Beruflich Dokumente
Kultur Dokumente
78 Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012
1 000 shaped rather than circular, as used by many
other researchers (Liu & Weyers 1998; El
Transverse and vertical strains (micro strains)
Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012 79
et al 1998) to 10400 μA/cm2 (Almusallam does not give partial surface steel corrosion side face. Interestingly, when the level of
et al 1996)] between researchers. Assuming which Yuan & Ji (2009), Yuan et al (2007) steel corrosion was increased, a third crack
proportional structural damage due to accel- and Jang & Oh (2010) contended to be more appeared next to a corroding bar, but on a
erated tests, a level of damage caused by a representative of in-service conditions. face that was uncracked. This indicates that
current density of 3 μA/cm 2 over a period of El Maaddawy & Soudki (2003) conducted the pattern of corrosion cracks changed with
one year can be obtained within two hours another research to find an impressed an increase in the level of steel corrosion.
when a current density of 10400 μA/cm 2 is current density that can produce desired Similar change in crack patterns as steel cor-
used. Should accelerated tests give propor- structural damage in a short time without rosion increased was reported by Malumbela
tional damage then there is little value in excessively altering structural response et al (2010a). They found that when a second
using low rates of steel corrosion. under natural steel corrosion. In their work crack appeared, the first crack ceased to
To assess the effect of varying current the researchers used rate of widening of cor- widen. Certainly, specimens that exhibit
densities on the proportion of structural rosion cracks and average mass loss of steel this crack pattern will have narrower cracks
damage, Mangat & Elgarf (1999) measured at the end of corrosion tests as parameters than those that only exhibit a single crack.
slopes of load-deflection curves of RC that indicate corrosion damage. Crack widths This was also found by Zhang et al (2010).
specimens that were corroded with corrosion were measured at one point on side faces of This argument certainly questions the use of
rates from 1000 to 4000 μA/cm2. The con- specimens using a demountable mechanical crack widths without full understanding of
crete mix that they used for RC specimens (demec) gauge with a gauge length of 50 mm. crack patterns to specify the level of current
had 1% NaCl salt by cement weight, and dur- Depassivation of steel was accelerated by density to be used in accelerated corrosion
ing the accelerated test their specimens were mixing concrete with 5% NaCl by weight of tests. Bear in mind that current density used
immersed in a 3.5% NaCl solution. At lower cement. Current densities assessed ranged by Malumbela et al (2010a) was 189 μA/cm2
levels of steel corrosion (<10% mass loss), from 100 to 500 μA/cm2. which is within the limit proposed by
Mangat & Elgarf (1999) found little influence They found that at corrosion crack widths El Maaddawy & Soudki (2003).
of corrosion rates on the stiffness of cor- below 0.03 mm (which corresponded to a Contrary to findings by El Maaddawy &
roded specimens. However, at mass losses theoretical mass loss of steel from Faraday’s Soudki (2003), Alonso et al (1998) found that,
of steel above 10%, slopes of load-deflection Law of 0.8%), specimens corroded using for a chosen crack width, higher penetration
curves or stiffness of corroded RC specimens various current densities which exhibited depth (more than three times larger) of steel
were severely influenced by the rate of steel a similar rate of expansion of the cover was needed when applying a current density
corrosion. For example, at a mass loss of steel concrete. At larger crack widths (>0.03 mm), of 100 μA/cm2 than when applying a density
of 15%, a specimen that was corroded with they found specimens subjected to cur- of 3 μA/cm2. This implies that crack widths
a current density of 4000 μA/cm 2 needed rent densities above 350 μA/cm 2 to exhibit increase faster with a lower corrosion rate
a load of 26 kN to cause a deflection of a significantly larger rate of widening of (3 μA/cm2) than with a higher corrosion rate
3 mm (stiffness = 8.7 kN/mm). At the same corrosion cracks (up to four times) than (100 μA/cm2). Somewhat similar to findings
level of steel corrosion, a similar specimen specimens that were subjected to current by Alonso et al (1998), Malumbela et al
that was corroded with a current density of densities below 200 μA/cm 2. Interestingly (2010b) found that, at high sustained loads,
1000 μA/cm2 exhibited a deflection of 3 mm though, they found that regardless of the when steel corrosion is firstly accelerated
at a load of 38 kN (stiffness = 12.7 kN/mm). level of impressed current density used, and then allowed to run naturally, the rate
This indicates that at the same level of steel average mass losses of steel at the end of of the widening of corrosion cracks does not
corrosion, there was a larger loss in stiffness corrosion tests were within 4% of theoretical change, but the rate of steel corrosion reduc-
of specimens that were corroded under a mass losses of steel predicted from Faraday’s es significantly. Malumbela et al (2010b)
higher current density. Therefore, if loss in Law. These results indicate that if corrosion attributed this to natural steel corrosion
stiffness were to be used to predict levels of crack widths were to be used to predict levels producing dryer products that are not easily
steel corrosion, its relation with corrosion of steel corrosion, relations between crack exuded to the exterior faces of concrete.
