Jess Worsley WRSJES002

MAM 2OD Project

June 2017
1 Overview

In the following article we will be analysing the physics of bungee jumping through an applied
maths perspective, using ordinary differential equations (ODEs) to predict the trajectory of
the jumper and determine the effect of the cord’s spring constant.

In this situation the cords’ lengths are all 100 ft, so they will only begin to extend beyond
their natural length after they have travelled that distance. We have made this point to be
x = 0 in our system, and have applied these coordinates to Figure 1.

For the first 100 ft of the fall, the only forces acting upon the jumper are the gravitational
force mg and the air resistance −βv, since the cord is falling freely with the jumper. Air
resistance increases as the velocity of the jumper increases and always acts in the opposite
direction to the velocity. The constant β varies according to how aerodynamic the falling
body is, and in our example is made equal to 1.0 (Cooper, 2013).

Since we are treating the bungee cord as a massless Hookean spring, we can make the as-
sumption that the cord will exert an upwards force, −kx, on the jumper which is proportional
to the spring constant and the displacement beyond its natural length. The spring constant
depends on the stiffness, thickness and length of the cord.

0 x≤0
F (x) = (1)
−kx x > 0

From Equation 1 we can see that the spring force only comes into effect when x is positive.
It is this upward force overcoming the downward force of gravity that creates the bouncing
motion of the bungee jumper. The bouncing motion quickly subsides due to the air resistance,
which, characteristic of frictional forces, withdraws energy from the system. It is easy to see
after graphing the values that this represents a damped harmonic oscillator.

Since we are using Dr Kevin Cooper’s setup, we have kept his imperial units as our goal is
to correctly analyse ODEs rather than worry too much about constants. We have a bungee
jumper with a height of 6.0 ft and a weight of 160 lbs. The distance from the bridge to the
water is 174 ft, meaning that we are looking for a cord that stops within 68 ft past the point
at which it begins to extend. Remember that the jumper is 6 ft tall and we need to take that
into account. It is also important that it stops gently.

At the end of this document is a small piece on ’localising the setup’, in which we convert
values to SI units as well as finding values that better represent the South African experience.

1
 Figure 1: Diagram showing bungee system

2 Calculations

2.1

The following equation describes the motion of the bungee jumper for x ≤ 0, before passing
the point at which the cord is fully extended:

mẍ + β ẋ = mg (2)

where x(0) = −100 and ẋ(0) = 0. Using reduction of order and variation of parameters we
can solve the equation for x(t).

Letting ω = ẋ and ω̇ = ẍ
d2 x dx
m 2
+β − mg = 0
dt dt
β
ω̇ + ω = g (3)
m

dω β
=− ω
Z dt Zm
1 β
dω = − dt
ω m
β
ln ω = − t + ln k
m

2
 β β
ω = ke− m t = k(t)e− m t (4)
β β β
ω̇ = k 0 (t)e− m t − k(t)e− m t (5)
m

Substituting equations (3) and (4) into (2) gives:

β β β β β
k 0 (t)e− m t − k(t)e− m t + k(t)e− m t = g
m m
β
0 −m t
∴ k (t)e =g
g β
∴ k 0 (t) = β = ge m t
e− m t
m β
k(t) = g e m t + C
β
After integrating we obtain an expression for k(t) which we can substitute into (3) to solve
for ω, thereby solving for ẋ.
m β
ω=g + Ce− m t = ẋ
β
To find a value for C we use the initial condition ẋ(0) = 0.
m
ẋ(0) = C + g =0
β
m
∴ C = −g
β
m −βt m
∴ ẋ = −g e m +g (6)
β β
Integrating equation (5) provides an expression for x(t).
Z Z
m β
−m t m
x(t) = −g e dt + g dt
β β
m2 β m
x(t) = g 2 e− m t + g t + D
β β
We now use the initial condition x(0) = −100 to calculate the constant of integration, D.

m2
x(0) = g + D = −100
β2
m2
∴ D = −g 2 − 100
β

m2 − β t m m2
∴ x(t) = g e m + g t − g − 100 (7)
β2 β β2
Having obtained an expression for x(t), we can substitute the values for g, m and β.

g = 32 f t.s−2

3
 m = 5 lbs

β=1

Which gives us
t
x(t) = 800e− 5 + 160t − 900 (8)

2.2

We can now use the equation found in the previous section to approximate the length of time,
t1 , that the jumper freefalls i.e. the time taken to fall the unextended length of the cord, 100
ft. At this point in the fall, x = 0 (See Figure 1). We can therefore say that:
t
800e− 5 + 160t − 900 = 0

