Final Report Summary - HOTSCAN (Long range ultrasonic system for continuous in service inspection and structural health monitoring of high temperature superheated steam pipes in power generation plant with 100% coverage)
Executive Summary:
High temperatures and pressures in superheated steam pipes in nuclear power plants can lead to the formation of flaws and defects due to material degradation mechanisms such as corrosion, creep and fatigue over time. If undetected in advance, these critical features can cause catastrophic failures which in turn will result in significant environmental and financial consequences. Inspection using conventional Non-Destructive Testing (NDT) techniques such as visual testing, eddy current testing and ultrasonic testing are employed to assess the integrity of the steam pipelines during planned outages for nuclear power plant maintenance. However, if the condition of a pipeline is not monitored between outages, this may present a problem for an aging power plant as defects such as creep may become more prevalent. In situ condition monitoring techniques therefore need to be developed to retain reliability and extend the lifetime of nuclear power plants.
The Hotscan project concept was to develop a Long Range Ultrasonic Testing (LRUT) system for continuous in service inspection and structural health monitoring of high temperature superheated steam pipes in power generation plant with 100% coverage. Performing LRUT at elevated temperatures can be complicated. The effect of temperature on wave propagation was modelled to understand how the selected wave modes behave at elevated temperatures. The optimum wave mode selected was the Torsional T(0,1) wave mode. The main challenge was to develop transducers that could withstand high temperatures. Lithium Niobate (LN) was selected as the most suitable piezoelectric material for further investigation. Owing to the significant risk of 580°C transducers not being feasible, a staged approach was adopted to prove some of the development strategies using the PZT transducers. Existing PZT type transducers were also further studied and improvements were made to extend their operating temperatures and were shown to work successfully up to 250°C. Prototype LN transducers were manufactured and tested ultrasonically for temperatures up to 600°C. The ageing behaviour of both PZT and LN transducers was investigated and it was shown that the PZT transducers could successfully operate at 250°C for over 30 days and the LN transducers were tested up to 580°C for 12 days. A collar array was designed and manufactured using high temperature materials suitable for high temperature operation. Its performance was compared with a commercial collar array and was shown to give comparable results at ambient temperatures, and was also shown to work at elevated temperatures in combination with the ultrasonic pulser-receiver unit during the laboratory trials. The integrated Hotscan system was installed in an electrical power plant to assess its performance in service conditions. The site trial was carried out in an Iberdrola coal power plant in Spain. Room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed. Data was collected during the rise in pipe temperature up to 339°C, but the pipe temperature exceeded the specified operation temperature of the system and hence it stopped working. Further work needs to be carried out to realise a commercial system. Although initial field trials have proven the concept further site trials are necessary to convince plant operators. The consortium will maintain communications after the project, and will review and exploit all potential commercialisation opportunities.
The Hotscan project has enabled a number of SME’s together with experienced RTDs with extensive expertise of innovation and research to define and explore a novel concept. The Hotscan system has the potential of providing benefit to the participants, to the power and oil and Gas industry in terms of safety management, the EU in terms of new Jobs and society through improved safety.
Project Context and Objectives:
High temperatures and pressures in superheated steam pipes in power plants can lead to the formation of flaws and defects due to material degradation mechanisms such as corrosion, creep and fatigue over time. If undetected in advance, these critical features can cause catastrophic failures which in turn will result in significant environmental and financial consequences. Inspection using conventional Non-Destructive Testing (NDT) techniques such as visual testing, eddy current testing and ultrasonic testing are employed to assess the integrity of the steam pipelines during planned outages for nuclear power plant maintenance. However, if the condition of a pipeline is not monitored between outages, this may present a problem for an aging power plant as defects such as creep may become more prevalent. In situ condition monitoring techniques therefore need to be developed to maintain reliability and extend the lifetime of nuclear power plants.
The Hotscan project concept was to develop a Long Range Ultrasonic Testing (LRUT) system for continuous in service inspection and structural health monitoring of high temperature superheated steam pipes in power generation plant with 100% coverage. LRUT is a type of ultrasonic testing but it is fundamentally different as it uses lower frequencies, normally in the kHz range, and inspects over a range of metres as opposed to MHz frequencies and centimetre ranges of conventional ultrasonics. There are commercial LRUT systems for inspection at ambient temperatures, such as the Teletest® system developed by Plant Integrity Ltd. The system consists of a transducer collar array that is driven by an ultrasonic pulser-receiver and a laptop PC to display the data.
LRUT employs Ultrasonic Guided Waves (UGW) to inspect long length of pipe from a single location by using the pipe as a wave guide. The system can detect internal and external corrosion as it is sensitive to volumetric wall loss. It can reliably detect 9% pipe wall cross-section loss and using focusing techniques it can detect up to 3% cross-section loss. Sound energy propagates through a structure in specific types of vibration pattern, known as wave modes. Many wave modes are generated when the transducer rings are excited. The key wave modes that are used in LRUT of pipes are the longitudinal L(0,2) and torsional T(0,1) wave modes. During inspection conditions are chosen to ensure the wave has a constant velocity over the frequency range. Performing LRUT at elevated temperatures can be complicated. The pipe geometry is likely to expand or contract due temperature and material properties such as Young's modulus will also change and as a result the velocity of the wave modes will vary. Hence, in the Hotscan project the effect of temperature on wave propagation was modelled to understand how the selected wave modes behave at elevated temperatures. The optimum wave mode selected was the torsional T(0,1) wave mode due to its non-dispersive nature.
Based on the modelling results a two ring torsional transducer collar array was designed to generate the desired T(0,1) wave mode. The collar was designed for an 8 inch (OD = 219.1mm and WT = 8.18mm) pipe as it is a commonly used pipe size. The collar array was designed using high temperature materials and unlike the commercial systems, which are used for inspection, this collar was designed for permanent installation and consisted of a main stainless steel frame and modules that were populated with transducers. The modules have a spring mechanism that could be calibrated to a desired load to allow the transducers to be coupled to the pipe. This prototype collar array was compared with a commercial collar array and was shown to give comparable results.
However, the main challenge was to develop transducers that could withstand high temperatures. There are many types of transducers available for ultrasonic testing which can be categorised as contact or non-contact type. There are advantages for using non-contact type transducers for ultrasonic inspection at elevated temperatures because the transducers will not experience the same temperatures as the specimen under test. Electromagnetic Acoustic Transducers (EMAT) and laser ultrasonics are two of the common non-contact techniques used for NDT at elevated temperatures. However, due to issues such as access, instrumentation sensitivity, testing costs and inspection resolution that are associated with these techniques, contact type piezoelectric transducers were investigated instead.
The problem often associated with the contact type piezoelectric transducers at high temperature is their Curie temperature. When piezoelectric materials operate above their Curie temperature they cease to be piezoelectric. Many piezoelectric materials that possess high Curie temperatures have been reported. A common problem associated with materials that poses high Curie temperatures is their piezoelectric response, which is generally low in comparison to Lead Zirconate Titanate, Pb(Zr,Ti)O3 (PZT), ceramics which is the most widely used piezoelectric material for conventional ultrasound NDT transducers.
Work has been carried out by researchers on new materials such as aluminium nitride thin films and also new fabrication techniques such as sol-gel deposited piezoelectric films. However, Lithium Niobate, LiNbO3, is a commercially available piezoelectric material that has Curie temperature of 1210°C, and Baba et al have shown that it can work at temperatures up to 1000°C. Therefore, in this project Lithium Niobate (LN) was selected as the most suitable piezoelectric material for further investigation. As well as LN transducers existing PZT type transducers were also further studied and improvements were made to extend their operating temperatures and were shown to work successfully up to 250°C. Prototype LN transducers were manufactured and tested ultrasonically for temperatures up to 600°C. The LN type transducers were expensive and time consuming to manufacture due to the cost of components and the high temperature joining technique involved. Therefore, for the laboratory trials high temperature testing of the transducer collar array the collar array was populated with less expensive high temperature PZT type transducers.
A prototype ultrasonic pulser-receiver (UPR) able to drive the high temperature collar arrays was developed to drive the transducer collar array at ambient and elevated temperatures. In its current form, the prototype UPR system can drive up to three collars, but is of a modular design which can be extended to drive more collars by adding more units. A 3dB (22%) improvement in signal to noise ratio was obtained when the prototype UPR was used in combination with the prototype high temperature transducer collar array when compared to existing commercially available inspection hardware. The UPR was controlled via flexible windows based software, which can interface with (for example) MATLAB to allow pre- and post-processing techniques to be incorporated. The electronics were designed to withstand the environmental conditions present in power stations.
The UPR was integrated with the high temperature collar array and used to collect data for development of the signal processing routines and the pattern recognition algorithms. Laboratory experiments were carried out using an 8 inch (OD = 219.1mm and WT = 8.18mm) representative steel pipe, P91 type, with a weld. UGW testing was carried out at ambient and then during the rise in temperature. Once the system reached the 250°C target temperature it was left to stabilise and baseline that represented a defect-free condition at elevated temperature was collected. Subsequently, a defect was introduced in the Heat Affected Zone (HAZ) of the weld and data was collected again, then the defect size was gradually increased, and data was collected at each defect size. This data was then used to perform advanced signal processing in order to detect the defect and also to monitor the increase in defect size. Advanced signal processing techniques such as normalization, Hilbert transform, signal smoothing, correlation, feature extraction, selection and classification based on Support Vector Machines were used. The signals were also used for the task of crack detection using the correlation-based pre-processing analysis. The proposed flaw detection technique was able to (i) identify with 100% confidence probability if a signal belonged to the baseline or not and (ii) identify the severity of the recognized defect with an average accuracy of 93.84%. An efficient defect mapping tool was developed that could be utilized on-line to provide a very useful real-time representation of the structural health of the pipe and/or to estimate the distance (in meters) of the recognized defects from the measurement unit. Trend analysis was conducted on the acquired signals, providing the relation between the peak-to-peak amplitude value of the ultrasonic signals and the progression of the defect size. A user-friendly Graphical User Interface (GUI) was developed to integrate all the proposed technologies (pre-processing, feature extraction/selection and pattern recognition) and allow NDT personnel limited experience to be able to use and exploit the Hotscan hardware/software functionalities.
The integrated Hotscan system was then installed in an electrical power plant to assess its performance in service conditions. The site trial was carried out in an Iberdrola coal power plant in Spain. The system sub-components were tested prior to shipment. The system was installed during the outage period and left to automatically collected data five times per day and record the temperature at the same time. Room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed. Data was collected during the rise in pipe temperature up to 339°C, but the pipe temperature exceeded the specified operation temperature of the system and hence it stopped working.
The Hotscan project has enabled a number of SME’s together with experienced RTDs with extensive expertise of innovation and research to define and explore a novel concept. The Hotscan system has the potential of providing benefit to the participants, to the power and oil and Gas industry in terms of safety management, the EU in terms of new Jobs and society through improved safety. Further work needs to be carried out to realise a commercial system. Although initial field trials have proven the concept further site trials are necessary to convince plant operators. The consortium will maintain communications after the project, and will review and exploit all potential commercialisation opportunities.
Project Results:
The initial stage of the Hotscan project was concentrated in reviewing the latest developments in UGW inspection and condition monitoring systems for structural health monitoring of high temperature pipelines in power and oil and gas sector. In order to clarify the industrial requirements and focus on the key objectives of the project a questionnaire was prepared and circulated to SME and end-user consortium members. The feedback from the questionnaire was incorporated in the project specification document (deliverable D1.2). In this document the following key aspects of the Hotscan system were addressed in detail:
• Propagation of ultrasonic guided waves (UGW) in pipes at high temperatures
• Transducer development
• Transducer collar design
• Development of a pulser-receiver system for serving multiple transducer collars
• Signal processing for trend analysis and noise reduction
The consortium was in agreement that development of high temperature transducers was the most challenging aspect of the research. The ultimate aim of the Hotscan project was to develop transducers that could withstand temperatures up to 580°C in order to meet the requirement for in-situ condition monitoring of the ageing electrical power plants. However, transducers that could withstand temperatures up to 250°C were considered to be commercially attractive in the oil and gas industry. Owing to the significant risk of 580°C transducers not being feasible, a staged approach was adopted to prove some of the development strategies using the PZT transducers. Therefore, a parallel programme of work was adopted to develop transducers for operation at 250°C and 580°C.
