这是用户在 2024-10-12 17:34 为 https://doc2x.noedgeai.com/pdfs/01925bf8-2657-7d34-9285-42302e2452c6?whetherToReadStore=1 保存的双语快照页面,由 沉浸式翻译 提供双语支持。了解如何保存?

Boomerang Flight Dynamics
Boomerang 飞行动力学

Conference Paper June 2012
会议论文 2012 年 6 月
DOI: 10.2514/6.20122650 CITATIONS READS
DOI: 10.2514/6.20122650 引文读取
10 9,080 1 author: J. Vassberg The Boeing Company Retired 96 PUBLICATIONS 4,124 CITATIONS SEE PROFILE
10 9,080 1 作者: J. Vassberg 波音公司已退休 96 出版物 4,124 次引用 查看个人简介
AIAA 2012-2650 美国建筑师协会 2012-2650

Boomerang Flight Dynamics
Boomerang 飞行动力学

John C. Vassberg* 约翰·瓦斯伯格*
Long Beach, California 90803
加利福尼亚州长滩 90803
Aerodynamics Lecture 空气动力学讲座
30th  AIAA Applied Aerodynamics Conference
30th  AIAA 应用空气动力学会议
New Orleans, Louisiana 路易斯安那州新奥尔良
28 June, 2012 28 六月, 2012
Abstract 抽象
Aerodynamic characteristics and flight dynamics of boomerangs are investigated. A basic aerodynamic model, developed in the 1960's, is expanded upon using Blade Element Theory. The new aerodynamic model is coupled with a gyroscope model for rudimentary analyses. Some significant findings are made regarding the radius of a boomerang's circular flight path, the required inclination angle of its axis-of-rotation, its trim state, as well as its dynamic stability. These discoveries provide a basic understanding of how the interplay between aerodynamic forces and moments, and gyroscopic precession combine to return the boomerang to its rightful owner by way of a circular flight path.
研究了回旋镖的空气动力学特性和飞行动力学。1960 年代开发的基本空气动力学模型在使用叶片元件理论的基础上进行了扩展。新的空气动力学模型与陀螺仪模型相结合,用于基本分析。关于回旋镖圆形飞行路径的半径、旋转轴所需的倾斜角、配平状态以及动态稳定性,取得了一些重要的发现。这些发现为空气动力和力矩以及陀螺进动之间的相互作用如何结合,通过圆形飞行路径将回旋镖送回其合法所有者提供了基本理解。
A traditional V-shaped boomerang design is developed as a case study for further detailed analyses. Unsteady Reynolds-averaged Navier-Stokes solutions provide accurate aerodynamic characteristics of the subject boomerang. The high-fidelity aerodynamic model is coupled with the equations of motion to provide accurate six-degree-of-freedom simulations of boomerang flight dynamics. Boomerang orientation during its flight trajectory is described by the classical Euler angles.
传统的 V 形回旋镖设计被开发为案例研究,用于进一步的详细分析。非稳态 Reynolds 平均 Navier-Stokes 解可提供对象回旋镖的精确空气动力学特性。高保真空气动力学模型与运动方程相结合,以提供回旋镖飞行动力学的精确六自由度仿真。回旋镖在其飞行轨迹中的方向由经典的欧拉角描述。
Copyright (C) 2012 by John C. Vassberg.
版权所有 (C) 2012 John C. Vassberg。
Published by the American Institute of Aeronautics and Astronautics with permission.
经许可由美国航空航天学会出版。

*Boeing Technical Fellow, Fellow AIAA, Boomerang Novice
*波音技术研究员、AIAA 研究员、Boomerang 新手

a,b,c,d Arbitrary Coefficients
a,b,c,d 任意系数
b Coordinate Vectors of Body Frame
b Body Frame 的坐标向量
C Blade Reference Chord C Blade 参考和弦
2D Lift Coefficient =liftqC
2D 提升系数 =liftqC
Cl0 2D Lift Coefficient at α=0
Cl0 2D 提升系数 α=0
Clα 2D Lift-Curve Slope
Clα 2D 提升曲线坡度
3D Lift Coefficient =LiftqS
3D 提升系数 =LiftqS
3D Rolling-Moment Coefficient
3D 滚动力矩系数
CP Pressure Coefficient =PPq
CP 压力系数 =PPq
g Gravitational Acceleration 32ftsec2
g 引力加速度 32ftsec2
i Coordinate Vectors of Inertial Frame
i 惯性系的坐标矢量
I Moment of Inertia
I 转动惯量
L Lift  L 电梯
L0 Basic Lift  L0 基本提升
Lα Lift due to Angle-of-Attack η Fraction of Blade Radius =rR .
Lα 由于叶片半径 =rR 的攻角 η 分数 .
LE Airfoil Leading Edge
LE 翼型前沿
m Mass  m 质量
M Rolling-Moment
M Rolling-Moment (滚动力矩)
M0 Basic Rolling-Moment
M0 基本滚动力矩
Mα Rolling-Moment due to Angle-of-Attack χ ND Rotational Velocity =ωRV
Mα 由攻角 χ ND 旋转速度 =ωRV 引起的滚动力矩
n Node Vector of Body Frame
n Body Frame 的节点向量
q Dynamic Pressure =12ρV2
q 动态压力 =12ρV2
r Radial Coordinate
r 径向坐标
R Reference Radius of Blade
R 叶片的参考半径
Rp Radius of Circular Flight Path
Rp 圆形飞行路径半径
s Vector in Rotation Plane n
s 旋转平面 n 中的向量
S Reference Area =RC
S 参考区域 =RC
Transformation Matrix 变换矩阵
TE Airfoil Trailing Edge
TE 翼型后缘
V Velocity  V 速度
x Streamwise Cartesian Coordinate
x Streamwise Cartesian Coordinate (流式笛卡尔坐标)
Lateral Cartesian Coordinate
横向笛卡尔坐标
z Vertical Cartesian Coordinate
z 垂直笛卡尔坐标
ND Non-Dimensionalized
ND 无量纲
2D Two Dimensional 2D 二维
2 Three Dimensional 2 三维
α Angle-of-Attack
α 攻角
Euler Angle: Orbital-Plane Inclination
Euler Angle: 轨道-平面倾斜
3.141592654
ρAir Density=1.3755×106 slugs/in3
ϕ Euler Angle: Right Ascension
ϕ Euler Angle: Right Ascension (欧拉角):赤经
χ^ Lessor of χ or 1;=min(χ,1)
χ^ χ1;=min(χ,1) ' 的出租人
ψ Euler Angle: Argument of Perigee
ψ Euler Angle: Argument of Perigee (欧拉角): Argument of Perigee (近地点参数)
ψc Cut-Off Angle =cos1(χ)
ψc 截止角 =cos1(χ)
ω Rotational Velocity
ω 旋转速度

I. Introduction I. 引言

Boomerangs are deceptively simple-looking devices. But make no mistake, boomerang aerodynamics are rather complex. Not only is the flow unsteady and very three dimensional; it also reverses itself across the wing such that the leading and trailing edges swap roles twice every cycle. Furthermore, the Reynolds number is very low, O(50,000) .
回旋镖是看似简单的设备。但不要误会,回旋镖空气动力学相当复杂。不仅流动不稳定且非常三维;它还会在伞体上反转自己,这样前缘和后缘在每个周期中交换角色两次。此外,雷诺数非常低, O(50,000)
In spite of these challenges and by trial and error, Australian Aborigines managed to develop boomerang technology some 10,000 years ago; see Lorenz, 1Mauro2 and Hanson. 3 In fact,several civilizations developed straight-flying "boomerangs" for hunting and war many millenia ago. A gold-tip throwing-stick was found in King Tutankhamun's tomb. A boomerang was found in the sand dunes of The Netherlands that dates back 2,400 years. An exquisite mammoth-tusk throwing stick found in Poland is about 23,000 years old. This unique artifact has a mass of about 800 grams,a span of almost 70cm ,a chord of about 6cm and is 1.5cm thick. The ivory of this relic has a density of about 1,900 kg-per-cubic-meter. These ancient throwing sticks have aerodynamic and gyroscopic properties that stabilize their flights and extend their range. However, it was only the Aborigines who created the returning boomerang which they used primarily for recreation and sport, just as we do today. Indirectly, they also used the returning boomerang to aid in hunting fowl. Here, they would throw a boomerang over a flock of ducks floating on a pond. The waterfowl would mistake the boomerang for a hawk and would take flight, yet remain low to avoid the "predator". The low-flying game would then be well within range of hunters waiting nearby. Another unusual, but very utilitarian, use of the boomerang is when the Aborigines rubbed the leading edge against another piece of wood in a sawing motion to ignite a fire.
尽管面临这些挑战,并通过反复试验,澳大利亚原住民还是在大约 10,000 年前成功开发了回旋镖技术;见 Lorenz 和 1Mauro2 Hanson。 3 事实上,许多文明在几千年前就发展出了用于狩猎和战争的直飞“回旋镖”。在图坦卡蒙国王的坟墓中发现了一根金尖投掷棍。在荷兰的沙丘中发现了一个回旋镖,其历史可以追溯到 2,400 年前。在波兰发现的一根精美的猛犸象牙投掷棒大约有 23,000 年的历史。这种独特的文物质量约为 800 克,跨度几乎 70cm 为,弦约为 6cm ,而且很 1.5cm 厚。这件文物的象牙密度约为每立方米 1,900 公斤。这些古老的投掷棒具有空气动力学和陀螺仪特性,可以稳定它们的飞行并扩大它们的射程。然而,只有原住民创造了回归的回旋镖,他们主要用于娱乐和运动,就像我们今天所做的那样。间接地,他们还使用返回的回旋镖来帮助狩猎家禽。在这里,他们会把回旋镖扔到漂浮在池塘上的一群鸭子上。水禽会把回旋镖误认为是鹰,然后飞起来,但为了躲避“捕食者”而保持低位。然后,低空飞行的猎物将完全在附近等待的猎人的射程内。回旋镖的另一个不寻常但非常实用的用途是,原住民以锯切动作将前缘与另一块木头摩擦以点燃火。
Figure 1 provides a small sampling of early 20th  -century Aboriginal boomerangs. The boomerangs in this figure are probably straight-flying configurations. However, it is hard to know without throwing them - it is possible they are of the returning type. Figure 2 illustrates a wide variety of modern-day boomerangs; all of these examples are returning designs. Of particular note are the Tri-Fly and LED boomerangs, as these are utilized by the author for flight tests and further analysis in this body of work. The middle V-type boomerang of the modern Aboriginal group is similar to a design developed herein for the purpose of high-fidelity analysis using computational fluid dynamics.
图 1 提供了本世纪早期 20th  土著回旋镖的一小部分样本。这个图中的回旋镖可能是直线飞行的配置。但是,如果不抛出它们就很难知道 - 它们可能是 return 类型。图 2 说明了各种各样的现代回旋镖;所有这些示例都是返回的设计。特别值得注意的是 Tri-Fly 和 LED 回旋镖,因为作者将这些用于飞行测试和本工作中的进一步分析。现代原住民群体的中间 V 型回旋镖类似于本文开发的设计用于使用计算流体动力学进行高保真分析的设计。
A number of boomerang world records as of June 2010 are included; see Wikipedia. 4 The most number of catches made within 5 minutes is 81 by Manuel Schutz, Switzerland, 2005. The quickest five-throw-catch sequence is 14.6 seconds by Adam Ruhf, USA, 1996. The most number of consecutive catches is 2,251 by Haruki Taketomi, Japan, 2009. The maximum time aloft in the unlimited class is 380.6 seconds by Billy Brazelton, USA, 2010. The farthest distance out for a returning flight is 238 meters by Manuel Schutz, Switzerland, 1999. The longest hand-held throw of any object was set in 2005 by David Schummy when he threw a boomerang 427.2 meters (over a quarter-mile). All of these are incredible statistics; little did the Aborigines know just how far their returning sticks could go.
包括截至 2010 年 6 月的许多 boomerang 世界纪录;参见 维基百科. 4 5 分钟内捕获的最多渔获数是瑞士 Manuel Schutz 于 2005 年完成的 81 次。最快的五次投球序列是 14.6 秒,由美国 Adam Ruhf 于 1996 年创作。连续捕捞次数最多的是 Haruki Taketomi,日本,2009 年,2009 年。无限级的最高高空飞行时间为 380.6 秒,由美国 Billy Brazelton 于 2010 年提供。返程航班最远的距离是 238 米,由 Manuel Schutz, Switzerland, 1999 提供。大卫·舒米 (David Schummy) 在 2005 年创造了任何物体中最长的手持投掷,当时他投掷了 427.2 米(超过四分之一英里)的回旋镖。所有这些都是令人难以置信的统计数据;原住民几乎不知道他们返回的棍子能走多远。
This paper is organized in the following manner. Section II introduces Euler's classical orientation angles. Section III derives the equations of angular motion associated with boomerang dynamics. Section IV provides a discussion regarding the kinematics of an over-hand boomerang throw. Section V revisits work done in the 1960's regarding the aerodynamics of a boomerang. Section VI develops a new aerodynamic model for boomerangs based on blade element theory. Section VII develops general formulas for the radius of a boomerang's circular flight path, as well as the rotation-plane inclination angle. Section VIII develops a general condition for static trim and proves that dynamic stability is an intrinsic property of the boomerang. Section IX develops a traditional V-type boomerang design for further analysis. Section X provides the two-dimensional steady-state CFD results for the boomerang airfoil. Section XI discusses the three-dimensional unsteady CFD results for the boomerang subjected to a fixed cyclic motion. Section XII provides a full 6- DOF numerical simulation of a boomerang in flight. Section XIII is a summary of conclusions. Throughout the text, references are given as positive superscripts while footnotes are indicated by negative superscripts. Tables of data are embedded within the text while all figures are appended to the end of the paper.
本文的组织方式如下。第 II 节介绍了欧拉的经典取向角。第 III 节推导出了与回旋镖动力学相关的角运动方程。第四部分讨论了上手回旋镖投掷的运动学。第五部分回顾了 1960 年代关于回旋镖空气动力学所做的工作。第六部分基于叶片元件理论开发了一种新的回旋镖空气动力学模型。第七节开发了回旋镖圆形飞行路径半径以及旋转平面倾斜角的一般公式。第八节制定了静态配平的一般条件,并证明了动态稳定性是回旋镖的内在特性。第九部分开发了传统的 V 型回旋镖设计以供进一步分析。第 X 部分提供了 Boomerang 翼型的二维稳态 CFD 结果。第十一节讨论了在固定循环运动作用下回旋镖的三维非稳态 CFD 结果。第 XII 部分提供了飞行中回旋镖的完整 6 自由度数值模拟。第十三节是结论的总结。在整个文本中,参考文献以正上标形式给出,而脚注以负上标表示。数据表嵌入在文本中,而所有图表都附加到论文的末尾。

