U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

 
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-11-060    Date:  November 2011
Publication Number: FHWA-HRT-11-060
Date: November 2011

 

Multiple Corrosion-Protection Systems for Reinforced Concrete Bridge Components

APPENDIX B. CORROSION LOSS REQUIRED TO CRACK CONCRETE CONTAINING CONVENTIONAL, EPOXY-COATED, AND GALVANIZED REINFORCEMENT

INTRODUCTION

The corrosion of steel reinforcement in concrete is a destructive process for both the steel and the concrete. The corrosion products of steel occupy several times the volume of solid steel, resulting in cracking and spalling of the concrete cover once a sufficient amount of corrosion loss has occurred. Several prior studies have worked to establish a relationship between corrosion loss and cracking of concrete cover for uncoated conventional reinforcement.(68-71) In addition, Torres-Acosta and Sagues examined the effects of localized corrosion, although the corroding areas were much larger than the area typically exposed due to damage to ECR.(71)

Limited research has been performed on the amount of corrosion loss required to crack concrete for galvanized reinforcement. Sergi, Short, and Page found that the corrosion product of zinc is often zinc oxide.(73) The volume of zinc oxide is only 1.5 times that of solid zinc, whereas the volume of ferric oxide is 3 times that of solid steel, indicating that the corrosion loss required to crack concrete for specimens with galvanized reinforcement should be greater than the corrosion loss required to crack concrete with conventional reinforcement.(74,75) However, under certain conditions, zinc can also form zinc hydroxychloride II, which has 3.6 times the volume of solid zinc.(73,74) The formation of zinc hydroxychloride II will result in corrosion losses for galvanized reinforcement at the onset of cracking similar to those observed for conventional reinforcement. Rasheeduzzafar et al. studied conventional and galvanized reinforcement cast in concrete with chloride contents at casting ranging from 2.4 to 19.2 kg/m3 (4 to 32 lb/yd3).(76) Rasheeduzzafar et al. found specimens containing galvanized reinforcement took longer to crack concrete than specimens containing conventional reinforcement; however, the corrosion loss at crack initiation was not determined.

The research described in this appendix examines the corrosion losses required to crack concrete cover for conventional, galvanized, and damaged epoxy-coated reinforcement. Specimens with conventional and galvanized reinforcement were tested at varying covers to establish a relationship between corrosion loss and cracking for conventional and galvanized reinforcement. ECR was tested at 25-mm (1-inch) cover with varying damage patterns to determine the effect of the damaged area on corrosion loss required to crack concrete. Two- and three-dimensional finite element models were created to test the corrosion loss to crack concrete for multiple combinations of cover and damaged area. The results from the finite element models are compared with experimental results from this and other studies, and an expression is developed relating damaged area, concrete cover, and corrosion loss to cause cracking.

EXPERIMENTAL PROCEDURE

Mixture Proportions

The mixture proportions used in the concrete for all specimens are shown in table 57. The materials used are as follows:

The mixture includes salt equivalent to 2 percent chlorides by weight of cement to destabilize the passive layer of the reinforcement and increase the ionic conductivity of the concrete. The salt is dissolved in the mix water prior to casting.

Table 57. Mix proportions for cracking specimens.

Cement, kg/m3
(lb/yd3)

Water, kg/m3
lb/yd3

Fine Aggregate, kg/m3
(lb/yd3)

Coarse Aggregate, kg/m3
(lb/yd3)

Sodium Chloride, kg/m3
(lb/yd3)

Air Entraining Agent,
L/m3 (oz/yd3)

356 (598)

160 (269)

854 (1,435)

883 (1,484)

11.7 (19.8)

2.66 (68.9)

Materials

The following materials are used in the cracking tests described in this appendix:

Specimens

A schematic of the cracking specimens is shown in figure 222. Cracking specimens are beam specimens, 152 mm (6 inches) wide by 305 mm (12 inches) long. Specimen height is dependent on the concrete cover. The top bar is the test bar and consists of conventional, galvanized, or epoxy-coated reinforcement. The bottom bars are pickled 2205 duplex stainless steel. All bars are No. 16 (No. 5) reinforcing steel. Specimens are connected to a power supply to drive corrosion on the test bar and are kept ponded with deionized water.

The cracking specimen is 152 by 305 mm (6 by 12 inches), with one bar on top and two bars on the bottom. An integral dam is used around the upper surface of the specimen. Top cover varies from 13 to 51 mm (0.5 to 2 inches). Bottom bar cover is 25 mm (1 inch). The mats are connected via a power supply that drives corrosion on the top bar.

Figure 222. Illustration. Cracking specimen.

A total of 34 specimens were tested in five series. Series 1 consisted of beams with conventional and galvanized reinforcement with 25-mm (1-inch) concrete cover. Series 2 consisted of beams with conventional and galvanized reinforcement with 13-mm (0.5-inch) concrete cover. Series 3 tested conventional and galvanized reinforcement with 51-mm (2-inch) cover. Series 4 tested damaged ECR with 25-mm (1-inch) concrete cover. Series 5 tested two specimens with galvanized reinforcement and 25-mm (1-inch) cover, with specimens removed from testing at crack initiation. Testing continued on series 1, 2, and 3 until the crack reached a width of 0.508 mm (0.02 inches). Testing continued on series 4 until the crack spanned the full length of the specimen because the lower corrosion rate of specimens containing ECR made it impractical to continue the test until the crack reached a width of 0.508 mm (0.02 inches).

Test Procedure

The test begins 14 days after the specimens are cast. During the test, the current to each specimen is measured daily. Dividing the measured current by the surface area of the test bar (or the damaged area for ECR) gives the corrosion current density, which is used to determine corrosion rate using Faraday's equation (see chapter 2). Specimens are monitored daily for staining and cracking. The corrosion loss at staining, crack initiation, and propagation of the crack to the full specimen length are recorded. In addition, the crack width as a function of corrosion loss is tracked for specimens with conventional and galvanized reinforcement.

Fabrication

Specimen fabrication for cracking specimens follows the preparation procedure for bench-scale specimens outlined in chapter 2, with two exceptions. ECR is damaged in either a two-hole or two half-ring pattern, as shown in figure 223 and figure 224. Specimens are also cured in molds for 14 days as opposed to the curing procedure used for bench-scale specimens.

Epoxy-coated reinforcement (ECR) was damaged in one of two ways. The two-hole damage pattern (2h) consists of two 3-mm (0.125-inch)-diameter holes through the epoxy, spaced at 102 mm (4 inches) on center.

Figure 223. Illustration. Damage patterns for ECR with two holes.

 

Epoxy-coated reinforcement (ECR) was damaged in one of two ways. The two half-ring damage pattern (2r) consists of two half-rings, 3 mm (0.125 inches) wide, running from longitudinal rib to longitudinal rib, spaced 102 mm (4 inches) on center.

Figure 224. Illustration. Damage patterns for ECR with two half-rings.

Test Program

The test program is summarized in table 58. The conventional and galvanized reinforcement were tested with 13-mm (0.5-inch), 25-mm (1-inch), and 51-mm (2-inch) concrete covers. The galvanized reinforcement had a nominal coating thickness of 0.15 mm (6 mil). The ECR, with coating thickness ranging from 0.20 to 0.27 mm (8 to 10.5 mil) and an average of 0.25 mm (9.7 mil), was tested using a 25-mm (1-inch) cover with two damage patterns, as previously described.

