Sigma Xi, The Scientific Research Society

Engineering: History and Failure

Author(s): Henry Petroski
Reviewed work(s):
Source: American Scientist, Vol. 80, No. 6 (November-December 1992), pp. 523-526
Published by: Sigma Xi, The Scientific Research Society
Stable URL: http://www.jstor.org/stable/29774773 .
Accessed: 01/10/2012 19:14

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Sigma Xi, The Scientific Research Society is collaborating with JSTOR to digitize, preserve and extend access
to American Scientist.

http://www.jstor.org
 Engineering

and Failure
History

Henry Petroski

is a profession focused on the future, looking And great new engineering achievements do not come to be
back mainly to calibrate progress; engineers are typical? merely by inference from an extrapolation of successful
Engineering
ly drafting plans for the next generation of artifacts, seeking precedents. Indeed, the history of civil engineering is littered
to achieve what has not been done before. On one hand, we with thewreckage of famous bridges thatwere designed in a
are ever conceiving larger, faster,more powerful structures tradition of success: the Tay Bridge in 1879, the firstQuebec
and systems; on the other, we are ever devising smaller, Bridge in 1907, the Tacoma Narrows Bridge in 1940.
lighter,more economical machines and devices. Every new In an open discussion of the relevance of history held in
thing has meaning in comparison with that which it 1975 before the Institution of Structural Engineers, R. J.M.
the past, as itwere, and the Sutherland expressed the opinion, which was echoed fre?
supersedes?bettering usually
recent
past. quently, thatmajor engineering disasters "are much more
Ifpast engineering achievements have any relevance, it likely to be avoided if future designers, individually, de?
would thus appear to be principally as standards against velop a habit of looking back and questioning how each
which the latest designs are judged. From this point of view, concept grew." Unfortunately, this is seldom done. Ironi?
only themost recent of technological history is really rele? cally, signal successes in engineering have tended to arise
vant formodern engineering, and then only insofar as it not out of a steady and incremental accumulation of suc?
presents data to be extrapolated or a challenge to be over? cessful experience but rather in reaction to the failures of
come. Any extended history of engineering would appear the past?from the minor annoyances accompanying ex?
tobe mostly irrelevant technically. The history of engineer? isting artifacts to the shock of realization that the state of
a succession
ing as of achievements, of incremental the artwas seriously wanting. Thus, the collapse of the Tay
progress, might motivate young engineers and give them led directly to the Firth of Forth Bridge, which celebrated its
pride in their profession, but it is not commonly expected to centennial in 1990; the collapse of the firstQuebec Bridge
add directly to their technical prowess. led to the redesigned second, now a symbol of Canada;
This is a shortsighted view. The history of engineering, and the colossal collapse of the Tacoma Narrows taught
when embedded in a social and cultural context, does in bridge engineers overnight an appreciation for aerody?
facthave considerable potential fordemonstrating the true namics that has led to such new suspension-bridge designs
nature of engineering as practiced in the real world, but as the Severn and Humber. An efficacious history of engi?
only if it is presented in a way thatmakes it engineering as neering should incorporate a treatment of engineering fail?
well as history. For no matter how profoundly engineers ures not only for their value in adding a measure of humil?
know that their problems have more than a technical di? ity to the innate hubris of engineers but also for the
mension, the solutions ultimately will suffer ifengineers do essential features of the engineering method that they can
not stand on solid technical ground. Thus one of themost so effectively reveal.
promising uses of history in engineering education is to add
to a fundamental understanding not of how obsolete arti? Sketches in the Sand
factsworked but of the timeless aspects of the engineering One very interesting study of engineering failures was con?
process itself, while at the same time providing an appreci? ducted by P. G. Sibly and A. C. Walker some years ago. They
ation for the,past and process of civilization and engineer? found thatmajor bridge disasters have occurred with sur?
ing's role in it. prising regularity over the past two centuries. If our engi?
The history of engineering, as of civilization, is clearly neering science and experience are cumulative, how can
one of both successes and failures, and the failuresmay be such cyclic behavior be explained? The explanation lies in
themore useful component of themix. Although examples the nature of engineering design, which begins in a most
of good engineering practice can certainly serve as para? primitive and nonrational way. Engineers literally dream
an
digms of good judgment, great people do not become so up designs, and usually in graphic form. It is only when
merely by reading biographies of great men and women. engineer has the equivalent of a sketch or drawing that the
engineering sciences can be called upon to analyze the prac?
Henry Petroski is professor of civil engineering and chairman of the ticality of a design, or that other engineers can be consulted
civil and environmental at Duke for their experience. The process is often convoluted, of
department of engineering University,
Durham, NC 27706. course, and an engineer's sketch can be informed, con

