Engineering Failure Analysis 11 (2004) 857872

www.elsevier.com/locate/engfailanal

Analysis of a fatal bungee-jumping accident

D.R.H. Jones *

Department of Engineering, University of Cambridge, Trumpington Street, Cambridge CB2 1PZ, UK

Received 1 March 2004; accepted 15 March 2004
Available online 5 May 2004

Abstract

The sport of bungee jumping has become very popular worldwide over the past decade. The author was recently
commissioned to investigate a fatal accident, which occurred during a bungee jump. As a result of this investigation, it
has emerged that the equipment used for bungee jumping has evolved in an empirical way, using a mixture of braided
rubber cords originally intended for aeronautical applications, and ancillary equipment designed for climbing pro-
tection. Many sporting organisations and government agencies have established codes of practice for bungee jumping.
However, these codes are essentially empirical, and are not based on a quantitative materials engineering analysis of the
forces generated in the load train in relation to the strength of the components. The fatal accident is presented as a
detailed case study, in which the load/extension characteristics of the bungee rope and the end attachment webbing are
measured, and used as the critical inputs to an energy-based analysis of the complete jumping process. It is shown that
the bungee rope was unable to absorb all the potential energy of the falling jumper, with the result that the jumper
broke away from the bottom end of the rope. The paper also discusses the urgent need for a quantitative code for the
design and use of bungee jumping equipment, based on rigorous materials engineering analysis.
2004 Elsevier Ltd. All rights reserved.

Keywords: Bungee jumping; Rope failures; Webbing failures; Overload; Fitness for purpose

1. Introduction

The sport of bungee jumping has become very popular worldwide over the past decade. The author was
recently commissioned to investigate a fatal accident, which occurred during a bungee jump. As a result of
this investigation, it emerged that the equipment used for bungee jumping has evolved in an empirical way,
using a mixture of braided rubber cords originally intended for aeronautical applications [1], and ancillary
equipment designed for climbing protection. Many sporting organisations and government agencies have
established codes of practice for bungee jumping [24]. However, these codes are essentially empirical, and
are not based on a quantitative materials engineering analysis of the forces generated in the load train in
relation to the strength of the components. The fatal accident will be presented as a detailed case study, in

*
Tel.: +44-1223-332600/366136; fax: +44-1223-332662/366136.
E-mail address: drj1000@eng.cam.ac.uk (D.R.H. Jones).

1350-6307/$ - see front matter 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2004.03.002
858 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

which the load/extension characteristics of the bungee rope and the end attachment webbing are measured,
and used as the critical inputs to an energy-based analysis of the complete jumping process. It will be shown
that the bungee rope was unable to absorb all the potential energy of the falling jumper, with the result that
the jumper broke away from the bottom end of the rope. The paper also discusses the urgent need for a
quantitative code for the design and use of bungee jumping equipment, based on rigorous materials
engineering analysis.

