Determination of Permissible Chloride Levels in Prestressed Concrete
Determination of Permissible Chloride Levels in Prestressed Concrete
Determination of Permissible Chloride Levels in Prestressed Concrete
Determination of
Permissible Chloride Levels
in Prestressed Concrete
David Stark
Principal Research Petrographer
Construction Technology Laboratories
A Division of the Portland Cement Association
Skokie, Illinois
106
stress level in the steel. Water-cement
ratio of the concrete influences corro-
Synopsis
sion as it affects moisture, oxygen, and
chloride diffusion rates. This report describes a labora-
The above findings can be extended tory investigation to determine
to field structures to explain purported chloride concentrations and expo-
corrosion-free performance of pre- sure conditions required to induce
stressed concrete members made with corrosion of prestressed tendons in
high levels of chloride ion. In these concrete beams.
cases, either drying has increased the Variables included source and
electrical resistivity of the concrete suf- concentration of chloride ion, wet-
ficiently to prevent flow of galvanic cur- ting and drying, C3A content of the
rent between potential anodic and cement, water-cement ratio of the
cathodic sites on the steel, or, under concrete, stress level in the steel,
uniform moist conditions, sufficient dis- and concrete curing method.
solved oxygen has not been present to Performance was evaluated by
sustain active corrosion. It should be electrical potential measurements,
noted that unforeseen changes in these visual examination of tendons, and
environmental conditions can introduce measurement of chloride contents.
a high risk of development of active cor- Recommendations are made for
rosion and should be considered. permissible chloride levels in pre-
To eliminate the risk of active corro- stressed concrete as they are af-
sion, it is recommended that the permis- fected by exposure conditions.
sible water soluble chloride ion content
in prestressed concrete not exceed 0.10
percent by weight of portland cement.
cement corresponds to about 1.0 percent
total chloride ion by weight of cement.
INTRODUCTION The ACI limit is based primarily on
estimates by George J. Verbeck* and ap-
Under certain exposure conditions, an
environment may develop that is con- pears to include a "safety factor" be-
ducive to galvanic corrosion of pre- cause of possible catastrophic results if
stressing steel in pretensioned concrete corrosion occurs.
members. This type of corrosion re- In view of the uncertain basis for
quires moisture, oxygen, and sufficient chloride limitations, an investigation
chloride ion at the surface of the steel to was undertaken to address two pertinent
sustain electrochemical corrosion reac- questions:
tions. The required chloride ion level 1. What is the permissible chloride
may be reached through the use of level in concrete, above which corrosion
chloride-bearing admixtures such as of prestressing steel occurs?
calcium chloride, or by migration 2. Under what exposure conditions
through concrete of chloride from exter- can calcium chloride admixture in con-
nal sources such as seawater and deicing crete induce corrosion of prestressing
salts. steel? Or conversely, under what condi-
The American Concrete Institute tions does it not induce corrosion?
(ACI) states that water soluble chloride A two-phase program was undertaken
in prestressed concrete should not ex- to obtain answers to these questions.
ceed 0.06 percent by weight of cement.'
Putting this figure into more practical *Personal communication. Mr. Verbeck was for-
terms, a 2.0 percent calcium chloride merly Director of Research, Portland Cement Associa-
(CaCl 2 • 2H20) addition by weight of tion, Skokie, Illinois.
108
C)
C- Tnhlo I Vnrinhla rnmhinatinns in test Droaram.