1992 November-December 523

 plauds are certainly not news, the lessons learned and
caveats embedded inhis stories of human error and techni?
cal failure sound modern because the same sorts of embar?
rassing mistakes and oversights have continued to plague
engineering through the ages. Human error, being part of
human nature, cannot be expected ever to cease, but we
might be able to reduce it,at least in engineering, by looking
to the distant rather than to themore recent past. Being root?
ed in human error, classical case studies of engineering fail?
ures are never obsolete. Furthermore,
being generally re?
moved frommodern controversy and debate, older case
studies can more easily reveal aspects of the engineering
method with incontrovertible clarity.
Among the stories told by Vitruvius is one about the en?
gineer Diognetus, who had been retained by the Rhodians
to design machines as needed to defend the city against
siege. One day another engineer, named Callias, came to
Rhodes and gave a public lecture,which was accompanied
by a model of a revolving crane claimed to be capable of
Figure 1. A successful move of an obelisk: the raising of the seizing an enemy's helepolis, or siege machine, and lifting it
Vatican obelisk in 1586. (Plate from Templum Vaticanum, inside thewall of the city.The Rhodians were so impressed
by Carlo
Fontana, 1694.) with the demonstration that they transferred Diognetus's

^^^^^^^ grant toCallias.

sciously or unconsciously, by a wealth of knowledge of en? In themeantime, according toVitruvius, a 180-ton hele?
gineering science and of experience with past designs, suc? polis had been designed and constructed by the enemy, and
cessful and unsuccessful. Such information helps themind's Callias was asked to scale up his model tomeet the chal?
eye select and reject features of a tentative design. lenge.He replied that itwas impossible to do so, because not
Eugene Ferguson has elaborated on the role of nonver? all things thatwork on a small scale are equally feasible on a
bal thought in engineering and has put it in a historical con? larger one. Vitruvius, as a good engineer would, illustrates
text.Although his work demonstrates that there has been a the principle with a familiar example: A hole of about an
definite evolutionary pattern in the quality and clarity of inch indiameter can readily be bored with a wood auger, but
graphic representation of artifacts, for example, the unmis? to bore a three-inch hole by a similar device is impractical,
takable lesson is that the engineering design process does and a six-inch one "seems not even conceivable." (Although
not proceed today in any fundamentally differentway than we may not have direct experience with augers, all of us
it did millennia ago. Egyptian pyramids, Roman arch must have experienced a similar frustration with screw?
bridges, Greek temples and every other engineering project drivers. Indeed, this iswhy slotted screws are made only so
of ancient civilizations began in themind's eye of some in? large; above a certain size they come with bolt heads.)
dividual engineer,whether or not the individual was so des? Not surprisingly, Callias was summarily dismissed by
ignated. And only when that conceptual designer articulat? the Rhodians, who then begged Diognetus to help them
ed it by a sketch in the sand, a word picture or a model capture the approaching siege machine. Vitruvius's contin?
could a conceptual design be praised or criticized by the uation of the story is full of human dimensions, which
conceiver's colleagues, accepted or rejected by the commis? makes itall themore valuable forputting the engineering in
sioners, or built or bungled by the contractors. a real-world context. According to Vitruvius, Diognetus
For all of engineering's real progress in technical knowl? agreed to help only if the captured machine would become
edge and analytical technique, the underlying nature of de? his property.When thiswas agreed to,he did not proceed to
sign is fundamentally the same today as ithas always been, construct a largermachine but "pierced thewall in the place
and therein lies the true potential of a properly conceived where themachine was going to approach it, and ordered
history course for engineering students: Not only can it all to bring forth fromboth public and private sources all the
bridge the gap between the nontechnical and the technical water, excrement, and filth, and to pour it in front of the
aspects of the curriculum, but it can also provide at the same wall through pipes projecting through the opening." This
time a theoretical framework within which the nature of en? action continued throughout the night, and the next day the
gineering itself can be articulated. Furthermore, thewealth giant helepolis became so bogged down in theman-made
of wisdom about the seemingly ineffable engineering swamp that it could neither advance nor retreat. The great
method that exists in countless classical case studies can be machine was abandoned, and Diognetus had it set up in a
brought out of the library, into the classroom and, ultimate? public place in the city and dedicated it to the people of
ly, into the design office.Good stories about the engineering Rhodes. Vitruvius's conclusion that "in works of defense,
design process are never obsolete, for they reveal the true not merely machines, but, most of all, wise plans must be
nature of engineering, warts and all. prepared," clearly remains valid today, because the design
Vitruvius is often cited as being among the earliest of process is no less dependent on a proper anticipation of the
writers on engineering, but that is not where his relevance conditions under which a design is possible and will be
ends. Some of the stories he relates in the last of his Ten used. Furthermore, the story ofDiognetus and the Rhodians
Books onArchitecture sound as fresh as those in the latest is? demonstrates very forcefully, ifmetaphorically, how hu?
sue of the Engineering News-Record.
Although the technolo? man-scale obstacles can check the advance of even themost
gy of which Vitruvius wrote and the achievements he ap powerful products of high technology.

