The National Steel Bridge NDE Working Group has identified a need to transition away from workmanship criteria to a fitness-for-service (FFS) criteria. To reach this holistic goal requires development of inspection criteria and agreement on the level of reliability required during an inspection. Another research need statement addresses the determination of the reliability level for structural performance with a given flaw size and design stress. However, the next critical piece of this holistic FFS approach relates to the inspection workforce and establishing performance criteria an inspector must meet. For example, the following questions must be answered through research:

• What is the required detection rate (e.g., 50 percent? 95 percent?)?
• What is the probability of detection (POD) associated with various flaw sizes and types?
• What is the reliability associated with characterization of various flaw sizes and types?

Prior research has established that the inspector is a great source of variability in inspection results. Jessop et. al.⁽¹⁾ noted this with just three UT inspectors in 1981 and recommended performance qualification of inspectors. More recently, Connor et. al.⁽²⁾conducted a small round-robin study of conventional ultrasonic testing (UT), phased array ultrasonic testing (PAUT), and time of flight diffraction (TOFD) operators. The results were also not optimistic considering conventional UT operators could detect and size 72 percent of actual reflectors. This decreased to 56 percent for PAUT and TOFD. Those authors also overwhelmingly recommended that performance-based qualifications are required for ultrasonic inspectors. General information on the POD associated with various forms of nondestructive testing (NDT) of complete joint penetration (CJP) welds is needed.  The vision is to obtain this information by conducting a statistically significant round-robin study of inspectors that pass the draft performance qualifications developed in the research.

Scope of work:

1. Develop performance-based qualification criteria for ultrasonic inspectors.
2. Establish the number, geometry, and type(s) of specimens.
3. Establish the number, size, and type of flaws to be included in the specimens.
4. Establish the acceptable and required detection rate (i.e., pass/fail criteria for the performance testing).
5. Establish the standard methodologies associated with conducting such performance tests.
    

References

1. T. Jessop, P. Mudge, and H. David, “Ultrasonic Measurement of Weld Penetration,” 1981.
2. R. J. Connor, C. J. Schroeder, B. M. Crowley, G. A. Washer, and P. E. Fish," Acceptance Criteria of Complete Joint Penetration Steel Bridge Welds Evaluated Using Enhanced Ultrasonic Methods". 2019. doi: 10.17226/25494.

The frequency of nondestructive tests used in steel bridge fabrication were established decades ago, and are rudimentary (i.e., are not statistically based). Also, the quality of work has increased greatly since testing frequencies were adopted. Significant cost savings could be realized in assessing what quality control measures could be trimmed from fabrication requirements due to having little impact on the overall quality and the performance of the bridge in service.

One opportunity for savings is reduced nondestructively testing of complete joint penetration (CJP) groove welds subjected to compression forces (and therefore not subject to tension). The reason welds are tested is to ensure there are not defects present in the weld that will grow into cracks when the bridge is in service, but cracks grow when subjected to a tensile force. Therefore, eliminating NDT of compression joints would be a low risk cost measure, particularly if other measures remain in place to ensure that the fabricators’ overall quality of performing complete joint penetration welds remains high. The objective of this research is to assess the feasibility of eliminating nondestructive testing of CJP welds that are in compression. Care should be taken regarding joints in structural redundant members (SRM), internally redundant members (IRM) and fracture critical member (FCM) although compression joints are not likely to be in these categories. The research should develop a threshold at which a fabricator could apply to eliminate testing of compression joints with consideration of the fabricator’s quality of work. For example, the fabricator might be eligible to apply for reduction in testing based on satisfying a certain maximum of splice rejects and repairs for a given period of time.

Scope of work:

1. Perform a risk-based analysis assessing the feasibility of eliminating NDT of CJP welds in compression
2. Determine overall cost savings to owners
3. Poll the fabricators for current data related to repairs.
4. Develop a minimum threshold from repair data at which testing of compression joints would be required.