Table 58. Corrosion loss to cause cracking, number of specimens in test program.

System

Cover

13 mm
(0.5 inches)

25 mm
(1 inch)

51 mm
(2 inches)

Uncoated bars (Conv.)

4

4

4

Galvanized bars (Zn)

4

6a

4

ECR-2 hole damage pattern

Horizontal alignment (ECR-2h-H)

-

2

-

Vertical alignment (ECR-2h-V)

-

2

-

ECR-2 ring damage pattern

Horizontal alignment (ECR-2r-H)

-

2

-

Vertical alignment (ECR-2r-V)

-

2

-

- No data.
a Two specimens removed from testing at crack initiation.

Finite Element Modeling of Corrosion Loss and Cracking

To further study the relationship between corrosion loss and cracking, two- and three-dimensional finite element models were created using ABAQUS 6.9.(77) The two-dimensional models were used to model uniform corrosion of a reinforcing bar. The three-dimensional models were tested with uniform corrosion over the entire bar, as well as with areas of localized corrosion. The model represents a slab with mirror symmetry about the axis of the reinforcement (see figure 225). The crack is assumed to propagate along the vertical boundary of the model centered on the reinforcing bar. A series of nonlinear springs were used to provide horizontal restraint along the plane of the crack to represent the nonlinear behavior of the concrete as it cracks.

The finite element model consisted of a concrete slab with a single reinforcing bar, cut vertically through the center of the rebar. The cut edge of the concrete was connected to ground with nonlinear springs to model the behavior of cracking concrete.

Figure 225. Illustration. Two-dimensional finite element model of concrete to measure cracking behavior.

The properties of the springs were based on measurements of fracture energy of concrete. Darwin et al. tested notched beams in center-point loading.(78) Fracture energy was calculated by determining the area under the load-deflection curves for each specimen. Darwin et al. found that for concretes older than 5 days, fracture energy is governed by coarse aggregate properties and is independent of w/c ratio, compressive strength, and age of concrete. The spring properties were adjusted to provide a fracture energy of 61 N/m (0.35 lb/in), comparable to the value reported by Darwin et al.(78) The initial stiffness of the springs provided an elastic modulus of 27.6 GPa (4,000 ksi) and a peak tensile stress of 2.76 MPa (400 psi). The spring behavior for a spring density of 6,200 springs/m2 (4 springs/in2) is shown in figure 226.

The initial stiffness of the springs provides an elastic modulus of 27.6 GPa (4,000 ksi) and a peak tensile stress of 2.76 MPa (400 psi). After reaching the peak stress, the springs’ load carrying capacity decreases nonlinearly with increasing deflection, based on research by Darwin et al.

1 lb = 2.228 N
1 inch = 25.4 mm

Figure 226. Graph. Load-deflection behavior for nonlinear spring model, spring density of 6,200 springs/m2 (4 springs/in2).

Material away from the plane of the crack was assumed to be linear and elastic, with an elastic modulus of 27.6 GPa (4,000 ksi) and a Poisson's ratio of 0.2. Corrosion was assumed to occur uniformly over the entire surface of the conventional and galvanized bars and over the localized damaged regions of the epoxy-coated bars. The buildup of corrosion products was represented by applying a uniform deflection normal to the reinforcing bar surface. The volume ratio of corrosion products to corrosion loss n was assumed to be 3.0 based on work by Suda et al.(75) A visible crack was assumed to have formed when the horizontal deflection at the top surface of the model (point A in figure 225) reached 25 Mum (1.0 mil). With the model symmetry, this corresponds to a 50-Mum (2.0-mil)-wide crack. The displacement at the surface of the concrete at the location of the reinforcing bar required to cause the formation of the crack (crit) was converted to a corrosion loss (xcrit) using the equation in figure 227. The term in the denominator (n-1) accounts for the volume of the reinforcing bar that is converted to corrosion product.

x subscript crit. x subscript crit equals delta subscript crit divided by the quantity: n minus 1, closed quantity.

Figure 227. Equation. Corrosion loss conversion.

Figure 228 and figure 229 show typical finite element model meshes used for the two- and three-dimensional models. The model dimensions are shown in figure 225. For the two-dimensional finite element models, concrete covers of 6.4, 13, 19, 25, 38, 51, 76, and 102 mm (0.25, 0.5, 0.75, 1, 1.5, 2, 3, and 4 inches) and bar diameters of 13, 19, and 25 mm (0.5, 0.75, and 1 inches) were evaluated. For the three-dimensional finite element models, concrete covers of 51 and 76 mm (2 and 3 inches) and bar diameters of 13, 19, and 25 mm (0.5, 0.75, and 1 inches) were used. The two-dimensional model had a unit length, and the three-dimensional model had a length of 508 mm (20 inches).

The two-dimensional finite element model uses a 6.4-mm (0.25-inch) quad-dominated mesh.

Figure 228. Illustration. Two-dimensional finite element analysis model.

 

The three-dimensional finite element model uses a 6.4-mm (0.25-inch) hex mesh.

Figure 229. Illustration. Three-dimensional finite element analysis model (end view).

Trial models were run for both the two- and three-dimensional models to determine the effect of mesh type on model performance (see table 59 and table 60). In both cases, the mesh type had no significant effect on the model performance, so the default meshes for the two- and three-dimensional models (quad-dominated and hex, respectively) were used.

Table 59. Effect of two-dimensional element type on corrosion loss.

Mesh Type

Corrosion Loss to Produce a 50- Mum Crack, Mum

Quad

38.99

Quad-dominated

38.48

Tri

38.74

1 Mum = 0.0394 mil

Table 60. Effect of three-dimensional element type on corrosion loss.

Mesh Type Corrosion Loss to Produce a 50- Mum Crack, Mum

Hex

64.5

Hex-dominated

64.2

Wedge

64.8

Tet

65.4

1 Mum = 0.0394 mil

Trial models were also run to determine the effect of mesh seed size on model performance. The mesh seed size was increased until the finite element model results were affected (see figure 230). A 6.4 mm (0.25-inch) mesh was chosen, as it was the largest mesh size for which the finite element model results were not affected.

Models with mesh sizes less than or equal to 6.4 mm (0.25 inches) required 39.4 µm (1.55 mil) of loss to crack the concrete cover. Models with mesh sizes greater than 6.4 mm (0.25 inches) required less corrosion loss to crack the concrete cover, with the model with a 25-mm (1-inch) mesh size requiring only 38.3 µm (1.51 mil) of loss to crack concrete. Since mesh sizes less than or equal to 6.4 mm (0.25 inches) produce constant results, a 6.4-mm (0.25-inch) mesh size is selected to minimize computing time without sacrificing accuracy.

1 Mum = 0.0394 mil

Figure 230. Graph. Effect of mesh seed size on corrosion loss required to produce a 50-Mum (2-mil) crack.

