Golf Ball Flight Dynamics

James Barber III Cornell University A&EP 434 Final Project Professor Lovelace

Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . Terminology . . . . . . . . . . . . . . . . . . . . . . . . . Idealizing a Golf Ball: Playing with a Smooth Sphere . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . The Model . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Eects of Dimples . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . The Eect of an Additional Lift Term . . . . . . . . . . The Eect of Reduced Drag . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A - Constants Used for Numerical Calculations . . Appendix B - Mathematica Code used to Model Ball Flights Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 3 3 5 5 5 6 7 10 11 12 14

James Barber III

Golf Ball Flight Dynamics

Introduction
Although poorly documented, golf is believed to have originated in Scotland in 1456[1]. It was rst played as a very casual game for which no standard rules existed. A wooden ball was used in conjunction with wooden clubs prior to 1618[2], when the featherie (a ball made of stitched leather and tightly packed with feathers) was introduced. The featherie was favored for its more forgiving feel on the hands of players when it was struck and was used until 1848 when the invention of the Gutta surpassed the featherie in both durability and cost. The Gutta was made of gutta-percha packing material which was not brittle and became soft and moldable at 100C. The Guttas pliability made it necessary to roll the ball on a smoothing board in order to maintain its shape and keep it free of imperfections which were created during normal play of the game. The smooth Gutta was used for only a few years before players began to realize that balls that had not been well maintained and had many nicks and scratches had a much more favorable ight. Thus began the practice of hammering the Gutta with a sharp-edged hammer in a regular pattern to increase the consistency of the balls play. In 1898 the rst Balata ball was created by wrapping rubber thread around a solid rubber core which was then covered by a solid layer of rubber that later became known as the ball cover. The Balata was the rst sign of a modern age of golf technology for it allowed molds to be used to create consistent cover patterns. In 1908 makers discovered the superiority of a regular dimple pattern over the haphazard grid pattern favored by players at the time. Dimples are small indentations on the exterior of the golf ball. They are typically round in shape and vary in diameter from 2-5mm in diameter and are about .2mm deep. Modern golf balls pack anywhere from 300-450 dimples of varying size arranged in a regular pattern on the outside of every ball[3]. Dimples have been one of the most inuential developments in golf ball design because they alter the dynamics of the balls ight in such a way that gives golfers a signicant amount of control over the height and shape of their shots. Today, golf balls are highly regulated by the United States Golf Association (USGA) for compliance with rules governing the design and capabilities of a golf ball. Modern materials, such as Surlyn and Urethane as well is dierent core designs, have given ball designers the ability to create golf balls with almost any property desired (higher spin, softer feel at impact, lower trajectory, etc...). It is well known that due to the complexities involved with the aerodynamics of a golf ball, ball designers have taken a design, test, and modify approach to ball design as an alternative to computer modeling and CAD techniques. In this paper I investigate a recent mathematical derivation for the aerodynamics of a smooth sphere and attempt to extend it to golf balls by incorporating the eects of dimples on air ow and comparing the results of simulations to observations taken from modern professional golfers. Although this technique will hopefully one day lead to the derivation of equations governing the ight of balls with arbitrary dimple patterns in pursuit of optimizing dimple design for maximum distance, the scope of this paper is limited to comparing two key eects that dimples have on ball ight.

James Barber III

Golf Ball Flight Dynamics

Terminology
Golfers use a signicant vocabulary that may not be recognized by those unfamiliar with the game. This section gives an introduction to basic terminology that will be used throughout this document. Loft - The angle between the club face and the shaft. The more loft a club has, the higher it will launch the ball at impact.

Figure 1: Loft[4]

Impact - The instant in time when a player strikes the ball with the club. Ball ight - The path a ball takes after impact, while it is in the air. There are also two broad types of shot shapes: a hook and a slice, and two directional descriptors: a pull and a push. A hook is a ball ight in which the ball curves from right to left due to a small amount of sidespin being imparted to it at impact. Similarly, a slice is a ball ight in which the ball curves from left to right, due to sidespin imparted to the ball at impact in the opposite direction from the hook result. A pull is when the ball starts its path to the left of its intended destination. Likewise, push is when the ball starts its path to the right of its intended destination. Please note that each of these terms is dened for a player using right-handed clubs, and changes its meaning if the player uses left-handed clubs.

Figure 2: Ball Flights[5]

Carry - The distance a ball travels in the air, after being struck. Note that this excludes the distance that the ball bounces and rolls after it rst strikes the ground (this is called roll ).

