Anthony J Alongi, President
Chloride infiltration into concrete is the major cause of corrosion induced delamination in steel reinforced bridge decks and repairing delaminated concrete is a major cost factor in bridge deck rehabilitation. Knowledge of the quantity and location of chlorides in bridge deck concrete is an important factor in decisions relating to the type and extent of repairs. Traditionally, the measurement of chlorides involves core sampling and laboratory testing of concrete samples. While this gives an accurate measure of chloride at the location of the core, it cannot readily determine the distribution of chlorides throughout the deck or the maximum or minimum chloride amounts without taking a great number of cores. A GPR method based on the measurement of signal attenuation was developed by Penetradar research staff as part of an NCHRP Idea project, which has the potential to provide that information and provide bridge owners and engineers with detailed information on the quantity and location of chlorides in their bridge decks, thereby improving the effectiveness of repairs, reducing the cost of maintenance and repairs, and potentially increasing the lifespan of the bridge deck.
GPR signal attenuation in bridge deck concrete occurs as a result of the conductive nature of the concrete when water and chlorides are introduced. Research conducted by Penetradar staff several years earlier under SHRP C101 found that a relationship existed between radar signal attenuation, chloride levels in concrete and moisture content. The results of that research showed that the signal attenuation experienced by the radar wave in concrete is affected by the quantity of chloride and moisture. While that research developed the GPR method of detecting delamination, which is in widespread use today, the current development provides a method for measuring the quantity of chloride in the concrete, based upon a detailed GPR scan of the deck along with limited coring (two or three core samples) and laboratory chloride measurement of those samples to produce an accurate and quantitative, spatial mapping of chlorides in bridge decks.
In the future, with further development it is believed that GPR will be able to predict chloride levels in bridge deck concrete independently and without the need of calibrating core samples.
Bridge decks are a vital component in a nation’s transportation system and represent a major cost factor for transportation engineers and DOT maintenance staff. It is well known that chlorides from deicing salts can attack the steel reinforcement in bridge decks. This occurs primarily in northern, cold weather climates but can also occur in coastal areas. Chloride intrusion into reinforced concrete structures, while sometimes taking years to occur can eventually cause corrosion of the reinforcement and then delamination, which is the laminar fracturing of the concrete. In the USA, the repair cost from these defects are estimated to exceed $5B per year and make up between 50% – 85% of bridge maintenance budgets. While the choice of remediation methods is limited, the removal and replacement of chloride contaminated concrete is the most long-lasting and generally the most cost-effective method in the long run. However, the problem is that few methods exist to determine chloride content in bridge decks. The most widely used method involves closing traffic lanes, extraction of large numbers of core samples and testing the samples in the laboratory for chloride ion content. While providing quantitative information, this method is expensive and time consuming, since it creates traffic slow-downs and a potential safety hazard, and because cores are discrete samples often produce inadequate information on the bridge deck condition and chloride quantities.
What was needed is a better method, which is faster, more accurate and lower cost, that provides quantitative information on chloride content over the entire bridge deck. Such a method would permit improved repair strategies by identifying specific areas of chloride contaminated concrete and thereby improving the effectiveness of repairs.
To address this need, a high-speed, non-contacting, ground penetrating radar (GPR) technique was developed, that provides a deck-wide topographical mapping of chloride concentration at the rebar level. This method utilizes a GPR scan of the entire bridge deck along with a minimal number (3 or more) of core samples and laboratory chloride measurements of the samples to calibrate the GPR measurements.
This entirely new method for determining chloride quantity in bridge deck concrete uses existing air-coupled GPR technology. The underlying research establishes and quantifies the relationship between chlorides in concrete (which cause corrosion of reinforcing steel and delamination of concrete) and the effect on GPR signal propagation. A chloride mapping method was developed that produces a complete and quantitative mapping of chlorides in bridge decks, the extent of which cannot be achieved in the same detail with any other method.
Ground Penetrating Radar has been utilized for several years in nondestructive testing on bridge decks. In the early days of the development of this technology the focus was on detection of delamination in concrete bridge decks. Research in this area took different approaches, including the investigation of waveform features and waveform signatures specific to bridge decks where delamination was present. SHRP C-101 research conducted in the late 1980’s found that a strong correlation existed between the level of signal loss or attenuation in a radar signal and the presence of delaminated or deteriorated concrete. It was also found that the conditions in the concrete causing corrosion of reinforcing steel and subsequent delamination, i.e. chlorides and moisture, were the same conditions that caused signal loss or attenuation of the radar signal. This research showed that GPR measurement of attenuation could be used to identify areas of delaminated bridge deck concrete and was the basis for GPR detection of delamination in bridge decks that is in use today.