II. Euler's Classical Orientation Angles
II. 欧拉经典取向角

The classical Euler angles ϕ,θ and ψ are shown in Figure 3. Also depicted in Figure 3 are the inertial and body-fixed coordinate systems,labeled as (i1,i2,i3) and (b1,b2,b3) ,respectively. In celestial mechanics, the angles (ϕ,θ,ψ) are used to specify the orientation of an orbit. They are known as the right ascension of the ascending node, the inclination of the orbital plane, and the argument of perigee, respectively. For the boomerang problem,as set-up herein,the axis of rotation is about the b3 vector, θ describes the bank angle that the plane of rotation makes with the horizon,while the precession rate of b3 about i3 is given by ϕ .
经典的欧 ϕ,θ 拉角 和 ψ 如图 3 所示。图 3 中还描绘了惯性坐标系和物体固定坐标系,分别标记为 (i1,i2,i3)(b1,b2,b3) 。在天体力学中,角度 (ϕ,θ,ψ) 用于指定轨道的方向。它们分别被称为升交点的赤经、轨道平面的倾角和近地点的参数。对于回旋镖问题,如本文所述,旋转轴是关于矢量的 b3θ 描述了旋转平面与地平线形成的倾斜角,而大约 i3b3 进动率由下 ϕ 式给出。
Transforming from the inertial frame to the body frame can be accomplished with three rotations in series. Pay attention to the order of rotations as this is crucial. First rotate by ϕ about the i3 -axis,then by θ about the node vector n and finally with a rotation of ψ about the b3 -axis. The complete transformation from the inertial frame to the body frame using this three-rotation sequence can be described as
从惯性坐标系到体坐标系的转换可以通过串联三次旋转来完成。注意轮换的顺序,因为这很关键。首先 ϕi3 -轴旋转,然后绕 θ 节点向量 n 旋转,最后绕 b3 -轴旋转 ψ 。使用这个 3 次旋转序列从惯性系到体系的完整变换可以描述为
(1)Tib=R(ψ;b3)R(θ;n)R(ϕ;i3).
Here, R(α;v) represents an elementary rotation of angle α about vector v . This sequence of rotations is graphically illustrated by Figure 4. Completing the multiplication symbolized by Eqn (1), one obtains the transformation matrix
其中, R(α;v) 表示绕向量 v 的角度 α 的基本旋转。图 4 以图形方式说明了这种旋转顺序。完成由 Eqn (1) 符号化的乘法,得到变换矩阵
(2)Tib=[cϕcψsϕcθsψsϕcψ+cϕcθsψsθsψcϕsψsϕcθcψsϕsψ+cϕcθcψsθcψsϕsθcϕsθcθ].
Here, c indicates cosine and s indicates sine. Since Eqn (2) is a pure rotation,its inverse is just its transpose. Hence, to transform from the body frame to the inertial frame, one uses
这里, c 表示余弦和 s 正弦。由于 Eqn (2) 是一个纯旋转,它的逆只是它的转置。因此,要从体系变换到惯性系,可以使用
(3)Tbi=[cϕcψsϕcθsψcϕsψsϕcθcψsϕsθsϕcψ+cϕcθsψsϕsψ+cϕcθcψcϕsθsθsψsθcψcθ].
The next section discusses Euler's equations of angular motion.
下一节讨论 Euler 的角运动方程。

III. Euler's Equations of Angular Motion
III. 欧拉角运动方程

In this section,we begin with the equations of angular motion as defined by Wiese 15 and then adapt them for boomerang dynamics. We start the discussion with a very generalized form of Euler's equations of angular motion, then progress through a series of simplifications which retain only essential elements of the equations needed to describe a level circular flight path typical of a returning boomerang. Further simplifications regarding mass properties are introduced which appropriately continue to capture the low-frequency content of dynamic motion for boomerang flight. While this may filter or blur some of the high-frequency motion within a spin cycle, it greatly reduces the complexity of the ensuing derivations presented throughout this body of work.
在本节中,我们从 Wiese 15 定义的角运动方程开始,然后根据回旋镖动力学对它们进行调整。我们从欧拉角运动方程的非常概括的形式开始讨论,然后通过一系列简化进行讨论,这些简化只保留了描述返回回旋镖典型的水平圆形飞行路径所需的方程的基本元素。引入了有关质量属性的进一步简化,这些属性适当地继续捕获回旋镖飞行的动态运动的低频内容。虽然这可能会过滤或模糊自旋周期中的一些高频运动,但它大大降低了整个工作中呈现的后续推导的复杂性。
For the boomerang problem, the principal-axis frame naturally aligns with an appropriate body-fixed coordinate system. This is fortuitous as the equations of angular motion are simplest when expressed in the principal-axis body frame because the off-diagonal terms of the moment-of-inertia matrix vanish. In the present work, the origin of the principal-axis frame coincides with the center-of-mass of the boomerang. When the Euler angles are zeroed, the body and inertial frames are aligned; see initial state of Figure 4. In this position,the Z -axis is the axis-of-rotation, Y is in the spanwise direction and X is streamwise. Here,the boomerang reference plane coincides with the XY plane where Z=0 ,the boomerang’s centerline coincides with the X -axis,the right wing resides in the half-space where Y0 and the boomerang upper-surface resides in the Z0 half-space. For this system,Euler’s equations of angular motion are given as:
对于回旋镖问题,主轴坐标系自然会与适当的身体固定坐标系对齐。这是偶然的,因为在主轴体坐标系中表示时,角运动方程是最简单的,因为惯性矩矩阵的非对角线项消失了。在本工作中,主轴框架的原点与回旋镖的质心重合。当 Euler 角度归零时,物体和惯性坐标系对齐;参见图 4 的初始状态。在这个位置, Z -轴是旋转轴, Y 在翼展方向上, X 是流向的。在这里,回旋镖参考平面与 XY 平面重合,其中 Z=0 回旋镖的中心线与 X 轴重合,右翼位于半空间中 Y0 ,回旋镖上表面位于 Z0 半空间中。对于该系统,欧拉角运动方程为:
M1=I11ω˙1+(I33I22)ω2ω3
(4)M2=I22ω˙2+(I11I33)ω1ω3
M3=I33ω˙3+(I22I11)ω1ω2.
Here,the ω -vector describes the angular rates in the principal body-axis frame. These equations provide the time-rate-of-change of the ω -vector,yet only constitute one half of the rotational equations of motion. Another set of equations is needed which link the angular rates of the body frame to the orientation angles of the body in the inertial frame. The remaining set of equations is given as:
在这里, ω -vector 描述了主体轴坐标系中的角速率。这些方程提供了 ω -向量的时间变化率,但只构成了旋转运动方程的一半。需要另一组方程,将体系的角速率与惯性系中体的方向角联系起来。其余的方程组如下:
ω1=ϕ˙sinθsinψ+θ˙cosψ
(5)ω2=ϕ˙sinθcosψθ˙sinψ
ω3=ψ˙+ϕ˙cosθ
These equations transform the ω -vector given by Eqn (4) to provide the time-rate-of-change of the Euler angles (ϕ˙,θ˙,ψ˙) . In practice,these angular velocities are used to update the orientation angles (ϕ,θ,ψ) for the body as a function of time.
这些方程变换 ω 方程 (4) 给出的向量,以提供欧 (ϕ˙,θ˙,ψ˙) 拉角 的时间变化率。在实践中,这些角速度用于更新物体的取向角 (ϕ,θ,ψ) 作为时间的函数。

