Condition and Performance of Epoxy-Coated Rebars in Bridge Decks
Fusion-bonded epoxy coating technology has been the primary defense used by most state highway agencies, including the Pennsylvania Department of Transportation (PennDOT) and the New York State Department of Transportation (NYSDOT), to protect reinforcing steel in concrete from corrosion. However, results from recent research activities cast doubt on the ability of epoxy coatings to provide long-term corrosion protection to steel in concrete exposed to chlorides.1-3 The problem has been attributed to the number and size of breaks or defects in the coating and to reduction in adhesion between the epoxy coating and steel substrate.
Coating breaks or defects are represented by narrow cracks; mashed areas; holidays (an area that is inadvertently skipped during the application of the coating); and other bare areas that occur during fabrication bending, transportation, handling, and concrete vibration.
Reduction of the adhesion bond between the epoxy coating and underlying steel has been widely observed. This phenomenon is independent of the amount of chlorides to which the reinforcing steel is exposed, and it may or may not be associated with corrosion underneath the coating.
Because there is much controversy about the effects of these problems on the long-term performance of epoxy-coated reinforcing steel in concrete bridges and structures exposed to salt, PennDOT and NYSDOT decided to conduct a joint research project, using a statistically based sampling plan, to evaluate in-service bridge decks constructed with epoxy-coated reinforcing steel. Project funding was provided through a regional pooled fund and by the Federal Highway Administration.
This article is a summary of the project findings. A detailed report on the project is available from PennDOT.4
The objectives of this project were to: (1) investigate the field performance of epoxy-coated reinforcing steel in selected bridge decks in Pennsylvania and New York, (2) determine if ongoing or progressive corrosion and/or a reduction in coating adhesion were occurring, with a 95-percent confidence level, in more than 3 percent of the deck area in either state, and (3) define variables associated with exposure conditions, concrete properties, and epoxy- coated rebar (reinforcing bar) properties that predict the presence or absence of corrosion and/or adhesion reduction.
"Performance" means the effectiveness of the coating in providing corrosion protection after the bar has been exposed to a corrosive environment for a minimal period of time. The existing condition of an epoxy-coated bar may be excellent, but if it has not been exposed to a corrosive environment for a sufficient length of time, its performance cannot be assessed.
"Progressive corrosion" is a process that is presently ongoing. For example, while visible rust indicates that a steel element has corroded, the element may or may not still be corroding.
|Table 1. Description of Salt Strata.|
|Salt Strata||New York tons/lane-mile/year||Pennsylvania tons/lane-mile/year|
|2||15 < x < 20||> 15|
|3||< 15||9 < x < 15|
|Table 2. Distribution of Spans by Salt Strata, ADT, Age Strata, and DCR.|
|Average Daily Traffic (ADT)|
|0 to 9,999||26||38|
|10,000 to 19,999||6||2|
|20,000 to 29,999||3||0|
|30,000 to 39,999||4||0|
|40,000 to 49,999||0||0|
|50,000 to 60,000||1||0|
|less than 5||8||9|
|6 to 10||14||18|
|11 to 15||14||13|
|16 to 20||4||0|
|Deck Condition Rating (DCR)|
|Table 3. Statistics for Bridge Information.|
Approach and Scope of Project
Systems for bridge inventory, inspection, and management were examined to identify each state's study population - bridge decks constructed with epoxy-coated reinforcing steel between 1977 and 1993, inclusively. Construction dates preceding 1977 were omitted to eliminate decks constructed with "Flintflex" epoxy coating. In New York, 1,425 bridge decks with a total of 3,360 spans and 16.6 million square feet (1.54 million square meters) of deck area were identified. The analysis in Pennsylvania found 2,290 bridge decks with a total of 6,062 spans and 26 million square feet (2.42 million square meters) of deck area.