0
C
Cement, Water- Stress Exposure conditions
Z
D percent cement Percent level,
C- Beam C3A ratio CaCl2 • 2 =O* Cure ksi Initial 10 to 12 months Final 11 months
C
C
2 11 0.35 2.0 Steam 0 Continuous damp ½ beam dry,' beam damp
C
3 11 0.35 1.0 Steam 80 Continuous damp Remain as is
4 11 0.35 2.0 Steam 80 Continuous damp Remain as is
5 11 0.35 1.0 Steam 160 Continuous damp 10 days dry + 4 days damp
6 11 0.35 2.0 Steam 160 Continuous damp 10 days dry + 4 days damp
7 11 0.35 2.0 14 days moist 160 Continuous damp Remain as is
8 8 0.35 2.0 Steam 160 Continuous damp 10 days dry + 4 days damp
9 11 0.50 2.0 Steam 160 Continuous damp 10 days dry + 4 days damp
11 0.50 2.0 14 days moist 160 Continuous damp Continuous dryt
10
0.35 None Steam 0 Pond -4 percent NaCI soln. 10 days dry + 4 days 8 percent NaCI
11 11
12 11 0.35 None Steam 80 Pond -4 percent NaCl soln. 10 days dry + 4 days 8 percent NaCl
13 11 0.35 None Steam 160 Pond - 4 percent NaCl soln. 10 days dry + 4 days 8 percent NaCl
14 8 0.35 None Steam 160 Pond -4 percent NaCI soln. 10 days dry + 4 days 8 percent NaCI
15 11 0.35 2.0 Steam 160 Pond -4 percent NaCI soln. Continuous dryt
16 11 0.35 Aggr. sat. Steam 160 Continuous damp Remain as is
4 percent NaC1
17 11 0.35 2.0 Steam 160 Continuous dry-So percent RH 10 days dry + 4 days 8 percent NaCI
18 8 0.50 0.17 Steam 160 Continuous damp 10 days dry + 4 days 8 percent NaCI
19 11 0.35 None 14 days moist 160 Pond -4 percent NaCl soln. 10 days dry + 4 days 8 percent NaCI
*Expressed as percent by weight of cement. Metric (5I) convention factor: 1 ksi = 6.895 MPa.
tTwo 1V2-in. (38 mm) diameter holes were drilled to two tendons and filled with water or 4 percent NaC1 solution for short-teem study of localized shifts in potential.
Details are given in full report.
0
CD
Table 2. Summary of concrete mix design and strength data.
Compressive
strength, psi*
Cement
C
factor, Water- Calcium
Beam lb per cement chloride, Method Percent
No. cu yd ratio percent of cure air 1 day 14 days
1 705 0.35 1.0 Steam 5.2 5510 -
2 705 0.35 2.0 Steam 5.5 5240 -
3 705 0.35 1.0 Steam 5.4 3910 -
4 705 0.35 2.0 Steam 6.2 4290 -
5 705 0.35 1.0 Steam 5.0 4790 -
6 705 0.35 2.0 Steam 5.7 4010 -
7 705 0.35 2.0 Moist 5.5 - 5440
8 705 0.35 2.0 Steam 5.5 5910 -
9 493 0.50 2.0 Steam 6.3 3510 -
10 493 0.50 2.0 Moist 6.0 - 4980
11 705 0.35 0 Steam 6.2 4270 -
12 705 0.35 0 Steam 5.6 4300 -
13 705 0.35 0 Steam 5.8 4730 -
14 705 0.35 0 Steam 5.6 5550 -
15 705 0.35 2.0 Steam 6.5 4190 -
16 705 0.35 Sat. Agg.t Steam 5.0 4290 -
17 705 0.35 2.0 Steam 6.0 4070 -
18 493 0.50 0.17 Steam 5.0 3460 -
19 705 0.35 0 Moist 5.2 - 5830
*Corrected to 6 x 12 in. cylinder strengths. All data are the average of two companion cylinders.
tAggregate saturated with 4 percent NaCI solution.
Metric (SI) conversion factors: 1 lb per cu yd = 0.5933 kg/m 3 ; 1 psi = 0.006895 MPa; 1 lb = 4.448 N; 1 in. _
25.4 mm.
110
and polyethylene for 14 days. Each day (76 x 152 mm) concrete cylinders that
during this period, top surfaces of the were cast and cured with the beams.
beams were wetted to minimize desic- Dry powder samples were obtained
cation due to cement hydration. from the beams at depths of 3/4 to 1 1/4 in.
After curing, and following load re- (19 to 32 mm) from the top surface to
lease and removal of forms, all beams determine chloride concentration at
were transferred to a test room main- about the level of the upper tendons.
tained at 73 ± 3 F (22.8 ± 1.7 C) and 50 Chemical analyses were made accord-
± 5 percent relative humidity. Steam ing to procedures described by Ber-
cured beams were stored under this man. 4 Nonevaporable water contents
condition for 27 days, while 14-day were measured on each powder sample
moist cured beams were stored under to correct for nonuniformities in paste-
the same condition for 14 days prior to aggregate ratios among the samples.
start of the test period.
7 2 percent Continuous damp Small areas 2 in. long with film of cor-
rosion product scattered along entire
length. Film of corrosion product 2 ft
long near one end of each tendon.
