Since our total investment in the nation's highway infrastructure amounts in the billions of dollars, it is extremely important that all possible methods applicable to controlling corrosion in existing concrete bridges be developed so that these structures will not deteriorate prematurely. Equally important is developing methods to avoid this costly corrosion problem in all new concrete bridges to be constructed in the future. Accordingly, the research program undertaken in this HPA was divided into two major areas:

• Corrosion Control in New Concrete Constructions.
• Corrosion Control for Rehabilitation of Existing Concrete Structures.

To be effective, this research program involved not only the participation of FHWA, but other partners including another federal agency (National Institute of Standards and Technology), some state highway agencies, the academia, and the private industry. The following is a presentation of results or accomplishments achieved to date in these two areas.

A. Corrosion Control in New Concrete Constructions

Given the very harsh service environments that many bridges will be exposed to typically, it is extremely difficult - but not impossible - to build reinforced concrete bridges that would be free of steel reinforcement corrosion. The achievement of this goal would require the adoption of system approach, i.e., using a combination of different measures, such as adequate depth of concrete cover, quality concrete, corrosion inhibitors admixture, and corrosion-resistant reinforcement.

1. Adequate Concrete Cover

It is now widely accepted that, for a concrete structure to be durable in a corrosive environment, it is absolutely necessary to provide an adequate layer of concrete or depth of concrete cover over the first layer of reinforcing steel so that it would not be easy for chloride ions to reach the steel. This adequate depth of cover can be determined by application of Fick's second law of diffusion, which adequately models the intrusion or diffusion of chloride through a porous material such as concrete (17):

(C x/ t) = D( 2C x/ 2)

(12)

where, C _x is the concentration of chloride at depth c at time t, D is the diffusion coefficient of chloride. The solution to this equation for a semi-infinite slab is:

C(,t) = Co { 1—erf [ / (2 (Dt)½)] }

(13)

which is applicable to concrete structures where the chloride ions enter from one direction, such as concrete bridge decks and piers. Using this relationship, the minimum depth of concrete cover over the reinforcing steel that would be required so that the total amount of chloride ions that will accumulate at the depth of steel will not exceed the corrosion threshold—before a desired service life is reached—can be determined from the measured chloride diffusivity of the selected concrete mix design and the anticipated level of exposure of the structure to deicing salts. Finally, this minimum depth of cover must be incorporated with expected construction tolerances (that accounts for typical construction variance) to achieve a rational depth of cover specification (18). For example, if a minimum concrete cover of 50 mm is required and statistical surveys of the cover on recently built bridge decks show that the cover on a deck typically assumes normal distribution with a standard deviation of approximately 10 mm, then the specified cover must be either 67 mm or 73 mm for 95-percent or 99-percent compliance, respectively.

However, even with adequate concrete cover, corrosion of reinforcement can still occur because, invariably, concrete will crack. In addition, presence of variances in the concrete cover and in the density of the placed concrete across a structure will eventually create corrosion micro-cells (consisting of cathodes and anodes), which drive steel corrosion. Therefore, other supplementary protective measures also need to be adopted in a new construction.

2. Quality Concrete

Recently, there is also a heightened awareness that the quality of concrete is of utmost importance in determining the durability of reinforced concrete bridge members exposed to chlorides and subjected to intermittent wetting. Although concrete is outwardly a dense material, it contains pores; and many of these pores are interconnected to form a network of channels that allows water and oxygen, both important to steel corrosion, to permeate into the concrete. And, as a general rule, low water/cement ratio and good consolidation lead to either a lower number of pores or smaller pores in the concrete, both of which can lead to reduce penneability. In addition, reduced permeability also leads to reduction in the electrical conductivity of the concrete by reducing not only the amount of moisture intrusion but also the amount of chloride ions carried by the moisture into the concrete. A low water/cement ratio also offers higher strength to the concrete, which would extend the time before stresses resulting from steel corrosion cause the concrete to crack. Therefore, mixture proportions must be carefully selected to the keep the water/cement ratio to an absolute minimum.

In addition to making state transportation agencies become aware of the importance of using low water/cement ratio in concrete mixes, some recent FHWA research efforts were aimed at identifying concrete materials that can consistently provide superior performance when used in construction of bridges. These efforts included investigation into how concrete material and mix variables, such as water-cement ratio, air content, coarse aggregate type, fine aggregate type, mineral admixture, and cement type, affect the corrosion behavior of steel (19). The test matrix adopted in this investigation is shown in Table 1.

Because only 30 concrete mix designs were used, which were insufficient to test all the possible combinations of the large number of variables, only weak correlation between corrosion rate or potential and these variables/types were obtained, for both moderate and severe environments. Nevertheless, it was possible to draw some general conclusions, the primary of which was that the corrosion rate of reinforcing steel varies significantly depending on the concrete mix components. Among the variables investigated, water/cement ratio, cement type, mineral admixture, and fine aggregate exhibited the most significant effects on the durability of the concrete. The other factors, such as coarse aggregate and air content, also affect behavior of steel in concrete, but to a lesser extent. The data indicated that the Type I cement (low C₃A), quartz fine and coarse aggregates, and silica fume appeared to be the best materials for improving resistance of concrete to deterioration (20).

The use of blended cements might, under certain circumstances, be detrimental because of a reduction in alkalinity in the concrete. However, this adverse effect may be more than offset by the beneficial effects that blended cements can offer, which include a substantial reduction in permeability and also a reduction in conductivity—especially where a reduction in the water-cement ratio is made possible.

In addition, the investigation also attempted to quantify the corrosive conditions fostering concrete bridge deterioration. The effects of environmental variables such as chloride concentration, temperature, and humidity, on the corrosion behavior of reinforcing steel in a typical concrete mix were examined (21). The adopted test matrix is shown in Table 2.

From the data collected, a statistical regression model was developed to permit prediction of the corrosion rate as a function of these three variables. Even though the model has only a correlation coefficient of 0.50, it serves to demonstrate that, given any chloride concentration, the environmental factors of temperature and humidity also significantly affect the corrosion behavior of steel in concrete. This explains why the problem of reinforcing steel corrosion is considerably more severe in the marine environment of states like Florida than in some other states.

The following are the most important findings resulting from these investigations:

• Chloride concentrations from 0.59 to 5.9 kg/m ³ (1 to 10 lb/yd ³) can produce a wide range of corrosion behavior of steel in concrete.



• Environmental variables such as temperature and humidity are also significant in influencing the corrosion behavior of steel. Intermediate levels of humidity or moisture provide aggressive environments than either low or high levels.



• The interactions of chloride ion concentration, temperature, and relative humidity on the corrosion behavior of steel are complex.



• The corrosion rate of reinforcing steel can vary significantly based on the concrete mix components. The most significant beneficial effects were obtained from use of low water-cement ratio and proper selection of cement type, mineral admixture, and fine aggregate. Factors such as coarse aggregate, and air content had a lesser effect.

3. Alternative Reinforcements

Even if concrete can be made to be extremely impermeable (by the addition of pozzolans like microsilica, etc.) and not conducive to steel corrosion (by the addition of an effective inhibitor), it still would not completely solve the corrosion problem, since concrete has a tendency to crack eventually, especially the high-performance concrete if proper curing is not observed. The final line of defense against corrosion of reinforcing steel would still lie with the reinforcing steel itself. Unfortunately, the resistance of mild steel to corrosion can not be significantly improved by just modifying its composition, grade, or the level of stress (22). Therefore, to prevent corrosion of steel reinforcement in concrete located in corrosive environment, either the conventional mild steel reinforcement must be coated with an effective and economical barrier to prevent contact with chloride, moisture, and oxygen, or reinforcement made of corrosion-resistant materials must be used.

Among the above two options, application of a suitable coating on the mild steel reinforcement may be the most economical. The coated reinforcing steel must be resistant to damages during transport from a plant to a construction site, storage at construction site, and placement in the structure. It must also be durable in severe service environment and capable of maintaining its structural function throughout the service life of the structure, and be economical.

a. Steel Bars With Organic Coating

After roadway deicing salts was identified as the cause of the premature deterioration of many concrete bridges, solutions to control this problems have to be developed since banning of deicing salts would not be acceptable to the motoring public. A solution that was identified as potentially viable and economical is the application of a suitable, stable organic coating on the reinforcing steel to serve as a barrier for isolating the steel from moisture, chloride ions, and oxygen, thereby preventing corrosion. Therefore, in the early 1970s, the Federal Highway Administration sponsored a project at the National Institute of Standards and Technology, then the National Bureau of Standards, to search for organic coatings suitable for this application. In that study, 47 coatings—including 36 epoxies (in both liquid and powder forms)—were evaluated in the laboratory (23). Four fusion-bonded epoxy powders emerged as the most promising of the coatings studied.

The first bridge utilizing epoxy powder-coated reinforcing steel bars (hereafter called epoxy-coated rebars) was built in 1973 in West Conshohocken, Pennsylvania. Subsequent trial of some of these coatings in bridge decks, under the National Experimental and Evaluation Program, revealed some problems such as damage to the coating during transport and handling, and cracking of coating (in the bend areas) rising from bending of bars at construction sites. To eliminate or alleviate these problems, measures such as bending the bars before coating, increasing bar supports during shipping (to prevent abrasion between bars), and using padded bundling bands and nylon slings during loading and unloading, were adopted.