In order to approximate t1 we use Newton’s Method:

x(tn )
tn+1 = tn − (9)
x0 (tn )
t
800e− 5 + 160t − 900
tn+1 = t0 − t (10)
−160e− 5 + 160
In order to calculate the best approximation to at least six decimal places, we use the following
code in Python:
import numpy a s np

def x ( t ) :
return 800∗ np . exp(− t /5)+160∗ t −900

def dx ( t ) :
return −160∗np . exp(− t /5)+160

def dt ( x , t ) :
return abs(0−x ( t ) )

def newtons method ( x , dx , t0 , n , p r i n t r e s=True ) :

d e l t a = dt ( x , t 0 )
while d e l t a > n :
t 0 = t 0 − x ( t 0 ) / dx ( t 0 )
d e l t a = dt ( x , t 0 )
if print res :
print ’ Root i s a t : ’ , t 0
print ’ x ( t ) a t r o o t i s : ’ , x ( t 0 )
return t 0

4
 if name == ’ m a i n ’ :
t 0 s= [ 0 , 0 . 5 , 2 . 5 , 5 ]

f o r t 0 in t 0 s :
newtons method ( x , dx , t0 , 1 e −10 , True )

The console output is:

Root is at: nan

x(t) at root is: nan
Root is at: 2.72688233348
x(t) at root is: 0.0
Root is at: 2.72688233348
x(t) at root is: 0.0
Root is at: 2.72688233348
x(t) at root is: 0.0

Therefore t1 ≈ 2.7269 seconds.

2.3

Now that we have the expressions for x(t) and t1 , we can calculate x0 (t1 ), which gives us the
downward velocity at x = 0.
t
x(t) = 800e− 5 + 160t − 900
t
∴ x0 (t) = −160e− 5 + 160
t1
x0 (t1 ) = −160e− 5 + 160
∴ x0 (t1 ) = 67.2602346714 = v(t1 )

Therefore the downward velocity at x = 0 is v1 ≈ 67.2602 f t.s−1 .

2.4

Having passed the point at which the cord is fully extended i.e. x > 0, we must take the
spring force into account, Fspring = −kx, and solve the following initial value problem:

mẍ + β ẍ + kx = mg (11)

Where x(t1 ) = 0 and x0 (t1 ) = v1 . This is a non-homogeneous equation, so we can start by

finding the solution to the complementary homogeneous linear differential equation:

mẍ + β ẍ + kx = 0 (12)

5
To do this we estimate a solution of the form eλt and substitute it into Equation 11.
 2   
d λt d λt
m e + β e + keλt = 0
dt2 dt
mλ2 eλt + βλeλt + keλt = 0
mλ2 + βλ + k = 0

We then substitute into the quadratic equation.

p
β 2 − 4mk −β ±
λ=
p p 2m
β β 2 − 4mk β |β 2 − 4mk|i
∴ λ1 = − + =− +
2m p 2m 2m p 2m
β 2
β − 4mk β |β 2 − 4mk|i
λ2 = − − =− −
2m 2m 2m 2m
Solving for the roots gives two complex conjugate roots, since β 2 − 4mk < 0. Coincidentally,
this is the case for an underdamped harmonic oscillator, which approaches zero quickly and
oscillates about y = 0 (Nave, 2016). We can see this later in Figures 2, 4 and 6. The roots
will give two linearly independent solutions x1 = eλ1 t and x2 = eλ2 t . Subsequently, due to
the General Principle of Superposition, we can expect a general solution of the form:

x(t) = C1 eλ1 t + C2 eλ2 t (13)

√
β |β 2 −4mk|
Let a = − 2m and b = 2m , for simplification.

∴ eλ1 t = e(a+bi)t = eat .ebit

eλ2 t = e(a−bi)t = eat .e−bit

We can now apply Euler’s Formula:

For any real number θ,

eθi = cos θ + i sin θ (14)

∴ eλ1 t = eat cos bt + i sin bt

eλ2 t = eat cos (−bt) + i sin (−bt)
= eat cos bt − i sin bt

We can now substitute these roots into Equation 12.

x(t) = C1 eat (cos bt + i sin bt) + C2 eat (cos bt − i sin bt) (15)

This expression still contains complex values which we need to remove in order for us to
express the general form for solutions with only real number coefficients. We can split our
general solution into real and imaginary parts and see whether they satisfy the equation.