The two-dimensional model was used to analyze uniform corrosion over the entire bar surface. For the three-dimensional model, three damage patterns were analyzed for each combination of cover and bar diameter, as shown in figure 231. The first damage pattern simulates corrosion along the entire circumference of the bar. Models with this damage pattern are designated FR (full ring corrosion pattern). The second damage pattern simulates corrosion along half the bar circumference and is designated HR. The third damage pattern simulates corrosion along one-quarter of the bar circumference and is designated QR. The length of the FR damage pattern along the bar ranged from 3.2 to 508 mm (0.125 to 20 inches). The length of the HR damage pattern along the bar ranged from 3.2 to 203 mm (0.125 to 8 inches), and the length of the QR damage pattern along the bar ranged from 1.6 to 203 mm (0.0625 to 8 inches).

Ring patterns are oriented so the centroid of the damage area coincides with the horizontal plane passing through the center of the reinforcement.

Figure 231. Illustration. Cross section of bar damage patterns for three-dimensional finite element models: full ring, half ring, and quarter ring.

EXPERIMENTAL RESULTS

Conventional and Galvanized Reinforcement

The values of corrosion loss to initiate cracking for conventional and galvanized reinforcement are shown in figure 232, with the standard deviation represented by error bars. For all concrete covers, galvanized reinforcement required significantly greater corrosion losses to crack the concrete cover than did conventional reinforcement. For 13-mm (0.5-inch) cover, conventional reinforcement required an average corrosion loss of 10.6 Mum (0.417 mil) to crack the concrete cover, compared to 45.9 Mum (1.81 mil) for galvanized reinforcement. For 25-mm (1-inch) cover, conventional reinforcement required an average corrosion loss of 22.4 Mum (0.882 mil) to crack the concrete cover, compared to 49.7 Mum (1.91 mil) for galvanized reinforcement, and for 51‑mm (2-inch) cover, conventional reinforcement required an average corrosion loss of 29.7 Mum (1.17 mil) to crack the concrete cover, compared to 68.0 Mum (2.68 mil) for galvanized reinforcement. For conventional reinforcement, increasing the cover from 13 to 51 mm (0.5 to 2 inches) nearly tripled the corrosion loss required to crack concrete from 10.6 to 29.7 Mum (0.417 to 1.17 mil), an increase of 19.1 Mum (0.752 mil). For galvanized reinforcement, the loss increased by 48 percent from 45.9 to 68.0 Mum (1.81 to 2.68 mil), an increase of 22.1 Mum (0.870 mil).

For all concrete covers, galvanized reinforcement exhibits over twice the corrosion loss required to crack concrete compared to conventional reinforcement. For 13-mm (0.5-inch) cover, conventional reinforcement requires an average corrosion loss of 10.6 µm (0.417 mil) to crack the concrete cover, compared to 45.9 µm (1.81 mil) for galvanized reinforcement. For 25-mm (1-inch) cover, conventional reinforcement requires an average corrosion loss of 22.4 µm (0.882 mil) to crack the concrete cover, compared to 49.7 µm (1.96 mil) for galvanized reinforcement, and for 51-mm (2-inch) cover, conventional reinforcement requires an average corrosion loss of 29.7 µm (1.17 mil) to crack the concrete cover, compared to 68.0 µm (2.68 mil) for galvanized reinforcement.

1 Mum = 0.0394 mil

Figure 232. Graph. Average corrosion loss required to crack concrete for specimens with conventional and galvanized reinforcement.

Autopsy results from all specimens with conventional reinforcement showed heavy corrosion losses over the entire bar surface (see figure 233 and figure 234). Staining was apparent in the concrete surrounding the reinforcement. Figure 235 shows a side view of the concrete around the reinforcement split along the plane of the crack. Orange corrosion products are visible in regions where the staining reached the surface. In figure 236, greenish-black corrosion products are visible in regions isolated from the atmosphere. All photos were taken immediately after autopsy.

Severe corrosion is observed over the entire bar surface, with some sections exhibiting visible reductions in rib area.

Figure 233. Photo. Top side of bar in specimen Conv.-3, 51-mm (2-inch) cover, after autopsy.

 

Severe corrosion is observed over the entire bar surface, with some sections exhibiting visible reductions in rib area.

Figure 234. Photo. Bottom side of bar in specimen Conv.-3, 51-mm (2-inch) cover, after autopsy.

 

Severe staining is observed along the plane of the crack and around the reinforcement. The corrosion products exposed to air in the crack have an orange-brown color; corrosion products not exposed to air during testing had a greenish-black color.

Figure 235. Photo. Side view of specimen Conv.-2, 25-mm (1-inch) cover, after autopsy (plane of crack visible above reinforcement).

 

Greenish-black corrosion products are observed extending greater than 25 mm (1 inch) into the concrete around the reinforcement.

Figure 236. Photo. Top view of specimen Conv.-2, 25-mm (1-inch) cover, after autopsy.

The autopsy found that galvanized reinforcement exhibited signs of pitting corrosion. Some regions of the test bar exhibited heavy corrosion products, while in other sections, the galvanized coating was unaffected (see figure 237 and figure 238). Most of the uncorroded regions were located on the top face of the bar, a result of the bottom side of the bar having more even exposure to the ions migrating from the bottom bars. Measurements with a coating thickness gauge showed no significant loss in the areas that appear uncorroded. Visual estimations of uncorroded surface areas were performed on all bars after autopsy, and results appear in table 61. The bars with 13-mm (0.5‑inch) cover showed the greatest average uncorroded area, 29 percent, likely due to the decreased cover interfering with ion transport to the top side of the bar. The bars with 25- and 51-mm (1- and 2-inch) cover showed average uncorroded areas of 6 and 13 percent, respectively. The corrosion products on the concrete surrounding the galvanized reinforcement resembled those seen in specimens with conventional reinforcement, indicating that the bulk of corrosion products applying pressure to the surrounding concrete are corrosion products of iron and not those of zinc (see figure 239 and figure 240).

Galvanized reinforcement shows heavy orange corrosion products over most of the bar surface with isolated regions of intact zinc present.

Figure 237. Photo. Top side of bar in specimen Zn-2, 25-mm (1-inch) cover, after autopsy.

 

Galvanized reinforcement shows heavy orange corrosion products over most of the bar surface with isolated regions of intact zinc present.

Figure 238. Photo. Bottom side of bar in specimen Zn-2, 25-mm (1-inch) cover, after autopsy.

 

Table 61. Estimated uncorroded surface area of galvanized reinforcement.

Specimen Estimated Uncorroded Area, percent
Cover
13 mm (0.5 inch) 25 mm (1 inch) 51 mm (2 inch)

Zn-1

30

8

5

Zn-2

30

5

10

Zn-3

40

5

50

Zn-4

15

5

30

Average

29

6

13

The plane of the crack is visible above the reinforcement. Along the plane, heavy orange-colored staining is observed.

Figure 239. Photo. Side view of specimen Zn-4, 25-mm (1-inch) cover, after autopsy.

 

The specimen split at the depth of reinforcement. As observed on concrete containing conventional reinforcement, greenish-black corrosion products are observed extending greater than 25 mm (1 inch) into the concrete around the galvanized reinforcement.

Figure 240. Photo. Top view of specimen Zn-4, 25-mm (1-inch) cover, after autopsy.