In a subsequent FHWA study using relatively large concrete slabs, the performance of some epoxy-coated rebars that failed specifications—by having excessive holidays and surface damage (in excess of 0.8 percent) and failing bend tests—was compared with that of black steel (24). It was found that even though the coating on those rebars failed specifications, it was effective in reducing steel corrosion in salted concrete. Based on the data, it was estimated that, relatively, if it required 1 year to consume a given amount of black steel, then 12 years would be required to consume the same amount of the coated rebars, when the epoxy-coated rebars were used only in the top mat bars; or, 46 years would be required, if the coated rebars were used in both top and bottom mats. (The shorter protection accrued when the coated rebars were used only in the top mat was due to the formation of macro-cells between the different types of rebars in the two mats, especially when electrical couplings exist.) It appeared that the presence of the insulating epoxy coating caused significant increase in the electrical resistance between the top- and the bottom-mat of steel bars, which typically behaved as macro-anode and the macro-cathode, thereby retarding steel corrosion.

These findings had lessened the concern that a small exposed area of steel would be susceptible to intense corrosion and, thereby, enabled the requirement for patching or repair of coating damages to be waived when the damage is less than 2 percent in straight bars, 5 percent in bent bars, and 3 percent after placing. This led to decrease in the cost of epoxy-coated steel bars and their wider use in bridges by many states. However, in 1992, a final report for the Canadian Strategic Highway Research Program, "Effectiveness of Epoxy-Coated Reinforcing Steel," raised some questions about the long-term corrosion and structural performances of this type of steel bars, based on observation that some of the epoxy films became disbonded, blistered, and cracked (25). Since then, both ASTM and AASHTO specifications for epoxy-coated rebars, on the presence of holidays and bare area and on coating thickness, were revised and tightened. The Canadian report also urged that the effect of coating adhesion loss on the structural bond and creep properties of concrete members reinforced with epoxy-coated bars be investigated.

To study this important issue, the FHWA conducted a series of short-term experiments to assess the possible effects of disbondment of the epoxy coating from the steel bars on critical bond stress and flexural strength (26). For comparison, a series of pull-out concrete specimens and flexural reinforced concrete slabs were fabricated with: (a) epoxy-coated bars with different levels (20 to 30 percent) of intentionally induced disbondment in coating, (b) epoxy-coated bars with good coating, and (c) black steel bars. Although the investigation was limited in the number of tests conducted and in some other aspects (including bar size, deformation pattern, grade of steel, and slab design used), it was adequately designed to detect any significant differences in the behaviors of concrete fabricated with coating-disbonded bars and normal epoxy-coated bars.

The positive moment tests showed that the mere presence of epoxy coating on the bars, with either well- or partially-bonded coating, appeared to reduce the average ultimate diagonal tension capacity by 14 percent. When the positive-moment flexural behavior of different slabs were compared, no significant difference was observed between bars with partially bonded coating and bars with well bonded epoxy coating. Furthermore, no significant differences were observed in the comparison of the negative-moment flexural behaviors of the concrete slabs cast with the three types of bars. Results from the pull-out tests showed that the critical bond strengths for the bars with partially disbonded coatings were measurably lower than those for good epoxy-coated bars. However, even with 20 to 30 percent coating disbonded, the coated bars still developed 80 percent of the mean critical bond strength for bare bars. Essentially, the study indicated that even with 20- to 30-percent disbondment in the coating, the structural capacity of concrete reinforced with epoxy-coated steel bars was not compromised.

In response to the concerns raised in some portions of the United States, especially in Florida, about potential problem with the long-term durability of present epoxy-coated rebars, a series of investigations were initiated by FHWA in 1993 with the following objectives:

• To reexamine the effectiveness of epoxy coatings on the steel rebars in substructural bridge members exposed to a simulated marine environment and to identify the cause of the problem, if any.
• To determine the up-to-date overall performance of epoxy-coated rebars in concrete bridge decks exposed to actual severe service conditions.
• To search for the most corrosion-resistant organic coating systems for steel bars.

The following are presentation of the various findings resulting from these investigations.

(1) Reexamination of the Performance of Epoxy-Coated Rebars in Concrete Structures Exposed to a Marine Environment.

Following some early success of epoxy-coated rebars in controlling corrosion in bridge decks, some state highway agencies started using them in substructure concrete members in the marine environments, despite some major differences between the exposure conditions existing in bridge deck and marine applications. Reinforced concrete in a marine environment is subjected to:

• Relatively higher temperature throughout the year.
• A continuous supply of moisture and chloride.
• More wetting and drying cycles to provide penetration of oxygen and chloride.
  

All these conditions contribute toward lowering the resistivity of the concrete and thereby facilitating the flow of corrosion current between anodes and cathodes in both micro and macro levels.

Therefore, it was not totally surprising when in the late 1980s, reports about the premature deterioration of a number of substructure concrete members in the splash zones of three bridges located in the Florida Keys began to circulate. Even though there were different opinions about the quality of materials and construction practices employed in those bridges, nevertheless it was not expected to encounter in such a short period the severity of corrosion observed. This report led to claims that: (a) when exposed to marine environments, epoxy-coated rebars are more susceptible to corrosion than bare rebars; (b) epoxy coatings are, therefore, not effective in providing long-term protection to rebars in salt-contaminated concrete, even in bridge decks; and (c) the technology of organic-coating of rebars, as practiced in North America, is (probably) flawed. Responding to this concern, a study was initiated to reexamine the effectiveness of epoxy coatings on the steel rebars in substructure bridge members exposed to a simulated marine environment and to identify the cause of the problem, if any (27).

In this study, two parallel sets of experiments were conducted to investigate: (a) anodic growth as a function of immersion time; and (b) cathodic disbondment plus wet-adhesion loss. Quantification of anodic growth was made using (nondestructive) infrared thermography; whereas, quantification of disbondment was made with a specially developed peel test apparatus. The variables included in each experiment were: (a) two commercial epoxy coatings — designated as Coating 1 and Coating 2; (b) two coating thicknesses—approximately 139 and 190 (5 to 8 mils); (c) two coating conditions—unscribed and scribed; (d) one immersion solution - saturated Ca(OH) ₂solution with 3.5 percent (by wt.) of NaCl; and (e) two immersion-solution temperatures—35°C and 50°C (95°F and 122°F). Approximately 200 steel panels coated with the two coatings were used in the experiments. The following observations were noted:

• Regardless of the coating used, none of the unscribed panels exhibited any visible sign of corrosion, even after 3,072 hours in the immersion solution at 35°C.

• As early as 2 to 5 hours after immersion at either 35 or 50°C, some of the scribed panels started to have anodic sites, within 2 to 4 mm from the edge of the scribes. Then, within 24 hours, all the scribed panels started to corrode and produced a black pasty material, that was thought to be magnetite (Fe ₃O ₄), underneath the coating. This confirmed that the corrosion rates of epoxy-coated steel panels depend largely on the presence or absence of defects through the thickness of the coating, since the diffusivity of chloride and sodium ions through intact coatings is known to be low.

• An increase in the temperature of the immersion solution from 35 to 50°C greatly accelerated the rate of corrosion of the scribed panels. The scribes acted like cathodes and remained free of corrosion, while corrosion take place around the scribes.

• Anodic sites have tendency to grow in size. When the number of anodic sites and the total area covered by the anodic sites were measured for panels immersed in the 50°C solution, it was found that after approximately 100 hours of immersion, the accumulative number of anodic sites did not change but that the total anodic area increased. After 500 hours of immersion at 35°C (or 100 to 500 h at 50°C), liquid- filled blisters started to form above each anodic site. Analysis of the blister fluid indicated that the concentration of Cl ^- was 4 to 6 times greater than that of the bulk electrolyte, while the concentration of Na ⁺ was about the same as in the bulk solution. The pH of the fluid ranged from 5 to 5.5.

• This increased activity at the anodic sites was accompanied by an increase in cathodic reaction at the edge of the anodic sites, which produces OH^- and causes the coating to disbond from the steel. This process of cathodic disbondment contributed to a reduction in the strength of the bond between the coating and the steel. It was also found that the rate of cathodic disbondment was not affected by the type and the thickness of the coating, but was affected by the temperature of the immersion solution.

• At beyond the disbondment front, loss of bond strength occurred through wet-adhesion loss, whereby water molecules accumulated between the coating and the steel substrate. This process was influenced by coating type and immersion temperature, but not by coating thickness. However, unlike cathodic disbondment, wet-adhesion loss was found to be recoverable once the coating is dried.

• Eventually through the combined action of cathodic disbondment and wet-adhesion loss, the bond strength of the coating to the steel panels was lost so that the coating could be easily removed.

The basic implication of all these findings is that for epoxy-coated rebars to protect reinforcing steel from corrosion, it's extremely important that the coating is free of any significant damage before being embedded in concrete and special care be observed to avoid damaging the coating during concrete pouring and consolidation. Hence, strict compliance with specifications and use of vibrators equipped with rubber-lined heads were strongly recommended.