Develop a final report supporting findings and make recommendations for AWS D1.5. ⁽¹⁾
 

References

1. Bridge welding code, AWS D1.5, 8th ed. American Association of State Highway and Transportation officials, 2020.

DICONDE is an American Society for Testing and Materials (ASTM) standard (E2339 ⁽¹⁾ which mirrors the Digital Imaging and Communications in Medicine (DICOM) standard for the medical field, to help “facilitate the interoperability of NDE imaging and data acquisition equipment by specifying the image data in commonly accepted terms. This standard represents a harmonization of NDE imaging systems, or modalities, with the National Electrical Manufacturers Association (NEMA) standards publication entitled Digital Imaging and Communications in Medicine, an international standard for image data acquisition, review, storage and archival. In addition, this standard provides a set of industrial NDE-specific information object definitions, which travel beyond the scope of standard DICOM modalities. The goal of this standard is to provide means by which NDE image/signal data may be displayed on any system conforming to the ASTM DICONDE format, regardless of which NDE modality was used to acquire the data” (ASTM E2339 ⁽¹⁾ scope). This standard covers multiple NDE modules such as ultrasound testing (UT), radiographic testing (RT), computed radiography (CR), and computed tomography (CT), but others could be further developed. However, this standard has only really been adopted by the digital radiography committee, and thus puts an asset owner at risk for potential loss of access to vital records at a later date through the use of these other methods.

Research is needed to understand the actual application of DICONDE as it applies to each owner’s business model. By having a standardized format such as DICONDE, in which NDE data can be acquired, reviewed and stored, owners can expect following benefits:

Engineers and inspectors who work with multiple fabricators will only need equipment and software that handles one file format.

NDE equipment companies will know what format is needed by the community and thereby design their systems to produce output in this format.

Less effort will be required by fabricators and contractors seeking approval for use of use for advanced NDE methods using digital imaging, and owners will be more accepting of the advanced methods.

The research will include a complete review of the DICONDE standards, how they may be implemented in a DOT environment and whether or not they are needed.

Scope of work:

1. Research how asset owners are currently store NDE data in digital format.
2. Perform a scan of the United States NDE industry to determine which technologies are supported by DICONDE and who is using them.
3. Develop a report with a high-level review of DICONDE standards and how DOT’s could implement (e.g., neural network).
4. Develop standardized formats and criteria for standardized formats of different NDE modalities for proposal to include in AASHTO documents.
    

References

1. Standard Practice for Digital Imaging and Communication in Nondestructive Evaluation. 2021. doi: 10.1520/E2339-21.

Any nondestructive test or evaluation can be generalized as a test method, comparative to the calibration of the equipment or limitations of the physic principles of the method. The comparative nature is lost with small variations in material properties resulting from heat treatments, steel producing methods, material densities and more.

In the design of steel structures, small changes to the Elastic Modulus (E) have very little effect and have been accepted to be a statistically standardized value because of that. However, in the nondestructive testing scheme of things, small changes in the microstructure of steel can have enormous influences. These microstructural changes come in the form of new alloys and heating or producing processes and have often been forgotten as an essential variable for nondestructive testing purposes when allowing new materials for use in design/fabrication.  

There are many different reasons for developing a good understanding of a material’s properties. Some reasons include: knowing whether or not anisotropy is present; correlating material properties to other variables; and the understanding of attenuation, absorption and other characteristics for a statistical group of materials to better, or more simplistically, perform calibrations.

Each nondestructive testing method is largely limited by the physical principles of the given method and each is influenced differently by the material properties. Anisotropic materials properties can cause beam bending, skewing, and nonlinear attenuations in the ultrasonic testing method without any operator knowledge resulting in unknowingly performing an invalid test. Absent of anisotropic materials, acoustic attenuation values are somewhat unknown for different steels and whether or not current practices are conservative or unconservative.  

There are benefits for understanding the material property influences for other, traditional nondestructive test methods as well. For example, when radiographing a dissimilar weld metal joint is there a need to change or require more than one type of penetrometers comparable to Naval Sea Systems Command (NAVSEA) requirements? Do anisotropic materials influence radiographs like a coarse grain material?

