Rockwell D. 1995. Quantitative ronaut. Astronaut. Aerosp.科學。 Meet.,
characteristics of laminar, un- steady necklace th, Reno Thomas ASW. 1987. The unsteady
vortex system at a rectangu- lar block-flat aracter- istics of laminar juncture flow. ​物
plate juncture. ​J. Fluid Mech. 2​ 86:117–35 學。 Fluids 3​ 0:283–85 Visbal MR. 1991.
Seal CV, Smith CR, Walker JDA. 1997. Dy- ructure of laminar juncture
namics of the vorticity distribution in endwall
flows. ​AIAA J. 2​ 9:1273–82 Walter
junctions. ​AIAA J. 3​ 5:1041–47 Shinpaugh
, Patel VC. 1996. Measurements in
KA. 1994. ​Measurements in the bi- modal
ee-dimensional wake of a surface- mounted
region of a wing-body junction flow with a
nglike symmetrical ellipsoid. 進 ​ 出口。
rapidly-scanning two-velocity-comp- onent
erm. Fluid Sci. ​13:266–91 Wood DH,
laser-Doppler anemometer​. PhD diss. Va.
estphal RV. 1992. Measurements of the
Polytech. Inst. State Univ., Blacksburg
w around a lifting-wing/body junc- tion.
Shizawa T, Honami S, Yamamoto M. 1996.
AA J. ​30:6–12 Wroblewski DE, Eibeck PA.
Experimental study of horseshoe vortex at
92 Turbulent heat transport in a boundary
wing/body junction with attack angle by triple
yer behind a junction of a streamlined
hot-wire. AIAA-96-0323.​ Presented at Am.
linder and a wall. ​TASME J. Heat Transf.
Inst. Aeronaut. Astronaut. Aerosp.科學。
4:840–49
Meet., 34th, Reno Sieverding CH. 1985.
Recent progress in the
November 30, 2000 17:54 Annual Reviews AR117-COLOR
(​a)​

Figure 10 ​Surface oil-flow visualizations of unseparated flow on the strake used by


Deven- port et al (1992) in front of a Rood wing for zero angle of attack: (​a)​ front
perspective view showing attachment line along the wing plane of symmetry and smooth
skin-friction lines around side of the strake and wing, (​b)​ (next page) side view showing
laminar separation and turbulent reattachment on wing near maximum thickness.
November 30, 2000 17:54 Annual Reviews AR117-COLOR
(​b)​

Figure 10 ​(Continued.)
November 13, 2000 19:27 Annual Reviews AR117-FM

Annual Review of Fluid Mechanics Volume


33, 2001
C​ONTENT
S

James Lighthill and His Contributions to Fluid Mechanics, ​TJ Pedley 1​ ​Steady
Streaming, ​N Riley 4​ 3 ​On the Fluid Mechanics of Fires, ​Sheldon R Tieszen ​67
Experiments on Thermocapillary Instabilities, ​Michael F Schatz and
G Paul Neitzel ​93 ​Robert Legendre and Henri Werl ́e: Toward the Elucidation of
Three-Dimensional Separation, J​ ean MD ́ elery 129 ​Surface Pressure Measurements
Using Luminescent Coatings,
James H Bell, Edward T Schairer, Lawrence A Hand, and Rabindra D Mehta 1​ 55
Rossby Wave Hydraulics, ​ER Johnson and SR Clarke 2​ 07 ​Spin-Up of
Homogeneous and Stratified Fluids, ​PW Duck and MR Foster ​231 ​Extrusion
​ orton M Denn 2​ 65 ​Turbulent Relative Dispersion,
Instabilities and Wall Slip, M
Brian Sawford ​289 ​Early Work on Fluid Mechanics in the IC Engine, J​ ohn L
Lumley ​319 ​Mechanics of Coastal Forms, P ​ aolo Blondeaux 3​ 39 ​Aerodynamics of
High-Speed Trains, J​ oseph A Schetz ​371 ​Junction Flows, R​ oger L Simpson ​415
Modeling of Fluid-Structure Interaction, ​Earl H Dowell and
Kenneth C Hall ​445 ​Compression System Stability and Active Control, ​JD Paduano,
EM Greitzer, and AH Epstein 4​ 91 ​Spilling Breakers, ​JH Duncan ​519 ​Shelterbelts and
Windbreaks: Mathematical Modeling and Computer
Simulations of Turbulent Flows, ​Hao Wang, Eugene S Takle, and Jinmei Shen 5​ 49
Drag Due to Lift: Concepts for Prediction and Reduction, I​ lan Kroo 5​ 87 ​Inertial
Effects in Suspension and Porous-Media Flows, ​Donald L Koch
and Reghan J Hill ​619

v
i
November 13, 2000 19:27 Annual Reviews AR117-FM

CONTENTS ​vii

I​NDEXE
S
Subject Index 649 Cumulative Index of Contributing Authors, Volumes 1–33 675
Cumulative Index of Chapter Titles, Volumes 1–33 682

You might also like