524 American Scientist, Volume 80

Obelisks and O-Rings
The reason Callias could not scale up his model to a full
size defense machine was understood in Vitruvius's day
only by analogy Indeed, themysterious problem of scale
continued to frustrate designers well into the Renaissance,
and still does today in cases ranging from supertankers to
massive process plants. Large obelisks were known to
have broken under their own weight during erection, great
cathedrals collapsed during construction, and colossal
Galileo opened his
ships broke up upon being launched.
seminal work on the strength of materials by recounting
similar engineering horror stories, which he used to argue Figure 2. Illustration of broken columns from Galileo's Dialogues
that geometry was not the only tool a designer needed. Concerning Two New Sciences.
has evolved considerably since
Analytical mechanics
Galileo's time, of course, and the stories he related tomo? as the Britannia, Firth of Forth, Niagara Gorge, and Brook?
of their
tivate his work remain more relevant for engineering than lyn Bridges) that tested the analytical capabilities
his errant analysis of the cantilever beam (although how time. Perhaps precisely because of this tradition ofwriting,
even a genius likeGalileo could fall into analytical error is the engineers of such projects tended to be more open about
itselfwell worth the attention of the modern instructor what motivated, drove and checked their designs than are
can mask their reasoning in computer
and student). engineers today,who
written that
For example, early in the firstday of dialogues on his graphics and printouts. Sir Alfred Pugsley has
new sciences, Galileo related a story of a column thatwas he found iteasier to "get under the skin of... earlier design?
on two piles of timbers. A mechanic, seeing ers... than to do the same in the case of themore typical de?
being stored
this and knowing how obelisks and ships could break under sign teams of today."
their own weight, feared that the column itself might be There are features of engineering design thatwill for?
close to He therefore, that ever remain independent of the state of the art and, there?
dangerously breaking. suggested,
a third support be added under themiddle of the column, fore, are independent of whether we use abacuses, slide
and everyone consulted agreed that itwas a good idea. So it rules, or computers to analyze and flesh out the conceptu?
was done, and the safety of the column was thought to be al designs fromwhich all engineering flows. Even as the
more and more scrutiny,
improved. Some months later, however, the column was design process is coming under
found to be broken in two, but not in theway themechanic in a computer context, more temporally distant
especially
had feared. What had happened was that the freshly in? writing about engineering remains one of themost illumi?
stalled support did not settle as readily as the original ones, nating sources of insight into the often ineluctable engi?
and the column was unable to support its own weight bal? neering method.
anced on the single central support.
The story of the fallen column was clearly significant
toGalileo, and it remains even today a forceful reminder
that a seemingly innocuous or even beneficial design
cause of a failure, albeit in
change can actually be the root
a different mode, that itwas intended to obviate. Within
recent years, not dissimilar minor engineering design
to such visible disasters as the
changes have ultimately led
collapse of the skywalks in the Kansas City Hyatt Regency
Hotel and the explosion of the space shuttle Challenger. Al?
did not have near?
though Renaissance engineers certainly
the even of today's students
ly analytical sophistication
as sopho?
(who are expected tomaster Galileo's problem
mores), that is not to say that classical experiences with
do not re?
conceptual or pre-analytical engineering design
main relevant. Indeed, itmight be argued that had engi?
neers involved in designing modern skywalks or solid
rocket boosters been familiar with Galileo's story of de?
changes made on the fallen column, theymight not
sign
have accepted so uncritically the promised efficacy ofmul?
or extra O-rings.
tiple support rods
Design anecdotes told by Vitruvius or Galileo can have
value well beyond merely mastering the texts of these clas?
sic authors or the nature of Roman or Renaissance engi?
there are
neering. The stories briefly related here?and
countless like them inmajor and minor classics and even in
obscure engineering texts and memoirs with no pretensions
to being classics?show us the timeless features of engi?
neering. The written record of 19th-century engineering, for Figure 3. Illustration of a cantilever beam from Dialogues Concerning
contains quite explicit discussions of projects (such Two New Sciences.
example,