The SHRP C-101 research not only showed that GPR was responsive to the chlorides and moisture content in concrete but also suggested that there was a deterministic relationship between GPR signal attenuation and the combined amount of moisture and chlorides in concrete. In other words, radar signal attenuation was a function of both the quantity of chloride and moisture content in concrete. This past research set the stage for the investigation described here which builds upon those results, and shows that GPR can be used as an analytical tool to identify the quantity of chlorides in concrete based on measurement of signal loss or signal attenuation.
Radar signal attenuation in bridge deck concrete depends on several factors, including the composition or thickness of the concrete, the reflective nature of the reinforcement, and the quantity of moisture and chlorides. Of these, the main variable factors are moisture and chloride. It is well known that corrosion of the embedded reinforcement is likely to occur when chlorides and moisture reach the reinforcement in sufficiently high concentration. To utilize GPR for measurement of chlorides it is necessary to relate signal attenuation to the factors that cause corrosion, i.e. chloride and moisture, and develop what we have defined as the attenuation-chloride relationship.
The development of the GPR method for measurement of chloride involved a three-pronged approach. First, in order to understand the physical principles an analytical model was developed that defines the electromagnetic interaction between the GPR wave and the concrete material, and describes through mathematical derivation the GPR losses or attenuation in chloride contaminated concrete. Modeling is important in that it provides an analytical foundation to support the concept. The second approach involved laboratory experimentation using test samples to validate the model and finally, the third approach involved a GPR field test on a bridge deck to predict chloride quantities, where chloride levels were accurately known. In each of these approaches the attenuation-chloride relationship was derived and quantified.
Initially an analytical model was developed to mathematically define the signal loss experienced by a radar wave in concrete with varying levels of chloride and moisture. The model was developed based on conventional electromagnetic theory for plane waves in lossy media and shows the theoretical relationship, defined in mathematics, between GPR signal attenuation, chlorides and moisture in concrete, i.e. the attenuation-chloride relationship. It relates the level of radar signal attenuation (in dB per inch) that would occur based on a predefined chloride quantity (lb/yd3) and moisture level (% by weight) in a material such as concrete.
The results of the model are shown in Figure (1). It was found that GPR is in fact responsive to chloride levels in concrete, and that there is a direct relationship between chloride quantity, when in the presence of moisture, and the conductivity and complex dielectric constant of the material. This means that greater levels of chloride, when in water solution, cause an increase in conductivity which in turn results in greater signal loss or attenuation. The model also revealed that signal attenuation (as measured in dB) at 1GHz and 2GHz, which are the center frequencies of a typical transmitted GPR pulse, is almost linearly related to the quantity of chloride in the material at concentrations normally found in bridge deck concrete, for a given moisture content. The linear relationship suggests that the chloride content can be adequately defined with a minimum of two or three calibrated GPR attenuation measurements when assuming relatively constant moisture. This also shows that from the attenuation-chloride relationship, chloride quantities can be extrapolated for the remaining attenuation measurements for the remainder of a bridge deck. In practice, chloride calibration measurements and corresponding GPR signal attenuation measurements will identify the specific moisture curve in Figure (1). Once the curve is identified, the attenuation-chloride relationship is established.
To confirm the analytical model, a series of experiments were then conducted with parameters similar to those used in the analytical model. Several test boxes were constructed containing a mix of compacted sand and gravel to simulate PCC. A predefined amount of chloride and water were added to replicate the conditions that would be encountered in bridge deck concrete in-situ, with chloride ranging from 0-10 lb/yd3 and moisture content ranging from 0% to 10% by weight. Radar testing was conducted using 1GHz and 2GHz (center frequency) air-coupled antennas to measure signal properties including relative dielectric constant and signal loss. The result of these experiments was the development of the attenuation-chloride relationship for different moisture concentrations. The experimentally derived attenuation-chloride relationship for a 1GHz antenna is shown in Figure (2). The experimental results confirmed the analytical modeling and while not identical were remarkably similar. The experiments also showed that the relative dielectric constant of the material is unaffected by the quantity of chloride introduced, at the frequencies tested, and that the presence of chlorides without moisture in the material had no measurable effect on signal attenuation relative to the case without chloride. It was only when moisture and chlorides were present together that the radar signal experienced attenuation. For chloride and moisture contents typically found in bridge deck concrete, the experiments determined that attenuation is directly, and in most cases linearly related to the quantity of chloride in the material for a given moisture content. This confirms what was found by the analytical model and also means that the attenuation-chloride relationship can be adequately defined with a minimum of two or three calibrated GPR attenuation measurements (cores).