James Barber III

Golf Ball Flight Dynamics

Idealizing a Golf Ball: Playing with a Smooth Sphere

Introduction
A golf ball is a sphere about 4.2cm in diameter, with hundreds of small niches (dimples) carved in its outer face. While the dimples play a key role in the ight dynamics of the ball, I refrain from covering their eects until the next section. In this section we use a complex mathematical model derived and published by Karl I. Borg[6] in 2003 to model the ight of a golf ball under conditions typical for a professional golfer when striking the ball with a driver (see table 1). While this model is intended for use at low Reynolds Numbers, it has been experimentally shown to be reasonable as well for higher Reynolds Numbers. The Reynolds Number is a quantity that indicates the relative importance of inertial forces compared to viscous forces in determining its path through a gas or uid. High Reynolds Numbers indicate that viscous forces are less important. Re = dV (1)

Equation (1) is used to compute the Reynolds Number (Re), where d is the diameter of the object, V is the objects velocity, and is kinematic viscosity of the uid or gas, given by (2), = (2)

where is the Absolute Viscosity of the uid or gas (xed value for a given compound) and is its density. For a golf ball ying through 68 F air at sea level, the associated Reynolds Number falls in the range of 2.15x105 to 9.0x104 , which means that viscosity is much less important than the balls inertia.

The Model
Borgs model incorporates drag, heat exchange between the gas and the sphere (ball), and the Magnus Eect. The Magnus Eect (occasionally referred to as the Robins Eect for spheres) is a lift force that results from the rotation of a cylinder or sphere as it moves through a uid or gas and was rst described by German physicist Heinrich Magnus in 1853[7]. The lift force occurs as a result of spin creating a region of lower pressure above a ball with backspin (due to Bernoullis Principle). The net force on the sphere as derived by Borg et al is given by (3) F = pR2 12 m 2 1 (z) (z)v R3 mn v R3 mn ( v) 2kB T 3 3 || (3)

where is the accommodation coecient of tangential momentum (fraction of reected gas particles that are reected diusely), R is the radius of the sphere, m is mass of a single molecule of the gas, kT is Boltzmanns Constant, T is the temperature of the gas, is the angular velocity of spheres rotation, v is the velocity of the sphere, and p, , z, and are given by equations (4) through (8)

James Barber III

Golf Ball Flight Dynamics

p = nkB T = nkB T kB T 2m ||R2 k

(4) (5) (6) (7)

1i B= 2 z=

j1 (B) j1 (B) + Bj1 (B)

= 1 (z) + 4

2kB T (z) m 4||R

(8)

where n is the number density of the gas, is the heat conductivity of the sphere, j1 are the spherical Bessel functions of the rst kind (with j1 its derivative), i = 1, and k is given by k= Cp (9)

where Cp is the heat capacity of the sphere and is the density of the sphere. Values used in the simulation for each of these constants can be found in appendix A. Many of the parameters required by this model are properties of materials whose values are readily available, while others are determined based on launch conditions. For purposes of comparison and the availability of data, launch conditions for tour professionals were used where required - launch conditions for various top tour professionals are given in table 1
Player Vijay Singh Robert Allenby Peter Lonard Phil Mickelson Ernie Els Test Conditions V (MPH) 176 161 168 178 174 170 V (m/s) 78.68 71.97 75.10 79.57 77.78 76.00 Angle (deg) 10.7 8.5 11.7 13.0 11.5 12.0 Spin (rpm) 2,600 2,390 2,673 2,200 2,400 2,626 Spin (sec1 ) 272.3 250.3 279.9 230.4 251.3 275.0 Carry (m) 282.46 274.59 268.38 281.64 292.24 ????

Table 1: Lanch conditions and statistics for top PGA Tour Professionals[8]

James Barber III

Golf Ball Flight Dynamics

Results
Although the amount of spin varies signicantly with the skill of the player and club selection, this paper assumes as valid data taken from current tour professionals for modeling purposes. A tour pro generates about 12,000 RPM of backspin at impact when hitting with a sand wedge (loft 56) and anywhere from 2,200-2,800 RPM with a driver (loft 10). Using this information and values given in appendix A, a recursive program (see appendix B for code) was utilized to model the trajectory of a smooth golf ball in ight.

Figure 3: Simulated ight of a smooth sphere using golf tour professional launch conditions and a mathematical derivation of ight forces presented by Borg[6] While the shape of this curve appears reasonable, anyone familiar with golf will immediately notice the short carry distance predicted in this simulation, as well as the low peak height obtained. From table (1) we could estimate that a modern golf ball launched under similar conditions would carry about 280m and reach an altitude of about 30m. These large dierences between the model and observation highlight the critical role of dimples in determining golf ball trajectory.