(2) Assessment of the Performance of Epoxy-Coated Rebars in Bridge Decks

Recently, 11 state highway agencies in the U. S. and the Canadian SHRP conducted investigations to evaluate the performance of epoxy-coated rebars on 92 bridge decks, 2 bridge baffler walls, and 1 noise barrier wall, which were exposed to harsh service conditions. To shed some light on the controversy regarding the performance of epoxy-coated rebars, FHWA assembled and analyzed the data available from these investigations (28). The bridge decks evaluated included some that were constructed with epoxy-coated rebars used in the top mat of reinforcement only and in both the top and the bottom mats. In addition, 7 different types of epoxy powders were involved in these structures, where the epoxy-coated rebars had been in service up to 20 years at the time of the investigations. In-situ and laboratory evaluations of the concrete and the epoxy-coated rebars were typically included in these investigations. The methodology used in the field or in-situ evaluations included some or all of the following:

• Visual examination of the concrete for cracking, spalling, and patches.
• Chain dragging to locate concrete delamination.
• Use of pachometer to determine concrete cover and to locate reinforcing steel.
• Drilling of concrete powder samples for determination of chloride content.
• Coring for determination of the quality and the chloride content of the concrete.
• Overall rating of deck condition.
• Survey of half cell potential.
• Survey of concrete resistivity.
• Determination of corrosion rate through measurement of linear polarization resistance.

The laboratory evaluations used some or all of the following procedures:

• Visual examination of the concrete in the extracted cores.
• Measurement of the concrete cover over the ECR in the extracted cores.
• Evaluation of the segments of ECR extracted.
• Measurement of the thickness of epoxy coating on the extracted ECR segments.
• Determination of the total or water-soluble chloride content in the concrete.
• Determination of the permeability of the concrete (in the extracted cores).
• Determination of the pH of the concrete adjacent to the ECR in the extracted cores.
• Determination of the compressive strength of the concrete.
• Determination of the unit weight of the concrete.

The results from the various investigations were summarized according to:

• Concrete Cover and Chloride Content. All the structures investigated appeared to have adequate concrete cover—with only 10 (5.4 percent) out of 184 extracted cores that were measured had concrete covers of less than 51mm (2 in). The concrete surrounding the rebars in many of the structures had accumulated sufficient chloride ions to exceed 0.6 kg/m ³(1.0 lb/yd ³), which is considered by many as the corrosion threshold level. In fact, of the 40 bridge decks, wherein the average total chloride contents in the concrete were measured, 33 (83 percent) decks had chloride content equal to or greater than this threshold level. In addition, in 11(28 percent) decks the chloride content was > 3.0 kg/m ³(5.0 lb/yd ³), with the highest content being 6.8 kg/in ³(11.5 lb/yd ³). In another 16 bridge decks where the water-soluble chloride contents were determined, the chloride content was >0.6 kg/m ³ (1.0 lb/yd ³) in 5 (31 percent) of these decks.
  
  
• Condition of Epoxy Coating on the Rebars. Some of the segments of epoxy-coated rebars extracted from the cores were examined for holidays, thickness of the epoxy coating, and, on some selected bars, the blast profiles. Most, if not all, of the segments examined contained holidays or bared areas. The thickness of the coatings was generally within the limits specified at the time of construction. When the coating thickness did not meet specifications, it exceeded the upper limit, in most cases. (In present construction, the coating thickness is higher at the lower limit and is about the same at the upper limit. Overall, the distribution of coating thickness is much narrower now. Field data indicate that increasing the thickness at the lower end is more beneficial, as far as resistant to corrosion of the bars is concerned.)
  
  
• Condition of the Steel. The segments of coated rebars were also examined to determine the condition of the steel itself Of the approximately 202 rebar segments extracted from some of the bridge decks, 81 percent did not have any corrosion present. For some of the remaining rebar segments, it was suspected that the corrosion was already present at the time of construction, since the chloride contents in the surrounding concrete were below the corrosion threshold level. Only 4 (2 percent) rebar segments were reported to have significant corrosion. Similarly, 8 (80 percent) of the 10 rebar segments extracted from the barrier and noise walls did not have any corrosion present; while 1 rebar segment was reported as having significant corrosion. In general, the corroded segments came from locations in the structures where concrete cover was relatively shallow, the chloride contents were high and/or there were cracks in the concrete. Furthermore, the corrosion found on the rebar segments were typically located underneath visible holidays in the coating or where the coating was missing completely.
  
  
• Adhesion of the Coating to the Steel. With the exception of California, Indiana, and Michigan, none of the other states had reported any significant reduction in coating adhesion on the extracted rebar segments. In California, where a total of 32 ECR segments were extracted from bridge decks that were 7 to 10 years old, reduction in coating adhesion of various extents was found on 12 (38 percent) segments, on both non-corroded and corroded areas. On these 12 segments, the loss of coating adhesion varied from 3 to 100 percent of the rebar surface—with 6 segments having adhesion loss on more than 75 percent of rebar surface. And, visible holidays were found in 11 of these segments.
  
  In the Indiana investigations, which involved 6 bridge decks and slabs that were 8 to 17 years old, none of the extracted epoxy-coated rebar segments showed any signs of loss in coating adhesion. In fact, it was difficult to strip the coatings, even with a knife, that mechanical means was used for stripping the coatings from some steel rebars to allow examination of the coating underside.
  
  Michigan evaluated 12 bridge decks that were 10 to 15 years and found that the epoxy coating on the rebar segments extracted from moist concrete in 3 experimental decks was easily removed by fingernail. However, on almost all of these rebar segments, steel corrosion was non-existent or very minor.
  
  Preliminary results from an on-going investigation in Ontario, Canada, where rebar segments were extracted from some 12 bridges and examined, indicated that coating adhesion appears to decrease with time. On 73 percent of rebars segments from bridges built between 1979 and 1980, the coating can either be pried up in small pieces or peeled off from the steel. In contrast, only 40 percent and 12 percent of the rebar segments extracted from bridges built during in the period of 1982 to 1985 and 1990, respectively, had coatings in the same adhesion conditions. In addition, using the dry-knife adhesion test, the Canadian SHRP study found that the coating on 54 percent of the rebar segments, extracted from structures that were in service for 3 to 16 years, was very well bonded. The coating on the remaining 46 percent of the segments varied from somewhat easy to remove to totally disbonded.
  
  
• Condition of the Concrete. Examination of the structures indicated that the condition of the concrete was generally good. Some concrete cracks, ranging in severity from "little or none" to extensive, were found in some of the decks. Such cracking was generally transverse in nature, and was not attributed to corrosion of epoxy-coated rebars. Some concrete delamination was detected in only 10 of the bridge decks. Approximately half of the delaminations were about 0.1 m ² (1 ft ²) in size. The rest varied from 0.3 m ²(3 ft ²) to approximately 2.8 m ² (28 ft ²).
  
  Interestingly, a 19-year old bridge deck in West Virginia with a total deck area of 1,654 m ²(17,800 ft ²), had concrete delamination of approximately 3.7 m² (40 ft ²), or 0.25 percent of the total area. The largest of these delaminations was centered on a construction joint and was likely not corrosion related. According to transportation personnel in West Virginia, it is typical to expect 5-percent to 20-percent concrete delamination in a deck of the same age and design, using black steel bars. Even more interesting was an analysis conducted by Kansas of the most recent deck ratings for 757 bridges built from 1977 to 1994 using epoxy-coated rebars. It showed that only 6 decks (0.8 percent), built mostly between 1980 and 1984, had any concrete deterioration that may be associated with corrosion of the reinforcing steel. All of the older decks were in good condition with minor or no concrete deterioration.
  
  A comparison of the performances of epoxy-coated bars among these structures suggested that better performance was obtained when the coated bars were used in both top and bottom reinforcement mats than when used only in top mat. The overall results of these investigations indicated that, in general, the epoxy coating did not perform as well when the concrete was cracked as when the concrete was not cracked. In fact, no visible or neglible corrosion was found on rebar segments extracted from uncracked concrete, even when the concrete chloride contents were up to 7.6 kg/m ³ (12.8 lb/yd ³). This is not surprising, since cracks in the concrete provide moisture and chloride an easy and direct access to the epoxy-coated rebars as opposed to the normally slow diffusion process through sound concrete; perhaps more detrimental, the cracks probably allow the coating to remain wet longer than otherwise. The investigations also suggested that the use of quality concrete and adequate cover, proper finishing and curing of concrete, and proper manufacturing and handling of ECR are important in ensuring effective corrosion protection in concrete bridge decks.

In summary, the following conclusions can be made on the performance of epoxy-coated rebars in the bridge decks surveyed:

• ECR has provided effective corrosion protection of concrete bridge decks for up to 20 years. No maintenance had yet been performed on thousands of bridge decks constructed with ECR.

• A bridge deck in West Virginia had only 0.25-percent concrete delamination after 19 years of service life. The largest delamination was centered at a construction joint and was not attributed to rebar corrosion.

• No evidence of corrosion has been found on 81 percent of the ECR segments extracted from deck cores.

• Some of the corrosion was observed on ECR segments in concrete where the chloride concentrations were below the corrosion threshold level. This corrosion was attributed to superficial corrosion that was already present on the rebars at the time of construction.



• Most of the corrosion was observed on ECR extracted from cracked concrete, where chloride concentrations were high.