Scope of work:

1. Obtain samples of currently accepted materials for use on highways and bridges.
2. Perform nondestructive evaluations by the ultrasonic testing method to:
   1. Understand what materials/processes might be categorized as acoustically anisotropic.
   2. Understand the influences of testing acoustically anisotropic materials.
   3. Understand the influences of testing dissimilar weld metals.
   4. Measure and perform a statistical analysis to group materials by attenuation values.
3. Perform nondestructive evaluations by the radiographic method to:
   1. Assess the effects of testing dissimilar weld metals, if any.
   2. Assess if anisotropic materials/welds influence radiographs.
4. Develop a written report with findings and publish with public access.

Any nondestructive test or evaluation performed can be classed as a comparative test method, comparative to the calibration of the equipment or limitations of the physic principles of the method. The comparative nature is lost with small variations in material properties resulting from heat treatments/welding, steel producing methods, material densities and more.   

There are many different reasons for developing a thorough understanding of a material’s properties outside of the influences on its comparative nature, including knowing whether anisotropy is present, being able to correlate the material properties to fracture toughness, and having the understanding of attenuation, absorption, and magnetic permeability characteristics for a statistical group of materials to better and more efficiently perform calibrations.

Each nondestructive testing method is largely limited by the physical principles of the given method and is influenced differently by the material properties. Ultrasonic testing has been researched in other industries to measure fracture toughness due to its sensitive ability to assess change in the microstructure.   

Magnetic particle testing for the bridge industry has traditionally been the dry powder electromagnetic yoke technique. On rare occasions, other techniques are used for increased sensitivity, part geometry, or for the ability to test specimens in a laboratory environment. Understanding a material’s magnetic permeability could aid in developments of new techniques for testing through greater coating thicknesses or to measure fracture toughness.

Scope of work:   

1. Obtain samples of currently accepted industry approved materials.
2. Perform nondestructive evaluations by the ultrasonic method to:
   1. Assess the feasibility for measuring fracture toughness via ultrasonic means.
3. Perform nondestructive evaluations by the magnetic particle method to:
   1. Measure material magnetic permeability for potential of statistically grouping materials.
   2. Assess the feasibility for measuring fracture toughness via magnetic means.

The goal of this research is to improve the quality of welding processes used for the fabrication of steel bridges and other transportation structures.  The objective of the study is to explore new innovations for sensing and monitoring welding processes. This project is not intended for implementation in two years, but is intended to seek solutions to be developed over the next 15 years.    

Weld quality is dependent on many variables and is currently defined using a variety of measures. The satisfactory performance of welds depends on processes that are currently controlled with specifications, as well as control of the process inputs. Some variables that are not currently controlled, and their resulting characteristics, are currently being investigated, though some have not currently been evaluated. We   currently control volts, amps, and travel speed. Through training, the intent is to control intermediate bead shape. To my knowledge   we do not explicitly control whether the arc leads the puddle or is in the puddle. We seek cracks, inclusions, porosity, and lack of fusion with varying degrees of success. We have no way to evaluate residual stresses or mechanical properties of the weld or heat affected zone (HAZ).

Some innovation is possible through new applications of current technology, but much more innovation is possible through the development and application of new technology and new materials. The goals of this research are to survey new technologies to find those that can be adapted to identifying any of the process variables or performance-related resulting characteristics, and to provide reliability in welds.  

Illustrative examples include:   

Can we combine artificial intelligence and high-speed photography to see the puddle form, freeze the image, and recognize where there has not fused to the substrate?

Do lack of fusion (LOF) defects leave a different thermal gradient in the near vicinity of the bead, either as the bead freezes or if the weld is locally heated after welding?

Can digital microscopy be used to sense weld discontinuities as the bead fuses to the substrate?

Could we embed microscopic markers   that we could sense and track their relative movement as the weld cools to measure residual strains?

Could we capture local spectroscopy emissions to sense small-scale composition changes?  

The goal is to find new sensing technologies that improve the reliability or economy of finding the characteristics being found today as well as to better sense more directly related issues that affect performance.  

Scope of work:

1. Perform a scan for high temperature sensors and their present use.
2. Perform a scan for high-speed photography equipment and if AI has been incorporated with these instruments.
3. Assess the feasibility in developing nanoparticles that could be contained in the molten weld pool to measure characteristics such as flow and solidification among other small scale composition changes.
4. Assess the feasibility of developing a method to determine thermal gradients.