The Eects of Dimples

Introduction
As was shown in the previous section that an idealized sphere could serve as an adequate recreational ball, in golf it is desirable to be able to make the ball y further and allow golfers to have more control over their shot shape (hook or slice) and trajectory. Dimples (or their predecessors) have served this dual purpose in golf for over 100 years. While it is undisputed that dimples have had a dramatic eect on ball ight for two key reasons - lift generation and drag reduction - little progress has been made in determining which eect is more signicant. In this section I attempt to show, using a simplied model of each of these eects, that drag reduction from the addition of dimples has made a signicantly larger contribution to the ight of a golf ball than the addition of a lift force.

James Barber III

Golf Ball Flight Dynamics

The Eect of an Additional Lift Term

Dimples generate a lift force on the golf ball because of the signicant amount of backspin imparted to the ball at impact. This spin creates an asymmetry in the speeds at which air ows over the dimples in the top and bottom of the ball. Assuming the ball is traveling at velocity V0 and the ball is spinning with angular velocity , the air speeds relative to the top and bottom of the ball are given by (10) Vtop = V0 R Vbottom = V0 + R (10)

where R is the radius of ball. Since, in air, lift is proportional to velocity squared[9], this gives a net lift on the ball from the dimples of
2 2 Lif t Vbottom Vtop = (V0 + R)2 (V0 R)2 = 4V0 R V0 R

(11)

Using this relation and estimates of the lift coecient (CL ) in the range of 1.5x105 to 3.0x105 kg/m, the results in gure (4) were obtained when this term was incorporated into the existing model.

Figure 4: Simulated ball ight with the addition of a lift term When this result is compared to the ball ight in the previous section without the lift term, it is easy to see that the lift term did little to increase the distance that the ball traveled. It should be noted, however, that the lift term did make two important changes to the result: 1) the ball traveled higher than without this term (9m at its peak), and 2) the ball stayed aloft for a signicantly longer period of time (4.9 seconds vs. 3.1 seconds without lift). The shape of this trajectory is similar to that of a golf ball hit with more than an optimal amount of backspin - this shape is known as a blow up in golf (gure 5). This simulation showed several encouraging signs that seemed to indicate that the approximate lift force used gave realistic results.

James Barber III

Golf Ball Flight Dynamics

Figure 5: Observed golf ball trajectories based on spin rate at launch[10]

The Eect of Reduced Drag

After investigating the aects of the addition of lift and observing the results of its application to the model of a smooth sphere in ight it is clear that drag is still a signicant factor aecting the distance that the ball carries. Dimples have a counter-intuitive eect on drag, although they roughen the surface of the ball, this adds turbulence to the boundary layer, which reduces drag by delaying boundary layer separation on the trailing side of the ball. This delayed separation of a turbulent boundary layer results in a narrower vortex street, decreasing the drag force that vortex formation exerts on the ball. In an attempt to investigate the eects of reduced drag we make use of Whites model for drag on a sphere[10]. This simplied model neglects the eect of the balls spin in its ight. The model can be summarized by equations (12) and (13) Cd = 8 Re 0.5 + log10
8 Re

(12)

Fdrag = Cd RV 2 =

4V 0.5 + log10
8 Re

(13)

where is the kinematic viscosity of air, V is the velocity of the ball, is the Euler-Mascheroni Constant (.5772), R is the radius of the ball, Re is the Reynolds Number, and Cd is the drag coecient. This model is intended to incorporate important drag considerations that the addition of dimples requires. The most important of these considerations is the alteration of the drag curve imposed by the turbulent ow created by the dimples.

James Barber III

Golf Ball Flight Dynamics

Figure 6: Flow modications imposed by the addition of dimples[11] This lowered Reynolds Number required for the transition to turbulent ow (steep drop in drag coecient) is necessary because even professionals are only able to give the golf ball an initial velocity high enough to create a Reynolds Number of about 4x105 , which would not quite reach the value required for the large drop in drag coecient for a smooth sphere. With the dimples, however, a Reynolds Number of only 5x104 is necessary to achieve a signicant drop in drag coecient; this value can be reached by a golf ball moving at just 8.9 m/s - a speed above which the ball travels for the entirety of its ight, thereby gaining maximum possible gains from the modied drag curve.

