ICOLD Bulletin: The Physical Properties of Hardened Conventional Concrete in Dams
Section 2 (Strength properties)
As submitted for ICOLD review, march 2008 Section 2-17
placement starts. This is not always possible. In order to overcome the impasse, it is
often decided to take 28/90 day as characteristic strength or to conservatively
extrapolate these early strength data. In most cases, such decisions lead to
excessive cement content, which is uneconomic and, as aside effect, contributes to
unfavourable thermal conditions. Of course, the obvious remedy is to start early with
concrete testing. However, this needs planning, discipline and is often not done
because of unknown final concrete constituents.
However the characteristic age have to be always related to the specific dam under
examination, taking into account the scale of the structure and the times of
construction and service life.
This Bulletin would like to encourage the dam profession to also view the problem
from a more pragmatic standpoint. Research, published data from case histories,
and preliminary series with similar than the final mix design from the dam in question
provide a host of strength-gain values to judge strength gain up to the age of the
characteristic strength. Certainly, such a judgement has to be accompanied by
assessing the consequences of an anticipated error between assessed and later
measured characteristic strength: What, if the required strength is not met? Are the
stipulated factors of safety (FS) still acceptable? But: isn't FS=3 just an arbitrary
chosen number and FS=2.8 also acceptable? Can a local redistribution of stresses
be assumed beneficial in mitigating a lacking safety margin? How much is the gain
of strength between one year and the age when the calculated stresses are really
occur, e.g. during extreme temperatures, floods or an earthquake?
In many cases the limiting strength is not compression but shear or tension, mostly
along local zones, and shear/tensile strength is generally taken as a percentage of
compressive strength, not measured per se. This percentage is highly speculative
(particularly for dynamic loading) and uncertain in a similar order of magnitude than
speculating a particular strength gain between 90 days and one year.
What this all means is that engineering judgement and peer review, based on an
increasing data-base, often is a remedy to justify acceptance of uncertainties. This
can be beneficial for economy and safety. The shortcut of immediate conservative
solutions is not necessarily the only safe way to go.
• Evaluating strength data, the most common design criterion, is based on the
following formula [2.23]:
f
m
= f
c
+
k.σ
f
m
the mean required strength for which the mix has to be
designed
f
c
the characteristic strength (specified design strength to meet
required safety against stresses)
k a factor derived from the assumed strength distribution
which depends on a stipulated proportion of tests (fractile)
to fall below the level of f
c
, and on the number of tests carried out.
σ standard deviation of strength test series
The k-factor is derived from fractiles of a Normal distribution. Commonly used k
values (for high number of tests, generally n≥30) are reported in Tab. 2.8.