• In uncracked concrete, where moisture levels were typically nominal, ECR had tolerated higher concentrations of chloride. In fact, little or no corrosion was observed in uncracked concrete with chloride concentrations as high as 7.6 kg/m ³ (12.8 lb/yd ³).

• The data from these field investigations, similar to those from previous and on-going research, have indicated that better corrosion performance is obtained when ECR was used in both mats of reinforcement than when it was used only in the top mat.

• Defects and holidays reduce the effectiveness of ECR in protecting steel bars from corrosion.

(3) Search for New Corrosion-Resistant Organic-Coated Bars

In 1993, following reports of contradictory performances of ECR in piles exposed to marine environment and in bridge decks exposed to deicing salts, FHWA initiated a 5-year research study to develop or identify corrosion resistant rebars that could provide a corrosion-free service life of 75 to 100 years (29,30,31). This study investigated 60 different rebars, including rebars with organic, inorganic, ceramic, and metallic coatings, as well as solid metallic bars.

Among these 60 different rebars were steel bars with 33 different organic coatings, of which 22 were bendable and 11 were non-bendable coatings. The bars were obtained from 15 organizations in the United States, Canada, Japan, England, and Germany. Of these, 17 coatings were utilized in conjunction with new improved steel surface cleaning processes and/or chemical treatments. In addition, since reinforcing bars are typically subjected to various severe conditions during their transportation, storage, installation, and service, the testing was designed to simulate field conditions.

In the Phase I prescreening tests, these rebars were subjected to accelerated solution-immersion screening and cathodic disbonding (CD) tests, using both straight and bent (to 4 D) specimens, each with one 6-mm hole drilled into the coating to simulate severed damage in the coating (29). The adopted test conditions are presented in Table 3. For the immersion tests, four solutions, of different compositions and pHs, were selected to produce corrosion and disbondment that the researchers believed to represent what could be expected during storage at construction sites, installation operations, and in-service behavior within mature concrete (Table 3). The specimens were immersed in these solutions at 55°C for up to 28 days. After immersion for 1, 3, 7, and 28 days the coated bars were visually examined for blisters, cracks, corrosion, and adhesion loss. The adhesion of the coatings was evaluated on both the straight and the bent sections of the bars, using the knofe-peel adhesion tests described in ASTM G1, while the bars were wet and again after 1 and 7 days of air drying.

To assess coating quality, the coating industry has been utilizing cathodic disbondment tests, which are described in AASHTO M284, ASTM A775, ASTM D3963, ASTM G8, and ASTM G42. To make the CD test in this study more severe than similar tests conducted by others, this test was conducted on 4-D bent bars (instead of typically straight bars) to introduce bending stresses on the coatings. As Table 3 indicated, the tests were conducted at a potential of -1,000 mV (versus static potential) over a period of 28 days at 23°C in a solution of 0.3N KOH + 0.05N NaOH at pH 13.3. This solution was used because it had previously been shown to produce more disbondment of organic coatings than other solutions. After 1 hr, 7 days, and 28 days of CD testing, impedance measurements were performed on the specimens; and, coating adhesion evaluation of the specimens was conducted only after the 28-day period.

In Phase II, seven of the best performing coatings from Phase I and three other new coatings (the 3M 213 bendable epoxy coating and two non-bendable coatings—an epoxy and a vinyl) were vigorously screened (30). The adhesion of these 10 coating systems on straight, 4D, 6D, and SD bent bars were tested, after solution immersion tests and cathodic disbonding tests, which were conducted under the conditions listed in Table 4.

• In the immersion test using deionized water (pH 7), any corrosion on the bars did not extend under the film, even if a coating was poorly bonded. Thus, loss of adhesion may not necessarily be a precursor to corrosion under the coating.

• Bending of coated bars induced significant strain on the coatings. This accounted for the observation of significantly better adhesion on straight bars than on bent bars, for both bendable and non-bendable coatings.

• In the immersion test, the non-bendable coatings exhibited significantly better adhesion than the bendable coatings. Furthermore, the adhesion of all coatings was significantly better when they were dry than when wet.

• In the cathodic disbondment test, the adhesion of all coatings appeared to be significantly better at locations away from the drilled hole (in the bar specimens) than at the hole. Overall, the non-bendable coatings exhibited excellent adhesion on more bars than the bendable coatings did—97 percent of the bars with the non-bendable coatings had marginal to excellent adhesion, while only 30 percent of the bars with bendable coatings had the same ratings. Furthermore, in contrast to what has been observed about coating adhesion in the immersion test, whenever adhesion is lost in the cathodic disbondment test, it was unlikely to be regained even when the coating was subsequently dried. This indicates that different adhesion deterioration mechanisms occur in each test.

• Two bendable epoxy coatings.
• Two non-bendable epoxy coatings.
• Scotchkote 213 epoxy coating (bendable).
• Post-baked epoxy coating (nonbendable).

These organic coatings—one from Phase I and five from Phase II—represent a very broad range of adhesion performances. The scotchkote 213 was also chosen since it is in most of the bridges currently in service. For control, a conventional black steel is also included. To more accurately represent real bridge structures, a new and even more severe test method that incorporates a new specimen configuration and cyclic salt ponding is being used. The following is a brief outline of this in-concrete testing of the rebars with the selected coatings:

This in-concrete testing is complete. It is anticipated that analysis of the data and preparation of a report will be completed by the end of 1998 for 75 percent of the concrete slabs. The remaining slabs will be kept for long-term observation.
So far, the following are some of the major conclusions that have been drawn on the performance of ECR in simulated adverse environments in outdoor and indoor laboratory research:

(4) Cost Impact and Potential Pay-Off from Use of Epoxy-Coated Rebars

(a) Cost Impact. For bridge engineers, cost and benefits are always important factors to consider when making materials selection. To illustrate the cost-benefit aspect of using ECR in bridge decks, consider the actual cost data for the construction of three bridge decks recently in Illinois using epoxy-coated rebars in lieu of black steel. The delivered on-site cost of uncoated black steel was $0.44/kg ($0.20/lb) and ECR was $0.62/kg ($0.28/lb). The cost associated with using the different rebars and the total costs of the decks are presented in Table 5. From this example, it is evident that the cost of decks constructed with epoxy-coated rebars averages about 1 percent more than if constructed with black steel.

(b) Potential Pay-off The potential pay-off can be estimated conservatively on the basis that the ECR deck will provide at least twice the corrosion-free service life of decks built with black steel. On the federal-aid highway system, there are about 2.3 billion square feet of deck surface. So far, ECR has been used on only 100 million square feet of deck. Based upon the average replacement cost of $40/sq. ft. for bridge deck, the potential savings are: $40/sq.ft. x 100,000,000 sq.ft. = $4 billion in 50 years, or earlier. Considering that eventually the remaining decks built with black steel will need replacement, the estimated future savings on the decks will be: $40/sq.ft. x 2,200,000,000 sq.ft. = $88 billion in that same amount of time. Since the above estimated figures are for decks on the federal-aid highway system only, the potential saving is greater if the off-system bridges are also considered in the above scenario.

At present, epoxy-coated rebar is the most common protection system used by 48 state highway agencies. To date, fusion-bonded ECR has been used as the preferred protection system in about 20,000 bridge decks, which represent roughly 95 percent of the new deck construction since the early 1980s. And, the use of ECR has extended to other structures, such as continuously reinforced concrete pavements, parking garages, nuclear power plants, coal plants, aquariums, buildings exposed to the marine environments, and wastewater treatment tanks. At present, there are approximately 100,000 structures containing ECR. Due to the success and the confidence gained by using fusion-bonded ECR over the last 20 years, there are about 35 coating plants. In addition, there are a significant number of people employed in the manufacture of epoxy powders and fabrication of the epoxy-coated reinforcement.

b. Steel Bars with Metallic Coating and Alternative Solid Metal Bars

The success enjoyed by metallic coatings on protecting steel from corrosion in other environments has raised the prospect of similar success in concrete. Metallic coatings that can be applied on reinforcing steel to provide protection against chloride-induced corrosion in concrete can be classified into two categories: sacrificial or non-sacrificial (noble). Coatings made of metal such zinc, which has more negative potentials or less noble than iron, can provide sacrificial protection to steel. When this sacrificial coating on a steel is broken, a galvanic cell or couple is formed whereby the coating is slowly sacrificed. Noble metals such as copper and nickel can also be coated on steel; however, the protection exist only as long as the coating is unbroken, since any exposed steel is anodic to the coating.

Nickel and zinc claddings began to receive attention in the late 1960s and have been shown to be capable of delaying, and in some cases preventing, the corrosion of reinforcing steel in concrete (32,33,34,35). The nickel-clad bar is produced by applying a heavy layer of nickel to a billet before hot rolling to its final form. This results into a continuous surface barrier of wrought nickel, at least 0,025-mm (0.001-in) thick, with an underlying diffusion zone of alloyed nickel and iron, which provides additional corrosion protection in the event of a break in the wrought nickel. The corrosion resistance of nickel is high in alkaline chloride solution, and even if breaks occur in the nickel coating, corrosion of steel is not appreciably accelerated, even though steel is relatively less noble than nickel (32). The results of an 11-year period testing of nickel-coated bars in a marine environment showed that the coating was effective in delaying, or sometime completely preventing, corrosion of the rebars (33). However, steel bars cladded with adequate thickness of nickel are still expensive. In addition, additional research is required to ensure their effectiveness as a corrosion protection system.

