Environmental Engineering Reference
In-Depth Information
scaling to field-sized properties is a statistical problem, but the profession has not
assimilated it.
5.
Can we develop reliability models for internal erosion
? About one-third of the failures
of modern earth dams occur because of internal erosion. However, practical design
methods are deterministic, and we do not have adequate methods to detect piping in
existing structures. The major problem is that we lack a physical model for predict-
ing internal erosion from first principles. It is a problem for which the basic physics
of failure has not been described adequately because they are not well understood. It
is also a problem dominated by local conditions of soil gradation, density, strength,
and permeability to such an extent that a realistic model is difficult to formulate with
meaningful parameters.
6.
Can we more strongly connect the observational method to quantitative Bayesian
updating?
In a paper near the end of his life, Terzaghi wrote, “Soil engineering proj-
ects […] require a vast amount of effort and labor securing only roughly approximate
values for the physical constants that appear in the equations. The results of the compu-
tations are not more than working hypotheses, subject to confirmation or modification
during construction. […]. The elements of this method are 'learn-as-you-go:' Base the
design on whatever information can be secured. Make a detailed inventory of all the
possible differences between reality and the assumptions. Then compute, on the basis
of the original assumptions, various quantities that can be measured in the field. On the
basis of the results of such measurements, gradually close the gaps in knowledge, and
if necessary modify the design during construction.” This is Bayesian logic.
7.
We need better ways of communicating risk to stakeholders
. There are actually two
types of risk communication involved: communication with owners and communica-
tion with the public. In both cases, the problem is that the engineer must convert tech-
nical knowledge into language that can be understood by stakeholders not conversant
with the technology. An underlying issue is that it is difficult for the public, owners,
and even engineers to deal with and plan for extremely rare events that have severe
consequences. How should resources be allocated? Which of many dire scenarios
should be taken seriously?
8.
Can we improve and validate the application of LRFD to geotechnical problems?
LRFD was developed as a rational way to quantify the contributions of different load-
ings in the analysis of steel and concrete structures. Resistance was and remains a
single variable. The factors are primarily applied to the loads. LRFD has been applied
with some success to design procedures for piles, which are essentially structural mem-
bers embedded in the ground. The approach has been less satisfactory when applied to
retaining structures, and even then effects such as strength dependent on normal stress
and varying water pressures are ignored.
9.
Can we develop practical guidance on statistical considerations in exploration strate-
gies
? There is a large literature on sampling strategy, but it has had little impact on site
characterization. Part of the problem is that the results are not very useful unless they
are combined with Bayesian thinking and subjective probability. There is a need for
serious study of what can be done in a realistic environment for communicating these
insights to practitioners.
10.
Can we improve on rules for multiple-failure modes
? Calculating the failure prob-
ability of a system from the computed failure probabilities of the individual modes
often leads to unreasonably conservative results. The current practice is to assume the
extreme cases that the behaviors of the individual failure modes are either perfectly
correlated or perfectly uncorrelated. The probability of system failure, presuming all
the modes are mutually independent, is an upper bound. The lower bound on the
Search WWH ::
Custom Search