levels from accelerated tests would result in widths and mass loss of steel from highly Results by El Maaddawy & Soudki (2003),
engineers underestimating levels of steel cor- accelerated tests would underestimate steel Mangat & Elgarf (1999), Malumbela et al
rosion in in-service structures. corrosion in in-service structures. Assuming (2010a,b) and Alonso et al (1998) indicate
Mangat & Elgarf (1999) asserted that a linear increase in mass loss of steel and that the effect of the level of current density
for accelerated corrosion tests in laboratory crack widths, under highly accelerated tests on structural behaviour is contentious.
specimens, especially when the target level (500 μA/cm2), they found a crack width of Further research to clarify this is therefore
of steel corrosion is high, the lowest practical 1 mm to correspond to mass loss of steel of necessary.
corrosion rate should be used to accelerate 7.3%. However, at lower corrosion rates (100
reinforcement corrosion. Since they used μA/cm2 and 200 μA/cm2) a crack width of
corrosion rates that ranged from 1000 to 1 mm was found to correspond to a mass TYPE OF CATHODES
4000 μA/cm2, it is reasonable to assume loss of steel of 13.3%. El Maaddawy & Soudki Whilst anodes are simply steel bars that
from their results that a corrosion rate that (2003) concluded that an impressed current are required to corrode, various types of
is below 1000 μA/cm2 is appropriate for density below 200 μA/cm 2 does not exces- cathodes have been used in accelerated
laboratory tests. One of the drawbacks with sively alter the structural performance of corrosion tests. In some research, metal
this work is that it used parameters that were corrosion-affected RC specimens that would bars embedded in concrete were used
only measured at the end of the corrosion be observed under natural steel corrosion. (El Maaddawy et al 2005b; El Maaddawy
process. It therefore does not provide the It should be mentioned that El Maaddawy & Soudki 2003; Badawi & Soudki 2005).
influence of the rate of steel corrosion on the & Soudki (2003) initially observed two corro- In others, metal bars were placed on
much-needed rate of change of parameters sion cracks near each corroding bar and each external surfaces and inside a chloride salt
with an increase in the level of steel corro- crack propagated parallel to the bar. These electrolyte (Malumbela et al 2009; Ballim
sion. In addition, the corrosion process used cracks were either on the top face or on the et al 2001). These bars were of different
80 Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012
compounds, are often found in in-service
Pure iron α – Fe structures, as well as in laboratory specimens
Wustite FeO where steel corrosion is natural (François &
Magnetite Fe3O4 Arliguie 1998; Castel et al 2003; Vidal et al
Corrosion products of iron
Hæmatite α – Fe2O3
2007; Zhang et al 2009a,b; Zhang et al 2010;
Malumbela et al 2010b). Since these products
Maghemite γ – Fe2O3
are of different volume densities, the rate of
Feroxyhite δ – FeOOH
widening of corrosion cracks is expected to
Goethite α – FeOOH be greatly influenced by the procedure used
Lepidocrocite γ – FeOOH to accelerate steel corrosion. It is important
Akaganeite β – FeOOH to observe that large densities belong with
more soluble products. Therefore, at the same
Ferrous Hydroxide Fe(OH)2
level of steel corrosion, specimens that exhibit
Ferric Hydroxide Fe(OH)3
unstable corrosion products are expected
Fe2O3 – 3H 2O to be more severely cracked than those with
0 1 2 3 4 5 6 7 more stable products. This is in agreement
Unit volume with results from El Maaddawy & Soudki et
al (2003). On the same note, soluble products
Figure 2 Corrosion products of iron (Jaffer & Hansson 2009) easily exude the corroding area and therefore
relieve the cover concrete of applied pressure.