(1) Galvanized Rebars

Zinc-coated, or galvanized, bars are produced by a hot-dip process, which consists of cleaning the steel by pickling it and then immersing it in molten zinc. Galvanized bars are typically dipped in a chromate bath to passivate the zinc surface and prevent it from reacting with the hydroxide in fresh concrete. Through this process, zinc is metallurgically bonded to the steel, providing a coating (usually not less than 0.086 mm or 3.4 mils) that is composed of an outer layer of pure zinc and a number of transitional zones of zinc-rich alloys. Since galvanized bars are commercially available, it was the subject of numerous laboratory (32,36,37,38) and field studies (35,39,40). The laboratory results of the performance of galvanized bars in concrete have been conflicting. (Perhaps the conflict pointed to possible influence of experimental techniques on the results and that there is a need to develop an evaluation method that can be standardized.) For example, it was found in one study that when concrete specimens were alternately exposed to 4-percent NaCl in the stressed state, those reinforced with galvanized bars cracked less and later than those reinforced with black steel (38). In another study, data obtained when lollipop concrete specimens were partially immersed in saturated NaCl solution showed corrosion began at roughly the same time for specimens made with galvanized and with black steel, suggesting that there was no benefit from galvanizing the steel bars (36).

A subsequent 10-year FHWA comparative testing of concrete slabs made with two different water-cement ratios and containing either conventional black steel or galvanized bars, wherein corrosion rate was estimated indirectly by measuring macrocell current between top- and bottom-mat bars, indicated that galvanized bars were subject to the same type of macroscopic corrosion as black steel bars (41). Furthermore, in concrete with 0.40 w/c, both the long-term exposure data and the rate-of corrosion data indicated that use of galvanized bars did not provide extra benefit over using black. However, in concrete with 0.50 w/c, when galvanized bars were used in both mats, the corrosion rate and the corresponding metal loss were about 30 percent and 22 percent, respectively, in comparison to black steel. (However, 86-percent reduction in corrosion rate and metal loss was achieved just by using black steel in concrete with 0.40 w/c.) And, in the same concrete, when galvanized bars were used only in the top mat, the corrosion rate was twice of that observed when only black steel was used in both mats. Essentially, these results suggested that the use of galvanized bars would not provide extra benefit over black steel bars, if water-cement ratio in the concrete is kept low. The results also pointed to the possibility of galvanic reaction between galvanized bars and black steel bars, when these bars are used in the same structure.

Field studies of the performance of galvanized bars in concrete structures exposed either to deicing salts or seawater have yielded conflicting results too. For example, accelerated field studies in Michigan over a period of six years have shown that galvanized bars will retard concrete delamination and spalling but will not prevent them, especially where cover over the reinforcement is shallow (42). Meanwhile, evaluation of salt-contaminated bridges that had been in service in Bermuda for up to 20 years noted no corrosion damage (43).

To resolve the discrepancy, a panel made a critical examination of literature on the performance of galvanized rebars from: (a) studies conducted in laboratory environment, (b) studies of model specimens exposed outdoor to either natural or artificial weathering, and (c) studies of reinforced concrete structures under service conditions (44). Briefly, the panel reported that these studies had found that:

• Laboratory studies with aqueous solutions had showed that zinc tolerates a higher chloride content than steel before corrosion begins, which at least partly explains why galvanizing delays the onset of corrosion in the outdoor specimens.

• However, there is indication from the outdoor exposure studies that once corrosion of galvanized steel bars begin, the rate at which corrosion-induced distress on the concrete occurs is more rapid than for black steel.
 
• Due to a lack of standardization in the methodology (such as specimen design, quality of concrete, exposure conditions, and criteria for distress) used in the various laboratory studies, it is difficult to extrapolate the laboratory results to field conditions for predicting service life.

• It was deemed impossible to predict their service life since most of the structures were not in service sufficiently long (at the time of the study) for chloride to accumulate in the concrete to a level at which corrosion of the reinforcement would be expected.
  
  
• In the case of some marine structures (where chloride levels are high), there is the possibility that high moisture contents cause insufficient oxygen to be available at the reinforcement to sustain significant corrosion activity.
  
  
• There is evidence that rapid corrosion will occur when galvanized and black steel are (used in the same structures and are electronically) interconnected in a salt-containing environment.
  
  
• When galvanizing is preceded by chromate treatment, the chemical reaction of zinc in the fresh concrete does not impair bonding of the steel to the concrete nor the mechanical properties of the reinforcing steel. Therefore, the same design criteria used for black steel bars can be used for galvanized steel bars.

Finally, based on the literature and the collective experience of its members, the panel estimated that, for a new concrete bridge deck with a 5.1-cm (2-in.) cover of 0.45 w/c concrete and assuming that normal construction practices are used, the use of galvanized steel bars may add 5 more years to the 10 to 15 years that is typically required for corrosion-induced distress to be manifested in unprotected bridge decks.

(2) Other Cladded Bars and Alternative Solid Metal Bars

With the ultimate goal of developing corrosion resistant reinforcement that will result in a 75- to 100-year design life for concrete structures, The FHWA sponsored a 5-year research project to evaluate corrosion-resistant bars. As part of that project, 24 different bars of various types were tested with newly developed accelerated tests that are considered to be more severe than most available test procedures and much more severe than typical field conditions (45). The 24 bars included 14 ceramic-, inorganic-, and metallic-clad bars:

(2) Ceramic-clad bars (using a micro-infiltrated macro-laminated coating),
(1) Inorganic zinc silicate-clad bar,
(1) Hot-dip galvanized bar,
(1) Zinc coated bar (using the Delot process),
(3) Zinc-rich cladded bars,
(1) Nickel-clad bar,
(1) Copper-clad bar,
(1) Copper alloy-clad bar,
(1) 304 stainless steel-clad bar,
(1) Galvalum (aluminum and zinc) clad bar,
(1) Reactive copper in an organic coating bar,

and 10 different types of solid metallic bars:

(1) Black bar,
(1) Titanium bar,
(1) Type 304 stainless steel bar,
(1) Type 316 stainless steel bar,
(1) Type 317 stainless steel bar,
(1) Type 304N stainless steel bar,
(1) Type XM-19 stainless steel bar,
(1) Nitronic 33 stainless steel bar,
(1) Corrosion-resistant steel alloy bar,
(1) Type C613000 aluminum bronze bar.

Initially, the 14 cladded bars, in both straight and bent forms, were tested for 28 days (in 112 cycles of 1.25 hours dipping in specified solutions and 4.75 hours drying in air) in two solutions (a 3-percent NaCl solution, which has a pH 7, and a solution of 0.3N KOH + 0.05N NaOH + 3-percent NaCl, which has a pH of 13). Both straight and bent bars were tested in three conditions: (1) as received, (2) with a 6-mm (0.25-in) hole intentionally drilled through the cladding, and (3) with cladding purposely abraded with black slag sprayed on the bars at a specified distance and then pressured for 5 minutes. Based on the polarization resistant data, along with visual examination of the condition of these 14 different cladded bars after 28 days of testing in the two solutions, the following five cladded bars were selected for additional testing:

• 304 Stainless steel cladded bar.
• Nickel cladded bar.
• Copper cladded bar.
• Ceramic cladded bar A.
• Ceramic cladded bar B.

These five cladded bars, along with the above-mentioned 10 different solid metallic bars, were further tested for longer periods and under more severe conditions. The very corrosive solutions and severe exposure conditions used sequentially in this test are tabulated in Table 6.

Table 6. Corrosive Solutions and Exposure Conditions

The 3-percent NaCl solution was selected to provide the exposure to seawater that may be encountered by bars prior to installation into the concrete or at cracks after placement in concrete. So, after 90 days of exposure to this solution, the bars are subjected to a total of 168 days of additional exposure to the three more-severe solutions, thereby providing a total of 1,032 wet-and-dry cycles.

The measurements indicated that the polarization resistance of the black bars in the various NaCl solutions (at pH 13) averaged about 0.90, 0.51, and 0.26 ohm·m ², respectively. These correspond to corrosion current densities of approximately 29, 50, and 100 mAm ² (2.7, 4.7, and 9.3 mA/ft ²), respectively. Assuming that concrete cracking occurs after a metal loss of 0.0254 mm (0.001 in), the calculated metal losses for the current densities observed for the black bars at 56 to 168 day (d) accelerated tests indicated that the concrete could crack in 1 year or less.

Using the same assumption, the approximate time-to-cracking for concrete using the various reinforcements were predicted and shown in Table 7. And, based on review of the polarization resistance data and visual assessment for the 15 types of bars, the 10 best performing bars were determined to be: the 7 solid stainless steel bars, the stainless steel-clad bar, the aluminum bronze bar, and the titanium bar. However, after consideration of cost and other factors (including availability), only 4 bars were selected for additional testing, with 6 different epoxy-coated bars identified earlier. These 4 bars were the Type 304 stainless steel bars, the copper-clad bar, the galvanized bar, and a zinc alloy-clad bar (with a newer zinc alloy).