A range of samples for transducer development and laboratory validation of the overall system were considered as part of deliverable D1.1. The sample used to evaluated the high temperature ultrasonic performance of transducers was a square (12mm2) steel rod. For laboratory testing of the Hotscan system a number of pipe sized were considered, but it was agreed that 6 and 8 inch pipes were most commonly used. Therefore, a 6 metre long, 8 inch (OD = 219.1mm and WT = 8.18mm) representative P91 steel (CrMoV) pipe sample with a weld was produced (Figure 1). A sample weld was produced using the P91 steel pipe. The entire sample weld was removed from the pipe and a section of it was polished and examined using optical microscopy to identify the Heat Affected Zone (HAZ) of the weld (Figure 2). In order to simulated a growing detect in the HAZ, which is where the defects normally initiate, a saw cut was introduced and its size was gradually increased. Other pipes with various types of defects such as circumferential slots, fatigue and corrosion were also identified.
Figure 1 Shows the P91 steel pipe with a weld.
Figure 2 Shows a) cross section of the weld and b) the length of the HAZ.
Change in temperature can expand and contract the pipes geometry and also influence the material properties. This can affect how ultrasound propagates on a solid media at different temperatures. Therefore, the effect of temperature on propagation of ultrasound in pipes was studied as part of WP2 and reported in D2.1. The main objectives were to study the feasibility of defect detection using UGW, selection of optimum wave modes and feasibility of detection of creep type defects in the HAZ of the weld. High temperature mechanical properties such as Young's modulus, density and wave speeds (both shear and bulk) from literature were used in the ABAQUS model. Wave propagation simulations were performed for a 6 and 8 inch pipe (Figure 3).
Figure 3 Shows an example of wave propagation simulations performed.
Dispersion curves were calculated for the 8 inch (OD = 219.1mm and WT = 8.18mm) steel pipe at different temperatures using high temperature material properties reported in literature (Figure 4). The general trend observed was a decrease in velocity of the wave modes at elevated temperatures. Transient analyses of defect detection were also performed at ambient and elevated temperatures. In one model a weld with HAZ was simulated to study the possibility of detecting creep. It was not possible to identify creep in the primary or secondary phase. Only using a significant variation of the Young Modulus (around 50%), a measurable echo could be reproduced representing the area of creep. Regarding wave mode selection, the optimum wave mode selected was the torsional T(0,1) wave mode due to its non-dispersive nature. This information was useful for design of the transducer collar array.
Figure 4 Shows the dispersion curves for the 8 inch pipe at different temperatures.
A prototype high temperature transducer collar array (Figure 5) was developed as part of WP3 and reported in deliverable D3.1. In this report the existing state-of-the art long range ultrasonic system was studied and the high temperature limitations of its subcomponents, especially the transducers, were reviewed. Alternative high temperature materials were selected, and transducers for for operation at 250˚C and 580˚C were developed (Figure 6). An experimental procedure was designed to test the prototype transducers that were developed. Initially the existing transducers were tested, and were shown to work near the Curie temperature of the piezoelectric material (350˚C), but with a reduced ultrasonic performance (Figure 7). The Ageing behaviour was also investigated at 250°C and was shown to work successfully for up to 35 days (Figure 8). The prototype transducers developed for operation at 250˚C were assembled in the same manner as the existing transducers, but high temperature adhesives were used instead. This resulted in improved ultrasonic performance at elevated temperatures. These transducer were able to withstand operation at 250˚C, and were simple and cost effective to manufacture. Prototype transducers for operation at 580˚C was designed, manufactured and tested and shown to work at these temperatures (Figure 9). These transducers were tested from 450°C to 580°C for up to 12 days (Figure 10). These transducers were manufactured using commercially available Lithium Niobate (LN) piezoelectric elements which were bonded to the backing block using a high temperature joining technique. Therefore, the LN type transducers were expensive and time consuming to manufacture. An initial prototype collar array to house the transducers was developed. Improved materials used in design of this collar were selected to withstand the target temperatures. The collar array was manufactured and demonstrated to work at ambient temperatures as well as the exiting collars. Additional work was carried out to characterise candidate piezoelectric materials at elevated temperatures via impedance measurements.
Figure 5 Prototype high temperature transducer collar array mounted on an 8 inch steel pipe.
Figure 6 Shows the PZT and LN type transducers designed for operation at a) 250°C and b) 580°C, respectively.
Figure 7 Shows average transmission and reception responses of ten HT and LT PZT transducers from RT to approximately 350°C.
Figure 8 Shows the transmission and reception quality of the transducer at 250°C for up to 35 days.
Figure 9 Shows the fastest arriving wave mode in (left) reception and (right) transmission mode for the LN transducer at 20°C and 600°C.
Figure 10 Effect of Ageing on the Transmission Quality of LN Transducer
A prototype ultrasonic pulser-receiver (UPR) able to drive multiple high temperature collar arrays was developed as part of WP4 and reported in deliverable D4.1. In its current form, the prototype system (Figure 11) can drive up to three collars, but is of a modular design which can be extended to drive more collars by adding more units. There is also the possibility of multiplexing additional channels if necessary. Improved signal to noise ratio can be achieved using signal averaging, increasing the gain and using filters. A 3dB (22%) improvement in signal to noise ratio was obtained when the prototype ultrasonic pulser-receiver was used in combination with the prototype high temperature transducer collar array when compared to existing commercially available inspection hardware. The UPR is controlled via flexible windows based software, which can interface with (for example) MATLAB to allow pre- and post-processing techniques developed elsewhere in the project to be incorporated. The electronics were designed to withstand the environmental conditions present in power stations.
Figure 11 Shows the prototype Hotscan UPR system.
In WP5 advanced signal processing routines were developed for continuous condition monitoring and trend analysis. There were two deliverables in this WP. In the first deliverable D5.1 software for signal pre-processing and signal focusing using Time-Reversal Focusing (TRF) and Time-Delay Focusing (TDF) techniques was developed. The implemented software was developed under the MATLAB programming environment. The developed software routines were integrated in a simple GUI (Figure 12) which allows signal visualization to validate the required functionality. In the second deliverable D5.2 advanced signal processing techniques were developed and integrated, such as normalization, Hilbert transform, signal smoothing, correlation, feature extraction, selection and classification based on Support Vector Machines. For the training and validation of the system, an extensive experimental investigation was carried to generate defect-free and defective signals to develop and test the signal processing algorithms, the experimental set-up can be seen in Figure 13. The results (Figure 14) show that the proposed method is able to effectively detect flaws under various conditions and experimental scenarios in laboratory conditions.
Figure 12 The Hotscan GUI.
Figure 13 Image showing the test set-up used for high temperature data collection.
Figure 14 Trend analysis based on the most informative features that were selected by the proposed feature selection approach
The main components of the Hotscan system such are the transducer collar array, pulser-receiver unit and the signal processing software which were integrated in WP6 and reported in deliverable D6.1. In this report it was shown that; the prototype transducers were able to function at 250°C and 580°C, the high temperature collar array was able to generate the desired wave modes and operate in identical manner to the commercial collar array system, a pulser-receiver that can drive multiple collars was produced and finally a pattern recognition and trend analysis software that could monitor crack propagation and display it in a GUI was also developed. In D6.2 report the laboratory testing and results of the Hotscan system in ambient and elevated temperatures were presented. The main subcomponents such as the transducers were individually tested at 250°C and 580°C. The high temperature collar array was populated with the high temperature PZT type transducers, and used with the integrated Hotscan system to perform ultrasonic guided wave testing. The guided wave testing was performed on an 8 inch (OD = 219.1mm and WT = 8.18mm) Steel (P91) pipe at up to 250°C in laboratory conditions, as shown in Figure 12.
The integrated Hotscan system was installed in an electrical power plant as part of WP7 and results of this trial was reported in deliverable D7.1. The site trial was carried out in an Iberdrola coal power plant in Spain (Figure 15). The system sub-components were tested prior to shipment and then again on site prior to installation (Figure 16). The system was installed during the outage period and left to automatically collected data five times per day and record the temperature at the same time (Figures 17 and 18). Room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed (Figure 19). Data was collected during the rise in temperature from 28°C to 339°C. The system stopped working when the pipe temperature reached 365°C, beyond the specified operating temperature of the transducers.
Figure 15 Location of the Hotscan site trials in an Iberdrola coal power plant located in Langreo, Austrias, Spain.
Figure 16 Testing and assembly of the collar array.
Figure 17 Shows a) the installation and b) the assembled collar on the pipe.
Figure 18 Installed Hotscan system automatically collecting data.
Figure 19 Weld reflection for all of the 27kHz signals collected at room temperature.
A detailed breakdown of the work performed and main results for each WP is shown in the following table.
WP no. Work progress
1 System functional design, hardware and software architecture to meet industrial requirements and provision of samples
The main objectives if this WP were to confirm the industrial requirements that the Hotscan system was intended to solve and to then use this information and compose a specification document which included the functional design and the architecture of the hardware and software components. Therefore, a questionnaire was prepared and circulated to SME and end-user consortium members. The feedback from the questionnaire highlighted the dimension, type of material and the temperatures if the pipes used as well as the types of defects that normally occur. This information was incorporated in the project specification document (deliverable D1.2). In this document the following key aspects of the Hotscan system were also addressed:
• Propagation of ultrasonic guided waves (UGW) in pipes at high temperatures - Developing high temperature transducers was very important, but this was not sufficient when performing LRUT inspection at high temperatures. In order to interpret the signals during inspection, understanding of propagation characteristics of UGWs is essential. When performing conventional LRUT on pipes, material properties such as Young’s modulus, poisons ratio and density, and pipe geometry (thickness and diameter) are important as they influence the velocity of the UGWs. As a result the use of various software such as Disperse and FEA modelling techniques were considered to produce dispersion curves for pipes at elevated temperatures.
• Transducer development - The consortium was in agreement that development of high temperature transducers was the most challenging aspect of the research. The ultimate aim of the Hotscan project was to develop transducers that could withstand temperatures up to 580°C in order to meet the requirement for in-situ condition monitoring of the ageing electrical power plants. However, transducers that could withstand temperatures up to 250°C were considered to be commercially attractive in the oil and gas industry. Therefore, a parallel programme of work was adopted to develop transducers for operation at 250°C and 580°C. The pros and cons of contact and non-contact type transducers were considered and it was decided to opt for piezoelectric contact type transducers. The components that make up the piezoelectric transducers such as the piezoelectric element, transducers housing/backing material, high temperature cables, wear plate and the possible joining techniques were discussed. Lithium Niobate was selected as the most suitable piezoelectric element for further investigation due to its high Curie temperature, which is reported to be approximately 1150°C for certain compositions
• Transducer collar design - Regarding the collar array, the modular collar design of existing LRUT system (Teletest®) was considered most appropriate to for generation of the desired wave modes. Functionality and components of the existing collar design were considered unsuitable for use at elevated temperatures as it's inflatable design meant that the bladder and modules were manufactured using polymers and composite materials that had low melting temperatures. Therefore, it was decided to manufacture a high temperature collar array using metal components.
• Development of a pulser-receiver system for serving multiple transducer collars - The pulser-receiver system was also based on the current Teletest© system, but iterative developments were to be carried out to build a system with enhanced signal to noise ratio capability. To collect data using the Hotscan pulser-receiver it was decided to use the standard Teletest© software (version 1). Additional graphical user interfaces (GUI) were to be designed specific to the Hotscan application to aid data collection, analysis and presentation.
• Signal processing for trend analysis and noise reduction - The Hotscan system was designed to continuously collecting data, which was then going to be processed using advanced signal processing routines and pattern recognition methods in MATLAB. Key features from the received signals were to be selected, extracted and classified. To build the software algorithms representative signal were required, hence an experimental procedure was designed to simulate defect-free and defective signals for an 8 inch pipe in laboratory conditions.