To determine if the pitting observed on galvanized reinforcement was also present at crack initiation, two additional specimens with galvanized reinforcement and 25-mm (1-inch) cover were cast and autopsied at the onset of cracking. Greenish-black corrosion products were visible along the crack at the upper surface of the specimens (see figure 241). The corrosion products turned orange about 2 h after exposure to air. The autopsy revealed pitting and localized corrosion on the bars similar to that observed in the specimens autopsied after the crack had propagated and widened (see figure 242 and figure 243). As previously discussed, uncorroded regions were more common on the top than the bottom side of the test bar (see figure 244). These results suggest that cracking of the concrete due to corrosion of the galvanized reinforcement did not result due to the buildup of zinc corrosion products but rather due to the formation of corrosion products from the intermetallic steel-zinc layers or from the underlying steel.

At crack initiation, a thin line of greenish-brown corrosion products is observed on the surface of the concrete at the location of the crack.

Figure 241. Photo. Staining on surface at crack initiation in galvanized reinforcement specimen, 25-mm (1‑inch) cover.

 

Galvanized reinforcement removed from concrete at crack initiation shows large regions covered by orange-brown corrosion products with isolated regions of intact zinc.

Figure 242. Photo. Top side of galvanized reinforcement, 25-mm (1-inch) cover, after autopsy at crack initiation.

 

Galvanized reinforcement removed from concrete at crack initiation shows large regions covered by orange-brown corrosion products with isolated regions of intact zinc.

Figure 243. Photo. Detail of top side of galvanized reinforcement, 25-mm (1-inch) cover, after autopsy at crack initiation.

 

Galvanized reinforcement removed from concrete at crack initiation shows large regions covered by orange-brown corrosion products with isolated regions of intact zinc.

Figure 244. Photo. Bottom side of galvanized reinforcement, 25-mm (1-inch) cover, after autopsy at crack initiation.

ECR

The corrosion losses required to crack concrete cover for specimens containing damaged ECR are shown in table 62. The losses are presented based on both the total area of the bar and the damaged (exposed) area in the epoxy. The bars with two half-rings had a nominal exposed area ten times greater than the bars with two holes in the epoxy; however, autopsy results revealed significant blistering on bars with holes in the epoxy (see figure 245). Blistering was also present on the bars with the half-rings but was less severe and exposed less than the area exposed by the rings. Table 62 reflects an estimate of the increased exposed area from the blistered regions for all specimens. Ignoring the blistered regions, specimens with two half-rings had an exposed area of 150.8 mm2 (0.234 in2) and specimens with two holes had an exposed area of 15.8 mm2 (0.024 in2).

Table 62. Average corrosion loss to crack concrete cover for specimens with ECR.

Specimen Exposed Area (Including Blisters), mm2 Corrosion Loss Based on Total Area, Mu m Corrosion Loss Based on Exposed Area, Mu m

ECR-2 hole damage pattern

ECR-2h-H-1

188.3

10.14

730

ECR-2h-H-2

233.5

10.10

587

ECR-2h-H Average

10.12

659

ECR-2h-V-1

181.8

11.67

874

ECR-2h-V-2

201.1

7.58

510

ECR-2h-V Average

9.70

692

ECR-2 ring damage pattern

ECR-2r-H-1

208.9

6.07

421

ECR-2r-H-2

234

5.87

363

ECR-2r-H Average

5.97

392

ECR-2r-V-1

254

6.22

354

ECR-2r-V-2

208.9

6.15

426

ECR-2r-V Average

6.18

390

1 mm2 = 0.00155 in2
1 Mum = 0.0394 mil

Epoxy-coated bars show significant blistering and staining on the surface of the epoxy, both at and away from intentional damage sites.

Figure 245. Photo. Test bar from specimen ECR-2h-V-2.

Table 62 shows no significant difference in the corrosion loss required to crack the concrete cover between the specimens with the damage pattern oriented horizontally or vertically. The corrosion losses on both the total and exposed areas indicate that the corrosion loss required to crack the concrete cover increases as exposed area decreases. Based on total and exposed area including blisters, the specimens with two holes in the epoxy required somewhat less than twice the corrosion loss to crack the concrete cover as the specimens with two half-rings in the epoxy.

Figure 246 shows a specimen at crack initiation. Figure 247 shows a specimen with the crack spanning the length of the specimen. No specimens containing ECR showed signs of surface staining during the test. After testing, however, staining was observed on the concrete surrounding the damaged regions in the epoxy when the specimens were autopsied (see figure 248 and figure 249).

A faint white crack, less than 0.25 mm (0.01 inch) wide, is observed on the concrete surface.

Figure 246. Photo. Crack initiation in specimen ECR-2r-H-2.

 

A faint white crack is observed over the length of the specimen. No staining on the surface is observed.

Figure 247. Photo. Crack propagation in specimen ECR-2h-V-1.

 

Staining is observed on the concrete around the regions where the epoxy was intentionally damaged, but much less staining is observed in the concrete around ECR than the concrete around conventional and galvanized reinforcement.

Figure 248. Photo. Concrete surrounding test bar from specimen ECR-2h-V-1.

 

. Light staining is observed on the concrete around the regions where the epoxy was intentionally damaged.

Figure 249. Photo. Concrete surrounding test bar from specimen ECR-2r-H-2.

FINITE ELEMENT RESULTS

The corrosion losses that cause a 50-Mum (2-mil)-wide crack to form at the surface of the specimen based on the two-dimensional finite element analyses are shown in table 63. The corrosion losses are plotted as a function of concrete cover in figure 250. The results suggest a linear relationship (for covers between 6 and 102 mm (0.25 and 4 inches)) between concrete cover and corrosion loss required to cause cracking. A slight dependence on bar diameter is also noted; figure 251 shows best-fit lines for each of the three bar diameters over the range of covers from 19 to 76 mm (0.75 to 3 inches).

Table 63. Finite element results for two-dimensional model.

Cover, mm Bar Diameter, mm Corrosion Loss to Crack Concretea, Mu m

6.4

12.7

19.7

13

12.7

20.3

25

12.7

26.0

38

12.7

35.6

51

12.7

45.1

76

12.7

59.5

9.5

19

19.7

19

19

25.4

38

19

33.7

57

19

45.1

76

19

56.5

13

25.4

20.3

25

25.4

26.7

51

25.4

40.6

76

25.4

51.1

102

25.4

76.2

1 mm = 0.039 inches
1 Mum = 0.0394 mil
a 50-Mum (0.002-inch) crack width.

A linear relationship exists between cover and corrosion loss. Models with less than 12.7 mm (0.5 inches) of cover require approximately 20 µm  (0.79 mil) of loss to crack concrete, while models with 76-mm (3-inch) cover require losses between 50 and 55 µm  (2 and 2.2 mil) to crack concrete.

1 Mum = 0.0394 mil
1 inch = 25.4 mm

Figure 250. Graph. Corrosion loss to crack concrete versus cover for two-dimensional finite element model.

For a given increase on cover, bars with a smaller diameter show a slightly greater increase in corrosion loss required to crack concrete than bars with a larger diameter; however, the effect is small.

1 Mum = 0.0394 mil
1 inch = 25.4 mm

Figure 251. Graph. Corrosion loss to crack concrete versus cover showing effect of bar diameter for two-dimensional finite element model.