For logistic purposes, the identified spans were reduced to a smaller sample. The decks in the smaller sample were representative of the study population in terms of age, salt usage, and deck condition. From the smaller sample, 40 spans in New York and 40 spans in Pennsylvania were randomly selected. The statistical sampling plan required the collection of three cores from each of the 80 spans.
|Table 4. Deck Condition Ratings|
|N||Not applicable: Concrete surface is not visible, for example, due to an overlay or sealer|
|9||Excellent Condition: No problems noted, generally used for a new structure|
|8||Very Good Condition: No problems noted, generally used for an old structure|
|7||Good Condition: Some minor problems|
|6||Satisfactory Condition: Less than 2% spalls or sum of all deteriorated deck concrete less than 20%|
|5||Fair Condition: Less than 5% spalls or sum of all deteriorated deck concrete 20% to 40%)|
|4||Poor Condition: Greater than 5% spalls or sum of all deteriorated deck concrete 40% to 60%|
|3||Extensive Deterioration: Greater than 5% spalls or sum of all deteriorated deck concrete greater than 60%)|
|Table 5. Statistics for Rebar Cover and EIS|
|Top Bar Cover, in.||238||1.6||4||2.7||2.7||0.4|
|Bot. Bar Cover, in.||235||2.2||4.5||3.3||3.3||0.4|
A detailed methodology for randomly locating and collecting three cores from each of the 80 spans (total of 240 cores) was developed. Prior to collecting the cores, delamination and crack surveys were performed in a 1- square-foot (0.093-square-meter) area surrounding each core sampling location. Although an effort was made to avoid collecting cores from areas that contained cracks, 16 percent of the extracted cores contained cracks. In each core, two intersecting rebars were extracted for evaluation. The condition of the bridge deck in each span was visually evaluated by the researchers using a deck-condition rating scheme similar to the National Bridge Inspection Standards. The methodology also included procedures for shipping and storing cores prior to testing in the "as extracted" condition as much as possible. The field evaluations and sampling plan were conducted from Sept. 4 to 19, 1996, in New York and from Sept. 30 to Oct. 15, 1996, in Pennsylvania.
The project also had a detailed laboratory methodology for documenting the condition and specific properties of the cores, concrete, and epoxy-coated rebars. Standard test methods were used when available. Measurements and tests were performed on each core and its two extracted rebars over a two-year period. Core evaluations and concrete property evaluations were performed on 240 cores, and epoxy-coating condition and property evaluations were conducted on 473 rebars.
Each core contained two intersecting epoxy-coated rebars from the top mat reinforcement; thus, a total of 480 rebars were extracted for evaluation. Of these, seven were too small to perform the prescribed laboratory procedures. As a result, the database generated in this study consists of 473 sets of data; one for each rebar evaluated.
The data obtained in this study were grouped into four categories: (1) bridge information, (2) core information, (3) concrete properties, and (4) epoxy-coated rebar condition and properties.
With the exception of the bridge deck rating, all the bridge information was obtained from the local agencies responsible for the bridge structures or from the appropriate state department of transportation (DOT). The visual condition of the deck in each span was rated during the field evaluation.
The selected spans were located in 13 counties in five districts in New York and in 16 counties in four districts in Pennsylvania. To account for variations in deicing salt applications, the four salt strata shown in table 1 were defined and used in the selection of spans for sampling. In some cases, variations in salt usage between the two states resulted in diferent levels of salt usage for the same salt stratum.
|Table 6. Statistics for Concrete Properties.|
|Unit Weight, pcf||240||64||206||134||135||10|
|Volume Pore Space, %||146||8.3||21.7||14.1||14.2||2|
|Rebar Trace pH||459||11||13||11||11||0.1|
|Total Chloride Content, pcy||457||0||11.37||0.95||0.47||1.34|
|Chloride Exposure, yrs||473||0||16||1||0||3|
Average daily traffic (ADT) information was obtained from state DOT databases and the resulting distribution is presented in table 2.