A range of samples for transducer development and laboratory validation of the overall system were considered as part of deliverable D1.1. The sample used to evaluated the high temperature ultrasonic performance of transducers was a 12mm2 steel rod. For laboratory testing of the Hotscan system a number of pipe sized were considered, but it was agreed that 6 and 8 inch pipes were most commonly used. Therefore, a 6 metre long, 8 inch (OD = 219.1mm and WT = 8.18mm) representative P91 steel (CrMoV) pipe sample with a weld was produced (Figure1). A sample weld was produced using the P91 steel pipe. The entire sample weld was removed from the pipe and a section of it was polished and examined using optical microscopy to identify the Heat Affected Zone (HAZ) of the weld (Figure 2). In order to simulated a growing detect in the HAZ, which is where the defects normally initiate, a saw cut was introduced and its size was gradually increased. Other pipes with various types of defects such as circumferential slots, fatigue and corrosion were also identified.
2 Modelling of guided wave propagation in pipes at normal and elevated temperatures for optimum mode selection
The main objectives were to study the feasibility of defect detection using UGW, selection of optimum wave modes and feasibility of detection of creep type defects in the HAZ of the weld. Temperature can expand and contract the pipes geometry and also influence the material properties. This can affect how ultrasound propagates on a solid media at different temperatures. Therefore, the effect of temperature on propagation of ultrasound in pipes was studied (Figure 3) as part of this WP and reported in D2.1.
Dispersion curves plot the phase speed of an elastic wave travelling through a specific waveguide against frequency. Dispersion curves are important for LRUT to determine the frequencies which are within a range where the wave modes become “dispersive”. This is represented in the dispersion curves by a high gradient for the specific frequency. Dispersion curves are used to select an optimal frequency for a given wave mode as less “dispersive” as possible which can then be used for selective excitation. Dispersion curves can be created by solving the analytical equations or by solving Eigen-Problems using numerical methods. PCdisp (Pochhammer-Chree dispersion), a small toolbox of MATLAB routines which uses a numerical simulation the generate propagation of ultrasonic waves in a cylindrical waveguide, was mainly used to produce the dispersion curves. Calculations of dispersion curves for three temperature ranges were carried out to allow for a better overview of the material's behaviour to ultrasonic waves due to the temperature change. To match available material data from open literature with the target temperatures of interest instead of 250°C and 580°C higher temperatures of 260oC and 595oC were used, respectively (Figure 4), and for ambient temperature 21°C was assumed. Dispersion curves were calculated for the 8 inch (OD = 219.1mm and WT = 8.18mm) steel pipe at these temperatures. The general trend observed was a decrease in velocity of the wave modes at elevated temperatures. For example for a frequency of 70 kHz the phase velocity decreases 20% when increasing the temperature from 21°C to 595°C.
Theoretical limits of defect detection using UGW to detect defects in pipes were calculated for L(0,1) and T(0,1) wave modes at different frequencies. it was concluded that even at 250kHz (upper range of UGW testing) it is theoretically not possible to detect a crack size of 1mm x 1mm in inspection mode.
Transient analyses of defect detection were also performed at ambient and elevated temperatures. Simulations that took into account only
the weld and not the HAZ were produced in Abaqus. Amplitude of the reflections from the crack, interaction of the reflected waves with the weld makes defect detection difficult when amplitude was used as a detection feature. As reference a crack of 10 degrees (of circumference = 19mm length) 1mm wide, 4.16mm deep (half thickness) was used. Simulations targeted to explore the effect variation of the geometrical characteristics of the crack (width, depth ) and the frequency of the ultrasonic waves would have to the response levels. It was shown that the reference defect is on the limits of detection using reflected amplitude as the detection feature. The modelling has shown that the targeted defect size of 1mm x 1mm as stated in the DoW cannot be detected.
Thereafter a feature was introduced to represent the reflection of a weld, and a crack was introduced in the immediate vicinity of the weld. The target was to identify the response (reflection) from the crack, despite the presence of the weld response. It has been demonstrated that under circumstances there is a distinguishable signal. Nevertheless, it has to be taken into account that field measurements will have additional noise (both due to electronics, as well as imperfections of the structure, material etc); those are general much higher than those due to numerical calculation. It was also mentioned that positive identification of defects can be enhanced by comparing data collected at previous inspections to current one to identify changes (using them as a possible flaw indication).
In one model a weld simulated to study the possibility of detecting creep. It was not possible to identify creep in the primary or secondary phase. Only using a significant variation of the Young Modulus (around 50%), a measurable echo could be reproduced representing the area of creep. Regarding wave mode selection, the optimum wave mode selected was the torsional T(0,1) wave mode due to its non-dispersive nature. This information was useful for design of the transducer collar array.
3 Modelling, design and prototyping of the high temperature LRU transducer collar array
The main objective of this WP was to develop a LRUT transducer collar array (Figure 5) that could be permanently air coupled to high temperature pipe work, temperatures up to 580°C and could also generate the desired wave modes selected in WP2. A prototype high temperature transducer collar array was developed as part of this WP and reported in deliverable D3.1. In this report the existing state-of-the art long range ultrasonic system was studied and the high temperature limitations of its subcomponents, especially the transducers, were reviewed. Alternative high temperature materials were selected, and transducers for operation at 250˚C and 580˚C were developed (Figure 6). Additional work was carried out to characterise candidate piezoelectric materials at elevated temperatures via impedance measurements.
Limitations of the existing transducers and their main sub-components were investigated. An experimental procedure was designed to test the prototype transducers that were developed. Initially the existing transducers were tested, and were shown to work near the Curie temperature of the piezoelectric material (350˚C), but with a reduced ultrasonic performance. The prototype transducers developed for operation at 250˚C were assembled in the same manner as the existing transducers, but high temperature adhesives were used instead. This resulted in improved ultrasonic performance at elevated temperatures. These transducer were able to withstand operation at 250˚C (Figure 7), and were simple and cost effective to manufacture. These transducers tested at 250°C for up to 35 days (Figure 8) and it was observed that their ultrasonic output increased significantly during the rise in temperature from ambient to 250°C and then stabilised.
Prototype transducers for operation at 580˚C was designed, manufactured and tested and shown to work successfully up to 600°C (Figure 9). The ageing behaviour of these transducers was tested and shown to work from 45°C to 580°C for 12 days (Figure 10). These transducers were manufactured using commercially available Lithium Niobate (LN) piezoelectric elements which were bonded to the backing block using a high temperature joining technique. To select the suitable material and size of the backing block modelling work was performed using ABAQUS. In this model a range of metals and ceramics were considered and eventually Aluminium oxide was selected as it had one of the closest matched acoustic impedances to LN as well as a fairly close value of CTE. The CTE was not a perfect match to LN, but it had a high Young’s Modulus which would allow the material to withstand greater stresses. Aluminium oxide is denser than steel, which along with the closer acoustic impedance made it a much more efficient dissipater of acoustic energy than steel. Aluminium oxide is a widely used ceramic and is cheaper compared to other advanced ceramics.
A new transducer collar array was designed to withstand high temperatures. The new collar array was manufactured using mainly stainless steel, which has a melting temperature of 1510˚C, and was designed to be fitted onto an 8 inch (OD = 219.1mm and WT = 8.18mm) steel pipe. The collar consists of two rings of transducers to enable operation in torsional mode. The key components in the collar are the frame and the modules. The frame is made up of two halves which are joined together using four clamp blocks. There are 24 equally spaced slots where the modules were mounted. There were 24 modules and each had two transducers which were aligned in the torsional orientation, and behind each transducer was a steel compression spring. The springs were calibrated to exert the required force when fitted to pipe, the springs were made from stainless steel (melting temperature of 1325-1530˚C) and were designed to withstand temperatures up to 250˚C with minimal change in the spring constant. For operation at higher temperatures springs made from other materials such as Tungsten could be used. To apply the force onto the pipe via the transducer would require unscrewing the cap screw above each transducer. This is a more time consuming method of coupling the transducers to the pipe compared with the inflatable collar, but it ensures that uniform pressure can be applied at the elevated temperatures, as polymeric materials are not required. The Ceramic block in the modules was designed to steer the transducers to their position and keep them at a fixed position. A ceramic material was used so that it could also act as a heat barrier to protect the cables and connectors.
The collar array was manufactured and demonstrated to work at ambient temperatures as well as the exiting collars. Initial trials were carried out in ambient conditions to compare the performance of the High Temperature (HT) transducer collar array with a commercially specified system. Both collars were used to perform UGW testing on a defect-free steel pipe. The collars were placed at the same location (2m from the pipe end), and the inspection was performed at 37kHz. The A-scans obtained for the two systems were compared. It could be seen that the high temperature transducer collar array performed as well as the commercial system, since the A-scan traces were almost identical. The prototype collar array was tested at elevated temperatures on an 8 inch pipe using the Hotscan system. The experimental set-up and dimensions of the pipe used can be seen in the deliverable report (D3.1).
According to the decision at the kick-off meeting the system was first tested at 250˚C. A script was used to automatically collect data in a fast manner at each temperature. This was carried out to avoid major temperature fluctuation during the data collection process. The experimental set-up can be seen in the deliverable report (D3.1). The collar was heated to the required temperatures using heating mats, which were placed on either side of the collar. The heating mats were covered with high temperature insulating materials to ensure the pipe reached the required temperatures faster. The heating mats and a thermocouple (resistance welded to the pipe) were connected a CooperHeat power supply and temperature regulator unit. Other thermocouples were used to verify the temperature of the pipe. The collar was then connected to the Hotscan pulser-receiver unit and operated using a laptop. The script written specifically for this data collection was used to collect the data. Data was collected at room temperature then in 50˚C steps between 50˚C and 250˚C. The results showed that the collar array was working up to 250°C. The pipe end signal amplitude decreases as temperature increases. This will make defect detection more difficult. However, when the signal to noise ratio was measured at the different temperatures it could be seen that there was not a significant decline in signal to noise ratio with respect to temperature. As a result it should still be possible to detect defects at elevated temperatures.
4 Development of prototype high power pulser-receiver for high signal attenuation conditions
The main objective if this WP was to develop a pulser-receiver unit (Figure 11) that could be permanently installed in power plants to drive the transducer collar array developed in WP3. A prototype ultrasonic pulser-receiver (UPR) able to drive multiple high temperature collar arrays was developed as part of this WP and reported in deliverable D4.1. Inside the UPR there are 24 transmission (Tx) and 8 reception (Rx) channels, which can be routed to a combination of transducers in one or more collar, as appropriate. The Rx channels are multiplexed sequentially between transducers as required to receive the transmitted signals, which is more cost effective than having 24 individual Rx channels. All of the channels are connected to the control unit which is connected to a Windows PC. The control unit is a FPGA chip that has been programmed to manage the Tx/Rx channels and to collect and communicate the data to the PC.
The UPR can be connected to mains power supply for constant use in condition monitoring mode, but it also has a rechargeable battery that can last up to eight hours. This has two benefits. Firstly, redundancy is added, as the battery acts as an uninterruptable power supply system such that the hardware will not fail or require resetting/maintenence in the event of a power failure. Secondly, it allows the UPR to function in inspection mode if required, making the device multifunctional and more widely useful.
The UPR will need to be permanently installed in power plants within 30m of the pipes under test. The UPR and the laptop running the software developed in WP5 will be housed inside a rack, which acts as a protective casing. The rack can be mounted on a wall or kept in a suitable location in the power plant. The rack can be locked and only authorised personnel can access the system to access the data and maintain the system. If necessary, fan trays can also be installed to protect the laptop and the UPR electronics from overheating. This will be assessed during the field trials and if necessary will be installed. As the software is Microsoft Windows, it could be controlled remotely if a suitable network infrastructure is available, but it is envisaged that the UPR will only ever be controlled ‘locally’ by the laptop within the cabinet.
In its current form, the prototype system can drive up to three collars, but is of a modular design which can be extended to drive more collars by adding more units. There is also the possibility of multiplexing additional channels if necessary. Improved signal to noise ratio can be achieved using signal averaging, increasing the gain and using filters. A 3dB (22%) improvement in signal to noise ratio was obtained when the prototype ultrasonic pulser-receiver was used in combination with the prototype high temperature transducer collar array when compared to existing commercially available inspection hardware. The UPR is controlled via flexible windows based software, which can interface with (for example) MATLAB to allow pre- and post-processing techniques developed elsewhere in the project to be incorporated. The electronics were designed to withstand the environmental conditions present in power stations.