8 2 percent Continuous damp-* Few spots, about' in. long, with film
10 days dry plus 4 days damp of corrosion product scattered full
length on each tendon.
the test period. One tendon in Beam 11 tained coarse aggregate presoaked in 4
displayed corrosion products. In this percent NaCI solution, displayed local-
case, a crack had formed over the tendon ized corrosion after 79 weeks under the
prior to indications of active corrosion. continuous damp exposure condition. In
Tendons from Beam 16 which con- Beam 18, which was made with 0.17
112
Table 3 (cont.). Summary of observations of prestressing tendons.
Beam CaCl2•
No. 2H20* Exposure conditions Observations of upper tendons
15 2 percent Pond 4 percent NaCl soln.–s Localized areas up to 6 in. long with
Continuous dryt film of corrosion product 2 ft from one
end. Occasional pitting scattered full
length of tendons.
0 0.30
4-
0.20
2-0
a
WIC - 0.35
0.10
Z C3A- I I %
o Stress-80 ksi
o 0 Cure-Steam
IDay 10 20 30 40 50
AGE, WEEKS
Fig. 1. Potential curves for beams made with 0, 1, or 2 percent calcium chloride
admixture.
cal potential measurements are sum- soaking the coarse aggregate in 4. per-
marized in Figs. 1 to 4. In Fig. 1, com- cent NaCI solution to simulate seawater
parisons are made for beams made with immersion. The potential curve shows
0, 1, or 2 percent calcium chloride ad- the development of an initial period of
mixture, by weight of cement. Results active corrosion that terminated at about
indicate that, where 1 or 2 percent cal- 15 weeks. This illustrates that water sol-
cium chloride was used, active corrosion uble chloride as low as 0.17 percent,
developed within 1 day of fabrication of even though introduced with the aggre-
the beam. However, it terminated gate, can initiate active corrosion.
within 3 to 10 weeks. This is indicated Fig. 2 also shows the potential curve
by shifts in potential to values less for Beam 18, which was made with 0.13
negative than –0.30V. percent total (0.11 percent water sol-
Potentials for the beam made without uble) chloride, by weight of cement. In
calcium chloride remained less negative this case, active corrosion was not initi-
than –0.30V, thus indicating that active ated until the period of ponding with
corrosion had not developed during the NaCI solution. Thus, 0.13 percent
test period. Results for this beam also chloride by weight of cement was not
indicate that sufficient chloride ion from sufficient to initiate corrosion.
the solution ponded on the top surface The effect of prolonged drying, at 73
of the beam did not reach the tendons to ± 3 F (23 ± 1.7 C) and 50 ± 5 percent
later initiate active corrosion. relative humidity, on corrosion resis-
Fig. 2 shows results for Beam 16, tance is indicated in Fig. 3 by the po-
which contained 0.20 percent total (0.17 tential curve for Beam 17, which was
percent water soluble) chloride by made using 2 percent calcium chloride
weight of cement. Virtually all of the admixture. Initial potential was –0.48V
chloride was introduced through pre- which indicates the development of ac-
114
BEAM NOS. 16 AND 18
0.50
w No. 16 Continuous
(n
0 Continuous Damp No.1 1 8 10 Days Dry
U) 040 I^ 4 Days Pond-8% NaCI
0.30 -- ------- --
0
No. 16-0.20% CI
, Pre-Soaked Agg^
0.20
No. 16 - 11%C3A
Z 0.10 No. 18-0.13% CI 0.35 W/C
IDay 20 40 60 80 o
AGE, WEEKS
Fig. 2. Potential curves for beams with 0.13 and 0.20 percent total chloride ion.
tive corrosion. During the ensuing 50- dons. This observation does not indicate
week continuous dry period, potentials whether corrosion occurred during the
progressively shifted to less negative first 20 weeks of drying, during the
values, and reached –0.18V as drying cycling period, or during both periods.
was terminated. However, potentials indicate that active
This long-term shift indicates that ac- corrosion occurred during two different
tive corrosion had stopped within about periods of time.
20 weeks. The tendons then remained Fig. 4 reveals the effects of differen-
in the passive state for the remaining tial drying on corrosion resistance in
30-week duration of the drying period. Beams 1 and 2. Beam 2 was made using
Following the period of continuous a 2 percent calcium chloride admixture.
drying, the beam was subjected to cy- After the initial 50-week period of con-
cles of 10 days drying and 4 days pond- tinuous damp exposure, one-half [6 ft
ing with 8 percent NaCI solution. The (1.83 m)] of the top surface of the beam
curve in Fig. 3 indicates that, within 1 or was exposed continuously to 50±5 per-
2 weeks of this change, potentials cent relative humidity, while the other
shifted to approximately –0.35V. This half remained in the continuous damp
indicates recurrence of active corrosion. exposure.