2. Case study

2.1. Background

A bungee-jumping accident occurred in the UK in 2002, which resulted in the death of the jumper. The
following items were made available for the investigation: the bungee-jumping equipment used in the
accident jump, together with similar items of equipment as used in other jumps; videotapes, including
a videotape of the accident itself; photographs taken at the scene of the accident; and supporting
documentation.
The essential details of the accident were as follows. According to the pathology report, the jumper
weighed 132 kg and was 1.83 m high (see Fig. 1). He jumped from a crane-mounted cage, which had its
oor approximately 53 m above ground level. The bungee rope consisted of three nominally identical cords
used in parallel, and taped together with insulating tape at regular intervals. The inboard end of the rope
was secured to a pair of snap hooks mounted on the vertical centreline of the cage, and positioned ap-
proximately 1.35 m above the cage oor (see Fig. 1). The rope passed vertically down through a large
circular hole in the oor of the cage. Before the jump commenced, the rope would have turned back up
again so that the outboard end would have entered the cage through the access gate and lain on the cage
oor.
The jumper was attached to the outboard end of the rope by means of a pair of cus pulled tight around
the lower legs. Each cu was attached to the end of the rope by a webbing strap. Measurements of the cus
indicated that, when a jumper was hanging upside-down from the outboard end of the rope, the soles of his
feet would have been approximately 0.36 m below the end of the rope (see Fig. 1).
As a safety measure, the jumper was independently attached to the end of the rope by webbing (see
Fig. 2) secured to a body harness. The webbing was labelled MAMMUT 2500 daN UIAA and Made in
SWITZERLAND. Mammut are a well-known Swiss manufacturer of climbing equipment. 2500 daN
means 2500 deca-newtons, i.e. 2500
10 N 25,000 N (25 kN) (2.55 tonne). UIAA is the acronym for
the Union Internationale des Associations dAlpinisme, which accredits mountaineering equipment as
part of its activities. The strength rating applies to an endless sling (with a factory-sewn lapped joint), which
is a standard item of climbing equipment. The nominal strength of a single length of webbing tape is half
this gure, i.e. 12,500 N. The piece of webbing attached to the body harness was a single length of tape,
which appeared to have been cut and opened-out from an endless sling. The piece of webbing attached to
the bungee rope was an endless sling. The endless sling and the single length of webbing were knotted
together near the jumpers feet. The breaking strength of this assembly would have been less than 12,500 N
because of the weakening eect of the knot. The total length of the webbing assembly was 2.13 m, of which
1.28 m was single. Measurements indicated that, when a jumper was hanging upside-down supported by the
cus, there would have been approximately 0.60 m of slack in the webbing.
In the accident, the jumper jumped o the edge of the cage through the opened gate, taking the outboard
end of the bungee rope with him in the normal way. During the descent, he moved from an upright to an
inverted position (see Fig. 1), and began to apply force to the rope. However, with the rope vertical and
under tension, his legs pulled out of the cus. Tension was then applied to the safety webbing, the single
 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872 859

Fig. 1. Schematic geometry of the initial free-fall phase of the bungee-jumping accident.

Fig. 2. Webbing assembly identical to that which failed in the accident (after tensile test to fracture).
860 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

length of which snapped at the knot near his feet. The jumper then descended by free fall to the ground
below. He was wearing long trousers (made from thin smooth fabric) and short socks, but no shoes.

2.2. Examination and testing of bungee rope

The bungee rope was laid out at on the oor and pulled straight but free of tension. The unstretched
length of the assembly from the webbing loop at the inboard end to the karabiner at the outboard end
measured 15.60 m (see Fig. 1). The bungee rope consisted of three apparently identical cords set side-
by-side and taped together with insulating tape at regular intervals. The outboard end was protected with a
padded sleeve covered with a fabric jacket. Each cord had a braided sheath (19 mm OD) containing a large
number of ne, parallel rubber laments as the load-bearing elements.

Fig. 3. Force/extension curve for the bungee rope, showing both loading and unloading curves. The force/extension characteristics for
a new rope consisting of three 19 mm cords, as specied by BS 3F 70, are shown as triangular data points.
 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872 861

The tapes and the padded sleeve were then removed, and the three cords separated. One of the cords was
tested to an extension of 16.0 m (100% strain). At this extension, the measured force was 135 kg (1324 N).
The force/extension curve for the complete assembly of three cords was obtained by multiplying the forces
measured in the single cord by a factor of three. As shown in Fig. 3, the curve is highly non-linear, and is
typical of that for rubber. Note that, at the extension of 16.0 m, the force rises almost vertically with ex-
tension. The breaking strength of the rope was not determined on safety grounds, but is likely to be much
higher than the maximum force of 3972 N obtained from the test. Nevertheless, the rope had become a very
sti structural element at an extension of 16.0 m, and a small additional extension would probably have
caused failure. The extension of 16.0 m can therefore be considered as being the limit of extension in
practical terms.
The area under the force/extension curve represents the stored strain energy in the complete rope as-
sembly. The fact that the unloading curve falls well below the loading curve means that only a proportion
of the stored strain energy is released on unloading. This hysteresis energy loss explains why in a normal
bungee jump, the jumper rebounds to a height signicantly less than the height of the cage.