dimensions and made from different metals. The type of corrosion product was However, drier products do not easily egress
The most used metals, such as stainless found to be primarily dependent on pH and the corrosion region and hence sustain the
steel, copper and titanium, had good electri- availability of oxygen (Roberge 1999; Jaffer pressure. This argument is in agreement with
cal conductivity. Rather than using bars, & Hansson 2009; Broomfield 1997). These results from Malumbela et al (2010b) and
other researchers chose to use metal plates factors (pH and quantity of oxygen) are very from Alonso et al (1998). Research on the
or mesh that covered the external faces of variable and difficult to quantify in a corrod- chemical composition of corrosion products
concrete (Azad et al 2007; Rio et al 2005; ing RC structure. Many researchers contend from accelerated corrosion and from natural
Gadve et al 2009; Fang et al 2004). that for corrosion of steel that is embedded corrosion tests, and how they affect cover
Placing cathodes inside concrete means in concrete, ferrous hydroxide is the fun- cracking, is needed.
hydroxyl ions are produced inside the con- damental corrosion product (Liu & Weyers
crete and then moved to the anodic steel. 1998; El Maaddawy & Soudki 2007; Bhargava
The rate of corrosion here is dependent on 2006; Roberge 1999). With an increase in the ACTUAL LOSS OF STEEL
how well oxygen can penetrate the concrete, supply of oxygen (especially after cracking DURING CORROSION TESTS
as well as how well hydroxyl ions penetrate of the cover concrete), more stable corrosion As already mentioned, structural engineers
the concrete to reach the anodic steel. When products such as haematite and magnetite and asset managers often rely on measur-
cathodes are placed externally, hydroxyl ions are formed. able parameters of corroding RC structures,
are no longer produced inside the concrete. Varying procedures of accelerated corro- such as corrosion crack widths and stiffness,
It was pointed out by Poursaee & Hansson sion tests is therefore likely to influence types to predict levels of steel corrosion, as well
(2009) that, from an electrochemical view- of corrosion products formed. For example, as their residual load-bearing capacities.
point, this is not acceptable, as hydroxyl ions when specimens are fully immersed in NaCl This involves using relations developed by
under natural corrosion are produced inside solution, Hussain (2010) has shown that mois- researchers such as a relation between cor-
the concrete. In the situation where cathodes ture blocks the pores of concrete, and hence rosion crack widths and mass loss of steel.
are placed externally, the rate of corrosion prevents oxygen from diffusing into the con- To confirm these relations, some researchers
is dependent on how well hydroxyl ions can crete to reach the anode. More soluble prod- have measured the actual level of steel cor-
penetrate the concrete. How cathodes should ucts, such as ferrous hydroxide, are therefore rosion at the end of accelerated corrosion
be designed to better represent in-service expected. In addition, when the rate of steel tests. This was done by removing corroded
conditions is unclear and most certainly corrosion is high (as in accelerated corrosion steel bars from concrete specimens, cleaning
requires further research. It is, however, tests), the rate of ingress of oxygen into the them, and measuring levels of steel corro-
reasonable to follow a recommendation by concrete might not be adequate to produce sion as mass losses of steel or as corrosion
Poursaee & Hansson (2009) that they should stable compounds. This helps to explain why pit depths. In real structures, however, it
be placed inside the concrete. in accelerated corrosion tests where speci- is uncommon for corroded steel bars to be
mens are immersed in salt solution, corrosion removed from structures. Faraday’s Law is
products that exude the concrete are often therefore often used to estimate the level of
TYPE OF CORROSION PRODUCTS greenish-black in colour, indicating a large steel corrosion. It is also extensively used in
DURING STEEL CORROSION presence of ferrous hydroxide (Malumbela modelling other parameters of corroding RC
One more parameter that needs discussion et al 2010b,c). On reaching the surface, they structures, such as time to first cover crack-
in designing corrosion tests in laboratories is immediately turn reddish-brown, indicating ing (El Maaddawy & Soudki 2007) and stiff-
the type of corrosion products. Researchers a conversion to the more stable compounds ness of corroded structures (El Maaddawy et
have detected various corrosion products in such as haematite and magnetite. When steel al 2005a). To relate measurable parameters
corrosion-affected RC structures, all with corrosion is slow and concrete is drier, oxygen of RC structures with the level of steel cor-
different densities and volume expansion is expected to be in abundance to form the rosion accurately, the suitability of Faraday’s
as shown in Figure 2 (Liu & Weyers 1998; stable products. Reddish-brown products, Law to estimate the level of steel corrosion
Roberge 1999; Jaffer & Hansson 2009). indicating a large presence of stable corrosion needs to be understood.
Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012 81
Figure 3 shows a plot of mass loss of steel 50
measured at the end of corrosion tests with
predicted mass loss of steel from Faraday’s Line of equality
Law. The data in Figure 3 was obtained
from various researchers in the literature. A 40
summary of conditions of the experiments
82 Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012
of steel and Faraday’s Law. Their corrosion corrosion region. On the other hand, more capacity was, however, found to be 78%.
process involved a current density of 189 μA/ stable products, such as haematite and mag- Azad et al (2007) found average mass loss of
cm2 and two different cycles of wetting of netite (which occupy less volume) could have steel of 1% to relate to loss in load-bearing
cover concrete with 5% NaCl solution, and formed during the four-day drying periods. capacity of 1.4%. The corresponding relation
natural drying under laboratory conditions. The formation of these lesser volumetric between mass loss of steel and theoretical
One accelerated process entailed four-day compounds could have allowed for more load-bearing capacity varied with the level of
wetting followed by two-day drying cycles access of corrosion agents to the rebar, which steel corrosion. At a corrosion level of 31%,
whilst in the other, cycles were all four days. could have led to larger corrosion rates. This theoretical load-bearing capacity exceeded
They only contaminated the tensile face of notion is in agreement with discussions by the measured capacity by 30%, but at lower
RC beams with salt solution. Their target Hussain (2010) on the effect of moisture levels of corrosion (around 5%) theoretical
mass loss of steel from Faraday’s Law was variation on rate of steel corrosion. capacity was found to be similar to the
10%. This meant 44 wetting days when the Figure 3 clearly indicates that Faraday’s measured capacity. The researchers (Mangat
current was impressed. At the end of the Law is not adequate to predict levels of steel & Elgarf 1999; Azad et al 2007) attributed
test, they measured both average mass loss of corrosion, particularly where sufficient the poor predictions of ultimate capacity of
steel and maximum loss. drying of cover concrete is permitted. Since beams at high mass losses of steel to losses
It is clear from Figure 3 that for beams natural steel corrosion often occurs under in the bond between corroded steel bars and
with two-day drying cycles, maximum mass drier conditions than most accelerated cor- the surrounding concrete. They therefore
losses of steel were largely greater than pre- rosion tests, Faraday’s Law is likely to under- developed necessary correction factors.
dicted losses. The largest loss in those beams estimate levels of steel corrosion in in-service According to Azad et al (2007), the residual
was 12.1% compared to 10% from Faraday’s structures. It is therefore recommended that load-bearing capacity of corroded RC beams
Law. Despite these larger mass losses of steel, further research be carried out to model the should be calculated using Equations 1 and
Figure 3 shows that they are still within the interaction between dryness of cover con- 2. In line with their experimental findings,
range of values that were observed by other crete and rate of steel corrosion. these Equations indicate that the needed-
researchers who measured average mass loss. correction-factor, α, reduces with an increase
However, their consistency, which did not in the level of steel corrosion (it).