5 Advanced signal processing techniques for continuous monitoring, trend analysis and crack image recognition
In this WP the main objective was to develop advanced signal processing routines for continuous condition monitoring and trend analysis. There were two deliverables in this WP. In the first deliverable D5.1 a software for signal pre-processing and signal focusing using Time-Reversal Focusing (TRF) and Time-Delay Focusing (TDF) techniques was developed. The implemented software was developed under the MATLAB programming environment. The developed software routines were integrated in a simple GUI (Figure 12) which allows signal visualization to validate the required functionality.
In the second deliverable D5.2 advanced signal processing techniques were developed and integrated, such as normalization, Hilbert transform, signal smoothing, correlation, feature extraction, selection and classification based on Support Vector Machines. For the training and validation of the system, an extensive experimental investigation was carried (Figure 13) to generate defect-free and defective signals to develop and test the signal processing algorithms. A novel defect detection system was developed based on a successful synergy of complementary advanced processing tools (Figure 14). The Hotscan system can be efficiency utilized as an integrated condition monitoring system based on a modular design, which enables the accurate and reliable evaluation of the structural condition of pipes. The technology provides information of the condition of the pipe checking for any degradation or changes over time. As the defects of interest tend to occur over sustained periods of time, the monitoring may only need to be conducted intermittently, perhaps weekly or monthly.
The validity of the system was tested over an extensive experimental setup. An overview of the detection capabilities of the proposed system is given in the following:
• Experiments were conducted at ambient temperatures for the task of crack detection using data from separate channels and different frequencies. The proposed flaw detection was capable to identify with 100% confidence probability if a testing signal belongs to the baseline or not.
• Trend analysis was conducted on the acquired signals at ambient temperature, providing the relation between the peak value of the ultrasonic signals and the progression of the defect size.
• Experiments were conducted at elevated temperatures for the task of crack detection using the correlation-based pre-processing analysis applied in all the channels. The proposed flaw detection was capable to (i) identify with 100% confidence probability if a testing signal belongs to the baseline or not and (ii) identify the severity of the recognized defect with an average accuracy of 93.84%.
• An efficient defect mapping tool was developed that can be utilized on-line (i) providing a very useful real-time representation of the structural health of the pipe and (ii) estimating the distance (in meters) of the recognized defects from the measurement unit.
• Trend analysis was also conducted on the acquired signals at ambient temperature providing a clear straight-forward relation between extracted informative features of the ultrasonic signals and the progression of the defect size.
• A user-friendly Graphical User Interface (GUI) was developed that integrates all the proposed technologies (pre-processing, feature extraction/selection and pattern recognition) and allows anybody with limited experience to be able to use and exploit the HOTSCAN hardware/software functionalities.
6 Prototype Hotscan integration and laboratory validation
The main objective of this WP was to integrate the hardware and software components of the Hotscan system. The main components of the Hotscan system such are the transducer collar array, pulser-receiver unit and the signal processing software which were integrated in this WP and reported in deliverable D6.1. Majority of the system components were developed, tested and reported in previous deliverables . Therefore, in this report the functionality of the overall Hotscan system was reported, and the main conclusions of the report were as follows:
• Transducers that can function at 250°C and 580°C have been developed and tested individually in laboratory conditions.
• High temperature collar array that can generate the desired wave modes and operate in almost identical manner to the commercial collar array has been developed and tested at ambient and up to 250°C.
• A pulser-receiver that can drive the transducer collar array has been developed and used to test the transducer collar array at ambient and high temperatures.
• Pattern recognition and trend analysis software that can monitor the defect propagation and display it in a GUI has been developed.
• The hardware and software components have been integrated.
In D6.2 results of the Laboratory trials carried out at ambient and high temperatures on the Hotscan system and its subcomponents were presented. The transducers developed for operation at 250°C using Lead Zirconate Titanate as the piezoelectric material, and transducers developed for operation at 580°C using Lithium Niobate, were tested at laboratory conditions from room temperature up to the target temperatures. The PZT type transducers were then used with the collar array to perform Ultrasonic Guided Wave testing on a pipe sample from ambient up to 250°C.
The main conclusions of this report were as follows:
• The PZT and LN type transducers have been tested in laboratory conditions from ambient up to 250°C and 580°C, respectively. Both type of transducers were able to transmit and receive UGW signals at the target temperatures.
• Due to cost of the manufacture of LN transducers the collar array was populated with the less expensive PZT type transducers.
• The integrated Hotscan system was tested in laboratory conditions using PZT type transducers from ambient up to 250°C.
• The effect of thermal cycling on the Hotscan system was studied for up to 4 thermal cycles at 250°C. A 20% drop in performance of the PZT transducers was observed after the first cycle, but the performance stabilised after the subsequent thermal cycles.
7 Demonstration
The main objective of this WP was to install the Hotscan system on a steam line in a power plant (Figure 15) and assess its performance during rise in temperature and then over time at elevated temperatures. The integrated Hotscan system was installed in an electrical power plant as part of this WP and results of this trial was reported in deliverable D7.1.
In previous deliverables (D6.1 and D6.2) the integrated Hotscan system was shown to work in ambient temperatures, and its performance was demonstrated for operation at elevated temperatures in laboratory conditions. It was shown that the system was able to detect defects at 250°C on a representative P91 pipe sample. Transducers that can operate at 580°C have been developed and shown to work in laboratory conditions. However, in this project it was not cost effective to manufacture 580°C transducers at the scale required for site trials. In addition it would have taken a significant amount of time to source the components, manufacture and test these transducers, which would have further delayed the project. At the project review meeting (Month 24) the case for performing site trials at 250°C instead of 580°C was presented to the European Commission’s independent project reviewer Dr Michael Scheerer. He was in agreement and recommended the field trials be carried out at ~ 250C. As a result the consortium tried to find a suitable power plant with pipes operating at around 250C. When selecting the most suitable location for the site trials the following criteria had to be considered, which further narrowed down the search for potential sites where the system could be installed.
• Installation during an outage for safety reasons.
• Outage taking place during lifetime of the project.
• Identifying an 8 inch pipe as a single prototype collar was manufactured.
• Pipe location had to be indoors to protect the pulser-receiver and rest of the system and also somewhere that did not require erection of additional scaffolding.
• Maximum pipe temperature of 250°C.
There were extensive discussions by the consortium members with various electrical power plant owners across Europe to enable the site trials to take place in their facilities. Dr Marko Budimir (INETEC) had been in contact with Krsko, a nuclear power plant in Slovenia. They were keen to assist with the site trials but they had very strict security checks which made it difficult. Marko subsequently spoke to Jertovec, a combined cycle power plant in Croatia. They were also keen to assist but they did not have any 8 inch pipes. A Turkish company called Tupras were approached by Dr Cem Selcuk (Brunel University). They wanted to be involved in the project and assist with the site trials, but their company policy required them to have a formal involvement in the project. This was discussed in a project review meeting (Month 18) and the consortium decided that Tupras could only be involved in an end user group. Abbas Mohimi (Brunel University) visited the EPR Ely, which is world’s largest straw fired power plant. They were interested in the Hotscan system, but their next planned outage was in July 2013, which was beyond the time scales of the Hotscan project. Rafael Delgado (Tecnitest) had been in contact with Iberdrola and they had a potential site suitable for the site trials. The Lada coal power plant, capacity of 513 MW, was going through an outage period, therefore the site trials were carried out at this location.
The system sub-components were tested prior to shipment and then again on site (Figure 16). Wall thickness measurements were performed at the location where the collar was going to be mounted using manual ultrasonics. The collar array components such as the transducers, wires and connectors were tested again and then assembled. The collar array was assembled in two halves and carefully fitted onto the pipe (Figure 17).
The assembly procedure was difficult due to the tight space available and required two people to perform this task. Mounting the existing collar whilst on scaffolding could be even more difficult. A more user friendly design is desirable for any future improvements. Once installed on the pipe the transducers were coupled by removing the screw caps from the back of the transducers. The transducers were then connected to the pulser-receiver using a tool lead, and a test was carried out to check that everything was functioning properly. A thermocouple (TC) was used to measure the ambient temperature, and four other TCs were placed around the pipe circumference at the cardinal points to measure the pipe temperature. The thermocouples were connected to the Pico temperature logger.
The wires and thermocouples were neatly assembled and the lagging material was placed either side of the collar and a high temperature resistant sheet was placed above the collar and secured using steel wires. A casing was placed near the installation and the pulser-receiver, laptop, Pico temperature logger, and an external hard drive were placed inside it. An extension lead was used to provide power to the pulser-receiver and the laptop. The Pico temperature logger was powered via the laptop. An automated data collection software was designed and installed in the laptop to collect ultrasonic and temperature measurements five times per day and store the data in the external hard drive. The software was set-up to generate the T(0,1) wave mode in pulse-echo mode from 20-200kHz frequencies at 1kHz steps, using a 10 cycle hann windowed input signal, and 32 averages. The software was tested twice and then left to collect data (Figure 18).
A second visit was arranged by Brunel University and Tecnitest on 11 July 2013 to inspect the system and collect the data by replacing the external hard drive and allowing the system to continue collecting data. The results showed that room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed (Figure 19).
The pipe temperature was not expected to exceed 250°C according to the plant operators. However, the data collected during the rise in temperature from 28°C to 339°C showed otherwise. The system stopped working when the pipe temperature reached 365°C due to the depolarisation of the transducers. Had the pipe temperature not exceeded the specified target temperature of 250°C the Hotscan system would have carried on collecting data.
8 Dissemination and exploitation
The main objectives of this WP were to disseminate the technology developed in the Hotscan project as widely as possible and particularly the EU community via marketing materials, attending conferences and publishing journal papers. The dissemination and exploitation activities were captured in this WP and reported in the Final PUDF (deliverable D8.2). The report sets out the details of the types of dissemination activities undertaken during the Project lifetime. It also captures all dissemination and exploitation activities which have been undertaken by the Project partners. This document also sets out the aspirations of the SME Partners for the commercial exploitation of the project foreground knowledge. It provides a complete picture of all dissemination activities undertaken and, most importantly, provides information on the future route to exploitation that will benefit the SME’s, industry, society and the EU economy through new job creation.
During the life of the project 2 peer-review journals papers have been published (one is currently under the review process) and 7 conference papers have been presented, as detailed in Tables A1 and A2.
The Hotscan website has been updated during lifetime of the project.
9 Project management
A progress review was carried out at half way stage (Month 9) of the programme and the findings were reported. Following the review process a number of deliverable reports were revised and re-submitted and the DoW was amended following the discussions with the independent project reviewer. All deliverables and progress report including management, finance and technical details were submitted for review at the ninth month stage, and the subsequent deliverables were also completed and submitted.
Despite the difficulty and high development risks in the Hotscan project, the consortium has made good progress towards meeting the milestones set out in the project Description of work.
Potential Impact:
It is anticipated that the Hotscan system has the potential to significantly change the nature of structural integrity assessment of high temperature pipe work in oil refineries and electrical power plant by enabling in-situ ultrasonic guided wave inspection and condition monitoring to be carried out. It is likely to contribute in reduction of failures in high temperature pipe work, especially in the ageing power plants. This may also help avoid unplanned outages and therefore significantly reduce the overall maintenance cost.
Project concept and some of the results have been published in peer reviewed journals and presented in various conferences (See Tables 1 and 2, respectively) by project partners.
The prototype system requires further development before it can be introduced to the market. The Hotscan partners, in particular the SMEs will continue to disseminate information about the project to the potential end users after the project has finished to secure funding for further development, this with be carried in collaboration between the SMEs as agreed in the Hotscan project.