2.3. Testing of safety webbing

A tensile test to fracture was carried out on a webbing assembly identical to that which failed in the
accident. Fig. 4 is a graph of the force/extension curve. The sling broke at only 5000 N near a knot.

2.4. Analysis of data

The mechanics of the fatal jump was analysed using an energy-based approach (see Fig. 5). In the rst
stage of the descent, the jumper is in free fall. He progressively loses potential energy, which goes into
progressively increasing his kinetic energy (and hence his speed). However, once the rope becomes taut and
then stretches, it absorbs strain energy. In this second stage of the descent, the loss in the potential energy of

Fig. 4. Force/extension curve obtained from tensile test of webbing assembly shown in Fig. 2.
862 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

Fig. 5. Energy diagram for the bungee jumping process (accident case).

the jumper is converted into strain energy as well as kinetic energy. This has the eect of slowing the jumper
down. However, the jumper will only be arrested safely if all his potential energy can be absorbed as strain
energy in the rubber.
The calculations (see Table 1) show that the maximum strain energy which the bungee rope can absorb
before it reaches the limit of extension (the area under the loading curve in Fig. 3) is 28,660 J. However, the
potential energy released by the jumper as he falls to the limit of extension is 42,344 J. This leaves a surplus
of 13,684 J as kinetic energy, so he continues to travel downwards (at a speed of 14.4 m/s) rather than being
arrested by the rope.
Fig. 6 shows a graph of the speed of the jumper as a function of distance fallen, obtained from the energy
calculations (see Table 1). At the end of the rst stage of the descent (free fall) the jumper has fallen 16.7 m
and has reached 18.1 m/s. In the second stage of the descent (tensile extension of the rope) the jumper is
slowed down progressively as the rope stretches, reaching a minimum speed of 14.4 m/s after falling a total
distance of 32.7 m. Although this minimum speed of 14.4 m/s is much less than the speed which would have
been reached in free fall (25.3 m/s) it is well in excess of the zero speed required for a safe arrest.
Fig. 7 shows a graph of elapsed time as a function of distance fallen, obtained from the energy calcu-
lations (see Table 1). The initial free fall lasts 1.85 s, and the rope extension lasts a further 0.95 s. The total
time of 2.8 s compares very well with the average time of 3.0 s, which was timed from the videotape of the
accident. The small (7%) undershoot is probably due to neglecting wind resistance in the calculations.

2.5. Release from leg cus

The shape of the force/extension curve in Fig. 3 indicates that the force in the rope at the moment of
release could have been anywhere between 3972 N (405 kg) and the breaking strength of the rope assembly.
In this context, it is hardly surprising that the cus were pulled o the jumpers legs.
One should question the eectiveness of cus when used without footwear, since they are retained only
by friction. Friction is unreliable, because it depends on the compressive force between cu and leg (which
cannot easily be controlled) and the coecient of friction (which depends on many factors, such as con-
dition of skin, type of clothing if worn, etc.). A more geometrical resistance to cu release can be provided
 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872 863

Table 1
Energy calculations (accident case)
Distance Rope extension (m) PE (J) SE (J) KE (J) Velocity (m/s) Free fall Time (s)
fallen (m) velocity (m/s)
0 0 0 0 0 0
5 6475 6475 9.9 9.9 1.01
10 12,949 12,949 14.0 14.0 1.43
15 19,424 19,424 17.2 17.2 1.75
16.7 0 21,625 0 21,625 18.1 1.85
17.7 1 22,920 850 22,070 18.3 1.905
18.7 2 24,215 1940 22,275 18.4 1.960
19.7 3 25,510 3150 22,360 18.4 2.014
20.7 4 26,805 4460 22,345 18.4 2.068
21.7 5 28,100 5860 22,240 18.4 20.6 2.123
22.7 6 29,395 7360 22,035 18.3 2.177
23.7 7 30,690 8950 21,740 18.2 2.232
24.7 8 31,985 10,640 21,345 18.0 2.287
25.7 9 33,279 12,420 20,859 17.8 2.343
26.7 10 34,574 14,290 20,284 17.5 22.9 2.400
27.7 11 35,869 16,250 19,619 17.2 2.458
28.7 12 37,164 18,310 18,854 16.9 2.516
29.7 13 38,459 21,480 17,979 16.5 2.576
30.7 14 39,754 22,800 16,954 16.0 24.5 2.638
31.7 15 41,049 25,390 15,659 15.4 24.9 2.701
32.7 16 42,344 28,660 13,684 14.4 25.3 2.768
33.0 16 42,732 29,860 12,872 14.0 25.5 2.790
p
Notes: PE (J) 132
9.81
distance fallen (m); PE SE + KE; velocity (m/s) [2KE (J)/132]; strain energy absorbed at start of
cu release 28,660 J; work done in cu release 1200 J; energy absorbed at end of cu release 28,660 + 1200 29,860 J; free fall
p
time (s) [2
distance (m)/9.81]; time during rope extension phase calculated incrementally from distance/velocity.