The goal of a nondestructive test on a groove weld inspection during bridge fabrication is to ensure that unacceptable defects are not permitted to remain in a weld in order to prevent a failure during the life of the structure from fatigue or brittle fracture. However, while current workmanship rejection criteria have been in place for many years, they do not explicitly consider the factors that influence whether a given flaw is or is not actually critical.  For example, the current American Welding Society (AWS) criteria do not consider the magnitude of the stress range, mean stress, toughness of the material, height, or location of the flaw within the weld. This will certainly result in some unnecessary repairs while also permitting some flaws that should be repaired to remain.    

The engineer controls the stress range at the weld in the structural design process by using codified fatigue performance curves based upon 97.5% survival.  Using accepted fatigue and strength design loading and frequency of loading is intended to produce the required design life of the structure.  Ideally, the size of a given flaw would be measured during inspection and compared to the allowable size based on the operational characteristics of the weld.  Unfortunately, the acceptable flaw sizes that correspond to the specific conditions to which the design life applies are not well defined. In almost all girder and other member design, short attachments corresponding to Category C control the design by reducing the stress at the groove welds in the structure. These lower stresses increase the flaw tolerance of the weld. Confounding this issue is the fact that the reliability of an inspection technique to locate and correctly characterize a weld flaw which will provide the desired survival has also not been established.  

The goal of the proposed research is to relate the reliability of flaw inspection to the resulting reliability of the structure. The available worldwide fatigue data bases, including documented in-service failures, shall be used to correlate fatigue life with initial flaw size. A parallel computational fracture mechanics evaluation will be performed using such data to provide a means of extending the effect of size, type, and geometry of potential flaws upon expected fatigue performance. The work should consider both finite life fatigue and infinite life design. The research will result in the development of a framework which establishes inspection methods and flaw acceptance standards which, when combined with selected design stress and expected stress history, will provide the required design life at the desired reliability. It is expected that the research will result in a matrix of solutions of flaw size and design stress. The required reliability of the inspection methods for a given flaw size providing the desired structural reliability will provide guidance for the evaluation of inspection techniques and standard.

Scope of work:

1. Perform a literature review of fatigue data bases to correlate fatigue life with initial flaw size.
2. Develop the framework for inspection methods and flaw acceptance standards to provide the design life at a desired reliability.
3. Perform computational fracture mechanics on the effect of size, type, and geometry of flaws with expected fatigue performance; consider finite and infinite life designs.
4. Develop a matrix of solutions for flaw size with design stress.

The world-renowned quality expert Edward Deming said, “Routine 100 percent inspection to improve quality is equivalent to planning for defects, acknowledgement that the process has not the capability required for specifications. Inspection to improve quality is too late, ineffective, costly.” Deming promoted “process control” versus after-the-fact inspection. The principle as described in this document is to integrate process controls and NDE with the goal of improving the overall process, reducing reliance on after-the-fact NDE.

Sensors are available to detect, measure, and control nearly every variable that affects the quality of bridge welds. Preheat temperature and joint tolerances can be measured immediately before a weld is made. During welding, the voltage, current, travel speed, and gas flow rates can all be measured and verified. Heat input can be instantaneously calculated. Importantly, the instantaneous voltage/amperage relationship of an ideally balanced arc can be compared to the actual welding output; deviations from the ideal conditions represent conditions that may result in weld quality problems. After the weld is made, the weld profile can be instantaneously read with lasers, which can identify visually discernible weld quality problems, including undercuts, undersized welds, poor weld profiles, and so on. Finally, post weld NDE (including the newer phased array ultrasonic testing (PAUT) outputs) can be linked and compared to the input characteristics.

The research goal of this proposal is to determine which quality-related problems can and cannot be detected with process controls as compared to after-the-fact NDE. For example, the loss of shielding gas is readily detected with an imbalance in the voltage/amperage output characteristics that are part of modern welding power sources. Some forms of cracking are related to inadequate preheat and preheat sensors should enable detection of these problems. However, it is not currently known whether all bridge quality concerns can be detected with process sensors.