List of Websites:
www.hotscan.eu
Relevant contact details:
Dr Paul Jackson, Plant Integrity Ltd
Tel: +44 (0) 1223 893994
Fax: +44 (0) 1223 893994
E-mail: paul.jackson@plantintegrity.co.uk
High temperatures and pressures in superheated steam pipes in nuclear power plants can lead to the formation of flaws and defects due to material degradation mechanisms such as corrosion, creep and fatigue over time. If undetected in advance, these critical features can cause catastrophic failures which in turn will result in significant environmental and financial consequences. Inspection using conventional Non-Destructive Testing (NDT) techniques such as visual testing, eddy current testing and ultrasonic testing are employed to assess the integrity of the steam pipelines during planned outages for nuclear power plant maintenance. However, if the condition of a pipeline is not monitored between outages, this may present a problem for an aging power plant as defects such as creep may become more prevalent. In situ condition monitoring techniques therefore need to be developed to retain reliability and extend the lifetime of nuclear power plants.
The Hotscan project concept was to develop a Long Range Ultrasonic Testing (LRUT) system for continuous in service inspection and structural health monitoring of high temperature superheated steam pipes in power generation plant with 100% coverage. Performing LRUT at elevated temperatures can be complicated. The effect of temperature on wave propagation was modelled to understand how the selected wave modes behave at elevated temperatures. The optimum wave mode selected was the Torsional T(0,1) wave mode. The main challenge was to develop transducers that could withstand high temperatures. Lithium Niobate (LN) was selected as the most suitable piezoelectric material for further investigation. Owing to the significant risk of 580°C transducers not being feasible, a staged approach was adopted to prove some of the development strategies using the PZT transducers. Existing PZT type transducers were also further studied and improvements were made to extend their operating temperatures and were shown to work successfully up to 250°C. Prototype LN transducers were manufactured and tested ultrasonically for temperatures up to 600°C. The ageing behaviour of both PZT and LN transducers was investigated and it was shown that the PZT transducers could successfully operate at 250°C for over 30 days and the LN transducers were tested up to 580°C for 12 days. A collar array was designed and manufactured using high temperature materials suitable for high temperature operation. Its performance was compared with a commercial collar array and was shown to give comparable results at ambient temperatures, and was also shown to work at elevated temperatures in combination with the ultrasonic pulser-receiver unit during the laboratory trials. The integrated Hotscan system was installed in an electrical power plant to assess its performance in service conditions. The site trial was carried out in an Iberdrola coal power plant in Spain. Room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed. Data was collected during the rise in pipe temperature up to 339°C, but the pipe temperature exceeded the specified operation temperature of the system and hence it stopped working. Further work needs to be carried out to realise a commercial system. Although initial field trials have proven the concept further site trials are necessary to convince plant operators. The consortium will maintain communications after the project, and will review and exploit all potential commercialisation opportunities.
The Hotscan project has enabled a number of SME’s together with experienced RTDs with extensive expertise of innovation and research to define and explore a novel concept. The Hotscan system has the potential of providing benefit to the participants, to the power and oil and Gas industry in terms of safety management, the EU in terms of new Jobs and society through improved safety.
Project Context and Objectives:
High temperatures and pressures in superheated steam pipes in power plants can lead to the formation of flaws and defects due to material degradation mechanisms such as corrosion, creep and fatigue over time. If undetected in advance, these critical features can cause catastrophic failures which in turn will result in significant environmental and financial consequences. Inspection using conventional Non-Destructive Testing (NDT) techniques such as visual testing, eddy current testing and ultrasonic testing are employed to assess the integrity of the steam pipelines during planned outages for nuclear power plant maintenance. However, if the condition of a pipeline is not monitored between outages, this may present a problem for an aging power plant as defects such as creep may become more prevalent. In situ condition monitoring techniques therefore need to be developed to maintain reliability and extend the lifetime of nuclear power plants.
The Hotscan project concept was to develop a Long Range Ultrasonic Testing (LRUT) system for continuous in service inspection and structural health monitoring of high temperature superheated steam pipes in power generation plant with 100% coverage. LRUT is a type of ultrasonic testing but it is fundamentally different as it uses lower frequencies, normally in the kHz range, and inspects over a range of metres as opposed to MHz frequencies and centimetre ranges of conventional ultrasonics. There are commercial LRUT systems for inspection at ambient temperatures, such as the Teletest® system developed by Plant Integrity Ltd. The system consists of a transducer collar array that is driven by an ultrasonic pulser-receiver and a laptop PC to display the data.
LRUT employs Ultrasonic Guided Waves (UGW) to inspect long length of pipe from a single location by using the pipe as a wave guide. The system can detect internal and external corrosion as it is sensitive to volumetric wall loss. It can reliably detect 9% pipe wall cross-section loss and using focusing techniques it can detect up to 3% cross-section loss. Sound energy propagates through a structure in specific types of vibration pattern, known as wave modes. Many wave modes are generated when the transducer rings are excited. The key wave modes that are used in LRUT of pipes are the longitudinal L(0,2) and torsional T(0,1) wave modes. During inspection conditions are chosen to ensure the wave has a constant velocity over the frequency range. Performing LRUT at elevated temperatures can be complicated. The pipe geometry is likely to expand or contract due temperature and material properties such as Young's modulus will also change and as a result the velocity of the wave modes will vary. Hence, in the Hotscan project the effect of temperature on wave propagation was modelled to understand how the selected wave modes behave at elevated temperatures. The optimum wave mode selected was the torsional T(0,1) wave mode due to its non-dispersive nature.
Based on the modelling results a two ring torsional transducer collar array was designed to generate the desired T(0,1) wave mode. The collar was designed for an 8 inch (OD = 219.1mm and WT = 8.18mm) pipe as it is a commonly used pipe size. The collar array was designed using high temperature materials and unlike the commercial systems, which are used for inspection, this collar was designed for permanent installation and consisted of a main stainless steel frame and modules that were populated with transducers. The modules have a spring mechanism that could be calibrated to a desired load to allow the transducers to be coupled to the pipe. This prototype collar array was compared with a commercial collar array and was shown to give comparable results.
However, the main challenge was to develop transducers that could withstand high temperatures. There are many types of transducers available for ultrasonic testing which can be categorised as contact or non-contact type. There are advantages for using non-contact type transducers for ultrasonic inspection at elevated temperatures because the transducers will not experience the same temperatures as the specimen under test. Electromagnetic Acoustic Transducers (EMAT) and laser ultrasonics are two of the common non-contact techniques used for NDT at elevated temperatures. However, due to issues such as access, instrumentation sensitivity, testing costs and inspection resolution that are associated with these techniques, contact type piezoelectric transducers were investigated instead.
The problem often associated with the contact type piezoelectric transducers at high temperature is their Curie temperature. When piezoelectric materials operate above their Curie temperature they cease to be piezoelectric. Many piezoelectric materials that possess high Curie temperatures have been reported. A common problem associated with materials that poses high Curie temperatures is their piezoelectric response, which is generally low in comparison to Lead Zirconate Titanate, Pb(Zr,Ti)O3 (PZT), ceramics which is the most widely used piezoelectric material for conventional ultrasound NDT transducers.
Work has been carried out by researchers on new materials such as aluminium nitride thin films and also new fabrication techniques such as sol-gel deposited piezoelectric films. However, Lithium Niobate, LiNbO3, is a commercially available piezoelectric material that has Curie temperature of 1210°C, and Baba et al have shown that it can work at temperatures up to 1000°C. Therefore, in this project Lithium Niobate (LN) was selected as the most suitable piezoelectric material for further investigation. As well as LN transducers existing PZT type transducers were also further studied and improvements were made to extend their operating temperatures and were shown to work successfully up to 250°C. Prototype LN transducers were manufactured and tested ultrasonically for temperatures up to 600°C. The LN type transducers were expensive and time consuming to manufacture due to the cost of components and the high temperature joining technique involved. Therefore, for the laboratory trials high temperature testing of the transducer collar array the collar array was populated with less expensive high temperature PZT type transducers.
A prototype ultrasonic pulser-receiver (UPR) able to drive the high temperature collar arrays was developed to drive the transducer collar array at ambient and elevated temperatures. In its current form, the prototype UPR system can drive up to three collars, but is of a modular design which can be extended to drive more collars by adding more units. A 3dB (22%) improvement in signal to noise ratio was obtained when the prototype UPR was used in combination with the prototype high temperature transducer collar array when compared to existing commercially available inspection hardware. The UPR was controlled via flexible windows based software, which can interface with (for example) MATLAB to allow pre- and post-processing techniques to be incorporated. The electronics were designed to withstand the environmental conditions present in power stations.
The UPR was integrated with the high temperature collar array and used to collect data for development of the signal processing routines and the pattern recognition algorithms. Laboratory experiments were carried out using an 8 inch (OD = 219.1mm and WT = 8.18mm) representative steel pipe, P91 type, with a weld. UGW testing was carried out at ambient and then during the rise in temperature. Once the system reached the 250°C target temperature it was left to stabilise and baseline that represented a defect-free condition at elevated temperature was collected. Subsequently, a defect was introduced in the Heat Affected Zone (HAZ) of the weld and data was collected again, then the defect size was gradually increased, and data was collected at each defect size. This data was then used to perform advanced signal processing in order to detect the defect and also to monitor the increase in defect size. Advanced signal processing techniques such as normalization, Hilbert transform, signal smoothing, correlation, feature extraction, selection and classification based on Support Vector Machines were used. The signals were also used for the task of crack detection using the correlation-based pre-processing analysis. The proposed flaw detection technique was able to (i) identify with 100% confidence probability if a signal belonged to the baseline or not and (ii) identify the severity of the recognized defect with an average accuracy of 93.84%. An efficient defect mapping tool was developed that could be utilized on-line to provide a very useful real-time representation of the structural health of the pipe and/or to estimate the distance (in meters) of the recognized defects from the measurement unit. Trend analysis was conducted on the acquired signals, providing the relation between the peak-to-peak amplitude value of the ultrasonic signals and the progression of the defect size. A user-friendly Graphical User Interface (GUI) was developed to integrate all the proposed technologies (pre-processing, feature extraction/selection and pattern recognition) and allow NDT personnel limited experience to be able to use and exploit the Hotscan hardware/software functionalities.
The integrated Hotscan system was then installed in an electrical power plant to assess its performance in service conditions. The site trial was carried out in an Iberdrola coal power plant in Spain. The system sub-components were tested prior to shipment. The system was installed during the outage period and left to automatically collected data five times per day and record the temperature at the same time. Room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed. Data was collected during the rise in pipe temperature up to 339°C, but the pipe temperature exceeded the specified operation temperature of the system and hence it stopped working.
The Hotscan project has enabled a number of SME’s together with experienced RTDs with extensive expertise of innovation and research to define and explore a novel concept. The Hotscan system has the potential of providing benefit to the participants, to the power and oil and Gas industry in terms of safety management, the EU in terms of new Jobs and society through improved safety. Further work needs to be carried out to realise a commercial system. Although initial field trials have proven the concept further site trials are necessary to convince plant operators. The consortium will maintain communications after the project, and will review and exploit all potential commercialisation opportunities.
Project Results:
The initial stage of the Hotscan project was concentrated in reviewing the latest developments in UGW inspection and condition monitoring systems for structural health monitoring of high temperature pipelines in power and oil and gas sector. In order to clarify the industrial requirements and focus on the key objectives of the project a questionnaire was prepared and circulated to SME and end-user consortium members. The feedback from the questionnaire was incorporated in the project specification document (deliverable D1.2). In this document the following key aspects of the Hotscan system were addressed in detail:
• Propagation of ultrasonic guided waves (UGW) in pipes at high temperatures
• Transducer development
• Transducer collar design
• Development of a pulser-receiver system for serving multiple transducer collars
• Signal processing for trend analysis and noise reduction
The consortium was in agreement that development of high temperature transducers was the most challenging aspect of the research. The ultimate aim of the Hotscan project was to develop transducers that could withstand temperatures up to 580°C in order to meet the requirement for in-situ condition monitoring of the ageing electrical power plants. However, transducers that could withstand temperatures up to 250°C were considered to be commercially attractive in the oil and gas industry. Owing to the significant risk of 580°C transducers not being feasible, a staged approach was adopted to prove some of the development strategies using the PZT transducers. Therefore, a parallel programme of work was adopted to develop transducers for operation at 250°C and 580°C.