(4) Copper-Clad Steel Bars

Copper-clad reinforcing steel bar is another of the metallic coated bars that were recently selected for additional evaluation. This type of reinforcing bars was initially tested in concrete in 1980 and 1984. In 1980, FHWA fabricated an extensive series of large test concrete slabs that contained various types of reinforcement—including black steel bars (in concrete with and without admixed calcium nitrite), non specification epoxy-coated bars, and copper-clad bars—and had various amounts of chloride added to the top lift of concrete. Since then, these slabs have been inspected several times (48). The conditions of these slabs at different ages are shown in Table 9.

As the inspection results indicated, the copper-clad bars were far more resistant to corrosion than any of the other types of bars in the identical test concrete slabs, and providing protection even better than calcium nitrite inhibitor. The slabs containing black bars started to crack severely after 2 to 7 years, even when the concrete had calcium nitrite. The slab reinforced with an epoxy-coated bar that had uncountable holidays and less than 0.80-percent damage performed better, but finally showed symptoms of corrosion after 7 years. The slabs containing copper-clad bars in both mats and in top mat only, were still in good condition after more than 13 years of outdoor exposure. It must be noted that the average total chloride contents in the slabs containing the copper-clad bars were between 8.50 to 10.32 kg/m ³ (14.33 to 17.40 lb/yd ³), which is at least 14 times of the corrosion threshold level. Examination of cores taken from these two slabs revealed that the cladded bars, with cladding thickness of about 0.5 mm (0.02 in.), had discolored the surrounding concrete to a gray-green color. Petrographic examinations indicated that there was significantly higher amount of unhydrated cement around the copper-clad bars—extending 0.25 to 0.5 mm (0.01 to 0.02 in.) into the concrete—than at other locations. It is well known that copper, lead, and zinc salts can be retard hydration of cement. Despite the presence of unhydrated cement, the cement paste surrounding the bar appeared to be sufficiently hard in the concrete.

These findings correlated well with the results from an earlier 1984 study, which involved 48-week cyclic wetting (in 15-percent sodium chloride solution) and drying tests of concrete specimens reinforced with copper-clad bars. The copper-clad bars showed similar good corrosion resistance—during the 48-week ponding test, none of the specimens that contained the copper-clad bars exhibited signs of chloride-induced corrosion activity. Similar to the 1980 study with the same type of bars, discoloration of the concrete and retardation of cement hydration in concrete surrounding the copper-clad bars were observed at the end of the testing. In addition, the copper cladding was generally blackish in color—probably the normal copper oxide film formed after the fabrication of the specimen.

Due to the exceptional corrosion resistance shown by copper-clad bars in these studies, along with successful performance of these bars in the recent 168-day screening tests that was described earlier, it was decided that these bars should be tested further in concrete. These tests, as mentioned earlier, are under way and the results will be available in December 1998. It is clear that this type of reinforcing bar has the potential of becoming a cost-effective option for corrosion protection system, since the cost of copper-clad bars could be under $1.20/kg ($0.54/lb). However, further study on the structural effect of the retardation of cement hydration is required, prior to using such bars in bridge structures.

4. Corrosion Inhibiting Admixtures

In the past decade, a promising new approach to controlling steel corrosion in concrete structures is the incorporation of a corrosion inhibitor in the concrete mixes. Corrosion inhibitors are materials, both inorganic and organic, that are added to water or other liquids or gases in small amounts to reduce or completely stop corrosion. Inhibitors have been classified in many ways, including by composition and mechanism of action. Substances that retard corrosion by forming protective precipitates or by removing an aggressive constituent from the environment are also considered as inhibitors. The mechanism of inhibition is often complex and varies according to the type of inhibitors. For simplicity, the major inhibitor classifications are:

• Anodic Inhibitors. These inhibitors passivate the metal by forming an insoluble protective film on anodic surfaces or by adsorption on the metal. Examples of this type of inhibitors are chromates, nitrites, molybdates, alkali phosphates, silicates, and carbonates. Certain anodic inhibitors, e.g., nitrites, can cause accelerated corrosion and pitting attack if used in insufficient concentrations.
  
  
• Cathodic Inhibitors. These inhibitors are generally less effective but safer than the anodic inhibitors, and function by forming an insoluble film or adsorbed on cathodic surfaces of a metal. Examples of cathodic inhibitors are zinc, and salts of antimony, magnesium, manganese, and nickel.
  
  
• Organic Inhibitors. These inhibitors function by blocking both anodic and cathodic reactions by adsorption on the entire surface of the metal. This type of inhibitors includes amines, esters, and sulfonates.

With the widely marketed calcium nitrite, inhibition occurs only at addition dosage that is sufficiently high to counteract the effect of chlorides. In addition, this inhibitor is water-soluble and is, therefore, subject to leaching from concrete and hence would become less effective after some time. Some of the inhibitors may cause adverse effects in concrete, such as low concrete strength, erratic setting times, efflorescence, and enhanced susceptibility to alkali aggregate reaction. Finally, the real long-term benefit provided by these materials in concrete - the commercial corrosion inhibitors in particular - remains uncertain. Despite the risks and uncertainty, the use of corrosion inhibitors, both inorganic and organic, as admixtures for the corrosion protection of steel in reinforced and prestressed concrete has increased in the last fifteen years. A recent survey (49) indicated that, at present, there are four commercial corrosion inhibiting admixtures. These are:

• Darex Corrosion Inhibitor (DCI and DCI-S) (W. R. Grace). The active ingredient in this admixture is calcium nitrite, in approximately 30 percent. The typical dosage is approximately 10 to 30 liters per cubic meter (2 to 6 gallons per cubic yard) of concrete, depending on the expected extent of chloride exposure. This inhibitor provides "active" protection of reinforcing steel by facilitating the formation of a passive oxide film on the steel surface, through the following reaction:
  
  

  2 Fe ++ + 2 0H - + 2 NO 2 - 2 NO + Fe 2O 3 + H 2O

  (14)

  
  
• Rheocrete 222 ⁺ (Master Builder). The active ingredient of this corrosion inhibiting admixture is a water based combination of amines and esters. According to its manufacturer, this organic mixture forms a protective film, by chelation process, on the reinforcing steel to serve as a barrier for actively inhibiting both the anodic and cathodic reactions. In addition, this mixture also lines the pores in the concrete with hydrophobic chemical compounds that provides a screen to "passively" reduce the ingress of chloride ions into the concrete (50). The recommended dosage is 5 l/m ³(1 gal/yd ³) of concrete, irrespective of the severity of the exposure environment.
  
  
• Ferrogard 901 (Sika). This is a water-based blend of surfactants and an amino alcohol (or alkanolamine), specifically dimethyl ethanolamine (DMEA). Using x-ray photoelectron spectroscopy (XPS) and secondary ion mass spectroscopy (SIMS), its manufacturer showed that the amino alcohol absorb on the steel bars by forming bond between the its amino functional group and the hydroxide group on the steel surface. This leads to the formation of insoluble iron oxide complexes, which stabilize the oxide surface and inhibit further corrosion. The recommended dosage is 10 l/m ³(2 gal/yd ³) of concrete, irrespective of the severity of the corrosive environment, irrespective of the severity of the exposure environment.
  
  
• Catexol 1000 CI (Axim Concrete Technologies). It appears that this corrosion inhibiting admixture combines both an organic film-forming inhibitor, specifically an amine derivative, and a nitrite-based inhibitor. (No specific data is available for this admixture.)
  

The bulk of the performance data for these corrosion inhibiting admixtures are obtained mostly from laboratory tests conducted by their manufacturers - using simulated environment and/or small concrete specimens; the rest come from investigations conducted by FHWA, some of the state transportation agencies, and the academia. Most of these data pertain to DCI, because this commercial inhibitor was introduced in 1978 while the others were introduced in the last five years. Due to the long time required to assess their performance and the lack of follow-up studies once a structure is built, field performance data on any of these admixtures are still very limited. Therefore, it is not yet possible to accurately define the actual benefits that can be expected from the use of a specific corrosion inhibiting admixture other than to state that delay of corrosion initiation is anticipated.

Investigations of the use of corrosion inhibitors in concrete started in the 1960s and included sodium nitrite, and the sodium and potassium salts of chromate and benzoate. It was found that the sodium and potassium salts of chromate and benzoate gave mixed results on inhibition and reduced concrete strength. In one early study, it was shown that nitrite and chromate were effective inhibitors (51). A subsequent study indicated that only sodium nitrite was effective but it exhibited deleterious effects on the strength of the concrete (52). In the late 1970s, calcium nitrite was introduced commercially as an alternate form of nitrite (53).

In 1980, the FHWA began an outdoor exposure study to assess the effectiveness of calcium nitrite as an admixture for inhibiting steel corrosion in concrete (54). Eighteen reinforced concrete slabs (0.6 m x 1.52 m x 0.15 m) were built to simulate bridge decks using bared black steel bars. Of these slabs, 13 slabs contained various amount of admixed chloride and 2.75-percent calcium nitrite (by weight of cement) in the top lift and 5 control slabs without admixed nitrite. These slabs were tested by periodic measurements of macro-cell corrosion current, half cell potential, driving voltage (i.e., potential difference between the top- and bottom mat reinforcing steel), and concrete electrical resistivity (i.e., resistance between the top and bottom mats), as well as visual inspection. Based on data at the time, it was concluded that:

• Calcium nitrite inhibitor appeared to function by not allowing a large electrical potential difference to develop between the bars, thereby preventing any voltage difference that may drive corrosion, and keeping the electrical potential of all steel bars at or near the passive range.
  