6
First, we choose C1 = C2 = 12 , making Equation 14 equal to:

1 1
x(t) = eat (cos bt + i sin bt + eat (cos bt − i sin bt))
2 2
x(t) = eat cos bt

Then we choose C1 = − 12 i and C2 = 12 i:

1 1
x(t) = − ieat (cos bt + i sin bt + ieat (cos bt − i sin bt))
2 2
x(t) = −(−eat sin bt) = eat sin bt

Therefore both the real and imaginary parts below satisfy the differential equation.

x(t)[Re] = eat cos bt

x(t)[Im] = eat sin bt

In order to determine whether or not this pair forms a fundamental set of solutions, we must
confirm that the Wronskian does not equal zero.
 
 x(t)[Re] x(t)[Im] 
W [x(t)] =  0
x (t)[Re] x0 (t)[Im]
 

eat cos bt eat sin bt

 
 
=  at

at at at

ae cos bt − be sin bt ae sin bt + be cos bt

∴ W [x(t)] = (ae2at cos bt sin bt + be2at cos2 bt) − (ae2at cos bt sin bt − be2at sin2 bt)
= be2at (cos2 bt + sin2 bt) = be2at 6= 0

Since the Wronskian can never be zero, x(t)[Re] and x(t)[Im] are linearly independent solu-
tions of our equation. Therefore, our general, or complementary, solution is:

xc (t) = C1 eat cos bt + C2 eat sin bt (16)

Before we can derive our final answer, we must find the particular solution. Since g(t) in our
non-homogeneous equation
mẍ + β ẋ + kx = mg = g(t)
is a constant, we can expect our particular solution to be of the form

xp = C.

Therefore ẋp = ẍp = 0. Our particular solution then becomes simple to solve.

kxp = mg
mg
∴ xp (t) = (17)
k

7
Now, from Equations 15 and 16,

x(t) = xc (t) + xp (t) (18)

becomes
mg
x(t) = C1 eat cos bt + C2 eat sin bt + (19)
k
We substitute our initial values and rearrange to find C1 and C2 .

x(t1 ) = 0 and x0 (t1 ) = v1 .

C2 eat1 sin bt1

C1 = − = −C2 tan bt1
eat1 cos bt1
C2 (aeat1 sin bt1 + beat1 cos bt1 ) − C2 tan bt1 (aeat1 cos bt − beat1 sin bt) = v1
C2 [(aeat1 sin bt1 + beat1 cos bt1 ) − tan bt1 (aeat1 cos bt − beat1 sin bt)] = v1 = C2 γ
v1
∴ C2 =
γ
v1
C1 = − tan bt
γ
We can finally plug in our values:

g = 32 f t.s−2

m = 5 lbs

β=1

k = 14
√
β |β 2 −4mk|
Recall that a = − 2m and b = 2m , and that our original equation (constants included)
is
5ẍ + ẋ + 14x = 160. (20)
We obtain values for C1 and C2 ,

C1 = 52.2357118433
C2 = −8.30105586495,

which we can plug into Equation 18, giving us our final solution to the differential equation.
√ √
t
− 10 3 31 t
− 10 3 31
x(t) = 52.2357e cos t − 8.3011e sin t + 11.4286 (21)
10 10

8
2.5

In order to find the times when the velocity is zero after passing x = 0, we need to find the
derivative of Equation 20 and once again use Newton’s Method to approximate t2 . We need
to find the lowest value below x = 0 i.e. the most positive x(t) value, as the bungee bounces
up and down a few times before coming to rest.
√ √ √ !
0 1 −t 3 31 3 31 − t 3 31
x (t) = 52.2357 − e 10 cos t− e 10 sin t
10 10 10 10
√ √ √ ! (22)
1 −t 3 31 3 31 − t 3 31
−8.3011 − e 10 sin t+ e 10 cos t
10 10 10 10

We use Python to find the approximation to within at least six decimal places, and make
sure our t2 gives a reasonable output when substituted into Equation 21.
import numpy a s np

C1=52.2357118433
C2= −8.30105586495
a=−0.1
b=(np . s q r t ( 2 7 9 ) ) / 1 0

def x ( t ) :
return C1∗ ( a∗np . exp ( a∗ t ) ∗ np . c o s ( b∗ t )−b∗np . exp ( a∗ t ) ∗ np . s i n ( b∗ t ) )
+C2∗ ( a∗np . exp ( a∗ t ) ∗ np . s i n ( b∗ t )−b∗np . exp ( a∗ t ) ∗ np . c o s ( b∗ t ) )

def dx ( t ) :
return C1∗ ( a ∗ ∗ ( 2 ) ∗ np . exp ( a∗ t ) ∗ np . c o s ( b∗ t )−2∗a ∗b∗np . exp ( a∗ t ) ∗ np . s i n ( b∗ t )−b
+C2( a ∗ ∗ ( 2 ) ∗ np . exp ( a∗ t ) ∗ np . s i n ( b∗ t )+2∗a∗b∗np . exp ( a∗ t ) ∗ np . c o s ( b∗ t )−b ∗ ∗ ( 2 ) ∗ n