A range of samples for transducer development and laboratory validation of the overall system were considered as part of deliverable D1.1. The sample used to evaluated the high temperature ultrasonic performance of transducers was a square (12mm2) steel rod. For laboratory testing of the Hotscan system a number of pipe sized were considered, but it was agreed that 6 and 8 inch pipes were most commonly used. Therefore, a 6 metre long, 8 inch (OD = 219.1mm and WT = 8.18mm) representative P91 steel (CrMoV) pipe sample with a weld was produced (Figure 1). A sample weld was produced using the P91 steel pipe. The entire sample weld was removed from the pipe and a section of it was polished and examined using optical microscopy to identify the Heat Affected Zone (HAZ) of the weld (Figure 2). In order to simulated a growing detect in the HAZ, which is where the defects normally initiate, a saw cut was introduced and its size was gradually increased. Other pipes with various types of defects such as circumferential slots, fatigue and corrosion were also identified.
Figure 1 Shows the P91 steel pipe with a weld.
Figure 2 Shows a) cross section of the weld and b) the length of the HAZ.
Change in temperature can expand and contract the pipes geometry and also influence the material properties. This can affect how ultrasound propagates on a solid media at different temperatures. Therefore, the effect of temperature on propagation of ultrasound in pipes was studied as part of WP2 and reported in D2.1. The main objectives were to study the feasibility of defect detection using UGW, selection of optimum wave modes and feasibility of detection of creep type defects in the HAZ of the weld. High temperature mechanical properties such as Young's modulus, density and wave speeds (both shear and bulk) from literature were used in the ABAQUS model. Wave propagation simulations were performed for a 6 and 8 inch pipe (Figure 3).
Figure 3 Shows an example of wave propagation simulations performed.
Dispersion curves were calculated for the 8 inch (OD = 219.1mm and WT = 8.18mm) steel pipe at different temperatures using high temperature material properties reported in literature (Figure 4). The general trend observed was a decrease in velocity of the wave modes at elevated temperatures. Transient analyses of defect detection were also performed at ambient and elevated temperatures. In one model a weld with HAZ was simulated to study the possibility of detecting creep. It was not possible to identify creep in the primary or secondary phase. Only using a significant variation of the Young Modulus (around 50%), a measurable echo could be reproduced representing the area of creep. Regarding wave mode selection, the optimum wave mode selected was the torsional T(0,1) wave mode due to its non-dispersive nature. This information was useful for design of the transducer collar array.
Figure 4 Shows the dispersion curves for the 8 inch pipe at different temperatures.
A prototype high temperature transducer collar array (Figure 5) was developed as part of WP3 and reported in deliverable D3.1. In this report the existing state-of-the art long range ultrasonic system was studied and the high temperature limitations of its subcomponents, especially the transducers, were reviewed. Alternative high temperature materials were selected, and transducers for for operation at 250˚C and 580˚C were developed (Figure 6). An experimental procedure was designed to test the prototype transducers that were developed. Initially the existing transducers were tested, and were shown to work near the Curie temperature of the piezoelectric material (350˚C), but with a reduced ultrasonic performance (Figure 7). The Ageing behaviour was also investigated at 250°C and was shown to work successfully for up to 35 days (Figure 8). The prototype transducers developed for operation at 250˚C were assembled in the same manner as the existing transducers, but high temperature adhesives were used instead. This resulted in improved ultrasonic performance at elevated temperatures. These transducer were able to withstand operation at 250˚C, and were simple and cost effective to manufacture. Prototype transducers for operation at 580˚C was designed, manufactured and tested and shown to work at these temperatures (Figure 9). These transducers were tested from 450°C to 580°C for up to 12 days (Figure 10). These transducers were manufactured using commercially available Lithium Niobate (LN) piezoelectric elements which were bonded to the backing block using a high temperature joining technique. Therefore, the LN type transducers were expensive and time consuming to manufacture. An initial prototype collar array to house the transducers was developed. Improved materials used in design of this collar were selected to withstand the target temperatures. The collar array was manufactured and demonstrated to work at ambient temperatures as well as the exiting collars. Additional work was carried out to characterise candidate piezoelectric materials at elevated temperatures via impedance measurements.
Figure 5 Prototype high temperature transducer collar array mounted on an 8 inch steel pipe.
Figure 6 Shows the PZT and LN type transducers designed for operation at a) 250°C and b) 580°C, respectively.
Figure 7 Shows average transmission and reception responses of ten HT and LT PZT transducers from RT to approximately 350°C.
Figure 8 Shows the transmission and reception quality of the transducer at 250°C for up to 35 days.
Figure 9 Shows the fastest arriving wave mode in (left) reception and (right) transmission mode for the LN transducer at 20°C and 600°C.
Figure 10 Effect of Ageing on the Transmission Quality of LN Transducer
A prototype ultrasonic pulser-receiver (UPR) able to drive multiple high temperature collar arrays was developed as part of WP4 and reported in deliverable D4.1. In its current form, the prototype system (Figure 11) can drive up to three collars, but is of a modular design which can be extended to drive more collars by adding more units. There is also the possibility of multiplexing additional channels if necessary. Improved signal to noise ratio can be achieved using signal averaging, increasing the gain and using filters. A 3dB (22%) improvement in signal to noise ratio was obtained when the prototype ultrasonic pulser-receiver was used in combination with the prototype high temperature transducer collar array when compared to existing commercially available inspection hardware. The UPR is controlled via flexible windows based software, which can interface with (for example) MATLAB to allow pre- and post-processing techniques developed elsewhere in the project to be incorporated. The electronics were designed to withstand the environmental conditions present in power stations.
Figure 11 Shows the prototype Hotscan UPR system.
In WP5 advanced signal processing routines were developed for continuous condition monitoring and trend analysis. There were two deliverables in this WP. In the first deliverable D5.1 software for signal pre-processing and signal focusing using Time-Reversal Focusing (TRF) and Time-Delay Focusing (TDF) techniques was developed. The implemented software was developed under the MATLAB programming environment. The developed software routines were integrated in a simple GUI (Figure 12) which allows signal visualization to validate the required functionality. In the second deliverable D5.2 advanced signal processing techniques were developed and integrated, such as normalization, Hilbert transform, signal smoothing, correlation, feature extraction, selection and classification based on Support Vector Machines. For the training and validation of the system, an extensive experimental investigation was carried to generate defect-free and defective signals to develop and test the signal processing algorithms, the experimental set-up can be seen in Figure 13. The results (Figure 14) show that the proposed method is able to effectively detect flaws under various conditions and experimental scenarios in laboratory conditions.
Figure 12 The Hotscan GUI.
Figure 13 Image showing the test set-up used for high temperature data collection.
Figure 14 Trend analysis based on the most informative features that were selected by the proposed feature selection approach
The main components of the Hotscan system such are the transducer collar array, pulser-receiver unit and the signal processing software which were integrated in WP6 and reported in deliverable D6.1. In this report it was shown that; the prototype transducers were able to function at 250°C and 580°C, the high temperature collar array was able to generate the desired wave modes and operate in identical manner to the commercial collar array system, a pulser-receiver that can drive multiple collars was produced and finally a pattern recognition and trend analysis software that could monitor crack propagation and display it in a GUI was also developed. In D6.2 report the laboratory testing and results of the Hotscan system in ambient and elevated temperatures were presented. The main subcomponents such as the transducers were individually tested at 250°C and 580°C. The high temperature collar array was populated with the high temperature PZT type transducers, and used with the integrated Hotscan system to perform ultrasonic guided wave testing. The guided wave testing was performed on an 8 inch (OD = 219.1mm and WT = 8.18mm) Steel (P91) pipe at up to 250°C in laboratory conditions, as shown in Figure 12.
The integrated Hotscan system was installed in an electrical power plant as part of WP7 and results of this trial was reported in deliverable D7.1. The site trial was carried out in an Iberdrola coal power plant in Spain (Figure 15). The system sub-components were tested prior to shipment and then again on site prior to installation (Figure 16). The system was installed during the outage period and left to automatically collected data five times per day and record the temperature at the same time (Figures 17 and 18). Room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed (Figure 19). Data was collected during the rise in temperature from 28°C to 339°C. The system stopped working when the pipe temperature reached 365°C, beyond the specified operating temperature of the transducers.
Figure 15 Location of the Hotscan site trials in an Iberdrola coal power plant located in Langreo, Austrias, Spain.
Figure 16 Testing and assembly of the collar array.
Figure 17 Shows a) the installation and b) the assembled collar on the pipe.
Figure 18 Installed Hotscan system automatically collecting data.
Figure 19 Weld reflection for all of the 27kHz signals collected at room temperature.
A detailed breakdown of the work performed and main results for each WP is shown in the following table.
WP no. Work progress
1 System functional design, hardware and software architecture to meet industrial requirements and provision of samples
The main objectives if this WP were to confirm the industrial requirements that the Hotscan system was intended to solve and to then use this information and compose a specification document which included the functional design and the architecture of the hardware and software components. Therefore, a questionnaire was prepared and circulated to SME and end-user consortium members. The feedback from the questionnaire highlighted the dimension, type of material and the temperatures if the pipes used as well as the types of defects that normally occur. This information was incorporated in the project specification document (deliverable D1.2). In this document the following key aspects of the Hotscan system were also addressed:
• Propagation of ultrasonic guided waves (UGW) in pipes at high temperatures - Developing high temperature transducers was very important, but this was not sufficient when performing LRUT inspection at high temperatures. In order to interpret the signals during inspection, understanding of propagation characteristics of UGWs is essential. When performing conventional LRUT on pipes, material properties such as Young’s modulus, poisons ratio and density, and pipe geometry (thickness and diameter) are important as they influence the velocity of the UGWs. As a result the use of various software such as Disperse and FEA modelling techniques were considered to produce dispersion curves for pipes at elevated temperatures.
• Transducer development - The consortium was in agreement that development of high temperature transducers was the most challenging aspect of the research. The ultimate aim of the Hotscan project was to develop transducers that could withstand temperatures up to 580°C in order to meet the requirement for in-situ condition monitoring of the ageing electrical power plants. However, transducers that could withstand temperatures up to 250°C were considered to be commercially attractive in the oil and gas industry. Therefore, a parallel programme of work was adopted to develop transducers for operation at 250°C and 580°C. The pros and cons of contact and non-contact type transducers were considered and it was decided to opt for piezoelectric contact type transducers. The components that make up the piezoelectric transducers such as the piezoelectric element, transducers housing/backing material, high temperature cables, wear plate and the possible joining techniques were discussed. Lithium Niobate was selected as the most suitable piezoelectric element for further investigation due to its high Curie temperature, which is reported to be approximately 1150°C for certain compositions
• Transducer collar design - Regarding the collar array, the modular collar design of existing LRUT system (Teletest®) was considered most appropriate to for generation of the desired wave modes. Functionality and components of the existing collar design were considered unsuitable for use at elevated temperatures as it's inflatable design meant that the bladder and modules were manufactured using polymers and composite materials that had low melting temperatures. Therefore, it was decided to manufacture a high temperature collar array using metal components.
• Development of a pulser-receiver system for serving multiple transducer collars - The pulser-receiver system was also based on the current Teletest© system, but iterative developments were to be carried out to build a system with enhanced signal to noise ratio capability. To collect data using the Hotscan pulser-receiver it was decided to use the standard Teletest© software (version 1). Additional graphical user interfaces (GUI) were to be designed specific to the Hotscan application to aid data collection, analysis and presentation.
• Signal processing for trend analysis and noise reduction - The Hotscan system was designed to continuously collecting data, which was then going to be processed using advanced signal processing routines and pattern recognition methods in MATLAB. Key features from the received signals were to be selected, extracted and classified. To build the software algorithms representative signal were required, hence an experimental procedure was designed to simulate defect-free and defective signals for an 8 inch pipe in laboratory conditions.
