31
International RILEM Conference on Materials, Systems and Structures in Civil Engineering
Conference segment on Service Life of Cement-Based Materials and Structures
22-24 August 2016, Technical University of Denmark, Lyngby, Denmark
(2)
3.
Reliability concepts to model the corrosion initiation due to carbonation
The safety margin (M) or limit state function for corrosion initiation can be defined as Eq. (3),
where ‘t’ is the exposure period, ‘X’ is random variables.
M = G(X, t)= T
i
– t (3)
Assuming that a system fails when corrosion initiation takes place, the probability of
corrosion initiation (
P
f
) is evaluated by integrating G(X, t) 0 over the failure domain,
considering the statistical distribution of each random variable. Eq. (4) shows how to estimate
’P
f
’ using the joint density function
fx(x
i
) with random variables [5].
(4)
In addition, Monte Carlo simulation can be used to calculate ‘P
f
’ by simulating the limit state
function for a range of sampling. In this approach, the mean value of I (x
i
) can be an estimator
for the probability of corrosion initiation as given in Eq. (5).
(5)
where
In estimating ‘P
f
’, it is vital to identify the basic set of random variables, of which
uncertainties have to be considered. Then, the randomness of all the variables is modelled,
recognizing the probability distributions of the variables. These probability distributions can
be defined by physical observations, statistical studies, laboratory analysis, and expert opinion
[5, 6, 7].
4.
Case study: condition assessment of reinforced concrete consoles
4.1 Statistical quantification of random variables and prediction of corrosion initiation
Three consoles, with almost no visible corrosion damage, were chosen to be tested 53 years
into their service life. They are located on the fourth floor in a residential building in Bergen,
Norway (about 2-3 km away from sea and 50 m above sea level); one of the consoles is
shown in Figure 1. While assessing the present condition of the consoles, it is necessary to
collect existing data. The existing data includes general information about the structure,
material properties and exposure condition, documentation of former inspections/monitoring
32
International RILEM Conference on Materials, Systems and Structures in Civil Engineering
Conference segment on Service Life of Cement-Based Materials and Structures
22-24 August 2016, Technical University of Denmark, Lyngby, Denmark
and documentation of former maintenance and repair. In the absence of information about the
actual cement type and water/cement ratio, cement type CEM I R and w/c 0.5 are chosen for
this analysis. The probability distributions for the random variables (R
-1ACC
,,
t
, RH
real
, C
s,atm
,
b
c
, b
w
, X
cover
) for Eq. (2) were chosen, taking into consideration the actual exposure condition,
as given in Tab. 1. Based on the exposure condition to rain, the probability of corrosion
initiation has been found for two cases: sheltered from rain and partially exposed to rain, as
shown in Figure 1.
Sheltered from rain
Partially exposed to rain
Figure 1: Reinforced concrete Console 1 in the building
Table 1: Statistical characteristics of variables
Parameter
Distribution
Mean
Standard
deviation
Reference
R
-1
ACC,
(10
-11
*mm
2
/s/kg/m
3
)
Normal
6.8(w/c=0.50
and CEM I R)
2.5
fib Bulletin 34[1]
t
(mm
2
/s/kg/m
3
)
Normal
315.5 48
fib Bulletin 34[1]
k
t
Normal
1.25 0.35
fib Bulletin 34[1]
ke =
RH
real
RH
ref
f
e
g
e
WBmax(w=1)
Constant
Constant
Constant
0.825
0.65
2.5
5.0
0.032
-
-
-
From weather
station
fib Bulletin 34[1]
C
s
=C
s,atm
+C
s,em
(kg/m
3
)
C
s,atm
C
s,em
Normal
-
0.0008
0
0.0001
fib Bulletin 34[1]
k
c
b
c
t
c
(days)
Normal
Constant
-0.567
1
0.02
-
fib Bulletin 34[1]
W(t) Exposed to rain
t
0
(years)
ToW
b
w
Constant
Constant
Normal
0.0767
0.010
0.446
-
-
0.163
fib Bulletin 34[1]
fib Bulletin 34[1]
Sheltered from rain
W(t) Constant 1
X
cover
(mm)
Normal
25
8