Proceedings of the International rilem conference Materials, Systems and Structures in Civil Engineering 2016

75

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 

 

coefficients were measured on concrete specimens under standard curing conditions (20°C 

and RH > 95%) for given ages, 28d and 56d, by RCM method.  

 

For tunnel segments, the target mean values for D

nssm

 are respectively 6.5×10

-12

m

2

/s and 

4.5×10

-12

m

2

/s for 28d and 56d. The mean values of measured D

nssm

 are 4.68×10

-12

m

2

/s (28d) 

and 2.95×10

-12

m

2

/s (56d), much lower than the required values. The variation coefficient of 

this D

nssm

 is 9.9% (28d) and 8.7% (56d), smaller than the expected level of 20% in design 

phase, especially for 56d age. For concretes in bridge structures, much lower mean values, 

compared to the required values, are obtained for both prefabricated and cast-in-place 

elements. The D

nssm

 dispersion of prefabricated bridge elements is very near 20% for both 

28d and 56d ages, but the variance ranges from 4% to 44% for cast-in-place elements. Most 

diffusivity values conform to the complaince critera (6) and (7). 

 

3.3  Concrete cover thickness 

The concrete cover was measured through the common midpoint (CMP) method and the 

principle of CMP method detects the depth of reinforcement bars through transmitting an 

electromagnetic wave pulse and receiving the reflected waves from the steel bars by two 

adjacent antennas. The measurement error of the device used in HZM project is evaluated as 

around 3mm through the comparison with the measurement on extracted cores from test 

segments of tunnel. For information, the compliance criteria for the construction errors of 

concrete cover are within -5mm/+18mm. 

 

For the prefabricated tunnel segments, the required values for thickness are respectively 

50mm (intrados) and 70mm (extrados). The measurements show that the mean values are 

53.5mm (intrados) and 73.4(extrados) and the construction error, corresponding to the 

distance between 5% percentile value and mean value, is 5.8mm (intrados) and 6.4mm 

(extrados), slightly larger than the expected construction tolerance, 5.0mm, for prefabricated 

elements. For the cast-in-place concrete elements in bridge structures, the design value for 

concrete cover thickness for bridge decks is 45mm, and the measured mean value attains 

50.5mm with only 1.11mm as standard deviation (SD) value; the design thickness for piers, in 

splashing zones, is 70mm and the measured values have 80.0mm as mean value and 6.74mm 

as SD value. The design values for bearing platforms, in immerged and splashing zones, are 

respectively 60mm and 80mm, and their measured values have mean values of 71.0mm and 

95.5mm and SD values of 6.85mm and 4.22mm.  

 

 

4. Durability 

assessment 

4.1 Assessment model 

The assessment model for chloride-induced corrosion is also adapted from the analytical 

model of Fick’s second law, similar to the design model in (1). With the corrosion initiation 

specified as DLS, the assessment equation writes, 

 

cr

0

0

0

Cl

0

1 erf

0

2

,

d

s

x

G

C

C

C

C

D

t t t

  

(8) 

---

76

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 

 

with C

0

 standing for the initial chloride concentration in concrete (%binder). Hereafter, this 

model is to be used through full probabilistic approach to evaluate the failure probability of 

durability, 

 

f

SL

SL

prob

0

p t

G t

  

(9) 

 

The service life is judged to be satisfied or not by comparing the failure probability at t

SL

 with 

the target failure probability p

target

. Thus, the assessment model contains six parameters: C

cr

, 

C

s

, C

0

, x

d

,  D

Cl

0

 and n(or  ). The statistical properties of the parameters C

cr,s

 and ageing 

exponent n have been investigated in depth on the basis of the long-term exposure tests in 

design phase [7], and the same statistical properties are retained in durability assessment. For 

the initial concentration C

0

, a rectangle, or uniform, distribution is calibrated for the structural 

concretes from the chemical analysis of raw materials. For the chloride diffusivity D

Cl

0

, the 

collected data are used to update the statistical properties, and the collected data are also used 

to update the statistical properties of concrete cover thickness.  

 

In the construction phase, different protection measures are adopted, including the silane 

impregnation on concrete surface, epoxy-coating on steel bars, stainless steel bars, coating on 

steel surface and cathode protection of steel bars in concrete elements. These measures are 

adopted to increase the durability safety margin and/or to compensate the possible adverse 

effect from construction defects. In durability assessment, these measures are taken into 

account after the following assumptions: (1) the silane impregnation is assumed to delay the 

accumulation of chloride on concrete surface to its stabilized value by 10 years in the 

assessment; (2) the epoxy-coatings have no impact on the corrosion initiation of 

reinforcement steel bars in the durability assessment; (3) the C

cr

 of stainless bars is 

conservatively taken as 5 times the value of the conventional carbon steel bars; (4) the 

cathodic protection passivates completely the reinforcement steel once it is put into function, 

keeping the steel in passivation during its active duration.  

 

4.2  Assessment through full probabilistic approach 

The fully probabilistic analysis for the durability assessment of RC/PC elements is realized 

through Monte-Carlo simulations. A computer-based program is developed especially to 

perform the probabilistic assessment. In the simulation, six parameters are considered as joint 

occurrence random variables. For a given exposure age, the Monte-Carlo simulations are 

performed to calculate the failure probability of (9), and 1,000,000 samplings are used to 

ensure the solution of “real” probability. Accordingly, the failure probability is solved with 

time from t=0 to t=120 years. Some assessment results are presented in Figure 2 for bridge 

elements in splashing and tidal zones. 

 

Globally the concrete elements all achieve reliability index above 2.0 at 120 years. The bridge 

elements in atmospheric zones have very low failure probability, i.e. p

f

 <10

-3

 at 120 years and 

the corresponding reliability index  >3.0. The silane impregnation is generally adopted as 

protection measure for elements, the stainless bars are adopted for pylons and the epoxy-

coated bars are used respectively for decks and box girders. The bridge elements in 

splashing/tidal zone, piers and bearing platforms, have satisfactory durability safety margin: 

---