Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Public Roads
Featuring developments in Federal highway policies, programs, and research and technology.
This magazine is an archived publication and may contain dated technical, contact, and link information.
Federal Highway Administration > Publications > Public Roads > Vol. 68 · No. 2 > Preventing Corrosion in Steel Bridges

Sept/Oct 2004
Vol. 68 · No. 2

Preventing Corrosion in Steel Bridges

by Shuang-Ling Chong

A painter applies a protective coating to a bridge to protect it from corrosion.

(Above) A painter applies a protective coating to a bridge to protect it from corrosion.

FHWA researchers evaluate the accuracy and reliability of three chloride test kits to determine their performance and accuracy.

Each year the Federal Government and State departments of transportation (DOTs) spend billions of dollars on bridge rehabilitation and maintenance due to corrosion. On bridges, corrosion is most often caused when steel is exposed to atmospheric conditions, such as salt, moisture, and oxygen. To prevent corrosion on bridges, transportation agencies apply a protective coating to the steel.

But according to Dr. Bernard Appleman, a consultant at KTA-Tator, Inc. and former executive director of the Society for Protective Coatings, if the steel has a corrosive agent on it before painting, the protective coating may fail prematurely. "Soluble salts, especially chloride salts that are not removed before painting, are a major source of early and often catastrophic paint failure," says Appleman. If the paint fails prematurely, the resultant corrosion will eventually compromise the structural integrity of the metal. "Ultimately, this paint failure can require extensive bridge maintenance, which is not only costly but also an inconvenience to the driving public," he adds. Therefore, before the bridge painter applies the protective coating to either new steel or a rehabilitated bridge, the surface needs to be evaluated for cleanliness.

Presently, painting specifications almost all rely on visual (or qualitative) measurements to determine readiness for applying protective coatings. However, researchers at the Federal Highway Administration (FHWA) are looking for a more accurate, quantitative measurement that can be used again and again to determine if corrosive elements are on steel prior to applying a coating. One such method may be to test for chloride.

To help bridge coating inspectors better assess the condition of steel prior to painting, FHWA recently evaluated three commercially available chloride test kits that are used to determine the cleanliness of steel surfaces. The objectives were to assess the accuracy and precision of the tests and to identify the factors that influence their performances.

From Visual to Quantitative Assessments

Because contaminants can affect the performance of bridge coatings, inspectors need accurate techniques to assess the cleanliness of the steel surfaces prior to painting. Most methods used today, however, are qualitative or semiqualitative at best.

"All of the cleaning standards today are visual," says Bob Kogler, team leader for bridge design and construction research at FHWA. According to Kogler, assessing steel cleanliness using visual standards can lead to disputes. "An inspector may look at the steel and see indications that it is not clean enough, while the contractor may argue that it is clean enough," he says. "To some degree, even though we have standards, it is almost a matter of opinion because the standards themselves are qualitative."

In 2001, the FHWA Nondestructive Evaluation Validation Center completed a study that evaluated the accuracy of the visual inspection method for determining the condition of bridges. The study showed that inspectors vary considerably in how they complete routine inspections. In particular, they vary in how they assign condition ratings.

"Eventually we need to make our evaluation of steel surface cleanliness a quantitative measure, because it would clear up a big area of disputes on bridge painting jobs," Kogler says. "The measurements [derived from the testing kits] will tell us how chemically clean [the surface] is, not just how clean it looks. And that will give us a much better measure of the potential performance of the paint."

Some applications, such as those in the marine industry, already are moving toward quantitative methods to assess chloride concentrations on steel surfaces.

The Problem with Chloride

Because chloride is the primary surface contaminant and is usually the most corrosive agent to steel, inspectors may be able to test for it before painting steel surfaces. High concentrations of chloride can cause early coating failures, such as rust and delamination, a process in which the coating begins to separate from the steel. Ultimately, the rust and coating delamination can destroy the structural integrity of the metal. Chloride is of particular concern for structures that are salted during deicing operations or are located in a marine environment, where the concentration of chloride salts can be high in seawater and spray.