1992 November-December 525

 The foundation engineer Ralph Peck has written wide?
and articulately on engineering judgment, that attribute
U.S. DEPARTMENT OF ENERGY
ly
RESEARCH
OFFICE OF HEALTHANDENVIRONMENTAL
generally thought most difficult to teach. In discussing the
as 1981 that "nine
reliability of dams, Peck observed as late
Alexander Hollaender out of ten recent failures occurred not because of inadequa?
DISTINGUISHED cies in the state of the art, but because of oversights that
could and should have been avoided." Peck has also point?
Postdoctoral ed out that,when it comes to design errors and failures,
"problems are essentially nonquantitative" and, further?
Fellowship Program more, that "the solutions are essentially nonquantitative."
Research OpportunitiesinEnergy-Related He has acknowledged that improvements in analysis and
Life, Biomedical, and Environmental Sciences testingmight certainly be profitable, but it is also likely that
? "the concentration of effortalong these linesmay dilute the
Research inOHER-sponsored programs
? effort that could be expended in investigating the factors
Tenable at various national laboratories
? entering into the causes of failure." Among Peck's prescrip?
Stipends $37,500 tions for restoring good judgment to engineering practice is
? a historical perspective; he has deplored, for example, the
Doctoral degree received after
fact that engineers taking the examination leading to regis?
May 1, 1991
? tration as a structural engineer in Illinois could not properly
U.S. citizens or PRA eligible
identify such significant structural achievements as theEads
and applications:
Information Bridge across theMississippi.
The practice of engineering is truly as old as civilization,
Hollaender Postdoctoral Fellowships
and indeed civilization as we know it is hard to conceive of
Science/Engineering Education Division without the work of engineers. But whereas many of the
Oak Ridge InstituteforScience and Education
classic works of civilization, from poetry to pyramids, have
P.O. Box 117
long been taken on their own timeless terms, themethods of
Oak Ridge, TN 37831-0117
engineers are often thought to be continually superseded.
(615)576-9975 Although helicopters may have replaced ramps for erecting
obelisks, and computers may have made possible designs
a genera?
Deadline January 15, 1993 beyond the calculational reach of engineers only
tion ago, there are fundamental aspects of the conceptual
engineering design process that have changed little (if any)
CARDS!
BURNYOURREFERENCE over millennia. In fact, the ready availability and power of
the tools ofmodern technology may even threaten to erode
TM somewhat themore basic conceptual engineering skills.
Ironic as itmay seem, we might find an antidote to the ero?
REF-11 sion of fundamental design skills and critical engineering
on en?
Computerizes your REFERENCES judgment more readily in some of the oldest volumes
and prepares your BIBLIOGRAPHIES gineering than in themost modern textbooks. Such promise
argues strongly for a proper infusion of history into engi?
? Maintains a data base of references curricula.
neering
? Searches forany combination of authors, years of
publication, reference title,publication title, Acknowledgment
keywords or abstract This column is adapted froma position paper preparedfor thecon?
? Formats bibliographies exactly as you want them
ference on thehistory of engineering held at StanfordUniversity
? Reads your paper, inserts citations into the paper, on July 9-11.1 am grateful toWalter Vincentifor invitingme to
and prepares a bibliography of the references cited which was supported by theAlfred P.
participate in theconference,
(optional) Sloan Foundation. A conferencereport is available fromRobert E.
? Downloads references fromanyOnline data base McGinn, Program inValues, Technology, Science, and Society,
including NLM, BRS, DIALOG and Laserdisc StanfordUniversity, Stanford,CA 94305-2120.
formats (optional)

MS-DOS, CP/M 80
IBM PC/XT/AT, $195.?? Bibliography
Ferguson, Eugene S. 1992. Engineering and theMind's Eye. Cambridge,
Mass.: MIT Press.
RT-11, TSX-Plus, RSX-11, P/OS ... $250<x> Galileo. 1638. Dialogues Concerning Two New Sciences. Translated by H.
Crew and A. de Salvio (1914 edition). New York: Dover Publications.
VAX/VMS (nativemode) . .'650.00 Peck, Ralph. 1981. Where Has All the Judgement Gone? Oslo: Norges geot
eckiske institutt, Publikasjon No. 134.
00 Pugsley, Sir Alfred, Mainstone, R. J., and Sutherland, R. J.M. 1974. The
ANY MANUAL & DEMO $20. relevance of history. The Structural Engineer 52:441^145. See also the
discussion, ibid. 53:387-398.
322 Prospect Ave., Hartford, CT 06106
and their causes.
Sibly, R G., and Walker, A. C. 1977. Structural accidents
m\ ISYSTEMSJ (203) 247-8500 add 8% sales tax.
Connecticut residents Proceedings of the Institution of Civil Engineers 62, Part 1:191-208.
Vitruvius. c. 30 B.C.E. The Ten Books on Architecture. Translated by M. H.
Circle No. 11 on Reader Service Card Morgan (1914 edition). New York: Dover Publications.

526 American Scientist, Volume 80