Table 63 and figure 251 show that as cover increases, bars with smaller diameters require somewhat greater corrosion losses to crack concrete than bars with larger diameters for covers between 25 and 76 mm (1 and 3 inches). An analysis of the data suggests the equation in figure 252 as a best-fit expression. The English equivalent is presented in figure 253.

x subscript crit equals 1.8 times by the quantity: C divided by D raised to the 0.38 power, that entire quotient plus 7.5, closed quantity

Figure 252. Equation. Suggested best-fit in SI units.

x subscript crit equals 0.53 times by the quantity: C divided by D raised to the 0.38 power, that entire quotient plus 1, closed quantity.

Figure 253. Equation. Suggested best-fit in English units.

Where:

xcrit = Corrosion loss at crack initiation, Mum or mil.

C = Cover, mm or inches.

D = Bar diameter, mm or inches.

To verify the accuracy of the two-dimensional finite element model, the results obtained from the model are compared with experimental results presented in this appendix along with experimental results obtained by Saeki et al., Rasheeduzzafar et al., Alonso et al., and Torres-Acosta and Sagues for corrosion along the full length of conventional reinforcement. (See references 68, 69, 71, and 79.) The data from these sources are shown in table 64. The experimental data are plotted along with the finite element results in figure 254. Data presented in this appendix are labeled "KU." Other data are identified by the first author. A best-fit line for the results from the finite element model is also shown. While there is much scatter in the experimental data, the corrosion loss required to crack concrete, as predicted by the finite element model, provides an excellent representation of the bulk of the experimental data.

Table 64. Corrosion loss to crack concrete (corrosion along entire bar length).

Study Cover, mm Diameter, mm Corrosion Loss, Mum

Torres-Acosta and Sagues(71)

39

13

35.9

39

13

31.1

Alonso et al.(69)

20

16

15

15

8

20

30

16

25

30

16

28

30

16

30

50

16

31

50

12

51

70

16

55

70

10

68

20

16

25

20

16

18

Rasheeduzzafar et al.(68)

19

13

33

19

13

26

19

13

34

38

13

32

38

13

30

38

13

47

38

13

38

38

13

27

38

13

27

50

13

70

50

13

71

50

13

74

50

13

61

60

13

67

Saeki et al.(79)

31.75

9.5

42

31.75

12.7

34

31.75

19

23

31.75

25

13

1 mm = 0.039 inches
1 Mum = 0.0394 mil

Plotting experimental data alongside the finite element model data shows that while there is much scatter in prior experimental results, the finite element model results generally agree with past experimental data.

1 Mum = 0.0394 mil
1 inch = 25.4 mm

Figure 254. Graph. Corrosion loss to crack concrete versus cover in two-dimensional finite element model with experimental data.

To determine its accuracy, the results predicted by the equation in figure 252 are compared with experimental results and finite element model results in figure 255.

Plotting experimentally obtained corrosion losses versus the corrosion losses predicted by figure 252 shows figure 252 generally provides an accurate prediction of corrosion losses required to crack concrete but is slightly unconservative.

1 Mum = 0.0394 mil

Figure 255. Graph. Corrosion loss to crack concrete for uniform general corrosion based on experimental and finite element results versus predicted corrosion losses using figure 252.

The proposed equation overestimates the corrosion loss required to crack concrete for most cases in which the actual corrosion loss required to crack the concrete is less than 45 Mum (1.78 mil). An alternate equation is proposed in figure 256 and figure 257 that provides a somewhat more conservative estimate of the corrosion loss required to crack concrete (see figure 258).

x subscript crit equals 1.8 times by the quantity: C divided by D raised to the 0.38 power, that entire quotient plus 4.5, closed quantity.

Figure 256. Equation. Alternate best-fit in SI units.

 

x subscript crit equals 0.53 times by the quantity: C divided by D raised to the 0.38 power, that entire quotient plus 0.6, closed quantity.

Figure 257. Equation. Alternate best-fit in English units.

Where:

xcrit = Corrosion loss at crack initiation, Mum or mil.

C = Cover, mm or inches.

D = Bar diameter, mm or inches.

Plotting experimentally obtained corrosion losses versus the corrosion losses predicted by figure 256 shows that figure 256 provides more conservative predictions of corrosion losses required to crack concrete while remaining reasonably accurate.

1 Mum = 0.0394 mil

Figure 258. Graph. Corrosion loss to crack concrete for uniform general corrosion based on experimental and finite element results versus predicting corrosion losses using figure 256.

The corrosion losses to cause cracking based on the three-dimensional finite element model are shown in table 65 and table 66. The models with a 51-mm (2-inch) cover and a 203-mm (8-inch) and 508-mm (20-inch) length of bar corroding show similar corrosion losses at crack initiation. The behavior of the three-dimensional finite element model under full-bar-length (508 mm (20 inches)) corrosion is compared to that of the two-dimensional finite element model in figure 259. The corrosion losses to cause cracking obtained from these models are similar. The differences in corrosion loss to crack concrete between two- and three-dimensional models is less than 1 Mum (0.039 mil), with the exception of models with a 25-mm (1-inch)-diameter bar and 76-mm (3‑inch) cover, which show a 2.9-Mum (0.11-mil) or 5.7 percent difference in corrosion loss to cause cracking.

Table 65. Finite element results for three-dimensional model, 51-mm (2-inch) cover.

Corrosion Pattern 13-mm (0.5-in.) Bar Diameter 19-mm (0.75-in.) Bar Diameter 25-mm (1-in.) Bar Diameter
Exposed
Area,
mm2 (in2)
Corrosion Loss at Cracking,
Mu m
Exposed
Area,
mm2 (in2)
Corrosion Loss at Cracking,
Mu m
Exposed Area,
mm2 (in2)
Corrosion Loss at Cracking, Mu m

Full Ring

508 mm (20 in.) length

10,134 (15.7)

46

15,201 (23.6)

43

20,268 (31.4)

40

203 mm (8 in.) length

4,054 (6.28)

57

6,080 (9.42)

44

8,107 (12.6)

41

102 mm (4 in.) length

2,027 (3.14)

79

3,040 (4.71)

57

4,054 (6.28)

56

51 mm (2 in.) length

1,013 (1.57)

159

1,520 (2.36)

133

2,027 (3.14)

80

25 mm (1 in.) length

507 (0.785)

330

760 (1.18)

254

1,013 (1.57)

144

13 mm (0.5 in.) length

253 (0.392)

483

380 (0.589)

381

507 (0.785)

281

6.4 mm (0.25 in.) length

127 (0.196)

659

190 (0.295)

508

253 (0.392)

361

3.2 mm (0.125 in.) length

63.3 (0.098)

851

95 (0.147)

658

127 (0.196)

502

Half Ring

102 mm (4 in.) length

1,013 (1.57)

178

1,520 (2.36)

152

2,027 (3.14)

88

51 mm (2 in.) length

507 (0.785)

273

760 (1.18)

229

1,013 (1.57)

150

25 mm (1 in.) length

253 (0.392)

425

380 (0.589)

347

507 (0.785)

279

13 mm (0.5 in.) length

127 (0.196)

635

190 (0.295)

502

253 (0.392)

418

6.4 mm (0.25 in.) length

63.3 (0.098)

904

95.0 (0.147)

704

127 (0.196)

572

3.2 mm (0.125 in.) length

31.7 (0.049)

1,228

47.5 (0.074)

973

63.3 (0.098)

784

Quarter Ring

102 mm (4 in.) length

507 (0.785)

216

760 (1.18)

191

1,013 (1.57)

170

51 mm (2 in.) length

253 (0.392)

337

380 (0.589)

292

507 (0.785)

259

25 mm (1 in.) length

127 (0.196)

546

190 (0.295)

470

253 (0.392)

404

13 mm (0.5 in.) length

63.3 (0.098)

861

95.0 (0.147)

737

127 (0.196)

622

6.4 mm (0.25 in.) length

31.7 (0.049)

1,293

47.5 (0.074)

1,090

63.3 (0.098)

890

3.2 mm (0.125 in.) length

15.8 (0.025)

1,969

23.8 (0.037)

1,562

31.7 (0.049)

1,226

1.6 mm (0.0625 in.) length

7.9 (0.012)

2654

11.8 (0.018)

2,223

15.8 (0.025)

1,930

1 Mum = 0.0394 mil

Table 66. Finite element results for three-dimensional model, 76-`mm (3-inch) cover.