  
• Calcium nitrite inhibitor was effective in reducing the rate of corrosion of black steel bars, up to a chloride-to-nitrite ratio, [Cl ^-] /[NO ^-] , of 1.79. At the ratio equal to or less than 1.25, 10 or more years would be required to consume the same of iron as that consumed in 1 year in a concrete with equal amount of chloride.

However, after testing of the same 18 concrete slabs were continued for 7 years, it was found that calcium nitrite was effective in reducing the rate of corrosion of black steel up to a [Cl ^-] /[NO ^-] of only 0.90, instead of the higher [Cl ^-] /[NO ^-] of 1.79 indicated earlier (55). In fact, the slabs fabricated with higher ratios, even up to only 1.11, showed some cracking and rust spots on their surfaces after 7 years, despite exhibiting a reduction in corrosion current by a factor of 10. Since the chloride and the nitrite were added at the same time when the test slabs were being fabricated, it is reasonable to assume that a [Cl ^-] / [NO ^-] higher than 1 may still be able to suppress steel corrosion, if the chloride ions penetrates instead naturally into the concrete as in a structure. However, the researcher still recommended that, if selected as the protection system in a new construction, sufficient quantity of calcium nitrite should be added to the fresh concrete mix so that the [Cl ^-] / [NO ^-] will not exceed 1.0 at the level of steel—throughout the expected design life of a structure.

Recently, several new inhibitors and the first three commercial inhibitor admixtures were examined in a preliminary study of inhibitors conducted in Virginia (56). Using electrochemical impedance spectroscopy and visual observation, the investigators measured and ranked the overall inhibition performance of these inhibitors in simulated pore solution. It was determined that the top three inhibitors were barium metaborate, disodium -glycerophosphate, and one of the commercial inhibitors. A follow-up study is under way to allow: (a) additional testing of these inhibitors in concrete to determine if their use has any adverse effects on setting and other properties of concrete, (b) development of an unique method of inhibitor delivery for alleviating such adverse effects, if there is any, and (c) expansion of the search to other new inhibitors (57). Another study is under way in Florida to (a) evaluate the ability of commercial inhibitors to stay in the concrete, which would influence their effectiveness, (b) estimate the long-term effectiveness of inhibitors, (c) determine possible adverse side effects on the corrosion process and the concrete properties, (d) quantitatively assess extension of concrete durability, and (e) establish their suitability for applications in new construction and in rehabilitation (58).

Recently, the commercial inhibitors—particularly the first three products—have been promoted for use in repair of corrosion-damaged concrete in bridges, either by: (a) addition as admixture to patching concrete, (b) spray application onto affected areas, or (c) saturation treatment to affected areas. Unfortunately, data on the effectiveness of this approach of using inhibitors are relatively limited. Therefore, an on-going study in Virginia is attempting to develop some data from actual bridge rehabilitation projects and concrete slabs fabricated to simulate bridge situations for assessing the effectiveness of these commercial inhibitors when used in rehabilitation (59).

Due to lack of widely accepted standard evaluation test that can truly be applied to all types of inhibitors, regardless of their inhibition mechanisms, comparison of the effectiveness of different inhibiting admixtures present difficult challenges. It's likely that a system of different acceptable tests would have to be developed. Furthermore, the effectiveness of all inhibitors should be evaluated based on per unit weight of the active ingredient(s) - the commercial corrosion inhibiting admixtures in particular. Unfortunately, most of the manufacturers are reluctant to share information on the compositions, or even just the concentrations of the active ingredients, of their products with the users.

In summary, corrosion inhibitors are increasingly being employed as a part of multiple corrosion protection systems in conjunction with epoxy-coated rebars and low-permeability concrete. As yet, epoxy-coated seven-wire strands are not usually employed for prestressed concrete bridge members. In lieu of coated seven-wire strands, corrosion inhibitors have found their niche in the prestressed highway construction industry. In addition, corrosion inhibitors are finding use in cementitious grouts for filling the post-tensioning ducts of bridge members or the sheating of cable stays to protect the highly tensioned and uncoated black steel. Due to the premature failure of epoxy-coated rebars in the splash zone of piles on the Florida Keys bridges, some state highway agencies are relying on corrosion inhibitors as one of the alternative corrosion protection systems for marine application. For bridge deck exposed to excessively aggressive environment, the use of a corrosion inhibitor, in conjunction with ECR, as a redundant corrosion protection system, can enhance the service life of a structure with respect to corrosion. (Industry-sponsored research is under way on the combined use of ECR and corrosion inhibitors as a dual corrosion-protection system for bridge decks.) In addition, the corrosion inhibitors can play an important role in protecting uncoated high-strength steel in PS/C bridge members and cable-stayed bridges. Research is continuing to identify other corrosion inhibitors that will provide long lasting corrosion protection to bridge members.

5. Corrosion Protection of Prestressed Concrete Bridge Members

In the fall of 1992, the United Kingdom Ministry of Transportation imposed a temporary ban on the commissioning of grouted, bonded post-tensioned bridges. This resulted from the collapse of two footbridges in 1960, the collapse of a single-span, segmental post-tensioned bridge in Wales in 1985, and an examination of nine other segmental bridges. United Kingdom is not the only place with problems of voided ducts and insufficient coverage over prestressing steel in ducts, as evident by the collapse in 1992 of the post-tensioned Melle Bridge across the Schelde in Belgium, which was constructed in 1956. This last failure was traced to the corrosion of the ducted post-tensioning wires even though the bridge had been inspected, load tested, and rated satisfactorily. The UK moratorium was lifted in 1996 with the publication of advisory report "Post tensioned Concrete Bridges: Planning Organization and Methods for Carrying Out Special Inspections" by the construction industry and owners of this type of bridges.

The underlying difficulty is that there are no reliable, cost-effective, and rapid nondestructive methods for providing assurance to the owners that the built structures have met the construction specification. One of the major inspection concern is determining whether the ducts in the post-tensioned bridge members have been completely filled with the grouts and that there is uniform coverage over the prestressing steel. It has been found many times that, invariably, the ducts had big voided sections and were filled only partially. In addition, it is very difficult to assess the condition of anchorage areas.

Chloride-bearing water can find its way through anchorage into the ducts and eventually initiate corrosion at the anchorage and subsequently of the prestressing steel inside the ducts, especially if the ducts have voids and are partially filled. The salty water can also access the ducts through faulty and leaky joints. In time, the chloride ions also penetrate through concrete cover and corrode the prestressing steel in the ducts, either after the corrosion of the metallic ducts or through defective plastic ducts. Beside causing pitting of the prestressed strands, corrosion reactions lead to the evolution of atomic hydrogen, which is subsequently absorbed into the steel leading to hydrogen embrittlement of the steel. This embrittlement can cause the strands to fail much more readily under normal loads. Since prestressed concrete bridge members rely on the tensile strength of the strands to resist loads, loss of even few strands per member can prove catastrophic. In addition, due to the high stresses the strands are subjected to, corrosion effects are accelerated. Even though corrosion initiate on prestressed steel in the same manner as on mild reinforcing steel, there is one significant difference between the two: since prestressed wires have smaller cross sections than rebars, sections of prestressed wires are more readily broken or completely rusted away.

Corrosion protection methods adopted at present in the construction of prestressed concrete members included: (1) the use of highly impermeable concrete through the use of fly ash and silica fume and controlled curing at the fabrication site, and (2) the use of corrosion inhibitor admixtures. The use of epoxy-coated steel is not yet common in prestressed concrete members and research on this application of coated strands is needed.

The main thrust of FHWA field and laboratory investigations in this area of prestressed concrete consisted of the following:

• Determination of the condition of oldest prestressed concrete bridges located in adverse environments. This is to ascertain whether the same problems observed in Europe exist in the United States.
• Develop methodology for inspection of grout coverage over prestressing steel in the ducts of post-tensioned bridge members.
• Develop corrosion resistant grouts and methods to test them in the laboratory before their use in the field.

a. Field Investigations

Many problems associated with deterioration of prestressed concrete members can be traced to improper designs in a structure. To help identify some common examples of improper design, a total of 17 prestressed concrete bridges were surveyed between 1988 and 1992 and the collected data were analyzed (60,61). Built between 1954 to 1979, eleven of these bridges were pre-tensioned and six were post-tensioned—with some of the bridges containing both types of members. These bridges were located both in northern climates exposed to application of roadway deicing salts and in southern climates exposed to seawater spray. The procedures used in the examination included: visual examination, concrete delamination survey, chloride content survey, concrete cover survey, half-cell potential survey, rapid chloride permeability, and petrographic and metallurgical analysis of the prestressed steel.