exist in results from other researchers, points INFLUENCE OF CORROSION TEST
to the need to be cautious when predicting ON LOAD-BEARING CAPACITY Mu_actual = α Mu_theoretical (1)
maximum mass loss of steel using Faraday’s Mangat & Elgarf (1999) found that, at
Law. mass losses of steel due to steel corrosion up
14.7
Mass losses of steel in beams with four- to 7%, the level of current density had little α= ≤1 (2)
d(it)0.5
day drying cycles were certainly much larger effect on the load-bearing capacity of RC
than corresponding losses in beams with beams. However, at mass losses of 10% and Where
two-day drying cycles. The most obvious beyond, load-bearing capacity of RC beams Mu_actual = measured capacity of beams
reason that can be attributed to beams with decreased significantly with increase in the (kN-m)
longer drying cycles having larger mass level of the impressed current density. For Mu_theoretical = theoretical capacity of beams
losses of steel, is that longer drying cycles example, at a mass loss of steel of 20%, cur- based on reduced average
could have allowed for more natural corro- rent density of 1000 μA/cm2 induced a loss cross-sectional area of steel
sion to occur because of the extended time of load-bearing capacity of 60% compared to (kN-m)
required to reach the desired time of elec- 78% when a current density of 4000 μA/cm 2 α = correction factor
trolysis. As already mentioned, for beams to was used. They attributed this to a larger loss d = bar diameter (mm)
have the target level of steel corrosion of 10%, in the interfacial bond at the steel/concrete i = corrosion current density
beams corroded using two-day drying cycles interface caused by the high corrosion rates. (mA/cm2)
were tested for 64 days (44 wetting days + Contrary to findings by Mangat & Elgarf t = duration of corrosion (days)
20 drying days). Beams tested with four-day (1999), Azad et al (2007) reported that it
drying cycles were, however, tested for 80 was not the current density that caused a Torres-Acosta et al (2007) found a poor
days (44 wetting days + 36 drying days). This larger reduction in load-bearing capacities at relation between average penetration depth
implies that beams under the four-day dry- higher levels of steel corrosion, but rather the on steel bars (calculated from average mass
ing cycle had 16 days of additional natural product of current density with time. They loss of steel), due to steel corrosion and
corrosion compared to beams under the further asserted that a higher value of cor- the residual capacity of RC specimens. A
two-day drying cycles. It was later shown by rosion current density for a lesser period of cross-sectional loss of steel of 1% was found
Malumbela et al (2010b) that the natural cor- time would be as damaging as a lesser value to be equivalent to a loss in capacity of
rosion rate in beams was too low (30.4 μA/ of current density for a longer corrosion 1.6%. This relation is similar to the relation
cm2) to have resulted in the recorded large period. found by Azad et al (2007) where average
mass losses of steel in beams with four-day Where Mangat & Elgarf (1999) and Azad mass loss of steel was used. Torres-Acosta
drying cycles. The large differences in mass et al (2007) agreed was that, at large mass et al (2007), however, found a good relation
losses here could be ascribed to the set-up losses of steel (>10%), calculated values of (R2 ≈ 1) between the load-bearing capacity
with two-day drying cycles not allowing the load-bearing capacity, using measured aver- and maximum pit depths. From this relation,
complete dryness of the concrete cover at age mass losses of steel, had little relation it can be shown that a 1% maximum loss in
the corroded rebar depth. Therefore, after with experimental results. For example, area of steel yields a 0.6% loss in load-bearing
the drying period, less stable products such according to Mangat & Elgarf (1999), a capacity. Note that Torres-Acosta et al (2007)
as ferrous hydroxide (which according to mass loss of steel of 19% corresponded to presented their results using radius loss
Figure 2 occupy a larger volume than dryer a predicted loss in load-bearing capacity of instead of loss in cross-sectional area of steel.
products) would still be available within the 20%. The measured loss in the load-bearing They were converted here to allow them to be
Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012 83
compared with those from other researchers. when corrosion occurs on a drier cover Azad, A K, Ahmad, S & Azher, S A 2007. Residual
Interestingly, the relation found by Torres- concrete. Since corrosion in in-service strength of corrosion-damaged reinforced concrete
Acosta et al (2007) is similar to a theoretical structures involves long drying periods, beams. ACI Materials Journal, 104(1): 40–47.