Potentials then remained at about Beam 1, which was similar to Beam 2
–0.33V for the duration of the 42-week except that it was made using 1 percent
cycling period. This indicates that active calcium chloride admixture, was sub-
corrosion continued throughout this jected to the same change in exposure
period. conditions. No additional chloride was
Visual examination revealed the introduced into either beam.
presence of a layer of corrosion product Two potential curves are shown for
along the entire length of the upper ten- each beam after the change in exposure
0.20
W/C-0.35
Calcium Chloride-2%
z 0.10 Stress-160 ksi
w Cure-Steam
0
a- n
I Day 20 40 60 80
AGE, WEEKS
Fig. 3. Potential curve for beam subjected to 50 weeks of drying followed by alternate
ponding and drying.
45 50 60 70 80 90
AGE, WEEKS
Fig. 4. Potential curves for two beams subjected to continuous half dry-half damp
exposure conditions.
116
condition; one for the dry half of each 3915 lbs per cu yd (2342 and 2322 kg/m3)
beam, and one for the damp half. Unlike were used to convert percent chloride
the similarity of potentials measured by weight of concrete to percent by
along tendons under uniform exposure weight of cement for the 0.35 and 0.50
conditions, differences in potential be- water-cement ratios, respectively.
tween the two halves of each beam de- From these data, an estimate can be
veloped soon after the change in expo- made of minimum chloride concen-
sure condition. In Beam 2, maximum trations required to induce corrosion of
potential differences reached 0.1OV prestressed tendons during and im-
(-0.28V vs –0.18V) in 5 to 10 weeks. A mediately following curing. The lowest
similar pattern developed for Beam 1, water soluble chloride level measured
except that maximum potential differ- at which corrosion was initiated during
ences reached only 0.06V. this period was 0.17 percent by weight
Visual examination of tendons from of cement. In this case, chloride ion was
these two beams revealed, in addition to introduced into the fresh concrete
the few localized deposits of corrosion through prior absorption by the coarse
product present along the full length of aggregate.
tendons from the two beams, a heavier, In contrast, corrosion did not develop
continuous film of corrosion product initially when measured water soluble
along only the 6-ft (1.83 m) length of chloride levels were 0.11 percent by
tendons embedded in concrete exposed weight of cement. Thus, the water sol-
continuously to the damp condition in uble chloride level required to induce
Beam 2. This was not seen along the corrosion of prestressed tendons during
tendons from Beam 1. and immediately following curing was
Thus, when 2 percent calcium in the range of 0.11 to 0.17 percent by
chloride admixture was used, prolonged weight of cement. Respective total
periods of damp exposure followed by chloride concentrations were 0.13 per-
differential drying of the concrete in- cent and 0.20 by weight of cement.
duced corrosion of prestressed tendons. The data also indicate that chloride
Potential curves (not shown) for other levels required to induce corrosion at
beams reveal that C 3 A content of the later ages were strongly dependent
cement, stress level in the steel, water- upon type of exposure condition. For
cement ratio of the concrete, and curing example, under uniform drying condi-
method, had little effect on initial corro- tions of 50 percent relative humidity at
sion resistance where 1 or 2 percent cal- 70 to 75 F (21 to 24C), corrosion failed to
cium chloride admixture had been used. develop in concrete containing 0.87
In all cases, only initial periods of active percent water soluble chloride ion by
corrosion, similar to those shown in weight of cement. Most of this chloride
Figs. 1 and 2, developed, which were was introduced into the concrete
terminated in less than 30 weeks. through use of 2 percent calcium
chloride admixture. Continuous expo-
sure of companion concrete beams to
Chloride Contents uniformly damp conditions following
Table 4 summarizes the results of ini- curing and 1 month of drying also failed
tial and final chloride content mea- to sustain active corrosion.
surements. Values are expressed as per- In contrast, exposure to nonuniform
cent by weight of cement, and have drying conditions permitted develop-
been corrected for differences in paste- ment of corrosion. In this instance, 2
aggregate ratios by averaging percent calcium chloride admixture, by
nonevaporable water contents. Mea- weight of cement, had been utilized.
sured concrete unit weights of 3950 and Corrosion did not develop under similar