Axi-Symmetric Body 轴对称体

If the body is axi-symmetric about one of the principal axes, then further simplifications can be made to the equations of angular motion. So how can a boomerang be axi-symmetric? This question is best answered if one views the boomerang when it is spinning fast, rather than when statically displayed. Just as this motion blurs the visual of the boomerang to appear as a spinning disk, the moments of inertia can be time-averaged in the XY plane. Although this simplification alters high-frequency content,it retains sufficient mass properties to accurately describe low-frequency boomerang dynamics.
如果物体绕其中一个主轴轴对称,则可以进一步简化角运动方程。那么回旋镖怎么可能是轴对称的呢?如果人们在回旋镖快速旋转时查看它,而不是在静态显示时查看它,则可以最好地回答这个问题。正如这种运动使回旋镖的视觉效果变得模糊,看起来像一个旋转的圆盘一样,惯性矩可以在 XY 平面上进行时间平均。尽管这种简化改变了高频内容,但它保留了足够的质量特性来准确描述低频回旋镖动力学。
For the purpose of this discussion,let’s assume that the body is symmetric about the b3 -axis and a rolling-moment exists purely about the n -axis. Let’s designate this torque about the nodal line as Mn . See Figure 3 to get familiar with this situation. Note that the n -axis is perpendicular to the b3 -axis and thus can be described as a linear combination of b1 and b2 .
为了进行本讨论,我们假设物体绕 b3 -轴对称,并且滚动力矩纯粹围绕 n -轴存在。让我们将关于节点线的扭矩指定为 Mn 。请参阅 图 3 以熟悉这种情况。请注意, n -axis 垂直于 b3 -axis,因此可以描述为 b1b2 的线性组合。
(6)n=cosψb1sinψb2
Also define an s-axis which is perpendicular to both the n -axis and the b3 -axis.
还要定义一个垂直于 n -轴和 -轴的 b3 s 轴。
(7)s=sinψb1+cosψb2
Note that the s-axis is not shown in Figure 3.
请注意,图 3 中未显示 s 轴。
Because of symmetry,let’s define a moment of inertia (Irr) that is constant for any radial in the (n,s) plane. Hence,
由于对称性,让我们定义一个惯性矩 (Irr) ,该矩对于平面中的任何径向都是恒定的 (n,s) 。因此
(8)IrrI11=I22
Combining Eqns (4-8) gives:
组合方程 (4-8) 可得到:
Mn=Irrθ¨+(I33Irr)ϕ˙2sinθcosθ+I33ψ˙ϕ˙sinθ
(9)Ms=Irrϕ¨sinθ+2Irrθ˙ϕ˙cosθI33θ˙(ψ˙+ϕ˙cosθ)
M3=I33ddt(ψ˙+ϕ˙cosθ)

Axi-Symmetric Disk Axi-对称盘

If an axi-symmetric body is also planar and homogeneous (i.e., a disk), then the principal moments of inertia are simply:
如果轴对称体也是平面和均匀的(即圆盘),那么主惯性矩就是:
(10)I33=12mR2=2Irr
where m is the mass of the disk and R its radius. Now,the equations of angular motion reduced to:
其中 m 是圆盘的质量及其 R 半径。现在,角运动方程简化为:
Mn=14mR2[θ¨+ϕ˙2sinθcosθ+2ψ˙ϕ˙sinθ]
(11)Ms=14mR2[ϕ¨sinθ2θ˙ψ˙]
M3=12mR2ddt[ψ˙+ϕ˙cosθ].
For the moment,let’s assume that the angular motion is such that θ is constant,then θ˙=θ¨=0 . Now, if the rolling-moment is purely about the n -axis,then Ms=0 . Let’s also assume that there is no moment being applied about the spin axis,i.e., M3=0 . Under these conditions,the equations of angular motion reduce further to:
目前,我们假设角运动是 θ 恒定的,那么 θ˙=θ¨=0 。现在,如果滚动力矩纯粹是关于 -轴 n 的,那么 Ms=0 。我们还假设没有关于自旋轴的弯矩,即 M3=0 。在这些条件下,角运动方程进一步简化为:
Mn=14mR2[ϕ˙2sinθcosθ+2ψ˙ϕ˙sinθ]
(12)0=ϕ¨
0=ψ¨
One may recognize these equations as those that describe the motion of an inclined toy gyroscope, with its tip resting on a table. Here,the high spin rate of the gyroscope disk remains at a constant ψ ,with the gyroscope slowly rotating around its pivot-point-vertical-axis with a constant precession rate of ϕ ,all the while,maintaining a constant inclination angle θ . This precessional motion is driven by the moment Mn created by the gyroscope's weight, multiplied by the lateral off-set distance between the pivot point and the gyroscope’s center-of-mass. For boomerang dynamics,the moment (Mn) is supplied aerodynamically.
人们可能会认出这些方程式是描述倾斜玩具陀螺仪的运动的方程式,陀螺仪的尖端放在桌子上。在这里,陀螺仪盘的高自旋速率保持恒定 ψ ,陀螺仪以恒定的进动率绕其枢轴-垂直-轴缓慢旋转 ϕ ,同时保持恒定的倾角 θ 。这种进动运动由陀螺仪重量产生的力矩 Mn 乘以枢轴点与陀螺仪质心之间的横向偏移距离驱动。对于回旋镖动力学,力矩 (Mn) 由空气动力学提供。
If Mn,θ and ψ are known,then ϕ can be determined using the quadratic formula applied to Eqn (12).
如果 Mn,θψ 已知,则可以使用 ϕ 应用于方程 (12) 的二次公式来确定。
(13)ϕ˙=1cosθ[ψ˙±ψ˙2+4cosθMnsinθmR2].
From experience, we know that the direction of the precession is the same as that of the disk spin. Hence, the positive branch of Eqn (13) is physically correct. Further,if ϕ˙ψ˙ ,then solution of Eqn (13) can be
根据经验,我们知道岁差的方向与圆盘自旋的方向相同。因此,方程 (13) 的正分支在物理上是正确的。此外,如果 ,则 ϕ˙ψ˙ 方程 (13) 的解可以是
approximated by: 近似值:
(14)ϕ˙2MnsinθmR2ψ˙=MnsinθI33ψ˙
where the second form is included for the more general case of constant θ motion. Eqn (14) can be directly derived from Eqn (12) by dropping the ϕ˙2 term; the intermediate step of Eqn (13) is included as a formality.
其中第二种形式包括在恒定 θ 运动的更一般情况下。Eqn (14) 可以通过删除 ϕ˙2 项直接从 Eqn (12) 派生出来;Eqn (13) 的中间步骤作为形式包括在内。
Eqns (12-14) will be used to develop a simplified analysis of a boomerang's flight dynamics to determine the radius of its circular flight path as well as other pertinent relationships.
方程 (12-14) 将用于开发对回旋镖飞行动力学的简化分析,以确定其圆形飞行路径的半径以及其他相关关系。
The next section reviews the kinematics of an over-hand boomerang throw.
下一节回顾了上手回旋镖投掷的运动学。

IV. Kinematics of a Boomerang Throw
IV. 回旋镖投掷的运动学

A brief discussion regarding the kinematics of a boomerang throw is included in this section. The primary objective is to understand how the initial conditions of the boomerang's kinetics are somewhat predetermined by the kinematic motion of an over-hand throw.
本节包括有关回旋镖投掷运动学的简要讨论。主要目标是了解回旋镖动力学的初始条件是如何在某种程度上由上手投掷的运动学预先决定的。
Consider holding a basketball in the cup of your hand and throwing it under-handed as if bowling. The natural motion associated with this toss has the basketball rolling off the fingertips just as the ball is being released. This establishes both a forward velocity as well as a rotational velocity for the basketball. When the basketball makes contact with the ground, its spin rate will remain essentially unchanged. Hence, the natural kinematics of bowling a basketball creates an initial relationship between the ball's velocity and its spin rate.
考虑将篮球放在手杯中,然后像打保龄球一样用手下扔。与这次抛球相关的自然运动使篮球在球被释放时从指尖滚落。这确定了篮球的前进速度和旋转速度。当篮球与地面接触时,其旋转速率将基本保持不变。因此,打篮球保龄球的自然运动学在球的速度和旋转速率之间产生了初始关系。
(15)ωRV
Here, V is the translational velocity of the basketball, ω is its rotational velocity and R its radius. While the relationship defined in Eqn (15) is only an approximate one and the person bowling the basketball can bias it one way or the other, it is quite difficult to violate it entirely.
这里, V 是篮球的平移速度, ω 是它的旋转速度和 R 半径。虽然 Eqn (15) 中定义的关系只是一个近似的关系,并且打篮球的人可能会以某种方式偏向它,但完全违反它是相当困难的。
Let's now consider a boomerang throw and draw similarities with the above example. To start, one holds a blade of the boomerang near its tip, grasping it in the palm of his hand. Because of the width of the hand,the center of this grasp is located around 85%90% of the blade’s radius. Traditionally,a V-type configuration is held by the lifting blade,with the dingle blade- 1 pointed forward at the far end of the boomerang. The boomerang throw proceeds in a manner analogous to throwing a small axe or a hatchet over-handed. The plane of rotation is nearly vertical, with only a slight lay-over angle. Hence, the axis of rotation is nearly horizontal. In terms of Euler's orientation angles, this corresponds to an initial inclination angle of the rotation plane in the range:
现在让我们考虑一个 boomerang throw and draw 与上述示例的相似之处。首先,一个人将回旋镖的刀片靠近其尖端,将其握在手掌中。由于手的宽度,这种抓握的中心位于刀片半径的周围 85%90% 。传统上,V 型配置由提升刀片保持,dingle 刀片 - 1 指向回旋镖的远端。回旋镖投掷的方式类似于双手投掷小斧头或斧头。旋转平面几乎是垂直的,只有一个轻微的放置角度。因此,旋转轴几乎是水平的。就 Euler 的方向角而言,这对应于 范围内旋转平面的初始倾角:
(16)70θ080
Since the boomerang blade must be firmly held, it does not roll out of the hand as efficiently as does the finger-tip roll of the basketball. Hence, the relationship between the initial translational and rotational velocities is slightly different than before. In the case of the boomerang throw, we draw from experience, make an educated guess, and assume this relationship to be:
由于必须牢牢握住回旋镖刀片,因此它不会像篮球的指尖滚动那样有效地从手中滚出。因此,初始平移速度和旋转速度之间的关系与以前略有不同。在回旋镖投掷的情况下,我们从经验中汲取灵感,做出有根据的猜测,并假设这种关系是:
(17)ω0R0.85V0
Here, ω0 and V0 are the boomerang’s initial rotational and translational velocities,respectively. Also define the initial rotational velocity (ω0) of Eqn(17) in a nondimensionalized form; this form will be used throughout the body of this work.
这里, ω0V0 分别是回旋镖的初始旋转和平移速度。还要定义 的初始旋转速度 (ω0) Eqn(17) 以无量纲形式;此表格将贯穿本作品的整个正文。
(18)χ0ω0RV00.85
The initial condition of Eqns (17-18) has the forward-going tip moving at 1.85V0 and the bottom tip also moving forward but only at a speed of 0.15V0 . In practice,a typical throw of a boomerang will yield a translational speed of about 60mph ,with a rotational cadence of about 10Hz .
方程 (17-18) 的初始条件是向前的尖端移动 1.85V0 ,底部尖端也向前移动,但速度为 0.15V0 。在实践中,典型的回旋镖投掷将产生大约 60mph 的平移速度,旋转节奏约为 10Hz