Fig. 6. Graph of speed of jumper versus distance fallen, obtained from the energy calculations (accident case).
864 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

Fig. 7. Graph of elapsed time versus distance fallen, obtained from the energy calculations (accident case).

by wearing high-laced climbing boots. But even with properly anchored cus, it is dicult to see how it is
possible to arrest a jumper moving at 14.4 m/s over a very small distance without causing high g-force
damage to the body.
Assuming that the cus release at a force of 4000 N, and require a movement of 0.3 m to come o the
legs, then release will require 4000 N
0.3 m 1200 J of work. Descent of the jumper by this additional
0.3 m provides a potential energy of 388 J (see Table 1). The decrease in kinetic energy is therefore 812 J (see
Table 1). Thus, at the end of the release event, the jumper has kinetic energy of 12,872 J and a speed of
14.0 m/s (see Table 1). This means that pulling the cus o the legs only slows the jumper by 0.4 m/s.

2.6. Snapping of safety webbing

Once the legs have pulled out of the cus, the remaining kinetic energy of the falling jumper (12,872 J)
must be absorbed by the safety webbing if arrest is to occur. The energy absorbed by the safety webbing
was calculated from the area under the force/extension curve (see Fig. 4), and was found to be 940 J-only
7% of the 12,872 J required for arrest. In other words, the webbing was totally inadequate in terms of static
strength and, most crucially, energy absorbing capacity. A dedicated energy-absorbing device should have
been used instead, capable of absorbing the kinetic energy of the falling jumper without applying excessive
force to his body.

2.7. Force/extension curve of bungee rope

In spite of the manifest defects of the cus and the safety webbing, the root cause of the accident was the
use of a bungee rope with an unsuitable force/extension curve. To bring the falling jumper to rest at the
limit of extension, the weight of the jumper should have been no more than 90 kg (see Fig. 8 and Table 2).
 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872 865

Fig. 8. Energy diagram for the bungee jumping process (accident case, plus variants for reduced body weights of 90 and 70 kg).

Table 2
Energy calculations (accident case, variants for 90 and 70 kg body weights)
Distance fallen (m) Rope extension (m) SE (J) PE for 90 kg (J) PE for 70 kg (J)
26.7 10 14,290 23,573 18,335
27.7 11 16,250 24,456 19,022
28.7 12 18,310 25,339 19,708
29.7 13 20,480 26,222 20,395
30.7 14 22,800 27,105 21,082
31.7 15 25,390 27,988 21,768
32.7 16 28,660 28,871 22,455