Two age strata were defined and used in the span-selection process. Spans constructed from 1985 to 1993 were categorized as Age Stratum 1 and those constructed from 1977 to 1984 were categorized as Age Stratum 2. The distribution of bridges by age stratum is provided in table 2.
The age of each selected bridge was obtained from state DOT databases. Some of the older structures had undergone deck replacement. In such cases, the age was calculated from the date of deck replacement. Table 2 presents a summary of the age distribution for all spans included in this study, and statistical information is shown in table 3.
For each of the 79 spans, visual evaluation of the deck surface was performed during the sample collection effort. One span in New York was not rated because it had an asphalt overlay. The distribution of spans by deck- condition rating is presented in table 2. Table 4 explains the deck-condition rating scheme.
Data in this category were collected from within the circular sampling area and from the extracted core samples.
A delamination survey was performed within each circular sampling area. No delaminations were found at any of the sampling locations.
Cracks within the circular sampling area were observed. The project panel decided that it would be more prudent to evaluate the performance of epoxy-coated rebars in uncracked concrete areas, and although an effort was made to avoid cracks, 16 percent (39 cores) of the extracted cores contained cracks.
Due to the drilling depth used to extract cores, several cores unintentionally contained bottom-mat reinforcement - not the bottom bar of the top mat, but the bar from the second layer of steel in the deck. Each core was visually observed for the presence of rust staining or honeycombing. Rust staining was observed on seven cores, and honeycombing was noted on one core. Rust staining on two of the seven cores was directly related to corrosion of the epoxy-coated bars.
The clear concrete cover over the top rebar was measured. The concrete cover over the bottom rebar was calculated by adding the diameter of the top rebar to the cover measurement for the top rebar. Three cores had a concrete overlay, and three other cores had a thin asphalt overlay. Statistical information for top and bottom rebar cover is shown in table 5.
Electrochemical impedance spectroscopy (EIS) testing was conducted on 178 cores, and the impedance obtained at 0.1 Hz was documented.2 The EIS technique is a nondestructive test method that has been successfully used to study the deterioration of coated metals. The primary purpose of conducting EIS tests in this study was to determine whether EIS test results could accurately provide information regarding the condition of epoxy-coated reinforcing steel in concrete. Statistical EIS test results are presented in table 5.
|Table 7. Statistics for Epoxy Coated Rebars.|
|# of Mashed Areas||473||0||20||2.1||2||2.2|
|# of Bare Areas||473||0||21||2.4||2||2.6|
|# of Holidays||473||0||156||7.7||3||15.8|
|Coating Thickness, mils||473||2.4||21.9||11.2||11.1||2.8|
|Corrosion Condition Rating||473||1||4||1.1||1||0.4|
Measurement of concrete properties included the specific gravity, permeability, absorption, percentage of volume pore space, acid-alkali balance (pH), and chloride ion content. These measurements are presented in table 6.
Chloride ion content data were obtained for a total of 457 rebars. In some cores, sufficient concrete was not available to collect the required sample. Figure 1 presents the percentage of rebar that had a chloride content in the surrounding concrete at or greater than a given level. The percentage of chloride content was converted to pounds per cubic yard (pcy) using the measured bulk dry specific gravity of the concrete in each core.
Determination of the time that rebars have been exposed to at least 1.2 pcy of chloride concentration (for brevity, this is termed chloride exposure in this report) is difficult as this is a function of concrete quality and exposure conditions (salt application rate, environment, etc.). Chloride exposure can best be estimated using Fick's law of diffusion. A diffusion constant was calculated for each rebar, and the same equation was used to calculate the age at which the chloride ion content at the rebar depth reached 1.2 pcy. If the results of this calculation were less than the age of the structure, the result was subtracted from the age of the structure to obtain the chloride exposure for 1.2 pcy. If the calculation resulted in a value greater than the age of the structure, the chloride exposure was considered to be zero. Statistical information for chloride exposure is presented in table 6, and the resulting continuous cumulative distribution is presented in figure 2. The ordinate of this plot shows the median chloride exposure for each range except for the chloride exposure of zero.