Figure 7: Simulated ball ight for a sphere with reduced drag Several observations can be made from the results of the simulation using the simplied, reduced drag model. First, the shape of the ight is much more ideal because the model neglects the eects of spin on drag, and also excludes lift forces due to spin. The most dramatic dierence to notice in this simulation is large increase in carry (over 150m). This is most likely due to a higher velocity being maintained throughout the balls ight, which allows the ball to travel much further in the horizontal direction in the same amount of time. It is also important to note that the maximum height achieved is 60% (5 meters) higher than that of the smooth ball. While this is still quite far from the heights typical of actual golf balls (30m), it is not dicult to see the aggregate eects of the reduced drag and increased lift combining to yield a signicantly higher peak height.

James Barber III

Golf Ball Flight Dynamics

Combining the Eects of Additional Lift and Reduced Drag

Figure 8: Simulated ball ight for a sphere with reduced drag and an additional lift term As a nal exhibition, gure (8) represents a simulated golf ball ight by using both the reduced drag force and the addition of a lift term. These two very simplistic force models combined harmoniously to produce a simulated ball ight with parameters that very nearly match observed values. An unfortunate by-product of using the simplied drag model is loss of shape in the balls path. This eect is noticeable when comparing the simplied drag model results (gures 7 & 8) to results from using Borgs full drag equation (gures 3 & 4). Both the angle of impact as well as the speed of the ball at landing should be subject to scrutiny in this model. While the ball is moving at about 37m/s just before it hits the ground in this model, observation has the ball impacting the ground at just under 32m/s. It should be noted that this is a signicant improvement (in terms of modeling a golf ball) over Borgs model which predicts a smooth sphere to hit the ground at around 10m/s.

James Barber III

Golf Ball Flight Dynamics

10

Conclusion
While it was clear that neither increasing the lift nor reducing the drag on the ball are enough alone to change the trajectory to match what is observed in actuality, it is not dicult to see that the reduction in drag has a much larger eect on the trajectory of the ball. While Borgs derivation of the forces on a smooth spinning sphere in ight is quite impressive and proves very successful in modeling the ight of a smooth sphere, a signicant amount of work remains to incorporate the eects of dimples on the balls ight to a level where the model could be of use to ball designers. This appears to be a clear next step in the technological advancement of the golf industry. While practical methods and designs have brought signicant advances in ball technology, computers and modern aerodynamic theory should allow for true optimization of golf ball design.

James Barber III

Golf Ball Flight Dynamics

11

Appendix A - Constants Used for Numerical Calculations

= accommodation coecient of tangential momentum = R = radius of the sphere (golf ball) = 2.13cm m = mass of a single molecule of the gas (air, average) = 28.98 amu = 4.812 x 1026 kg kB = Boltzmanns Constant = 1.38 x 1023 J K T = temperature of the gas (air) = 20 C = 293.15 K = angular velocity of spheres rotation = 76
rad sec 360 40 360

8 9

(at launch)
15

m v = velocity of the sphere (golf ball) = 76 sec cos

, sin

15

, 0 (at launch)

Launch angle = 12 =

15

radians

n = number density of the gas (air) = 2.503 x 1025 molecules m3

J = heat conductivity of the sphere (assumed to be made of hard rubber) = 0.4 msec K cal Cp = heat capacity of the sphere = 124 K = 519.16 J K

