Above: The nuclear industry has moved beyond manual ultrasonic testing for volumetric examinations
In the early days of nuclear power plant inspection, there were only the basic methods of radiography (RT) and manual ultrasonic testing (UT) for volumetric examinations. Inspectors used cumbersome single-wave UT tools that had limited capabilities to detect flaws in nuclear reactor components and piping. Today, the once analogue practice of nuclear plant inspections has undergone a digital transformation. Technological advances in predictive maintenance, such as AI, sensors and other automated tools, have led to improved inspection quality to enhance nuclear safety.
Non-destructive Evaluation (NDE), or Non-destructive Testing (NDT), refers to the inspection of materials without affecting their useability and the overall integrity of the asset in which they are installed. NDE is applied in many industries. In the case of nuclear power generation, it is used to evaluate components in both the initial manufacturing phase and throughout the reactor’s lifecycle.
NDE helps maintain the very strict quality control standards in place to ensure safe operation of nuclear power plants. Engaging certified NDE inspectors in routine evaluations helps reduce the risk of manufacturing defects and in-service flaws going undetected, which in a worst-case scenario, may lead to operational failure.
Changing NDE in the nuclear industry
With the integration of new technology, NDE is shaping the future of nuclear safety. The American Society of Mechanical Engineers (ASME) and the Nuclear Regulatory Commission (NRC) heavily regulate nuclear power plants in the United States. For both Boiling Water Reactors (BWR) and Pressurized Water Reactors (PWR), several safety-related systems are required to undergo regular inspections, in addition to routine assessments of primary reactor components and containment vessels. The rules dictate when and how often inspectors must test each asset using NDE and which inspection methods should be applied.
Frequently used NDE techniques in the nuclear power industry include ultrasonic testing and Phased Array Ultrasonic Testing (PAUT), Liquid Penetrant Testing (PT), Eddy current Testing, (ET), Magnetic Particle testing (MP) and Visual Testing (VT). Of these, PAUT represents an advanced method of UT inspection, and uses a set of probes made up of multiple segments that can be individually computer-activated by the examiner. Each of the probes allows separate, staggered pulses, enabling NDE professionals to create a guided sound beam to sweep across the component collecting data and visualising the component’s inspected area.
In contrast, PT is an NDE method that requires minimal training. It is used widely for inspecting nuclear power plant piping and components. PT inspections check components for material flaws visible at the surface by flowing a very thin liquid known as a penetrant into any potential discontinuities and then drawing the liquid out with a chalk-like developer to reveal if an actual flaw exists. Because most of the components in nuclear reactors contain non-ferromagnetic materials PT can easily and quickly reveal surface breaking flaws.
NDE in the digital age
As the concept of a fourth industrial revolution takes hold in the energy industry, NDE technology leaders are going beyond traditional tools and procedures by integrating automation and artificial intelligence (AI) into inspection techniques. The goal of these efforts is to leverage new technological advancements to improve inspection efficiency and detection probabilities.
Adopting new, advanced technologies and tools has become necessary for NDE inspectors as they support plant engineers in meeting the needs of the nuclear energy industry. As technology such as automation and AI evolves and grows in adjacent sectors, that technology can be adapted to bring enhancements to NDE methodologies as well. NDE experts in the power generation industry are working to tap into this potential.
In the nuclear energy field, integrating advanced technology and adequately training inspectors in advanced inspection procedures can improve the reliability and efficiency of inspections. Evolving toward a monitoring approach rather than traditional inspection may offer real benefits to utilities as well, since inspection personnel can spend less time in high-temperature, high-pressure, and potentially dangerous conditions. For emerging reactor designs such as Small Modular Reactors (SMRs) and non-light water reactors, monitoring of components at high temperatures could be paramount.
When any flaw is detected, the examiner must look at various aspects, such as thermal shock and fatigue. Conventional ultrasonic sensory technology is built to survive a maximum temperature of about 200°C (392°F). As nuclear plants install advanced reactor vessels with much higher temperatures, sensors also will need to advance to survive. Extensive thermal testing trials are currently underway with sensor prototypes to demonstrate leading-edge sensor prototypes, and to find adhesive alternatives to clamps that can secure permanent sensory technology in place to monitor hot reactor components.
In the modern age of digitalization, emerging technologies in machine learning and AI are improving workplace efficiency and making automated industry capabilities possible. From the introduction of UT and PT to the emergence of PAUT, digitalisation of NDE has been steering advances in automation. Organisations such as EPRI – the Electric Power Research Institute – are leading the way in exploring and testing emerging technology. EPRI’s team of researchers is currently examining how advancing automation technology such as AI, drones and sensors can benefit the nuclear and NDE industries. Additionally, other leading-edge technology being evaluated supports the implementation of permanent sensors in the nuclear industry that can monitor and alert personnel if maintenance is needed on structures or components.