Epoxy-Coated Rebar Condition and Properties
Tests and observations on the epoxy-coated rebars, including the color of the epoxy; defects such as mashed areas, blisters, bare areas, and holidays; deformation pattern; coating thickness; coating hardness; corrosion-condition rating; and adhesion rating are summarized in table 7.
The number of mashed areas, blisters, bare areas, and holidays was documented for each extracted rebar. No blisters were detected on any of the rebars tested. The deformation pattern on each rebar was classified.
Coating thickness was measured on four different sections of each rebar. Each bar was divided into a top half (closest to the deck surface) and a bottom half. In each of these sections, coating thickness was measured at three locations on the ribs and in the valleys between the ribs. The data were then averaged to provide a total of four coating thickness results for each bar. No variation in coating thickness between the two bar sections (top half and the bottom half) was expected, and none was detected. However, minor variations in coating thickness on ribs compared to valleys were noted; coating thickness on the ribs averaged 1.2 mils more than in the valleys.
The pencil-hardness test method was used to determine the hardness of the coating. This test uses pencils varying in hardness from 6B to 9H. Of the 473 bars tested, 98 percent had a pencil hardness of HB, which is about in the middle of 6B to 9H pencil-hardness spectrum. These results are not significantly different from those obtained on epoxy-coated bars collected from job sites prior to concrete placement.4
Table 8 explains the corrosion-condition rating scheme, and figure 3 presents photographs depicting the visual condition of bars with corrosion ratings of 1, 2, and 4. Of the 473 rebars rated, 86.5 percent showed no evidence of corrosion (rating 1), 13.1 percent had a small number of spots of corrosion (rating 2), and 0.4 percent were corroded more than 20 percent to 60 percent of their surface area (rating 4).
Figure 4 presents photographs showing the typical condition observed for adhesion ratings 1, 3, and 5. These ratings are explained in table 9.
|Table 8. Corrosion Ratings|
|1||No evidence of corrosion|
|2||A number of small, countable corrosion spots|
|3||Corrosion area less than 20% of total ECR surface area|
|4||Corrosion area between 20% to 60%c of total ECR surface area|
|5||Corrosion area greater than 60% of total ECR surface area|
Coating adhesion was measured in the wet (as-received condition) and dry (after exposure for 7 days in a desicator) conditions. Test locations included areas with no visible coating defects and areas directly adjacent to visible coating defects. In each case, the data were averaged to provide a single adhesion rating. Distribution of adhesion ratings was not significantly different for wet versus dry conditions.
The distribution of wet/no defect adhesion ratings for the 473 bars tested was: 46.5 percent at rating 1, 22.2 percent at rating 2, 8 percent at rating 3, 10 percent at rating 4, and 13.3 percent at rating 5.
It has been reported in recent investigations on existing structures that adhesion reduction is a function of age.1,3 An analysis of all individual adhesion ratings versus age strata 1 and 2 was performed. Results from these tests showed that adhesion exhibited by rebars from structures in the two age strata investigated was significantly different and that age stratum 1 (bridges constructed from 1985 to 1993) exhibited much lower (better) adhesion ratings compared to age stratum 2 (bridges constructed from 1977 to 1984). These analyses suggest that adhesion reduction or loss is related to some threshold age or range of ages.
|Table 9. Adhesion Test Ratings|
|1||Well-adhered coating that cannot be peeled or lifted from the steel substrate|
|3||Coating that can be pried from the steel substrate in small pieces, but cannot be peeled off easily|
|5||Coating that can be peeled from the steel surface easily, without residue|
Further study to analyze this relationship was conducted. The probability of some adhesion reduction was 24 percent for bridges less than five years old, 54 percent for bridges from six to 10 years old, 63 percent for bridges from 11 to 15 years old, and 70 percent for bridges from 16 to 20 years old.