This proposal involves the integration of the pre-weld sensor output, during welding output, post welding quality surface measurements, and NDE PAUT results in a way that will enable better quality welds on a first pass basis, eliminating the creation of welds that require repair.

Examples of weld flaw identification/elimination include:

Sensors that detect pre-welding conditions such as joint fitup, profile, and preheat measurements can create “no-go” conditions that preclude welding until the conditions are corrected.

Sensors that measure arc instability can be used to shut down welding operations until corrected.

Sensors that detect weld surface profiles a foot or two behind a completed weld can stop welding operations if the output characteristics are non-conforming. For long welds, such as web-to-flange fillet welds, production can be stopped instead of the continued creation of less than ideal welds.

Linkage of NDE results to the production conditions (pre-weld, during weld, and post welding sensor outputs) can help identify the conditions under which weld discontinuities are created, enabling adjustments to the acceptable pre-weld and during welding outputs.

The welding process controls described above can be used to identify portions of welds that are most likely to contain weld discontinuities, and therefore constitute the best use of our NDE resources. The output of the sensors has the potential of “pointing fingers” at suspect welds.

As data is collected, it is probable that the need for new sensors will be identified that would further reduce the need for after-the-fact NDE.

A closed loop system could be developed to adjust welding output in real time, eliminating conditions that result in weld flaws. Currently, seam tracking and joint volume measurement tools are being used to adjust robotic welding conditions to eliminate underfill and incomplete fusion problems.

Process control has the potential of significantly improving first-pass weld quality and reducing the reliance on NDE that is currently a major means by which bridge owners obtain confidence that their structures will perform as intended in service.

Scope of work:  

1. Research sensors capable of measuring joint fit up, profile, preheat and arc instability that can be linked to the welding equipment for correction.
2. Research sensors capable of measuring acceptable surface profiles as welds are being placed and link back to the welding equipment for correction.
3. Perform a series of verification testing that correlates sensor outputs with suspected defects and confirm with NDE.
4. Develop standard procedures when NDE should be performed based on verification testing results of sensor outputs.

Use of nondestructive testing for the steel fabrication community is heavily governed by standardized procedures and requirements contained within code books or industry-specific specifications. In the absent of standardization for a specified testing method, it is unlikely for an industry to adopt the use of newer technologies, or even overlook existing technologies, that might contain some benefit. Using antiquated techniques, when there is the potential for improved quality, reliability, and speed of testing, is a current problem facing the advancement of nondestructive technologies and the steel fabrication community.

A review of existing nondestructive testing technologies and techniques not included in the current steel bridge codes or specifications is needed to see what the possible uses are, and if efforts are needed to standardize a particular method or refine current practices. Understanding that there will be some needs to standardize, additional research of procedure development for a particular application would follow the synthesis. This would include the development and/or adoption of acceptance criteria for critical discontinuities.  

Known examples would include:

1. Eddy current testing.
2. Full matrix capture ultrasonic testing.
3. Refracted longitudinal wave testing with dual matrix array (DMA) (dissimilar welds or antistrophic base metals and welds).
4. Remote visual inspections.
5. Laser scanning of weldments.

These nondestructive test procedures would then be further defined through specification/code parameters for use on common joint configurations, base/weld materials, or a specific application. By having these parameters standardized and used from one fabrication shop to another will ensure consistency in the accept/reject levels.

Scope of work:

1. Perform a literature review of NDE methods and techniques not included in AWS D1.5 ⁽¹⁾ but has use in testing highway infrastructure    
2. Determine the intended measurement of each method not incorporated in AWS D1.5 ⁽¹⁾ and develop standardized procedures and acceptance criteria
3. Present results in a format approved for adopting into AWS D1.5 ⁽¹⁾
    

References:

1. Bridge welding code, AWS D1.5, 8th ed. American Association of State Highway and Transportation officials, 2020.

Current standards for weld quality are based on workmanship. These standards are found in AWS D1.5 ⁽¹⁾, and reflect acceptance criteria for visual inspection and NDE, including magnetic particle testing (MT) and the three volumetric methods, radiography (RT), ultrasonic testing (UT), and phased array ultrasonic testing (PAUT).