After the steel surface is blasted clean with abrasives or cleaned with high-pressure water, and before a coating is applied, the inspector should assess or test the steel surface for chloride. If the visual inspection or testing indicates high chloride concentrations, the metal must be cleaned again and retested.

Three Chloride Test Kits Evaluated

Currently where specified, coating inspectors use one of three commercial test kits to evaluate chloride levels quantitatively. Generically, the kits are the swab test, the patch test, and the sleeve test.

All three tests use a liquid, either acidic fluid or de-ionized water, to dissolve or extract chlorides on the surface of the steel into a solution. The inspector then tests the solution for chloride concentrations. The swab test relies on wet cotton balls to extract the chloride from the surface of the steel. The patch test uses a syringe containing extraction fluid to draw chloride from the patch test area. And the sleeve test extracts the chloride in a fluid-containing sleeve that is attached to the steel.

According to State DOTs and bridge inspectors, all three tests have shown inconsistent and highly variable results. These inconsistencies may be due to different extraction efficiencies and detection sensitivities in the tests, as well as operator variability.

Therefore, FHWA researchers investigated the variability and limitations of the test methods to establish techniques that may be used to obtain reliable and accurate chloride concentration test results.

The steel has rust shown on the underside of this bridge. If the bridge is not rehabilitated, the corrosion eventually will compromise its structural integrity.

If the steel surface of a bridge is not cleaned adequately before painting, the protective coating can fail prematurely. The steel then will develop corrosion, such as the rust shown on the underside of this bridge. If the bridge is not rehabilitated, the corrosion eventually will compromise its structural integrity.

Experimental Procedures

The researchers analyzed steel panels in a vertical position. Four different levels of chloride concentration, ranging from 3 to 30 micrograms per centimeter squared (mg/cm2), were applied to the panels to determine if the chloride concentration affected the validity of the results. An industry rule of thumb is that after blasting, a bridge should be painted within 4 hours. Therefore, the researchers performed tests under three conditions that fell within this timeframe: within 1 minute after panels were doped (that is, artificially contaminated with chloride), after aging doped panels at high heat and moderate humidity for 4 hours, and after aging doped panels at high heat and high humidity for 4 hours.

The detectors for the swab, patch, and sleeve tests are an ion detection strip, four bottles of titration liquids, and an ion detection tube, respectively. Because the patch test can use two different fluids, acidic fluid or de-ionized water, the researchers conducted additional tests to determine which fluid recovered the most chloride. Since the researchers found that acidic fluid extracted more chloride than de-ionized water, acidic fluid was used in the patch test.

In all, the researchers evaluated each kit under 12 different conditions (4 chloride concentrations and 3 aging conditions) to determine how chloride concentrations and aging affect the accuracy of the test. Each test was performed three times by three different operators at the Paint and Corrosion Laboratory at FHWA's Turner-Fairbank Highway Research Center (TFHRC) in McLean, VA.

Three Chloride Test Methods

Following are detailed descriptions of how the researchers conducted tests using the three kits.

Swab Test
The researcher extracted chloride from a surface area of 150 cm2 (23.25 in2)-a 15-cm-long by 10-cm-wide steel panel-using 15 milliliters, ml (0.51 fluid ounce, fl oz) of de-ionized water for cotton swabbing. To reduce dripping, only one-third of each of the four cotton balls used for swabbing was soaked with de-ionized water. The researcher then absorbed the remaining liquid with an additional cotton ball and used an ion detection strip to measure the concentration of chloride in the extracted solution.

Patch Test
The researcher glued a patch securely onto the steel surface, covering an area of 12.25 cm2 (1.89 in2). The researcher then injected 1.5 ml (0.05 fl oz) of extraction fluid into the patch, then extracted two-thirds of the fluid from the patch and reinjected it to mix the fluid more thoroughly. The researcher then rubbed the patch with a finger for 1 minute to promote chloride solubility. Next the patch was rinsed with an additional 1.5 ml (0.05 fl oz) of extraction fluid. Finally, the researcher combined the two extractions and titrated the resulting 3 ml (0.1 fl oz) of extract with reagents included in the kit.