The sampling plan was designed to detect progressive corrosion even if the frequency of occurrence was less than 3 percent of the global population. Results from the study showed that the frequency of occurrence of progressive corrosion is less than 3 percent in Pennsylvania and at least 3 percent in New York.
A total of 409 rebars showed no evidence of corrosion; 62 bars had a number of small, countable corrosion spots; and only two rebars, both in New York, exhibited significant visible corrosion. In addition, corrosion products were observed under the coating on approximately 7 percent of the bars tested. However, in most cases, corrosion products found under the coating were not the result of ongoing corrosion. Therefore, the existing condition of epoxy-coated rebars - that is, the condition at the time the bar was examined - in bridge decks in Pennsylvania and New York State was found to be very good from a corrosion point of view.
Coating adhesion reduction or loss was found to be more prevalent and extensive. Only 47 percent of the bars tested had no reduction in adhesion. More than 13 percent exhibited a complete loss of adhesion, and the remaining 40 percent had varying degrees of adhesion reduction. Probability distribution analyses showed that more than 50 percent of epoxy-coated rebars in bridge decks in Pennsylvania and New York exhibit some degree of adhesion reduction within six to 10 years of placement in concrete.
It should be pointed out that, although progressive corrosion must be accompanied by complete adhesion loss, coating adhesion alone was not found to be a good predictor of corrosion condition in this study.
No correlation was found with corrosion-condition rating and chloride exposure time or chloride content. This is most likely attributable to the age and chloride content distribution of the study population and/or satisfactory performance of the epoxy-coated bars.
Among all the variables included in the analyses, logarithm of EIS, number of holidays, and number of bare spots were found to be the best predictors of corrosion-condition rating, but in all cases, the correlations were weak. These same parameters were also found to have statistically significant relationships with adhesion reduction, but again, the correlations were weak.
Corrosion-condition rating did not correlate with coating thickness, clear concrete cover, color of epoxy, or bridge deck-condition rating.
Adhesion reduction or loss is irreversible at least after a seven-day drying period. There is a higher probability of adhesion reduction adjacent to areas with visible coating defects compared to those with no visible defects. In addition, the deformation pattern on the bars has some impact on adhesion reduction.
A good correlation between concrete resistivity and coulombs passed was found and the following equation can be used to describe the relationship: Coulombs Passed = 2E+09*A/C Resistivity-1.4539.
Results of pH testing in rebar traces and pencil-hardness testing on the coating did not provide any useful information.
1. A.A. Sagues. Corrosion of Epoxy-Coated Rebar in Florida Bridges, Final Report to Florida Department of Transportation, WPI No. 0510603, University of South Florida, Tampa, Fla., May 1994.
2. K.C. Clear, W.H. Hartt, McIntyre, and S.K. Lee. Performance of Epoxy-Coated Reinforcing Steel in Highway Bridges, NCHRP Report No. 370, Transportation Research Board, Washington, D.C., 1995.
3. R.E. Weyers, W. Pyc, J. Zemajtis, Y. Liu, D. Mokarem, and M.M. Sprinkel. "Field Investigation of Corrosion-Protection Performance of Bridge Decks Constructed With Epoxy-Coated Reinforcing Steel in Virginia," Transportation Research Record No. 1597, Transportation Research Board, Washington, D.C., 1997, pp. 82-90.
4. A. Sohanghpurwala, and W.T. Scannell. Verification of Effectiveness of Epoxy-Coated Rebars, Final Report to Pennsylvania Department of Transportation, Project No. 94-05.
Ali Akbar Sohanghpurwala and William T. Scannall are the principals of CONCORR Inc., the largest firm in the United States solely devoted to corrosion and corrosion protection of metals embedded in concrete. The authors have been involved in research and evaluation of many of the corrosion-mitigation technologies used for reinforced- concrete highway structures. The authors are also co-inventors of a modified graphite reference electrode for controlling cathodic protection systems.