Analysis of a Night Flight
夜航分析

In order to verify the assumption made by Eqns (17-18), we analyze a strobed image of a boomerang in night flight; 6 see Figures 5-6. In Figure 5,note the insert at the bottom-right corner. This insert shows a close-up view of a three-bladed boomerang with an LED strobe light located on each blade at 75% of its radius. Also in Figure 5 is a red rectangular box that traps a portion of the flight path approximately perpendicular to the camera's view. This segment of the flight path is depicted in Figure 6 and is further analyzed. Since there are three strobed paths, focus on the one centered about the solid bold vertical aqua line. To extract the relationship between translational and rotational velocities of the boomerang at this stage of its flight, a hypocycloid is fit through the strobed curve of interest. The result is shown as the bold gray line that fits the strobed-light path. The hypocycloid corresponding to this path is graphically given by the very-large-radius circular contact surface depicted by the orange line at the bottom of the image and the rolling wheel given by the yellow circle. The green circle in this figure represents the swept area of the boomerang while the gray circle coincides with the location of the LED light. Finally, the ratio of the radii of the gray-to-green circles provides a derived relationship of rotational-to-translational velocities for this flight segment. In this particular case, the initial condition of Eqns (17-18) is exactly verified. In the author's experience, the translational velocity decays slightly faster than does the rotational velocity during the course of a boomerang flight. Hence,the nondimensionalized rotational velocity (χ) of a boomerang increases during the course of its mission, from start to finish, so we have:
为了验证 Eqns (17-18) 的假设,我们分析了夜间飞行中回旋镖的频闪图像; 6 参见图 5-6。在图 5 中,请注意右下角的插入。该插页显示了三叶片回旋镖的特写视图,每个叶片在其半径处 75% 都有一个 LED 频闪灯。图 5 中还有一个红色矩形框,它捕获了大致垂直于相机视图的飞行路径的一部分。图 6 描绘了这段飞行路径,并进行了进一步分析。由于有三条叠放路径,因此请专注于以实线粗体垂直浅绿色线为中心的路径。为了提取回旋镖在飞行的这个阶段的平移速度和旋转速度之间的关系,通过感兴趣的选通曲线拟合下摆线。结果显示为适合频闪光路径的粗灰色线。与该路径对应的近摆线由图像底部的橙色线和黄色圆圈表示的滚动轮以图形方式给出。图中的绿色圆圈代表回旋镖的扫掠区域,而灰色圆圈与 LED 灯的位置相吻合。最后,灰色与绿色圆圈的半径比提供了该飞行段的旋转速度与平移速度的推导关系。在这种特殊情况下,方程 (17-18) 的初始条件得到了精确验证。根据作者的经验,在回旋镖飞行过程中,平移速度的衰减速度比旋转速度略快。因此,回旋镖的无量纲旋转速度 (χ) 在其任务过程中从开始到结束都会增加,因此我们有:
(19)dχdt>0

1 The dingle blade is the blade most-influenced by the wake of the other for a V-type boomerang.
1 对于 V 型回旋镖,dingle 刀片是受另一个尾流影响最大的刀片。

This property will be used in Section VII to explain why the inclination angle of a properly-design boomerang reduces during flight.
这一特性将在第七节中用于解释为什么设计合理的回旋镖在飞行过程中的倾角会减小。

Inducing a High-Spin-Rate Flight
诱导高旋转速率飞行

Due to the natural kinematics of a boomerang throw, it is almost impossible to induce an extremely-high rotational velocity at point of launch. However, if one throws a boomerang slightly downward (like throwing a skipping stone) and bounces it off the ground, a high-spin-rate flight can be achieved. In essence, some of the translational kinetic energy gets impulsively converted to rotational kinetic energy. To establish a rough order-of-magnitude estimate, assume that the instantaneous bounce event conserves kinetic energy. Using the moment of inertia of a disk from Eqn (10), the states before and after the bounce are related as:
由于回旋镖投掷的自然运动学,几乎不可能在发射点产生极高的旋转速度。但是,如果将回旋镖稍微向下投掷(就像投掷跳石)并将其从地面上弹起,则可以实现高旋转速率的飞行。从本质上讲,一些平移动能被冲动转化为旋转动能。要建立一个粗略的数量级估计,假设瞬时反弹事件动能守恒。使用 Eqn (10) 中磁盘的惯性矩,反弹前后的状态相关如下:
12mV02+12Iω0212mV12+12Iω12
2V02+(ω0R)22V12+(ω1R)2
(20)[2+χ02]V02[2+χ12]V12.
where R is the radius of the boomerang,and subscripts 0 and 1 indicate the states before and after the bounce event, respectively. With some rearrangement, substitution of Eqn (17) into Eqn (20) gives:
其中 R 是回旋镖的半径,下标 0 和 1 分别表示 bounce 事件前后的状态。经过一些重新排列,将 Eqn (17) 代入 Eqn (20) 得到:
(21)(ω1ω0)2[2.72252(V1V0)20.7225]
(22)(χ1χ0)2[2.7225(V0V1)220.7225].
Hence,if the bounce event reduces the translational velocity (V) by 50% ,the rotational velocity (ω) will increase by about 75% . However in nondimensionalized terms,the ratio between rotational and translational velocities change by a factor of about 3.5 due to the impulse of the bounce,to give χ13.0 .
因此,如果弹跳事件使平移速度 (V) 降低 50% ,则旋转速度 (ω) 将增加约 75% 。然而,在无量纲化的术语中,由于反弹的冲量,旋转速度和平移速度之间的比率发生了大约 3.5 倍的变化,得到 χ13.0
The next section provides some background on work done in the field of boomerang flight dynamics nearly 50 years ago.
下一节提供了近 50 年前在回旋镖飞行动力学领域所做的一些工作的背景。

V. Felix Hess Model V. Felix Hess 模型

In the 1960’s,the aerodynamics of boomerangs was studied by Felix Hess. 7 Hess concluded that for sufficiently rapid rotation, the time-averaged lift and rolling-moment of an advancing boomerang are directly proportional to ω2 and ωV ,respectively. Hence,
1960 年代,Felix Hess 研究了回旋镖的空气动力学。 7 赫斯得出结论,对于足够快的旋转,前进的回旋镖的时间平均升力和滚动力矩分别与 ω2ωV 正比。因此
(23)L¯=aω2,M¯=bωV
where a and b are constants of proportionality, ω is the rotational speed of the boomerang,and V is its flight speed. Note that the coefficients a and b lump together both aerodynamic characteristics and mass properties of a given boomerang. The units of a and b are lbssec2 .
其中 ab 是比例常数, ω 是回旋镖的转速,是 V 它的飞行速度。请注意,系数 ab 将给定回旋镖的空气动力学特性和质量特性归为一谈。和 ba 单位是 lbssec2
Hess utilized his aerodynamic model, given by Eqn (23), in conjunction with the gyroscope model given by Eqn (11), to develop representative flight paths for a family of boomerangs, simply by varying the coefficients a and b . A deficiency of Hess’ aerodynamics model is that it cannot self trim. Further,if manually trimmed, it is neutrally stable. However from practice, we know that boomerangs are dynamically stable over a wide variety of shapes and flight conditions; there must exist an inherent self-trimming capability in their designs. Hence, further work is needed to develop an aerodynamics model that can accommodate this characteristic.
Hess 利用 Eqn (23) 给出的空气动力学模型和 Eqn (11) 给出的陀螺仪模型,只需改变系数 ab ,就可以为一系列回旋镖开发具有代表性的飞行路径。赫斯空气动力学模型的一个缺陷是它不能自我修剪。此外,如果手动修剪,它是中性稳定的。然而,从实践中我们知道,回旋镖在各种形状和飞行条件下都是动态稳定的;他们的设计中必须存在固有的 self-trimning 功能。因此,需要进一步的工作来开发能够适应这一特性的空气动力学模型。
For easier comparison later, Hess' model, given by Eqn (23), is reformulated. In the current work, the rotational speed (ω) is nondimensionalized as follows.
为了便于以后比较,我们重新制定了 Eqn (23) 给出的 Hess 模型。在当前工作中,转速 (ω) 是无量纲的,如下所示。
(24)χ=ωRV=Vtip V
where R is the radius of boomerang blades,and Vtip is the blade tip speed due to rotation. We also refer to χ as the tip-speed ratio. Substitution of Eqn (24) into Eqn (23) gives Hess’ aerodynamic model in terms of
其中 R 是回旋镖叶片的半径,是 Vtip 旋转引起的刀尖速度。我们也称 χ 之为叶尖速度比。将 Eqn (24) 代入 Eqn (23) 可得出 Hess 的空气动力学模型,即
χ as:  χ 如:
(25)L¯=(aR2)χ2V2,M¯=(bR)χV2,
where the bracketed terms are now the constants of proportionality.
其中,括号内的项现在是比例的常数。
Time-averaged relationships for lift and rolling-moment will be significantly enhanced in the next section using Blade Element Theory.
在下一节中,将使用叶片元理论显著增强升力矩和滚动力矩的时间平均关系。

VI. Blade Element Theory VI. 刀片元件理论

An aerodynamic model is developed herein using Blade Element Theory (BET). This model uses the lift-curve of a boomerang's two-dimensional airfoil section and it is assumed that the boomerang operates at angles-of-attack close to zero. 2 Hence,the relationship between lift and alpha is essentially linear.
本文使用叶片元件理论 (BET) 开发了空气动力学模型。该模型使用回旋镖二维翼型截面的升力曲线,并假设回旋镖以接近零的攻角运行。 2 因此,提升和 alpha 之间的关系基本上是线性的。
(26)Cl(α)=Cl0+αClα
Here, α is the angle-of-attack of the airfoil, Cl0 is its lift coefficient at α=0 and Clα is the slope of the lift curve. These values will be determined numerically by running computational fluid dynamics (CFD) on a representative airfoil section of the boomerang blade over a range of angles-of-attack.
其中, α 是翼型的攻角, Cl0 是它的升力系数, α=0 Clα 是升力曲线的斜率。这些值将通过在一定迎角范围内对回旋镖叶片的代表性翼型截面运行计算流体动力学 (CFD) 来确定。
Figure 7 provides a top-view schematic of a single,non-tapered blade that rotates about the Z -axis in a counter-clockwise manner. In this figure, R is the radius of the blade, C is its chord and the freestream velocity (V) is aligned with the X -axis. Also in this schematic, ψ is the angular position of the blade,where ψ=0 represents the position when the advancing blade is perpendicular to the freestream flow. Vn is the local normal on-set velocity component to a blade section; it is a function of radial and angular positions and is given by:
图 7 提供了一个俯视图示意图,该叶片以逆时针方式绕 Z 轴旋转。在此图中, R 是叶片的半径, C 是其弦,自由流速度 (V)X -轴对齐。在此示意图中, ψ 还有叶片的角度位置,其中 ψ=0 表示前进叶片垂直于自由流时的位置。 Vn 是叶片截面的局部法线起始速度分量;它是径向和角度位置的函数,由下式给出:
(27)Vn(r,ψ)=Vcosψ+ωr
Here, ω=ψ˙ is the rotational speed of the blade spinning about the Z -axis. In the present work,this rotational speed is nondimensionalized as follows.
这里, ω=ψ˙ 是刀片绕 Z -轴旋转的转速。在本工作中,该转速是无量纲的,如下所示。
(28)χ=ωRV
Figures 8-10 illustrate how Vn varies within the disk area swept by the blade for rotational speeds of χ=(0.7,1.0,1.5) ,respectively. In these images,the freestream velocity is from left-to-right. Note that Vn<0 is depicted by the blue-shaded regions which only occur when 90ψ270 . As an aside,the Vn=0 contour is a circle of diameter 1χ .
图 8-10 分别说明了当转速为 χ=(0.7,1.0,1.5) 时,叶片扫过的圆盘区域内 Vn 的变化情况。在这些图像中,自由流速度是从左到右。请注意, Vn<0 它由蓝色阴影区域表示,该区域仅在 时 90ψ270 出现。顺便说一句, Vn=0 轮廓是一个直径 1χ 为 的圆。
In general,the rotational plane of the blade can be at an angle-of-attack (α) to the freestream velocity (V) . The velocity component normal to the rotational plane (Vz) can be approximated as:
通常,叶片的旋转平面可以处于自由流速度 (V) 的攻角 (α) 。垂直于旋转平面 (Vz) 的速度分量可以近似为:
(29)Vz=VsinααV,|α|1.
As the blade sweeps in ψ ,the airfoil sections along its span are subjected to a cyclic variation of both normal velocity (Vn) and apparent angle-of-attack. Designate this effective angle-of-attack as αn and approximate
当叶片扫入 ψ 时,沿其跨度的翼型截面会受到法向速度和 (Vn) 视攻角的循环变化。将此有效攻角指定为 αn 和 近似值
with: 跟:
(30)αn=tan1(Vz|Vn|)Vz|Vn|,|Vz||Vn|.