To have a safe arrest, the rope must absorb all the potential energy of the falling jumper well before the
limit of extension of the rope is reached. This will also limit the maximum force in the load-bearing system
to calculable levels, reducing the risk of cu release. If the cus do release, a well designed safety webbing
should not break in these circumstances. As an example, Fig. 8 and Table 2 show the energy balance for a
body weight of 70 kg. In this case, it is clear that there is signicant spare energy-absorbing capacity in the
rope. The rope is only extended by 13 m, 3 m short of the limit of extension. This produces a force of 2250
N (229 kg) (see Fig. 3). However, even this force is 3.27 times body weight, and a g-force of 3.27 may well
cause physical damage to the jumper.
Cords should be tested on a periodic basis to ensure that there is no drift in the force/extension curve
with time and usage. In this connection, it is instructive to compare the force/extension characteristics
specied in BS 3F 70 [1] for new 19 mm cord. For 10% extension (1.56 m in the case study) the force must
be at least 340 N (1020 N for three cords in parallel). For 30% extension (4.68 m) the force must be
between 500 and 650 N (1500 and 1950 N for three cords). For 70% extension (11.70 m) the force must be
between 850 and 1100 N (2550 and 3300 N for three cords). The minimum total extension must be 105%
(16.38 m). These data points are plotted on Fig. 3. Comparison of the specied and actual data shows
that the bungee rope has suered considerable degradation of energy-absorbing capacity as the result of
repeated use.
866 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

2.8. Deadweight drop testing

It should be noted that HELA [3] refer to the need to undertake deadweight drop tests to demonstrate
that the cage height is properly set for the rope used and the jumpers weight, and to ensure the integrity of
the complete rope system. However, the above analysis shows that such an approach is not only concep-
tually awed, but is likely to overload the rope and cause permanent damage. It is essential to use the force/
extension curve for the rope and perform an energy-based mechanics analysis to determine the limit of safe
operating parameters. In this connection, it is noteworthy that an incident report from Central Bungee [5]
refers to tests involving a four-cord bungee rope loaded to 115 kg. The load was in the form of sandbags
supported in a canvas bag reinforced with webbing straps. On two consecutive occasions the bag burst at
full extension, conrming the predictions of very high forces produced at the limit of extension.

2.9. Conclusions from case study

The root cause of the accident was the use of a bungee rope with an unsuitable force/extension curve,
only able to absorb 68% of the potential energy of the falling jumper at the limit of extension. This was
partly due to degradation of the cords, presumably as the result of repeated use.
As a result, the rope did not arrest the falling jumper, who was consequently subjected to a large force,
sucient to pull the cus o his legs. Detachment of the cus was facilitated by the absence of high-laced
climbing boots and the consequent reliance on friction alone for security. The wearing of smooth-textured
trousers may also have contributed to cu detachment.
Subsequent to cu release, the safety webbing took the full force of the rope. Because of its low strength
and totally inadequate energy absorbing capacity, the webbing broke and the jumper fell to the ground with
a speed of impact equivalent to a free fall from a height of 31 m.

3. The need for a code

On the basis of the above analysis, it would appear to be only a matter of time before another similar
fatality occurs. In the authors opinion, all bungee jumping which is open to the public should be banned
until the operators do the following:
1. Prove by energy calculations, using the minimum properties for the ropes, and making an allowance for
degradation in service, that the energy of the heaviest jumper can be absorbed by the rope signicantly
before it reaches the limit of extension, so that the jumper is brought fully to rest.
2. Prove by energy calculations that the maximum deceleration force on the jumper does not exceed a spec-
ied value so as to avoid physical damage to the person.
3. Fit a dedicated energy-absorbing device between the end of the bungee rope and the jumpers body har-
ness, capable of absorbing any residual energy with an ample safety margin should the bungee rope pull
the ankle cus o the jumper.
If 1 and 2 are carried out properly, there should be little risk of the ankle cus being pulled o, so the
backup attachment should hardly ever be used. Thus cost cannot be used as an objection to tting a
dedicated energy-absorbing device. Such devices are standard in the climbing world, and are also required
when workers are suspended from ropes. They consist of webbing doubled back and forth many times and
cross stitched together. Under excessive load, the stitching progressively pulls out, absorbing large amounts
of energy in the process.
Calculations 1 and 2 only have to be done once for each conguration of cords. The question of
degradation requires cords to be tested periodically for load-extension characteristics. When enough
 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872 867

information has been gathered, it would be a simple matter to lay down a discard time (measured in number
of jumps completed) after which the cords should be destroyed.

4. Some calculations for BS ropes

4.1. Lower bound to BS force data

Fig. 9 shows the force/extension data for a new rope consisting of three 19 mm cords, as specied by BS
3F 70. The unstretched length of the rope is 15.6 m, as in the case study. It is evident that the straight line in
Fig. 9 gives a reasonable lower bound to the force data for a new rope. The equation for the line is

F u 800 150u; 1

where F is the force in N and u is the extension in m. The stored strain energy is given by

Fig. 9. Force/extension data for a new rope consisting of three 19 mm cords, as specied by BS 3F 70. The unstretched length of the
rope is 15.6 m, as in the case study. The straight line gives an approximate lower bound to the force data.
868 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

Fig. 10. Energy diagram for the bungee jumping process (linear lower bound to BS, 15.6 m unstretched length).

Z u Z u
U F udu 800 150udu 800u 75u2 ; 2
0 0

where U is in J. The curve of U u is plotted in Fig. 10. The maximum value of U occurs at the specied
105% extension (u 16:4 m) and is calculated to be 33,292 J from Eq. (2).
The distance fallen by the jumper in this case is 16.7 + 16.4 33.1 m. The weight which is just arrested is
given by

33; 292
m 102 kg; 3
9:81
33:1

and this is shown in Fig. 10.

It should be noted that the free fall distance of 16.7 m is 1.1 m greater than the unstretched length of
15.6 m (see Fig. 1). We call this extra vertical movement of the centre of mass the component slack. In the
absence of any slack, the jumper does not fall as far, and the rope can therefore arrest a slightly heavier
jumper. The weight which is just arrested when there is no slack is given by

33; 292
m 106 kg; 4
9:81
32

and this is shown in Fig. 10.

4.2. Changing the unstretched length

Fig. 11 shows the force/extension data again, but this time plotted for an unstretched length of 31.2 m,
twice that in the case study. The equation of the lower-bound line is now
 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872 869

Fig. 11. Force/extension curve for a new rope consisting of three 19 mm cords, as specied by BS 3F 70. The unstretched length of the
rope is 31.2 m, twice that in the case study. The straight line gives an approximate lower bound to the force data.

F u 800 75u; 5

and the strain energy is given by

U 800u 37:5u2 : 6

The curve of U u is plotted in Fig. 12. The maximum value of U occurs at the specied 105% extension
(u 32:8 m) and is calculated to be 66,584 J from Eq. (6).
Assuming the same slack of 1.1 m, the free fall distance now becomes 31.2 + 1.1 32.3 m, and the total
distance fallen is 32.3 + 32.8 65.1 m. The weight which is just arrested is

66; 584
m 104 kg; 7
9:81
65:1

and this is shown in Fig. 12. In the absence of any slack, the weight which is just arrested is
870 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

Fig. 12. Energy diagram for the bungee jumping process (linear lower bound to BS, 31.2 m unstretched length).

66; 584
m 106 kg: 8
9:81
64

Comparison of Eqs. (4) and (8) shows that, when there is no component slack, the weight which is just
arrested is independent of unstretched length. This is also intuitively obvious: if the unstretched length
is doubled, the jumper will fall twice as far, and will generate twice as much PE; but the strain-energy
absorbed by the rope will also double, so the two eects balance.
Comparison of Eqs. (3) and (7) shows that the eect of the xed 1.1 m of slack decreases as the un-
stretched length increases, tending to zero at innite unstretched length. The opposite is true as the
unstretched length decreases, as the examples of Figs. 13 and 14 show. However, even when the unstretched
length is only 5 m (Fig. 13), the weight which is just arrested falls by only 10 kg to 96 kg.

4.3. Changing the number of cords

It is intuitively obvious that if the number of cords in the rope is changed, then the jumper weight which
can be arrested will change in the same proportion. Thus, a rope of six cords instead of three will absorb
twice as much energy at the limit of extension, and will arrest twice the body weight.