A range of samples for transducer development and laboratory validation of the overall system were considered as part of deliverable D1.1. The sample used to evaluated the high temperature ultrasonic performance of transducers was a 12mm2 steel rod. For laboratory testing of the Hotscan system a number of pipe sized were considered, but it was agreed that 6 and 8 inch pipes were most commonly used. Therefore, a 6 metre long, 8 inch (OD = 219.1mm and WT = 8.18mm) representative P91 steel (CrMoV) pipe sample with a weld was produced (Figure1). A sample weld was produced using the P91 steel pipe. The entire sample weld was removed from the pipe and a section of it was polished and examined using optical microscopy to identify the Heat Affected Zone (HAZ) of the weld (Figure 2). In order to simulated a growing detect in the HAZ, which is where the defects normally initiate, a saw cut was introduced and its size was gradually increased. Other pipes with various types of defects such as circumferential slots, fatigue and corrosion were also identified.
2 Modelling of guided wave propagation in pipes at normal and elevated temperatures for optimum mode selection
The main objectives were to study the feasibility of defect detection using UGW, selection of optimum wave modes and feasibility of detection of creep type defects in the HAZ of the weld. Temperature can expand and contract the pipes geometry and also influence the material properties. This can affect how ultrasound propagates on a solid media at different temperatures. Therefore, the effect of temperature on propagation of ultrasound in pipes was studied (Figure 3) as part of this WP and reported in D2.1.
Dispersion curves plot the phase speed of an elastic wave travelling through a specific waveguide against frequency. Dispersion curves are important for LRUT to determine the frequencies which are within a range where the wave modes become “dispersive”. This is represented in the dispersion curves by a high gradient for the specific frequency. Dispersion curves are used to select an optimal frequency for a given wave mode as less “dispersive” as possible which can then be used for selective excitation. Dispersion curves can be created by solving the analytical equations or by solving Eigen-Problems using numerical methods. PCdisp (Pochhammer-Chree dispersion), a small toolbox of MATLAB routines which uses a numerical simulation the generate propagation of ultrasonic waves in a cylindrical waveguide, was mainly used to produce the dispersion curves. Calculations of dispersion curves for three temperature ranges were carried out to allow for a better overview of the material's behaviour to ultrasonic waves due to the temperature change. To match available material data from open literature with the target temperatures of interest instead of 250°C and 580°C higher temperatures of 260oC and 595oC were used, respectively (Figure 4), and for ambient temperature 21°C was assumed. Dispersion curves were calculated for the 8 inch (OD = 219.1mm and WT = 8.18mm) steel pipe at these temperatures. The general trend observed was a decrease in velocity of the wave modes at elevated temperatures. For example for a frequency of 70 kHz the phase velocity decreases 20% when increasing the temperature from 21°C to 595°C.
Theoretical limits of defect detection using UGW to detect defects in pipes were calculated for L(0,1) and T(0,1) wave modes at different frequencies. it was concluded that even at 250kHz (upper range of UGW testing) it is theoretically not possible to detect a crack size of 1mm x 1mm in inspection mode.
Transient analyses of defect detection were also performed at ambient and elevated temperatures. Simulations that took into account only
the weld and not the HAZ were produced in Abaqus. Amplitude of the reflections from the crack, interaction of the reflected waves with the weld makes defect detection difficult when amplitude was used as a detection feature. As reference a crack of 10 degrees (of circumference = 19mm length) 1mm wide, 4.16mm deep (half thickness) was used. Simulations targeted to explore the effect variation of the geometrical characteristics of the crack (width, depth ) and the frequency of the ultrasonic waves would have to the response levels. It was shown that the reference defect is on the limits of detection using reflected amplitude as the detection feature. The modelling has shown that the targeted defect size of 1mm x 1mm as stated in the DoW cannot be detected.
Thereafter a feature was introduced to represent the reflection of a weld, and a crack was introduced in the immediate vicinity of the weld. The target was to identify the response (reflection) from the crack, despite the presence of the weld response. It has been demonstrated that under circumstances there is a distinguishable signal. Nevertheless, it has to be taken into account that field measurements will have additional noise (both due to electronics, as well as imperfections of the structure, material etc); those are general much higher than those due to numerical calculation. It was also mentioned that positive identification of defects can be enhanced by comparing data collected at previous inspections to current one to identify changes (using them as a possible flaw indication).
In one model a weld simulated to study the possibility of detecting creep. It was not possible to identify creep in the primary or secondary phase. Only using a significant variation of the Young Modulus (around 50%), a measurable echo could be reproduced representing the area of creep. Regarding wave mode selection, the optimum wave mode selected was the torsional T(0,1) wave mode due to its non-dispersive nature. This information was useful for design of the transducer collar array.
3 Modelling, design and prototyping of the high temperature LRU transducer collar array
The main objective of this WP was to develop a LRUT transducer collar array (Figure 5) that could be permanently air coupled to high temperature pipe work, temperatures up to 580°C and could also generate the desired wave modes selected in WP2. A prototype high temperature transducer collar array was developed as part of this WP and reported in deliverable D3.1. In this report the existing state-of-the art long range ultrasonic system was studied and the high temperature limitations of its subcomponents, especially the transducers, were reviewed. Alternative high temperature materials were selected, and transducers for operation at 250˚C and 580˚C were developed (Figure 6). Additional work was carried out to characterise candidate piezoelectric materials at elevated temperatures via impedance measurements.
Limitations of the existing transducers and their main sub-components were investigated. An experimental procedure was designed to test the prototype transducers that were developed. Initially the existing transducers were tested, and were shown to work near the Curie temperature of the piezoelectric material (350˚C), but with a reduced ultrasonic performance. The prototype transducers developed for operation at 250˚C were assembled in the same manner as the existing transducers, but high temperature adhesives were used instead. This resulted in improved ultrasonic performance at elevated temperatures. These transducer were able to withstand operation at 250˚C (Figure 7), and were simple and cost effective to manufacture. These transducers tested at 250°C for up to 35 days (Figure 8) and it was observed that their ultrasonic output increased significantly during the rise in temperature from ambient to 250°C and then stabilised.
Prototype transducers for operation at 580˚C was designed, manufactured and tested and shown to work successfully up to 600°C (Figure 9). The ageing behaviour of these transducers was tested and shown to work from 45°C to 580°C for 12 days (Figure 10). These transducers were manufactured using commercially available Lithium Niobate (LN) piezoelectric elements which were bonded to the backing block using a high temperature joining technique. To select the suitable material and size of the backing block modelling work was performed using ABAQUS. In this model a range of metals and ceramics were considered and eventually Aluminium oxide was selected as it had one of the closest matched acoustic impedances to LN as well as a fairly close value of CTE. The CTE was not a perfect match to LN, but it had a high Young’s Modulus which would allow the material to withstand greater stresses. Aluminium oxide is denser than steel, which along with the closer acoustic impedance made it a much more efficient dissipater of acoustic energy than steel. Aluminium oxide is a widely used ceramic and is cheaper compared to other advanced ceramics.
A new transducer collar array was designed to withstand high temperatures. The new collar array was manufactured using mainly stainless steel, which has a melting temperature of 1510˚C, and was designed to be fitted onto an 8 inch (OD = 219.1mm and WT = 8.18mm) steel pipe. The collar consists of two rings of transducers to enable operation in torsional mode. The key components in the collar are the frame and the modules. The frame is made up of two halves which are joined together using four clamp blocks. There are 24 equally spaced slots where the modules were mounted. There were 24 modules and each had two transducers which were aligned in the torsional orientation, and behind each transducer was a steel compression spring. The springs were calibrated to exert the required force when fitted to pipe, the springs were made from stainless steel (melting temperature of 1325-1530˚C) and were designed to withstand temperatures up to 250˚C with minimal change in the spring constant. For operation at higher temperatures springs made from other materials such as Tungsten could be used. To apply the force onto the pipe via the transducer would require unscrewing the cap screw above each transducer. This is a more time consuming method of coupling the transducers to the pipe compared with the inflatable collar, but it ensures that uniform pressure can be applied at the elevated temperatures, as polymeric materials are not required. The Ceramic block in the modules was designed to steer the transducers to their position and keep them at a fixed position. A ceramic material was used so that it could also act as a heat barrier to protect the cables and connectors.
The collar array was manufactured and demonstrated to work at ambient temperatures as well as the exiting collars. Initial trials were carried out in ambient conditions to compare the performance of the High Temperature (HT) transducer collar array with a commercially specified system. Both collars were used to perform UGW testing on a defect-free steel pipe. The collars were placed at the same location (2m from the pipe end), and the inspection was performed at 37kHz. The A-scans obtained for the two systems were compared. It could be seen that the high temperature transducer collar array performed as well as the commercial system, since the A-scan traces were almost identical. The prototype collar array was tested at elevated temperatures on an 8 inch pipe using the Hotscan system. The experimental set-up and dimensions of the pipe used can be seen in the deliverable report (D3.1).
According to the decision at the kick-off meeting the system was first tested at 250˚C. A script was used to automatically collect data in a fast manner at each temperature. This was carried out to avoid major temperature fluctuation during the data collection process. The experimental set-up can be seen in the deliverable report (D3.1). The collar was heated to the required temperatures using heating mats, which were placed on either side of the collar. The heating mats were covered with high temperature insulating materials to ensure the pipe reached the required temperatures faster. The heating mats and a thermocouple (resistance welded to the pipe) were connected a CooperHeat power supply and temperature regulator unit. Other thermocouples were used to verify the temperature of the pipe. The collar was then connected to the Hotscan pulser-receiver unit and operated using a laptop. The script written specifically for this data collection was used to collect the data. Data was collected at room temperature then in 50˚C steps between 50˚C and 250˚C. The results showed that the collar array was working up to 250°C. The pipe end signal amplitude decreases as temperature increases. This will make defect detection more difficult. However, when the signal to noise ratio was measured at the different temperatures it could be seen that there was not a significant decline in signal to noise ratio with respect to temperature. As a result it should still be possible to detect defects at elevated temperatures.
4 Development of prototype high power pulser-receiver for high signal attenuation conditions
The main objective if this WP was to develop a pulser-receiver unit (Figure 11) that could be permanently installed in power plants to drive the transducer collar array developed in WP3. A prototype ultrasonic pulser-receiver (UPR) able to drive multiple high temperature collar arrays was developed as part of this WP and reported in deliverable D4.1. Inside the UPR there are 24 transmission (Tx) and 8 reception (Rx) channels, which can be routed to a combination of transducers in one or more collar, as appropriate. The Rx channels are multiplexed sequentially between transducers as required to receive the transmitted signals, which is more cost effective than having 24 individual Rx channels. All of the channels are connected to the control unit which is connected to a Windows PC. The control unit is a FPGA chip that has been programmed to manage the Tx/Rx channels and to collect and communicate the data to the PC.
The UPR can be connected to mains power supply for constant use in condition monitoring mode, but it also has a rechargeable battery that can last up to eight hours. This has two benefits. Firstly, redundancy is added, as the battery acts as an uninterruptable power supply system such that the hardware will not fail or require resetting/maintenence in the event of a power failure. Secondly, it allows the UPR to function in inspection mode if required, making the device multifunctional and more widely useful.
The UPR will need to be permanently installed in power plants within 30m of the pipes under test. The UPR and the laptop running the software developed in WP5 will be housed inside a rack, which acts as a protective casing. The rack can be mounted on a wall or kept in a suitable location in the power plant. The rack can be locked and only authorised personnel can access the system to access the data and maintain the system. If necessary, fan trays can also be installed to protect the laptop and the UPR electronics from overheating. This will be assessed during the field trials and if necessary will be installed. As the software is Microsoft Windows, it could be controlled remotely if a suitable network infrastructure is available, but it is envisaged that the UPR will only ever be controlled ‘locally’ by the laptop within the cabinet.
In its current form, the prototype system can drive up to three collars, but is of a modular design which can be extended to drive more collars by adding more units. There is also the possibility of multiplexing additional channels if necessary. Improved signal to noise ratio can be achieved using signal averaging, increasing the gain and using filters. A 3dB (22%) improvement in signal to noise ratio was obtained when the prototype ultrasonic pulser-receiver was used in combination with the prototype high temperature transducer collar array when compared to existing commercially available inspection hardware. The UPR is controlled via flexible windows based software, which can interface with (for example) MATLAB to allow pre- and post-processing techniques developed elsewhere in the project to be incorporated. The electronics were designed to withstand the environmental conditions present in power stations.