2 Actually,very large angles-of-attack exist,but only at very low dynamic pressures,so the error introduced is minimal.
2 实际上,存在非常大的迎角,但仅限于非常低的动态压力下,因此引入的误差很小。

The reason the absolute value of Vn is needed above is because,even when Vn is negative,the sign of lift due to angle-of-attack is solely determined by the sign of Vz . As will be seen later,this complicates matters in the derivations for lift and rolling-moment due to angle-of-attack.
上面需要绝对值 的 Vn 原因是,即使为负数 Vn ,攻角引起的升力符号也完全由 Vz 的符号决定。正如稍后将看到的,这使得攻角引起的升力和滚动力矩的推导变得复杂。
While all boomerangs have more than one blade, with blade element theory each are treated identically; there is no distinction between the force and moments of the lifting and dingle blades. As a consequence, deriving the coefficients of lift and rolling-moment for a single blade is equivalent to doing so for an N -bladed configuration. For example, a 3-bladed boomerang will have 3 times the lift of one of its blades but its lift will also be nondimensionalized by 3 times the reference area of a blade.
虽然所有回旋镖都有不止一个刀片,但在刀片元素理论中,每个刀片的处理方式都是相同的;Lifting 和 Dingle 刀片的力和力矩没有区别。因此,推导出单个叶片的升力和滚动力矩系数等同于推导出 N -bladed 配置的升力和滚动力矩系数。例如,一个 3 叶片的回旋镖将具有其其中一个叶片的 3 倍的升力,但其升力也将由叶片参考面积的 3 倍进行无量纲化。

Lift Derivation Lift 派生

The next step is to develop a relationship for the dimensional lift generated by the blade as a function of its angular position (ψ) .
下一步是建立叶片产生的尺寸升力与其角位置的 (ψ) 函数的关系。
L(ψ)=0R12ρVn2Cl(αn)Cdr
=12ρC0RVn2[Cl0+Vz|Vn|Clα]dr
(31)=12ρC0R[Vn2Cl0+|Vn|VzClα]dr.
Here,the integration is over the span of the blade, ρ is air density, Cl(αn) is the 2D sectional lift coefficient given by Eqn (26),but now as a function of αn and C is chord length. Eqns (26-30) have also been incorporated into Eqn (31). By observation, note that the total lift of Eqn (31) can be split into a basic lift and a lift due to angle-of-attack. To help keep the remaining derivations manageable, we will develop these lifts separately, then combine later.

Basic Lift

In the context of this work, basic lift is defined as that which is generated by an airfoil acting at no angle-of-attack to its onset flow. Starting with the first term of the integrand of Eqn (31), and substituting Eqns (27-28), the basic-lift derivation continues as:
L0(ψ)=12ρCCl00R[Vcosψ+ωr]2dr
=12ρV2CCl00Rcos2ψ+2cosψχ(rR)+χ2(rR)2dr
=12ρV2RCCl0[cos2ψ+cosψχ+13χ2]0R
(32)=qSCl0[cos2ψ+cosψχ+13χ2].
where q=12ρV2 is the freestream dynamic pressure and S=RC is the blade planform area.
To determine a time-averaged value for the basic lift, L0(ψ) is integrated over ψ . However,since L0 is symmetric about the Y -axis,this integration can be performed from 0 to π .
L¯0=1π0πL0(ψ)dψ
=1πqSCl0[12(ψ+sinψcosψ)+χsinψ+13χ2ψ]0π
(33)=qSCl0[12+13χ2]
Comparing Eqn (33) with Eqn (25) illustrates that Hess' aerodynamic model is insufficient for small rotational speeds where χ<1.5 ,even for basic lift.
Before proceeding to the derivation of the lift due to angle-of-attack, it will be necessary to briefly touch on a special integration topic. Consider a line segment on the interval [0,1] of the form y(x)=a+bx . If this segment is either completely above or below the X -axis,then an integration of its absolute value is trivial. However,let’s consider the situation where the line starts below the X -axis at x=0 and ends above it at x=1 . For this case,it can be shown that:
(34)01|a+bx|dx=[a2b+a+b2],a0|a|b.
Here, a and b are arbitrary constants. Note that the first term in the square brackets of Eqn (34) is an addition to the result one would get if the integrand had not been made an absolute value.

Lift due to Angle-of-Attack

We now proceed with the derivation of lift due to angle-of-attack, starting with the second term of the integrand of Eqn (31).
Lα(ψ)=12ρC0R|Vn|VzClαdr
=12ρC0R|Vcosψ+ωr|αVClαdr
=12ρV2αClαC0R|cosψ+χ(rR)|dr
(35)=qSαClα01|cosψ+χη|dη
Here, η=rR is the nondimensional spanwise coordinate along the blade. Because of the absolute-valued term inside the integral of Eqn (35),the following four cases exist for 0ψπ .
For 0ψπ2 ,and any value of χ>0 ,the term [cosψ+χη]0 over the interval of integration,so:
(36)Lα(ψ)=qSαClα[cosψ+χ2].
For π2<ψπ ,and χ1 ,the term [cosψ+χη] switches sign,so we apply Eqn (34) to get:
(37)Lα(ψ)=qSαClα[cos2ψχ+cosψ+χ2].
For π2<ψcos1(χ) ,and χ<1 ,the term [cosψ+χη] also switches sign,and again we have:
(38)Lα(ψ)=qSαClα[cos2ψχ+cosψ+χ2].
For cos1(χ)<ψπ ,and χ<1 ,the term [cosψ+χη]<0 over the interval,so:
(39)Lα(ψ)=qSαClα[cosψχ2].
In order to determine a time-averaged value for lift due to angle-of-attack, Lα(ψ) is integrated over ψ from 0 to π . It is possible to do this considering only two cases; one with χ1 and the other with χ<1 . However, this does require a little manipulation of Eqns (36-39) when folded into the integration over ψ . For χ1 ,
L¯α=1πqSαClα{0π[cosψ+χ2]dψ+π2π[cos2ψχ]dψ}
=1πqSαClα{[sinψ+χψ2]0π+[12χ(ψ+sinψcosψ)]π2π}
(40)=qSαClα[χ2+14χ].
For χ<1 ,
L¯α=1πqSαClα{0ψc[cosψ+χ2]dψ+π2ψc[cos2ψχ]dψψcπ[cosψ+χ2]dψ}
=1πqSαClα{[sinψ+χψ2]0ψc+[12χ(ψ+sinψcosψ)]π2ψc[sinψ+χψ2]ψcπ}
=1πqSαClα{321χ2+χ[ψcπ2]+12χ[ψcπ2]}
(41)=1πqSαClα{321χ2+[χ+12χ][ψcπ2]}.
where ψc=cos1(χ) .
As it turns out, boomerangs require a spin rate sufficient to be stable in flight. Furthermore, the kinematics of a boomerang throw naturally place the spin rate such that χ=O(1) . Depending on the person,this typically results in spin rates of χ0.7 . With a lower limit on χ being known,let’s expand the arc-cosine about χ0=1 as follows.
(42)ψc=cos1(χ)[π21χ]162(1χ)32
Retaining only the bracketed two terms of this expansion is sufficient for our purposes and plugging these into Eqn (41) gives an approximate form that is more analogous with that of Eqn (40). For 0.7<χ<1 ,
(43)L¯αqSαClα{[χ2+14χ][122π1χ]+32π1χ2}.
By observation,Eqns (40 &43) exhibit a smooth transition across χ=1 for the lift due to angle-of-attack; this is expected and serves as a sanity check that the derivation for lift due to angle-of-attack may be correct.

Coefficient of Total Lift

Combining Eqns (33,40,43) ,gives an expression for total lift,and nondimensionalizing by (qS) provides a time-averaged total-lift coefficient for a boomerang based on blade element theory of:
(44)C¯L=Cl0[12+13χ2]+αClα{[χ2+14χ][122π1χ^]+32π1χ^2},
where χ^=min(χ,1) and Eqn (44) is appropriate for χ0.7 .

Rolling-Moment Derivation

The next step is to develop a relationship for the dimensional rolling-moment about the X -axis generated by the blade as a function of its angular position (ψ) .
M(ψ)=0R12ρVn2Cl(αn)Crcosψdr
=12ρC0RVn2[Cl0+Vz|Vn|Clα]rcosψdr
(45)=12ρC0R[Vn2Cl0+|Vn|VzClα]rcosψdr.
Here,the integration is over the span of the blade, ρ is air density, Cl(αn) is the 2D sectional lift coefficient given by Eqn (26) and rcosψ is the y -arm on which a sectional lift acts to generate an incremental rolling-moment. Eqns (26-30) have been incorporated into Eqn (45). By observation, note that the total rolling-moment of Eqn (45) can be split into a moment due to basic lift and one due to angle-of-attack. To help keep the remaining derivations manageable, we will develop these rolling-moments separately, then combine later.
In the context of this work, basic rolling-moment is defined as that which is generated by an airfoil acting at no angle-of-attack to its onset flow. Starting with the first term of the integrand of Eqn (45) and substituting Eqns (27-28), the derivation for the basic rolling-moment continues as:
M0(ψ)=12ρCCl00R[Vcosψ+ωr]2rcosψdr
=12ρV2RCCl00Rcos3ψ(rR)+2cos2ψχ(rR)2+χ2(rR)3cosψdr
=qSRCl001cos3ψη+2cos2ψχη2+χ2η3cosψdr
(46)=qSRCl0[12cos3ψ+23cos2ψχ+14cosψχ2],
where q=12ρV2 is the freestream dynamic pressure, η=rR is the nondimensional span along the blade, dr=Rdη and S=RC is the blade planform area.
To determine a time-averaged value for the basic rolling-moment, M0(ψ) is integrated over ψ . However, since M0 is symmetric about the Y -axis,this integration can be performed from 0 to π .
M¯0=1π0πM0(ψ)dψ
=1πqSRCl0[124(9sinψ+sin3ψ)+13χ(ψ+sinψcosψ)+14χ2sinψ]0π
(47)=13qSRCl0χ
Comparing Eqn (47) with Eqn (25) illustrates that Hess' aerodynamic model for a time-averaged rolling-moment is analogous to the basic rolling-moment derived by blade element theory.