Sleeve Test
The researcher poured 10 ml (0.34 fl oz) of extraction fluid into a sleeve and then attached the sleeve firmly to the steel panel. The researcher then lifted the free end of the test sleeve and held it upright with one hand to allow the extraction fluid to make contact with the test surface. With the other hand, the researcher massaged the solution through the test sleeve against the steel surface for 2 minutes. The researcher then removed the test sleeve and used an ion detection tube to test the solution for chloride concentration.



A technician uses highpressure water to clean a section of steel.

A technician uses high-pressure water to clean a section of steel.

Comparing the Results

The researchers identified strengths and shortcomings for each test. The research team acknowledges that since these tests were conducted in a laboratory, where procedures were carefully controlled, the experimental results may be better than would be expected in the field.

Of the three tests, the swab test recovered the highest amount of chloride and also offered the most reproducible data. For freshly doped steel (that is, a specific amount of chloride applied to the steel for testing), the swab test recovered approximately 70 to 100 percent of the chloride. This test also had the least variability.

The swab test uses an ion detection strip, which can detect chloride concentration only above 30 parts per million (ppm). The large extraction area, however, compensates for the detection limit. The researchers found that the lowest level of chloride detection possible under the laboratory conditions employed was 3 mg/cm2.

One shortcoming of the swab test is that it must be conducted very carefully, which may be challenging in the field where a test operator or inspector may be high on a ladder testing steel overhead. Because it is conducted in an open environment, water can drip and evaporate easily, which will result in reduced chloride recovery and therefore imprecise results.

Unlike the swab test, the patch test is a closed extraction system, which prevents fluid evaporation and loss. However, the operator may still lose fluid if either the patch is not adhered to the steel surface firmly, or if the syringe, which is used to extract fluid, is improperly inserted into the patch. In either case, the loss of even a small amount of extraction fluid will result in inaccurate chloride measurements.

The patch test, with titration liquids used as a detector, also provides high chloride recovery. But the results were found to be the most unreliable of the three tests, as indicated by a higher margin of error. One potential cause for the error is the variability in drops needed to reach the titration end point (that is, color change). If the color change falls between two drops, some operators will use an extra drop while others will use one drop less. The number of drops used may vary by operator, or the same operator may use a different number of drops for each test conducted. This variation in the number of drops will affect chloride concentrations.

An additional shortcoming of the patch test is that it only indicates minimum and maximum values rather than actual values. However, a coating inspector could be conservative and use the maximum value to determine whether to proceed with a painting job.

A final shortcoming of the patch test is that the acidic fluid requires mercury nitrate as one of the titrants. Because mercury is a hazardous waste, the operators or inspectors must follow strict guidelines when disposing of the fluid.

The sleeve test, like the patch test, is a closed system with little risk of fluid loss, but extraction fluid can be lost if the sleeve is not adhered to the steel surface firmly. If fluid is lost, the test will generate unreliable chloride measurements.

The sleeve test was more effective at recovering chloride at higher concentrations than at lower concentrations. The low rate of chloride recovery at lower concentrations may be due to the low sensitivity and unclear color separation at the low reading end of the ion detection tube. The sensitivity can be increased if the extraction volume is reduced or extracted area is increased. The sleeve test had a margin of error that fell between that of the swab and patch tests.

For the patch test, a researcher injects chloride extraction fluid into a patch on a vertical steel plate.

For the patch test, a researcher injects chloride extraction fluid into a patch on a vertical steel plate.

For the swab test, a researcher uses a cotton ball and deionized water to detect and extract chloride from the steel plate

For the swab test, a researcher uses a cotton ball and deionized water to detect and extract chloride from the steel plate.



Parameters of the Three Chloride Extraction Test Methods
Swab Test Patch Test Sleeve Test
System type Open Closed Closed
Detection method Ion detection strip Titration Ion detection tube
Area, cm2 150 12.25 10
Extraction fluid, ml 15 3 10
Area (cm2)/volume (ml) 10 4.1 1
pH value of extraction fluid De-ionized water, pH = 6.7 Acidic, pH = 3.9 Acidic, pH = 4.2
The table shows the system type, detection method, area of extraction, extraction fluid volume, ratio of area to fluid volume, and pH value of the extraction fluid for each of the three test kits. Source: FHWA.