= density of the gas (air) at 20 C and 1 atm = 1.2047

kg m3

kg = Absolute Viscosity of the gas (air) = 1.82 x 105 msec

James Barber III

Golf Ball Flight Dynamics

12

Appendix B - Mathematica Code used to Model Ball Flights

>with(LinearAlgebra): deltaT:=.01; M:=.04593; BallPosition:=<0,0,0>; GravForce:=<0,-9.8*M,0>; > with(LinearAlgebra): B:=(1-I)*sqrt(VectorNorm(omega, 2)*(R^)/(2*k)); 2 chi:=(alphaT*n*KB*T/kappa)*sqrt(KB*T/(2*Pi*m)); k:=kappa/(rho*Cp); z:=BesselJ(1,B)/(chi*BesselJ(1,B)+B*(BesselJ(0,B)-BesselJ(1,B)/B)); p:=n*KB*T; xi:=1-(chi/4)*(Re(z)+sqrt(2*Pi*KB*T/m)*Im(z)/(4*VectorNorm(omega, 2)*R)); > refreshV := proc(Fnet) global V, deltaT, M: V:=V+ScalarMultiply(Fnet, deltaT/M): end proc: > refreshBallPos := proc() global deltaT, V, BallPosition: BallPosition:=BallPosition+ScalarMultiply(V, deltaT): end proc: > computeFnet := proc() global V, omega, GravForce, B, chi, z, xi, p, k, Force, DragForce, NetForce, LiftForce, Clift, R, CD, nu: R:=.04267/2: nu:=.000015: with(LinearAlgebra); ^ Force:=-alphaT*(Pi/12*p*R2*sqrt(Pi*m/(2*KB*T))*chi*Re(z)*V-2/3*xi*Pi*R^*m*n* 3 ^*m*n*chi*Im(z)/VectorNorm(omega, 2)* CrossProduct(omega, V)-Pi/3*R3 CrossProduct(omega, CrossProduct(omega, V))): DragForce:=eval(Force, [alphaT=280/360, n=2.503e25, KB=1.38e-23, T=293.15, R=.04267/2, m=4.812e-26, kappa=.4, rho=1.1347e6, Cp=519.16]): #The following lines alter the drag force to a less precise model #CD:=4*Pi*nu/(.5-.577215665+log10(4*nu/(R*VectorNorm(V, 2)))): #DragForce:=ScalarMultiply(V, -CD): #The following lines alter the drag force to the most basic model (CD=const.) #CD:=.29; #DragForce:=ScalarMultiply(V, -CD*R):

LiftForce:=<0,0,0>:

James Barber III

Golf Ball Flight Dynamics

13

# The following adds a lift term to the forces #Clift:=.000015; #LiftForce:=Clift*CrossProduct(omega, V): NetForce:=evalf(DragForce+GravForce+LiftForce): end proc:

> # this subroutine updates the amount of spin on the ball (Cds is an arbitrary Cefficient) refreshSpin := proc() global omega, Cds, M, omegaI, deltaT: Cds:=-24*M/(35*omegaI): omega:=omega+ScalarMultiply(omega, deltaT*Cds*VectorNorm(omega, 2)): end proc: > # Main simulation routine VI:=76; omegaI:=275; V:=<VI*cos(2*Pi*12/360), VI*sin(2*Pi*12/360), 0>: StrikeAngle:=0; #defines a non-square hit that alters the spin axis, causing a hook omega:=<0,omegaI*sin(StrikeAngle),omegaI*cos(StrikeAngle)>: BallPosition:=<0,0,0>: MaxInt:=1000; with(LinearAlgebra): Ball:=[[0,0], [0,0]]: for i from 1 by 1 to MaxInt do refreshBallPos(): Ball:=[op(Ball), [evalf(BallPosition[1]), evalf(BallPosition[2])]]; computeFnet(): refreshV(NetForce): #refreshSpin(): # Stop calculating the ball path if it has gone below the ground if ( evalf(BallPosition[2]) < 0 ) then break end if: end do: FlightTime:=i*deltaT; Carry:=evalf(BallPosition[1]); Vfinal:=evalf(V); plot(Ball,style=LINE);

Bibliography
[1] Golf. Encyclopedia Britannica. 2007. Encyclopedia Britannica Online. http://secure.britannica.com/eb/article-222218/golf. Accessed May 13, 2007. Available:

[2] History of Golf Balls. Online. Available: http://www.golfjoy.com/golf physics/history.asp. Accessed 5/13/07. [3] Golf Ball. Online. Available: http://en.wikipedia.org/wiki/Golf balls. Accessed 5/13/07. [4] Original image obtained from: cessed 5/13/07. http://www.angelre.com/pa/TWGOLF/tinfo.html. Ac-

[5] Original image obtained from: http://golfdirt.com/cgi-bin/f.cgi?url=http://news.bbc.co.uk/sport2/hi/golf/skills/4245518.stm. Accessed 5/13/07. [6] Borg, Karl et al. Forces on a spinning sphere moving in a rareed gas. Physics of Fluids Volume 15, Number 3 (March 2003): 736-41. [7] Tritton, D.J. Physical Fluid Dynamics. Second Edition. Oxford: Clarendon Press, 1988. pp159-161. [8] Special Report: The Growing Gap. Online. Available: http://www.golfdigest.com/equipment/index.ssf?/equipment/gd200305growinggap.html. Accessed: 5/9/07. [9] Cross, Rod. Ball Trajectories. University of Sydney. August 1, 2006. Online. Available: http://www.physics.usyd.edu.au/ross/TRAJECTORIES/Trajectories.pdf. Accessed: c 5/11/07. [10] White, Frank M. Viscous Flow. First edition. McGraw-Hill. Eqn. 2-265. [11] Flight Dynamics of Golf Balls. Online. Available: http://www.golfjoy.com/golf physics/dynamics.asp. Accessed: 5/11/07.