Another Special Integration Topic

Before proceeding to the derivation of the rolling-moment due to angle-of-attack, it will be necessary to briefly touch on another special integration topic. Consider a parabola on the interval [0,1] of the form y(x)=ax+bx2 . If this curve segment is either completely above or below the X -axis,then an integration of its absolute value is trivial. However,let’s consider the situation where the curve moves below the X -axis at x=0 and ends up above it at x=1 . For this case,it can be shown that:
(48)01|ax+bx2|dx=[a33b2+a2+b3],a0|a|b.
Here, a and b are arbitrary constants. Note that the first term in the square brackets of Eqn (48) is a positive value and is an addition to the result one would get if the integrand had not been made an absolute value.

Rolling-Moment due to Angle-of-Attack

We now continue with the derivation of rolling-moment due to angle-of-attack, starting with the second term of the integrand of Eqn (45).
Mα(ψ)=12ρC0R|Vn|VzrcosψClαdr
=12ρC0R|Vcosψ+ωr|αVrcosψClαdr
=12ρV2αClαC0R|cosψ+χ(rR)|rcosψdr
(49)=qSαClαcosψ01|cosψ+χη|ηdη.
Here, η=rR is the nondimensional spanwise coordinate along the blade. Because of the absolute-valued term inside the integral of Eqn (49),the following four cases exist for 0ψπ .
For 0ψπ2 ,and any value of χ>0 ,the term [cosψ+χη]0 over the interval of integration,so:
(50)Mα(ψ)=qSRαClα[12cos2ψ+χ3cosψ].
For π2<ψπ ,and χ1 ,the term [cosψ+χη] switches sign,so we apply Eqn (48) to get:
(51)Mα(ψ)=qSRαClα[cos4ψ3χ2+12cos2ψ+χ3cosψ].
For π2<ψcos1(χ) ,and χ<1 ,the term [cosψ+χη] also switches sign,and again we have:
(52)Mα(ψ)=qSRαClα[cos4ψ3χ2+12cos2ψ+χ3cosψ].
For cos1(χ)<ψπ ,and χ<1 ,the term [cosψ+χη]<0 over the interval,so:
(53)Mα(ψ)=qSRαClα[12cos2ψχ3cosψ].
To determine a time-averaged value for rolling-moment due to angle-of-attack, Mα(ψ) is integrated over ψ from 0 to π . It is possible to do this considering only two cases; one with χ1 and the other with χ<1 . However,this does require a little manipulation of Eqns (50-53) when folded into the integration over ψ . For χ1 ,
M¯α=1πqSRαClα{0π[12cos2ψ+χ3cosψ]dψ+π2π[cos4ψ3χ2]dψ}
=1πqSRαClα{[14(ψ+sinψcosψ)+χ3sinψ]0π[12ψ+8sin2ψ+sin4ψ96χ2]π2π}
(54)=qSRαClα[14116χ2]
For χ<1 ,
M¯α=1πqSRαClα{0ψc[12cos2ψ+χ3cosψ]dψ+π2ψc[cos4ψ3χ2]dψψcπ[12cos2ψ+χ3cosψ]dψ}
=1πqSRαClα{[ψ+sinψcosψ4+χ3sinψ]0,πψc[12ψ+8sin2ψ+sin4ψ96χ2]π2ψc}
=1πqSRαClα{[ψcπ2+sinψccosψc2+2χ3sinψc][12ψc6π+8sin2ψc+sin4ψc96χ2]}
=1πqSRαClα{ψcχ1χ22+2χ31χ2π412ψc6π(8χ3+12χ)1χ296χ2}
=qSRαClα{116χ2ψc8πχ2+1χ28πχ14+ψc2π+χ1χ24π}
(55)=qSRαClα{(π2ψc)16πχ2+1χ28πχ(π2ψc)4π+χ1χ24π}.
where ψc=cos1(χ) . Note the nomenclature above: []0,πχc([]0χc+[]πχc)([]0χc[]χcπ) .
Substituting the bracketed two terms of the arc-cosine expansion of Eqn (42) into Eqn (55) gives an approximate form more analogous to that of Eqn (54).
For 0.7<χ<1 ,
(56)M¯αqSRαClα{[14116χ2]+21χ2π[14χ21]+14π1χ2[12χ+χ]}.
By observation,Eqns (54 &56) exhibit a smooth transition across χ=1 for the rolling-moment due to angle-of-attack; this is expected and serves as a sanity check that the derivation for rolling-moment due to angle-of-attack may be correct.

Coefficient of Total Rolling-Moment

Combining Eqns (47,54,56) ,gives an expression for total rolling-moment and nondimensionalizing by (qSR) provides a time-averaged total rolling-moment coefficient for a boomerang based on blade element theory of:
(57)C¯M13Cl0χ+αClα{[14116χ2][1221χ^π]+1χ^24π[12χ^+χ^]},
where χ^=min(χ,1) and Eqn (57) is appropriate for χ0.7 .

Pitching-Moment Derivation

Since cyclic lifting forces associated with BET are symmetric about the Y -axis,an integration for the time-averaged pitching-moment will always vanish. Hence,
(58)C¯Y=0
Here C¯Y is the moment coefficient about the Y -axis (pitch direction).
Before we leave this section, we remind the reader that the BET derivations developed herein are based on non-tapered blades with zero off-set to the point of rotation. If the present boomerang theory is ever to be expanded upon, it should be done so with consideration to both of these geometric properties. However, even then,Eqn (58) will still hold. In order to achieve a non-zero C¯Y ,downwash and/or wake effects must be included in the analysis. Until then, the lay-over phenomenon of boomerang dynamics cannot be explicitly studied. Nonetheless, the present work provides a good foundation in the understanding of boomerang aerodynamics. And although the investigations of Sections VII-VIII focus on time-averaged properties, the aerodynamic formulations of this section have been developed such that they can be utilized for time-accurate flight simulations of boomerangs with their actual mass properties of I11I22 .
In the next section, we will study various flight-path characteristics of a returning-type boomerang.

VII. Radius of a Boomerang's Flight Path

Let's assume that it is possible for a boomerang to reach a time-averaged steady-state motion of level flight which forms,when viewed from above,a circular path of radius Rp . Further,assume the boomerang rotation plane maintains a constant bank-angle of θ during its flight. In this scenario,the precession rate of the boomerang rotation plane (ϕ) ,times the radius of the circular flight path (Rp) ,must be equal to the velocity of the advancing boomerang (V) .
(59)V=ϕ˙Rp
For this same circular level flight path, the centrifugal acceleration is balanced by the horizontal component
of the lift vector.
(60)Lsinθ=m(V2Rp)
To sustain level flight, the vertical component of lift must balance the boomerang's weight.
(61)Lcosθ=mg
Here m is mass and g is gravitational acceleration. Strictly speaking,the equality of Eqn (61) must be enforced for level flight. In practice however, a fairly large departure from this state can occur. For example, consider playing catch with someone where a baseball is thrown back and forth. The path of the baseball is essentially a parabolic arc, as no lift is produced by the ball. The players of this game must throw the ball with initial velocity vectors appropriate for the path of the ball to intersect the reciever's position. In the case of a boomerang throw, while some lift in the vertical direction is being generated, it may not exactly balance the weight. Here, the thrower compensates the lift-weight imbalance simply by launching the boomerang slightly upwards to give just the right amount of time aloft for the boomerang to return. Becuase of this extra degree of freedom, Eqn (61) will not be enforced as a constraint in the current body of work.
Recall that Eqn (14) provides an approximation for the precession rate (ϕ) in terms of rolling-moment (Mn) . If the boomerang is designed to fly at a negligible angle-of-attack,then the basic rolling-moment of Eqn (47) applies. Equating these relationships of rolling-moment and substitution of Eqn (59), gives the following sequence of equalities.
(12mR2)sinθψ˙ϕ˙=(12ρV2)SRCl0[χ3]
mR2sinθ(χVR)(VRp)=ρV2SRCl0[χ3]
(62)msinθRp=ρSCl03
Rearranging Eqn (62) gives a compact derivation for the radius of the circular flight path. 3
(63)Rp=[3msinθρSCl0]
Remarkably,note that Rp is independent of the boomerang’s velocity (V) or its rotational speed (ω) . As a consequence, the radius of the boomerang's flight path can be set by design.
If we rearrange Eqn (60) for Rp and substitute the basic lift given by Eqn (33),we can solve for an equilibrium value of θ for a level circular flight path.
(64)sin2θ=[34+χ22]1
Amazingly, Eqn (64) draws a general conclusion which holds for any boomerang, regardless of mass or dimensional size! The only assumptions used to lead to this result are that the boomerang trims at α=0 , its moment of inertia is I=12mR2 ,that its flight path is level and that BET provides a reasonable aerodynamic model.
For a real solution of Eqn (64) to exist, the following condition regarding the boomerang's rotational velocity must hold.
(65)χ220.707
Recall the predetermined initial state of a boomerang throw as described by Eqn (17). This relationship stated that χ00.85 . Now substitute this value of χ0 into Eqn (64) to derive an initial inclination angle for the boomerang's orientation as it is being thrown.
(66)θ071.6
This result fits nicely within the range of applied experience as given by Eqn (16). Experience also has shown that χ is not constant during boomerang flight; see Eqn (19). Considering this,it may be advantageous to combine Eqns (63-64) to define another relationship for the radius of a level, circular flight path.
(67)Rp=[6mρSCl03+2χ2]
3A similar conclusion can be drawn using Hess’ aerodynamic model.
Therefore,from start to finish,as χ increases,the flight-path radius (Rp) reduces. If the boomerang ever ends up near a hover mode where θ0 and χ ,it can rapidly precess such that this segment of its flight path will be a tight loop when viewed from above.
In Section IV is a discussion on bouncing a boomerang off the ground to induce a high-spin-rate flight. The example there showed that a value of χ=3 can be achieved if the velocity (V) is reduced by 50% due to the impulse of the bounce. Through some experimentation, the author was able to achieve several such flights and managed to catch them on high-definition video. Qualitatively,the radius of these flight paths (Rp) is the same as that for a regular throw of the same boomerang when θ70 . This observation is consistent with the implications of Eqn (63). During one of these flights, the bounce itself caused the boomerang to initially climb, then fall; from the thrower's view the path looked like a parabolic arc. However, about half-way down from its apex, the boomerang began to flare and adjust its attitude until it established an equilibrium flight path with a nominal sink rate. Measuring the inclination angle of the high-spin boomerang off video footage showed θ22 . Although estimates for these velocities are rough,Eqn (64) nonetheless gives θ125.9 for χ3 ,which is in close agreement to that observed.
Conditions for static trim and dynamic stability are developed in the next section.