relation between loss of steel cross-sectional it was recommended that laboratory Badawi, M & Soudki, K 2005. Control of corrosion-
area and load-bearing capacity of RC beams corrosion tests should also entail long induced damage in reinforced concrete beams using
developed by Ting & Nowak (1991). Even drying periods. Probably more research carbon fiber-reinforced polymer laminates. Journal
more intriguing, no correction factors, as is needed to standardise the duration of of Composites for Construction, 9(2): 195–201.
recommended by Mangat & Elgarf (1999) drying periods. Ballim, Y & Reid, J C 2003. Reinforcement corrosion
and Azad et al (2007) were needed in Ting & 3. Various types of cathodes are often used and deflection of RC beams – An experimental cri-
Nowak’s model. It therefore suggests that the when accelerating steel corrosion embed- tique of current test methods. Cement & Concrete
correction factors are limited to theoretical ded in concrete. Placing cathodes on Composites, 25: 625–632.
models of load-bearing capacity which use exterior surfaces of concrete was shown Ballim, Y, Reid, J C & Kemp, A R 2001. Deflection of RC
average mass loss of steel. More importantly, not to represent natural steel corrosion beams under simultaneous load and steel corrosion.
it implies that the loss in bond between steel well. It was recommended that cathodes Magazine of Concrete Research, 53(3): 171–181.
and concrete may not be the reason for the should be placed inside the concrete. Bhargava, K, Ghosh, A K, Mori, Y & Ramanujam, S
failure of the theoretical models. Further research on this was, however, 2006. Analytical model for time to cover cracking
Malumbela et al (2010c) showed that recommended. in RC structures due to rebar corrosion. Nuclear
load-bearing capacity of corroded RC beams 4. The level of impressed current density to Engineering & Design, 23(6):1123–1139.
against maximum mass loss of steel was be used to accelerate steel corrosion was Broomfield, J. 1997. Corrosion of steel in concrete.
closely related to theoretical results from a found to be contentious between research- London: E & FN Spon.
basic model of load-bearing capacity of RC ers. For example, El Maaddawy & Soudki Cabrera, J G 1996. Deterioration of concrete due to
beams. This was without applying factors (2003) and Mangat & Elgarf (1999) found reinforcement steel corrosion. Cement & Concrete
of bond between steel and concrete, as sug- that, at the same level of steel corrosion, Composites, 18: 47–59.
gested by Azad et al (2007). Malumbela et higher current densities cause more struc- Cairns, J, Du, Y & Law, D 2008. Structural performance
al (2010c) further demonstrated that the use tural damage than lower densities, while of corrosion-damaged concrete beams. Magazine of
of average mass loss to predict load-bearing Alonso et al (1998) and Malumbela et al Concrete Research, 60(5): 359–370.
capacity of RC beams at high mass losses of (2010b) found that a lower current density Castel, A, Vidal, T, Francois, R & Arliguie, G 2003.
steel will overestimate it. caused more structural damage. It was Influence of steel–concrete interface quality on rein-
Results from Torres-Acosta et al (2007) therefore recommended that this should forcement corrosion induced by chlorides. Magazine
and from Malumbela et al (2010c) against be researched further. of Concrete Research, 55(2): 151–159.
those from Azad et al (2007) and from 5. Except for results from Mangat & Elgarf El Maaddawy, T, Chahrour, A & Soudki, K 2006. Effect
Mangat & Elgarf (2007) suggest that the (1999), many researchers found the of fiber-reinforced polymer wraps on corrosion
load-bearing capacity of corroded RC beams load-bearing capacity of corroded RC activity and concrete cracking in chloride-contam-
is not related to the level of current density, specimens to be related to actual loss inated concrete cylinders. Journal of Composites for
but to the actual maximum mass loss of in area of steel and not to the level of Construction, 10(2): 139–147.
steel. However, more test results are needed current density used. Torres-Acosta et El Maaddawy, T & Soudki, K 2003. Effectiveness of
to confirm this. al (2007) and Malumbela et al (2010c) impressed current technique to simulate corrosion of
further showed that load-bearing capacity steel reinforcement in concrete. Journal of Materials
of corroded RC beams was related more in Civil Engineering, 15(1): 41–47.