AI and sensors: The key to NDE automation
Artificial intelligence is a powerful new tool that can help expedite and improve the data and image collection process of NDE by taking on monotonous, tedious and repetitive tasks now handled by examiners. It will not replace the work of an NDE inspector but will allow them to efficiently analyse the most pertinent data collected by new technologies.
Another potentially important use for AI is to analyse new types of flaws that may arise from new reactor designs and manufacturing techniques. Everything the industry knows about fabrication flaws is based on what examiners have seen for decades and which is then used to form a general expectation of what each kind of fabrication looks like. With new SMR and non-light water reactor designs as well as new manufacturing techniques, it can be more difficult for engineers to know what to expect.
Implementing new manufacturing techniques can lead to new types of flaws unfamiliar even to experienced inspectors. For example, fabrication techniques like electron beam welds are not expected to produce the kinds of flaws typical of conventional welding methods. Any new kinds of flaws will have to be addressed by the industry as they are encountered using enhanced inspection methods and comprehensive training courses.
Integrating AI into the NDE inspection process will increase reliability and efficiency by helping inspection professionals make well-informed, data-driven decisions. Recently, EPRI conducted NDE field trials on partial penetration J-groove welds and dissimilar metal welds (DMWs) at various nuclear stations. The AI programme collected a vast amount of data and raised flags for further review by examiners. AI automated much of the confirmation process and shaved off substantial time throughout this process. In one instance, AI cut the data analysis process down from four days to just four hours without loss of assessment integrity. This advancement can enhance nuclear safety by allowing examiners to focus more time assessing areas of welds that require a more detailed evaluation. It can also ease staffing issues for NDE personnel with the added advantage of the AI program’s ability to significantly shorten inspection time.
For example, as a part of its training for a Reactor Vessel Upper Head (RVUH) application, an AI model was fed a large volume of UT data from a two-unit PWR that had been slated for decommissioning. Experts instructed the model to evaluate the data and alert the examiner if further review and consideration were needed. Although most of the data volume yielded no flagged indications of interest, the AI tool was able to identify benign conditions such as fabrication-related responses that require review from a qualified examiner to confirm their status. As the AI application performed the monotonous task of searching through all the UT data looking for indications of interest, the UT examiners could spend more time evaluating each pre-identified indication by comparing the response in the new UT data against the long history of archived data files from the previous examinations. If any changes to the UT indication were noted to have occurred over time, any changes to the UT responses were further scrutinised as potentially originating from service-induced degradation. Such a detailed historical comparison requires significant time, and the use of AI provided time savings in other areas that then afforded examiners the ability to expend this additional time where it mattered the most. In the future, experts plan to develop an AI program for manual and automated PAUT inspections that will be more heavily involved and complex with the ability to evaluate the data in real time.
As experts address areas of improvement for Reactor Vessel Upper Head AI tools, new program field trials for AI applications for other nuclear components have already begun to launch. Experts are now exploring a comparable program development and testing process for dissimilar metal welds that may enable them to build an AI application to examine core shroud welds in BWRs and core barrel welds in PWRs.
Through these field trials, the industry is seeking a way to close the technology gap. Online monitoring will perhaps then be able to replace NDE data collection where it is unsafe for human inspectors to evaluate the reactor components. With permanent sensors, examiners can continuously monitor for defects in the component and track flaws that might trigger a need for repairs.
In addition to AI, uncrewed inspection systems, such as drones and permanent sensors, are emerging automation tools. By integrating all these advanced technologies into NDE inspection, examiners can use remote inspection for a potentially fully automated analysis. Automation is the key to efficiency and reliability. It can also improve the quality of data and lower the cost of inspections.
Future evolution of NDE
As countries focus their attention on advancing nuclear power generation capabilities, it will become increasingly important for engineers to invest in effective and cost-efficient advanced monitoring and sensor technology. The integration of automation into NDE techniques will continue to develop across the nuclear energy industry. Although experts are still working out the kinks of using advanced technologies, it is paramount that inspection methodologies evolve as surrounding technologies evolve to ensure reliable and cost-effective asset integrity evaluation and monitoring. The success of today’s NDE digital transformation will also rely on comprehensive training for examiners to become proficient in the latest and newest tools and technology.
Given nuclear plant inspections continue to evolve from the cumbersome original tools to increasingly advanced, powerful, and efficient new toolsets, the transformation of NDE will continue to ensure the safe and sustainable future of nuclear energy.
Authors: Danny Keck, Independent NDE Consultant and Current Chair of the Board of ASNT; Randy Linden, Senior Level III for Sonic Systems International; Bret Flesner, Principal Technical Leader at EPRI; Luke Breon, Senior Technical Leader at EPRI