Corrosion Pattern 13-mm (0.5-in.) Bar Diameter 19-mm (0.75-in.) Bar Diameter 25-mm (1-in.) Bar Diameter
Exposed
Area,
mm2 (in2)
Corrosion
Loss at Cracking,
Mu m
Exposed
Area,
mm2 (in2)
Corrosion
Loss at
Cracking,
Mu m
Exposed Area,
mm2 (in2)
Corrosion Loss at Cracking,
Mu m

Full Ring

508 mm (20 in.) length

10,134 (15.7)

57

15,201 (23.6)

56

20,268 (31.4)

54

203 mm (8 in.) length

4,054 (6.28)

102

6,080 (9.42)

152

8,107 (12.6)

83

102 mm (4 in.) length

2,027 (3.14)

216

3,040 (4.71)

267

4,054 (6.28)

108

51 mm (2 in.) length

1,013 (1.57)

445

1,520 (2.36)

406

2,027 (3.14)

267

25 mm (1 in.) length

507 (0.785)

660

760 (1.18)

584

1,013 (1.57)

446

13 mm (0.5 in.) length

253 (0.392)

927

380 (0.589)

813

507 (0.785)

611

6.4 mm (0.25 in.) length

127 (0.196)

1,295

190 (0.295)

1,067

253 (0.392)

853

3.2 mm (0.125 in.) length

63.3 (0.098)

1,689

95 (0.147)

1,321

127 (0.196)

1,116

Half Ring

102 mm (4 in.) length

1,013 (1.57)

368

1,520 (2.36)

356

2,027 (3.14)

328

51 mm (2 in.) length

507 (0.785)

559

760 (1.18)

521

1,013 (1.57)

483

25 mm (1 in.) length

253 (0.392)

838

380 (0.589)

762

507 (0.785)

693

13 mm (0.5 in.) length

127 (0.196)

1,219

190 (0.295)

1,118

253 (0.392)

968

6.4 mm (0.25 in.) length

63.3 (0.098)

1,676

95.0 (0.147)

1,461

127 (0.196)

1,283

3.2 mm (0.125 in.) length

31.7 (0.049)

2,261

47.5 (0.074)

1,842

63.3 (0.098)

1,689

Quarter Ring

102 mm (4 in.) length

507 (0.785)

508

760 (1.18)

445

1,013 (1.57)

394

51 mm (2 in.) length

253 (0.392)

737

380 (0.589)

622

507 (0.785)

610

25 mm (1 in.) length

127 (0.196)

1,067

190 (0.295)

927

253 (0.392)

902

13 mm (0.5 in.) length

63.3 (0.098)

1,626

95.0 (0.147)

1,308

127 (0.196)

1,270

6.4 mm (0.25 in.) length

31.7 (0.049)

2,350

47.5 (0.074)

1,791

63.3 (0.098)

1,702

3.2 mm (0.125 in.) length

15.8 (0.025)

3,226

23.8 (0.037)

2,502

31.7 (0.049)

2,273

1.6 mm (0.0625 in.) length

7.9 (0.012)

4,343

11.8 (0.018)

3,543

15.8 (0.025)

3,226

1 Mum = 0.0394 mil

Three-dimensional finite element models designed to simulate uniform corrosion of a reinforcing bar provide results similar to those obtained from the two-dimensional model. All models show a difference in corrosion loss to crack concrete between two- and three-dimensional models of less than 1 µm (0.4 mil), with the exception of models with a 25-mm (1-inch)-diameter bar and 76-mm (3-inch) cover, which shows a 2.9-µm (0.11-mil) difference in corrosion loss to cause cracking between the two- and three-dimensional models, a 5.7 percent difference.

1 Mum = 0.0394 mil
1 inch = 25.4 mm

Figure 259. Graph. Corrosion loss to crack concrete for uniform general corrosion versus cover for two-and three-dimensional finite element model.

The number of variables studied in the three-dimensional finite element model makes plotting all data points on a single plot impractical. Instead, data subsets holding as many variables constant as possible are analyzed to determine the effect of a variable on the corrosion loss required to crack concrete. Furthermore, corroding area, bar diameter, length of corroding region, and damage pattern are not independent variables-specifying any three variables restricts the fourth to a single value. For this analysis, the effect of cover, bar diameter, corroding area, and corroding length are analyzed with the goal of creating an equation that reduces to figure 256 in the case of general corrosion. Corroding area is expressed as a fraction of the total area of the bar, Af(Af = Acorroding/Abar). Corroding length is expressed as a fraction of the total length of the bar, Lf(Lf = Lcorroding/Lbar).

The corrosion loss to crack concrete is plotted versus exposed area for a 13-mm (0.5-inch)-diameter bar with 51-mm (2-inch) cover in figure 260. A curve of the form xcrit = m(Af)b is fit to the data. Table 67 summarizes the values of m and b for all three-dimensional finite element models. Based on table 67, it may be reasonably assumed the constant b is equal to -0.6, while the constant A is dependent on other variables.

Comparing Af with corrosion loss suggests the corrosion loss required to crack concrete varies inversely with Af raised to the 0.6 power.

1 Mum = 0.0394 mil

Figure 260. Graph. Corrosion loss to crack concrete versus fraction of exposed area with best-fit line for 13-mm (0.5-inch)-diameter bar and 51-mm (2-inch) cover.

 

Table 67. Constants m and b for best-fit curve xcrit = m(Af)b to corrosion loss versus Af plots.

Bar Diameter, mm (inches) Cover, mm (inches) m b

13 (0.5)

51 (2)

41.11

-0.602

19 (0.75)

51 (2)

34.20

-0.597

25 (1)

51 (2)

26.60

-0.600

13 (0.5)

76 (3)

85.11

-0.587

19 (0.75)

76 (3)

78.97

-0.592

25 (1)

76 (3)

60.65

-0.589

Figure 261 shows the relationship between corrosion loss and cover for all bars with a corroding area of 1,013 mm2 (1.57 in2). Similar plots were analyzed for other exposed areas. For bars with a fixed damage pattern and diameter, increasing the cover from 51 to 76 mm (2 to 3 inches) approximately doubles the corrosion loss required to crack concrete. This suggests that for localized corrosion, the corrosion loss required to crack concrete is proportional to cover squared. For larger corroding areas, the relationship between corrosion loss and cover becomes linear, as shown in figure 252 and figure 256.

Comparing cover to corrosion loss shows when corrosion is limited to a localized region, the corrosion loss required to crack concrete varies approximately with the cover squared.