CONCLUSIONS to maximum mass loss of steel than to El Maaddawy, T & Soudki, K 2007. Model for prediction
1. This paper discussed various procedures average loss. More data to confirm this is, of time from corrosion initiation to corrosion crack-
that are often used to accelerate steel however, needed. ing. Cement & Concrete Composites, 29(3): 168–175.
corrosion in laboratory tests of RC El Maaddawy, T, Soudki, K & Topper T 2005a.
specimens. It clearly pointed out that, Analytical model to predict nonlinear flexural
to hurriedly de-passify steel, researchers ACKNOWLEDGEMENT behaviour of corroded reinforced concrete beams.
should avoid adding chlorides to concrete The support of the Concrete Materials ACI Structural Journal, 102(4): 550–559.
mixes or fully immersing their samples & Structural Integrity Research Unit El Maaddawy, T, Soudki, K & Topper T 2005b. Long-
in salt solutions. These procedures result (CoMSIRU) at the University of Cape Town term performance of corrosion-damaged rein-
in uniform steel corrosion that unfor- is greatly acknowledged. forced concrete beams. ACI Structural Journal,
tunately underestimates the effects of 102(5): 649–656.
partial surface steel corrosion, which is Fang, C, Lundgren, K, Chen, L & Chaoying, Z 2004.
often observed in in-service structures. REFERENCES Corrosion influence of bond in reinforced concrete.
It was instead recommended in the paper Almusallam, A A, Al-Gahtani, A S, Aziz, A F & Cement & Concrete Research, 34: 2159–2167.
that limited faces of specimens should Rasheeduzzafar, F 1996. Effect of reinforcement cor- François, R & Arliguie, G 1998. Influence of service
be contaminated with chlorides. This rosion on bond strength. Construction & Building cracking on reinforcement steel corrosion. Journal of
can easily be achieved in laboratories by Materials, 10(2): 123–129. Materials in Civil Engineering, 10(1): 14–20.
building ponds on surfaces to be con- Alonso, C, Andrade, C, Rodriguez, J & Diez, J M 1998. Gadve, S, Mukherjee, A & Malhotra, S N 2009.
taminated or by selectively spraying them Factors controlling cracking of concrete affected by Corrosion of steel reinforcements embedded in
with chlorides. reinforcement corrosion. Materials & Structures, FRP wrapped concrete. Construction & Building
2. To accelerate steel corrosion, continuous 31: 435–441. Materials, 23: 153–161.
immersing of specimens in salt solu- Andrade, C & Alonso, C 2001. On-site measurements Hussain, R R 2010. Effect of moisture varia-
tion was shown to underestimate the of corrosion rate of reinforcements. Construction & tion on oxygen consumption rate of corrod-
rate of steel corrosion that is observed Building Materials, 15: 141–145. ing steel in chloride-contaminated concrete.
84 Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012
Cement & Concrete Composites, doi:10.1016/j. service loads. Construction & Building Materials, reinforced concrete beams. Engineering Structures,
cemconcomp.2010.09.014. 23: 3346–3351. 29: 1145–1152.
Jaffer, S J & Hansson, C M 2009. Chloride-induced Mangat, P S & Elgarf, M S 1999. Flexural strength of Tuutti, K 1980. Service life of structures with regard to
corrosion products of steel in cracked-concrete concrete beams with corroding reinforcement. ACI corrosion of embedded steel. In: Performance of con-
subjected to different loading conditions. Cement & Structural Journal, 96(1): 149–158. crete in marine environment, ACI SP-65, Detroit, US:
Concrete Research, 39: 116–125. Masoudi, S, Soudki, K & Topper T 2005. Post-repair American Concrete Institute, pp 223–236.