VIII. Static Trim & Dynamic Stability

Consider a boomerang traveling in a level, circular flight path. Following a similar derivation as before for Eqns (62-64), one can determine a general relationship for the trim state to be:
(68)(CM:CL)CMCL=[ImR2]sinθχ.
Here, C¯L and C¯M are the three dimensional,time-averaged lift and rolling-moment coefficients,respectively, and I is the moment of inertia. In general,the aerodynamic coefficients can be determined with BET equations, CFD simulation, or by experimental testing; it does not matter. However, note that this ratio is also a function of angle-of-attack; we will show that the condition for dynamic stability is:
(69)d(C¯MCL)dα<0
To illustrate why this is the case, for example, assume that a boomerang is initially thrown at a condition of α=0 ,whereas its trim state may occur at α=2 . If the slope of the (CM:C¯L) curve is negative,then initially the (CM:C¯L) ratio will be too large for static trim. Because of this state of inequilibrium,the boomerang's precession rate will turn the rotation plane faster than the lateral component of lift turns the direction of flight; thus increasing the angle-of-attack of the rotation plane with respect to flight direction. This automatic adjustment will continue until the angle-of-attack reaches its trim value of α=2 in this example. If this adjustment slightly over-shoots the trim condition,then the (CM:C¯L) ratio will be too small for static trim, now the horizontal lift component will turn the flight path faster than the rotation plane precesses and the angle-of-attack will be nudged back accordingly.
For the moment,consider χ1 . This will greatly simplify the following derivation for a condition of dynamic stability based on the BET aerodynamics model. Now substitute Eqns (44 & 57) into Eqn (68) and multiply numerator and denominator by 48χ2 to obtain:
(70)CMCL=Cl0[χ3]+αClα[14116χ2]Cl0[12+χ23]+αClα[χ2+14χ]=16Cl0[χ3]+αClα[12χ23]Cl0[24χ2+16χ4]+αClα[24χ3+12χ].
Note that Eqn (70) is in the form:
(71)CMCL=a+bαc+dα
where, a,b,c and d are independent of α . Differentiating Eqn (71) and applying the stability condition gives:
(72)d(CM¯CL)dα=b(c+dα)d(a+bα)(c+dα)2=(bcad)(c+dα)2<0.
Hence,if [bc<ad] ,then the boomerang is dynamically stable. Assuming that Cl0 and Clα are non-zero and positive, we have:
[12χ23][24χ2+16χ4]<[16χ3][24χ3+12χ]
[192χ4+240χ272]<[384χ4+192χ2]
(73)0<[192χ448χ2+72]
Eqn (73) always holds true,provided Cl0>0 and Clα>0 . However,at the beginning of this derivation thread,we assumed χ1 to simplify the mathematics. Therefore,we have only proven that boomerangs are dynamically stable for χ1 thus far. Nonetheless,dynamic stability seems to be an intrinsic property of the boomerang.
In the next section, a detailed boomerang geometry is developed as a test case for CFD analysis.

IX. Boomerang Geometry

In this section, a boomerang design is developed for further CFD analysis. This configuration is representative of a starting point in a design process. Due to time limitations, no effort has been extended to improve upon this design... yet.
Figure 11 illustrates the general layout of the boomerang in planform view. Spanwise element lines are shown at quarter-chord intervals. Airfoil sections are depicted at five stations along the span of the boomerang. Note that the leading edge of the right blade is towards the top of the page, whereas the leading edge of the left blade is towards the bottom of the page. The twist distribution in this design is identically zero everywhere. This boomerang is designed to rotate counter-clockwise and is appropriate for a right-handed thrower. Because the airfoil sections flip direction, note that the centerline airfoil is symmetric front-to-back. The mid-chord lines of the trap planform are anchored to the origin at the centerline, their radial span is 12 inches and the angle between them is 100 . Note the curved planform near the centerline. Also shown in this figure is the center-of-mass location depicted by the asterisk symbol.
Figure 12 provides characteristics of the outboard airfoil sections, including thickness and camber distributions as well as airfoil geometry. The airfoil is 15% thick,with the maximum thickness occuring at 35% chord. The lower surface is flat from 2.5% to 97.5% chord. Both the leading and trailing edges have a radius of about 3% chord. The purpose of having a rounded trailing edge is that the airfoil is subjected to reversing onset flows during each cycle; refer back to Figures 8-10.
Reference quantities are provided in Table I. Mass properties are given in Table II.
Table I Table II Reference Quantities
Mass Properties
ρw0.0360slugs/in 3
m0.0207slugs
Xcg3.1780in
I110.4320slugs-in 2
I220.1020slugs-in 2
I330.5340slugs-in 2
Rg5.0780in
Cref2.00in
Rref12.00In
Sref59.80in2
λ0.67
A40.0°
Swet 125.60in2
Volume18.39in3
Reference quantities are used in CFD calculations, λ is the taper ratio of the trap wing, Λ is the mid-chord sweep of the trap wing, Swet is the surface-wetted area,Volume is the volume of the boomerang, ρw is the density of Australian mangrove wood, m is the boomerang’s mass, Xcg is the X -location of the center-of-mass,the I ’s are principal moments of inertia and Rg is the radius of gyration about the principal axis of rotation.
The results from 2D steady-state CFD solutions are discussed next.

X. 2D Steady-State CFD Solutions

In this section, the boomerang airfoil developed in the previous section is analyzed using the OVERFLOW code. OVERFLOW 8 is a Reynolds-averaged Navier-Stokes method develop by NASA. The primary purpose of this 2D analysis is to determine Cl0 and Clα values for the boomerang airfoil. These two quantities will be used in conjunction with BET Eqns (44 & 57) for comparison with the 3D CFD results in Section XI.
Figure 13 illustrates a close-up view of the O-mesh grid used for OVERFLOW calculations. Dimensions of this mesh are (257×97) . The viscous spacing at the wall is Δy1=0.0000125C which is approximately Δy1+=1 . Figure 14 provides pressure contours of the flowfield and depicts the pockets of flow separation with red streamlines. The flow condition shown here is Mach=0.0774 and α=0 . Note the large region of flow separation at the trailing edge of the airfoil. This indicates that the initial airfoil specified for the boomerang section is probably much too thick. Figures 15-16 provide the lift curve and the drag polar for the boomerang airfoil, respectively. A least-squares linear fit of the lift curve gives:
(74)Cl0.2D=0.164,Clα.2D=4.433
where the units of Clα are per radian.
As an alternative to the coefficients of Eqn (74), which are based on 2D CFD solutions, we also include a set of coefficients that use the form of BET, but are reverse-engineered from time-averaged 3D unsteady CFD simulations on the fine mesh,with χ=1 . Fitting these data gives:
(75)Cl0.3D=0.262,Clα.3D=3.056
Comparisons of 3D cyclic CFD calculations with BET estimates using both sets of coefficients will be provided in the next section.

XI. 3D Unsteady CFD Simulations

This section provides numerical results for the boomerang configuration subjected to a forced rotary cyclic motion,submersed in uniform freestream flow. The freestream flow corresponds to 60mph at sealevel conditions near Wagga Wagga,Australia; Mach =0.0774,Ren=43,572 and T=538.5K . Air density at sea level is ρ=1.3755×106slugs/in3 . The aerodynamic database developed includes variations in rotational velocity and angle-of-attack.
A sequence of three volumetric meshes was generated for CFD computations. The fine mesh, with dimensions (257×169×97) ,was generated using the globally-elliptic meshing method of Vassberg. 9,10 Viscous spacing at the wall is Δy1=0.000025 in which is approximately Δy1+=1 . Limited fine-mesh solutions were obtained due to their significant computational requirements, so medium and coarse meshes were also developed. Starting with the fine mesh, a medium mesh level was constructed by removing even-numbered computational planes of the fine mesh. This process is repeated using the medium mesh to create the coarse mesh. Dimensions of these meshes are (129×85×49) and (65×43×25) ,respectively.
Figure 17 illustrates a collection of truncated field planes near the boomerang surface. The predominately-vertical O-mesh planes wrapped around airfoil sections are J-planes. The horizontal planes at the boomerang leading and trailing edges are I-planes. The boomerang surface itself is the K=1 plane.
Figures 18-23 are a sequence of images which depict K-planes in ascending order. This grid is generated by starting with the boomerang surface (K=1) and extruding outwards to the farfield boundary. The farfield boundary is almost a perfect spherical surface and is about 100Crefs  away from the boomerang; see Figure 23.
Figures 24-25 provide cyclic lift and rolling-moment coefficients for Mach=0.0774,α=0 and χ=1 . These data were computed on the fine mesh. The flowfield was initialized to freestream conditions and the forced cyclic motion of spinning the boomerang about its center-of-mass location was instantaneously started. The transients of this start-up process were allowed to sufficiently damp out, requiring 6 cycles. Hence,data presented herein is for the 7th  cycle. To compare with BET,coefficients for the right and left blades are presented separately and have been shifted in phase by 40 and 140 ,respectively. In these figures, ψ=0 corresponds to the angular position of each advancing blade when its mid-chord-line is perpendicular to the freestream flow. The BET results are shown as solid lines, whereas the CFD results for the right and left blades are depicted as dashed lines with solid and open symbols, respectively. Note that BET captures the general trends of the CFD results, but not exactly. Further, the lifting (right) wing is shown to be more powerful than the dingle (left) wing, based on CFD. The reason is that the wake of the lifting wing has a much stronger influence on the dingle wing than vice-versa. One surprise revealed by these results is that BET, based on the 2D steady-state CFD coefficients of Eqn (74), under-predicts the lift and rolling-moment coefficients - usually the opposite occurs. However, in the case of 3D unsteady boomerang aerodynamics, cyclic reversals of flow direction over the airfoil sections, combined with large fluctuations of spanwise flow, work to keep the flow better attached to the boomerang's surface than predicted by 2D steady-state airfoil solutions,at least for α=0 . Using the coefficients of Eqn (75) yields a much better agreement with 3D unsteady CFD results; these are depicted as solid lines with × symbols in Figures 24-25.
Figures 26-38 provide a sequence of upper-surface isobars for the boomerang, again at flow conditions of Mach=0.0774,α=0 and χ=1 ,for the 7th  cycle. However,for these figures, ψ corresponds to the boomerang angular position,not the individual blades’ position. This sequence begins at ψ=0 and steps forward in time at 30 intervals until the boomerang completes one full cycle,returning it to its initial position at ψ=360 . Note that the color spectrum of Cp-contours is held at a constant range of [2,0] over the duration of the spin cycle. Here,dark-blue contours depict Cp0 ,while bright-red contours signify Cp2 . In these figures,the freestream flow is from left-to-right and the boomerang rotation is counter-clockwise. In general,the pressures on a retreating blade are close to Cp0 ,while an advancing blade exhibits regions of flow where the pressures approach Cp2 ,when it is nearly perpendicular to the freestream flow direction. Figures 27&31 show advancing blades near their peak lifting positions.
In order to develop an aerodynamic database for the boomerang, it was deemed necessary to run these cases on the coarse mesh. For reference, coarse-mesh transients sufficiently damp out within 3 cycles, so only a total of 4 cycles are required. This, in combination with the fact that the fine mesh is 64 times as large as the coarse mesh, reduces computational requirements by more than two-orders-of-magnitude.
Figure 39 illustrates the ratio of (C¯M:C¯L) ,as predicted by CFD on the coarse mesh. The significance of this relationship between the lift and rolling-moment coefficients is two-fold. Firstly, it establishes trimming conditions on α as a function of (χ,θ) ,per Eqn (68). Secondly,it shows the boomerang to be dynamically stable because its slope is negative. For example,consider a condition of χ=1.2 and θ=70 . Recall the boomerang’s values of I=0.534slug in2,m=0.0207 slugs and R=12 in. Plugging these into Eqn 68 gives a required (CM:C¯L) ratio of 0.202 . Hence,under these conditions,the boomerang will trim at α=2.5 . Furthermore, not only will it trim at this condition, it will be dynamically stable, per Eqn (69).
Figures 40-41 provide time-averaged values of lift and rolling-moment coefficients as a function of angle-of-attack, respectively. BET results are depicted by solid lines while CFD data (as computed on the coarse mesh) are given as solid symbols on dashed lines. Note that "BET-2D" (based on 2D steady-state CFD coefficients) under-predicts the lift at small angles-of-attack,but then over-predicts it at α>3.4 ,whereas it over-predicts the rolling-moment at all conditions run with CFD. The "BET 3D" data is based 3D unsteady CFD solutions on fine-mesh. BET 3D predicts a slightly-higer level of lift than CFD does on the coarse mesh. Further, it predicts a much larger rolling-moment than CFD does. This is probably due to blade taper and off-set effects which are not captured by BET derivations herein.