Humidity Affects Test Results

An important finding of the research is that heat and humidity will affect test results. When the doped panels were aged at a high temperature- 37 degrees Celsius (98.6 degrees Fahrenheit)-for 4 hours under two different humidity conditions, the chloride recovery was less than that of freshly doped panels. However, the researchers noted a considerable difference between moderate and high humidity. At 57 percent relative humidity, the chloride recovery was reduced only slightly. But for all three tests, chloride recovery decreased considerably at 78 percent relative humidity.

The bar graph compares the chloride recovery by the swab, patch, and sleeve tests at four different chloride levels: 3, 5, 9 and 30 micrograms per centimeter squared (mg/cm<sup>2</sup>). For all tests, the chloride recovery decreased with decreasing chloride concentrations; it decreased from 100 to 70 percent, 120 to 85 percent, and 100 to 25 percent for the swab test, patch test, and sleeve test, respectively. The graph also shows the error bar for each of the tests, indicating how reproducible the results are. A low standard error indicates high reproducible results, while a high standard error indicates low reproducible results. All three tests showed some standard error in the tests. The standard error was lowest for the swab test and highest for the patch test.

The bar graph compares the chloride recovery by the swab, patch, and sleeve tests at four different chloride levels: 3, 5, 9 and 30 micrograms per centimeter squared (mg/cm2). For all tests, the chloride recovery decreased with decreasing chloride concentrations; it decreased from 100 to 70 percent, 120 to 85 percent, and 100 to 25 percent for the swab test, patch test, and sleeve test, respectively. The graph also shows the error bar for each of the tests, indicating how reproducible the results are. A low standard error indicates high reproducible results, while a high standard error indicates low reproducible results. All three tests showed some standard error in the tests. The standard error was lowest for the swab test and highest for the patch test. Source: FHWA.

The researchers speculate that low levels of rust formed after the steel was exposed to high levels of heat and humidity for 4 hours. The invisible rust entrapped the chloride, making it more difficult to extract.

These results suggest that inspectors should determine the surface chloride concentration as soon as possible after blasting. Any delay, especially in hot and humid environments, may result in erroneously low chloride values. If these low chloride values fall within the acceptable limits for the protective coating, an inspector may decide to apply the coating to a steel surface that in reality has unacceptable levels of chloride trapped under invisible rust.

Using the Research Findings

Although additional research is warranted, the findings from the FHWA study may provide valuable information that coating inspectors can apply in the field today.

Summary of Test Results
Swab Test Patch Test Sleeve Test
Chloride concentration High (10)a Medium (4.1) Low (1)
Loss of extraction fluid Yes No, if patch adhered
to steel firmly
No, if patch adhered to steel firmly
Minimum threshold 3µg/cm2 ~ 1 µg/cm2 ~ 5 µg/cm2
Reproducibility of results High (for tests above 30 ppm) Low (only minimum and maximum values given) High (for tests above 9 ppm)
Detection sensitivity > 30 ppm > 1 µg/cm2 (4 ppm) > 5 ppm
Detection range 30 - 600 ppm 1 - 50 µg/cm2 1 - 50 ppm
(= 1 -50 µg/cm2
a: Ratio of extracted area to extraction fluid volume.

"A coating inspector could utilize the results of [the] research by selecting a technique based on the levels of chloride expected and the environmental conditions encountered," Appleman says. "For example, if the surface was exposed to the sun on a hot day, the coating inspector might choose to use a method other than swabbing, since that method would result in inaccurate chloride recovery due to water evaporation during extraction in open air. As another example, if the acceptance criterion was low-such as 3 micrograms per square centimeter-a coating inspector would avoid using a method with low sensitivity."

Tips to Improve the Accuracy Of Chloride Tests

  • Because all chloride detection tests are highly sensitive to the testing procedure, the research team developed the following suggestions to help improve the accuracy of the tests.