Under D1.5 ⁽¹⁾, the volumetric methods are applied to complete joint penetration (CJP) welds, which are commonly found in web and flange   butt splices, as well as other CJPs that are less common, like CJP corner welds in boxes and CJP tee joints on special web-to-flange   connections. Some of these welds are fracture critical (FC); in FC welds, the acceptance criteria are the same, but the frequency of testing is higher. The testing frequency also varies depending on loading condition; 100 percent of tension butt splices are tested by volumetric NDE, and 1025 percent of compression butt splices are testing by volumetric NDE. Failure rates vary among fabricators, but two percent is a typical number. Summing this up, under current acceptance criteria, perhaps two percent of the butt splices welded in steel bridges are repaired.

Repairing butt splices is normal and effective but undesirable. The process involves removal of weld metal and the defect, (perhaps to a depth of half the joint thickness or more) preparing the weld, and then welding it. Under AWS  D1.5 ⁽¹⁾, good practices are prescribed to ensure that weld repair is fine, but a repaired weld is less desirable   than one that has not been repaired. Also, repairs take time and are costly. Usually, excavation, preparation, welding, and retesting take the better part of a shift.  

The workmanship standards used for NDE acceptance were developed for RT and UT in the 1950s and 1960s based on reasonable expectations for the level of quality that welders could achieve (except for PAUT). The use of these standards has been effective. In the hundreds of thousands of bridge butt splices that have been accepted through these criteria since they were implemented, zero welds (that met the criteria) have failed or otherwise had any service problem.

While the performance of welds under current criteria is excellent, it suggests that the current criteria are more stringent than necessary. The NCHRP funded research to explore this question (Report 908 ) ⁽²⁾. However, that study found that the acceptance criteria should actually be more stringent. These findings were based on design fatigue limits. The report 908 study shows that the actual stress ranges seen by the subject bridge welds need to be discovered and considered and thereby establish what loads should be used for fitness-for-purpose evaluation. Report 908 also points an issue regarding the actual toughness of bridge steels. Material test reports on modern projects show that actual toughness is generally much higher than minimum requirements, which is another factor than can be considered in fitness-for-purpose evaluations.

Modern NDE technology also facilitates the opportunity to refine evaluation. PAUT allows improved flaw sizing and orientation, both of which can be used for better fitness-for-purpose evaluations. To improve productivity, one approach could be to use current or updated criteria as a first pass in the shop, and then use a refined evaluation on defects to establish whether or not they actually need repair.

Further research is needed to examine, once again, the acceptance criteria needed for bridge butt welds, including these factors:

To examine the applicability of the current design fatigue limits to this question.

To reconsider actual loads seen by bridge welds and the associated fatigue performance needed for bridge welds.

To examine the use of refined defect size evaluations given the improvement of NDE through modern methods.

Scope of work:

1. Perform a literature review of previous research associated with acceptance criteria for welded steel bridge groove weld.
2. Evaluate current design fatigue limits for welded steel bridge groove welds.
3. Obtain data for actual loads and fatigue performance on steel bridge groove welds from modern bridge designs.
4. Assess the feasibility and develop refined defect size limits that accommodate the design and fatigue and can reliability be detected with modern NDE methods.
    

References

1. Bridge welding code, AWS D1.5, 8th ed. American Association of State Highway and Transportation officials, 2020.
2. R. J. Connor, C. J. Schroeder, B. M. Crowley, G. A. Washer, and P. E. Fish, "Acceptance Criteria of Complete Joint Penetration Steel Bridge Welds Evaluated Using Enhanced Ultrasonic Methods". 2019. doi: 10.17226/25494.

Steel structure owners have traditionally assessed the condition of their assets using human assisted inspections consisting primarily of visual and physical examinations, occasionally supplemented with more advanced methods such as Magnetic Particle (MT) and Ultrasonic Testing (UT). These inspections have been shown to produce widely varying results due to the inability to detect all deficiencies and inaccurately quantify them or correctly evaluate their severity.