XII. Coupled CFD-6DOF Simulations

A very preliminary numerical simulation of a boomerang flight has been performed using the six-degree-of-freedom (6DOF) capability of OVERFLOW. For this flight, the environmental condition has a wind blowing in the X -direction at 5mph . The boomerang is launched from the origin with an initial velocity of V0=60mph and rotational speed of ω0=88rad/sec ,corresponding to χ0=1 . The heading of the throw is directly out the Y -axis and 30 upward. The initial inclination angle of the boomerang’s axis of rotation is θ=75 . Note that this throw is perpendicular to the wind direction. In addition,the thrower is right-handed and the expected path from his point-of-view is for the boomerang to curve left into the wind, make a full circle and return.
Figure 42 provides a four-view of the computed flight path depicted as solid lines with asterisk symbols and encapsulated by a box with 2,000-inch sides. For reference, the sole asterisk in the center of the box is located at (X,Y,Z)=(1000,1000,1000) inches. Now focus on the top-left quadrant of this figure where the image is a view from above. In this quadrant, the wind is coming from the top of the page and the boomerang is thrown to the right from the origin which is located at the bottom-left corner of the square. Also provided in this view is a circular arc of radius 2,783 inches; this is the value of Rp as estimated by Eqn (63). It is incredible how similar the flight path of the numerical simulation matches the predicted circular path at point of launch. Furthermore,as the boomerang flight progresses and χ increases,the radius of the 6DOF flight path is reduced. This characteristic of the flight path tightening is qualitatively predicted by Eqns (19 & 67). In this simulation, the boomerang reaches out almost 2,600 inches, climbs to a height of about 1,400 inches and returns about half-way before making contact with the ground. Although this numerical flight is far from perfect, it exhibits behavior that verifies BET predictions.
Figure 43 provides a stereoscopic pair of images for the oblique view of Figure 42. In order to view this figure in 3D,hold the page about 125 away from the eyes and stare deeply into the images until they merge in the center. Adjustment fore-and-aft may be necessary to pull the view into focus.

XIII. Summary and Conclusions

A new aerodynamic model has been developed using Blade Element Theory for the purpose of investigating aerodynamic characteristics and flight dynamics of boomerangs. This represents a substantial advancement of boomerang study. The new theory leads to some significant findings regarding the radius of a boomerang's circular flight path, the required inclination angle of its axis-of-rotation, its trim state, as well as its dynamic stability. Further,a lower bound on tip-speed-ratio exists (χ0.707) for a boomerang to trim statically. For a boomerang to be dynamically stable, its lift at no angle-of-attack must be positive. These discoveries provide a basic understanding of how the interplay between aerodynamic forces and moments, and gyroscopic precession combine to return the boomerang to its rightful owner by way of a circular flight path.
A V-shaped boomerang design has been used as a case study for detailed analyses. Unsteady Reynolds-averaged Navier-Stokes solutions provide accurate three-dimensional aerodynamic characteristics of the subject boomerang. These results are compared with BET predictions which illustrate that BET captures the essence of CFD results, but not exactly their full details. A high-fidelity CFD-6DOF simulation of a boomerang flight has been performed. Although preliminary, this result verifies that BET, coupled with gyroscope equations of motion, can accurately predict the radius and other characteristics of a boomerang's flight path.
Performing accurate BET-6DOF flight simulations should also be possible as BET captures the essential features of cyclic forces and moments, however, with the caveat that its cyclic pitching-moment integrates to zero. Nonetheless, these simulations can be performed faster than real time whereas CFD-6DOF cannot. Future work by the author will include developing BET-6DOF numerical flight simulation technology to study boomerang trajectories.
The theory developed in this body of work provides a formal methodology for customizing future boomerang designs. With knowledge of this theory, it is clear in retrospect that the case-study boomerang developed herein should be redesigned with the following features. Reducing thickness of its airfoil sections will reduce mass, flow separation and drag. Increasing curvature immediately at the leading and trailing edges will better facilitate the Kutta condition, whether the flow is reversed or not. Less flow separation and improved Kutta condition will effectively increase lift at no angle-of-attack. These will all combine to reduce the radius of the flight path which is currently much too large for practical use.

XIV. Acknowledgements

The author thanks Neal Harrison, Dylan Vassberg and Jessee Ramirez for their help in this endeavor. Neal volunteered significant personal time for CAD support and CFD processing. Jessee provided high-definition camera gear and personal time to film our boomerang flight tests. Dylan converted many takes of raw video into polished movie clips; he strategically edited out his bloopers, yet methodically retained those of the author.
The author also acknowledges and thanks the proof readers of this paper: Blaine Rawdon, Andrew Killeen, Antony Jameson, Neal Harrison, Anthony Sclafani, Brian Vassberg, Dylan Vassberg and Claire Vassberg.

References

1 R. D. Lorenz. Spinning Flight. Springer,New York,2006.
2 J. B. Mauro. An Introduction to Boomerangs. Smith &Flannery,Inc.,Richmond,VA,1989. Third Edition.
3 M. J. Hanson. The Boomerang Book. Puffin Books,Harmondsworth,Middlesex,England,1974.
4 Unknown. Boomerang. http://en.wikipedia.org/wiki/Boomerang,June 2012.
5 W. E. Wiesel. Spaceflight Dynamics. McGraw-Hill,New York,1997. Second Edition.
6 Unknown. light-up-boomerang.jpg. http://www.geekologie.com/2007/09/,June 2012.
7 F. Hess. The Aerodynamics of Boomerangs. Scientific American,219:124-136,September 1968.
8 P. G. Buning,D. C. Jespersen,T. H. Pulliam,W. M. Chan,J. P. Slotnick,S. E. Krist,and K. J. Renze. OVERFLOW user's manual, version 1.81. NASA Report, NASA Langley Research Center, Hampton, VA, 1999.
9 J. C. Vassberg. Multi-block mesh extrusion driven by a globally elliptic system. Conference Proceedings, 5th  U.S. National Congress on Computational Mechanics, 2nd Symposium on Trends in Unstructured Mesh Generation,Boulder,CO,August
10 J. C. Vassberg. Further study of globally elliptic meshing. Conference Proceedings, 7th  International Conference on Numerical Grid Generation in Computational Field Simulations, Whistler, British Columbia, Canada, September 2000.
Figure 1. Aboriginal Boomerangs.
Figure 2. Modern-Day Boomerangs.
Figure 3. Classical Euler Angles: ϕ,θ,ψ .
Figure 4. Rotation Sequence to Tranform Boomerang to (ϕ,θ,ψ)=(45,45,45) .
Figure 5. Boomerang Strobed Night Flight.
Figure 6. Hypo-Cycloid Analysis of Boomerang Night Flight; χ=0.85 .
Figure 7. Blade Element Theory Schematic.
Figure 8. Vn Contours in the Rotation Plane; χ=0.7 .
Figure 9. Vn Contours in the Rotation Plane; χ=1.0 .
Figure 10. Vn Contours in the Rotation Plane; χ=1.5 .
Figure 12. Boomerang Airfoil Geometry, Camber & Thickness Distributions.
Figure 13. Boomerang Airfoil Grid; (257x97) O-Mesh.
Figure 14. Boomerang Airfoil Field Cp-Contours and Separation Bubbles; Mach=0.0774 , Ren=43,572 , α=0 .
Boomerang Airfoil
Lift Curve
Figure 15. Boomerang Airfoil Lift Curve; M=0.0774,Re=43,572 .
Figure 16. Boomerang Airfoil Drag Polar; M=0.0774,Re=43,572 .
Figure 17. Boomerang Near-Field Grid with ILE-Plane, ITE-Plane & Several J-Planes.
Figure 18. Boomerang Surface Grid; K=1 Plane.
Figure 19. Boomerang Field Grid; K=25/49 Plane.
Figure 20. Boomerang Field Grid; K=29/49 Plane.
Figure 21. Boomerang Field Grid; K=33/49 Plane.
Figure 22. Boomerang Field Grid; K=37/49 Plane; 5-Crefs Out.
Figure 23. Boomerang Far-Field Boundary; K=49/49 Plane; 100-Crefs Out.
Boomerang Cyclic Lift
Figure 24. Boomerang Phase-Aligned Cyclic Lift; Mach=0.0774,Re=43,572,χ=1.0,α=0 .
Boomerang Cyclic Rolling-Moment
Figure 25. Boomerang Phase-Aligned Cyclic Rolling-Moment; Mach=0.0774,Re=43,572,χ=1.0,α=0 .
Figure 26. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=0 .
Figure 27. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=30 .
Figure 28. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=60 .
Figure 29. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=90 .
Figure 30. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=120 .
Figure 31. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=150 .
Figure 32. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=180 .
Figure 33. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=210 .
Figure 34. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=240 .
Figure 35. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=270 .
Figure 36. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=300 .
Figure 37. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=330 .
Figure 38. Boomerang Upper-Surface Contours; Mach=0.0774,Re=43,572,χ=1.0,α=0,ψ=360 .
Figure 39. Boomerang Trim-Stability Curve; Mach=0.0774,Re=43,572,χ=1.0 ,Coarse Mesh.
Figure 40. Boomerang Time-Averaged Lift Curve; Mach=0.0774,Re=43,572,χ=1.0 ,Coarse Mesh.
Figure 41. Boomerang Time-Averaged Rolling-Moment Curve; Mach=0.0774,Re=43,572,χ=1.0 , Coarse Mesh.
Figure 42. Preliminary Boomerang Flight Path in a 2,000-inch Box; V0=60mph,χ=1.0 ,Coarse Mesh.
Figure 43. Stereoscopic-View of Preliminary Boomerang Flight Path.