Jang, B S & Oh, B H 2010. Effects of non-uniform cor- fatigue performance of FRP-repaired corroded RC Vidal, T, Castel, A & François, R 2007. Corrosion proc-
rosion on the cracking and service life of reinforced beams: experimental and analytical investigation. ess and structural performance of a 17-year-old rein-
concrete structures. Cement & Concrete Research, Journal of Composites for Construction, 9(5): 441–449. forced concrete beam stored in a chloride environ-
40: 1441–1450. Poursaee, A & Hansson, C M 2009. Potential pitfalls in ment. Cement & Concrete Research, 37: 1551–1561.
Liu, Y & Weyers, R E 1998. Modelling the time-to- assessing chloride-induced corrosion of steel in con- Yoon, S, Wang, K, Weiss, J & Shah S 2000. Interaction
corrosion cracking in chloride-contaminated rein- crete. Cement & Concrete Research, 39(5): 391–400. between loading, corrosion, and serviceability of rein-
forced concrete structures. ACI Materials Journal, Rio, O, Andrade, C, Izquierdo, D & Alonso, C 2005. forced concrete. ACI Materials Journal, 97(6): 637–644.
95(6): 675–681. Behaviour of patch-repaired concrete structural Yuan, Y & Ji, Y 2009. Modeling corroded section configu-
Malumbela, G, Alexander, M G & Moyo, P 2010a. elements under increasing static loads to flexural ration of steel bar in concrete structure. Construction
Interaction between corrosion crack width and steel failure. Journal of Materials in Civil Engineering, & Building Materials, 23(6): 2461–2466.
loss in RC beams corroded under load. Cement & 17(2): 168–177. Yuan, Y, Ji, Y & Shah, S P 2007. Comparison of two
Concrete Research, 40: 1419–1428. Roberge, P R 1999. Handbook of corrosion engineering. accelerated corrosion techniques for concrete struc-
Malumbela, G, Alexander, M G & Moyo, P 2010b. New York: McGraw-Hill. tures. ACI Materials Journal, 104(3): 344–347.
Serviceability of corrosion-affected RC beams after Rodriguez, J, Ortega, LM & Casal, J 1997. Load carry- Zhang, R, Castel, A & François, R 2009a. Serviceability
patch repairs and FRPs under load. Materials & ing capacity of concrete structures with corroded limit state criteria based on steel–concrete bond loss
Structures, doi 10.1617/s11527-010-9630-9638. reinforcement. Construction & Building Materials, for corroded reinforced concrete in chloride envi-
Malumbela, G, Alexander, M G & Moyo, P 2010c. 11(4): 239–248. ronment. Materials & Structures, 42(10): 1407–1421.
Variation of steel loss and its effect on flexural capa- Ting, S & Nowak A S 1991. Effect of reinforcing steel Zhang, R, Castel, A & François, R 2009b. The corro-
city of RC beams corroded and repaired under load. area loss on flexural behaviour of reinforced con- sion pattern of reinforcement and its influence on
Construction & Building Materials, 24: 1051–1059. crete beams. ACI Structural Journal, 88(3): 309–314. serviceability of reinforced concrete members in
Malumbela, G, Alexander, M G & Moyo P 2011. Model Torres-Acosta, A A & Martinez-Madrid M 2003. chloride environment. Cement & Concrete Research,
for cover cracking of RC beams due to partial surface Residual life of corroding reinforced concrete struc- 39(11): 1077–1086.
steel corrosion. Construction & Building Materials, tures in marine environment. Journal of Materials in Zhang, R, Castel, A & François, R 2010. Concrete cover
25: 987–991. Civil Engineering, 15(4): 344–353. cracking with reinforcement corrosion of RC beam
Malumbela, G, Moyo, P & Alexander, M G 2009. Torres-Acosta, A A, Navarro-Gutierrez, S & Terán- during chloride-induced corrosion process. Cement
Behaviour of RC beams corroded under sustained Guillén, J 2007. Residual flexure capacity of corroded & Concrete Research, 40(3): 415–425.
Journal of the South African Institution of Civil Engineering • Volume 54 Number 2 October 2012 85