The detailed examinations of these bridges revealed that, since most common sources of chlorides in concrete in the substructures are water running from bridge decks and spray from seawater, the corrosion of prestressed steel observed in these bridges was due commonly to one or more of the following causes:

• Water-permeable deck overlay.
• Leaking expansion joints.
• Poorly designed and malfunctioning drains.
• Inadequate concrete cover over the prestressed steel.
• Improper protection of the anchorage systems provided by mortar.
• Access of water and chloride into the sheathing through the anchorages.
• Poor grouting.
• Lack of grease in unbonded tendons.
• Exposure to direct action of seawater spray.
• High chloride permeability of the concrete, mortar, and grout.

Overall, the conditions of the bridges surveyed were judged to be adequate. And, to add to the durability of these types of concrete structures located in adverse environments, the following recommendations were made for:

• Pretensioned Box Beams.
  • Installation of protective membrane systems on the top flange of the beams.
  • Installation of a rigid wearing course or composite deck slab (provided the structure could accommodate the additional dead load).
  • Addition of stiffeners to prevent the movement of the beam faces.
  • Transverse post-tensioning of the beams to hold beam faces together.
  • Design of proper slope and drainage details to minimize the dwell time of deicer solutions on the deck.
  
  
• Pretensioned I-Beams.
  • Minimizing exposure of beam ends to deicing salts runoff from the deck. This can be done by running beams through diaphragms over piers at the center of the bridge (allowing beams to be made continuous) or by running beam ends into concrete abutments at the bridge ends.
  • Install and maintain a functional gutter and drainage system using large-capacity gutters and wide drainage pipes to carry water off the deck quickly.
  • Install rigid concrete overlay to prevent leakage of water through deck on to the superstructure.
  • Reduce or eliminate joints in the new construction and replace deteriorating deck joints.
  
  
• Post-tensioned Beams.
  • Use of corrosion-resistant grouts containing corrosion inhibitors, in addition to low w/c with silica fume and fly ash.
  • Use of plastic ducts with adequate thickness to eliminate puncturing by prestressing steel during pulling and grouting.

Finally, the researchers emphasized that routine inspection and proper maintenance are extremely important in eliminating major deck-to-steel water routes for prestressed substructure concrete elements.

b. Laboratory Investigation

(1) Corrosion-Resistant Grouts

Grout is the final line of defense against corrosion of steel tendons. Based on review of state highway agencies' specifications on grouts, it was determined that the design mixtures were inadequate and that there was no suitable method for evaluating the corrosion performance of grouts before their use in the field. Hence a study was initiated with the following objectives: (1) to develop new mixture designs, and (2) to develop and perform accelerated test methods on the new grouts. In this study, in addition to varying w/c, several modifiers and additives for grouts were examined, including high-range water reducers, fly ash, silica fume, latex-polymer modifier, expansive agents, anti-bleed additives, and corrosion inhibitors (62,63,64,65,66).

A large number of grout design mixtures were tested with an accelerated corrosion test method (ACTM). ACTM uses a test specimen that simulates a post-tensioning strand embedded in grout, but with a section of duct missing (representing a break in the duct). Other features of the test included minimal cover on the tendons, a 5-percent sodium chloride solution in direct contact with the grout, and forcing of chloride ions through the grout by an applied voltage. The study resulted in the following conclusions:

• Reduction in w/c had a marginal beneficial effect.
• Silica fume combined with low w/c had a significant beneficial effect. A 10-percent silica fume provided a two-fold increase in time-to-corrosion initiation.
• 33 percent (by weight of cement) flyash provided a three-fold increase in time-to-corrosion initiation.
• Latex polymer modifier and anti-bleeding admixtures reduced pressure-induced bleeding.

In summary, use of fly ash or silica fume, in conjunction with proper dosage of a corrosion inhibitor with a suitable anti-bleeding agent, can provide a grout which can protect the prestressing steel even if it is contained in defective ducts. Based on these recommendations, the Post-Tensioning Institute has written a grouting specification, which will be published in 1998.

(2) Inspection of Grout Cover Over Prestressing Steel and Voids in the Ducts

Prior to the discovery of problems with grouting of ducts in existing prestress sing concrete structures in UK and other parts of the world, little attention was given to this subject. Review of literature offered the following relevant techniques:

• Drilling and monitoring air flow at suspected locations. The disadvantages with this method are that it is a destructive method and there is always a slight chance of damaging or cutting the prestressing steel. In addition, the drilled hole, even after filling, becomes an avenue for future corrosion.



• Boroscope. This technique allows viewing of the inside of a duct by insertion of a fiber optic cable through a 12.5-mm (0.5-in) diameter hole drilled through the duct at the location of suspected void. A camera is typically used in conjunction with the optic fiber to record images of the ducts. This has the same disadvantages of the air-flow technique.
  
  
• Radiography. The French use a radiography system (Scorpion), which utilizes a portable linear accelerator to provide the ionizing rays, to inspect duct systems in post-tensioned concrete bridge members. With this system, they can inspect the grout cover and also find voids in the ducts containing prestressing steel.
  

Neither of these techniques are widely used in the United States, if at all, because the first two are destructive and the third technique, even though it is nondestructive, involves a bulky equipment and is expensive. Hence, FHWA initiated a research study to develop an equipment which is in the range of $15,000 to $25,000 and also not bulky. After completion of the first phase of this study, the impact-echo technique was selected for further evaluation and development, since it showed the most promise (66,67).

In this method, a short-duration stress pulse is sent through the duct location under inspection by striking the outer concrete surface. The reflected pulses are received by a transducer (which is in close proximity to the impactor) and is displayed on a computer screen as a frequency spectrum. The frequencies of the pulses reflected from a voided and/or partially filled duct are much lower than those reflected from a filled duct. This technique has been successfully used on a few bridges. The concrete cores taken from the suspected locations showing lower frequency had either a void in the duct or honeycombing around duct area. The correlation between the nondestructive impact-echo technique and the destructive examination of concrete cores was excellent.

B. Corrosion Control for Rehabilitation of Existing Concrete Structures

There are different remedial methods that can be applied, as part of rehabilitation of existing concrete structures, for controlling chloride-induced reinforcing steel corrosion. Each of the methods, which are used after all damaged concrete in a structure have been removed and patched, functions by one of the following principles:

• Providing a barrier on the surface of the existing concrete to prevent fixture ingress of chloride. The conventional rehabilitation methods belong in this category.
• Controlling the electron flow within the reinforced concrete environment to halt metal loss. Cathodic protection functions in this manner.
• Modifying the concrete environment to make it less corrosive. This is how electrochemical chloride extraction can extend the service life of chloride-contaminated concrete structures.

1. Conventional Rehabilitation Methods

Methods that function by the first approach—such as application of an overlay of polymer concrete, low slump concrete, latex modified concrete, or high density concrete, on the existing concrete—provide a barrier that prevents continued intrusion of additional harmful chloride ions, moisture, and oxygen that are needed to sustain corrosion. In some cases, these methods have successfully extended the useful life of a structure. However, past experiences have also indicated that when such measures were used, without first decontaminating the existing concrete of the necessary corrosion ingredients, it was often that these substances become entrapped (in the old concrete) by the overlays in sufficient amounts to initiate new corrosion. Furthermore, the ensuing corrosion is often worse than before rehabilitation, due to arising electrochemical incompatibility between the new patching concrete and the surrounding old but physically sound concrete (68). The following is a brief summary of work performed on measures belonging to approach one.

FHWA undertook a long-term staff study in the 1980s to evaluate the effect of various conventional repair systems on the rate of corrosion of reinforcing steel (69). It had been debated whether the overlaying of a bridge deck with a relatively impermeable concrete or membrane would sufficiently arrest the corrosion to effect a permanent repair. The following overlays were placed on chloride-contaminated reinforced concrete slabs and tested:

• 50.8-mm (2-in) thick concrete overlay with 0.32 wlc.
• 50.8-mm (2-in) thick concrete overlay with 0.45 wlc.
• 31.7-mm (1.25-in) thick latex-modified (styrene-butadiene) concrete overlay, uncracked and badly cracked, with 0.40 wlc.
• 50.8-mm (2-in) thick internally-sealed concrete overlay with 0.50 w/c.
• 50.8-mm (2-in) AC overlay over preformed waterproof membrane.
• Silane surface treatment.
• Nominal 13-mm (0.5-in) thick built-up polymer concrete overlay (two with vinylesters, one with MMA, and one with polyester resins).
• Nominal 10-mm (0.4-in) thick acrylic (latex mortar with glass fabric overlay).
• Delamination patching only (as necessary; very little patching with salt-admixed concrete was actually done).

The concrete specimens were 1.2-m-wide x 1.5-m-long x 0.152-m-thick (4-ft. x 5-ft. x 0.5-ft.) slabs that were fabricated in 1971 using a top-mat reinforcing steel only. These slabs were subjected to daily salting for 8 years. All of the slabs in this phase of the study were fabricated using concrete with a 0.5 w/c and 25.4 mm (1.0 in) of clear concrete cover over the reinforcing steel, and exhibited some corrosion-induced concrete distress. Measured electrical potentials indicated the presence of corrosion on the rebars in all slabs. In order to make the slabs more closely simulate a bridge deck, a 63.5-mm (2.5-in) layer of chloride-free concrete containing a mat of reinforcing steel was placed on the bottom of each slab. Prior to concrete placement, the old bottom of each slab was sandblasted to exposed coarse aggregate. Care was exercised to insure the new bottom mat of steel was not electrically continuous to the top