The third approach involved field tests, which were carried out to determine the feasibility of the GPR method for predicting chlorides in-situ in bridge deck concrete. Good correlation was found between radar attenuation measurements and chloride levels in concrete, based on tests conducted on the Interstate 395 Southbound bridge over Sanger Avenue in Arlington, VA. Penetradar IRIS GPR equipment was used for data collection and bridge deck chloride information was provided by VDOT (NOVA District) from earlier in-depth coring that was performed. A comparison was made between the non-contacting, ground penetrating radar measurements of signal loss and laboratory measurements of chloride obtained from cores. A regression was performed between GPR signal attenuation and chloride quantity to determine the level of correlation that existed and the R-Squared value or quality of the relationship. R-Squared values can range from 0 for no correlation, to 1 for perfect correlation and generally the higher the R-Squared the better the ability of the independent variable, which in this case is GPR attenuation, to predict the dependent variable, chloride quantity. Based on the six laboratory chloride samples available, an R-squared value of (0.875) was determined to exist for this data set. For this bridge deck, the results suggest that GPR signal attenuation is a good predictor of chloride quantity in concrete.
Based on the GPR attenuation data it was possible to create a mapping of bridge deck chloride quantities, showing detailed levels of chloride throughout the deck. From the regression model and chloride mapping, the maximum and minimum chloride levels were extrapolated and found to vary between 0 lb/yd3 and approximately 7.4 lb/yd3 (at the 99th percentile), with an average chloride level for the entire bridge deck to be 2.0 lb/yd3. This compared favorably with average laboratory chloride measurements of 2.6 lb/yd3.
It was also found that the attenuation-chloride relationship could be reasonably defined with only three chloride sample measurements, one corresponding to low, intermediate and high attenuation, with an R-squared of 0.835, as shown by the GPR chloride mapping in Figure (3). In a direct comparison between chloride quantities based on laboratory testing of cores and GPR predictions of chloride, 53% of the laboratory samples exceeded a level of 2 lb/yd3 – which traditionally is in a range of chloride concentration that can cause corrosion of reinforcement in concrete, while 47.7% of the GPR measurements exceeded a 2 lb/yd3 threshold.
It was found that the selection of core and chloride sample locations are important to overall accuracy, as is locational accuracy in GPR data collection and ensuring that the core samples are taken in the correct locations. Deviations in location can produce significant errors and adversely affect GPR’s ability to predict chloride quantities.
Ground penetrating radar has been used in the past to measure material properties, however, radar measurement of chlorides in concrete has not been extensively studied. The method that was developed here represents an entirely new and novel approach to the measurement of chloride in concrete and the non-destructive evaluation of bridge decks.
This work defines the relationship between GPR measurement of signal loss based on variations in conductivity resulting from intrusion of dissolved chlorides and moisture into concrete, i.e. the attenuation-chloride relationship. This was demonstrated (1) analytically using radio frequency (RF) and microwave theory (2) experimentally with laboratory measurements, (3) with actual GPR bridge deck data using core samples.
Our results show that GPR technology is able to predict chloride quantities in bridge deck concrete using limited ground-truth information based on core samples and laboratory measurements of chloride. We have shown that this is possible with a high level of accuracy using only three core samples for calibration, which resulted in an R2 of 0.835 on a test bridge deck. The high degree of correlation implies that a measurement of radar signal attenuation in bridge deck concrete is a good predictor of chloride content. In developing the GPR attenuation-chloride relationship we were able to produce a chloride mapping that shows the chloride distribution throughout the bridge deck, and predict the average, maximum and minimum chloride levels.
Going forward, we believe the attenuation-chloride relationship can be further developed to eliminate the need for chloride calibration information and with additional research it may be possible to accurately predict chloride quantities in bridge deck concrete with radar, independent of cores and laboratory measurement of chlorides for calibration.
With the increased focus on infrastructure investment, this method will provide much needed information to bridge owners on the condition of bridge decks; it will improve the effectiveness of bridge deck repairs, reduce repair costs and increase the longevity of bridge decks.