4.4. Final remarks

The calculations in this section can be used to form the basis of a quantitative mechanics-based oper-
ating code. Provided the component slack is 1 m or less, then the weight which can be arrested by a given
rope will be xed, provided the unstretched length is at least 15 m. For a new rope consisting of 3
19 mm
braided cords to BS 3F 70 1991, the maximum weight is around 100 kg. However, a factor of safety is
required on this weight to ensure that, after allowing for in-service degradation, there is still signicant
residual energy-absorbing capacity after the arrest has taken place.
 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872 871

Fig. 13. Energy diagram for the bungee jumping process (linear lower bound to BS, 5.0 m unstretched length).

Fig. 14. Energy diagram for the bungee jumping process (linear lower bound to BS, 1.0 m unstretched length).

5. Additional comments

5.1. Redundancy

The BERSA and HELA codes [2,3] both place great emphasis on the need for structural redundancy.
The BERSA code states If the failure of a component in a system can cause a dangerous situation, that
component is backed up by the installation of a second component in parallel with the rst. If the rst fails,
872 D.R.H. Jones / Engineering Failure Analysis 11 (2004) 857872

the second takes over and danger is averted. The HELA code states All parts of the jump line must be
duplicated.
Redundancy is clearly easy to achieve in the end attachments of the bungee rope, with components such
as karabiners or shackles. With suciently strong components, it is a straightforward matter to ensure that
the failure of one component (e.g. a karabiner with an open gate) will shed load to a parallel component
without overloading it. However, the codes are unduly restrictive in specifying only two-fold redundancy.
In the case of components such as slings, a connection may require multiple parallel components to achieve
sucient combined strength. In such cases, there must be sucient elasticity in the components to ensure
that the load is spread evenly between them.
Redundancy in the rope itself (including the end splices of the cords) is unlikely to be eective. As can be
seen from this analysis, the failure of one cord in a three-cord rope could be disastrous, because the re-
maining two cords would in all probability not be able to absorb all the energy of the falling jumper. Thus
the structural integrity of each cord and its end splices are paramount to the safety of the operation. This is an
example of how blind reliance on the concept of redundancy without a detailed analysis of the load
shedding process can be dangerous.

5.2. End attachments

The BERSA code states All karabiners, pulleys and shackles shall have a minimum breaking load of
2000 kg, and shall have a safety factor of at least 4 times the dynamic load. 2000 kg is equal to 19,620 N.
In a 3
19 mm rope, the maximum dynamic load with residual energy-absorbing capacity at arrest could be
around 4000 N (see Fig. 3). A strength of 4 times the dynamic load would be 16,000 N, less than the
maximum requirement of 19,620 N. However, a 5
19 mm rope would require a strength of 16,000

5/3 26,667 N according to the code.
Because leg cus are unreliable points of attachment, the connection between the end of the rope and the
body harness should also satisfy an appropriate load criterion and should be duplicated in line with the
redundancy requirement. The BERSA and HELA codes do not specify either load ratings or redundancy in
this nal, critical link in the load train. Neither code considers a minimum energy-absorbing capacity for
harness connections.

5.3. Unbraided ropes

Unbraided ropes are generally used when much greater percentage elongations are required, as in the
more extreme sporting situations. The author is not aware of any manufacturing standard for such ropes.
Although the analysis described here can in principle be applied to unbraided ropes, test data for actual
service ropes is required as input to the calculations. The distance fallen by the jumper from jump-o to
arrest is often critical. The author has witnessed jumpers descending head rst from the deck of the Victoria
Falls railway bridge to within a few feet of the Zambesi river some 360 feet below. In such applications, very
minor changes in the force/extension characteristics of the rope can have tragic results.

References

[1] British Standards Institution. BS 3F 70: 1991: specication for heavy duty braided rubber cord. London: BSI; 1991.
[2] British Elastic Rope Sports Association. The BERSA code of safe practice, August 1998. Oxford: BERSA; 1998.
[3] Health and Safety Executive/Local Authorities Enforcement Liaison Committee (HELA). Local authority circular 47/2, bungee
jumping, August 2000. London: HELA; 2000.
[4] South Carolina General Assembly. Bill 3094-bungee jumping. Charleston: SCGA; 19931994.
[5] Central Bungee. Incident. Oxford: Central Bungee; undated.