def dt ( x , t ) :
return abs(0−x ( t ) )

def newtons method ( x , dx , t0 , n , p r i n t r e s=True ) :

d e l t a = dt ( x , t 0 )
while d e l t a > n :
t 0 = t 0 − x ( t 0 ) / dx ( t 0 )
d e l t a = dt ( x , t 0 )
if print res :
print ’ Root i s a t : ’ , t 0
print ’ x ( t ) a t r o o t i s : ’ , x ( t 0 )
return t 0

if name == ’ m a i n ’:
t 0 s= [ 1 , 2 , 3 , 4 , 5 ]

9
 f o r t 0 in t 0 s :
newtons method ( x , dx , t0 , 1 e −10 , True )

The code output provided several values of t2 , which were substituted into Equation 20 to
give the lowest position reached. It was found that

t2 = 3.72584508586 ≈ 3.7258 seconds

and x(t2 ) = 47.265323744 ≈ 47.2653 feet.

In other words, the lowest position is approximately 147 feet below the bridge, or 27 feet
above the surface of the water.

2.6

Equation 20 is re-evaluated with three different spring constants in order to determine which
coil is the most suitable. Calculations were carried out in Python to find new constants
C1 and C2 , keeping parameters and initial conditions the same and only varying the spring
constants. Conclusions can be drawn from graphs of the results.

2.7

The trajectory for each spring constant is graphed over 80 seconds so we can see the point
at which it converges. Included are the graphs of position against time and velocity against
position for each cord, k = 8.5, k = 10.7 and k = 16.4. We can immediately see that the
cord with k = 8.5 is inappropriate. Its lowest position is at x ≈ 80 feet i.e. you end up in
the river.

The cord with k = 10.7 is likely the safest cord out of the three, though ideally the spring
constant would be slightly higher since the bungee jumper ends up a few feet in the water.
The jumper needs at least 68 ft of space due to his height and this cord provided only ∼ 72 ft.
It is important, however, to avoid large spring constants, as increasing the spring constant
increases the force in the opposite direction to motion as you travel past the point of no
extension, x = 0. As a result of this combination of factors, the spring constant of 14 which
we used earlier is the best bet.

Large values could lead to serious injury due to the sudden stopping force in the upward
direction. As such, the rope with spring constant k = 16.4 is likely too stiff for bungee
jumping. A typical human can withstand no more than 9 g’s for a few seconds (Cavelos,
1999), which is approximately 88.2 m.s−2 . This is for the whole body, so considering the
force is focussed on the ankle joint, the g-force should be even less. If the upwards acceleration
at the lowest position is anywhere near this value it can be considered unsafe. We also notice
from the velocity/position graphs that the lower the spring constant, the lower the height to
which the position converges. The jumper’s mass pulls the cord below its natural extension
and the spring constant determines how much it stretches.

10
 Figure 2: Graph showing the position/time results for the cord with k=8.5

Figure 3: Graph showing the velocity/position results for the cord with k=8.5

11
 Figure 4: Graph showing the position/time results for the cord with k=10.7

Figure 5: Graph showing the velocity/position results for the cord with k=10.7

12
 Figure 6: Graph showing the position/time results for the cord with k=16.4

Figure 7: Graph showing the velocity/position results for the cord with k=16.4

13
3 References

[1] Nave, C. R. Damped Harmonic Oscillator.

http://hyperphysics.phy-astr.gsu.edu/hbase/oscda.html. Department of Physics
and Astronomy, Georgia State University, GA 30302. 2016.

[2] Cooper, K. Bungee Jumping.

http://www.idea.wsu.edu/Bungee/. Department of Mathematics, Washington State
University, WA 99164-3113. 2013.

[3] Q&A: Force, work, energy in bungee jumping.

https://van.physics.illinois.edu/qa/listing.php?id=350.
Department of Physics, University of Illinois, IL 61801-3080. 2007.

[4] Tseng, Z. S. Second Order Linear Differential Equations.

http://www.math.psu.edu/tseng/class/Math251/Notes-2nd%20order%20ODE%20pt1.pdf.
Pennsylvania State University, PA 16801. 2008, 2016.

[5] Dawkins, P. Second Order Linear Differential Equations.

http://tutorial.math.lamar.edu/Classes/DE/NonhomogeneousDE.aspx.
Lamar University, TX 77705. 2017.

[6] Cavelos, J. The Science of Star Wars. 1st ed. New York: St. Martin’s Press. 1999.
https://www.scientificamerican.com/article/star-wars-science-light-speed/.

14