5 Advanced signal processing techniques for continuous monitoring, trend analysis and crack image recognition
In this WP the main objective was to develop advanced signal processing routines for continuous condition monitoring and trend analysis. There were two deliverables in this WP. In the first deliverable D5.1 a software for signal pre-processing and signal focusing using Time-Reversal Focusing (TRF) and Time-Delay Focusing (TDF) techniques was developed. The implemented software was developed under the MATLAB programming environment. The developed software routines were integrated in a simple GUI (Figure 12) which allows signal visualization to validate the required functionality.
In the second deliverable D5.2 advanced signal processing techniques were developed and integrated, such as normalization, Hilbert transform, signal smoothing, correlation, feature extraction, selection and classification based on Support Vector Machines. For the training and validation of the system, an extensive experimental investigation was carried (Figure 13) to generate defect-free and defective signals to develop and test the signal processing algorithms. A novel defect detection system was developed based on a successful synergy of complementary advanced processing tools (Figure 14). The Hotscan system can be efficiency utilized as an integrated condition monitoring system based on a modular design, which enables the accurate and reliable evaluation of the structural condition of pipes. The technology provides information of the condition of the pipe checking for any degradation or changes over time. As the defects of interest tend to occur over sustained periods of time, the monitoring may only need to be conducted intermittently, perhaps weekly or monthly.
The validity of the system was tested over an extensive experimental setup. An overview of the detection capabilities of the proposed system is given in the following:
• Experiments were conducted at ambient temperatures for the task of crack detection using data from separate channels and different frequencies. The proposed flaw detection was capable to identify with 100% confidence probability if a testing signal belongs to the baseline or not.
• Trend analysis was conducted on the acquired signals at ambient temperature, providing the relation between the peak value of the ultrasonic signals and the progression of the defect size.
• Experiments were conducted at elevated temperatures for the task of crack detection using the correlation-based pre-processing analysis applied in all the channels. The proposed flaw detection was capable to (i) identify with 100% confidence probability if a testing signal belongs to the baseline or not and (ii) identify the severity of the recognized defect with an average accuracy of 93.84%.
• An efficient defect mapping tool was developed that can be utilized on-line (i) providing a very useful real-time representation of the structural health of the pipe and (ii) estimating the distance (in meters) of the recognized defects from the measurement unit.
• Trend analysis was also conducted on the acquired signals at ambient temperature providing a clear straight-forward relation between extracted informative features of the ultrasonic signals and the progression of the defect size.
• A user-friendly Graphical User Interface (GUI) was developed that integrates all the proposed technologies (pre-processing, feature extraction/selection and pattern recognition) and allows anybody with limited experience to be able to use and exploit the HOTSCAN hardware/software functionalities.
6 Prototype Hotscan integration and laboratory validation
The main objective of this WP was to integrate the hardware and software components of the Hotscan system. The main components of the Hotscan system such are the transducer collar array, pulser-receiver unit and the signal processing software which were integrated in this WP and reported in deliverable D6.1. Majority of the system components were developed, tested and reported in previous deliverables . Therefore, in this report the functionality of the overall Hotscan system was reported, and the main conclusions of the report were as follows:
• Transducers that can function at 250°C and 580°C have been developed and tested individually in laboratory conditions.
• High temperature collar array that can generate the desired wave modes and operate in almost identical manner to the commercial collar array has been developed and tested at ambient and up to 250°C.
• A pulser-receiver that can drive the transducer collar array has been developed and used to test the transducer collar array at ambient and high temperatures.
• Pattern recognition and trend analysis software that can monitor the defect propagation and display it in a GUI has been developed.
• The hardware and software components have been integrated.
In D6.2 results of the Laboratory trials carried out at ambient and high temperatures on the Hotscan system and its subcomponents were presented. The transducers developed for operation at 250°C using Lead Zirconate Titanate as the piezoelectric material, and transducers developed for operation at 580°C using Lithium Niobate, were tested at laboratory conditions from room temperature up to the target temperatures. The PZT type transducers were then used with the collar array to perform Ultrasonic Guided Wave testing on a pipe sample from ambient up to 250°C.
The main conclusions of this report were as follows:
• The PZT and LN type transducers have been tested in laboratory conditions from ambient up to 250°C and 580°C, respectively. Both type of transducers were able to transmit and receive UGW signals at the target temperatures.
• Due to cost of the manufacture of LN transducers the collar array was populated with the less expensive PZT type transducers.
• The integrated Hotscan system was tested in laboratory conditions using PZT type transducers from ambient up to 250°C.
• The effect of thermal cycling on the Hotscan system was studied for up to 4 thermal cycles at 250°C. A 20% drop in performance of the PZT transducers was observed after the first cycle, but the performance stabilised after the subsequent thermal cycles.
7 Demonstration
The main objective of this WP was to install the Hotscan system on a steam line in a power plant (Figure 15) and assess its performance during rise in temperature and then over time at elevated temperatures. The integrated Hotscan system was installed in an electrical power plant as part of this WP and results of this trial was reported in deliverable D7.1.
In previous deliverables (D6.1 and D6.2) the integrated Hotscan system was shown to work in ambient temperatures, and its performance was demonstrated for operation at elevated temperatures in laboratory conditions. It was shown that the system was able to detect defects at 250°C on a representative P91 pipe sample. Transducers that can operate at 580°C have been developed and shown to work in laboratory conditions. However, in this project it was not cost effective to manufacture 580°C transducers at the scale required for site trials. In addition it would have taken a significant amount of time to source the components, manufacture and test these transducers, which would have further delayed the project. At the project review meeting (Month 24) the case for performing site trials at 250°C instead of 580°C was presented to the European Commission’s independent project reviewer Dr Michael Scheerer. He was in agreement and recommended the field trials be carried out at ~ 250C. As a result the consortium tried to find a suitable power plant with pipes operating at around 250C. When selecting the most suitable location for the site trials the following criteria had to be considered, which further narrowed down the search for potential sites where the system could be installed.
• Installation during an outage for safety reasons.
• Outage taking place during lifetime of the project.
• Identifying an 8 inch pipe as a single prototype collar was manufactured.
• Pipe location had to be indoors to protect the pulser-receiver and rest of the system and also somewhere that did not require erection of additional scaffolding.
• Maximum pipe temperature of 250°C.
There were extensive discussions by the consortium members with various electrical power plant owners across Europe to enable the site trials to take place in their facilities. Dr Marko Budimir (INETEC) had been in contact with Krsko, a nuclear power plant in Slovenia. They were keen to assist with the site trials but they had very strict security checks which made it difficult. Marko subsequently spoke to Jertovec, a combined cycle power plant in Croatia. They were also keen to assist but they did not have any 8 inch pipes. A Turkish company called Tupras were approached by Dr Cem Selcuk (Brunel University). They wanted to be involved in the project and assist with the site trials, but their company policy required them to have a formal involvement in the project. This was discussed in a project review meeting (Month 18) and the consortium decided that Tupras could only be involved in an end user group. Abbas Mohimi (Brunel University) visited the EPR Ely, which is world’s largest straw fired power plant. They were interested in the Hotscan system, but their next planned outage was in July 2013, which was beyond the time scales of the Hotscan project. Rafael Delgado (Tecnitest) had been in contact with Iberdrola and they had a potential site suitable for the site trials. The Lada coal power plant, capacity of 513 MW, was going through an outage period, therefore the site trials were carried out at this location.
The system sub-components were tested prior to shipment and then again on site (Figure 16). Wall thickness measurements were performed at the location where the collar was going to be mounted using manual ultrasonics. The collar array components such as the transducers, wires and connectors were tested again and then assembled. The collar array was assembled in two halves and carefully fitted onto the pipe (Figure 17).
The assembly procedure was difficult due to the tight space available and required two people to perform this task. Mounting the existing collar whilst on scaffolding could be even more difficult. A more user friendly design is desirable for any future improvements. Once installed on the pipe the transducers were coupled by removing the screw caps from the back of the transducers. The transducers were then connected to the pulser-receiver using a tool lead, and a test was carried out to check that everything was functioning properly. A thermocouple (TC) was used to measure the ambient temperature, and four other TCs were placed around the pipe circumference at the cardinal points to measure the pipe temperature. The thermocouples were connected to the Pico temperature logger.
The wires and thermocouples were neatly assembled and the lagging material was placed either side of the collar and a high temperature resistant sheet was placed above the collar and secured using steel wires. A casing was placed near the installation and the pulser-receiver, laptop, Pico temperature logger, and an external hard drive were placed inside it. An extension lead was used to provide power to the pulser-receiver and the laptop. The Pico temperature logger was powered via the laptop. An automated data collection software was designed and installed in the laptop to collect ultrasonic and temperature measurements five times per day and store the data in the external hard drive. The software was set-up to generate the T(0,1) wave mode in pulse-echo mode from 20-200kHz frequencies at 1kHz steps, using a 10 cycle hann windowed input signal, and 32 averages. The software was tested twice and then left to collect data (Figure 18).
A second visit was arranged by Brunel University and Tecnitest on 11 July 2013 to inspect the system and collect the data by replacing the external hard drive and allowing the system to continue collecting data. The results showed that room temperature data were collected for up to 14 days. The data stability was good, and small variations due to change in pipe temperature were observed (Figure 19).
The pipe temperature was not expected to exceed 250°C according to the plant operators. However, the data collected during the rise in temperature from 28°C to 339°C showed otherwise. The system stopped working when the pipe temperature reached 365°C due to the depolarisation of the transducers. Had the pipe temperature not exceeded the specified target temperature of 250°C the Hotscan system would have carried on collecting data.
8 Dissemination and exploitation
The main objectives of this WP were to disseminate the technology developed in the Hotscan project as widely as possible and particularly the EU community via marketing materials, attending conferences and publishing journal papers. The dissemination and exploitation activities were captured in this WP and reported in the Final PUDF (deliverable D8.2). The report sets out the details of the types of dissemination activities undertaken during the Project lifetime. It also captures all dissemination and exploitation activities which have been undertaken by the Project partners. This document also sets out the aspirations of the SME Partners for the commercial exploitation of the project foreground knowledge. It provides a complete picture of all dissemination activities undertaken and, most importantly, provides information on the future route to exploitation that will benefit the SME’s, industry, society and the EU economy through new job creation.
During the life of the project 2 peer-review journals papers have been published (one is currently under the review process) and 7 conference papers have been presented, as detailed in Tables A1 and A2.
The Hotscan website has been updated during lifetime of the project.
9 Project management
A progress review was carried out at half way stage (Month 9) of the programme and the findings were reported. Following the review process a number of deliverable reports were revised and re-submitted and the DoW was amended following the discussions with the independent project reviewer. All deliverables and progress report including management, finance and technical details were submitted for review at the ninth month stage, and the subsequent deliverables were also completed and submitted.
Despite the difficulty and high development risks in the Hotscan project, the consortium has made good progress towards meeting the milestones set out in the project Description of work.
Potential Impact:
It is anticipated that the Hotscan system has the potential to significantly change the nature of structural integrity assessment of high temperature pipe work in oil refineries and electrical power plant by enabling in-situ ultrasonic guided wave inspection and condition monitoring to be carried out. It is likely to contribute in reduction of failures in high temperature pipe work, especially in the ageing power plants. This may also help avoid unplanned outages and therefore significantly reduce the overall maintenance cost.
Project concept and some of the results have been published in peer reviewed journals and presented in various conferences (See Tables 1 and 2, respectively) by project partners.
The prototype system requires further development before it can be introduced to the market. The Hotscan partners, in particular the SMEs will continue to disseminate information about the project to the potential end users after the project has finished to secure funding for further development, this with be carried in collaboration between the SMEs as agreed in the Hotscan project.
List of Websites:
www.hotscan.eu
Relevant contact details:
Dr Paul Jackson, Plant Integrity Ltd
Tel: +44 (0) 1223 893994
Fax: +44 (0) 1223 893994
E-mail: paul.jackson@plantintegrity.co.uk