Page 158 - Forensic Structural Engineering Handbook

STANDARD OF CARE 4.7

Error
The role of error and failure in structural engineering design is amply described in the
books of Henry Petroski, among others. This chapter does not delve into that most impor-
tant topic. The reader is encouraged to refer to Professor Petroski’s many excellent volumes
on the subject.
Error can be broadly defined as the failure to achieve an intended outcome. An error can
be categorized as accidental or systematic, or as a blunder, a slip, a lapse, a mistake, or a
violation.
Accidental error is also referred to as random error. Accidental error can be seen in a
target on a rifle range. A shooter aiming at the bull’s-eye is likely to not hit the exact cen-
ter every time. The bullet holes in the target will be randomly clustered around a central
area. The source of this scatter is the random variability of each shot. The size of the clus-
ter is a display of the precision of the shots. A shooter who is very precise will have a tar-
get with a closely-packed cluster of holes; a less precise shooter will have a wider cluster.
Random error can be reduced, or precision enhanced, by controlling the randomly varying
influences on each shot. Note that a tight cluster and a wide cluster can both be accurate.
That is, they can both be centered on the bull’s-eye. The accuracy of the cluster of shots is
the distance between the bull’s-eye and the average shot, or central point of the cluster.
Both the tight cluster and the wide cluster can have the same average, in which case they
would be equally accurate.
Systematic error can be illustrated using the same target on a rifle range. If the bullet
holes in the target are clustered around a spot other than the bull’s-eye at which the shooter
was aiming, then the distance between the center of the cluster and the bull’s-eye discloses
the systematic error. This error might be a characteristic of the rifle if the sights are not cor-
rectly aligned or of the shooter if the rifle is consistently jerked to one side at every shot.
This kind of error can be corrected by adjusting the rifle to eliminate the misalignment,
by aiming away from the bull’s-eye to compensate for the misalignment, or by learning
how to fire without jerking the rifle. Note that the cluster of shots can be very precise, or
tightly packed, yet they can still be off target, or far from accurate.
Again looking at the target at the rifle range, we see that a blunder, or a slip, a lapse, a
mistake, or a violation, can be analogous to the single stray shot that goes wide of the main
grouping of shots. There might be many reasons for such an error, including (but certainly
not limited to) inattention, distraction, fatigue, mechanical or material defect, organiza-
tional deficiency or human frailty, venality, and avarice.
To tie this discussion to structural engineering practice, rather than riflery, an example
of random error might be the variation between the design floor live load and the actual load
on any particular part of a floor in a building. There is no need to precisely forecast the exact
load on each square foot of the floor if the design live load is appropriately selected. The
accuracy of the design floor live load is adequate if the structure performs as intended with-
out being “overengineered.” The building codes that specify minimum design floor live
loads intend to guide engineers to this result. Those minimum loads were selected by the
engineers who wrote the codes, with an appreciation of the random error inherent in defin-
ing the load.
A systematic error in structural engineering can be seen in the use of small coupons of
wood in determining allowable stresses in wood. Because the small coupons are less likely
to have the full range of defects that a full-sized member has, the allowable stresses derived
from small coupons will be higher than those derived from full-sized members. This sys-
tematic error has in fact resulted in failures of some older wood structures, and a revision
downward of allowable stresses.
Blunders, slips, lapses, and mistakes in structural engineering practice occur all the
time to varying degrees and with varying consequences. A mislocated decimal point, a