  • Conduct extractions as soon as possible after the steel is blasted because exposure of clean steel to high heat and humidity can reduce the amount of chloride extracted.

  • Obtain complete training in the extraction method being used.

  • Verify the accuracy for each batch of detector using known chloride standards before performing field tests. Furthermore, extract steel panels that are freshly doped with a solution of known chloride concentration to test the operator's technique.

  • Use extreme caution to avoid losing fluid during the extraction.

  • Extract large areas using smaller amounts of extraction fluid to increase the sensitivity of the test.

From the Lab to the Field

The next step would be to evaluate the performance of the chloride recovery test kits in the field to determine whether quantitative tests should be incorporated into specifications for bridge painting. "This research [was conducted] in a laboratory under controlled conditions, which is appropriate for basic research," Kogler says. "[But] we need to evaluate these methods under field conditions, in cooperation with State bridge owners, to see what kind of improvements we need to make to these tests or procedures before we put [them] in our specifications."

The bar graph shows the effect of aging on the chloride recovery for the swab, patch, and sleeve tests. For the test panel that aged at 37 degrees Celsius (98.6 degrees Fahrenheit) and 57 percent relative humidity (RH) for 4 hours, the chloride recovery is almost 100 percent, similar to that from the freshly doped panels. However, after the test panels were aged at 37 degrees Celsius and 78 percent relative humidity for 4 hours, the chloride recovery decreased significantly for all  the tests. The recovery decreased from 100 to 40 percent, 100 to 80 percent, and 100 to 60 percent for the swab, patch, and sleeve tests, respectively. The graph also shows the error bar for each of the tests. All three tests showed a similar, moderate standard error for the aged tests.

The bar graph shows the effect of aging on the chloride recovery for the swab, patch, and sleeve tests. For the test panel that aged at 37 degrees Celsius (98.6 degrees Fahrenheit) and 57 percent relative humidity (RH) for 4 hours, the chloride recovery is almost 100 percent, similar to that from the freshly doped panels. However, after the test panels were aged at 37 degrees Celsius and 78 percent relative humidity for 4 hours, the chloride recovery decreased significantly for all the tests. The recovery decreased from 100 to 40 percent, 100 to 80 percent, and 100 to 60 percent for the swab, patch, and sleeve tests, respectively. The graph also shows the error bar for each of the tests. All three tests showed a similar, moderate standard error for the aged tests. Source: FHWA.

Future research will not only evaluate the accuracy and reliability of the tests but also the usability. "Imagine climbing up on a ladder to a bridge abutment, with traffic overhead on the bridge," Kogler says. "You have a syringe of water, and you need to glue this sticky thing on the bridge, and use a dropper. It takes a while to run the test. Even though [the tests] were designed to be used in the field, we need to find out how usable they really are."

The big question is whether bridge inspectors should be required to test for chlorides for all bridge painting jobs right now, and whether the tests are accurate and user-friendly enough. In laboratory tests, the researchers found that the testing must be conducted very carefully to ensure consistency of results. "This research provides some of the first real unbiased data on these test kits," Kogler says. "The analysis gives us a snapshot of where the technology is now and where we need to go to improve it."


Shuang-Ling Chong, Ph.D., has been a senior research chemist at FHWA since 1989. Chong's responsibilities have included managing the Paint and Corrosion Laboratory, studying accelerated testing of bridge coatings, and developing methods for characterizing coating materials and failures. Chong earned her Ph.D. in physical chemistry in 1969 from Rutgers, The State University of New Jersey.

Additional details regarding the experimental methods used in the study are available in Dr. Chong's article, "Intra-Laboratory Assessment of Commercial Test Kits for Quantifying Chloride on Steel Surfaces," published in the Journal of Protective Coatings and Linings, p. 42, August 2003. For more information, contact Shuang-Ling Chong at 202-493-3081.

The author would like to thank Yuan Yao and Muriel Rozario of Soil and Land Use Technology (SaLUT), Inc. for their input in preparing this article. The author also would like to acknowledge the vendors of the three test kits evaluated, who were very cooperative in this study.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration