Link to Report: Coming Soon
Background :
The proposed research will investigate the use of hyperspectral imaging for identifying corrosion of reinforced concrete and steel bridge components. Research outputs will comprise i) data characterizing the efficacy of hyperspectral imaging for identification of corrosion prior to corrosion products being visible to the human eye, ii) data characterizing the link between corrosion products that are visible via hyperspectral imaging and the extent of steel mass loss for reinforced concrete and steel bridge components, and iii) recommendations for using hyperspectral imaging as part of a comprehensive bridge inspection and maintenance program.
Objectives :
The proposed research will investigate the use of hyperspectral imaging for identifying the onset of corrosion and monitoring the corrosion process for structural steel bridge members and reinforcing steel embedded in concrete bridge members that are exposed to corrosive environments. The research plan is designed to definitively answer the following questions regarding corrosion of structural steel and reinforcing steel embedded in RC components:
- Can hyperspectral imaging improve on visual inspection methods for identification and characterization of the extent of corrosion? Specifically, can hyperspectral imaging identify the onset of significant corrosion earlier than visual inspection?
- If structural steel and reinforced concrete components are imaged at regular intervals, can hyperspectral imaging provide quantitative data characterizing the progression and/or extent of corrosion damage?
- What additional research is required to develop and validate protocols for using hyperspectral imaging to support corrosion detection as part of a comprehensive maintenance bridge program?
Scope :
Task 1 – Build an accelerated corrosion test setup
A corrosion chamber will be constructed for the project to enable accelerated corrosion of specimens via the “close-circuit impressed current method”. The chamber will be constructed to accommodate specimens subjected to zero load as well as subjected to sustained loading, as cracking of reinforced concrete members could be expected to accelerate the corrosion process.
Figure 1 shows the proposed accelerated corrosion test setup for steel beams, and Figure 2 shows the proposed test setup modified to include application of applied load for the reinforced concrete beams. This setup will accommodate both a 30 in. long W6x25 steel beam and a 30 in. long 6 in. by 6 in. RC beam. Beams will rest on a synthetic sponge and titanium mesh and be partially submerged in a corrosive solution (water with 3.5% NaCl). For steel beams, this setup will be used without axial loading (Figure 1). For concrete beams, axial load will be applied to the protruding ends of a reinforcing bar offset from the center of the beam to produce cracking on the top of the beam that is like the cracking that develops in a beam subjected to flexural loading (Figure 2). A closed electrical circuit will be created by connecting a power source, the steel beam or the steel reinforcing bar embedded in the concrete beam, and the titanium mesh. In this circuit, steel acts as an anode, titanium acts as a cathode, and the DC power source completes the circuit, thereby enabling the research team to quantify and/or control the extent of corrosion developed in the steel element by monitoring the change in electrical resistivity of the system as the steel loses mass due to corrosion.
The extent of corrosion and corrosion-induced damage will be monitored using i) visual inspection and ii) ASTM C876-22b and ASTM G101, which are standard methods for measuring corrosion potential for reinforced concrete and structural steel as well as using iii) hyperspectral imaging equipment that is available for use through the NSF/NIH-funded NHERI RAPID facility housed and administered by the UW Department of Civil and Environmental Engineering, and iv) for RC beams, high-energy CT scanning.
Figure 1: Steel Beam in Proposed Accelerated Corrosion Chamber
Figure 2: RC Beam in Proposed Accelerated Corrosion Chamber with Applied Load
Task 2 – Perform accelerated corrosion of steel elements with hyper spectral imaging of steel elements at multiple levels of corrosion
The first test specimens will be steel beams. To provide understanding of the uncertainty in the experimental results, three identical steel beams will be tested using the same accelerated corrosion protocol. Each beam will be subjected to accelerated corrosion, without load, to achieve the same prescribed set of target increases in electrical resistivity of 2%, 5%, 10%, 20%, and 50%. Prior to starting the accelerated corrosion process, the specimen will be weighed and scanned with the hyperspectral camera. Then, the specimen will be placed in the testing chamber shown in Figure 1 and subjected to the accelerated corrosion protocol to achieve target increases in electrical resistivity ranging from 2% to 50%. Each time a target increase in resistivity is achieved, the specimen will be i) removed from the tank ii) air dried, iii) imaged using visible light and hyperspectral imaging equipment, iv) weighed, and v) returned to the tank and subjected to further accelerated corrosion.
Task 3 – Process hyper-spectral data for steel specimens and assess efficacy of hyper spectral imaging for enhancing visible-light inspection protocols
For the steel beams, electrical resistivity will be used to establish the extent of corrosion for control of the experiment. Thus, at intervals during Task 2, when electrical resistivity increases by a prescribed percentage from the base levels (e.g., 2%, 5%, 10%, 20%, 50%), the accelerated corrosion protocol will be paused, hyperspectral data will be collected, processed and analyzed to assess the efficacy with which corrosion products can be detected and quantified using hyperspectral imaging. Here it is expected that hyperspectral image processing will be accomplished using Python scripts and Jupyter notebooks and will comprise quantification of the area (i.e. number of pixels) within the image in which the wavelength of the radiation measured by the hyperspectral camera matches that associated with steel corrosion products (1900 – 2500 nm). Processing of visible light images will follow a somewhat similar process, but with 50% of the images used to calibrate the visible light wavelengths associated with corrosion products and the remaining 50% of the images used to quantify the accuracy of corrosion product detection using the visible light spectrum. The primary questions to be answered using these data are
- What are the challenges associated with collecting and analyzing hyperspectral data to support corrosion detection for structural steel components.
- Can corrosion products be detected on the beam surface using hyperspectral imaging before and more accurately they can be detected using images in the visible light range.
Task 4 – Perform accelerated corrosion of reinforced concrete components under load
The second set of six test specimens will be reinforced concrete beams. Beams will be 6 in. by 6 in. by 30 in. long and reinforced with a single Grade 60 #3 reinforcing bar off-set from the vertical center of the test specimen to provide the ACI 318 Code minimum cover of 1.5 inches at the top of the beam and to develop concrete crack patterns consistent with flexural loading (see Figure 2). Reinforcing bars will be epoxy coated at the ends of the beam to limit corrosion to the interior of the beam. For each specimen the testing process will comprise: placing the specimen in the accelerated corrosion chamber, subjecting it to accelerated corrosion to achieve levels of reinforcing steel mass loss ranging from 5% to 50%, draining the corrosion chamber and disassembling to provide camera access to the specimen, imaging the specimen using a traditional camera while under load, imaging the specimen using the hyperspectral camera while under load, unloading and moving the specimens from the test setup to the high-energy CT scanner, returning the specimen to the corrosion chamber, reloading the specimen, and continuing the accelerated corrosion protocol. Beams must be unloaded, removed from the test setup, and placed in the relatively small, shielded, rotating CT imaging chamber for CT scanning. Table 1 presents the proposed hyperspectral and CT scanning plan for each of the six reinforced concrete beam specimens, as follows:
- Green fill indicates the percentage of reinforcing steel mass loss, as determined by change in steel resistivity, and the beam load or reinforcement stress level at which hyperspectral imaging will be done.
- Red fill red indicates steel mass loss and reinforcement stress level combinations that will not be considered; specifically, red fill shows that specimens with reinforcing steel loss of 20% to 50% in combination with steel stress levels of 0.5fy to 0.8fy will not be considered.
- Numbers in the table (1 through 6) identify specific specimens. The location in the table of specimen numbers indicates the level of load that will be applied to the specimen and the percentage of reinforcing steel mass loss, as determined from increase in electrical resistivity, at which the specimen will be unloaded, removed from the test rig (Figure 2), subjected to CT scanning, placed back in the test rig, reloaded, and subjected to additional mass loss.
Table 1: Scope of laboratory testing as defined by reinforcement stress level and mass loss. Note: numbers represent specific specimens.
| Beam load or beam reinforcement stress level | ||||
| Reinforcing steel mass loss (%) | No load | Concrete cracking | Reinforcing steel at 0.5fy | Reinforcing steel at 0.8fy |
| 0 | 1 | 2,3 | 4,5 | 6 |
| 5 | 1 | 3 | 5 | no CT scanning |
| 10 | 1 | 3 | 5 | 6 |
| 20 | 1 | 3 | 4,5 | |
| 50 | 1 | 2,3 | ||
Task 5 – Process visible light, hyperspectral image, and CT scan data for reinforced concrete specimens and assess efficacy of hyperspectral imaging for enhancing visible-light inspection protocols
As was the case for the steel beams, electrical resistivity provides a “ground truth” for the extent of mass loss due to corrosion for the reinforcing steel embedded in the concrete beams. Processing hyperspectral image data will be accomplished using Python scripts and Jupyter notebooks and will comprise quantification of the area within the image in which the wavelength of the radiation measured by the hyperspectral camera matches that associated with steel corrosion products (1900 – 2500 nm). Processing visible-light images will follow a somewhat similar approach with 50% of the images used to establish the visible-light wavelengths associated with corrosion products and 50% of the images used test the efficacy of this approach. This ground truth will be used to assess the efficacy of imaging using energy in the visible light and hyperspectral range to detect and quantify corrosion. High-energy CT scan data will provide additional understanding of how internal damage to the concrete matrix affects propagation of corrosion product to the surface of the component. Specific question of interest for reinforced concrete beams include the following:
1. Can use of hyperspectral imaging result in detection of corrosion products on the surface of a reinforced concrete component at an earlier point in the corrosion process than visual inspection or inspection using cameras that detect energy only in visible light range?
2. How does the internal concrete damage pattern affect propagation of corrosion products from the reinforcing bar to the surface of concrete?
3. What are the challenges associated with collecting and analyzing hyperspectral data to support corrosion detection for reinforced concrete components.
Task 6 – Recommendations for future research to support implementation
The proposed research represents a preliminary investigation of the potential for hyperspectral imaging to enhance bridge inspection and maintenance. Thus, a final project task will be the development of recommendations for specific additional research and field-testing activities required to develop prototype bridge inspection protocols.
Task 7 – Dissemination of research results
Research results will be disseminated via presentations at conferences and a journal paper.
Research Team :
Principal Investigator: Laura N Lowes
Co-Principal Investigator: Jeff Berman