Service-life modelling for chloride-induced corrosion - Concrete Repair - page 255

DuraCrete, ageing coefficient values under field conditions were determined
from profiles taken from structures and exposure tests, combined with
k -values (equation (15.2)) for different environments and lengths of wet
curing. Based on those data, the present analysis and additional work, in
particular on blast furnace slag cement (Polder and Rooij, 2005; Rooij and
Polder, 2005), n -values were chosen in two groups of environmental classes:
very wet (XD2/XS3; also XS2) and moderately wet (XD1/XD3/XS1), see
Table 15.1.
The time needed to reach the critical chloride content at a certain depth
can now be calculated for any given mix (with a composition covered by the
available database) in exposure classes XS and XD, using equations (15.1),
(15.2) and (15.3) indicatively. For a specific service-life design calculation,
D RCM of a particular mix should be measured and the test result should be
used for the calculation.
15.3.4 Reliability considerations and semi-probabilistic approach
For a given concrete cover depth, equation (15.2) can be used for calculating
the time needed for the critical chloride content to reach the reinforcement.
A matching diffusion coefficient for the intended concrete at reference time
D RCM,28 can be calculated. Such a calculation, however, is deterministic and
yields a mean value. This means that the probability of corrosion initiation at
that point in time is 50%. Consequently, the structure could have corrosion
initiation over 50% of its surface! In practice such a high probability is
unacceptable. An acceptable probability of failure for an MLS may be 10%
for reinforcing steel (Fluge, 2001).
To obtain such a lower probability of failure (10%) either the cover
depth can be increased or the maximum D RCM,28 can be decreased. The latter
option has serious disadvantages due to concrete technological limitations.
If the former option is chosen, the required amount of additional cover
can be calculated for each individual case using probabilistic calculations.
This option is provided by the guideline, among others, by specifying
mean values, standard deviations and distribution types for all parameters.
However, another option given by the guideline is based on adding a fixed
amount to the (deterministically determined) cover depth as a safety margin .
This is a semi-probabilistic approach, comparable to using a safety factor
for a materials property or a load. Calculations using TNO's probabilistic
software Prob2B TM have shown that a safety margin of 20 mm results in a
probability of failure of about 10%; a safety margin of 30 mm produces a
probability of about 5%. Such probabilities are considered appropriate for
corrosion initiation of reinforcing and prestressing steel, respectively.