Recently, agencies have been exploring the use of unmanned technologies to augment visual inspections. A more popular platform consists of sensors attached to small Unmanned Aerial Systems (sUAS). Typical sensors include high-definition photographic cameras, thermal cameras, and Lidar. Advantages of using sUAS include easier access to difficult to reach areas and minimizing traffic impacts. An additional advantage to inspection by sUAS is the ability to capture the data digitally which can be stored, enhanced, and integrated into 3D models of the bridge. A further advancement would be an AI based system that would recognize the anomalies on the structures and highlight them for evaluation.

Disadvantages include difficulties in accessing confined or congested locations, navigating in adverse weather conditions, collection of tactile data, efficient use of large sets of high-resolution data, and the inability to consistently detect all defects, particularly cracks.  

The Transportation Research Board (TRB) workshop 1008 in July 2021 provides a good summary of current applications. The FHWA report “Collection, Analysis, and Interpretation of Data Obtained from Unmanned Aerial Systems for Bridges provides best practices operating sUAS and visual bridge inspection including managing the data collected by the sUAS. As reported in the TRB workshop some owners do not use sUAS for inspection of fracture critical members. The purpose of the biannual inspection of fracture critical bridges is to detect subcritical cracks growing by fatigue or hydrogen assisted cracking before they become unstable. What is needed is development of technology to detect small cracks in steel bridges. Obviously, as was found in the I40 bridge, sUAS are more reliable than traditional inspection for finding large cracks.

The need is to identify and refine existing technologies to overcome the deficiencies encountered in existing technologies, particularly in the ability to identify and measure crack lengths. Visual detection of small, less than 1 inch in length, fatigue cracks is difficult particularly if they are growing from the toe of a weld. The inspection is further complicated by the coatings and corrosion debris which can mask the crack.  

There are a variety of technology advances used for other applications in other industries that may improve or even replace the human inspection. There are various deployment methods, both contact and non-contact, that can improve upon or supplement current sUAS capabilities.

Methods include:

1. Traditional sUAS with enhanced navigation capabilities in GPS denied environments (e.g. the Skydio X2D) and allowing access to congested locations.
2. Aerial Robotic NDE, sUAS deploying contact testing (e.g., UT thickness testing, see Microsoft Word - NDT.Net Dec 2020 - Aerial Robots and the War on Corrosion - NDE.docx)
3. sUAS with tilt mechanisms for aerial deployment and thrust assisted contact see IEEE Xplore Full-Text PDF:).
4. Magnetic crawler (see Multi-directional Bicycle Robot for Steel Structure Inspection - YouTube)
5. Spider Bots (e.g., The Spider-Bots Are Coming - TFOT (thefutureofthings.com))
6. Worm robots (see These earthworm-inspired robots may one day crawl through aircraft engines (interestingengineering.com))
7. Other adherence methods such as suction, adhesive action, or tarsal-like claws that grip tiny surface irregularities.  
8. Combined methods.

These can be used to deploy a variety of testing methods, conventional (VT, UT, MT, MFL) and newer and more advanced technologies, including crack detection using high resolution digital photogrammetry and thermography. China Jiliang University and Purdue University have jointly developed a thermal imaging system for crack detection that sees through corrosion using deep learning. Movable heating is required to produce thermal gradients for the scheme to work. Work by Runnemalm and Brogerg at University West in Sweden uses the time derivative of the temperature change using UV light as an excitation.  

These methods and others need to be evaluated and refined for application of contact and non-contact methods for inspecting steel bridges.

The research should also address efficient data collection, transmission, and evaluation and incorporation of Artificial Intelligence (AI) and Deep Learning techniques to enhance automation of all processes (deployment, data collection, data transmission, and data evaluation).

Scope of work:

1. Review of multiple deployment methods and identification of those with the most promise.
2. Review of testing and inspection methods compatible with the various deployment methods and identifying those with the most promise.
3. Procuring or developing deployment systems and incorporating testing and inspection equipment.
4. Evaluate detection and sizing capabilities of the selected system considering varying surface conditions.
5. Develop automating routines for each of the processes.
6. Evaluate the technology at the S-Brite center at Purdue which has documented cracks at typical bridge weld details.