Famous Failures:
The Vasa
The Story of the Vasa
Taken from http://www.cise.ufl.edu/~jnw/OOCourse/Lectures/01.05.html
•
•
•
•
1625: King Gustav of Sweden commissions construction of a new flagship,
the Vasa.
Hendrick Hybertszoon (master shipwright from Holland) is selected to build
it.
No written specifications
Shipbuilder assumes ship will be 108 feet in length.
– After first review, King Gustav requests a 135 foot ship.
– Timber is added to make the ship 120 feet.
•
While on vacation, King Gustav finds out that the Danish king is building a
ship with three gun decks.
– He asks that a third gun deck be added to the two already on the Vasa.
– 50 brass 24-lb cannons (at one ton each) will be added.
•
•
•
Stability tests are conducted (involving 30 sailors who run from one side of
the ship to the other). The ship appears to be unstable, but the problems
are ignored and not communicated up the chain.
On a Sunday in August 1628, the Vasa sets sail. One mile from harbor a
wind gust catches the main sail and the ship overturns and immediately
sinks.
The Vasa was later brought up from the briny deep at greater cost than was
expended in her manufacture.
Parallels between Shipbuilding in the
1600's and Software in the 1990's
• Shipbuilding in the 1600's was a craft based on engineering
practice.
• Specifications for ships were ad hoc and usually verbally
communicated.
• Designs for ships were inadequate to insure the finished product
would satisfy the needs of its users.
• The full implications of changes to the design were not always
understood.
• Too much time was wasted carving decorative sculpture prior to
determining if the ship would sail.
• Success breeds failure. Those who succeeded in building small
ships would believe they knew how to do it even though they had no
idea of the fundamental physical principles involved.
• Shipwrights in the 1600's (one in particular) did not know how to say
``no'' to customers.
Implications:
• Systems may last longer than we think
(sometimes hundreds of years).
• Maintenance costs may exceed the
original development cost.
• We can examine, in minute detail,
components of systems we have built
decades or even centuries later.
Famous System Failures 1
Name
Year
Hubble Space
Telescope
1990
poor
requirements?
No
Probable cause of failure
Lack of total system test. Mirror was too flat on
one edge by 1 / 50th of the width of human hair.
Ariane 5 missile
1996
No
Incorrect reuse of software
Faulty scaling up, Faulty software testing,
Software operand error
SuperConducting
SuperCollider
1995
No
Cost overruns, Failure to maintain public
support
GE rotary
compressor
refrigerator
1986
No
Inadequate testing of new technology
Motorola, Iridium
1999
No
Misjudged competition and miss-predicted
technology
IBM PCjr
1983
Yes
An attempt to impose customer needs
Space Shuttle
Challenger
1986
No
Failure to communicate with design,
Motivation to make client “happy”
War in Vietnam
1967-72
Yes
No problem statement,
Micromanagement
Edsel automobile
1958
Yes
Failure to discover customer needs
Titanic
1912
No
Poor quality control
Apollo-13
1970
No
http://tucson.sie.arizona.edu/sysengr/slides/failures.doc.
Famous System Failures 2
Name
Year
poor
requirements?
Probable cause of failure
Tacoma Narrows
Bridge
1940
No
Scaling up an old design (remind you of
something?)
New Coke
1988
Yes
Arrogance, The “wrong question”
questionnaire, Underestimating the effects of
social influence
A-12 airplane
1980s
No
Mismanagement ($2B for nothing)
Chernobyl Nuclear
Power Plant
1986
No
Bad design, Bad risk management,
Cost cutting
Lewis Spacecraft
1997
No
Design mistakes, Ineffective assurance
processes at NASA
Mars Climate
Orbiter
1999
No
Lack of training of Navigation team, Technical
error (use of English and not Metric units)
Mars Polar Lander
2000
No
Failure of middle management
Sept 11 attack on
WTT
2001
Yes
Flimsy cockpit door
Super-Conducting Super-Collider Failure
• In the 1980s a project got underway in the U.S.
to build a device in Texas.
• It had the support of 3 presidents from both
parties: Reagan, Bush (senior), and Clinton.
• Billions were spent on research and
development
• Land for the 'runway' was purchased and
excavation begun.
• And then Congress pulled the plug.
back
Challenger
• On January 28, 1986 the challenger space
shuttle exploded about 1 minute after launch
killing all 7 astronauts on board.
• The shuttle exploded because two rubber O-rings leaked after
losing their resiliency because the shuttle was launched on a
very cold day (less than 0 Celsius).
• On the day before the launch the engineers who designed the
rockets were opposed to launching the challenger because
they were concerned that the rings would not seal at such cold
temperatures. This was to be the EXACT cause of the
accident.
Edward R. Tufte: Visual Explanations.
Back
Challenger
• The engineers presented their case using 13 charts which they
faxed to NASA.
• A high level NASA official was “appalled” by the
recommendation, and asked that the rocket maker (Morton
Thiokol) should reconsider – even though this was the ONLY
no-launch recommendation in 12 years.
• Other NASA officials pointed out serious weaknesses in the
charts.
• Reassessing their the situation after such responses, the
Morton Thiokol managers changed their minds and favored a
launch.
• They explained their change of mind by saying that the
evidence presented by the engineers was inconclusive in
showing that O-ring problems were linked to cool temperatures.
Edward R. Tufte: Visual Explanations.
Back
Challenger
Back
Challenger
Edward R. Tufte: Visual Explanations.
Back
Tacoma Narrows Bridge
Collapsed in 1940
• If engineers like had solved the
same problem nearly a century
ago- why did it happen?
• Two reasons—By the 1920s
and 1930s, after
– many successful suspension
bridges had been built
– engineers became overly confident
in their ability to extrapolate from
earlier experience
– Preoccupation with with economy
and aesthetics
See movie
Back
Why Study Failures?
• There is probably more to learn from
failures than from successes
– As long as you are willing to look honestly at
the reasons for failure
– Too many times failures are brushed away
and not learned from- ensuring that more
failures will occur
 Those who do not remember the
past are condemned to repeat it