1 Mum = 0.0394 mil
1 inch = 25.4 mm

Figure 261. Graph. Corrosion loss to crack concrete xcrit versus cover C for 1,013-mm2 (1.57-in2) corroding area.

Figure 262 shows the relationship between corrosion loss and fractional corroding length Lf (Lf = Lcorroding/Lbar) for all bars with a corroding area of 1013 mm2 (1.57 in2). Similar plots are analyzed for other lengths. A best-fit power line to the data suggests a relationship between corrosion loss and fractional corroding length to the -0.1 power.

Comparing Lf  to corrosion loss shows the corrosion loss required to crack concrete varies inversely with Lf  raised to the 0.1 power.

1 Mum = 0.0394 mil

Figure 262. Graph. Corrosion loss to crack concrete xcrit versus Lf with best fit line for 1,013-mm2 (1.57-in2) corroding area.

Based on the data presented, the equations in figure 263 and figure 264 represent a potential relationship between corrosion loss and the variables in this study. The term 3Af -1 is required for localized corrosion and reduces to 1 for general corrosion.

x subscript crit equals 0.53 times by the quantity: C raised to the quantity 2 minus A subscript f, closed quantity, divided by the quantity D raised to the 0.38 power times L subscript f raised to the 0.1 power times A subscript f raised to the 0.6 power, closed quantity, that entire quotient plus 0.6, closed quantity, that quantity times 3 raised to the quantity A subscript f minus 1, closed quantity.

Figure 263. Equation. Potential relationship between corrosion loss and variables in English units.

 

x subscript crit equals 45 times by the quantity: C divided by 25.4, that quotient raised to the quantity 2 minus A subscript f, closed quantity, divided by the quantity D raised to the 0.38 power times L subscript f raised to the 0.1 power times A subscript f raised to the 0.6 power, closed quantity, that entire quotient plus 0.2, closed quantity, that quantity times 3 raised to the quantity A subscript f minus 1, closed quantity.

Figure 264. Equation. Potential relationship between corrosion loss and variables in SI units.

Where:

xcrit = Corrosion loss at crack initiation, mil or Mum.

C = Cover, inches or mm.

D = Bar diameter, inches or mm.

Lf = Fractional length of bar corroding, Lcorroding/Lbar.

Af = Fractional area of bar corroding, Acorroding/Abar.

Figure 265 compares the corrosion losses for the finite element models with the corrosion losses predicted by the equation. There is some scatter, but the equations in figure 263 and figure 264 provide a reasonable match with the results obtained from the finite element model.

The equations in figure 263 and figure 264 predict corrosion losses in line with those predicted by the finite element model.

1 Mum = 0.0394 mil

Figure 265. Graph. Corrosion loss to crack concrete for localized corrosion based on the finite element model results versus corrosion losses calculated by the equations in figure 263 and figure 264.

To further verify the accuracy of the equations in figure 263 and figure 264, the corrosion loss predicted by the equation is compared to the experimental data for localized corrosion of ECR, as well as experimental results presented by Rasheeduzzafar et al., Alonso et al., and Torres-Acosta and Sagues, which are summarized in table 68.(68,69,71) Data for generalized corrosion of steel are also included in the analysis to check the accuracy of the equation for bars with large corroding areas. The comparison is presented in figure 266 and figure 267 along with the comparison for the three-dimensional finite element model results shown in figure 265.

Table 68. Results from other research: corrosion loss to crack concrete (localized corrosion).

 

Cover, mm

Diameter, mm

Exposed Area, mm2

Bar Area, mm2

Exposed Length, mm

Bar Length, mm

Corrosion Loss, Mu m

Torres-Acosta and Sagues (72)

27.6

21

2,105

16,757

32

254

48.3

27.6

21

2,105

16,757

32

254

66.4

40.3

21

2,738

20,122

42

305

88.2

40.3

21

2,738

20,122

42

305

69.6

65.7

21

4,486

26,785

68

406

76.5

65.7

21

4,486

26,785

68

406

121.8

40.3

21

2,764

20,122

42

305

55.2

40.3

21

2,764

20,122

42

305

68.9

40.3

21

1,260

13,393

19

203

141.2

40.3

21

1,260

13,393

19

203

70.6

40.3

21

2,738

20,122

42

305

60.3

40.3

21

2,738

20,122

42

305

65.0

40.3

21

22,827

26,785

346

406

28.4

40.3

21

22,827

26,785

346

406

7.2

27.5

21

1,649

26,785

25

406

30.8

40.3

21

1,649

26,785

25

406

61.6

45

13

4,084

16,581

100

406

84.0

45

13

1,021

16,581

25

406

336.0

38

13

4,084

16,581

100

406

49.8

38

13

4,084

16,581

100

406

49.8

13

13

4,084

16,581

100

406

31.1

13

13

1,021

16,581

25

406

37.3

13

13

1,021

16,581

25

406

49.8

13

13

4,084

16,581

100

406

3B2

28.8

13

796

16,581

20

406

207.4

30.3

13

796

16,581

20

406

111.7

39

13

15,928

16,581

390

406

35.9

39

13

15,928

16,581

390

406

31.1

39

13

1,593

16,581

39

406

151.6

39

13

1,593

16,581

39

406

159.6

39

13

327

16,581

8

406

233.3

39

13

327

16,581

8

406

272.2

27.5

6.4

603

8,163

30

406

63.2

26.5

6.4

603

8,163

30

406

8B3

39

13

1,593

16,581

39

406

271.2

39

13

1,593

16,581

39

406

191.5

39

13

1,593

16,581

39

406

159.6

39

13

1,593

16,581

39

406

151.6

Table 68. Results from other research: corrosion loss to crack concrete (localized corrosion)-Continued.

 

Cover, mm

Diameter, mm

Exposed Area, mm2

Bar Area, mm2

Exposed Length, mm

Bar Length, mm

Corrosion Loss, Mu m

Alonso et al.(70)

19

1B6

17,448

17,448

381

381

15

15.2

8.0

9,550

9,550

381

381

20

30.4

16.0

19,101

19,101

381

381

25

30.4

16.0

19,101

19,101

381

381

28

30.4

16.0

19,101

19,101

381

381

30

49.4

15.9

19,024

19,024

381

381

31

49.4

11.8

14,041

14,041

381

381

51

68.4

15.5

18,558

18,558

381

381

55

68.4

9.8

11,665

11,665

381

381

68

19

1B6

17,448

17,448

381

381

25

19

1B6

17,448

17,448

381

381

18

29

38.1

91,207

91,207

381

381

3

Rasheeduzzafar et al.(69)

20

13

22,462

22,462

550

550

33

20

13

22,462

22,462

550

550

26

20

13

22,462

22,462

550

550

34

35

13

22,462

22,462

550

550

32

35

13

22,462

22,462

550

550

30

35

13

22,462

22,462

550

550

47

35

13

22,462

22,462

550

550

38

35

13

22,462

22,462

550

550

27

35

13

22,462

22,462

550

550

27

50

13

22,462

22,462

550

550

70

50

13

22,462

22,462

550

550

71

50

13

22,462

22,462

550

550

74

50

13

22,462

22,462

550

550

61

60

13

22,462

22,462

550

550

67

1 mm = 0.039 inches
1 mm2 = 0.00155 in2
1 Mum = 0.0394 mil

Plotting experimental data alongside the finite element model data shows that while there is much scatter in prior experimental results, the finite element model results generally agree with past experimental data, and figure 263 and figure 264 provide slightly conservative predictions for both.

1 Mum = 0.0394 mil

Figure 266. Graph. Corrosion loss in localized corrosion specimens versus corrosion loss predicted by the equation in figure 263 and figure 264 for three-dimensional finite element model and experimental data.

Even over the larger range of finite element model data, figure 263 figure 264 provide slightly conservative predictions for the corrosion loss required to crack concrete.

1 Mum = 0.0394 mil

Figure 267. Graph. Corrosion loss in localized corrosion specimens versus corrosion loss predicted by the equation in figure 263 and figure 264 for three-dimensional finite element model and experimental data (revised scale).

Figure 266 covers the range of the experimental data in table 68. There is a moderate degree of scatter for both the finite element model and experimental results, but the finite element model generally agrees with the experimental data. The equations in figure 263 and figure 264 provide a generally conservative estimate of the corrosion loss required to crack concrete based on both the experimental and finite element results; that is, in most cases, the equations in figure 263 and figure 264 underestimate the loss required to cause a crack to form.

The finite element models extend well beyond the range of experimental data (see figure 267); additional testing will be needed to verify the accuracy of the finite element model in this range. The KU specimens with actual corrosion losses between 350 and 900 Mum (14 and 35 mil) represent the epoxy-coated bars with half-rings and holes in the epoxy. Figure 263 and figure 264 are very conservative for these specimens, predicting losses of approximately 200 Mum (7.8 mil), compared to the 350 and 900 Mum (14 and 35 mil) range in actual losses. The equation is most conservative for the epoxy-coated specimens with two holes in the epoxy; these specimens are shown as open circles in figure 267, as the uncertainty in the exposed area due to blistering of the epoxy calls the accuracy of these data points into question.

DISCUSSION

Torres-Acosta and Sagues derived an expression, shown in figure 268, relating bar cover, bar diameter, and localized corrosion length with the corrosion loss required for crack initiation based on experimental results.(71)

x subscript crit equals 11 times the quotient C divided by D times by the quantity: C divided by L, that entire quotient plus 0.6, closed quantity, that quantity squared.

Figure 268. Equation. Torres-Acosta and Sagues' corrosion loss to crack initiation.

Where:

xcrit = Corrosion loss at crack initiation, Mum.

C = Cover, mm.

D = Bar diameter, mm.

L= Length of exposed steel, mm.

Figure 269 and figure 270 compare the corrosion losses predicted by the equation with the experimental data for localized corrosion of ECR presented in table 62, as well as the finite element results and the experimental results presented by Rasheeduzzafar et al., Alonso et al., and Torres-Acosta and Sagues, as done for the equations in figure 263 and figure 264 in figure 266 and figure 267.(68,69,71)

For specimens undergoing general corrosion, figure 268 provides predictions similar to but slightly less conservative than figure 263.

1 Mum = 0.0394 mil

Figure 269. Graph. Corrosion loss in localized corrosion specimens versus corrosion loss predicted by the equation in figure 268 with three-dimensional finite element model and experimental data.

For specimens undergoing localized corrosion, figure 268 becomes increasingly unconservative.

1 Mum = 0.0394 mil

Figure 270. Graph. Corrosion loss in localized corrosion specimens versus corrosion loss predicted by the equation in figure 268 with three-dimensional finite element model and experimental data (revised scale).

Comparing figure 266 and figure 269 shows that for bars that require less than 50 Mum (2 mil) of loss to crack concrete, the equations in figure 263 and figure 264 and the equation in figure 268 perform comparably. However, the equation developed by Torres-Acosta is less conservative based on both experimental and finite element model results for bars that require greater than 50 Mum (2 mil) of loss to crack concrete; that is, the corrosion loss required to crack concrete predicted by the equation in figure 268 is greater than the corrosion loss required to crack concrete in the test specimens and for many of the finite element results. The equations in figure 263 and figure 264, in contrast, are more conservative with respect to many of the experimental specimens. The equation in figure 268 overestimates a significant portion of the experimental results obtained by Torres-Acosta and Sagues, in one case predicting a corrosion loss of 173 Mum (6.81 mil) for a specimen that only required 63 Mum (2.48 mil) of loss to crack concrete. For all experimental specimens with actual losses greater than 60 Mum (2.4 mil), the equation in figure 268 overestimates the corrosion loss requited to crack concrete for over 75 percent of the specimens. In comparison, the equation in figure 263 and figure 264 overestimates the corrosion loss to crack concrete for only 14 percent of the specimens with actual losses greater than 60 Mum (2.4 mil).

Comparing the two expressions based on results from the finite element model suggests that the equation in figure 268 becomes increasingly inaccurate and unconservative for bars that require very large corrosion losses to crack concrete (see figure 270). Furthermore, results from the finite element models where greater than 1,200 Mum (47 mil) of loss is required to crack concrete do not appear in figure 270, as the equation in figure 268 predicts that greater than 2,000 Mum (79 mil) of loss is required to crack the concrete.

The ratio of experimentally obtained corrosion losses required to crack concrete to the corrosion losses obtained by the two equations is also used to judge the degree of conservatism in each equation. A ratio less than 1.0 indicates an unconservative estimate for that specimen. Figure 271 and figure 272 compare this ratio for each equation based on corrosion losses obtained from experimental and finite element model results, respectively. Over the range of available experimental data, the two equations perform comparably, with the equations in figure 263 and figure 264 being more conservative for systems where actual losses exceeded 50 Mum (2 mil). As previously discussed, the available experimental data involves exposed areas far larger than those typically observed on damaged ECR. The specimens with damaged ECR tested as part of this study developed blisters that greatly increased the exposed area; therefore, only finite element model results are available for small exposed areas. Over the range of finite element model data, the equation in figure 268 rapidly becomes unconservative, as noted by the large percentage of ratios of finite element model-predicted corrosion losses to figure 268-predicted losses that are much less than 1.0 for models with expected corrosion losses greater than 500 Mum (20 mil). In contrast, the equations in figure 263 and figure 264 do not exhibit this behavior. Therefore, the equations in figure 263 and figure 264 are used to determine the corrosion loss required to crack concrete for damaged ECR.

For specimens undergoing general corrosion, figure 268 provides predictions similar to but slightly less conservative than figure 263. A wide amount of scatter is seen in both data sets, reflecting the variability of the experimental results.

1 Mum = 0.0394 mil

Figure 271. Graph. Ratio of experimentally derived corrosion loss to predicted corrosion loss versus corrosion loss to crack concrete based on experimental data.

This scale reflects corrosion losses for specimens and models undergoing localized corrosion. For specimens undergoing localized corrosion, figure 263 provides somewhat accurate predictions, with the ratio of experimentally derived loss to predicted loss remaining near 1.0. Figure 268 provides increasingly unconservative predictions, with the ratio of experimentally derived loss to predicted loss dropping below 0.2 at larger losses.

1 Mum = 0.0394 mil

Figure 272. Graph. Ratio of finite element model-derived corrosion loss to predicted corrosion loss versus corrosion loss to crack concrete based on finite element model.

 

 

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101